1 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7 (vect_transform_stmt): ...here.
8 (vect_analyze_stmt): Use it instead of calling
9 vectorizable_live_operation directly.
11 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14 non-SIMT targets in acc vector loops.
16 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
18 * configure.ac: Add arc and check if assembler supports gdwarf2.
19 * configure: Regenerate.
21 2017-09-18 Richard Biener <rguenther@suse.de>
23 PR tree-optimization/82220
24 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
25 epilogue niters from the min_profitable_iters compute.
27 2017-09-18 Jakub Jelinek <jakub@redhat.com>
30 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
31 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
32 (ix86_init_pic_reg): Revert 2017-09-01 changes.
34 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
37 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
38 switching to a new text section.
40 2017-09-18 Richard Biener <rguenther@suse.de>
42 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
44 (build_alias_set): Reject aliases with no access function.
46 2017-09-18 Richard Biener <rguenther@suse.de>
48 PR tree-optimization/79622
49 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
51 (build_cross_bb_scalars_use): Likewise.
53 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
55 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
57 2017-09-18 Alan Modra <amodra@gmail.com>
60 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
61 stack_pointer_rtx for count 0. Update comments. Break up
64 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
66 config/i386/i386.c: (xlogue_layout::STUB_NAME_MAX_LEN): Increase to 20
68 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
69 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
70 or avx version of the stub.
72 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
75 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
76 compute the minimum stack alignment. Also update preferred stack
77 boundary for leaf functions.
79 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
81 PR tree-optimization/82228
82 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
85 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
87 * common/config/nds32/nds32-common.c
88 (nds32_option_optimization_table): Refine formatting.
89 (nds32_option_optimization_table): Use -fsched-pressure and
90 -fomit-frame-pointer for specific optimization level.
92 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
94 * config/nds32/nds32.c: Refine formatting and comments.
95 * config/nds32/nds32.h: Likewise.
96 * config/nds32/nds32.md: Likewise.
97 * config/nds32/nds32-cost.c: Likewise.
98 * config/nds32/nds32-isr.c: Likewise.
99 * config/nds32/nds32-md-auxiliary.c: Likewise.
100 * config/nds32/nds32-multiple.md: Likewise.
101 * config/nds32/nds32-predicates.c: Likewise.
103 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
104 Jakub Jelinek <jakub@redhat.com>
106 Add support for -std=c++2a.
107 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
109 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
111 2017-09-15 Steve Ellcey <sellcey@cavium.com>
114 * doc/extend.texi (Common Function Attributes): Add
115 references to ARM, AArch64, and S/390 specific attributes.
116 (Function Specific Option Pragmas): Add AArch64 and S/390
117 to list of back ends that support the target pragma.
119 2017-09-15 Nathan Sidwell <nathan@acm.org>
121 * doc/standards.texi: Fix C++17 description. Update URLs for
124 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
126 * common.opt (Wcast-align=strict): New warning option.
127 * doc/invoke.texi: Document -Wcast-align=strict.
129 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
131 * cgraph.h (cgraph_thunk_info): Add comments.
132 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
133 assert for VIRTUAL_* arguments stricter.
135 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
137 PR tree-optimization/71026
138 * match.pd: Move RDIV patterns from fold-const.c
139 * fold-const.c (distribute_real_division): Removed.
140 (fold_binary_loc): Remove calls to distribute_real_divison.
142 2017-09-15 Jakub Jelinek <jakub@redhat.com>
144 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
145 c++1z and gnu++1z as deprecated. Change other references to
146 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
147 Change -Wc++1z-compat to -Wc++17-compat.
148 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
149 * dwarf2out.c (highest_c_language): Handle C++17.
150 (gen_compile_unit_die): Likewise.
152 2017-09-15 Jakub Jelinek <jakub@redhat.com>
154 PR rtl-optimization/82192
155 * combine.c (make_extraction): Don't look through non-paradoxical
156 SUBREGs or TRUNCATE if pos + len is or might be bigger than
159 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
160 Alan Hayward <alan.hayward@arm.com>
161 David Sherwood <david.sherwood@arm.com>
163 * target.def (function_arg_offset): New hook.
164 * targhooks.h (default_function_arg_offset): Declare.
165 * targhooks.c (default_function_arg_offset): New function.
166 * function.c (locate_and_pad_parm): Use
167 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
168 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
169 (TARGET_FUNCTION_ARG_OFFSET): ...this.
170 * doc/tm.texi: Regenerate.
171 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
172 * config/spu/spu.c (spu_function_arg_offset): New function.
173 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
174 * system.h (FUNCTION_ARG_OFFSET): Poison.
176 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
177 Alan Hayard <alan.hayward@arm.com>
178 David Sherwood <david.sherwood@arm.com>
180 * target.def (truly_noop_truncation): New hook.
181 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
182 than TRULY_NOOP_TRUNCATION.
183 * hooks.h (hook_bool_uint_uint_true): Declare.
184 * hooks.c (hook_bool_uint_uint_true): New function.
185 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
186 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
187 * doc/tm.texi: Regenerate.
188 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
189 rather than TRULY_NOOP_TRUNCATION in comments.
190 (simplify_comparison): Likewise.
191 (record_truncated_value): Likewise.
192 * expmed.c (extract_bit_field_1): Likewise.
193 (extract_split_bit_field): Likewise.
194 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
195 instead of TRULY_NOOP_TRUNCATION.
196 * function.c (assign_parm_setup_block): Likewise.
197 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
198 * rtlhooks.c: Include target.h.
199 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
200 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
201 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
202 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
203 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
204 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
205 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
206 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
207 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
208 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
209 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
210 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
211 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
212 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
213 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
214 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
215 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
216 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
217 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
218 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
219 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
220 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
221 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
222 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
223 * config/mips/mips.c (mips_truly_noop_truncation): New function.
224 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
225 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
226 rather than TRULY_NOOP_TRUNCATION in comments.
227 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
228 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
229 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
230 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
231 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
232 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
233 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
234 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
235 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
236 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
237 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
238 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
239 rather than TRULY_NOOP_TRUNCATION in comments.
240 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
241 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
242 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
243 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
244 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
245 TRULY_NOOP_TRUNCATION condition.
246 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
247 (TRULY_NOOP_TRUNCATION): Delete.
248 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
249 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
250 * config/spu/spu.c (spu_truly_noop_truncation): New function.
251 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
252 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
253 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
254 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
255 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
256 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
257 rather than TRULY_NOOP_TRUNCATION in comments.
258 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
259 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
260 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
261 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
262 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
263 * system.h (TRULY_NOOP_TRUNCATION): Poison.
265 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
268 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
269 (*cmp_ior): Likewise.
270 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
271 (*ior_scc_scc_cmp): Likewise.
272 (*and_scc_scc): Likewise.
273 (*and_scc_scc_cmp): Likewise.
275 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
276 Alan Hayard <alan.hayward@arm.com>
277 David Sherwood <david.sherwood@arm.com>
279 * target.def (can_change_mode_class): New hook.
280 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
281 (hard_regno_nregs): Likewise.
282 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
283 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
284 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
285 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
286 (LOAD_EXTEND_OP): Update accordingly.
287 * doc/tm.texi: Regenerate.
288 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
289 CANNOT_CHANGE_MODE_CLASS.
290 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
291 (REG_CAN_CHANGE_MODE_P): ...this new macro.
292 * combine.c (simplify_set): Update accordingly.
293 * emit-rtl.c (validate_subreg): Likewise.
294 * recog.c (general_operand): Likewise.
295 * regcprop.c (mode_change_ok): Likewise.
296 * reload1.c (choose_reload_regs): Likewise.
297 (inherit_piecemeal_p): Likewise.
298 * rtlanal.c (simplify_subreg_regno): Likewise.
299 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
300 instead of CANNOT_CHANGE_MODE_CLASS.
301 (reload_cse_simplify_operands): Likewise.
302 * reload.c (push_reload): Use targetm.can_change_mode_class
303 instead of CANNOT_CHANGE_MODE_CLASS.
304 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
305 REG_CANNOT_CHANGE_MODE_P.
306 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
307 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
308 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
309 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
310 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
311 (arm_can_change_mode_class): New function.
312 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
313 than CANNOT_CHANGE_MODE_CLASS in comments.
314 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
315 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
316 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
317 (ix86_can_change_mode_class): ...this new function, inverting the
318 sense of the return value.
319 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
320 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
321 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
322 (ia64_can_change_mode_class): New function.
323 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
324 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
325 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
326 (m32c_can_change_mode_class): ...this new function, inverting the
327 sense of the return value.
328 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
329 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
330 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
331 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
332 (mips_can_change_mode_class): ...this new function, inverting the
333 sense of the return value.
334 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
335 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
336 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
337 (msp430_can_change_mode_class): New function.
338 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
339 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
340 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
341 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
342 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
343 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
344 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
345 (pa_cannot_change_mode_class): Replace with...
346 (pa_can_change_mode_class): ...this new function, inverting the
347 sense of the return value.
348 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
349 than CANNOT_CHANGE_MODE_CLASS in comments.
350 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
351 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
352 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
353 (pdp11_cannot_change_mode_class): Replace with...
354 (pdp11_can_change_mode_class): ...this new function, inverting the
355 sense of the return value.
356 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
357 * config/powerpcspe/powerpcspe-protos.h
358 (rs6000_cannot_change_mode_class_ptr): Delete.
359 * config/powerpcspe/powerpcspe.c
360 (rs6000_cannot_change_mode_class_ptr): Delete.
361 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
362 (rs6000_option_override_internal): Assign to
363 targetm.can_change_mode_class instead of
364 rs6000_cannot_change_mode_class_ptr.
365 (rs6000_cannot_change_mode_class): Replace with...
366 (rs6000_can_change_mode_class): ...this new function, inverting the
367 sense of the return value.
368 (rs6000_debug_cannot_change_mode_class): Replace with...
369 (rs6000_debug_can_change_mode_class): ...this new function.
370 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
371 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
372 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
373 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
374 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
376 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
377 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
378 (rs6000_option_override_internal): Assign to
379 targetm.can_change_mode_class instead of
380 rs6000_cannot_change_mode_class_ptr.
381 (rs6000_cannot_change_mode_class): Replace with...
382 (rs6000_can_change_mode_class): ...this new function, inverting the
383 sense of the return value.
384 (rs6000_debug_cannot_change_mode_class): Replace with...
385 (rs6000_debug_can_change_mode_class): ...this new function.
386 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
387 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
388 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
389 (s390_can_change_mode_class): ...this new function, inverting the
390 sense of the return value.
391 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
392 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
393 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
394 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
395 (sh_cannot_change_mode_class): Replace with...
396 (sh_can_change_mode_class): ...this new function, inverting the
397 sense of the return value.
398 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
399 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
400 (sparc_can_change_mode_class): New function.
401 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
402 * config/spu/spu.c (spu_can_change_mode_class): New function.
403 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
404 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
405 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
406 (visium_can_change_mode_class): New function.
407 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
409 2017-09-15 Richard Biener <rguenther@suse.de>
411 PR tree-optimization/82217
412 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
413 but not undefined case.
415 2017-09-15 Jakub Jelinek <jakub@redhat.com>
418 * postreload.c (reload_cse_simplify_operands): Skip
419 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
421 2017-09-15 Richard Biener <rguenther@suse.de>
423 PR tree-optimization/68823
424 * graphite-scop-detection.c (build_alias_set): If we have a
425 possible dependence check whether we can handle them by just
426 looking at the DRs DR_ACCESS_FNs.
427 (build_scops): If build_alias_set fails, fail the SCOP.
429 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
431 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
432 to support float128 built-in functions that require the ISA 3.0
434 (BU_FLOAT128_3_HW): Likewise.
435 (SQRTF128): Add support for the IEEE 128-bit square root and fma
439 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
440 support for built-in functions that need the ISA 3.0 IEEE 128-bit
441 floating point instructions.
442 (rs6000_invalid_builtin): Likewise.
443 (rs6000_builtin_mask_names): Likewise.
444 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
445 (RS6000_BTM_FLOAT128_HW): Likewise.
446 (RS6000_BTM_COMMON): Likewise.
447 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
449 * doc/extend.texi (RS/6000 built-in functions): Document the
450 IEEE 128-bit floating point square root and fused multiply-add
453 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
455 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
456 reg (r2) isn't in the set of registers defined in the prologue.
458 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
459 Alan Hayward <alan.hayward@arm.com>
460 David Sherwood <david.sherwood@arm.com>
462 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
463 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
464 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
465 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
466 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
468 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
469 max_vectorization_factor.
470 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
472 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
473 Alan Hayward <alan.hayward@arm.com>
474 David Sherwood <david.sherwood@arm.com>
476 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
477 (vect_worthwhile_without_simd_p): Declare.
478 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
479 (vectorizable_reduction): Use it.
480 * tree-vect-stmts.c (vectorizable_shift): Likewise.
481 (vectorizable_operation): Likewise.
483 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
484 Alan Hayward <alan.hayward@arm.com>
485 David Sherwood <david.sherwood@arm.com>
487 * tree-vectorizer.h (vect_get_num_copies): New function.
488 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
489 * tree-vect-loop.c (vectorizable_reduction): Likewise.
490 (vectorizable_induction): Likewise.
491 (vectorizable_live_operation): Likewise.
492 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
493 (vectorizable_bswap): Likewise.
494 (vectorizable_call): Likewise.
495 (vectorizable_conversion): Likewise.
496 (vectorizable_assignment): Likewise.
497 (vectorizable_shift): Likewise.
498 (vectorizable_operation): Likewise.
499 (vectorizable_store): Likewise.
500 (vectorizable_load): Likewise.
501 (vectorizable_condition): Likewise.
502 (vectorizable_comparison): Likewise.
503 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
505 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
506 Alan Hayward <alan.hayward@arm.com>
507 David Sherwood <david.sherwood@arm.com>
509 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
512 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
513 Alan Hayward <alan.hayward@arm.com>
514 David Sherwood <david.sherwood@arm.com>
516 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
517 an inline wrapper that provides a location.
518 (gimple_build_vector): Likewise.
519 * gimple-fold.c (gimple_build_vector_from_val): New function.
520 (gimple_build_vector): Likewise.
521 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
522 functions to build the initial value. Always return a gimple value.
523 (get_initial_defs_for_reduction): Likewise. Only compute
525 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
526 vect_init_vector on the results from get_initial_def(s)_for_reduction.
527 (vectorizable_induction): Use gimple_build_vector rather than
530 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
531 Alan Hayward <alan.hayward@arm.com>
532 David Sherwood <david.sherwood@arm.com>
534 * target.h (vec_perm_indices): New typedef.
535 (auto_vec_perm_indices): Likewise.
536 * optabs-query.h: Include target.h
537 (can_vec_perm_p): Take a vec_perm_indices *.
538 * optabs-query.c (can_vec_perm_p): Likewise.
539 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
540 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
541 * tree-vect-generic.c (lower_vec_perm): Likewise.
542 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
543 (vect_grouped_load_supported): Likewise.
544 (vect_shift_permute_load_chain): Likewise.
545 (vect_permute_store_chain): Use auto_vec_perm_indices.
546 (vect_permute_load_chain): Likewise.
547 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
548 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
549 Update uses of can_vec_perm_p.
550 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
551 mode with a number of elements. Take a vec_perm_indices *.
552 (vect_create_epilog_for_reduction): Update accordingly.
553 Use auto_vec_perm_indices.
554 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
555 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
556 (vect_transform_slp_perm_load): Likewise.
557 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
558 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
559 (vect_gen_perm_mask_checked): Likewise.
560 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
561 (vect_gen_perm_mask_checked): Likewise.
562 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
563 (vectorizable_store): Likewise.
564 (vectorizable_load): Likewise.
565 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
566 (vectorizable_bswap): Likewise.
568 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
569 Alan Hayward <alan.hayward@arm.com>
570 David Sherwood <david.sherwood@arm.com>
572 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
573 * tree.c (build_vector): Likewise.
574 (build_vector_from_ctor): Update accordingly.
575 (build_vector_from_val): Likewise.
576 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
577 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
578 * tree-vect-generic.c (add_rshift): Likewise.
579 (expand_vector_divmod): Likewise.
580 (optimize_vector_constructor): Likewise.
581 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
582 (vect_transform_slp_perm_load): Likewise.
583 (vect_schedule_slp_instance): Likewise.
584 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
585 (vectorizable_call): Likewise.
586 (vect_gen_perm_mask_any): Likewise. Add elements in order.
587 * expmed.c (make_tree): Likewise.
588 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
589 a vector passed to build_vector.
590 (fold_convert_const): Likewise.
591 (exact_inverse): Likewise.
592 (fold_ternary_loc): Likewise.
593 (fold_relational_const): Likewise.
594 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
595 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
596 (const_unop): Likewise. Store the reduction accumulator in a
597 variable rather than an array.
598 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
599 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
600 the new vector, rather than constructing it after the input arrays.
601 (native_interpret_vector): Use auto_vec<tree> when building
602 a vector passed to build_vector. Add elements in order.
603 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
604 auto_vec<tree> when building a vector passed to build_vector.
605 (vect_create_epilog_for_reduction): Likewise.
606 (vectorizable_induction): Likewise.
607 (get_initial_def_for_reduction): Likewise. Fix indentation of
609 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
611 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
612 passed to build_vector.
614 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
615 Alan Hayward <alan.hayward@arm.com>
616 David Sherwood <david.sherwood@arm.com>
618 * tree-core.h (tree_base::u): Add an "nelts" field.
619 (tree_vector): Use VECTOR_CST_NELTS as the length.
620 * tree.c (tree_size): Likewise.
621 (make_vector): Initialize VECTOR_CST_NELTS.
622 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
623 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
624 TYPE_VECTOR_SUBPARTS.
625 * expr.c (const_vector_mask_from_tree): Consistently use "units"
626 as the number of units, setting it from VECTOR_CST_NELTS.
627 (const_vector_from_tree): Likewise.
628 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
629 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
630 (fold_negate_expr_1): Likewise.
631 (fold_convert_const): Likewise.
632 (const_binop): Likewise. Differentiate the number of output and
634 (const_unop): Likewise.
635 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
636 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
637 in cases that did the opposite.
639 2017-09-14 Richard Biener <rguenther@suse.de>
641 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
644 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
646 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
648 2017-09-14 Jakub Jelinek <jakub@redhat.com>
651 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
652 if and where to split a bb, except for splitting before debug insn
653 sequences followed by non-label real insn. Delete debug insns
654 in between basic blocks.
656 * combine.c (make_compound_operation_int): Formatting fixes.
658 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
659 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
660 * config/netbsd.h (LINK_EH_SPEC): Likewise.
661 * config/sol2.h (LINK_EH_SPEC): Likewise.
662 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
663 * config/s390/linux.h (LINK_SPEC): Likewise.
664 * config/freebsd.h (LINK_EH_SPEC): Likewise.
665 * config/openbsd.h (LINK_EH_SPEC): Likewise.
666 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
667 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
668 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
669 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
670 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
671 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
673 2017-09-13 Jakub Jelinek <jakub@redhat.com>
675 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
677 (ENDFILE_LINUX_SPEC): Likewise.
678 (LINK_EH_SPEC): Likewise.
679 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
680 (LINK_OS_LINUX_SPEC32): Likewise.
681 (LINK_OS_LINUX_SPEC64): Likewise.
682 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
683 (LINK_OS_LINUX_SPEC): Likewise.
685 2017-09-13 Martin Liska <mliska@suse.cz>
688 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
689 CASE_HIGH is NULL_TREE.
691 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
692 Alan Hayward <alan.hayward@arm.com>
693 David Sherwood <david.sherwood@arm.com>
695 * target.def (secondary_memory_needed): New hook.
696 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
697 instead of SECONDARY_MEMORY_NEEDED.
698 (secondary_memory_needed_mode): Likewise.
699 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
700 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
701 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
702 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
703 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
704 * doc/tm.texi: Regenerate.
705 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
706 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
707 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
708 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
709 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
710 * config/i386/i386.c (inline_secondary_memory_needed): Put the
711 mode argument first and change the reg_class arguments to reg_class_t.
712 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
713 Make static. Update the call to inline_secondary_memory_needed.
714 (ix86_register_move_cost): Update the call to
715 inline_secondary_memory_needed.
716 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
717 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
719 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
720 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
722 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
723 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
724 * config/mips/mips.c (mips_secondary_memory_needed): Make static
725 and match hook interface. Add comment from mips.h.
726 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
727 * config/mmix/mmix.md (truncdfsf2): Refer to
728 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
730 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
731 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
732 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
733 (pa_secondary_memory_needed): New function.
734 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
735 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
736 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
737 (pdp11_secondary_memory_needed): Make static and match hook interface.
738 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
739 * config/powerpcspe/powerpcspe-protos.h
740 (rs6000_secondary_memory_needed_ptr): Delete.
741 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
743 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
744 (rs6000_option_override_internal): Assign to
745 targetm.secondary_memory_needed rather than
746 rs6000_secondary_memory_needed_ptr.
747 (rs6000_secondary_memory_needed): Match hook interface.
748 (rs6000_debug_secondary_memory_needed): Likewise.
749 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
750 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
751 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
752 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
753 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
754 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
756 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
757 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
758 (rs6000_option_override_internal): Assign to
759 targetm.secondary_memory_needed rather than
760 rs6000_secondary_memory_needed_ptr.
761 (rs6000_secondary_memory_needed): Match hook interface.
762 (rs6000_debug_secondary_memory_needed): Likewise.
763 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
764 * config/s390/s390.c (s390_secondary_memory_needed): New function.
765 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
766 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
767 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
768 (sparc_secondary_memory_needed): New function.
769 * lra-constraints.c (check_and_process_move): Refer to
770 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
772 (curr_insn_transform): Likewise.
773 (process_alt_operands): Use targetm.secondary_memory_needed
774 instead of TARGET_SECONDARY_MEMORY_NEEDED.
775 (check_secondary_memory_needed_p): Likewise.
776 (choose_split_class): Likewise.
777 * reload.c: Unconditionally include code that was previously
778 conditional on SECONDARY_MEMORY_NEEDED.
779 (push_secondary_reload): Use targetm.secondary_memory_needed
780 instead of TARGET_SECONDARY_MEMORY_NEEDED.
781 (push_reload): Likewise.
782 * reload1.c: Unconditionally include code that was previously
783 conditional on SECONDARY_MEMORY_NEEDED.
784 (choose_reload_regs): Use targetm.secondary_memory_needed
785 instead of TARGET_SECONDARY_MEMORY_NEEDED.
786 (gen_reload): Likewise.
787 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
789 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
790 Alan Hayward <alan.hayward@arm.com>
791 David Sherwood <david.sherwood@arm.com>
793 * target.def (secondary_memory_needed_mode): New hook:
794 * targhooks.c (default_secondary_memory_needed_mode): Declare.
795 * targhooks.h (default_secondary_memory_needed_mode): New function.
796 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
797 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
798 * doc/tm.texi: Regenerate.
799 * lra-constraints.c (check_and_process_move): Use
800 targetm.secondary_memory_needed_mode instead of
801 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
802 (curr_insn_transform): Likewise.
803 * reload.c (get_secondary_mem): Likewise.
804 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
805 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
807 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
808 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
809 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
810 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
811 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
813 * config/powerpcspe/powerpcspe-protos.h
814 (rs6000_secondary_memory_needed_mode): Delete.
815 * config/powerpcspe/powerpcspe.c
816 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
817 (rs6000_secondary_memory_needed_mode): Make static.
818 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
819 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
821 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
823 (rs6000_secondary_memory_needed_mode): Make static.
824 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
825 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
826 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
827 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
828 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
830 (sparc_secondary_memory_needed_mode): New function.
831 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
833 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
835 * config/aarch64/constraints.md (Umq): New constraint.
836 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
838 (mov<mode>): Update condition.
840 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
842 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
843 when bitposition is the same.
845 2017-09-13 Richard Biener <rguenther@suse.de>
847 * dwarf2out.c (output_die_symbol): Remove.
848 (output_die): Do not output a DIEs symbol.
850 2017-09-13 Richard Biener <rguenther@suse.de>
853 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
854 default-def to avoid breaking iterator update with the weird
855 interaction with cgraph_update_edges_for_call_stmt_node.
857 2017-09-13 Richard Biener <rguenther@suse.de>
859 * tree-cfg.c (verify_gimple_assign_binary): Add verification
860 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
861 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
862 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
864 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
866 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
867 Disable pc relative literal load irrespective of
868 TARGET_FIX_ERR_A53_84341 for default.
870 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
872 * config/sparc/sparc.c (output_return): Output the source location of
873 the insn in the delay slot, if any.
874 (output_sibcall): Likewise.
876 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
879 * common.opt (-static-pie): New alias.
880 (shared): Negate static-pie.
881 (-no-pie): Update help text.
883 (static-pie): New option.
884 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
886 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
887 (LINK_EH_SPEC): Likewise.
888 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
889 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
890 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
891 * gcc.c (LINK_COMMAND_SPEC): Likewise.
892 (init_gcc_specs): Likewise.
893 (init_spec): Likewise.
894 (display_help): Update help message for -pie.
895 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
898 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
900 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
901 (movdi_aarch64): Likewise.
902 (movti_aarch64): Likewise.
904 2017-09-12 Simon Wright <simon@pushface.org>
907 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
908 calculation of the minor version, always output as 0.
910 2017-09-12 Jakub Jelinek <jakub@redhat.com>
913 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
914 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
915 on it early, rather than manual conversion late. For
916 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
917 instead of performing manual conversion.
919 2017-09-12 Carl Love <cel@us.ibm.com>
921 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
922 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
924 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
925 VMULOSW): Add definitions.
926 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
927 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
928 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
929 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
930 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
932 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
934 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
936 (movti_aarch64): Likewise.
937 (movdf_aarch64): Likewise.
938 (movtf_aarch64): Likewise.
939 (load_pairdi): Likewise.
940 (store_pairdi): Likewise.
941 (load_pairdf): Likewise.
942 (store_pairdf): Likewise.
943 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
944 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
945 (ldr_got_small_<mode>): Likewise.
946 (ldr_got_small_28k_<mode>): Likewise.
947 (ldr_got_tiny): Likewise.
948 * config/aarch64/iterators.md (ldst_sz): New.
949 (ldpstp_sz): Likewise.
950 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
952 (thunderx_load): Split load_8 to load_16.
953 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
955 (thunderx2t99_storepair_basic): Split store_8 to store_16.
956 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
957 (xgene1_store_pair): Split store_8 to store_16.
958 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
959 (falkor_st_0_st_sd): Split store_8 to store_16.
961 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
963 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
964 and store1/2/3/4 to store_4/8/12/16.
965 * config/aarch64/aarch64.md: Update for rename.
966 * config/arm/arm.md: Likewise.: Likewise.
967 * config/arm/arm.c: Likewise.
968 * config/arm/thumb1.md: Likewise.
969 * config/arm/thumb2.md: Likewise.
970 * config/arm/vfp.md: Likewise.
971 * config/arm/arm-generic.md: Likewise.
972 * config/arm/arm1020e.md: Likewise.
973 * config/arm/arm1026ejs.md: Likewise.
974 * config/arm/arm1136jfs.md: Likewise.
975 * config/arm/arm926ejs.md: Likewise.
976 * config/arm/cortex-a15.md: Likewise.
977 * config/arm/cortex-a17.md: Likewise.
978 * config/arm/cortex-a5.md: Likewise.
979 * config/arm/cortex-a53.md: Likewise.
980 * config/arm/cortex-a57.md: Likewise.
981 * config/arm/cortex-a7.md: Likewise.
982 * config/arm/cortex-a8.md: Likewise.
983 * config/arm/cortex-a9.md: Likewise.
984 * config/arm/cortex-m4.md: Likewise.
985 * config/arm/cortex-m7.md: Likewise.
986 * config/arm/cortex-r4.md: Likewise.
987 * config/arm/exynos-m1.md: Likewise.
988 * config/arm/fa526.md: Likewise.
989 * config/arm/fa606te.md: Likewise.
990 * config/arm/fa626te.md: Likewise.
991 * config/arm/fa726te.md: Likewise.
992 * config/arm/fmp626.md: Likewise.
993 * config/arm/iwmmxt.md: Likewise.
994 * config/arm/ldmstm.md: Likewise.
995 * config/arm/marvell-pj4.md: Likewise.
996 * config/arm/xgene1.md: Likewise.
997 * config/aarch64/thunderx.md: Likewise.
998 * config/aarch64/thunderx2t99.md: Likewise.
999 * config/aarch64/falkor.md: Likewise.
1001 2017-09-12 Martin Liska <mliska@suse.cz>
1003 * attribs.c (private_lookup_attribute): New function.
1004 * attribs.h (private_lookup_attribute): Declared here.
1005 (lookup_attribute): Called from this place.
1007 2017-09-12 Richard Biener <rguenther@suse.de>
1009 PR tree-optimization/82157
1010 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
1011 stmts with side-effects.
1013 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1014 Alan Hayward <alan.hayward@arm.com>
1015 David Sherwood <david.sherwood@arm.com>
1017 * target.def (hard_regno_nregs): New hook.
1018 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
1019 * targhooks.h (default_hard_regno_nregs): Declare.
1020 * targhooks.c (default_hard_regno_nregs): New function.
1021 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
1022 (TARGET_HARD_REGNO_NREGS): ...this hook.
1023 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
1024 (CLASS_MAX_NREGS): Likewise.
1025 * doc/tm.texi: Regenerate.
1026 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
1027 instead of HARD_REGNO_NREGS.
1028 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
1029 HARD_REGNO_NREGS in the comment.
1030 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
1031 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
1032 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
1033 Return an unsigned int.
1034 (TARGET_HARD_REGNO_NREGS): Redefine.
1035 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
1036 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
1037 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1038 (arc_hard_regno_nregs): New function.
1039 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
1040 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
1041 (arm_hard_regno_nregs): New function.
1042 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
1043 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
1044 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
1045 (TARGET_HARD_REGNO_NREGS): Redefine.
1046 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
1047 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
1048 (HARD_REGNO_NREGS): Delete.
1049 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
1050 (cr16_hard_regno_nregs): New function.
1051 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
1052 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
1053 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
1054 (cris_hard_regno_nregs): New function.
1055 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
1056 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
1057 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1058 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
1059 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
1060 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
1061 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
1062 (frv_hard_regno_nregs): Make static. Take and return an
1064 (frv_class_max_nregs): Remove outdated copy of documentation.
1065 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
1066 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
1067 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
1068 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
1069 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
1070 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
1071 (TARGET_HARD_REGNO_NREGS): Redefine.
1072 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
1073 (CLASS_MAX_NREGS): Update comment.
1074 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
1075 (ia64_hard_regno_nregs): New function.
1076 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
1077 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
1078 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
1079 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
1080 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
1082 (m32c_hard_regno_nregs): Likewise. Make static.
1083 (TARGET_HARD_REGNO_NREGS): Redefine.
1084 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
1085 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
1086 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
1087 (m68k_hard_regno_nregs): New function.
1088 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
1089 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
1090 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
1091 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
1092 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
1093 Take and return an unsigned int.
1094 (TARGET_HARD_REGNO_NREGS): Redefine.
1095 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
1096 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1097 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
1098 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
1099 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
1100 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
1101 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
1102 (msp430_hard_regno_nregs): Make static. Take and return an
1104 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
1105 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
1106 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
1107 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
1108 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
1109 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
1110 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
1111 (TARGET_HARD_REGNO_NREGS): Redefine.
1112 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
1113 (PA_HARD_REGNO_NREGS): ...this.
1114 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
1115 (PA_HARD_REGNO_NREGS): ...this.
1116 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1117 (pa_hard_regno_nregs): New function.
1118 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
1119 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
1120 (pdp11_hard_regno_nregs): New function.
1121 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
1122 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
1123 (rs6000_hard_regno_nregs_hook): New function.
1124 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
1125 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1126 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
1127 Take and return an unsigned int. Move earlier in file.
1128 (TARGET_HARD_REGNO_NREGS): Redefine.
1129 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
1130 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
1131 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
1132 (rl78_hard_regno_nregs): Make static. Take and return an
1134 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
1135 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
1136 (rs6000_hard_regno_nregs_hook): New function.
1137 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
1138 * config/rx/rx.c (rx_hard_regno_nregs): New function.
1139 (TARGET_HARD_REGNO_NREGS): Redefine.
1140 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
1141 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
1142 instead of HARD_REGNO_NREGS.
1143 (s390_hard_regno_nregs): New function.
1144 (s390_hard_regno_mode_ok): Add comment from s390.h.
1145 (TARGET_HARD_REGNO_NREGS): Redefine.
1146 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
1147 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
1148 (sh_hard_regno_nregs): New function.
1149 (sh_pass_in_reg_p): Use it.
1150 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
1151 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1152 (sparc_hard_regno_nregs): New function.
1153 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
1154 * config/spu/spu.c (spu_hard_regno_nregs): New function.
1155 (spu_function_arg_advance): Use it, supplying a valid register number.
1156 (TARGET_HARD_REGNO_NREGS): Redefine.
1157 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
1158 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
1159 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
1160 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
1161 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
1162 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
1163 (CLASS_MAX_NREGS): Remove copy of old documentation.
1164 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
1165 (visium_hard_regno_nregs): New function.
1166 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
1167 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
1168 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1169 xtensa_hard_regno_nregs): New function.
1170 * system.h (HARD_REGNO_NREGS): Poison.
1172 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1174 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
1175 hard_regno_nregs instead of HARD_REGNO_NREGS.
1176 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
1177 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
1178 (c6x_expand_epilogue): Likewise.
1179 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
1180 (frv_read_iacc_argument): Likewise.
1181 * config/sh/sh.c: Include regs.h.
1182 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1183 (regs_used): Likewise.
1184 (output_stack_adjust): Likewise.
1185 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
1186 * expmed.c: Include regs.h.
1187 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1188 * ree.c: Include regs.h.
1189 (combine_reaching_defs): Use hard_regno_nregs instead of
1191 (add_removable_extension): Likewise.
1193 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1195 * regs.h (hard_regno_nregs): Turn into a function.
1196 (end_hard_regno): Update accordingly.
1197 * caller-save.c (setup_save_areas): Likewise.
1198 (save_call_clobbered_regs): Likewise.
1199 (replace_reg_with_saved_mem): Likewise.
1200 (insert_restore): Likewise.
1201 (insert_save): Likewise.
1202 * combine.c (can_change_dest_mode): Likewise.
1203 (move_deaths): Likewise.
1204 (distribute_notes): Likewise.
1205 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
1206 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
1207 (rs6000_split_multireg_move): Likewise.
1208 (rs6000_register_move_cost): Likewise.
1209 (rs6000_memory_move_cost): Likewise.
1210 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
1211 (rs6000_split_multireg_move): Likewise.
1212 (rs6000_register_move_cost): Likewise.
1213 (rs6000_memory_move_cost): Likewise.
1214 * cselib.c (cselib_reset_table): Likewise.
1215 (cselib_lookup_1): Likewise.
1216 * emit-rtl.c (set_mode_and_regno): Likewise.
1217 * function.c (aggregate_value_p): Likewise.
1218 * ira-color.c (setup_profitable_hard_regs): Likewise.
1219 (check_hard_reg_p): Likewise.
1220 (calculate_saved_nregs): Likewise.
1221 (assign_hard_reg): Likewise.
1222 (improve_allocation): Likewise.
1223 (calculate_spill_cost): Likewise.
1224 * ira-emit.c (modify_move_list): Likewise.
1225 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
1226 (ira_hard_reg_in_set_p): Likewise.
1227 * ira.c (setup_reg_mode_hard_regset): Likewise.
1228 (clarify_prohibited_class_mode_regs): Likewise.
1229 (check_allocation): Likewise.
1230 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1231 (lra_setup_reg_renumber): Likewise.
1232 (setup_try_hard_regno_pseudos): Likewise.
1233 (spill_for): Likewise.
1234 (assign_hard_regno): Likewise.
1235 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1236 * lra-constraints.c (in_class_p): Likewise.
1237 (lra_constraint_offset): Likewise.
1238 (simplify_operand_subreg): Likewise.
1239 (lra_constraints): Likewise.
1240 (split_reg): Likewise.
1241 (split_if_necessary): Likewise.
1242 (invariant_p): Likewise.
1243 (inherit_in_ebb): Likewise.
1244 * lra-lives.c (process_bb_lives): Likewise.
1245 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
1246 (get_hard_regs): Likewise.
1247 (do_remat): Likewise.
1248 * lra-spills.c (assign_spill_hard_regs): Likewise.
1249 * mode-switching.c (create_pre_exit): Likewise.
1250 * postreload.c (reload_combine_recognize_pattern): Likewise.
1251 * recog.c (peep2_find_free_register): Likewise.
1252 * regcprop.c (kill_value_regno): Likewise.
1253 (set_value_regno): Likewise.
1254 (copy_value): Likewise.
1255 (maybe_mode_change): Likewise.
1256 (find_oldest_value_reg): Likewise.
1257 (copyprop_hardreg_forward_1): Likewise.
1258 * regrename.c (check_new_reg_p): Likewise.
1259 (regrename_do_replace): Likewise.
1260 * reload.c (push_reload): Likewise.
1261 (combine_reloads): Likewise.
1262 (find_dummy_reload): Likewise.
1263 (operands_match_p): Likewise.
1264 (find_reloads): Likewise.
1265 (find_equiv_reg): Likewise.
1266 (reload_adjust_reg_for_mode): Likewise.
1267 * reload1.c (count_pseudo): Likewise.
1268 (count_spilled_pseudo): Likewise.
1269 (find_reg): Likewise.
1270 (clear_reload_reg_in_use): Likewise.
1271 (free_for_value_p): Likewise.
1272 (allocate_reload_reg): Likewise.
1273 (choose_reload_regs): Likewise.
1274 (reload_adjust_reg_for_temp): Likewise.
1275 (emit_reload_insns): Likewise.
1276 (delete_output_reload): Likewise.
1277 * rtlanal.c (subreg_get_info): Likewise.
1278 * sched-deps.c (sched_analyze_reg): Likewise.
1279 * sel-sched.c (init_regs_for_mode): Likewise.
1280 (mark_unavailable_hard_regs): Likewise.
1281 (choose_best_reg_1): Likewise.
1282 (verify_target_availability): Likewise.
1283 * valtrack.c (dead_debug_insert_temp): Likewise.
1284 * var-tracking.c (track_loc_p): Likewise.
1285 (emit_note_insn_var_location): Likewise.
1286 * varasm.c (make_decl_rtl): Likewise.
1287 * reginfo.c (choose_hard_reg_mode): Likewise.
1288 (init_reg_modes_target): Refer directly to
1289 this_target_regs->x_hard_regno_nregs.
1291 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1293 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
1294 instead of hard_regno_nregs.
1296 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1298 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
1299 end_hard_regno instead of hard_regno_nregs.
1300 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
1301 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
1302 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
1303 * ira-color.c (improve_allocation): Likewise.
1304 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1305 * lra-lives.c (mark_regno_live): Likewise.
1306 (mark_regno_dead): Likewise.
1307 * lra-remat.c (operand_to_remat): Likewise.
1308 * lra.c (collect_non_operand_hard_regs): Likewise.
1309 * postreload.c (reload_combine_note_store): Likewise.
1310 (move2add_valid_value_p): Likewise.
1311 * reload.c (regno_clobbered_p): Likewise.
1313 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1315 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
1317 * config/v850/v850.c (v850_reorg): Likewise.
1318 * reload.c (refers_to_regno_for_reload_p): Likewise.
1319 (find_equiv_reg): Likewise.
1320 * reload1.c (reload_reg_reaches_end_p): Likewise.
1322 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1324 * caller-save.c (add_used_regs): Use REG_NREGS instead of
1326 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
1327 * config/arm/arm.c (output_move_neon): Likewise.
1328 (arm_attr_length_move_neon): Likewise.
1329 (neon_split_vcombine): Likewise.
1330 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
1331 (c6x_mark_reg_written): Likewise.
1332 (c6x_dwarf_register_span): Likewise.
1333 * config/i386/i386.c (ix86_save_reg): Likewise.
1334 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
1335 (rws_access_reg): Likewise.
1336 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
1337 * mode-switching.c (create_pre_exit): Likewise.
1338 * ree.c (combine_reaching_defs): Likewise.
1339 (add_removable_extension): Likewise.
1340 * regcprop.c (find_oldest_value_reg): Likewise.
1341 (copyprop_hardreg_forward_1): Likewise.
1342 * reload.c (reload_inner_reg_of_subreg): Likewise.
1343 (push_reload): Likewise.
1344 (combine_reloads): Likewise.
1345 (find_dummy_reload): Likewise.
1346 (reload_adjust_reg_for_mode): Likewise.
1347 * reload1.c (find_reload_regs): Likewise.
1348 (forget_old_reloads_1): Likewise.
1349 (reload_reg_free_for_value_p): Likewise.
1350 (reload_adjust_reg_for_temp): Likewise.
1351 (emit_reload_insns): Likewise.
1352 (delete_output_reload): Likewise.
1353 * sel-sched.c (choose_best_reg_1): Likewise.
1354 (choose_best_pseudo_reg): Likewise.
1356 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1357 Alan Hayward <alan.hayward@arm.com>
1358 David Sherwood <david.sherwood@arm.com>
1360 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
1361 * target.def (slow_unaligned_access): New hook.
1362 * targhooks.h (default_slow_unaligned_access): Declare.
1363 * targhooks.c (default_slow_unaligned_access): New function.
1364 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
1365 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
1366 * doc/tm.texi: Regenerate.
1367 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
1368 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
1369 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
1371 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
1372 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
1374 (rs6000_slow_unaligned_access): New function.
1375 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1376 (expand_block_compare): Likewise.
1377 (expand_strn_compare): Likewise.
1378 (rs6000_rtx_costs): Likewise.
1379 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
1380 (riscv_slow_unaligned_access): Likewise.
1381 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
1382 (riscv_slow_unaligned_access_p): ...this and make static.
1383 (riscv_option_override): Update accordingly.
1384 (riscv_slow_unaligned_access): New function.
1385 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1386 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
1387 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1388 (rs6000_slow_unaligned_access): New function.
1389 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1390 (rs6000_rtx_costs): Likewise.
1391 * config/rs6000/rs6000-string.c (expand_block_compare)
1392 (expand_strn_compare): Use targetm.slow_unaligned_access instead
1393 of SLOW_UNALIGNED_ACCESS.
1394 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
1395 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
1396 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
1397 of SLOW_UNALIGNED_ACCESS.
1398 * expmed.c (simple_mem_bitfield_p): Likewise.
1399 * expr.c (alignment_for_piecewise_move): Likewise.
1400 (emit_group_load_1): Likewise.
1401 (emit_group_store): Likewise.
1402 (copy_blkmode_from_reg): Likewise.
1403 (emit_push_insn): Likewise.
1404 (expand_assignment): Likewise.
1405 (store_field): Likewise.
1406 (expand_expr_real_1): Likewise.
1407 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1408 * lra-constraints.c (simplify_operand_subreg): Likewise.
1409 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1410 * gimple-ssa-store-merging.c: Likewise in block comment at start
1412 * tree-ssa-strlen.c: Include target.h.
1413 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
1414 of SLOW_UNALIGNED_ACCESS.
1415 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
1417 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1419 PR rtl-optimization/82185
1420 * expmed.c (emit_store_flag_int): Only test tem if it has been
1423 2017-09-12 Richard Biener <rguenther@suse.de>
1426 * match.pd ((FTYPE) N CMP CST): Fix typo.
1428 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1430 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1432 (mips_near_type_p): Add 'short_call' attribute as a synonym
1434 * doc/extend.texi (short_call): Document new function attribute.
1436 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1439 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1440 assertion check that in the condition.
1441 (get_atomic_generic_size): Likewise. Before testing if parameter
1442 has pointer type, if it has array type, call for C++
1443 default_conversion to perform array-to-pointer conversion.
1445 2017-09-12 Richard Biener <rguenther@suse.de>
1447 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1448 for operations we cannot scalarize.
1450 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1452 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1453 vectors heap vectors. Clean up comments.
1454 Make visited_bbs a reference.
1455 (profitable_jump_thread_path): Make GC
1456 vectors heap vectors. Clean up comments.
1458 (convert_and_register_jump_thread_path): Make GC vectors heap
1460 (check_subpath_and_update_thread_path): Same. Clean up comments.
1461 Make visited_bbs a reference.
1462 (handle_phi): Abstract common code to to
1463 register_jump_thread_path_if_profitable.
1464 Rename VAR_BB to DEF_BB.
1466 Make GC vectors heap vectors.
1467 Make visited_bbs a reference.
1468 (handle_assignment): Same.
1469 (register_jump_thread_path_if_profitable): New.
1470 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1472 Make GC vectors heap vectors. Clean up comments.
1473 Make visited_bbs a reference.
1474 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1475 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1478 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1481 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1482 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1484 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1486 Revert r251800 and r251799.
1488 2017-09-11 Martin Jambor <mjambor@suse.cz>
1491 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1492 arguments in advance.
1493 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1494 use it to find predecessor edges.
1495 (naive_outof_ssa): Collect vector of predecessors.
1497 2017-09-08 Jason Merrill <jason@redhat.com>
1499 PR c++/70029 - ICE with ref-qualifier and -flto
1500 * langhooks.h (struct lang_hooks_for_types): Add
1501 copy_lang_qualifiers.
1502 * attribs.c (build_type_attribute_qual_variant): Use it.
1503 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1505 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1506 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1508 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1511 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1512 (*mulsi3_sp64): New instruction.
1513 (mulsi3): New expander.
1515 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1517 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1519 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1521 * sancov.c: Include memmodel.h.
1523 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1526 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1529 2017-09-07 Carl Love <cel@us.ibm.com>
1531 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1532 the sldi instruction.
1534 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1536 * sancov.c: Include tm_p.h.
1538 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1541 * output.h (switch_to_other_text_partition): New declaration.
1542 * varasm.c (switch_to_other_text_partition): New function.
1543 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1544 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1545 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1546 to the other text partition before emitting LCL label and switch back
1547 after emitting the word after it.
1549 2017-09-07 Richard Biener <rguenther@suse.de>
1551 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1552 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1553 critical edges here, after loop init.
1554 (pass_data_pre): Remove PROP_no_crit_edges flags.
1555 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1556 for valueization of call args to avoid leaking VN_TOP.
1557 (visit_use): Assert we do not visit default defs.
1558 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1559 Use error_mark_node to more easily detect leaking VN_TOP.
1560 All default-defs are varying, not VN_TOP. Mark them visited.
1561 (run_scc_vn): Make code match comment.
1563 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1565 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1566 OPTION_MASK_FLOAT128_KEYWORD.
1567 (POWERPC_MASKS): Likewise.
1568 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1569 support for the -mfloat128-type option, and make -mfloat128
1570 default on PowerPC Linux systems. Define or undefine
1571 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1572 Define __float128 to be __ieee128 if IEEE 128-bit support is
1573 enabled, or undefine it.
1574 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1575 Delete defining __FLOAT128_TYPE__.
1576 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1577 -mfloat128-type option and make -mfloat128 default on PowerPC
1579 (TARGET_FLOAT128_TYPE): Likewise.
1580 (-mfloat128-type): Likewise.
1581 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1582 Delete the -mfloat128-type option and make -mfloat128 default on
1583 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1584 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1585 128-bit floating point is enabled. Change tests from using
1586 -mfloat128-type to -mfloat128.
1587 (rs6000_mangle_type): Use the correct mangling for the __float128
1588 type even if normal long double is restricted to 64-bits.
1589 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1591 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1592 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1593 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1595 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1596 documentation for -mfloat128.
1598 2017-09-06 Olivier Hainque <hainque@adacore.com>
1600 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1602 2017-09-06 Wish Wu <wishwu007@gmail.com>
1603 Jakub Jelinek <jakub@redhat.com>
1605 * asan.c (initialize_sanitizer_builtins): Add
1606 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1607 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1608 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1609 BT_FN_VOID_UINT64_PTR variables.
1610 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1611 (BT_FN_VOID_UINT16_UINT16): Likewise.
1612 (BT_FN_VOID_UINT32_UINT32): Likewise.
1613 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1614 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1615 (BT_FN_VOID_UINT64_PTR): Likewise.
1616 * common.opt (flag_sanitize_coverage): New variable.
1617 (fsanitize-coverage=trace-pc): Remove.
1618 (fsanitize-coverage=): Add.
1619 * flag-types.h (enum sanitize_coverage_code): New enum.
1620 * fold-const.c (fold_range_test): Disable non-short-circuit
1621 optimization if flag_sanitize_coverage.
1622 (fold_truth_andor): Likewise.
1623 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1624 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1625 (coverage_sanitizer_opts): New array.
1626 (get_closest_sanitizer_option): Add OPTS argument, handle also
1627 OPT_fsanitize_coverage_.
1628 (parse_sanitizer_options): Adjusted to also handle
1629 OPT_fsanitize_coverage_.
1630 (common_handle_option): Add OPT_fsanitize_coverage_.
1631 * sancov.c (instrument_comparison, instrument_switch): New function.
1632 (sancov_pass): Add trace-cmp support.
1633 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1634 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1635 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1636 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1637 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1638 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1639 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1640 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1641 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1642 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1644 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1646 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1647 error. Only quit immediately if parsing is complete.
1648 (BEGIN): Initialize fatal_err and parse_done.
1649 (begin fpu, end fpu): Check number of arguments.
1650 (begin arch, end arch): Likewise.
1651 (begin cpu, end cpu): Likewise.
1652 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1653 (optalias): Likewise.
1655 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1657 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1658 * config/arm/arm-isa.h: Delete. Move definitions to ...
1659 * arm-cpus.in: ... here. Use new feature and fgroup values.
1660 * config/arm/arm.c (arm_option_override): Use lower case for feature
1662 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1663 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1664 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1666 (print_isa_bits_for): New function.
1667 (gen_isa): New function.
1668 (gen_comm_data): Use print_isa_bits_for.
1669 (define feature): New keyword.
1670 (define fgroup): New keyword.
1671 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
1672 (arm-isa.h): Add rule to generate file.
1673 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1674 case for feature bit names.
1676 2017-09-06 Richard Biener <rguenther@suse.de>
1678 * tree-ssa-pre.c (NECESSARY): Remove.
1679 (create_expression_by_pieces): Do not touch pass-local flags.
1680 (insert_into_preds_of_block): Likewise.
1681 (do_pre_regular_insertion): Likewise.
1682 (eliminate_insert): Likewise.
1683 (eliminate_dom_walker::before_dom_children): Likewise.
1684 (fini_eliminate): Do not look at inserted_exprs.
1685 (mark_operand_necessary): Remove.
1686 (remove_dead_inserted_code): Replace with simple work-list
1687 algorithm based on inserted_exprs and SSA uses.
1688 (pass_pre::execute): Re-order fini_eliminate and
1689 remove_dead_inserted_code.
1691 2017-09-06 Olivier Hainque <hainque@adacore.com>
1693 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1694 for VxWorks 7. Adjust surrounding comments.
1696 2017-09-06 Richard Biener <rguenther@suse.de>
1698 * gimple-ssa-strength-reduction.c
1699 (find_candidates_dom_walker::before_dom_children): Also allow
1702 2017-09-06 Richard Biener <rguenther@suse.de>
1704 PR tree-optimization/82108
1705 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1706 for gap in the non-permutation SLP case.
1708 2017-09-06 Martin Jambor <mjambor@suse.cz>
1710 PR tree-optimization/82078
1711 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1712 add_access_to_work_queue...
1713 (build_accesses_from_assign): ...here.
1714 (propagate_all_subaccesses): Make sure racc is the group
1715 representative, if there is one.
1717 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1720 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1723 2017-09-06 Richard Biener <rguenther@suse.de>
1725 * gimple-ssa-strength-reduction.c
1726 (find_candidates_dom_walker::before_doom_children): Use a
1727 type and not a mode check.
1729 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1732 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1734 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1736 2017-09-05 Jeff Law <law@redhat.com>
1738 PR tree-optimization/64910
1739 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1740 cases where we have 3 or more operands.
1742 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1745 * omp-low.c (lower_omp_for): Recompute tree invariant if
1746 gimple_omp_for_initial/final is ADDR_EXPR.
1749 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1750 into gimple val before gimplification fo the COND_EXPR.
1752 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1754 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1755 REGION_COPY argument.
1756 (thread_through_all_blocks): Remove unused argument to
1757 duplicate_thread_path.
1759 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1760 Alan Hayward <alan.hayward@arm.com>
1761 David Sherwood <david.sherwood@arm.com>
1763 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1764 Take a scalar_mode rather than a machine_mode.
1765 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1766 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1767 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1768 (aarch64_gen_adjusted_ldpstp): Likewise.
1769 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
1771 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1772 Alan Hayward <alan.hayward@arm.com>
1773 David Sherwood <david.sherwood@arm.com>
1775 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
1776 Take a scalar_int_mode instead of a machine_mode.
1777 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1778 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1779 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1780 (aarch64_simd_attr_length_rglist): Delete.
1781 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
1782 a scalar_int_mode instead of a machine_mode.
1783 (aarch64_add_offset): Likewise.
1784 (aarch64_internal_mov_immediate): Likewise
1785 (aarch64_add_constant_internal): Likewise.
1786 (aarch64_add_constant): Likewise.
1787 (aarch64_movw_imm): Likewise.
1788 (aarch64_rtx_arith_op_extract_p): Likewise.
1789 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1790 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1791 Remove assert that the mode isn't a vector.
1792 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1793 (aarch64_expand_mov_immediate): Update calls after above changes.
1794 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
1795 (aarch64_and_bitmask_imm): Check for scalar integer modes.
1796 (aarch64_move_imm): Likewise.
1797 (aarch64_can_const_movi_rtx_p): Likewise.
1798 (aarch64_strip_extend): Likewise.
1799 (aarch64_extr_rtx_p): Likewise.
1800 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
1801 a CONST_INT when the mode parameter is VOIDmode.
1802 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
1804 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1806 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
1807 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1808 (bitwise_type_for_mode): Update accordingly.
1810 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1812 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
1813 * stor-layout.c (mode_for_size_tree): Likewise.
1814 (mode_for_array): Update accordingly.
1815 (layout_decl): Likewise.
1816 (compute_record_mode): Likewise. Only set the mode once.
1818 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1820 * target.def (get_mask_mode): Change return type to opt_mode.
1822 * doc/tm.texi: Regenerate.
1823 * targhooks.h (default_get_mask_mode): Return an opt_mode.
1824 * targhooks.c (default_get_mask_mode): Likewise.
1825 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1826 * optabs-query.c (can_vec_mask_load_store_p): Update use of
1827 targetm.get_mask_mode.
1828 * tree.c (build_truth_vector_type): Likewise.
1830 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1832 * machmode.h (mode_for_vector): Return an opt_mode.
1833 * stor-layout.c (mode_for_vector): Likewise.
1834 (mode_for_int_vector): Update accordingly.
1835 (layout_type): Likewise.
1836 * config/i386/i386.c (emit_memmov): Likewise.
1837 (ix86_expand_set_or_movmem): Likewise.
1838 (ix86_expand_vector_init): Likewise.
1839 (ix86_get_mask_mode): Likewise.
1840 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
1842 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
1843 * expmed.c (extract_bit_field_1): Likewise.
1844 * expr.c (expand_expr_real_2): Likewise.
1845 * optabs-query.c (can_vec_perm_p): Likewise.
1846 (can_vec_mask_load_store_p): Likewise.
1847 * optabs.c (expand_vec_perm): Likewise.
1848 * targhooks.c (default_get_mask_mode): Likewise.
1849 * tree-vect-stmts.c (vectorizable_store): Likewise.
1850 (vectorizable_load): Likewise.
1851 (get_vectype_for_scalar_type_and_size): Likewise.
1853 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1855 * machmode.h (mode_for_int_vector): New function.
1856 * stor-layout.c (mode_for_int_vector): Likewise.
1857 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
1858 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
1859 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1860 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1861 (s390_expand_vcond): Likewise.
1863 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1865 * machmode.h (opt_machine_mode): New type.
1866 (opt_mode<T>): Allow construction from anything that can be
1868 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
1869 (mode_for_size): Return an opt_machine_mode.
1870 * stor-layout.c (mode_for_size): Likewise.
1871 (mode_for_size_tree): Update call accordingly.
1872 (bitwise_mode_for_mode): Likewise.
1873 (make_fract_type): Likewise.
1874 (make_accum_type): Likewise.
1875 * caller-save.c (replace_reg_with_saved_mem): Update call
1877 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1878 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1879 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1880 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1881 * expmed.c (extract_bit_field_1): Likewise.
1882 * reload.c (get_secondary_mem): Likewise.
1883 * varasm.c (assemble_integer): Likewise.
1884 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
1887 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1889 * machmode.h (decimal_float_mode_for_size): New function.
1890 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
1891 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
1892 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
1893 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
1894 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
1895 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
1897 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1899 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
1900 (get_builtin_sync_mode): Likewise.
1901 (expand_ifn_atomic_compare_exchange): Likewise.
1902 (expand_builtin_atomic_clear): Likewise.
1903 (expand_builtin_atomic_test_and_set): Likewise.
1904 (fold_builtin_atomic_always_lock_free): Likewise.
1905 * calls.c (compute_argument_addresses): Likewise.
1906 (emit_library_call_value_1): Likewise.
1907 (store_one_arg): Likewise.
1908 * combine.c (combine_instructions): Likewise.
1909 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
1910 * config/arm/arm.c (arm_function_value): Likewise.
1911 (aapcs_allocate_return_reg): Likewise.
1912 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
1913 * config/i386/i386.c (construct_container): Likewise.
1914 (ix86_gimplify_va_arg): Likewise.
1915 (ix86_expand_sse_cmp): Likewise.
1916 (emit_memmov): Likewise.
1917 (emit_memset): Likewise.
1918 (expand_small_movmem_or_setmem): Likewise.
1919 (ix86_expand_pextr): Likewise.
1920 (ix86_expand_pinsr): Likewise.
1921 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
1922 * config/microblaze/microblaze.c (microblaze_block_move_straight):
1924 * config/mips/mips.c (mips_function_value_1) Likewise.
1925 (mips_block_move_straight): Likewise.
1926 (mips_expand_ins_as_unaligned_store): Likewise.
1927 * config/powerpcspe/powerpcspe.c
1928 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1929 (rs6000_darwin64_record_arg_flush): Likewise.
1930 * config/rs6000/rs6000.c
1931 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1932 (rs6000_darwin64_record_arg_flush): Likewise.
1933 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
1934 (sparc_function_value_1): Likewise.
1935 * config/spu/spu.c (adjust_operand): Likewise.
1936 (spu_emit_branch_or_set): Likewise.
1937 (arith_immediate_p): Likewise.
1938 * emit-rtl.c (gen_lowpart_common): Likewise.
1939 * expr.c (expand_expr_real_1): Likewise.
1940 * function.c (assign_parm_setup_block): Likewise.
1941 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1942 * reload1.c (alter_reg): Likewise.
1943 * stor-layout.c (mode_for_vector): Likewise.
1944 (layout_type): Likewise.
1946 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1948 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
1949 (spu_convert_move): Likewise.
1950 * lower-subreg.c (resolve_simple_move): Likewise.
1952 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1955 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
1956 define_insn to a define_expand.
1957 (altivec_vsum2sws_direct): New define_insn.
1958 (altivec_vsumsws): Convert from a define_insn to a define_expand.
1960 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1962 * config/arm/arm.c (arm_option_params_internal): Improve setting of
1965 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
1970 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
1971 realign stack if stack alignment needed is less than incoming
1974 2017-09-05 Marek Polacek <polacek@redhat.com>
1977 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
1980 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1982 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
1984 2017-09-05 Richard Biener <rguenther@suse.de>
1986 PR tree-optimization/82084
1987 * fold-const.c (can_native_encode_string_p): Handle wide characters.
1989 2017-09-05 Richard Biener <rguenther@suse.de>
1991 PR tree-optimization/82102
1992 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
1994 2017-09-05 Martin Liska <mliska@suse.cz>
1996 PR tree-optimization/82032
1997 * tree-cfg.c (generate_range_test): New function.
1998 * tree-cfg.h (generate_range_test): Declared here.
1999 * tree-cfgcleanup.c (convert_single_case_switch): New function.
2000 (cleanup_control_expr_graph): Use it.
2001 * tree-switch-conversion.c (try_switch_expansion): Remove
2003 (emit_case_nodes): Use generate_range_test.
2005 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2008 * config/i386/i386.md (*<btsc><mode>_mask): Add
2009 TARGET_USE_BT to insn constraint.
2010 (*btr<mode>_mask): Ditto.
2012 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
2014 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
2015 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
2017 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2020 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
2021 TARGET_NEON and TARGET_IWMMXT.
2022 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
2023 TARGET_NEON and TARGET_IWMMXT.
2024 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
2026 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2028 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
2029 (ix86_rewrite_tls_address): Ditto.
2030 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
2031 (ix86_rewrite_tls_address_1): Ditto.
2032 (ix86_rewrite_tls_address): Ditto.
2033 * config/i386/predicates.md (tls_address_pattern): New predicate.
2034 * config/i386/i386.md (TLS address splitter): New splitter.
2036 2017-09-04 Richard Biener <rguenther@suse.de>
2038 PR tree-optimization/82084
2039 * fold-const.h (can_native_encode_string_p): Declare.
2040 * fold-const.c (can_native_encode_string_p): Factor out from ...
2041 (native_encode_string): ... here.
2042 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
2043 vectorizing stores from constants we later cannot handle.
2045 2017-09-04 Marek Polacek <polacek@redhat.com>
2048 * doc/invoke.texi: Update -Wtautological-compare documentation.
2050 2017-09-04 Jeff Law <law@redhat.com>
2052 PR tree-optimization/64910
2053 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
2054 swap the first and last operand if the last is a constant.
2056 2017-09-04 Marek Polacek <polacek@redhat.com>
2059 * convert.c (do_narrow): When sanitizing signed integer overflows,
2060 bail out for signed types.
2061 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
2063 2017-09-04 Richard Biener <rguenther@suse.de>
2065 PR tree-optimization/82060
2066 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
2067 Move devirtualization after stmt folding and before EH/AB/noreturn
2068 cleanup to get the stmt refs canonicalized. Use a bool instead
2069 of gimple_modified_p since that doesn't work for NOPs. Schedule
2070 NOPs generated by folding for removal.
2072 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2073 Alan Hayward <alan.hayward@arm.com>
2074 David Sherwood <david.sherwood@arm.com>
2076 * coretypes.h (pad_direction): New enum.
2077 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
2078 (FUNCTION_ARG_PADDING): Likewise.
2079 * target.def (function_arg_padding): New hook.
2080 * targhooks.h (default_function_arg_padding): Declare.
2081 * targhooks.c (default_function_arg_padding): New function.
2082 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
2083 (TARGET_FUNCTION_ARG_PADDING): ...this.
2084 * doc/tm.texi: Regenerate.
2085 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
2086 instead of direction.
2087 (compute_argument_addresses): Likewise.
2088 (load_register_parameters): Likewise.
2089 (emit_library_call_value_1): Likewise.
2090 (store_one_arg): Use targetm.calls.function_arg_padding instead
2091 of FUNCTION_ARG_PADDING.
2092 (must_pass_in_stack_var_size_or_pad): Likewise.
2093 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
2094 (emit_group_store): Likewise.
2095 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
2096 instead of FUNCTION_ARG_PADDING.
2097 (emit_push_insn): Likewise, and propagate enum change throughout
2099 * function.h (direction): Delete.
2100 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
2102 * function.c (assign_parm_find_stack_rtl): Likewise.
2103 (assign_parm_setup_block_p): Likewise.
2104 (assign_parm_setup_block): Likewise.
2105 (gimplify_parameters): Likewise.
2106 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
2107 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
2109 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
2110 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2111 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
2112 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
2113 (aarch64_function_arg_padding): ...this new function.
2114 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
2115 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2116 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
2117 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2118 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
2119 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2120 (arm_pad_arg_upward): Replace with...
2121 (arm_function_arg_padding): ...this new function.
2122 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
2124 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
2125 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
2126 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2127 (ia64_hpux_function_arg_padding): Replace with...
2128 (ia64_function_arg_padding): ...this new function. Use pad_direction
2129 instead of direction. Check for TARGET_HPUX.
2130 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
2131 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2132 (iq2000_function_arg_padding): New function.
2133 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
2134 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
2135 (mips_function_arg_padding): ...this new function.
2136 (mips_pad_reg_upward): Update accordingly.
2137 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2138 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
2139 targetm.calls.function_arg_padding.
2140 (FUNCTION_ARG_PADDING): Delete.
2141 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2142 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
2143 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2144 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
2145 (nios2_block_reg_padding): Return pad_direction instead of direction.
2146 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
2147 instead of direction.
2148 (nios2_function_arg_padding): Likewise. Make static.
2149 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2150 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
2151 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
2152 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
2153 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2154 (pa_function_arg_padding): Make static. Return pad_direction instead
2156 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
2157 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2158 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
2159 instead of direction. Use targetm.calls.function_arg_padding.
2160 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
2161 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2162 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
2163 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
2164 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
2166 (function_arg_padding): Rename to...
2167 (rs6000_function_arg_padding): ...this. Make static. Return
2168 pad_direction instead of direction.
2169 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2170 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
2171 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2172 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
2173 instead of direction. Use targetm.calls.function_arg_padding.
2174 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
2175 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2176 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
2177 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
2178 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2179 (function_arg_padding): Rename to...
2180 (rs6000_function_arg_padding): ...this. Make static. Return
2181 pad_direction instead of direction.
2182 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2183 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
2184 * config/s390/s390.c (s390_function_arg_padding): New function.
2185 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2186 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
2187 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
2188 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2189 (function_arg_padding): Rename to...
2190 (sparc_function_arg_padding): ...this. Make static. Return
2191 pad_direction instead of direction.
2192 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
2193 * config/spu/spu.c (spu_function_arg_padding): New function.
2194 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2195 * system.h (FUNCTION_ARG_PADDING): Poison.
2197 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2198 Alan Hayward <alan.hayward@arm.com>
2199 David Sherwood <david.sherwood@arm.com>
2201 * target.def (modes_tieable_p): New hook.
2202 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
2203 (TARGET_MODES_TIEABLE_P): ...this.
2204 * doc/tm.texi.in: Regenerate.
2205 * hooks.h (hook_bool_mode_mode_true): Declare.
2206 * hooks.c (hook_bool_mode_mode_true): New function.
2207 * combine.c (subst): Use targetm.modes_tieable_p instead of
2209 * dse.c (find_shift_sequence): Likewise.
2210 * expmed.c (extract_low_bits): Likewise.
2211 * lower-subreg.c: Include target.h.
2212 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
2214 * rtlanal.c (rtx_cost): Likewise.
2215 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
2216 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
2217 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
2218 (TARGET_MODES_TIEABLE_P): Redefine.
2219 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
2220 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
2221 (TARGET_MODES_TIEABLE_P): Redefine.
2222 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
2223 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
2224 (arc_modes_tieable_p): New function.
2225 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
2226 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
2227 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
2228 (arm_modes_tieable_p): Make static.
2229 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
2230 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
2231 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
2232 (TARGET_MODES_TIEABLE_P): Redefine.
2233 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
2234 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
2235 (TARGET_MODES_TIEABLE_P): Redefine.
2236 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
2237 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
2238 (cr16_modes_tieable_p): New function.
2239 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
2240 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
2241 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
2242 (TRULY_NOOP_TRUNCATION): Update comment.
2243 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
2244 (TRULY_NOOP_TRUNCATION): Update comment.
2245 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
2246 (frv_modes_tieable_p): New function.
2247 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
2248 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
2249 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
2250 (TARGET_MODES_TIEABLE_P): Redefine.
2251 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
2252 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
2253 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
2254 (TARGET_MODES_TIEABLE_P): Redefine.
2255 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
2256 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
2257 (ia64_modes_tieable_p): New function.
2258 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
2259 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
2260 (iq2000_modes_tieable_p): New function.
2261 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
2262 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
2263 (lm32_modes_tieable_p): New function.
2264 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
2265 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
2266 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
2267 (TARGET_MODES_TIEABLE_P): Redefine.
2268 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
2269 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
2270 (m32r_modes_tieable_p): New function.
2271 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
2272 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
2273 (m68k_modes_tieable_p): New function.
2274 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
2275 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
2276 (mcore_modes_tieable_p): New function.
2277 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
2278 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
2280 (TARGET_MODES_TIEABLE_P): Redefine.
2281 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
2282 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
2283 * config/mips/mips.c (mips_modes_tieable_p): Make static.
2284 (TARGET_MODES_TIEABLE_P): Redefine.
2285 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
2286 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
2287 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
2288 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
2289 (mn10300_modes_tieable_p): ...this and make static.
2290 (TARGET_MODES_TIEABLE_P): Redefine.
2291 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
2292 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
2293 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
2294 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
2295 (msp430_modes_tieable_p): Make static.
2296 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
2297 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
2298 (TARGET_MODES_TIEABLE_P): Redefine.
2299 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
2300 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
2301 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
2302 (TARGET_MODES_TIEABLE_P): Redefine.
2303 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
2304 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
2305 * config/pa/pa.c (pa_modes_tieable_p): Make static.
2306 (TARGET_MODES_TIEABLE_P): Redefine.
2307 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
2308 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
2309 (pdp11_modes_tieable_p): New function.
2310 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
2311 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
2312 (rs6000_modes_tieable_p): New function.
2313 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2314 * config/powerpcspe/powerpcspe.md: Update comment.
2315 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
2316 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
2317 (TARGET_MODES_TIEABLE_P): Redefine.
2318 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
2319 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
2320 (rl78_modes_tieable_p): New function.
2321 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
2322 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
2323 (rs6000_modes_tieable_p): New function.
2324 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2325 * config/rs6000/rs6000.md: Update comment.
2326 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
2327 * config/rx/rx.c (rx_modes_tieable_p): New function.
2328 (TARGET_MODES_TIEABLE_P): Redefine.
2329 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
2330 * config/s390/s390.c (s390_modes_tieable_p): New function.
2331 (TARGET_MODES_TIEABLE_P): Redefine.
2332 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
2333 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
2334 (sh_modes_tieable_p): New function.
2335 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
2336 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
2337 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
2338 (sparc_modes_tieable_p): Make static.
2339 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
2340 * config/spu/spu.c (spu_modes_tieable_p): New function.
2341 (TARGET_MODES_TIEABLE_P): Redefine.
2342 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
2343 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
2344 (TARGET_MODES_TIEABLE_P): Redefine.
2345 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
2346 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
2347 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
2348 * config/v850/v850.c (v850_modes_tieable_p): New function.
2349 (TARGET_MODES_TIEABLE_P): Redefine.
2350 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
2351 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
2352 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
2353 (visium_modes_tieable_p): New function.
2354 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
2355 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
2356 (xtensa_modes_tieable_p): New function.
2357 * system.h (MODES_TIEABLE_P): Poison.
2359 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2360 Alan Hayward <alan.hayward@arm.com>
2361 David Sherwood <david.sherwood@arm.com>
2363 * target.def (hard_regno_mode_ok): New hook.
2364 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
2365 (TARGET_HARD_REGNO_MODE_OK): ...this.
2366 * doc/tm.texi.in: Regenerate.
2367 * hooks.h (hook_bool_uint_mode_true): Declare.
2368 * hooks.c (hook_bool_uint_mode_true): New function.
2369 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
2371 * genpreds.c (write_insn_preds_c): Add an include of target.h.
2372 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
2373 instead of HARD_REGNO_MODE_OK.
2374 * caller-save.c: Include target.h.
2375 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
2377 * combine.c (can_combine_p): Likewise.
2378 (combinable_i3pat): Likewise.
2379 (can_change_dest_mode): Likewise.
2380 * expr.c (init_expr_target): Likewise.
2381 (convert_move): Likewise.
2382 (convert_modes): Likewise.
2383 * ira.c (setup_prohibited_class_mode_regs): Likewise.
2384 (setup_prohibited_mode_move_regs): Likewise.
2385 * ira.h (target_ira): Likewise.
2386 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2387 * lra-constraints.c (process_alt_operands): Likewise.
2388 (split_reg): Likewise.
2389 * recog.c (peep2_find_free_register): Likewise.
2390 * ree.c (combine_reaching_defs): Likewise.
2391 * regcprop.c (maybe_mode_change): Likewise.
2392 * reginfo.c (init_reg_sets_1): Likewise.
2393 (choose_hard_reg_mode): Likewise.
2394 (simplifiable_subregs): Likewise.
2395 * regrename.c (check_new_reg_p): Likewise.
2396 * reload.c (find_valid_class): Likewise.
2397 (find_valid_class_1): Likewise.
2398 (reload_inner_reg_of_subreg): Likewise.
2399 (push_reload): Likewise.
2400 (combine_reloads): Likewise.
2401 (find_dummy_reload): Likewise.
2402 (find_reloads): Likewise.
2403 * reload1.c (find_reg): Likewise.
2404 (set_reload_reg): Likewise.
2405 (allocate_reload_reg): Likewise.
2406 (choose_reload_regs): Likewise.
2407 (reload_adjust_reg_for_temp): Likewise.
2408 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
2409 (simplify_subreg_regno): Likewise.
2410 * sel-sched.c (init_regs_for_mode): Likewise.
2411 * varasm.c (make_decl_rtl): Likewise.
2412 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
2413 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
2415 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
2416 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
2417 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2418 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
2419 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
2420 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2421 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
2422 (arc_mode_class): Delete.
2423 (HARD_REGNO_MODE_OK): Delete.
2424 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2425 (arc_hard_regno_mode_ok): Rename old array to...
2426 (arc_hard_regno_mode_ok_modes): ...this.
2427 (arc_conditional_register_usage): Update accordingly.
2428 (arc_mode_class): Make static.
2429 (arc_hard_regno_mode_ok): New function.
2430 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2431 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2432 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2433 (arm_hard_regno_mode_ok): Make static.
2434 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2436 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2437 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2438 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2440 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2441 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2442 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2443 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2444 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2445 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2446 * config/bfin/predicates.md (valid_reg_operand): Use
2447 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2448 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2449 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2450 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2451 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2452 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2453 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2454 (cr16_hard_regno_mode_ok): Make static and return a bool.
2455 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2456 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2457 (cris_hard_regno_mode_ok): New function.
2458 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2459 (epiphany_mode_class): Delete.
2460 (HARD_REGNO_MODE_OK): Delete.
2461 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2462 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2463 (hard_regno_mode_ok): Rename to...
2464 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2465 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2466 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2468 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2469 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2470 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2471 (frv_hard_regno_mode_ok): Make static and return a bool.
2472 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2474 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2475 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2476 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2477 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2479 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2480 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2481 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2482 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2484 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2485 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2486 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2487 (ia64_hard_regno_mode_ok): New function.
2488 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2489 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2490 (iq2000_hard_regno_mode_ok): New function.
2491 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2492 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2493 (lm32_hard_regno_mode_ok): New function.
2494 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2495 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2496 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2497 instead of HARD_REGNO_MODE_OK.
2498 (m32c_hard_regno_ok): Rename to...
2499 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2500 (m32c_cannot_change_mode_class): Update accordingly.
2501 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2502 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2503 (m32r_mode_class): Delete.
2504 (HARD_REGNO_MODE_OK): Delete.
2505 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2506 (m32r_hard_regno_mode_ok): Rename to...
2507 (m32r_hard_regno_modes): ...this.
2508 (m32r_mode_class): Make static.
2509 (m32r_hard_regno_mode_ok): New function.
2510 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2511 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2512 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2513 (m68k_hard_regno_mode_ok): Make static.
2514 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2515 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2516 (mcore_hard_regno_mode_ok): New function.
2517 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2518 (HARD_REGNO_MODE_OK): Delete.
2519 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2521 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2522 (microblaze_hard_regno_mode_ok): New function.
2523 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2524 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2525 (mips_hard_regno_mode_ok): Delete.
2526 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2527 (mips_hard_regno_mode_ok_p): ...this and make static.
2528 (mips_hard_regno_mode_ok_p): Rename to...
2529 (mips_hard_regno_mode_ok_uncached): ...this.
2530 (mips_hard_regno_mode_ok): New function.
2531 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2532 of HARD_REGNO_MODE_OK.
2533 (mips_option_override): Update after above name changes.
2534 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2535 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2536 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2537 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2538 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2539 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2540 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2541 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2542 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2543 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2544 (msp430_hard_regno_mode_ok): Make static and return a bool.
2545 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2546 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2547 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2549 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2550 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2551 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2552 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2553 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2554 (PA_HARD_REGNO_MODE_OK): ...this
2555 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2556 (PA_HARD_REGNO_MODE_OK): ...this.
2557 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2558 (pa_hard_regno_mode_ok): New function.
2559 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2560 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2561 (pdp11_hard_regno_mode_ok): New function.
2562 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2563 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2565 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2567 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2568 (rs6000_hard_regno_mode_ok): Rename to...
2569 (rs6000_hard_regno_mode_ok_uncached): ...this.
2570 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2571 (rs6000_hard_regno_mode_ok): New function.
2572 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2573 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2574 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2575 (riscv_hard_regno_mode_ok): ...this and make static.
2576 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2577 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2578 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2579 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2580 (rl78_hard_regno_mode_ok): Make static and return bool.
2581 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2582 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2584 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2585 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2586 (rs6000_hard_regno_mode_ok): Rename to...
2587 (rs6000_hard_regno_mode_ok_uncached): ...this.
2588 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2589 (rs6000_hard_regno_mode_ok): New function.
2590 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2591 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2592 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2593 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2594 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2595 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2596 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2597 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2598 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2599 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2600 (sh_hard_regno_mode_ok): Make static.
2601 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2602 instead of HARD_REGNO_MODE_OK.
2603 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2604 (sparc_mode_class): Delete.
2605 (HARD_REGNO_MODE_OK): Delete.
2606 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2607 (hard_regno_mode_classes): Make static.
2608 (sparc_mode_class): Likewise.
2609 (sparc_hard_regno_mode_ok): New function.
2610 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2611 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2612 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2614 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2615 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2616 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2617 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2618 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2619 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2620 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2621 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2622 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2623 (visium_hard_regno_mode_ok): New function.
2624 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2625 instead of HARD_REGNO_MODE_OK.
2626 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2627 (HARD_REGNO_MODE_OK): Delete.
2628 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2629 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2630 (xtensa_option_override): Update accordingly.
2631 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2632 (xtensa_hard_regno_mode_ok): New function.
2633 * system.h (HARD_REGNO_MODE_OK): Poison.
2635 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2636 Alan Hayward <alan.hayward@arm.com>
2637 David Sherwood <david.sherwood@arm.com>
2639 * target.def (hard_regno_call_part_clobbered): New hook.
2640 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2641 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2642 * doc/tm.texi: Regenerate.
2643 * hooks.h (hook_bool_uint_mode_false): Declare.
2644 * hooks.c (hook_bool_uint_mode_false): New function.
2645 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2646 * cselib.c (cselib_process_insn): Use
2647 targetm.hard_regno_call_part_clobbered instead of
2648 HARD_REGNO_CALL_PART_CLOBBERED.
2649 * ira-conflicts.c (ira_build_conflicts): Likewise.
2650 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2651 * lra-constraints.c (need_for_call_save_p): Likewise.
2652 * lra-lives.c: Include target.h.
2653 (check_pseudos_live_through_calls): Use
2654 targetm.hard_regno_call_part_clobbered instead of
2655 HARD_REGNO_CALL_PART_CLOBBERED.
2656 * regcprop.c: Include target.h.
2657 (copyprop_hardreg_forward_1): Use
2658 targetm.hard_regno_call_part_clobbered instead of
2659 HARD_REGNO_CALL_PART_CLOBBERED.
2660 * reginfo.c (choose_hard_reg_mode): Likewise.
2661 * regrename.c (check_new_reg_p): Likewise.
2662 * reload.c (find_equiv_reg): Likewise.
2663 * reload1.c (emit_reload_insns): Likewise.
2664 * sched-deps.c (deps_analyze_insn): Likewise.
2665 * sel-sched.c (init_regs_for_mode): Likewise.
2666 (mark_unavailable_hard_regs): Likewise.
2667 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2668 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2669 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
2671 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2672 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2673 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
2675 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
2677 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2678 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2679 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
2681 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2682 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2683 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2685 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2686 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2688 * config/powerpcspe/powerpcspe.c
2689 (rs6000_hard_regno_call_part_clobbered): New function.
2690 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2691 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2692 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2694 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2695 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2696 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2698 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2699 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2700 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2702 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2703 Alan Hayward <alan.hayward@arm.com>
2704 David Sherwood <david.sherwood@arm.com>
2706 * rtl.h (subreg_memory_offset): Declare.
2707 * emit-rtl.c (subreg_memory_offset): New function.
2708 * expmed.c (store_bit_field_1): Use it.
2709 * expr.c (undefined_operand_subword_p): Likewise.
2710 * simplify-rtx.c (simplify_subreg): Likewise.
2712 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2714 PR rtl-optimization/57448
2717 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2718 using atomic_load pattern.
2719 (expand_atomic_store): Likewise.
2721 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2724 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2725 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2726 handling. Use replace_call_with_value with NULL instead of
2727 gsi_replace, unlink_stmt_vdef and release_defs.
2729 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2732 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2734 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2737 * rtl.h (emit_library_call_value_1): Declare.
2738 (emit_library_call): Replace declaration with a series of overloads.
2739 Remove the parameter count argument.
2740 (emit_library_call_value): Likewise.
2741 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2742 with an "rtx_mode_t *".
2743 (emit_library_call_value): Delete.
2744 (emit_library_call): Likewise.
2745 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2746 (asan_emit_allocas_unpoison): Likewise.
2747 * builtins.c (expand_builtin_powi): Likewise.
2748 (expand_asan_emit_allocas_unpoison): Likewise.
2749 * cfgexpand.c (expand_main_function): Likewise.
2750 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2751 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2752 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2753 * config/arm/arm.c (arm_trampoline_init): Likewise.
2754 (arm_call_tls_get_addr): Likewise.
2755 (arm_expand_divmod_libfunc): Likewise.
2756 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2757 (smulsi3_highpart): Likewise.
2758 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2759 (c6x_expand_compare): Likewise.
2760 (c6x_expand_movmem): Likewise.
2761 * config/frv/frv.c (frv_trampoline_init): Likewise.
2762 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2763 (ix86_expand_divmod_libfunc): Likewise.
2764 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2765 (ia64_expand_compare): Likewise.
2766 (ia64_profile_hook): Likewise.
2767 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2768 (nonlocal_goto): Likewise.
2769 (restore_stack_nonlocal): Likewise.
2770 * config/m32r/m32r.c (block_move_call): Likewise.
2771 (m32r_trampoline_init): Likewise.
2772 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
2773 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
2774 (m68k_call_m68k_read_tp): Likewise.
2775 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
2776 (microblaze_expand_divide): Likewise.
2777 * config/mips/mips.h (mips_args): Likewise.
2778 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
2779 (MIPS_ICACHE_SYNC): Likewise.
2780 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
2781 (nios2_trampoline_init): Likewise.
2782 * config/pa/pa.c (hppa_tls_call): Likewise.
2783 (pa_trampoline_init): Likewise.
2784 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
2785 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
2786 (expand_strn_compare): Likewise.
2787 (rs6000_generate_compare): Likewise.
2788 (rs6000_expand_float128_convert): Likewise.
2789 (output_profile_hook): Likewise.
2790 (rs6000_trampoline_init): Likewise.
2791 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
2792 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
2793 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
2794 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
2795 (rs6000_generate_compare): Likewise.
2796 (rs6000_expand_float128_convert): Likewise.
2797 (output_profile_hook): Likewise.
2798 (rs6000_trampoline_init): Likewise.
2799 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
2800 * config/sh/sh.c (sh_trampoline_init): Likewise.
2801 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
2802 (sparc_emit_float_lib_cmp): Likewise.
2803 (sparc32_initialize_trampoline): Likewise.
2804 (sparc64_initialize_trampoline): Likewise.
2805 (sparc_profile_hook): Likewise.
2806 * config/spu/spu.c (ea_load_store): Likewise.
2807 * config/spu/spu.md (floatunssidf2): Likewise.
2808 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
2809 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
2810 * config/visium/visium.c (expand_block_move_4): Likewise.
2811 (expand_block_move_2): Likewise.
2812 (expand_block_move_1): Likewise.
2813 (expand_block_set_4): Likewise.
2814 (expand_block_set_2): Likewise.
2815 (expand_block_set_1): Likewise.
2816 (visium_trampoline_init): Likewise.
2817 (visium_profile_hook): Likewise.
2818 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
2819 (xtensa_setup_frame_addresses): Likewise.
2820 (xtensa_trampoline_init): Likewise.
2821 * except.c (sjlj_emit_function_enter): Likewise.
2822 (sjlj_emit_function_exit): Likewise.
2823 * explow.c (allocate_dynamic_stack_space): Likewise.
2824 (probe_stack_range): Likewise.
2825 * expr.c (convert_mode_scalar): Likewise.
2826 * optabs.c (expand_binop): Likewise.
2827 (expand_twoval_binop_libfunc): Likewise.
2828 (expand_unop): Likewise.
2829 (prepare_cmp_insn): Likewise.
2830 (prepare_float_lib_cmp): Likewise.
2831 (expand_float): Likewise.
2832 (expand_fix): Likewise.
2833 (expand_fixed_convert): Likewise.
2834 (maybe_emit_sync_lock_test_and_set): Likewise.
2835 (expand_atomic_compare_and_swap): Likewise.
2836 (expand_mem_thread_fence): Likewise.
2837 (expand_atomic_fetch_op): Likewise.
2839 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
2841 * doc/generic.texi (OpenACC): Adjust URL.
2842 * doc/invoke.texi (C Dialect Options): Ditto.
2844 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
2846 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
2847 predicate for operand 1. Add (m,<S>) constraint.
2848 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
2849 Prevent memory operand 1 with register operand 2.
2851 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
2853 PR rtl-optimization/82024
2854 * combine.c (try_combine): If the combination result is a PARALLEL,
2855 and we only need to retain the SET in there that would be placed
2856 at I2, check that we can place that at I3 instead, before doing so.
2858 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2861 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
2863 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
2864 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
2867 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
2868 Jeff Law <law@redhat.com>
2870 * varasm.c (bss_initializer_p): Do not put constants into .bss
2871 (categorize_decl_for_section): Handle bss_initializer_p returning
2872 false when DECL_INITIAL is NULL.
2874 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2877 * config/s390/s390.c (s390_can_inline_p): New function.
2879 2017-09-01 Jeff Law <law@redhat.com>
2881 PR tree-optimization/82052
2882 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
2883 Always initialize the returned slot after a hash table miss
2884 when INSERT is true.
2886 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
2888 * config/s390/s390.md (mem_signal_fence): Remove.
2889 * doc/md.texi (mem_signal_fence): Remove.
2890 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
2892 * target-insns.def (mem_signal_fence): Remove.
2894 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2897 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
2900 * asan.c (create_odr_indicator): Strip name encoding from assembler
2901 name before appending it after __odr_asan_.
2903 2017-09-01 Martin Liska <mliska@suse.cz>
2905 PR tree-optimization/82059
2906 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
2907 frequency only when an edge is redirected.
2909 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2911 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
2912 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
2913 (arc_conditional_register_usage): Remove ARC600 lp_count
2915 (arc_file_start): Emit Tag_ARC_CPU_variation.
2916 (arc_can_use_doloop_p): New conditions to use ZOLs.
2917 (hwloop_fail): New function.
2918 (hwloop_optimize): Likewise.
2919 (hwloop_pattern_reg): Likewise.
2920 (arc_doloop_hooks): New struct, to be used with reorg_loops.
2921 (arc_reorg_loops): New function, calls reorg_loops.
2922 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
2923 (arc600_corereg_hazard): Remove ZOL checking, case handled by
2925 (arc_loop_hazard): Remove function, functionality moved into
2927 (arc_hazard): Remove arc_loop_hazard call.
2928 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
2929 into hwloop_optimize.
2930 (arc_label_align): Remove ZOL handling.
2931 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
2932 * config/arc/arc.md (doloop_begin): Remove pattern.
2933 (doloop_begin_i): Likewise.
2934 (doloop_end_i): Likewise.
2935 (doloop_fallback): Likewise.
2936 (doloop_fallback_m): Likewise.
2937 (doloop_end): Reimplement expand.
2938 (arc_lp): New pattern for LP instruction.
2939 (loop_end): New pattern.
2940 (loop_fail): Likewise.
2941 (decrement_and_branch_until_zero): Likewise.
2942 * config/arc/arc.opt (mlpc-width): New option.
2943 * doc/invoke.texi (mlpc-width): Document option.
2945 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2947 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
2948 (arc_ccfsm_advance): Fix checking for delay slots.
2949 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
2951 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2953 * config/arc/arc.md (movqi_insn): Add stores to save constant long
2955 (movhi_insn): Update store instruction constraint which are saving
2956 6-bit short immediates.
2957 (movsi_insn): Consider also short scaled load operations.
2958 (zero_extendhisi2_i): Use Usd constraint instead of T.
2959 (extendhisi2_i): Add q constraint.
2960 (arc_clzsi2): Add type and length attributes.
2961 (arc_ctzsi2): Likewise.
2962 * config/arc/constraints.md (Usc): Update constraint, the
2963 assembler can parse two relocations for a single instruction.
2965 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2967 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
2968 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
2970 2017-08-31 Olivier Hainque <hainque@adacore.com>
2972 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
2973 match as powerpc-wrs-vxworks*.
2975 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
2977 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
2978 register constraint for by-element operand.
2979 (aarch64_mls_elt_merge<mode>): Likewise.
2981 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2983 * config/arc/arc.c (arc_can_follow_jump): Check for short
2986 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2988 * config.gcc: Use g.opt for arc.
2989 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
2990 functionality moved to ...
2991 (legitimate_scaled_address_p): New function, ...here.
2992 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
2993 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
2994 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
2996 (arc_override_options): Handle G option.
2997 (arc_output_pic_addr_const): Correct function definition.
2998 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
2999 (arc_decl_anon_ns_mem_p): Delete.
3000 (arc_in_small_data_p): Overhaul this function to take into
3001 consideration the value given via G option.
3002 (arc_rewrite_small_data_1): Renamed and corrected old
3003 arc_rewrite_small_data function.
3004 (arc_rewrite_small_data): New function.
3005 (small_data_pattern): Don't use pic_offset_table_rtx.
3006 * config/arc/arc.h (CC1_SPEC): Recognize G option.
3007 * config/arc/simdext.md (movmisalignv2hi): Use
3008 prepare_move_operands function.
3010 (movmisalign*): Likewise.
3011 * doc/invoke.texi (ARC options): Document -G option.
3013 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
3015 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
3017 * config/arc/arc.c (arc_print_operand): Output scalled address for
3018 sdata whenever is possible.
3019 (arc_in_small_data_p): Allow sdata for 64bit datum when double
3020 load/stores are available.
3021 (compact_sda_memory_operand): Check for the alignment required by
3022 code density instructions.
3023 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
3025 * config/arc/constraints.md (Usd): Update constraint.
3026 (Us0): New constraint.
3027 (Usc): Update constraint.
3029 2017-08-31 Richard Biener <rguenther@suse.de>
3032 * dwarf2out.c (dwarf2out_early_global_decl): Process each
3035 2017-08-31 Tamar Christina <tamar.christina@arm.com>
3037 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
3038 Resize type_signature.
3040 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3041 Alan Hayward <alan.hayward@arm.com>
3042 David Sherwood <david.sherwood@arm.com>
3044 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
3045 subregs whose inner modes can be stored in GPRs.
3046 (aarch64_classify_index): Likewise.
3048 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3049 Alan Hayward <alan.hayward@arm.com>
3050 David Sherwood <david.sherwood@arm.com>
3052 * config/aarch64/iterators.md (V_cmp_result): Rename to...
3053 (V_INT_EQUIV): ...this.
3054 (v_cmp_result): Rename to...
3055 (v_int_equiv): ...this.
3056 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
3057 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
3058 (copysign<mode>3): Likewise.
3059 (aarch64_simd_bsl<mode>_internal): Likewise.
3060 (aarch64_simd_bsl<mode>): Likewise.
3061 (vec_cmp<mode><mode>): Likewise.
3062 (vcond<mode><mode>): Likewise.
3063 (vcond<v_cmp_mixed><mode>): Likewise.
3064 (vcondu<mode><v_cmp_mixed>): Likewise.
3065 (aarch64_cm<optab><mode>): Likewise.
3066 (aarch64_cmtst<mode>): Likewise.
3067 (aarch64_fac<optab><mode>): Likewise.
3068 (vec_perm_const<mode>): Likewise.
3069 (vcond_mask_<mode><v_cmp_result>): Rename to...
3070 (vcond_mask_<mode><v_int_equiv>): ...this.
3071 (vec_cmp<mode><v_cmp_result>): Rename to...
3072 (vec_cmp<mode><v_int_equiv>): ...this.
3074 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3075 Alan Hayward <alan.hayward@arm.com>
3076 David Sherwood <david.sherwood@arm.com>
3078 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
3080 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
3081 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
3082 (UNSPEC_LD4_DREG): New unspecs.
3083 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
3084 (aarch64_ld2<mode>_dreg_be): Replace with...
3085 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
3087 (aarch64_ld3<mode>_dreg_le)
3088 (aarch64_ld3<mode>_dreg_be): Replace with...
3089 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
3091 (aarch64_ld4<mode>_dreg_le)
3092 (aarch64_ld4<mode>_dreg_be): Replace with...
3093 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
3096 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3098 PR tree-optimization/81987
3099 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
3100 insert an initializer in a location not dominated by the stride
3103 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3105 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
3106 on the entire header of the finally block in the fallthru case.
3108 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3110 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
3112 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
3114 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
3115 rs6000_emit_move_from_cr and call renamed function.
3116 (rs6000_emit_prologue): Call renamed functions.
3117 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
3118 movesi_from_cr, remove volatile CRs.
3120 2017-08-30 Jon Beniston <jon@beniston.com>
3121 Richard Biener <rguenther@suse.de>
3123 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
3124 of VECTOR_MODE_P check.
3125 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
3126 element vector types.
3128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3130 * df.h (df_read_modify_subreg_p): Remove in favor of...
3131 * rtl.h (read_modify_subreg_p): ...this new function. Take a
3132 const_rtx instead of an rtx.
3133 * cprop.c (local_cprop_find_used_regs): Update accordingly.
3134 * df-problems.c (df_word_lr_mark_ref): Likewise.
3135 * ira-lives.c (mark_pseudo_reg_live): Likewise.
3136 (mark_pseudo_reg_dead): Likewise.
3137 (mark_ref_dead): Likewise.
3138 * reginfo.c (init_subregs_of_mode): Likewise.
3139 * sched-deps.c (sched_analyze_1): Likewise.
3140 * df-scan.c (df_def_record_1): Likewise.
3141 (df_uses_record): Likewise.
3142 (df_read_modify_subreg_p): Remove in favor of...
3143 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
3144 const_rtx instead of an rtx.
3146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3147 Alan Hayward <alan.hayward@arm.com>
3148 David Sherwood <david.sherwood@arm.com>
3150 * rtl.h (partial_subreg_p): New function.
3151 * caller-save.c (save_call_clobbered_regs): Use it.
3152 * calls.c (expand_call): Likewise.
3153 * combine.c (combinable_i3pat): Likewise.
3154 (simplify_set): Likewise.
3155 (make_extraction): Likewise.
3156 (make_compound_operation_int): Likewise.
3157 (gen_lowpart_or_truncate): Likewise.
3158 (force_to_mode): Likewise.
3159 (make_field_assignment): Likewise.
3160 (reg_truncated_to_mode): Likewise.
3161 (record_truncated_value): Likewise.
3162 (move_deaths): Likewise.
3163 * cse.c (record_jump_cond): Likewise.
3164 (cse_insn): Likewise.
3165 * cselib.c (cselib_lookup_1): Likewise.
3166 * expmed.c (extract_bit_field_using_extv): Likewise.
3167 * function.c (assign_parm_setup_reg): Likewise.
3168 * ifcvt.c (noce_convert_multiple_sets): Likewise.
3169 * ira-build.c (create_insn_allocnos): Likewise.
3170 * lra-coalesce.c (merge_pseudos): Likewise.
3171 * lra-constraints.c (match_reload): Likewise.
3172 (simplify_operand_subreg): Likewise.
3173 (curr_insn_transform): Likewise.
3174 * lra-lives.c (process_bb_lives): Likewise.
3175 * lra.c (new_insn_reg): Likewise.
3176 (lra_substitute_pseudo): Likewise.
3177 * regcprop.c (mode_change_ok): Likewise.
3178 (maybe_mode_change): Likewise.
3179 (copyprop_hardreg_forward_1): Likewise.
3180 * reload.c (push_reload): Likewise.
3181 (find_reloads): Likewise.
3182 (find_reloads_subreg_address): Likewise.
3183 * reload1.c (alter_reg): Likewise.
3184 (eliminate_regs_1): Likewise.
3185 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3187 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
3189 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
3190 back to if statements, including unpack.
3192 2017-08-30 Martin Liska <mliska@suse.cz>
3195 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
3197 (func_checker::compare_asm_inputs_outputs): ... this function.
3198 (func_checker::compare_gimple_asm): Use the function to compare
3199 also ASM constrains.
3200 * ipa-icf-gimple.h: Rename the function.
3202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3203 Alan Hayward <alan.hayward@arm.com>
3204 David Sherwood <david.sherwood@arm.com>
3206 * coretypes.h (complex_mode): New type.
3207 * gdbhooks.py (build_pretty_printer): Handle it.
3208 * machmode.h (complex_mode): New class.
3209 (complex_mode::includes_p): New function.
3210 (is_complex_int_mode): Likewise.
3211 (is_complex_float_mode): Likewise.
3212 * genmodes.c (get_mode_class): Handle complex mode classes.
3213 * function.c (expand_function_end): Use is_complex_int_mode.
3215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3216 Alan Hayward <alan.hayward@arm.com>
3217 David Sherwood <david.sherwood@arm.com>
3219 * coretypes.h (scalar_mode_pod): New typedef.
3220 * gdbhooks.py (build_pretty_printer): Handle it.
3221 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
3222 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
3223 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
3224 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
3227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3228 Alan Hayward <alan.hayward@arm.com>
3229 David Sherwood <david.sherwood@arm.com>
3231 * machmode.h (mode_for_vector): Take a scalar_mode instead
3233 * stor-layout.c (mode_for_vector): Likewise.
3234 * explow.c (promote_mode): Use as_a <scalar_mode>.
3235 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
3237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3238 Alan Hayward <alan.hayward@arm.com>
3239 David Sherwood <david.sherwood@arm.com>
3241 * target.def (preferred_simd_mode): Take a scalar_mode
3242 instead of a machine_mode.
3243 * targhooks.h (default_preferred_simd_mode): Likewise.
3244 * targhooks.c (default_preferred_simd_mode): Likewise.
3245 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
3246 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
3247 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
3248 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
3249 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
3250 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
3251 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
3252 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
3254 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
3255 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
3256 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
3257 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
3258 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
3259 * doc/tm.texi: Regenerate.
3260 * optabs-query.c (can_vec_mask_load_store_p): Return false for
3263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3264 Alan Hayward <alan.hayward@arm.com>
3265 David Sherwood <david.sherwood@arm.com>
3267 * target.def (scalar_mode_supported_p): Take a scalar_mode
3268 instead of a machine_mode.
3269 * targhooks.h (default_scalar_mode_supported_p): Likewise.
3270 * targhooks.c (default_scalar_mode_supported_p): Likewise.
3271 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
3272 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3273 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
3274 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
3275 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
3276 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
3277 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
3278 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
3279 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
3280 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
3281 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
3282 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
3284 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
3285 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
3286 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3287 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3288 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
3290 * doc/tm.texi: Regenerate.
3292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3293 Alan Hayward <alan.hayward@arm.com>
3294 David Sherwood <david.sherwood@arm.com>
3296 * coretypes.h (opt_scalar_mode): New typedef.
3297 * gdbhooks.py (build_pretty_printers): Handle it.
3298 * machmode.h (mode_iterator::get_2xwider): Add overload for
3300 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
3302 * expr.c (convert_mode_scalar): Likewise.
3303 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3304 * optabs.c (expand_float): Likewise.
3305 (expand_fix): Likewise.
3306 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3309 Alan Hayward <alan.hayward@arm.com>
3310 David Sherwood <david.sherwood@arm.com>
3312 * optabs.c (expand_float): Explicitly check for scalars before
3313 using a branching expansion.
3314 (expand_fix): Likewise.
3316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3317 Alan Hayward <alan.hayward@arm.com>
3318 David Sherwood <david.sherwood@arm.com>
3320 * expr.c (convert_mode): Split scalar handling out into...
3321 (convert_mode_scalar): ...this new function. Treat the modes
3324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3325 Alan Hayward <alan.hayward@arm.com>
3326 David Sherwood <david.sherwood@arm.com>
3328 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
3330 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
3332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3333 Alan Hayward <alan.hayward@arm.com>
3334 David Sherwood <david.sherwood@arm.com>
3336 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
3337 rather than a machine_mode.
3338 (fixed_from_string): Likewise.
3339 (fixed_convert): Likewise.
3340 (fixed_convert_from_int): Likewise.
3341 (fixed_convert_from_real): Likewise.
3342 (real_convert_from_fixed): Likewise.
3343 * fixed-value.c (fixed_from_double_int): Likewise.
3344 (fixed_from_string): Likewise.
3345 (fixed_convert): Likewise.
3346 (fixed_convert_from_int): Likewise.
3347 (fixed_convert_from_real): Likewise.
3348 (real_convert_from_fixed): Likewise.
3349 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
3351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3352 Alan Hayward <alan.hayward@arm.com>
3353 David Sherwood <david.sherwood@arm.com>
3355 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
3356 of separate mode class checks. Do not allow vector modes here.
3357 (immed_wide_int_const): Use as_a <scalar_mode>.
3358 * explow.c (trunc_int_for_mode): Likewise.
3359 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
3360 (wi::shwi): Likewise.
3361 (wi::min_value): Likewise.
3362 (wi::max_value): Likewise.
3363 * dwarf2out.c (loc_descriptor): Likewise.
3364 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
3367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3368 Alan Hayward <alan.hayward@arm.com>
3369 David Sherwood <david.sherwood@arm.com>
3371 * tree.h (SCALAR_TYPE_MODE): New macro.
3372 * expr.c (expand_expr_addr_expr_1): Use it.
3373 (expand_expr_real_2): Likewise.
3374 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
3375 (fold_convert_const_fixed_from_int): Likewise.
3376 (fold_convert_const_fixed_from_real): Likewise.
3377 (native_encode_fixed): Likewise
3378 (native_encode_complex): Likewise
3379 (native_encode_vector): Likewise.
3380 (native_interpret_fixed): Likewise.
3381 (native_interpret_real): Likewise.
3382 (native_interpret_complex): Likewise.
3383 (native_interpret_vector): Likewise.
3384 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
3385 (simd_clone_adjust_argument_types): Likewise.
3386 (simd_clone_init_simd_arrays): Likewise.
3387 (simd_clone_adjust): Likewise.
3388 * stor-layout.c (layout_type): Likewise.
3389 * tree.c (build_minus_one_cst): Likewise.
3390 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3391 * tree-inline.c (estimate_move_cost): Likewise.
3392 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
3393 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
3394 (vectorizable_reduction): Likewise.
3395 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
3396 (vect_recog_mixed_size_cond_pattern): Likewise.
3397 (check_bool_pattern): Likewise.
3398 (adjust_bool_pattern): Likewise.
3399 (search_type_for_mask_1): Likewise.
3400 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3401 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3402 (vectorizable_load): Likewise.
3403 (vectorizable_store): Likewise.
3404 * ubsan.c (ubsan_encode_value): Likewise.
3405 * varasm.c (output_constant): Likewise.
3407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3408 Alan Hayward <alan.hayward@arm.com>
3409 David Sherwood <david.sherwood@arm.com>
3411 * coretypes.h (scalar_mode): New class.
3412 * machmode.h (scalar_mode): Likewise.
3413 (scalar_mode::includes_p): New function.
3414 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
3415 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
3416 * genmodes.c (get_mode_class): Handle remaining scalar modes.
3417 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
3418 * expmed.c (store_bit_field_1): Likewise.
3419 (extract_bit_field_1): Likewise.
3420 * expr.c (write_complex_part): Likewise.
3421 (read_complex_part): Likewise.
3422 (emit_move_complex_push): Likewise.
3423 (expand_expr_real_2): Likewise.
3424 * function.c (assign_parm_setup_reg): Likewise.
3425 (assign_parms_unsplit_complex): Likewise.
3426 * optabs.c (expand_binop): Likewise.
3427 * rtlanal.c (subreg_get_info): Likewise.
3428 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3429 * varasm.c (output_constant_pool_2): Likewise.
3431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3432 Alan Hayward <alan.hayward@arm.com>
3433 David Sherwood <david.sherwood@arm.com>
3435 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3437 (expmed_mult_highpart_optab): Likewise.
3438 (expmed_mult_highpart): Likewise.
3440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3441 Alan Hayward <alan.hayward@arm.com>
3442 David Sherwood <david.sherwood@arm.com>
3444 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3445 instead of a machine_mode.
3446 (builtin_memset_read_str): Likewise.
3447 * builtins.c (c_readstr): Likewise.
3448 (builtin_memcpy_read_str): Likewise.
3449 (builtin_strncpy_read_str): Likewise.
3450 (builtin_memset_read_str): Likewise.
3451 (builtin_memset_gen_str): Likewise.
3452 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3453 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3454 instead of a machine_mode.
3455 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3457 (make_extraction): Likewise.
3458 (try_widen_shift_mode): Take and return scalar_int_modes instead
3460 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3461 a scalar_int_mode instead of a machine_mode.
3462 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3463 (avr_addr_space_pointer_mode): Likewise.
3464 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3465 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3466 (msp430_unwind_word_mode): Likewise.
3467 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3468 (spu_addr_space_pointer_mode): Likewise.
3469 (spu_addr_space_address_mode): Likewise.
3470 (spu_libgcc_cmp_return_mode): Likewise.
3471 (spu_libgcc_shift_count_mode): Likewise.
3472 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3473 (rl78_addr_space_pointer_mode): Likewise.
3474 (fl78_unwind_word_mode): Likewise.
3475 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3477 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3478 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3479 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3480 (mips_valid_pointer_mode): Likewise.
3481 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3482 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3483 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3485 (ft32_addr_space_address_mode): Likewise.
3486 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3487 scalar_int_mode instead of a machine_mode.
3488 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3490 (m32c_addr_space_address_mode): Likewise.
3491 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3492 (rs6000_eh_return_filter_mode): Likewise.
3493 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3494 (rs6000_eh_return_filter_mode): Likewise.
3495 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3496 (s390_libgcc_shift_count_mode): Likewise.
3497 (s390_unwind_word_mode): Likewise.
3498 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3500 * target.def (mode_rep_extended): Likewise.
3501 (valid_pointer_mode): Likewise.
3502 (addr_space.valid_pointer_mode): Likewise.
3503 (eh_return_filter_mode): Return a scalar_int_mode rather than
3505 (libgcc_cmp_return_mode): Likewise.
3506 (libgcc_shift_count_mode): Likewise.
3507 (unwind_word_mode): Likewise.
3508 (addr_space.pointer_mode): Likewise.
3509 (addr_space.address_mode): Likewise.
3510 * doc/tm.texi: Regenerate.
3511 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3513 (do_jump): Use scalar_int_mode for local variables.
3514 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3515 rather than a machine_mode.
3516 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3517 (scompare_loc_descriptor_wide): Likewise.
3518 (scompare_loc_descriptor_narrow): Likewise.
3519 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3521 * except.c (sjlj_emit_dispatch_table): Likewise.
3522 (expand_builtin_eh_copy_values): Likewise.
3523 * explow.c (convert_memory_address_addr_space_1): Likewise.
3524 Take a scalar_int_mode rather than a machine_mode.
3525 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3526 than a machine_mode.
3527 (memory_address_addr_space): Use scalar_int_mode for local variables.
3528 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3529 rather than a machine_mode.
3530 * expmed.c (mask_rtx): Likewise.
3531 (init_expmed_one_conv): Likewise.
3532 (expand_mult_highpart_adjust): Likewise.
3533 (extract_high_half): Likewise.
3534 (expmed_mult_highpart_optab): Likewise.
3535 (expmed_mult_highpart): Likewise.
3536 (expand_smod_pow2): Likewise.
3537 (expand_sdiv_pow2): Likewise.
3538 (emit_store_flag_int): Likewise.
3539 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3541 (extract_low_bits): Likewise.
3542 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3544 * expr.c (pieces_addr::adjust): Likewise.
3545 (can_store_by_pieces): Likewise.
3546 (store_by_pieces): Likewise.
3547 (clear_by_pieces_1): Likewise.
3548 (expand_expr_addr_expr_1): Likewise.
3549 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3550 (expand_expr_real_1): Likewise.
3551 (try_casesi): Likewise.
3552 * final.c (shorten_branches): Likewise.
3553 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3554 type of "mode" to machine_mode.
3555 * internal-fn.c (expand_arith_overflow_result_store): Take a
3556 scalar_int_mode rather than a machine_mode.
3557 (expand_mul_overflow): Use scalar_int_mode for local variables.
3558 * loop-doloop.c (doloop_modify): Likewise.
3559 (doloop_optimize): Likewise.
3560 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3561 than a machine_mode.
3562 (expand_doubleword_shift_condmove): Likewise.
3563 (expand_doubleword_shift): Likewise.
3564 (expand_doubleword_clz): Likewise.
3565 (expand_doubleword_popcount): Likewise.
3566 (expand_doubleword_parity): Likewise.
3567 (expand_absneg_bit): Use scalar_int_mode for local variables.
3568 (prepare_float_lib_cmp): Likewise.
3569 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3570 rather than a machine_mode.
3571 (convert_memory_address_addr_space): Likewise.
3572 (get_mode_bounds): Likewise.
3573 (get_address_mode): Return a scalar_int_mode rather than a
3575 * rtlanal.c (get_address_mode): Likewise.
3576 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3577 than a machine_mode.
3578 * targhooks.c (default_mode_rep_extended): Likewise.
3579 (default_valid_pointer_mode): Likewise.
3580 (default_addr_space_valid_pointer_mode): Likewise.
3581 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3582 than a machine_mode.
3583 (default_libgcc_cmp_return_mode): Likewise.
3584 (default_libgcc_shift_count_mode): Likewise.
3585 (default_unwind_word_mode): Likewise.
3586 (default_addr_space_pointer_mode): Likewise.
3587 (default_addr_space_address_mode): Likewise.
3588 * targhooks.h (default_eh_return_filter_mode): Likewise.
3589 (default_libgcc_cmp_return_mode): Likewise.
3590 (default_libgcc_shift_count_mode): Likewise.
3591 (default_unwind_word_mode): Likewise.
3592 (default_addr_space_pointer_mode): Likewise.
3593 (default_addr_space_address_mode): Likewise.
3594 (default_mode_rep_extended): Take a scalar_int_mode rather than
3596 (default_valid_pointer_mode): Likewise.
3597 (default_addr_space_valid_pointer_mode): Likewise.
3598 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3600 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3601 rather than a machine_mode.
3602 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3603 for local variables.
3604 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3605 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3606 than a machine_mode.
3608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3609 Alan Hayward <alan.hayward@arm.com>
3610 David Sherwood <david.sherwood@arm.com>
3612 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3613 the mode argument to scalar_int_mode.
3614 (do_jump_by_parts_zero_rtx): Likewise.
3615 (do_jump_by_parts_equality_rtx): Likewise.
3616 (do_jump_by_parts_greater): Take a mode argument.
3617 (do_jump_by_parts_equality): Likewise.
3618 (do_jump_1): Update calls accordingly.
3620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3621 Alan Hayward <alan.hayward@arm.com>
3622 David Sherwood <david.sherwood@arm.com>
3624 * is-a.h (safe_dyn_cast): New function.
3625 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3626 (jump_table_for_label): Likewise.
3627 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3628 instead of an rtx_insn *.
3629 (shorten_branches): Use dyn_cast instead of LABEL_P and
3630 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3631 rtx_jump_table_data::get_data_mode.
3632 (final_scan_insn): Likewise.
3634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3635 Alan Hayward <alan.hayward@arm.com>
3636 David Sherwood <david.sherwood@arm.com>
3638 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3639 trying to combine a full-register integer set with a subreg
3642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3643 Alan Hayward <alan.hayward@arm.com>
3644 David Sherwood <david.sherwood@arm.com>
3646 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3647 that is always either address_mode or pointer_mode.
3649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3650 Alan Hayward <alan.hayward@arm.com>
3651 David Sherwood <david.sherwood@arm.com>
3653 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
3654 when the two are known to be equal.
3656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3657 Alan Hayward <alan.hayward@arm.com>
3658 David Sherwood <david.sherwood@arm.com>
3660 * simplify-rtx.c (simplify_const_unary_operation): Use
3661 is_a <scalar_int_mode> instead of checking for a nonzero
3662 precision. Forcibly convert op_mode to a scalar_int_mode
3663 in that case. More clearly differentiate the operand and
3664 result modes and use the former when deciding what the value
3665 of a count-bits operation should be. Use is_int_mode instead
3666 of checking for a MODE_INT. Remove redundant check for whether
3667 this mode has a zero precision.
3669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3670 Alan Hayward <alan.hayward@arm.com>
3671 David Sherwood <david.sherwood@arm.com>
3673 * optabs.c (widen_leading): Change the type of the mode argument
3674 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
3675 (widen_bswap): Likewise.
3676 (expand_parity): Likewise.
3677 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
3678 (expand_ffs): Likewise.
3679 (epand_unop): Check for scalar integer modes before calling the
3682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3683 Alan Hayward <alan.hayward@arm.com>
3684 David Sherwood <david.sherwood@arm.com>
3686 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3688 (expand_expr_real_1): Update call accordingly.
3690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3691 Alan Hayward <alan.hayward@arm.com>
3692 David Sherwood <david.sherwood@arm.com>
3694 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3695 value_mode arguments. Use scalar_int_mode internally.
3696 (store_bit_field_1): Rename the new integer mode from imode
3697 to op0_mode and use it instead of GET_MODE (op0). Update calls
3698 to store_split_bit_field, store_bit_field_using_insv and
3699 store_fixed_bit_field.
3700 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3701 Use scalar_int_mode internally. Use a bit count rather than a mode
3702 when calculating the largest bit size for get_best_mode.
3703 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3704 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3705 Remove assertion that OP0 has a scalar integer mode.
3706 (store_split_bit_field): Add op0_mode and value_mode arguments.
3707 Update calls to extract_fixed_bit_field.
3708 (extract_bit_field_using_extv): Add an op0_mode argument.
3709 Use scalar_int_mode internally.
3710 (extract_bit_field_1): Rename the new integer mode from imode to
3711 op0_mode and use it instead of GET_MODE (op0). Update calls to
3712 extract_split_bit_field, extract_bit_field_using_extv and
3713 extract_fixed_bit_field.
3714 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3715 to extract_split_bit_field and extract_fixed_bit_field_1.
3716 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3717 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3719 (extract_split_bit_field): Add an op0_mode argument. Update call
3720 to extract_fixed_bit_field.
3722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3723 Alan Hayward <alan.hayward@arm.com>
3724 David Sherwood <david.sherwood@arm.com>
3726 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3727 * explow.c (hard_function_value): Likewise.
3728 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3729 convert_to_mode call outside the loop.
3730 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3731 for the mode iterator. Require the mode specified by max_pieces
3733 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3735 (copy_blkmode_to_reg): Likewise.
3736 (set_storage_via_setmem): Likewise.
3737 * optabs.c (prepare_cmp_insn): Likewise.
3738 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3739 * stor-layout.c (finish_bitfield_representative): Likewise.
3741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3742 Alan Hayward <alan.hayward@arm.com>
3743 David Sherwood <david.sherwood@arm.com>
3745 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3746 * expr.c (convert_move): Use them.
3747 (convert_modes): Likewise.
3748 (store_expr_with_bounds): Likewise.
3750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3751 Alan Hayward <alan.hayward@arm.com>
3752 David Sherwood <david.sherwood@arm.com>
3754 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3755 parameter for the mode of "x". Remove the "known_x", "known_mode"
3756 and "known_ret" arguments. Change the type of the mode argument
3758 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3759 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3760 (reg_num_sign_bit_copies_for_combine): Likewise.
3761 * rtlanal.c (nonzero_bits1): Likewise.
3762 (num_sign_bit_copies1): Likewise.
3763 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3764 (reg_num_sign_bit_copies_general): Likewise.
3765 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3766 (reg_nonzero_bits_general): Likewise.
3768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3769 Alan Hayward <alan.hayward@arm.com>
3770 David Sherwood <david.sherwood@arm.com>
3772 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
3773 than in subroutines. Return 1 for non-integer modes.
3774 (cached_num_sign_bit_copies): Change the type of the mode parameter
3776 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
3777 classes. Handle CONST_INT_P first and then check whether X also
3778 has a scalar integer mode. Check the same thing for inner registers
3779 of a SUBREG and for values that are being extended or truncated.
3781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3782 Alan Hayward <alan.hayward@arm.com>
3783 David Sherwood <david.sherwood@arm.com>
3785 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
3786 in subroutines. Return the mode mask for non-integer modes.
3787 (cached_nonzero_bits): Change the type of the mode parameter
3789 (nonzero_bits1): Likewise. Remove early exit for other mode
3790 classes. Handle CONST_INT_P first and then check whether X
3791 also has a scalar integer mode.
3793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3794 Alan Hayward <alan.hayward@arm.com>
3795 David Sherwood <david.sherwood@arm.com>
3797 * expr.c (widest_int_mode_for_size): Make the comment match the code.
3798 Return a scalar_int_mode and assert that the size is greater than
3800 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
3801 (op_by_pieces_d::op_by_pieces_d): Likewise.
3802 (op_by_pieces_d::run): Likewise.
3803 (can_store_by_pieces): Likewise.
3805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3806 Alan Hayward <alan.hayward@arm.com>
3807 David Sherwood <david.sherwood@arm.com>
3809 * combine.c (extract_left_shift): Add a mode argument and update
3811 (make_compound_operation_int): Change the type of the mode parameter
3812 to scalar_int_mode and update the call to extract_left_shift.
3814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3815 Alan Hayward <alan.hayward@arm.com>
3816 David Sherwood <david.sherwood@arm.com>
3818 * combine.c (simplify_and_const_int): Change the type of the mode
3819 parameter to scalar_int_mode.
3820 (simplify_and_const_int_1): Likewise. Update recursive call.
3822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3823 Alan Hayward <alan.hayward@arm.com>
3824 David Sherwood <david.sherwood@arm.com>
3826 * combine.c (simplify_compare_const): Check that the mode is a
3827 scalar_int_mode (rather than VOIDmode) before testing its
3829 (simplify_comparison): Move COMPARISON_P handling out of the
3830 loop and restrict the latter part of the loop to scalar_int_modes.
3831 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
3832 and when considering SUBREG_REGs. Use is_int_mode instead of
3833 checking GET_MODE_CLASS against MODE_INT.
3835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3836 Alan Hayward <alan.hayward@arm.com>
3837 David Sherwood <david.sherwood@arm.com>
3839 * combine.c (try_widen_shift_mode): Move check for equal modes to...
3840 (simplify_shift_const_1): ...here. Use scalar_int_mode for
3841 shift_unit_mode and for modes involved in scalar shifts.
3843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3844 Alan Hayward <alan.hayward@arm.com>
3845 David Sherwood <david.sherwood@arm.com>
3847 * combine.c (force_int_to_mode): New function, split out from...
3848 (force_to_mode): ...here. Keep xmode up-to-date and use it
3849 instead of GET_MODE (x).
3851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3852 Alan Hayward <alan.hayward@arm.com>
3853 David Sherwood <david.sherwood@arm.com>
3855 * optabs-query.h (extraction_insn::struct_mode): Change type to
3856 opt_scalar_int_mode and update comment.
3857 (extraction_insn::field_mode): Change type to scalar_int_mode.
3858 (extraction_insn::pos_mode): Likewise.
3859 * combine.c (make_extraction): Update accordingly.
3860 * optabs-query.c (get_traditional_extraction_insn): Likewise.
3861 (get_optab_extraction_insn): Likewise.
3862 * recog.c (simplify_while_replacing): Likewise.
3863 * expmed.c (narrow_bit_field_mem): Change the type of the mode
3864 parameter to opt_scalar_int_mode.
3866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3867 Alan Hayward <alan.hayward@arm.com>
3868 David Sherwood <david.sherwood@arm.com>
3870 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
3871 to a scalar_int_mode instead of a machine_mode.
3872 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
3873 (get_best_mode): Return a boolean and use a pointer argument to store
3874 the selected mode. Replace the limit mode parameter with a bit limit.
3875 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
3876 for the values returned by bit_field_mode_iterator::next_mode.
3877 (store_bit_field): Update call to get_best_mode.
3878 (store_fixed_bit_field): Likewise.
3879 (extract_fixed_bit_field): Likewise.
3880 * expr.c (optimize_bitfield_assignment_op): Likewise.
3881 * fold-const.c (optimize_bit_field_compare): Likewise.
3882 (fold_truth_andor_1): Likewise.
3883 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
3884 Update for new type of m_mode.
3885 (get_best_mode): As above.
3887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3888 Alan Hayward <alan.hayward@arm.com>
3889 David Sherwood <david.sherwood@arm.com>
3891 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
3892 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
3893 (store_bit_field): Check is_a <scalar_int_mode> before calling
3894 strict_volatile_bitfield_p.
3895 (extract_bit_field): Likewise.
3897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3898 Alan Hayward <alan.hayward@arm.com>
3899 David Sherwood <david.sherwood@arm.com>
3901 * target.def (cstore_mode): Return a scalar_int_mode.
3902 * doc/tm.texi: Regenerate.
3903 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
3904 * targhooks.h (default_cstore_mode): Likewise.
3905 * targhooks.c (default_cstore_mode): Likewise, using a forced
3907 * expmed.c (emit_cstore): Expect the target of the cstore to be
3910 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3911 Alan Hayward <alan.hayward@arm.com>
3912 David Sherwood <david.sherwood@arm.com>
3914 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
3916 (niter_desc): Likewise mode.
3917 (iv_analyze): Add a mode parameter.
3919 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
3920 and change its type to scalar_int_mode.
3921 * loop-iv.c: Update commentary at head of file.
3922 (iv_constant): Pass the mode paraeter before the rtx it describes
3923 and change its type to scalar_int_mode. Remove VOIDmode handling.
3924 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
3925 (iv_extend): Likewise.
3926 (shorten_into_mode): Likewise.
3927 (iv_add): Use scalar_int_mode.
3928 (iv_mult): Likewise.
3929 (iv_shift): Likewise.
3930 (canonicalize_iv_subregs): Likewise.
3931 (get_biv_step_1): Pass the outer_mode parameter before the rtx
3932 it describes and change its mode to scalar_int_mode. Also change
3933 the type of the returned inner_mode to scalar_int_mode.
3934 (get_biv_step): Likewise, turning outer_mode from a pointer
3935 into a direct parameter. Update call to get_biv_step_1.
3936 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
3937 iv_constant and get_biv_step.
3938 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
3939 and change its type to scalar_int_mode. Don't initialise iv->mode
3940 to VOIDmode and remove later checks for its still being VOIDmode.
3941 Update calls to iv_analyze_op and iv_analyze_expr. Check
3942 is_a <scalar_int_mode> when changing the mode under consideration.
3943 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
3944 Update call to iv_analyze_expr.
3945 (iv_analyze_op): Add a mode parameter. Reject subregs whose
3946 inner register is not also a scalar_int_mode. Update call to
3948 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
3949 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
3950 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
3951 separate mode class checks. Update calls to iv_analyze. Remove
3952 fix-up of VOIDmodes after iv_analyze_biv.
3953 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
3954 don't have a scalar_int_mode. Update call to biv_p.
3956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3957 Alan Hayward <alan.hayward@arm.com>
3958 David Sherwood <david.sherwood@arm.com>
3960 * cfgexpand.c (convert_debug_memory_address): Use
3961 as_a <scalar_int_mode>.
3962 * combine.c (expand_compound_operation): Likewise.
3963 (make_extraction): Likewise.
3964 (change_zero_ext): Likewise.
3965 (simplify_comparison): Likewise.
3966 * cse.c (cse_insn): Likewise.
3967 * dwarf2out.c (minmax_loc_descriptor): Likewise.
3968 (mem_loc_descriptor): Likewise.
3969 (loc_descriptor): Likewise.
3970 * expmed.c (init_expmed_one_mode): Likewise.
3971 (synth_mult): Likewise.
3972 (emit_store_flag_1): Likewise.
3973 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
3974 of a comparison with size.
3975 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
3976 (reduce_to_bit_field_precision): Likewise.
3977 * function.c (expand_function_end): Likewise.
3978 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
3979 * loop-doloop.c (doloop_modify): Likewise.
3980 * optabs.c (expand_binop): Likewise.
3981 (expand_unop): Likewise.
3982 (expand_copysign_absneg): Likewise.
3983 (prepare_cmp_insn): Likewise.
3984 (maybe_legitimize_operand): Likewise.
3985 * recog.c (const_scalar_int_operand): Likewise.
3986 * rtlanal.c (get_address_mode): Likewise.
3987 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3988 (simplify_cond_clz_ctz): Likewise.
3989 * tree-nested.c (get_nl_goto_field): Likewise.
3990 * tree.c (build_vector_type_for_mode): Likewise.
3991 * var-tracking.c (use_narrower_mode): Likewise.
3993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3994 Alan Hayward <alan.hayward@arm.com>
3995 David Sherwood <david.sherwood@arm.com>
3997 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
3998 * builtins.c (expand_builtin_signbit): Use it.
3999 * cfgexpand.c (expand_debug_expr): Likewise.
4000 * dojump.c (do_jump): Likewise.
4001 (do_compare_and_jump): Likewise.
4002 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
4003 * expmed.c (make_tree): Likewise.
4004 * expr.c (expand_expr_real_2): Likewise.
4005 (expand_expr_real_1): Likewise.
4006 (try_casesi): Likewise.
4007 * fold-const-call.c (fold_const_call_ss): Likewise.
4008 * fold-const.c (unextend): Likewise.
4009 (extract_muldiv_1): Likewise.
4010 (fold_single_bit_test): Likewise.
4011 (native_encode_int): Likewise.
4012 (native_encode_string): Likewise.
4013 (native_interpret_int): Likewise.
4014 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
4015 * internal-fn.c (expand_addsub_overflow): Likewise.
4016 (expand_neg_overflow): Likewise.
4017 (expand_mul_overflow): Likewise.
4018 (expand_arith_overflow): Likewise.
4019 * match.pd: Likewise.
4020 * stor-layout.c (layout_type): Likewise.
4021 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
4022 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
4023 * tree-ssanames.c (get_range_info): Likewise.
4024 * tree-switch-conversion.c (array_value_type) Likewise.
4025 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
4026 (vect_recog_divmod_pattern): Likewise.
4027 (vect_recog_mixed_size_cond_pattern): Likewise.
4028 * tree-vrp.c (extract_range_basic): Likewise.
4029 (simplify_float_conversion_using_ranges): Likewise.
4030 * tree.c (int_fits_type_p): Likewise.
4031 * ubsan.c (instrument_bool_enum_load): Likewise.
4032 * varasm.c (mergeable_string_section): Likewise.
4033 (narrowing_initializer_constant_valid_p): Likewise.
4034 (output_constant): Likewise.
4036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4037 Alan Hayward <alan.hayward@arm.com>
4038 David Sherwood <david.sherwood@arm.com>
4040 * machmode.h (NARROWEST_INT_MODE): New macro.
4041 * expr.c (alignment_for_piecewise_move): Use it instead of
4042 GET_CLASS_NARROWEST_MODE (MODE_INT).
4043 (push_block): Likewise.
4044 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4046 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4049 Alan Hayward <alan.hayward@arm.com>
4050 David Sherwood <david.sherwood@arm.com>
4052 * postreload.c (move2add_valid_value_p): Change the type of the
4053 mode parameter to scalar_int_mode.
4054 (move2add_use_add2_insn): Add a mode parameter and use it instead
4056 (move2add_use_add3_insn): Likewise.
4057 (reload_cse_move2add): Update accordingly.
4059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4060 Alan Hayward <alan.hayward@arm.com>
4061 David Sherwood <david.sherwood@arm.com>
4063 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
4065 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
4066 * optabs.c (expand_unop): Likewise.
4068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4069 Alan Hayward <alan.hayward@arm.com>
4070 David Sherwood <david.sherwood@arm.com>
4072 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
4073 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
4074 (popcount_loc_descriptor): Likewise.
4075 (bswap_loc_descriptor): Likewise.
4076 (rotate_loc_descriptor): Likewise.
4077 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
4078 calling the functions above.
4080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4081 Alan Hayward <alan.hayward@arm.com>
4082 David Sherwood <david.sherwood@arm.com>
4084 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
4086 (try_combine): Likewise.
4087 (simplify_if_then_else): Likewise.
4088 * cse.c (cse_insn): Likewise.
4089 * dwarf2out.c (mem_loc_descriptor): Likewise.
4090 * emit-rtl.c (gen_lowpart_common): Likewise.
4091 * simplify-rtx.c (simplify_truncation): Likewise.
4092 (simplify_binary_operation_1): Likewise.
4093 (simplify_const_relational_operation): Likewise.
4094 (simplify_ternary_operation): Likewise.
4095 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
4097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4098 Alan Hayward <alan.hayward@arm.com>
4099 David Sherwood <david.sherwood@arm.com>
4101 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
4102 * reload.c (push_reload): Likewise.
4103 (find_reloads): Likewise.
4105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4106 Alan Hayward <alan.hayward@arm.com>
4107 David Sherwood <david.sherwood@arm.com>
4109 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
4110 (make_compound_operation_int): Likewise.
4111 (change_zero_ext): Likewise.
4112 * expr.c (convert_move): Likewise.
4113 (convert_modes): Likewise.
4114 * fwprop.c (forward_propagate_subreg): Likewise.
4115 * loop-iv.c (get_biv_step_1): Likewise.
4116 * optabs.c (widen_operand): Likewise.
4117 * postreload.c (move2add_valid_value_p): Likewise.
4118 * recog.c (simplify_while_replacing): Likewise.
4119 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4120 (simplify_binary_operation_1): Likewise. Remove redundant
4121 mode equality check.
4123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4124 Alan Hayward <alan.hayward@arm.com>
4125 David Sherwood <david.sherwood@arm.com>
4127 * combine.c (combine_simplify_rtx): Add checks for
4128 is_a <scalar_int_mode>.
4129 (simplify_if_then_else): Likewise.
4130 (make_field_assignment): Likewise.
4131 (simplify_comparison): Likewise.
4132 * ifcvt.c (noce_try_bitop): Likewise.
4133 * loop-invariant.c (canonicalize_address_mult): Likewise.
4134 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4137 Alan Hayward <alan.hayward@arm.com>
4138 David Sherwood <david.sherwood@arm.com>
4140 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
4141 is_a <scalar_int_mode> instead of != BLKmode.
4143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4144 Alan Hayward <alan.hayward@arm.com>
4145 David Sherwood <david.sherwood@arm.com>
4147 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
4148 instead of != VOIDmode.
4149 * combine.c (if_then_else_cond): Likewise.
4150 (change_zero_ext): Likewise.
4151 * dwarf2out.c (mem_loc_descriptor): Likewise.
4152 (loc_descriptor): Likewise.
4153 * rtlanal.c (canonicalize_condition): Likewise.
4154 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
4156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4157 Alan Hayward <alan.hayward@arm.com>
4158 David Sherwood <david.sherwood@arm.com>
4160 * simplify-rtx.c (simplify_binary_operation_1): Use
4161 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
4163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4164 Alan Hayward <alan.hayward@arm.com>
4165 David Sherwood <david.sherwood@arm.com>
4167 * wide-int.h (int_traits<unsigned char>) New class.
4168 (int_traits<unsigned short>) Likewise.
4169 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
4170 Use GET_MODE_UNIT_PRECISION and remove redundant test for
4172 * combine.c (set_nonzero_bits_and_sign_copies): Use
4173 is_a <scalar_int_mode>.
4174 (find_split_point): Likewise.
4175 (combine_simplify_rtx): Likewise.
4176 (simplify_logical): Likewise.
4177 (expand_compound_operation): Likewise.
4178 (expand_field_assignment): Likewise.
4179 (make_compound_operation): Likewise.
4180 (extended_count): Likewise.
4181 (change_zero_ext): Likewise.
4182 (simplify_comparison): Likewise.
4183 * dwarf2out.c (scompare_loc_descriptor): Likewise.
4184 (ucompare_loc_descriptor): Likewise.
4185 (minmax_loc_descriptor): Likewise.
4186 (mem_loc_descriptor): Likewise.
4187 (loc_descriptor): Likewise.
4188 * expmed.c (init_expmed_one_mode): Likewise.
4189 * lra-constraints.c (lra_constraint_offset): Likewise.
4190 * optabs.c (prepare_libcall_arg): Likewise.
4191 * postreload.c (move2add_note_store): Likewise.
4192 * reload.c (operands_match_p): Likewise.
4193 * rtl.h (load_extend_op): Likewise.
4194 * rtlhooks.c (gen_lowpart_general): Likewise.
4195 * simplify-rtx.c (simplify_truncation): Likewise.
4196 (simplify_unary_operation_1): Likewise.
4197 (simplify_binary_operation_1): Likewise.
4198 (simplify_const_binary_operation): Likewise.
4199 (simplify_const_relational_operation): Likewise.
4200 (simplify_subreg): Likewise.
4201 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4202 * var-tracking.c (adjust_mems): Likewise.
4203 (prepare_call_arguments): Likewise.
4205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4206 Alan Hayward <alan.hayward@arm.com>
4207 David Sherwood <david.sherwood@arm.com>
4209 * machmode.h (is_int_mode): New fuction.
4210 * combine.c (find_split_point): Use it.
4211 (combine_simplify_rtx): Likewise.
4212 (simplify_if_then_else): Likewise.
4213 (simplify_set): Likewise.
4214 (simplify_shift_const_1): Likewise.
4215 (simplify_comparison): Likewise.
4216 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
4217 * cse.c (notreg_cost): Likewise.
4218 (cse_insn): Likewise.
4219 * cselib.c (cselib_lookup_1): Likewise.
4220 * dojump.c (do_jump_1): Likewise.
4221 (do_compare_rtx_and_jump): Likewise.
4222 * dse.c (get_call_args): Likewise.
4223 * dwarf2out.c (rtl_for_decl_init): Likewise.
4224 (native_encode_initializer): Likewise.
4225 * expmed.c (emit_store_flag_1): Likewise.
4226 (emit_store_flag): Likewise.
4227 * expr.c (convert_modes): Likewise.
4228 (store_field): Likewise.
4229 (expand_expr_real_1): Likewise.
4230 * fold-const.c (fold_read_from_constant_string): Likewise.
4231 * gimple-ssa-sprintf.c (get_format_string): Likewise.
4232 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
4233 * optabs.c (expand_binop): Likewise.
4234 (expand_unop): Likewise.
4235 (expand_abs_nojump): Likewise.
4236 (expand_one_cmpl_abs_nojump): Likewise.
4237 * simplify-rtx.c (mode_signbit_p): Likewise.
4238 (val_signbit_p): Likewise.
4239 (val_signbit_known_set_p): Likewise.
4240 (val_signbit_known_clear_p): Likewise.
4241 (simplify_relational_operation_1): Likewise.
4242 * tree.c (vector_type_mode): Likewise.
4244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4245 Alan Hayward <alan.hayward@arm.com>
4246 David Sherwood <david.sherwood@arm.com>
4248 * machmode.h (smallest_mode_for_size): Fix formatting.
4249 (smallest_int_mode_for_size): New function.
4250 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
4251 instead of smallest_mode_for_size.
4252 * combine.c (make_extraction): Likewise.
4253 * config/arc/arc.c (arc_expand_movmem): Likewise.
4254 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
4255 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4256 * config/s390/s390.c (s390_expand_insv): Likewise.
4257 * config/sparc/sparc.c (assign_int_registers): Likewise.
4258 * config/spu/spu.c (spu_function_value): Likewise.
4259 (spu_function_arg): Likewise.
4260 * coverage.c (get_gcov_type): Likewise.
4261 (get_gcov_unsigned_t): Likewise.
4262 * dse.c (find_shift_sequence): Likewise.
4263 * expmed.c (store_bit_field_1): Likewise.
4264 * expr.c (convert_move): Likewise.
4265 (store_field): Likewise.
4266 * internal-fn.c (expand_arith_overflow): Likewise.
4267 * optabs-query.c (get_best_extraction_insn): Likewise.
4268 * optabs.c (expand_twoval_binop_libfunc): Likewise.
4269 * stor-layout.c (layout_type): Likewise.
4270 (initialize_sizetypes): Likewise.
4271 * targhooks.c (default_get_mask_mode): Likewise.
4272 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
4274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4275 Alan Hayward <alan.hayward@arm.com>
4276 David Sherwood <david.sherwood@arm.com>
4278 * machmode.h (opt_mode::else_blk): New function.
4279 (int_mode_for_mode): Declare.
4280 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
4281 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
4283 * cfgexpand.c (expand_debug_expr): Likewise.
4284 * combine.c (gen_lowpart_or_truncate): Likewise.
4285 (gen_lowpart_for_combine): Likewise.
4286 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
4287 * config/avr/avr.c (avr_to_int_mode): Likewise.
4288 (avr_out_plus_1): Likewise.
4289 (avr_out_plus): Likewise.
4290 (avr_out_round): Likewise.
4291 * config/i386/i386.c (ix86_split_to_parts): Likewise.
4292 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4293 (s390_expand_vcond): Likewise.
4294 * config/spu/spu.c (spu_split_immediate): Likewise.
4295 (spu_expand_mov): Likewise.
4296 * dse.c (get_stored_val): Likewise.
4297 * expmed.c (store_bit_field_1): Likewise.
4298 (convert_extracted_bit_field): Use int_mode_for_mode instead of
4300 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
4301 (extract_low_bits): Likewise.
4302 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
4303 handling rather than repeating the check.
4304 (emit_group_store): Likewise.
4305 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
4306 * optabs.c (expand_absneg_bit): Likewise.
4307 (expand_copysign_absneg): Likewise.
4308 (expand_copysign_bit): Likewise.
4309 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4310 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
4311 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
4312 * var-tracking.c (prepare_call_arguments): Likewise.
4313 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
4314 int_mode_for_mode instead of mode_for_size.
4315 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4318 Alan Hayward <alan.hayward@arm.com>
4319 David Sherwood <david.sherwood@arm.com>
4321 * machmode.h (int_mode_for_size): New function.
4322 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
4323 instead of mode_for_size.
4324 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
4326 * combine.c (expand_field_assignment): Use int_mode_for_size
4327 instead of mode_for_size.
4328 (make_extraction): Likewise.
4329 (simplify_shift_const_1): Likewise.
4330 (simplify_comparison): Likewise.
4331 * dojump.c (do_jump): Likewise.
4332 * dwarf2out.c (mem_loc_descriptor): Likewise.
4333 * emit-rtl.c (init_derived_machine_modes): Likewise.
4334 * expmed.c (flip_storage_order): Likewise.
4335 (convert_extracted_bit_field): Likewise.
4336 * expr.c (copy_blkmode_from_reg): Likewise.
4337 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
4338 * internal-fn.c (expand_mul_overflow): Likewise.
4339 * lower-subreg.c (simple_move): Likewise.
4340 * optabs-libfuncs.c (init_optabs): Likewise.
4341 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4342 * tree.c (vector_type_mode): Likewise.
4343 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
4344 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
4345 * tree-vect-generic.c (expand_vector_parallel): Likewise.
4346 * tree-vect-stmts.c (vectorizable_load): Likewise.
4347 (vectorizable_store): Likewise.
4349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4350 Alan Hayward <alan.hayward@arm.com>
4351 David Sherwood <david.sherwood@arm.com>
4353 * coretypes.h (pod_mode): New type.
4354 (scalar_int_mode_pod): New typedef.
4355 * machmode.h (pod_mode): New class.
4356 (int_n_data_t::m): Change type to scalar_int_mode_pod.
4357 * genmodes.c (emit_mode_int_n): Update accordingly.
4358 * lower-subreg.h (target_lower_subreg): Change type to
4359 scalar_int_mode_pod.
4360 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
4361 scalar_int_mode_pod.
4363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4364 Alan Hayward <alan.hayward@arm.com>
4365 David Sherwood <david.sherwood@arm.com>
4367 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
4368 machine_mode to scalar_int_mode.
4369 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
4370 (rs6000_option_override_internal): Remove cast to int.
4371 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
4372 machine_mode to scalar_int_mode.
4373 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
4374 (rs6000_option_override_internal): Remove cast to int.
4375 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
4376 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
4378 * config/s390/s390.c (s390_expand_builtin): Likewise.
4379 * coretypes.h (scalar_int_mode): New type.
4380 (opt_scalar_int_mode): New typedef.
4381 * machmode.h (scalar_int_mode): New class.
4382 (scalar_int_mode::includes_p): New function.
4383 (byte_mode): Change type to scalar_int_mode.
4384 (word_mode): Likewise.
4385 (ptr_mode): Likewise.
4386 * emit-rtl.c (byte_mode): Likewise.
4387 (word_mode): Likewise.
4388 (ptr_mode): Likewise.
4389 (init_derived_machine_modes): Update accordingly.
4390 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
4391 and MODE_PARTIAL_INT.
4392 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
4393 opt_scalar_int_mode.
4395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4396 Alan Hayward <alan.hayward@arm.com>
4397 David Sherwood <david.sherwood@arm.com>
4399 * target.def (libgcc_floating_mode_supported_p): Take a
4401 * doc/tm.texi: Regenerate.
4402 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
4404 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4405 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
4408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4409 Alan Hayward <alan.hayward@arm.com>
4410 David Sherwood <david.sherwood@arm.com>
4412 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
4413 * doc/tm.texi: Regenerate.
4414 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
4415 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
4416 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
4417 * targhooks.h (default_floatn_mode): Likewise.
4418 * targhooks.c (default_floatn_mode): Likewise.
4419 * tree.c (build_common_tree_nodes): Update accordingly.
4421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4422 Alan Hayward <alan.hayward@arm.com>
4423 David Sherwood <david.sherwood@arm.com>
4425 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
4426 (mode_iterator::iterate_p): Likewise.
4427 (mode_iterator::get_wider): Likewise.
4428 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4430 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4431 Alan Hayward <alan.hayward@arm.com>
4432 David Sherwood <david.sherwood@arm.com>
4434 * coretypes.h (opt_scalar_float_mode): New typedef.
4435 * machmode.h (float_mode_for_size): New function.
4436 * emit-rtl.c (double_mode): Delete.
4437 (init_emit_once): Use float_mode_for_size.
4438 * stor-layout.c (layout_type): Likewise.
4439 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4442 Alan Hayward <alan.hayward@arm.com>
4443 David Sherwood <david.sherwood@arm.com>
4445 * output.h (assemble_real): Take a scalar_float_mode.
4446 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4447 * config/arm/arm.md (consttable_4): Likewise.
4448 (consttable_8): Likewise.
4449 (consttable_16): Likewise.
4450 * config/mips/mips.md (consttable_float): Likewise.
4451 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4452 * varasm.c (assemble_real): Take a scalar_float_mode.
4453 (output_constant_pool_2): Update accordingly.
4454 (output_constant): Likewise.
4456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4457 Alan Hayward <alan.hayward@arm.com>
4458 David Sherwood <david.sherwood@arm.com>
4460 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4461 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4462 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4463 (native_encode_real): Likewise.
4464 (native_interpret_real): Likewise.
4465 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4466 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4469 Alan Hayward <alan.hayward@arm.com>
4470 David Sherwood <david.sherwood@arm.com>
4472 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4473 <scalar_float_mode>. Simplify.
4474 (gen_extend_conv_libfunc): Likewise.
4476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4477 Alan Hayward <alan.hayward@arm.com>
4478 David Sherwood <david.sherwood@arm.com>
4480 * coretypes.h (scalar_float_mode): New type.
4481 * machmode.h (mode_traits::from_int): Use machine_mode if
4482 USE_ENUM_MODES is defined.
4483 (is_a): New function.
4485 (dyn_cast): Likewise.
4486 (scalar_float_mode): New class.
4487 (scalar_float_mode::includes_p): New function.
4488 (is_float_mode): Likewise.
4489 * gdbhooks.py (MachineModePrinter): New class.
4490 (build_pretty_printer): Use it for scalar_float_mode.
4491 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4492 (format_helper::format_helper): Turn into a template.
4493 * genmodes.c (get_mode_class): New function.
4494 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4495 or machine_mode if none.
4496 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4497 as_a <scalar_float_mode>.
4498 * dwarf2out.c (mem_loc_descriptor): Likewise.
4499 (insert_float): Likewise.
4500 (add_const_value_attribute): Likewise.
4501 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4502 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4503 (expand_unop): Update accordingly.
4504 (expand_abs_nojump): Likewise.
4505 (expand_copysign_absneg): Take a scalar_float_mode.
4506 (expand_copysign_bit): Likewise.
4507 (expand_copysign): Update accordingly.
4509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4510 Alan Hayward <alan.hayward@arm.com>
4511 David Sherwood <david.sherwood@arm.com>
4513 * coretypes.h (opt_mode): New class.
4514 * machmode.h (opt_mode): Likewise.
4515 (opt_mode::else_void): New function.
4516 (opt_mode::require): Likewise.
4517 (opt_mode::exists): Likewise.
4518 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4519 (GET_MODE_2XWIDER_MODE): Likewise.
4520 (mode_iterator::get_wider): Update accordingly.
4521 (mode_iterator::get_2xwider): Likewise.
4522 (mode_iterator::get_known_wider): Likewise, turning into a template.
4523 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4524 forcing a wider mode to exist.
4525 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4526 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4527 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4528 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4529 * lower-subreg.c (init_lower_subreg): Likewise.
4530 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4531 on the final iteration.
4532 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4533 a wider mode exists before asking for a move pattern.
4534 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4535 forcing a wider mode to exist.
4536 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4537 returning false if no such mode exists.
4538 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4539 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4540 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4541 Avoid checking for a MODE_INT if we already know the mode is not a
4543 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4544 forcing a wider mode to exist.
4545 (expmed_mult_highpart_optab): Likewise.
4546 (expmed_mult_highpart): Likewise.
4547 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4549 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4550 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4551 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4552 * internal-fn.c (expand_mul_overflow): Update use of
4553 GET_MODE_2XWIDER_MODE.
4554 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4555 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4556 GET_MODE_WIDER_MODE.
4557 (convert_plusminus_to_widen): Likewise.
4558 * tree-switch-conversion.c (array_value_type): Likewise.
4559 * var-tracking.c (emit_note_insn_var_location): Likewise.
4560 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4561 Return false inside rather than outside the loop if no wider mode
4563 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4564 and GET_MODE_2XWIDER_MODE
4565 (can_compare_p): Use else_void.
4566 * gdbhooks.py (OptMachineModePrinter): New class.
4567 (build_pretty_printer): Use it for opt_mode.
4569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4570 Alan Hayward <alan.hayward@arm.com>
4571 David Sherwood <david.sherwood@arm.com>
4573 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4574 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4577 Alan Hayward <alan.hayward@arm.com>
4578 David Sherwood <david.sherwood@arm.com>
4580 * machmode.h (mode_traits): New structure.
4581 (get_narrowest_mode): New function.
4582 (mode_iterator::start): Likewise.
4583 (mode_iterator::iterate_p): Likewise.
4584 (mode_iterator::get_wider): Likewise.
4585 (mode_iterator::get_known_wider): Likewise.
4586 (mode_iterator::get_2xwider): Likewise.
4587 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4588 (FOR_EACH_MODE): Likewise.
4589 (FOR_EACH_MODE_FROM): Likewise.
4590 (FOR_EACH_MODE_UNTIL): Likewise.
4591 (FOR_EACH_WIDER_MODE): Likewise.
4592 (FOR_EACH_2XWIDER_MODE): Likewise.
4593 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4594 * combine.c (simplify_comparison): Likewise
4595 * config/i386/i386.c (type_natural_mode): Likewise.
4596 * cse.c (cse_insn): Likewise.
4597 * dse.c (find_shift_sequence): Likewise.
4598 * emit-rtl.c (init_derived_machine_modes): Likewise.
4599 (init_emit_once): Likewise.
4600 * explow.c (hard_function_value): Likewise.
4601 * expmed.c (extract_fixed_bit_field_1): Likewise.
4602 (extract_bit_field_1): Likewise.
4603 (expand_divmod): Likewise.
4604 (emit_store_flag_1): Likewise.
4605 * expr.c (init_expr_target): Likewise.
4606 (convert_move): Likewise.
4607 (alignment_for_piecewise_move): Likewise.
4608 (widest_int_mode_for_size): Likewise.
4609 (emit_block_move_via_movmem): Likewise.
4610 (copy_blkmode_to_reg): Likewise.
4611 (set_storage_via_setmem): Likewise.
4612 (compress_float_constant): Likewise.
4613 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4614 * optabs-query.c (get_best_extraction_insn): Likewise.
4615 * optabs.c (expand_binop): Likewise.
4616 (expand_twoval_unop): Likewise.
4617 (expand_twoval_binop): Likewise.
4618 (widen_leading): Likewise.
4619 (widen_bswap): Likewise.
4620 (expand_parity): Likewise.
4621 (expand_unop): Likewise.
4622 (prepare_cmp_insn): Likewise.
4623 (prepare_float_lib_cmp): Likewise.
4624 (expand_float): Likewise.
4625 (expand_fix): Likewise.
4626 (expand_sfix_optab): Likewise.
4627 * postreload.c (move2add_use_add2_insn): Likewise.
4628 * reg-stack.c (reg_to_stack): Likewise.
4629 * reginfo.c (choose_hard_reg_mode): Likewise.
4630 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4631 * stor-layout.c (mode_for_size): Likewise.
4632 (smallest_mode_for_size): Likewise.
4633 (mode_for_vector): Likewise.
4634 (finish_bitfield_representative): Likewise.
4635 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4636 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4637 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4638 * var-tracking.c (prepare_call_arguments): Likewise.
4640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4641 Alan Hayward <alan.hayward@arm.com>
4642 David Sherwood <david.sherwood@arm.com>
4644 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4645 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4646 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4647 * machmode.h (mode_size): Move earlier in file.
4648 (mode_precision): Likewise.
4649 (mode_inner): Likewise.
4650 (mode_nunits): Likewise.
4651 (mode_unit_size): Likewise.
4652 (unit_unit_precision): Likewise.
4653 (mode_wider): Likewise.
4654 (mode_2xwider): Likewise.
4655 (machine_mode): New class.
4656 (mode_to_bytes): New function.
4657 (mode_to_bits): Likewise.
4658 (mode_to_precision): Likewise.
4659 (mode_to_inner): Likewise.
4660 (mode_to_unit_size): Likewise.
4661 (mode_to_unit_precision): Likewise.
4662 (mode_to_nunits): Likewise.
4663 (GET_MODE_SIZE): Use mode_to_bytes.
4664 (GET_MODE_BITSIZE): Use mode_to_bits.
4665 (GET_MODE_PRECISION): Use mode_to_precision.
4666 (GET_MODE_INNER): Use mode_to_inner.
4667 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4668 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
4669 (GET_MODE_NUNITS): Use mode_to_nunits.
4670 * system.h (ALWAYS_INLINE): New macro.
4671 * config/powerpcspe/powerpcspe-c.c
4672 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
4673 int for arg1_mode and arg2_mode.
4675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4676 Alan Hayward <alan.hayward@arm.com>
4677 David Sherwood <david.sherwood@arm.com>
4679 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
4680 Prefix mode names with E_ in case statements.
4681 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4682 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
4683 (aarch64_split_simd_move): Likewise.
4684 (aarch64_gen_storewb_pair): Likewise.
4685 (aarch64_gen_loadwb_pair): Likewise.
4686 (aarch64_gen_store_pair): Likewise.
4687 (aarch64_gen_load_pair): Likewise.
4688 (aarch64_get_condition_code_1): Likewise.
4689 (aarch64_constant_pool_reload_icode): Likewise.
4690 (get_rsqrte_type): Likewise.
4691 (get_rsqrts_type): Likewise.
4692 (get_recpe_type): Likewise.
4693 (get_recps_type): Likewise.
4694 (aarch64_gimplify_va_arg_expr): Likewise.
4695 (aarch64_simd_container_mode): Likewise.
4696 (aarch64_emit_load_exclusive): Likewise.
4697 (aarch64_emit_store_exclusive): Likewise.
4698 (aarch64_expand_compare_and_swap): Likewise.
4699 (aarch64_gen_atomic_cas): Likewise.
4700 (aarch64_emit_bic): Likewise.
4701 (aarch64_emit_atomic_swap): Likewise.
4702 (aarch64_emit_atomic_load_op): Likewise.
4703 (aarch64_evpc_trn): Likewise.
4704 (aarch64_evpc_uzp): Likewise.
4705 (aarch64_evpc_zip): Likewise.
4706 (aarch64_evpc_ext): Likewise.
4707 (aarch64_evpc_rev): Likewise.
4708 (aarch64_evpc_dup): Likewise.
4709 (aarch64_gen_ccmp_first): Likewise.
4710 (aarch64_gen_ccmp_next): Likewise.
4711 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4712 (alpha_emit_xfloating_libcall): Likewise.
4713 (emit_insxl): Likewise.
4714 (alpha_arg_type): Likewise.
4715 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4716 (arc_preferred_simd_mode): Likewise.
4717 (arc_secondary_reload): Likewise.
4718 (get_arc_condition_code): Likewise.
4719 (arc_print_operand): Likewise.
4720 (arc_legitimate_constant_p): Likewise.
4721 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4722 * config/arc/arc.md (casesi_load): Likewise.
4723 (casesi_compact_jump): Likewise.
4724 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4725 (cc_use_register): Likewise.
4726 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4727 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4728 (arm_init_iwmmxt_builtins): Likewise.
4729 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4730 (neon_expand_vector_init): Likewise.
4731 (arm_attr_length_move_neon): Likewise.
4732 (maybe_get_arm_condition_code): Likewise.
4733 (arm_emit_vector_const): Likewise.
4734 (arm_preferred_simd_mode): Likewise.
4735 (arm_output_iwmmxt_tinsr): Likewise.
4736 (thumb1_output_casesi): Likewise.
4737 (thumb2_output_casesi): Likewise.
4738 (arm_emit_load_exclusive): Likewise.
4739 (arm_emit_store_exclusive): Likewise.
4740 (arm_expand_compare_and_swap): Likewise.
4741 (arm_evpc_neon_vuzp): Likewise.
4742 (arm_evpc_neon_vzip): Likewise.
4743 (arm_evpc_neon_vrev): Likewise.
4744 (arm_evpc_neon_vtrn): Likewise.
4745 (arm_evpc_neon_vext): Likewise.
4746 (arm_validize_comparison): Likewise.
4747 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4748 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4749 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4750 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4751 (c6x_preferred_simd_mode): Likewise.
4752 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4753 (epiphany_rtx_costs): Likewise.
4754 * config/epiphany/predicates.md (proper_comparison_operator):
4756 * config/frv/frv.c (condexec_memory_operand): Likewise.
4757 (frv_emit_move): Likewise.
4758 (output_move_single): Likewise.
4759 (output_condmove_single): Likewise.
4760 (frv_hard_regno_mode_ok): Likewise.
4761 (frv_matching_accg_mode): Likewise.
4762 * config/h8300/h8300.c (split_adds_subs): Likewise.
4763 (h8300_rtx_costs): Likewise.
4764 (h8300_print_operand): Likewise.
4765 (compute_mov_length): Likewise.
4766 (output_logical_op): Likewise.
4767 (compute_logical_op_length): Likewise.
4768 (compute_logical_op_cc): Likewise.
4769 (h8300_shift_needs_scratch_p): Likewise.
4770 (output_a_shift): Likewise.
4771 (compute_a_shift_length): Likewise.
4772 (compute_a_shift_cc): Likewise.
4773 (expand_a_rotate): Likewise.
4774 (output_a_rotate): Likewise.
4775 * config/i386/i386.c (classify_argument): Likewise.
4776 (function_arg_advance_32): Likewise.
4777 (function_arg_32): Likewise.
4778 (function_arg_64): Likewise.
4779 (function_value_64): Likewise.
4780 (ix86_gimplify_va_arg): Likewise.
4781 (ix86_legitimate_constant_p): Likewise.
4782 (put_condition_code): Likewise.
4783 (split_double_mode): Likewise.
4784 (ix86_avx256_split_vector_move_misalign): Likewise.
4785 (ix86_expand_vector_logical_operator): Likewise.
4786 (ix86_split_idivmod): Likewise.
4787 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
4788 (ix86_build_const_vector): Likewise.
4789 (ix86_build_signbit_mask): Likewise.
4790 (ix86_match_ccmode): Likewise.
4791 (ix86_cc_modes_compatible): Likewise.
4792 (ix86_expand_branch): Likewise.
4793 (ix86_expand_sse_cmp): Likewise.
4794 (ix86_expand_sse_movcc): Likewise.
4795 (ix86_expand_int_sse_cmp): Likewise.
4796 (ix86_expand_vec_perm_vpermi2): Likewise.
4797 (ix86_expand_vec_perm): Likewise.
4798 (ix86_expand_sse_unpack): Likewise.
4799 (ix86_expand_int_addcc): Likewise.
4800 (ix86_split_to_parts): Likewise.
4801 (ix86_vectorize_builtin_gather): Likewise.
4802 (ix86_vectorize_builtin_scatter): Likewise.
4803 (avx_vpermilp_parallel): Likewise.
4804 (inline_memory_move_cost): Likewise.
4805 (ix86_tieable_integer_mode_p): Likewise.
4806 (x86_maybe_negate_const_int): Likewise.
4807 (ix86_expand_vector_init_duplicate): Likewise.
4808 (ix86_expand_vector_init_one_nonzero): Likewise.
4809 (ix86_expand_vector_init_one_var): Likewise.
4810 (ix86_expand_vector_init_concat): Likewise.
4811 (ix86_expand_vector_init_interleave): Likewise.
4812 (ix86_expand_vector_init_general): Likewise.
4813 (ix86_expand_vector_set): Likewise.
4814 (ix86_expand_vector_extract): Likewise.
4815 (emit_reduc_half): Likewise.
4816 (ix86_emit_i387_round): Likewise.
4817 (ix86_mangle_type): Likewise.
4818 (ix86_expand_round_sse4): Likewise.
4819 (expand_vec_perm_blend): Likewise.
4820 (canonicalize_vector_int_perm): Likewise.
4821 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
4822 (expand_vec_perm_1): Likewise.
4823 (expand_vec_perm_interleave3): Likewise.
4824 (expand_vec_perm_even_odd_pack): Likewise.
4825 (expand_vec_perm_even_odd_1): Likewise.
4826 (expand_vec_perm_broadcast_1): Likewise.
4827 (ix86_vectorize_vec_perm_const_ok): Likewise.
4828 (ix86_expand_vecop_qihi): Likewise.
4829 (ix86_expand_mul_widen_hilo): Likewise.
4830 (ix86_expand_sse2_abs): Likewise.
4831 (ix86_expand_pextr): Likewise.
4832 (ix86_expand_pinsr): Likewise.
4833 (ix86_preferred_simd_mode): Likewise.
4834 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
4835 * config/i386/sse.md (*andnot<mode>3): Likewise.
4836 (<mask_codefor><code><mode>3<mask_name>): Likewise.
4837 (*<code><mode>3): Likewise.
4838 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
4839 (ia64_expand_atomic_op): Likewise.
4840 (ia64_arg_type): Likewise.
4841 (ia64_mode_to_int): Likewise.
4842 (ia64_scalar_mode_supported_p): Likewise.
4843 (ia64_vector_mode_supported_p): Likewise.
4844 (expand_vec_perm_broadcast): Likewise.
4845 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
4846 (iq2000_function_arg_advance): Likewise.
4847 (iq2000_function_arg): Likewise.
4848 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
4849 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
4850 (m68k_libcall_value): Likewise.
4851 (m68k_function_value): Likewise.
4852 (sched_attr_op_type): Likewise.
4853 * config/mcore/mcore.c (mcore_output_move): Likewise.
4854 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
4856 (microblaze_function_arg): Likewise.
4857 * config/mips/mips.c (mips16_build_call_stub): Likewise.
4858 (mips_print_operand): Likewise.
4859 (mips_mode_ok_for_mov_fmt_p): Likewise.
4860 (mips_vector_mode_supported_p): Likewise.
4861 (mips_preferred_simd_mode): Likewise.
4862 (mips_expand_vpc_loongson_even_odd): Likewise.
4863 (mips_expand_vec_unpack): Likewise.
4864 (mips_expand_vi_broadcast): Likewise.
4865 (mips_expand_vector_init): Likewise.
4866 (mips_expand_vec_reduc): Likewise.
4867 (mips_expand_msa_cmp): Likewise.
4868 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
4869 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
4870 (cc_flags_for_mode): Likewise.
4871 * config/msp430/msp430.c (msp430_print_operand): Likewise.
4872 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
4873 (nds32_output_casesi_pc_relative): Likewise.
4874 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4875 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
4876 (nvptx_gen_unpack): Likewise.
4877 (nvptx_gen_pack): Likewise.
4878 (nvptx_gen_shuffle): Likewise.
4879 (nvptx_gen_wcast): Likewise.
4880 (nvptx_preferred_simd_mode): Likewise.
4881 * config/pa/pa.c (pa_secondary_reload): Likewise.
4882 * config/pa/predicates.md (base14_operand): Likewise.
4883 * config/powerpcspe/powerpcspe-c.c
4884 (altivec_resolve_overloaded_builtin): Likewise.
4885 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
4887 (rs6000_preferred_simd_mode): Likewise.
4888 (output_vec_const_move): Likewise.
4889 (rs6000_expand_vector_extract): Likewise.
4890 (rs6000_split_vec_extract_var): Likewise.
4891 (reg_offset_addressing_ok_p): Likewise.
4892 (rs6000_legitimate_offset_address_p): Likewise.
4893 (rs6000_legitimize_address): Likewise.
4894 (rs6000_emit_set_const): Likewise.
4895 (rs6000_const_vec): Likewise.
4896 (rs6000_emit_move): Likewise.
4897 (spe_build_register_parallel): Likewise.
4898 (rs6000_darwin64_record_arg_recurse): Likewise.
4899 (swap_selector_for_mode): Likewise.
4900 (spe_init_builtins): Likewise.
4901 (paired_init_builtins): Likewise.
4902 (altivec_init_builtins): Likewise.
4903 (do_load_for_compare): Likewise.
4904 (rs6000_generate_compare): Likewise.
4905 (rs6000_expand_float128_convert): Likewise.
4906 (emit_load_locked): Likewise.
4907 (emit_store_conditional): Likewise.
4908 (rs6000_output_function_epilogue): Likewise.
4909 (rs6000_handle_altivec_attribute): Likewise.
4910 (rs6000_function_value): Likewise.
4911 (emit_fusion_gpr_load): Likewise.
4912 (emit_fusion_p9_load): Likewise.
4913 (emit_fusion_p9_store): Likewise.
4914 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
4915 (fusion_gpr_mem_load): Likewise.
4916 (fusion_addis_mem_combo_load): Likewise.
4917 (fusion_addis_mem_combo_store): Likewise.
4918 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
4919 (fusion_gpr_mem_load): Likewise.
4920 (fusion_addis_mem_combo_load): Likewise.
4921 (fusion_addis_mem_combo_store): Likewise.
4922 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4924 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
4925 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4926 (rs6000_preferred_simd_mode): Likewise.
4927 (output_vec_const_move): Likewise.
4928 (rs6000_expand_vector_extract): Likewise.
4929 (rs6000_split_vec_extract_var): Likewise.
4930 (reg_offset_addressing_ok_p): Likewise.
4931 (rs6000_legitimate_offset_address_p): Likewise.
4932 (rs6000_legitimize_address): Likewise.
4933 (rs6000_emit_set_const): Likewise.
4934 (rs6000_const_vec): Likewise.
4935 (rs6000_emit_move): Likewise.
4936 (rs6000_darwin64_record_arg_recurse): Likewise.
4937 (swap_selector_for_mode): Likewise.
4938 (paired_init_builtins): Likewise.
4939 (altivec_init_builtins): Likewise.
4940 (rs6000_expand_float128_convert): Likewise.
4941 (emit_load_locked): Likewise.
4942 (emit_store_conditional): Likewise.
4943 (rs6000_output_function_epilogue): Likewise.
4944 (rs6000_handle_altivec_attribute): Likewise.
4945 (rs6000_function_value): Likewise.
4946 (emit_fusion_gpr_load): Likewise.
4947 (emit_fusion_p9_load): Likewise.
4948 (emit_fusion_p9_store): Likewise.
4949 * config/rx/rx.c (rx_gen_move_template): Likewise.
4950 (flags_from_mode): Likewise.
4951 * config/s390/predicates.md (s390_alc_comparison): Likewise.
4952 (s390_slb_comparison): Likewise.
4953 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
4954 (s390_vector_mode_supported_p): Likewise.
4955 (s390_cc_modes_compatible): Likewise.
4956 (s390_match_ccmode_set): Likewise.
4957 (s390_canonicalize_comparison): Likewise.
4958 (s390_emit_compare_and_swap): Likewise.
4959 (s390_branch_condition_mask): Likewise.
4960 (s390_rtx_costs): Likewise.
4961 (s390_secondary_reload): Likewise.
4962 (__SECONDARY_RELOAD_CASE): Likewise.
4963 (s390_expand_cs): Likewise.
4964 (s390_preferred_simd_mode): Likewise.
4965 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
4966 * config/sh/sh.c (sh_print_operand): Likewise.
4967 (dump_table): Likewise.
4968 (sh_secondary_reload): Likewise.
4969 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4970 * config/sh/sh.md (casesi_worker_1): Likewise.
4971 (casesi_worker_2): Likewise.
4972 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
4973 (fcc_comparison_operator): Likewise.
4974 * config/sparc/sparc.c (sparc_expand_move): Likewise.
4975 (emit_soft_tfmode_cvt): Likewise.
4976 (sparc_preferred_simd_mode): Likewise.
4977 (output_cbranch): Likewise.
4978 (sparc_print_operand): Likewise.
4979 (sparc_expand_vec_perm_bmask): Likewise.
4980 (vector_init_bshuffle): Likewise.
4981 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4982 (spu_vector_mode_supported_p): Likewise.
4983 (spu_expand_insv): Likewise.
4984 (spu_emit_branch_or_set): Likewise.
4985 (spu_handle_vector_attribute): Likewise.
4986 (spu_builtin_splats): Likewise.
4987 (spu_builtin_extract): Likewise.
4988 (spu_builtin_promote): Likewise.
4989 (spu_expand_sign_extend): Likewise.
4990 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4991 (tilegx_simd_int): Likewise.
4992 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
4993 (tilepro_simd_int): Likewise.
4994 * config/v850/v850.c (const_double_split): Likewise.
4995 (v850_print_operand): Likewise.
4996 (ep_memory_offset): Likewise.
4997 * config/vax/vax.c (vax_rtx_costs): Likewise.
4998 (vax_output_int_move): Likewise.
4999 (vax_output_int_add): Likewise.
5000 (vax_output_int_subtract): Likewise.
5001 * config/visium/predicates.md (visium_branch_operator): Likewise.
5002 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
5003 (visium_print_operand_address): Likewise.
5004 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
5005 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
5006 (xtensa_expand_conditional_branch): Likewise.
5007 (xtensa_copy_incoming_a7): Likewise.
5008 (xtensa_output_literal): Likewise.
5009 * dfp.c (decimal_real_maxval): Likewise.
5010 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
5012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5013 Alan Hayward <alan.hayward@arm.com>
5014 David Sherwood <david.sherwood@arm.com>
5016 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
5017 (mode_nunits_inline): Likewise.
5018 (mode_inner_inline): Likewise.
5019 (mode_unit_size_inline): Likewise.
5020 (mode_unit_precision_inline): Likewise.
5021 (emit_insn_modes_h): Likewise. Also emit a #define of the
5023 (emit_mode_wider): Add an E_ prefix to mode names.
5024 (emit_mode_complex): Likewise.
5025 (emit_mode_inner): Likewise.
5026 (emit_mode_adjustments): Likewise.
5027 (emit_mode_int_n): Likewise.
5028 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
5029 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
5030 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
5031 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
5032 (CRC32_BUILTIN, ENTRY): Likewise.
5033 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
5034 (aarch64_pop_regs): Likewise.
5035 (aarch64_process_components): Likewise.
5036 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
5037 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
5038 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
5039 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
5040 * config/arm/arm.c (arm_init_libfuncs): Likewise.
5041 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
5043 * config/i386/i386-builtin.def (pcmpestr): Likewise.
5044 (pcmpistr): Likewise.
5045 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
5046 * config/mmix/mmix.c (mmix_output_condition): Likewise.
5047 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
5049 * config/rl78/rl78.c (mduc_regs): Likewise.
5050 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5051 (htm_expand_builtin): Likewise.
5052 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
5053 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
5054 * config/xtensa/xtensa.c (print_operand): Likewise.
5055 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
5056 (NUM_MODE_VECTOR_INT): Likewise.
5057 * genoutput.c (null_operand): Likewise.
5058 (output_operand_data): Likewise.
5059 * genrecog.c (print_parameter_value): Likewise.
5060 * lra.c (debug_operand_data): Likewise.
5062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5064 * dwarf2out.c (scompare_loc_descriptor_wide)
5065 (scompare_loc_descriptor_narrow): New functions, split out from...
5066 (scompare_loc_descriptor): ...here.
5067 * expmed.c (emit_store_flag_int): New function, split out from...
5068 (emit_store_flag): ...here.
5070 2017-08-30 Richard Biener <rguenther@suse.de>
5072 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
5073 (dwarf2out_early_finish): Move setting of AT_pubnames from
5074 early debug output to early finish.
5076 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5078 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
5079 and -mdata-region to the assembler.
5081 2017-08-30 Richard Biener <rguenther@suse.de>
5083 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
5085 (gen_subprogram_die): Add DW_AT_object_pointer only early.
5086 (dwarf2out_early_global_decl): Only generate a DIE for the
5087 abstract origin if it doesn't already exist or is a declaration DIE.
5088 (resolve_addr): Do not add the linkage name twice when
5089 generating a stub DIE for the DW_TAG_GNU_call_site target.
5091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5093 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5094 Use machine_mode rather than int for arg1_mode.
5096 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
5099 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
5100 that the second argument of the built-in functions to unpack
5101 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
5102 switch statement instead a lot of if statements.
5103 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
5104 Allow 64-bit values to be in Altivec registers as well as
5105 traditional floating point registers.
5106 (pack<mode>, FMOVE128_VSX iterator): Likewise.
5108 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
5110 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
5111 MAX_REGS_PER_ADDRESS == 1.
5113 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
5115 * config/i386/i386.opt (flag_fentry): Do not init to -1.
5116 * config/i386/i386.c (ix86_option_override_internal): Simplify
5117 setting of opts->x_flag_entry.
5119 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5120 Jakub Jelinek <jakub@redhat.com>
5121 Richard Biener <rguenther@suse.de>
5123 PR tree-optimization/81503
5124 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
5125 folded constant fits in the target type; reorder tests for clarity.
5127 2017-08-29 Martin Liska <mliska@suse.cz>
5129 * passes.def: Include pass_lower_switch.
5130 * stmt.c (dump_case_nodes): Remove and move to
5131 tree-switch-conversion.
5132 (case_values_threshold): Likewise.
5133 (expand_switch_as_decision_tree_p): Likewise.
5134 (emit_case_decision_tree): Likewise.
5135 (expand_case): Likewise.
5136 (balance_case_nodes): Likewise.
5137 (node_has_low_bound): Likewise.
5138 (node_has_high_bound): Likewise.
5139 (node_is_bounded): Likewise.
5140 (emit_case_nodes): Likewise.
5141 (struct simple_case_node): New struct.
5142 (add_case_node): Remove.
5143 (emit_case_dispatch_table): Use vector instead of case_list.
5144 (reset_out_edges_aux): Remove.
5145 (compute_cases_per_edge): Likewise.
5146 (expand_case): Build list of simple_case_node.
5147 (expand_sjlj_dispatch_table): Use it.
5148 * tree-switch-conversion.c (struct case_node): Moved from
5149 stmt.c and adjusted.
5150 (emit_case_nodes): Likewise.
5151 (node_has_low_bound): Likewise.
5152 (node_has_high_bound): Likewise.
5153 (node_is_bounded): Likewise.
5154 (case_values_threshold): Likewise.
5155 (reset_out_edges_aux): Likewise.
5156 (compute_cases_per_edge): Likewise.
5157 (add_case_node): Likewise.
5158 (dump_case_nodes): Likewise.
5159 (balance_case_nodes): Likewise.
5160 (expand_switch_as_decision_tree_p): Likewise.
5161 (emit_jump): Likewise.
5162 (emit_case_decision_tree): Likewise.
5163 (try_switch_expansion): Likewise.
5164 (do_jump_if_equal): Likewise.
5165 (emit_cmp_and_jump_insns): Likewise.
5166 (fix_phi_operands_for_edge): New function.
5167 (record_phi_operand_mapping): Likewise.
5168 (class pass_lower_switch): New pass.
5169 (pass_lower_switch::execute): New function.
5170 (make_pass_lower_switch): Likewise.
5171 (conditional_probability):
5172 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
5173 * tree-pass.h: Add make_pass_lower_switch.
5175 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5178 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
5181 2017-08-29 Richard Biener <rguenther@suse.de>
5183 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
5184 we do not add a DW_AT_inline attribute twice.
5185 (gen_subprogram_die): Remove code setting DW_AT_inline on
5186 DECL_ABSTRACT_P nodes.
5188 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
5190 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
5191 calls to internal functions.
5192 (gimplify_modify_expr): Likewise.
5193 * tree-call-cdce.c (use_internal_fn): Likewise.
5194 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5195 (convert_to_divmod): Set the nothrow flag.
5196 * tree-if-conv.c (predicate_mem_writes): Likewise.
5197 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5198 (vectorizable_call): Likewise.
5199 (vectorizable_store): Likewise.
5200 (vectorizable_load): Likewise.
5201 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
5202 (vect_recog_mask_conversion_pattern): Likewise.
5204 2017-08-29 Martin Liska <mliska@suse.cz>
5207 * gcc.c (driver_handle_option): Add new argument.
5208 * opts-common.c (handle_option): Pass
5209 target_option_override_hook.
5210 * opts-global.c (lang_handle_option): Add new option.
5211 (set_default_handlers): Add new argument.
5212 (decode_options): Likewise.
5213 * opts.c (target_handle_option): Likewise.
5214 (common_handle_option): Call target_option_override_hook.
5215 * opts.h (struct cl_option_handler_func): Add hook for
5216 target option override.
5217 (struct cl_option_handlers): Likewise.
5218 (set_default_handlers): Add new argument.
5219 (decode_options): Likewise.
5220 (common_handle_option): Likewise.
5221 (target_handle_option): Likewise.
5222 * toplev.c (toplev::main): Pass targetm.target_option.override
5225 2017-08-29 Richard Biener <rguenther@suse.de>
5226 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5228 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
5229 life to the active subtree.
5231 2017-08-28 Jeff Law <law@redhat.com>
5233 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
5234 derive_equivalences.
5235 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
5237 (edge_info::derive_equivalences): New private member function
5239 * tree-ssa-dom.c (class edge_info): Changed from a struct
5240 to a class. Add ctor/dtor, methods and data members.
5241 (edge_info::edge_info): Renamed from allocate_edge_info.
5242 Initialize additional members.
5243 (edge_info::~edge_info): New.
5244 (free_dom_edge_info): Delete the edge info.
5245 (record_edge_info): Use new class & associated member functions.
5246 Tighten forms for testing for edge equivalences.
5247 (record_temporary_equivalences): Iterate over the simple
5248 equivalences rather than assuming there's only one per edge.
5249 (cprop_into_successor_phis): Iterate over the simple
5250 equivalences rather than assuming there's only one per edge.
5251 (optimize_stmt): Use operand_equal_p rather than pointer
5252 equality for mini-DSE code.
5254 2017-08-28 Nathan Sidwell <nathan@acm.org>
5256 * gcc.c (execute): Fold SIGPIPE handling into switch
5257 statement. Adjust internal error message.
5259 2017-08-28 Richard Biener <rguenther@suse.de>
5262 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
5263 Do nothing for removed DIEs.
5265 2017-08-28 Richard Biener <rguenther@suse.de>
5267 PR tree-optimization/81977
5268 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
5271 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
5274 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
5275 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
5276 using targetm.gen_mem_thread_fence.
5278 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
5281 * config/i386/i386.md (*<btsc><mode>): Change operand 2
5282 predicate to register_operand. Reorder operands.
5283 (*btr<mode>): Ditto.
5284 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
5285 (*btr<mode>_mask): Ditto.
5287 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
5289 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
5290 * config/rs6000/xmmintrin.h: New file.
5291 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
5293 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5296 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
5297 parameter and_insn and return it.
5298 (recombine_lvx_pattern): Insert a copy to ensure availability of
5299 the base register of the copied masking operation at the point of
5300 the instruction replacement.
5301 (recombine_stvx_pattern): Likewise.
5303 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5305 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
5306 undocumented switches.
5307 (-mpower9-dform-vector): Likewise.
5308 (-mpower9-dform): Likewise.
5309 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
5310 comments to delete references to -mpower9-dform* switches.
5311 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
5312 Delete reference to -mpower9-dform* switches, test for
5313 -mpower9-vector instead.
5314 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
5315 (OTHER_P9_VECTOR_MASKS): Likewise.
5316 (POWERPC_MASKS): Likewise.
5317 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
5318 tests against -mpower9-dform* to -mpower9-vector. Delete code
5319 that checked for -mpower9-dform* consistancy with other options.
5320 Add test for -mpower9-misc to enable other power9 switches.
5321 (rs6000_init_hard_regno_mode_ok): Likewise.
5322 (rs6000_option_override_internal): Likewise.
5323 (rs6000_emit_prologue): Likewise.
5324 (rs6000_emit_epilogue): Likewise.
5325 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
5326 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
5327 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
5329 (emit_fusion_p9_store): Likewise.
5330 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
5331 resetting these macros if the assembler does not support ISA 3.0
5333 (TARGET_P9_DFORM_VECTOR): Likewise.
5334 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
5335 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
5337 2017-08-25 Alan Modra <amodra@gmail.com>
5339 PR rtl-optimization/81747
5340 * cse.c (cse_extended_basic_block): Don't attempt to record
5341 equivalences for degenerate conditional branches that branch
5342 to their fall-through.
5344 2017-08-24 Martin Sebor <msebor@redhat.com>
5347 * gimple-fold.c (size_must_be_zero_p): New function.
5348 (gimple_fold_builtin_memory_op): Call it.
5350 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
5352 * config/rs6000/mm_malloc.h: New file.
5354 2017-08-24 Bin Cheng <bin.cheng@arm.com>
5356 PR tree-optimization/81913
5357 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
5358 analysis when either IVs in condition can wrap.
5360 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
5362 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
5363 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
5365 2017-08-24 Richard Biener <rguenther@suse.de>
5368 * targhooks.c (default_target_can_inline_p): Properly
5369 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
5370 is present and always compare.
5371 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
5372 infer -mfpmath=sse from TARGET_SSE_P.
5373 (ix86_can_inline_p): Properly use target_option_default_node when
5374 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
5376 2017-08-24 Richard Biener <rguenther@suse.de>
5379 * dwarf2out.c (output_die): Handle flag_generate_offload like
5381 (output_comp_unit): Likewise.
5382 (gen_array_type_die): Likewise.
5383 (dwarf2out_early_finish): Likewise.
5384 (note_variable_value_in_expr): Likewise.
5385 (dwarf2out_finish): Likewise. Adjust assert.
5386 * cgraphunit.c (symbol_table::compile): Move setting of
5387 flag_generate_offload earlier ...
5388 (symbol_table::finalize_compilation_unit): ... here, before
5389 early debug finalization.
5391 2017-08-24 Richard Biener <rguenther@suse.de>
5393 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
5394 and ipa-fnsummary.h.
5395 (ix86_can_inline_p): When ix86_fpmath flags do not match
5396 check whether the callee uses FP math at all.
5398 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
5401 * tree-ssanames.c (get_nonzero_bits): Use element_precision
5402 instead of TYPE_PRECISION.
5404 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
5405 Alan Hayward <alan.hayward@arm.com>
5406 David Sherwood <david.sherwood@arm.com>
5408 * combine.c (make_extraction): Use subreg_offset_from_lsb.
5410 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
5412 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
5414 (ix86_frame::stack_realign_allocate): New field.
5415 (struct machine_frame_state): Modify comments.
5416 (machine_frame_state::sp_realigned_fp_end): New field.
5417 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
5419 (sp_valid_at): Add assertion to assure no attempt to access invalid
5420 offset of a realigned stack.
5421 (fp_valid_at): Likewise.
5422 (choose_baseaddr): Modify comments.
5423 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
5424 ix86_expand_prologue.
5425 (ix86_expand_prologue): Modify stack realignment and allocation.
5426 (ix86_expand_epilogue): Modify comments.
5427 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5428 avx2_runtime, avx512f, and avx512f_runtime.
5430 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5432 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5433 (mstackrealign): Do not init to -1.
5434 * config/i386/i386.c (ix86_option_override_internal):
5435 Check opts_set, not opts when setting default value of
5436 opts->x_ix86_force_align_arg_pointer.
5438 2017-08-23 Richard Biener <rguenther@suse.de>
5440 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5441 lang_hooks.decl_printable_name.
5442 * print-rtl-function.c (print_rtx_function): Likewise.
5443 * tree-pretty-print.c (dump_function_header): Likewise.
5445 2017-08-23 Richard Biener <rguenther@suse.de>
5448 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5449 -g0 at compile-time.
5451 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5454 * doc/sourcebuild.texi (Other hardware attributes):
5457 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5460 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5461 Fix single-use check.
5463 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5465 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5467 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5469 * doc/install.texi: Modify to add more details on running selected
5472 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5474 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5475 is combined with -mabi=ms.
5476 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5479 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5481 PR tree-optimization/81488
5482 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5483 and cached_basis fields.
5484 (MAX_SPREAD): New constant.
5485 (alloc_cand_and_find_basis): Initialize new fields.
5486 (clear_visited): New function.
5487 (create_phi_basis_1): Rename from create_phi_basis, set visited
5488 and cached_basis fields.
5489 (create_phi_basis): New wrapper function.
5490 (phi_add_costs_1): Rename from phi_add_costs, add spread
5491 parameter, set visited field, short-circuit when limits reached.
5492 (phi_add_costs): New wrapper function.
5493 (record_phi_increments_1): Rename from record_phi_increments, set
5495 (record_phi_increments): New wrapper function.
5496 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5497 (phi_incr_cost): New wrapper function.
5498 (all_phi_incrs_profitable_1): Rename from
5499 all_phi_incrs_profitable, set visited field.
5500 (all_phi_incrs_profitable): New wrapper function.
5502 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5503 Alan Hayward <alan.hayward@arm.com>
5504 David Sherwood <david.sherwood@arm.com>
5506 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5507 that takes the outer and inner modes.
5508 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5509 comparison as the canonical test for a paradoxical subreg.
5510 * combine.c (simplify_set): Use paradoxical_subreg_p.
5511 (make_extraction): Likewise.
5512 (force_to_mode): Likewise.
5513 (rtx_equal_for_field_assignment_p): Likewise.
5514 (gen_lowpart_for_combine): Likewise.
5515 (simplify_comparison): Likewise.
5516 * cse.c (equiv_constant): Likewise.
5517 * expmed.c (store_bit_field_1): Likewise.
5518 * final.c (alter_subreg): Likewise.
5519 * fwprop.c (propagate_rtx): Likewise.
5520 (forward_propagate_subreg): Likewise.
5521 * ira-conflicts.c (ira_build_conflicts): Likewise.
5522 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5523 * lra-constraints.c (curr_insn_transform): Likewise.
5524 (split_reg): Likewise.
5525 * lra-eliminations.c (move_plus_up): Likewise.
5526 (lra_eliminate_regs_1): Likewise.
5527 * recog.c (general_operand): Likewise.
5528 * ree.c (combine_reaching_defs): Likewise.
5529 * reload.c (push_reload): Likewise.
5530 (find_reloads): Likewise.
5531 * reload1.c (elimination_effects): Likewise.
5532 (compute_reload_subreg_offset): Likewise.
5533 (choose_reload_regs): Likewise.
5534 * rtlanal.c (subreg_lsb_1): Likewise.
5535 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5536 (simplify_subreg): Likewise.
5537 * var-tracking.c (track_loc_p): Likewise.
5538 * emit-rtl.c (byte_lowpart_offset): Likewise.
5539 (paradoxical_subreg_p): Delete out-of-line definition.
5541 2017-08-22 Jeff Law <law@redhat.com>
5543 PR tree-optimization/81741
5544 PR tree-optimization/71947
5545 * tree-ssa-dom.c: Include tree-inline.h.
5546 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5547 equivalences if one is more expensive to compute than the other.
5548 * tree-ssa-scopedtables.h (class const_or_copies): Make
5549 record_const_or_copy_raw method private.
5550 (class avail_exprs_stack): New method simplify_binary_operation.
5551 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5552 avail_exprs_stack::simplify_binary_operation as needed.
5553 (avail_exprs_stack::simplify_binary_operation): New function.
5555 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5557 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5558 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5559 (DOT_SYMBOLS): Likewise.
5560 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5561 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5562 (RS6000_ABI_NAME): Likewise.
5563 (TARGET_CMODEL): Likewise.
5564 (TOC_SECTION_ASM_OP): Likewise.
5565 (SET_CMODEL): New macro.
5566 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5568 2017-08-22 Richard Biener <rguenther@suse.de>
5570 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5571 to allow for free-lang-data replacements similar to verify_type_variant.
5573 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5574 Alan Hayward <alan.hayward@arm.com>
5575 David Sherwood <david.sherwood@arm.com>
5577 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5578 VOIDmode for the LABEL_REF.
5580 2017-08-22 Richard Biener <rguenther@suse.de>
5582 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5584 2017-08-22 Richard Biener <rguenther@suse.de>
5586 * common.opt (feliminate-dwarf2-dups): Ignore.
5587 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5588 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5589 same_die_p_wrap, compute_section_prefix,
5590 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5591 (comdat_symbol_id, comdat_symbol_number): Likewise.
5592 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5594 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5595 (output_die): Mark unreachable path unreachable.
5596 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5597 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5598 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5599 (dwarf2out_early_finish): Likewise.
5601 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5603 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5605 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5608 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5609 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5610 (avr_attribute_table) <io, io_low, address>: Initialize
5611 .decl_required with true.
5613 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5615 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5616 undocumented debugging options.
5617 (-mvsx-scalar-double): Likewise.
5618 (-mallow-df-permute): Likewise.
5619 (-mvectorize-builtins): Likewise.
5620 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5621 (rs6000_builtin_vectorized_function): Likewise.
5622 (rs6000_builtin_md_vectorized_function): Likewise.
5623 (rs6000_opt_vars): Likewise.
5625 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5628 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5629 (*btrq_imm): Rename from *btrq.
5630 (*btcq_imm): Rename from *btcq.
5631 (btsc): New code attribute.
5632 (*<btsc><mode>): New insn pattern.
5633 (*btr<mode>): Ditto.
5634 (*<btsc><mode>_mask): New insn_and_split pattern.
5635 (*btr<mode>_mask): Ditto.
5637 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5638 Alan Hayward <alan.hayward@arm.com>
5639 David Sherwood <david.sherwood@arm.com>
5641 * function.c (pad_below): Simplify padding calculation.
5643 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5644 Alan Hayward <alan.hayward@arm.com>
5645 David Sherwood <david.sherwood@arm.com>
5647 * target.def (function_prologue): Remove frame size argument.
5648 (function_epilogue): Likewise.
5649 * doc/tm.texi: Regenerate.
5650 * final.c (final_start_function): Update call to function_prologue.
5651 (final_end_function): Update call to function_epilogue.
5652 (default_function_pro_epilogue): Remove frame size argument.
5653 * output.h (default_function_pro_epilogue): Likewise.
5654 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5655 (arm_output_function_prologue): Likewise.
5656 * config/frv/frv.c (frv_function_prologue): Likewise.
5657 (frv_function_epilogue): Likewise.
5658 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5659 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5660 (ia64_output_function_epilogue): Likewise.
5661 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5662 (m32r_output_function_epilogue): Likewise.
5663 * config/microblaze/microblaze.c (microblaze_function_prologue)
5664 (microblaze_function_epilogue): Likewise.
5665 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5666 (mips_output_function_epilogue): Likewise.
5667 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5668 (mmix_target_asm_function_epilogue): Likewise.
5669 * config/msp430/msp430.c (msp430_start_function): Likewise.
5670 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
5671 (nds32_asm_function_epilogue): Likewise.
5672 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
5673 * config/pa/pa.c (pa_output_function_prologue): Likewise.
5674 (pa_output_function_epilogue): Likewise.
5675 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
5676 (rs6000_output_function_epilogue): Likewise.
5677 * config/rl78/rl78.c (rl78_start_function): Likewise.
5678 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
5679 (rs6000_output_function_epilogue): Likewise.
5680 * config/rx/rx.c (rx_output_function_prologue): Likewise.
5681 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
5682 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
5683 (sparc_asm_function_epilogue): Likewise.
5685 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5687 * tree.h (type_has_mode_precision_p): New function.
5688 * convert.c (convert_to_integer_1): Use it.
5689 * expr.c (expand_expr_real_2): Likewise.
5690 (expand_expr_real_1): Likewise.
5691 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5692 * match.pd: Likewise.
5693 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5694 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5695 * tree-tailcall.c (process_assignment): Likewise.
5696 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5697 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5698 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5699 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5700 (vectorizable_assignment): Likewise.
5701 (vectorizable_shift): Likewise.
5702 (vectorizable_operation): Likewise.
5703 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5705 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5707 * match.pd: Add pow (C, x) simplification.
5709 2017-08-21 Richard Biener <rguenther@suse.de>
5711 PR tree-optimization/81900
5712 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5713 for blocks with abnormal predecessors.
5714 (compute_antic): Do not set visited flag prematurely.
5716 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5719 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5721 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5723 * stor-layout.h (vector_type_mode): Move to...
5724 * tree.h (vector_type_mode): ...here.
5725 * stor-layout.c (vector_type_mode): Move to...
5726 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5728 2017-08-21 Richard Biener <rguenther@suse.de>
5730 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5731 register_external_die hooks.
5732 (debug_false_tree_charstarstar_uhwistar): Declare.
5733 (debug_nothing_tree_charstar_uhwi): Likewise.
5734 * debug.c (do_nothing_debug_hooks): Adjust.
5735 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5736 (debug_nothing_tree_charstar_uhwi): Likewise.
5737 * dbxout.c (dbx_debug_hooks): Adjust.
5738 (xcoff_debug_hooks): Likewise.
5739 * sdbout.c (sdb_debug_hooks): Likewise.
5740 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5741 * dwarf2out.c (macinfo_label_base): New global.
5742 (dwarf2out_register_external_die): New function for the
5743 register_external_die hook.
5744 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5745 (dwarf2_debug_hooks): Use them.
5746 (dwarf2_lineno_debug_hooks): Adjust.
5747 (struct die_struct): Add with_offset flag.
5748 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5749 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5750 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5751 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5752 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5753 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5754 defining section names for the early LTO debug variants.
5755 (reset_indirect_string): New helper.
5756 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5757 (print_dw_val): Add support for offsetted symbol references.
5758 (get_ultimate_context): Split out from is_cxx.
5759 (is_cxx): Use get_ultimate_context.
5760 (is_fortran): Add decl overload.
5761 (compute_comp_unit_symbol): Split out worker from
5762 compute_section_prefix.
5763 (compute_section_prefix): Call compute_comp_unit_symbol and
5764 set comdat_type_p here.
5765 (output_die): Skip DIE symbol output for the LTO added one.
5766 Handle DIE symbol references with offset.
5767 (output_comp_unit): Guard section name mangling properly.
5768 For LTO debug sections emit a symbol at the section beginning
5769 which we use to refer to its DIEs.
5770 (add_abstract_origin_attribute): For DIEs registered via
5771 dwarf2out_register_external_die directly refer to the early
5772 DIE rather than indirectly through the shadow one we created.
5773 Remove obsolete call to dwarf2out_abstract_function for
5774 non-function/block origins.
5775 (gen_array_type_die): When generating early LTO debug do
5776 not emit DW_AT_string_length.
5777 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
5778 late when in LTO. As suggested place a gcc_unreachable for
5779 the DECL_ABSTRACT_P case.
5780 (gen_subprogram_die): Avoid another specification DIE
5781 for early built declarations/definitions for the late LTO case.
5782 (gen_variable_die): Add type references for late duplicated VLA dies
5784 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
5785 we have the abstract instance already.
5786 (process_scope_var): Adjust decl DIE contexts in LTO which
5787 first puts them in limbo.
5788 (gen_decl_die): Do not generate type DIEs late apart from
5789 types for VLAs or for decls we do not yet have a DIE. Do not
5790 call dwarf2out_abstract_function late.
5791 (dwarf2out_early_global_decl): Make sure to create DIEs
5792 for abstract instances of a decl first.
5793 (dwarf2out_late_global_decl): Adjust comment.
5794 (output_macinfo_op): With multiple macro sections use
5795 macinfo_label_base to distinguish labels.
5796 (output_macinfo): Likewise. Update macinfo_label_base.
5797 Pass in the line info label.
5798 (note_variable_value_in_expr): When generating LTO resolve
5799 all variable values here by generating DIEs as needed.
5800 (init_sections_and_labels): Add early LTO debug flag parameter
5801 and generate different sections and names if set. Add generation
5802 counter for the labels so we can have multiple of them.
5803 (reset_dies): Helper to allow DIEs to be output multiple times.
5804 (dwarf2out_finish): When outputting DIEs to the fat part of an
5805 LTO object first reset DIEs.
5806 (dwarf2out_early_finish): Output early DIEs when generating LTO.
5807 (modified_type_die): Check for decl_ultimate_origin being self
5809 (gen_type_die_with_usage): Likewise.
5810 (gen_typedef_die): Allow decl_ultimate_origin being self.
5811 (set_decl_abstract_flags): Remove.
5812 (set_block_abstract_flags): Likewise.
5813 (dwarf2out_abstract_function): Treat the early generated DIEs
5814 as the abstract copy and only add DW_AT_inline and
5815 DW_AT_artificial here and call set_decl_origin_self.
5816 If the DIE has an abstract origin don't do anything.
5817 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
5818 if we have none yet (Go fails to build one, PR78628).
5819 (variably_modified_type_p): Prevent endless recursion for Ada
5820 cyclic pointer types.
5821 * lto-streamer-in.c: Include debug.h.
5822 (dref_queue): New global.
5823 (lto_read_tree_1): Stream in DIE references.
5824 (lto_input_tree): Register DIE references.
5825 (input_function): Stream DECL_DEBUG_ARGS.
5826 * lto-streamer-out.c: Include debug.h.
5827 (lto_write_tree_1): Output DIE references.
5828 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
5829 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
5830 (output_function): Stream DECL_DEBUG_ARGS.
5831 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
5832 Stream DECL_ABSTRACT_ORIGIN.
5833 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
5834 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
5835 DECL_CONTEXT for file-scope decls.
5836 * lto-streamer.h (struct dref_entry): Declare.
5837 (dref_queue): Likewise.
5838 * cfgexpand.c (pass_expand::execute): Do not call the
5839 outlining_inline_function hook here.
5840 * lto-wrapper.c (debug_obj): New global.
5841 (tool_cleanup): Unlink it if required.
5842 (debug_objcopy): New function.
5843 (run_gcc): Handle early debug sections in the IL files by
5844 extracting them to separate files, partially linkin them and
5845 feeding the result back as result to the linker.
5846 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
5847 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
5848 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
5849 sections into a separate segment.
5850 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
5852 (darwin_asm_dwarf_section): Likewise.
5853 (darwin_asm_output_dwarf_offset): Likewise.
5854 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
5856 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5857 Alan Hayward <alan.hayward@arm.com>
5858 David Sherwood <david.sherwood@arm.com>
5860 * read-md.h (md_reader::record_potential_iterator_use): Replace
5861 pointer argument with an rtx and an index.
5862 * read-rtl.c (iterator_group::apply_iterator): Likewise.
5863 (apply_mode_iterator): Likewise.
5864 (apply_code_iterator): Likewise.
5865 (apply_int_iterator): Likewise.
5866 (apply_subst_iterator): Likewise.
5867 (record_iterator_use): Likewise.
5868 (record_attribute_use): Likewise.
5869 (md_reader::record_potential_iterator_use): Likewise. Update calls
5870 to record_iterator_use and apply_iterator.
5871 (iterator_use): Replace ptr with x and index.
5872 (attribute_use): Likewise.
5873 (apply_attribute_uses): Update calls to apply_iterator.
5874 (apply_iterators): Likewise. Update initialization of iterator_use.
5875 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
5876 and record_potential_iterator_use.
5877 (rtx_reader::read_rtx_operand): Likewise.
5879 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5880 Alan Hayward <alan.hayward@arm.com>
5881 David Sherwood <david.sherwood@arm.com>
5883 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
5886 2017-08-21 Richard Biener <rguenther@suse.de>
5889 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
5890 at struct end conservatively when comparing common bases.
5892 2017-08-21 Richard Biener <rguenther@suse.de>
5894 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
5895 (mem_ref_in_stmt): Remove.
5896 (determine_max_movement): Use ref index to get at the reference.
5897 (invariantness_dom_walker::before_dom_children): Deal with
5898 lim data already initialized.
5899 (gather_mem_refs_stmt): Initialize lim data and record ref index.
5901 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5903 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
5904 (TARGET_ISA_ROUND): Ditto.
5905 (TARGET_ROUND): Ditto.
5906 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
5907 * config/i386/i386.md: Ditto.
5908 * config/i386/sse.md: Ditto.
5909 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
5910 with OPTION_MASK_ISA_SSE4_1.
5912 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5915 * doc/extend.texi (x86 Built-in Functions): Correct the name of
5916 __builtin_ia32_lzcnt_u16.
5918 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5921 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
5922 (rs6000_set_current_function): Rewrite function to use it.
5924 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
5927 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
5928 and TYPE_WARN_IF_NOT_ALIGN.
5929 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
5930 (handle_warn_if_not_align): New.
5931 (place_union_field): Call handle_warn_if_not_align.
5932 (place_field): Call handle_warn_if_not_align.
5933 Copy TYPE_WARN_IF_NOT_ALIGN.
5934 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
5935 (layout_type): Likewise.
5936 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
5938 (tree_decl_common): Add warn_if_not_align.
5939 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
5940 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
5941 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
5942 (DECL_WARN_IF_NOT_ALIGN): Likewise.
5943 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
5944 * doc/extend.texi: Document warn_if_not_aligned attribute.
5945 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
5947 2017-08-17 Martin Liska <mliska@suse.cz>
5950 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
5951 (get_data_dependence): Use it as pointer type.
5952 (distribute_loop): Likewise.
5954 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5956 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
5957 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
5958 (p8_vmrgow_<mode>_direct): New define_insn.
5959 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
5960 handle endianness for vmrgew and vmrgow permute patterns.
5962 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5964 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
5965 * config/rs6000/rs6000-cpus.def: Remove comment.
5966 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
5967 (POWERPC_MASKS): Likewise.
5968 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
5969 use of TARGET_VSX_TIMODE.
5970 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
5971 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
5972 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
5973 (rs6000_option_override_internal): Remove dead code.
5974 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
5975 (rs6000_legitimize_reload_address): Likewise.
5976 (rs6000_legitimate_address_p): Likewise.
5977 (rs6000_opt_masks): Delete "vsx-timode".
5978 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
5979 from function comment.
5980 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
5981 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
5982 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
5984 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
5985 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
5986 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
5989 (define_peephole2): Likewise.
5991 2017-08-17 Martin Sebor <msebor@redhat.com>
5994 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
5995 past the end of an array.
5996 (test_pp_format): Add test cases.
5998 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
6000 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
6001 missing ECF_NOTHROW flags.
6003 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
6006 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
6007 operands residing in integer registers.
6008 (*vsx_le_perm_load_<mode>): Likewise.
6009 (*vsx_le_perm_store_<mode>): Likewise.
6010 (define_peephole2): Add peepholes to optimize the above.
6012 2017-08-17 Marek Polacek <polacek@redhat.com>
6015 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
6016 to mimic what shorten_compare did. Change the return type to bool.
6017 (fold_cond_expr_with_comparison): Update call to
6018 operand_equal_for_comparison_p.
6019 (fold_ternary_loc): Likewise.
6021 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
6023 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
6025 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
6027 2017-08-17 Richard Biener <rguenther@suse.de>
6029 * tree-ssa-structalias.c (solve_graph): When propagating
6030 to successors update the graphs succ edges and avoid duplicate work.
6032 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
6035 * config/i386/i386.c (ix86_option_override_internal): Save target
6036 specific options after ix86_stack_protector_guard_reg was changed.
6038 2017-08-17 Richard Biener <rguenther@suse.de>
6040 PR tree-optimization/81827
6041 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
6042 (new_var_info): Initialize it conservatively.
6043 (get_call_vi): Mark register vars.
6044 (new_scalar_tmp_constraint_exp): Likewise.
6045 (handle_rhs_call): Likewise.
6046 (handle_const_call): Likewise.
6047 (create_function_info_for): Likewise.
6048 (solve_constraints): Sort varinfos to separate register from
6049 non-register vars to pack points-to solution bitmaps during
6052 2017-08-17 Marek Polacek <polacek@redhat.com>
6054 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
6056 2017-08-17 Richard Biener <rguenther@suse.de>
6058 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
6059 to true when overflow is undefined and we saturated the result.
6061 2017-08-17 Alan Modra <amodra@gmail.com>
6064 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
6065 Don't use store multiple if only one reg needs saving.
6066 (interesting_frame_related_regno): New function.
6067 (rs6000_frame_related): Don't emit frame info for regs that
6069 (rs6000_emit_epilogue): Likewise.
6071 2017-08-16 Nathan Sidwell <nathan@acm.org>
6073 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
6074 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
6075 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
6076 * tree.c (free_lang_data_in_type): Use else-if chain. Always
6077 clear TYPE_LANG_1. Remove obsolete member-function stripping.
6078 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
6079 (verify_type): Adjust for TYPE_BINFO move.
6080 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
6081 process TYPE_BINFO directly.
6082 (hash_tree): Likewise.
6083 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6085 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6088 2017-08-16 David Malcolm <dmalcolm@redhat.com>
6090 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
6092 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
6095 * config/i386/i386.md (*anddi_1_btr): Change predicates of
6096 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
6097 Add ix86_binary_operator_ok to insn constraint.
6098 (*iordi_1_bts): Ditto.
6099 (*xordi_1_btc): Ditto.
6100 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
6101 Update corresponding peephole2 pattern.
6105 2017-08-16 Bin Cheng <bin.cheng@arm.com>
6107 PR tree-optimization/81832
6108 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
6109 copy loop header which has IFN_LOOP_DIST_ALIAS call.
6111 2017-08-16 Marek Polacek <polacek@redhat.com>
6114 * fold-const.c (fold_indirect_ref_1): Restore original behavior
6115 regarding size_zero_node.
6117 2017-08-16 Martin Liska <mliska@suse.cz>
6120 * config.gcc: Respect previously set extra_objs in case
6123 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
6125 PR tree-optimization/81835
6126 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
6127 the phi SSA_NAME. Check that the condition in a COND_EXPR does
6128 not depend on the phi.
6130 2017-08-16 Alan Modra <amodra@gmail.com>
6132 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
6135 2017-08-16 Alan Modra <amodra@gmail.com>
6137 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
6138 (save_reg_p): ..into this. Update all callers.
6139 (first_reg_to_save): Simplify.
6141 2017-08-16 Alan Modra <amodra@gmail.com>
6143 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
6146 2017-08-15 Joseph Myers <joseph@codesourcery.com>
6150 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
6151 constant count if that count is less than 32.
6153 2017-08-15 Nathan Sidwell <nathan@acm.org>
6155 * gcc.c (execute): Emit friendlier message if inferior is killed
6156 by an external cause.
6158 2017-08-15 Richard Biener <rguenther@suse.de>
6160 PR tree-optimization/81790
6161 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
6162 CONSTRUCTORs from simplifying and VN.
6164 2017-08-14 Martin Sebor <msebor@redhat.com>
6166 * builtin-attrs.def: Add comments.
6168 2017-08-14 Martin Sebor <msebor@redhat.com>
6171 * doc/extend.texi (attribute nonstring): Document new attribute.
6173 2017-08-14 Martin Sebor <msebor@redhat.com>
6176 * tree-diagnostic.c (default_tree_printer): Handle %G.
6177 * gimple-pretty-print.h (percent_G_format): Declare new function.
6178 * gimple-pretty-print.c (percent_G_format): Define.
6179 * tree-pretty-print.c (percent_K_format): Add argument.
6181 2017-08-14 Martin Sebor <msebor@redhat.com>
6183 PR translation/79998
6184 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
6185 Remove a stray space.
6187 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
6190 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
6191 (*iordi_1_bts): Ditto.
6192 (*xordi_1_btc): Ditto.
6194 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6197 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
6198 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6200 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
6201 quoted strings, and make more translator-friendly.
6202 (darwin_rs6000_override_options): Likewise.
6203 (rs6000_option_override_internal): Likewise.
6204 (rs6000_return_in_memory): Fix overlong line.
6205 (init_cmulative_args): Use quoted strings, and make more
6206 translator-friendly.
6207 (rs6000_pass_by_reference): Fix overlong line.
6208 (def_builtin): Use quoted strings.
6209 (altivec_expand_predicate_builtin): Use quoted strings, and make
6210 more translator-friendly.
6211 (htm_expand_builtin): Use quoted strings.
6212 (cpu_expand_builtin): Use quoted strings, and make more
6213 translator-friendly.
6214 (altivec_expand_builtin): Likewise.
6215 (paired_expand_predicate_builtin): Likewise.
6216 (rs6000_invalid_builtin): Likewise.
6217 (builtin_function_type): Use quoted strings.
6218 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
6219 more translator-friendly.
6220 (rs6000_trampoline_init): Likewise.
6221 (rs6000_handle_altivec_attribute): Likewise.
6222 (rs6000_inner_target_options): Use quoted strings.
6223 (rs6000_disable_incompatible_switches): Likewise.
6224 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
6225 strings, and make more translator-friendly.
6226 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
6228 2017-08-14 Bin Cheng <bin.cheng@arm.com>
6230 PR tree-optimization/81799
6231 * tree-loop-distribution.c (version_loop_by_alias_check): Force
6232 cond_expr to simple gimple operand.
6234 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
6237 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
6239 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
6243 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
6244 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
6245 TARGET_GASISR_PROLOGUES.
6246 * config/avr/avr.c (avr_option_override): Same.
6247 (avr_pass_pre_proep::execute): Same.
6249 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
6252 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
6253 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
6255 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
6257 * config/i386/i386.md (*load_tp_<mode>): Redefine as
6258 define_insn_and_split. Split to a memory load from 0 in
6259 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
6260 using PTR mode iterator.
6261 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
6262 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
6263 (*add_tp_<mode>): Redefine as define_insn_and_split.
6264 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
6265 address space. Merge with *add_tp_x32 using PTR mode iterator.
6266 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
6267 Split to an add with a memory load from 0 in
6268 DEFAULT_TLS_SEG_REG address space.
6270 2017-08-12 Andrew Pinski <apinski@cavium.com>
6272 * config/aarch64/aarch64-option-extensions.def (rdma):
6273 Fix feature string to what Linux prints out in /proc/cpuinfo.
6275 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
6278 * dwarf2out.c (modified_type_die): For C typedef types that have
6279 an ultimate origin, process the ultimate origin instead of the
6281 (gen_typedef_die): Assert that input DECLs have no ultimate
6283 (gen_type_die_with_usage): For typedef variants that have an
6284 ultimate origin, just call gen_decl_die on the original DECL.
6285 (process_scope_var): Avoid creating DIEs for local typedefs and
6286 concrete static variables.
6288 2017-08-12 Alan Modra <amodra@gmail.com>
6292 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
6293 match gnu-user.h startfile.
6294 (ENDFILE_LINUX_SPEC): Similarly.
6296 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
6299 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
6301 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
6303 2017-08-11 Tamar Christina <tamar.christina@arm.com>
6304 * config/aarch64/aarch64.md (mov<mode>): Change.
6305 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6306 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
6307 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
6309 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
6311 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
6312 for storage order barriers.
6314 2017-08-11 Martin Liska <mliska@suse.cz>
6317 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
6318 variables of void type.
6320 2017-08-11 Martin Liska <mliska@suse.cz>
6322 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
6323 TARGET_SUPPORTS_ALIASES.
6324 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
6325 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
6326 (optimize_weakref): Likewise.
6327 * symtab.c (symtab_node::noninterposable_alias): Likewise.
6328 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
6329 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
6331 2017-08-11 Martin Liska <mliska@suse.cz>
6334 * config/i386/i386.c (make_resolver_func): Do complete
6335 refactoring of the function.
6337 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6340 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
6341 * config/i386/i386.c (ix86_stack_protect_guard): Use
6342 ix86_stack_protect_guard_symbol_str to generate varible declaration.
6343 * doc/invoke.texi (x86 Options): Document
6344 -mstack-protector-guard-symbol= option.
6346 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6348 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
6349 * config/i386/i386.c (ix86_split_stack_guard): New function.
6350 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
6351 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
6352 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
6353 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
6354 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
6355 (split_stack_space_check): Call ix86_split_stack_guard.
6357 2017-08-10 Martin Sebor <msebor@redhat.com>
6359 * print-tree.c (print_node): Print location using the established
6361 Replace spaces with colons.
6362 (debug_raw, debug): Ditto.
6364 2017-08-10 Martin Sebor <msebor@redhat.com>
6367 * pretty-print.c (pp_format): Correct the handling of %s precision.
6369 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
6372 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6374 (ix86_finalize_stack_frame_flags): This. Also clear
6375 frame_pointer_needed if -fno-omit-frame-pointer is used without
6377 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6378 with ix86_finalize_stack_frame_flags.
6379 (ix86_expand_epilogue): Likewise.
6380 (ix86_expand_split_stack_prologue): Likewise.
6381 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
6383 2017-08-10 Martin Liska <mliska@suse.cz>
6386 * c-attribs.c (handle_target_attribute):
6387 Report warning for an empty string argument of target attribute.
6389 2017-08-09 Jakub Jelinek <jakub@redhat.com>
6392 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
6394 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
6395 or DECL_NONLOCAL labels.
6396 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
6397 or DECL_NONLOCAL labels here.
6399 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
6401 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
6402 to indicate when early gimple folding has been disabled.
6403 (rs6000_gimple_fold_builtin): Add debug content.
6404 (rs6000_invalid_builtin): Fix whitespace.
6405 (rs6000_expand_builtin): Fix whitespace.
6406 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
6408 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
6411 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
6412 SAVE_MULTIPLE if not all the registers that saves, should be saved.
6414 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
6416 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
6417 (qdf24xx): Likewise.
6418 * config/aarch64/aarch64.md: Include falkor.md.
6419 * config/aarch64/falkor.md: New.
6421 2017-08-09 Marek Polacek <polacek@redhat.com>
6424 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
6425 * diagnostic.c (emit_diagnostic): Add a comment.
6426 (emit_diagnostic_valist): New function.
6428 2017-08-09 Marek Polacek <polacek@redhat.com>
6431 * input.c (make_location): New overload.
6432 * input.h (make_location): Declare.
6434 2017-08-08 Alan Modra <amodra@gmail.com>
6435 H.J. Lu <hongjiu.lu@intel.com>
6438 * gcc.c (NO_PIE_SPEC): Delete.
6439 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6441 (LINK_PIE_SPEC): ..to here.
6442 (LINK_COMMAND_SPEC): Support -no-pie.
6443 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6444 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6446 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6447 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6448 (ENDFILE_CRTEND_SPEC): Similarly.
6450 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6453 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6454 (mstack-protector-guard-offset=): Ditto.
6455 * config/i386/i386.c (ix86_option_override): Handle
6456 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6458 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6459 ix86_stack_protect_guard_offset variables.
6460 (TARGET_STACK_PROTECT_GUARD): Always define.
6461 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6462 and -mstack-protector-guard-offset= options.
6464 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6466 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6467 boundary case for the last candidate.
6469 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6471 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6472 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6474 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6477 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6478 * config/aarch64/aarch64-builtins.c
6479 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6480 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6481 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6483 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6484 Andrew Pinski <pinskia@gmail.com>
6487 * internal-fn.def (XORSIGN): New.
6488 * optabs.def (xorsign_optab): New.
6489 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6490 (convert_expand_mult_copysign): New.
6491 (pass_optimize_widening_mul::execute): Call
6492 convert_expand_mult_copysign.
6494 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6496 PR tree-optimization/81354
6497 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6498 Insert on edges rather than explicitly creating landing pads.
6499 (analyze_candidates_and_replace): Commit edge inserts.
6501 2017-08-08 Richard Biener <rguenther@suse.de>
6504 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6505 (expand_simple_operations): Also look through ADDR_EXPRs with
6506 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6508 2017-08-08 Richard Biener <rguenther@suse.de>
6510 PR tree-optimization/81723
6511 * tree-vect-slp.c (struct bst_traits): New hash traits.
6512 (bst_fail): New global.
6513 (vect_build_slp_tree_2): New worker, split out from ...
6514 (vect_build_slp_tree): ... this now wrapping it with using
6515 bst_fail set to cache SLP tree build fails. Properly handle
6517 (vect_analyze_slp_instance): Allocate and free bst_fail.
6519 2017-08-08 Martin Liska <mliska@suse.cz>
6522 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6523 LABEL_DECLs that can be from a different function.
6525 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6527 PR tree-optimization/81744
6528 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6529 loop's number of iterations.
6531 2017-08-08 Martin Liska <mliska@suse.cz>
6533 * asan.c: Include header files.
6534 * attribs.c (build_decl_attribute_variant): New function moved
6536 (build_type_attribute_qual_variant): Likewise.
6537 (cmp_attrib_identifiers): Likewise.
6538 (simple_cst_list_equal): Likewise.
6539 (omp_declare_simd_clauses_equal): Likewise.
6540 (attribute_value_equal): Likewise.
6541 (comp_type_attributes): Likewise.
6542 (build_type_attribute_variant): Likewise.
6543 (lookup_ident_attribute): Likewise.
6544 (remove_attribute): Likewise.
6545 (merge_attributes): Likewise.
6546 (merge_type_attributes): Likewise.
6547 (merge_decl_attributes): Likewise.
6548 (merge_dllimport_decl_attributes): Likewise.
6549 (handle_dll_attribute): Likewise.
6550 (attribute_list_equal): Likewise.
6551 (attribute_list_contained): Likewise.
6552 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6553 (lookup_attribute_by_prefix): Likewise.
6554 * bb-reorder.c: Include header files.
6555 * builtins.c: Likewise.
6556 * calls.c: Likewise.
6557 * cfgexpand.c: Likewise.
6558 * cgraph.c: Likewise.
6559 * cgraphunit.c: Likewise.
6560 * convert.c: Likewise.
6561 * dwarf2out.c: Likewise.
6562 * final.c: Likewise.
6563 * fold-const.c: Likewise.
6564 * function.c: Likewise.
6565 * gimple-expr.c: Likewise.
6566 * gimple-fold.c: Likewise.
6567 * gimple-pretty-print.c: Likewise.
6568 * gimple.c: Likewise.
6569 * gimplify.c: Likewise.
6570 * hsa-common.c: Likewise.
6571 * hsa-gen.c: Likewise.
6572 * internal-fn.c: Likewise.
6573 * ipa-chkp.c: Likewise.
6574 * ipa-cp.c: Likewise.
6575 * ipa-devirt.c: Likewise.
6576 * ipa-fnsummary.c: Likewise.
6577 * ipa-inline.c: Likewise.
6578 * ipa-visibility.c: Likewise.
6580 * lto-cgraph.c: Likewise.
6581 * omp-expand.c: Likewise.
6582 * omp-general.c: Likewise.
6583 * omp-low.c: Likewise.
6584 * omp-offload.c: Likewise.
6585 * omp-simd-clone.c: Likewise.
6586 * opts-global.c: Likewise.
6587 * passes.c: Likewise.
6588 * predict.c: Likewise.
6589 * sancov.c: Likewise.
6590 * sanopt.c: Likewise.
6591 * symtab.c: Likewise.
6592 * toplev.c: Likewise.
6593 * trans-mem.c: Likewise.
6594 * tree-chkp.c: Likewise.
6595 * tree-eh.c: Likewise.
6596 * tree-into-ssa.c: Likewise.
6597 * tree-object-size.c: Likewise.
6598 * tree-parloops.c: Likewise.
6599 * tree-profile.c: Likewise.
6600 * tree-ssa-ccp.c: Likewise.
6601 * tree-ssa-live.c: Likewise.
6602 * tree-ssa-loop.c: Likewise.
6603 * tree-ssa-sccvn.c: Likewise.
6604 * tree-ssa-structalias.c: Likewise.
6605 * tree-ssa.c: Likewise.
6606 * tree-streamer-in.c: Likewise.
6607 * tree-vectorizer.c: Likewise.
6608 * tree-vrp.c: Likewise.
6610 * ubsan.c: Likewise.
6611 * varasm.c: Likewise.
6612 * varpool.c: Likewise.
6613 * tree.c: Remove functions moved to attribs.[ch].
6615 * config/aarch64/aarch64.c: Add attrs.h header file.
6616 * config/alpha/alpha.c: Likewise.
6617 * config/arc/arc.c: Likewise.
6618 * config/arm/arm.c: Likewise.
6619 * config/avr/avr.c: Likewise.
6620 * config/bfin/bfin.c: Likewise.
6621 * config/c6x/c6x.c: Likewise.
6622 * config/cr16/cr16.c: Likewise.
6623 * config/cris/cris.c: Likewise.
6624 * config/darwin.c: Likewise.
6625 * config/epiphany/epiphany.c: Likewise.
6626 * config/fr30/fr30.c: Likewise.
6627 * config/frv/frv.c: Likewise.
6628 * config/ft32/ft32.c: Likewise.
6629 * config/h8300/h8300.c: Likewise.
6630 * config/i386/winnt.c: Likewise.
6631 * config/ia64/ia64.c: Likewise.
6632 * config/iq2000/iq2000.c: Likewise.
6633 * config/lm32/lm32.c: Likewise.
6634 * config/m32c/m32c.c: Likewise.
6635 * config/m32r/m32r.c: Likewise.
6636 * config/m68k/m68k.c: Likewise.
6637 * config/mcore/mcore.c: Likewise.
6638 * config/microblaze/microblaze.c: Likewise.
6639 * config/mips/mips.c: Likewise.
6640 * config/mmix/mmix.c: Likewise.
6641 * config/mn10300/mn10300.c: Likewise.
6642 * config/moxie/moxie.c: Likewise.
6643 * config/msp430/msp430.c: Likewise.
6644 * config/nds32/nds32-isr.c: Likewise.
6645 * config/nds32/nds32.c: Likewise.
6646 * config/nios2/nios2.c: Likewise.
6647 * config/nvptx/nvptx.c: Likewise.
6648 * config/pa/pa.c: Likewise.
6649 * config/pdp11/pdp11.c: Likewise.
6650 * config/powerpcspe/powerpcspe.c: Likewise.
6651 * config/riscv/riscv.c: Likewise.
6652 * config/rl78/rl78.c: Likewise.
6653 * config/rx/rx.c: Likewise.
6654 * config/s390/s390.c: Likewise.
6655 * config/sh/sh.c: Likewise.
6656 * config/sol2.c: Likewise.
6657 * config/sparc/sparc.c: Likewise.
6658 * config/spu/spu.c: Likewise.
6659 * config/stormy16/stormy16.c: Likewise.
6660 * config/tilegx/tilegx.c: Likewise.
6661 * config/tilepro/tilepro.c: Likewise.
6662 * config/v850/v850.c: Likewise.
6663 * config/vax/vax.c: Likewise.
6664 * config/visium/visium.c: Likewise.
6665 * config/xtensa/xtensa.c: Likewise.
6667 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6670 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
6671 constraints since the -mupper-regs-* switches have been
6673 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
6674 into a vector from a double word element that was extracted from
6675 another vector, and eliminate extra XXPERMDI instructions.
6676 (vsx_concat_<mode>_2): Likewise.
6677 (vsx_concat_<mode>_3): Likewise.
6678 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
6679 concat to allow optimizing inserts from previous extracts.
6681 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
6683 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6684 memory reference to a SSP offset in TLS address space.
6685 (ix86_print_operand) <case '@'>: Remove.
6686 (ix86_print_operand_punct_valid_p): Remove '@' code.
6687 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6689 (stack_tls_protect_set_<mode>): Remove.
6690 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6691 (stack_tls_protect_test_<mode>): Remove.
6692 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6694 2017-08-07 Olivier Hainque <hainque@adacore.com>
6697 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6699 2017-08-07 Douglas Rupp <rupp@adacore.com>
6701 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6702 variable was referenced as multidir in command.
6704 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6707 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6710 2017-08-07 Martin Liska <mliska@suse.cz>
6712 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6713 * config/rl78/rl78.c: Add include of attribs.h.
6714 * config/sh/sh.c: Likewise.
6715 * config/v850/v850.c: Likewise.
6717 2017-08-07 Tom de Vries <tom@codesourcery.com>
6720 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6722 2017-08-07 Martin Liska <mliska@suse.cz>
6724 * config/mips/mips.c: Include attribs.h.
6726 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6729 * doc/invoke.texi: Document change in behvaior for -Ofast for
6732 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6734 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6736 (aarch64_pop_regs): Likewise.
6737 (aarch64_gen_load_pair): Likewise.
6738 (aarch64_save_callee_saves): Likewise.
6739 (aarch64_restore_callee_saves): Likewise.
6741 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6743 * config/i386/i386.c: Revert the last change.
6745 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6748 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6750 (ix86_finalize_stack_frame_flags): This. Also clear
6751 frame_pointer_needed if -fno-omit-frame-pointer is used without
6753 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6754 with ix86_finalize_stack_frame_flags.
6755 (ix86_expand_epilogue): Likewise.
6756 (ix86_expand_split_stack_prologue): Likewise.
6758 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6761 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6762 to P_AES for Westmere.
6764 2017-08-07 Jonathan Yong <10walls@gmail.com>
6766 * config/i386/mingw.opt (fset-stack-executable): Removed.
6767 * config/i386/cygming.opt (fset-stack-executable): Moved
6769 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
6771 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
6773 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
6775 2017-08-07 Marek Polacek <polacek@redhat.com>
6778 * fold-const.c (fold_indirect_ref_1): Check type_domain.
6780 2017-08-07 Martin Liska <mliska@suse.cz>
6782 * attribs.h (canonicalize_attr_name): New function.
6783 (cmp_attribs): Move from c-format.c and adjusted.
6784 (is_attribute_p): Moved from tree.h.
6785 * tree-inline.c: Add new includes.
6786 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
6787 (private_is_attribute_p): Remove.
6788 (private_lookup_attribute): Likewise.
6789 (private_lookup_attribute_by_prefix): Simplify.
6790 (remove_attribute): Use is_attribute_p.
6791 * tree.h: Remove removed declarations.
6793 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6796 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
6797 instead of computing it in the function. Formatting fix.
6798 (expand_case): Don't rely on default_edge being the first edge,
6799 clear it if removing it, pass default_edge to
6800 emit_case_dispatch_table.
6801 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
6804 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
6806 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
6807 insn in the function, emit NOP after the insn.
6809 2017-08-06 Tom de Vries <tom@codesourcery.com>
6811 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
6814 2017-08-06 Tom de Vries <tom@codesourcery.com>
6816 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
6819 2017-08-04 Yury Gribov <tetra2005@gmail.com>
6821 PR tree-optimization/57371
6822 * match.pd: New pattern.
6824 2017-08-04 Marek Polacek <polacek@redhat.com>
6827 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
6828 perform the computation in offset_int.
6830 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6832 PR tree-optimization/81136
6833 * tree-vectorizer.h: Include tree-hash-traits.h.
6834 (vec_base_alignments): New typedef.
6835 (vec_info): Add a base_alignments field.
6836 (vect_record_base_alignments): Declare.
6837 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
6839 (DR_IS_CONDITIONAL_IN_STMT): New macro.
6840 (create_data_ref): Add an is_conditional_in_stmt argument.
6841 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
6842 the is_conditional_in_stmt field.
6843 (data_ref_loc): Add an is_conditional_in_stmt field.
6844 (get_references_in_stmt): Set the is_conditional_in_stmt field.
6845 (find_data_references_in_stmt): Update call to create_data_ref.
6846 (graphite_find_data_references_in_stmt): Likewise.
6847 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
6848 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
6849 (vect_record_base_alignment): New function.
6850 (vect_record_base_alignments): Likewise.
6851 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
6852 for nested statements even if we fail to compute a misalignment.
6853 Use pooled base alignments for unconditional references.
6854 (vect_find_same_alignment_drs): Compare base addresses instead
6856 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
6857 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
6859 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6861 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
6862 Add an explicit name for the enum. Use auto_vec for slp_instances
6864 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
6866 (_bb_vec_info): Add a constructor and destructor.
6867 (vinfo_for_stmt): Return NULL for uids of -1 as well.
6868 (destroy_loop_vec_info): Delete.
6869 (vect_destroy_datarefs): Likewise.
6870 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
6871 (vec_info::vec_info): New function.
6872 (vec_info::~vec_info): Likewise.
6873 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
6874 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
6875 destroy_loop_vec_info.
6876 * tree-vect-loop.c (new_loop_vec_info): Replace with...
6877 (_loop_vec_info::_loop_vec_info): ...this.
6878 (destroy_loop_vec_info): Replace with...
6879 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
6880 the stmt_vec_infos. Leave handling of vec_info information to its
6881 destructor. Remove explicit vector releases.
6882 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
6883 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
6884 * tree-vect-slp.c (new_bb_vec_info): Replace with...
6885 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
6886 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
6887 (destroy_bb_vec_info): Replace with...
6888 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
6889 information to its destructor.
6890 (vect_slp_analyze_bb_1): Use new and delete instead of
6891 new_bb_vec_info and destroy_bb_vec_info.
6892 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
6895 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6897 * tree-data-ref.h (subscript): Add access_fn field.
6898 (data_dependence_relation): Add could_be_independent_p.
6899 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
6900 (same_access_functions): Move to tree-data-ref.c.
6901 * tree-data-ref.c (ref_contains_union_access_p): New function.
6902 (access_fn_component_p): Likewise.
6903 (access_fn_components_comparable_p): Likewise.
6904 (dr_analyze_indices): Add a reference to access_fn_component_p.
6905 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
6907 (constant_access_functions): Likewise.
6908 (add_other_self_distances): Likewise.
6909 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
6910 (initialize_data_dependence_relation): Use XCNEW and remove
6911 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
6912 of access functions that have the same type. Allow the
6913 subsequence to end with different bases in some circumstances.
6914 Record the chosen access functions in SUB_ACCESS_FN.
6915 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
6916 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
6917 (subscript_dependence_tester_1): Likewise dra and drb.
6918 (build_classic_dist_vector): Update calls accordingly.
6919 (subscript_dependence_tester): Likewise.
6920 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
6921 DDR_COULD_BE_INDEPENDENT_P.
6922 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
6923 comp_alias_ddrs instead of may_alias_ddrs.
6924 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6926 (vect_analyze_data_ref_dependence): Use it if
6927 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
6928 distance vectors if that fails.
6929 (dependence_distance_ge_vf): New function.
6930 (vect_prune_runtime_alias_test_list): Use it. Don't clear
6931 LOOP_VINFO_MAY_ALIAS_DDRS.
6933 2017-08-04 Richard Biener <rguenther@suse.de>
6936 * fold-const.c (fold_binary_loc): Properly restrict
6937 minus_var0 && minus_var1 case when associating undefined overflow
6940 2017-08-04 Tom de Vries <tom@codesourcery.com>
6942 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
6944 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6946 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6947 Don't start diagnostic messages with a capital letter.
6948 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6950 (rs6000_invalid_builtin): Likewise.
6951 (rs6000_trampoline_init): Likewise.
6953 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6956 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
6957 after setting changeable df flags.
6959 2017-08-03 Richard Biener <rguenther@suse.de>
6961 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
6962 up if the use is in USE - X.
6964 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
6966 * toplev.c (dumpfile.h): New include.
6967 (internal_error_reentered): New static function. Use it...
6968 (internal_error_function): ...here to handle reentered internal_error.
6970 2017-08-03 Richard Biener <rguenther@suse.de>
6973 * fold-const.c (split_tree): Add minus_var and minus_con
6974 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
6975 here but always use minus_*.
6976 (associate_trees): Assert we never associate with MINUS_EXPR
6977 and NULL first operand. Do not recurse for PLUS_EXPR operands
6978 when associating as MINUS_EXPR either.
6979 (fold_binary_loc): Track minus_var and minus_con.
6981 2017-08-03 Tom de Vries <tom@codesourcery.com>
6984 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
6985 ACCEL_COMPILER, apply finish_options on
6986 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
6988 2017-08-03 Tom de Vries <tom@codesourcery.com>
6991 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
6992 function_entry_patch_area_size > 0.
6994 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6997 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
6998 instead of 10??LU, perform unit multiplication in wide_int,
6999 don't change alloc_object_size_limit if the limit is larger
7002 PR tree-optimization/81655
7003 PR tree-optimization/81588
7004 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
7005 the case when ranges[i].low and high are 1 for unsigned type with
7009 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
7010 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
7012 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7014 * tree-vrp.h: Add include guard.
7016 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
7019 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
7020 (ud2): New insn pattern.
7021 * config/i386/i386.c (ix86_expand_epilogue):
7022 For naked functions, generate ud2 instead of trap insn.
7024 2017-08-02 Marek Polacek <polacek@redhat.com>
7027 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
7029 2017-08-02 Tom de Vries <tom@codesourcery.com>
7030 Cesar Philippidis <cesar@codesourcery.com>
7032 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
7033 Add missing edge probabilities.
7035 2017-08-02 Tamar Christina <tamar.christina@arm.com>
7037 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
7040 2017-08-02 Jakub Jelinek <jakub@redhat.com>
7043 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
7044 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
7045 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
7046 prologue_seq sequences - if any.
7048 2017-08-02 Richard Biener <rguenther@suse.de>
7050 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
7051 via vectors if supported, integer extracts via punning if supported
7052 or otherwise vector extracts.
7054 2017-08-02 Richard Biener <rguenther@suse.de>
7056 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
7058 (bitmap_insert_into_set): ... this.
7060 2017-08-02 Richard Biener <rguenther@suse.de>
7062 PR tree-optimization/81633
7064 2015-08-17 Alan Hayward <alan.hayward@arm.com>
7066 PR tree-optimization/71752
7067 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
7069 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
7071 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
7072 (machine_function::call_ms2sysv_pad_out): Remove field.
7073 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
7074 (ix86_compute_frame_layout): Likewise.
7076 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
7079 * config/i386/i386.c (ix86_set_func_type): Disallow naked
7080 attribute with interrupt attribute.
7082 2017-08-01 Andrew Pinski <apinski@cavium.com>
7084 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
7085 BIT_INSERT_EXPR's operand 1
7086 to see if the types precision matches.
7088 2017-08-01 Martin Liska <mliska@suse.cz>
7091 * builtins.c (expand_builtin_memcpy_args): Remove.
7092 (expand_builtin_memcpy): Call newly added function
7093 expand_builtin_memory_copy_args.
7094 (expand_builtin_memcpy_with_bounds): Likewise.
7095 (expand_builtin_mempcpy): Remove last argument.
7096 (expand_builtin_mempcpy_with_bounds): Likewise.
7097 (expand_builtin_memory_copy_args): New function created from
7098 expand_builtin_mempcpy_args with small modifications.
7099 (expand_builtin_mempcpy_args): Remove.
7100 (expand_builtin_stpcpy): Remove unused argument.
7101 (expand_builtin): Likewise.
7102 (expand_builtin_with_bounds): Likewise.
7104 2017-08-01 Martin Liska <mliska@suse.cz>
7107 Make mempcpy more optimal (PR middle-end/70140).
7109 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7112 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7113 __builtin_vec_cmpne verify both arguments are compatible vectors
7114 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
7115 verify arg1_type is a pointer or array type. For __builtin_vec_st,
7116 move computation of aligned to after checking the argument types.
7120 * config/rs6000/vsx.md (vextract_fp_from_shorth,
7121 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
7124 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
7125 Doug Rupp <rupp@adacore.com>
7126 Olivier Hainque <hainque@adacore.com>
7128 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
7129 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
7131 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
7132 for TARGET_OS_CPP_BUILTIN.
7133 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
7134 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
7136 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
7137 passing required abi options to the assembler for EABI configurations.
7138 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
7139 of .text.hot and .text.unlikely sections for kernel modules when
7140 using ARM style exceptions.
7141 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
7142 options. Add EXTRA_CC1_SPEC.
7143 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
7145 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
7147 (ARM_TARGET2_DWARF_FORMAT): Define.
7148 * config/arm/t-vxworks: Adjust multilib control to removal of the
7149 Diab command line options.
7151 2017-08-01 Martin Liska <mliska@suse.cz>
7153 PR gcov-profile/81561
7154 * gcov.c (unblock): Make unblocking safe as we need to preserve
7155 index correspondence of blocks and block_lists.
7157 2017-08-01 Richard Biener <rguenther@suse.de>
7159 PR tree-optimization/81181
7160 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7161 (compute_antic): ... end of iteration here.
7163 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
7165 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
7166 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
7167 (ftree-slp-vectorize): Likewise.
7168 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
7169 can no longer be set independent of flag_tree_loop_vectorize.
7170 * omp-general.c (emp_max_vf): Likewise.
7171 * opts.c (enable_fdo_optimizations): Remove references to
7172 flag_tree_vectorize, these are now implicit.
7173 (common_handle_option): Remove handling for OPT_ftree_vectorize,
7174 and leave it for the options machinery.
7176 2017-08-01 Martin Liska <mliska@suse.cz>
7179 * builtins.c (expand_builtin_memcpy_args): Remove.
7180 (expand_builtin_memcpy): Call newly added function
7181 expand_builtin_memory_copy_args.
7182 (expand_builtin_memcpy_with_bounds): Likewise.
7183 (expand_builtin_mempcpy): Remove last argument.
7184 (expand_builtin_mempcpy_with_bounds): Likewise.
7185 (expand_builtin_memory_copy_args): New function created from
7186 expand_builtin_mempcpy_args with small modifications.
7187 (expand_builtin_mempcpy_args): Remove.
7188 (expand_builtin_stpcpy): Remove unused argument.
7189 (expand_builtin): Likewise.
7190 (expand_builtin_with_bounds): Likewise.
7192 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7195 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
7196 print "ds:" only for immediates in generic address space.
7198 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7201 * config/i386/i386.c (ix86_funciton_naked): New prototype.
7202 (ix86_function_ok_for_sibcall): Return false for naked functions.
7204 2017-08-01 Richard Biener <rguenther@suse.de>
7206 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
7207 (compute_antic): Seed worklist with exit block predecessors.
7208 * cfganal.c (dfs_find_deadend): For a cycle return the source
7209 of the edge closing it.
7211 2017-08-01 Tamar Christina <tamar.christina@arm.com>
7213 * config/aarch64/aarch64.c
7214 (aarch64_can_const_movi_rtx_p): Move 0 check.
7216 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7218 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
7219 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
7221 * match.pd: Ditto in address comparison pattern.
7223 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7225 PR tree-optimization/81627
7226 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
7227 closed ssa form for store-store chain.
7229 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7231 PR tree-optimization/81620
7232 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
7233 for store-store chain.
7235 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7237 PR tree-optimization/81588
7238 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
7239 ranges[i].in_p, invert comparison code ccode. For >/>=,
7240 swap rhs1 and rhs2 and comparison code unconditionally,
7241 for </<= don't do that. Don't swap rhs1/rhs2 again if
7242 ranges[i].in_p, instead invert comparison code ccode if
7243 opcode or oe->rank is BIT_IOR_EXPR.
7246 * optabs.def (vec_extract_optab, vec_init_optab): Change from
7247 a direct optab to conversion optab.
7248 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7249 with GET_MODE_INNER as last argument instead of optab_handler.
7250 * expmed.c (extract_bit_field_1): Likewise. Use vector from
7251 vector extraction if possible and optab is available.
7252 * expr.c (store_constructor): Use convert_optab_handler instead
7253 of optab_handler. Use vector initialization from smaller
7254 vectors if possible and optab is available.
7255 * tree-vect-stmts.c (vectorizable_load): Likewise.
7256 * doc/md.texi (vec_extract, vec_init): Document that the optabs
7258 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
7259 of vec_init from half-sized vectors with the same element mode.
7260 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
7261 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
7262 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
7263 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
7264 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
7265 after mode in gen_vec_extract* calls.
7266 (vec_extract<mode>): Renamed to ...
7267 (vec_extract<mode><ssescalarmodelower>): ... this.
7268 (vec_extract<mode><ssehalfvecmodelower>): New expander.
7269 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
7270 element mode after mode in gen_vec_init* calls.
7271 (VEC_INIT_HALF_MODE): New mode iterator.
7272 (vec_init<mode>): Renamed to ...
7273 (vec_init<mode><ssescalarmodelower>): ... this.
7274 (vec_init<mode><ssehalfvecmodelower>): New expander.
7275 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
7276 (vec_extractv2sfsf): ... this.
7277 (vec_initv2sf): Renamed to ...
7278 (vec_initv2sfsf): ... this.
7279 (vec_extractv2si): Renamed to ...
7280 (vec_extractv2sisi): ... this.
7281 (vec_initv2si): Renamed to ...
7282 (vec_initv2sisi): ... this.
7283 (vec_extractv4hi): Renamed to ...
7284 (vec_extractv4hihi): ... this.
7285 (vec_initv4hi): Renamed to ...
7286 (vec_initv4hihi): ... this.
7287 (vec_extractv8qi): Renamed to ...
7288 (vec_extractv8qiqi): ... this.
7289 (vec_initv8qi): Renamed to ...
7290 (vec_initv8qiqi): ... this.
7291 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
7292 (vec_init<mode>): Renamed to ...
7293 (vec_init<mode><VEC_base_l>): ... this.
7294 (vec_extract<mode>): Renamed to ...
7295 (vec_extract<mode><VEC_base_l>): ... this.
7296 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
7297 (vec_initv2sfsf): ... this.
7298 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
7299 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7300 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
7301 element mode after mode in gen_vec_init* calls.
7302 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
7303 (vec_init<mode><Vel>): ... this.
7304 (vec_extract<mode>): Renamed to ...
7305 (vec_extract<mode><Vel>): ... this.
7306 * config/aarch64/iterators.md (Vel): New mode attribute.
7307 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
7308 Add element mode after mode in gen_vec_extract* calls.
7309 * config/s390/vector.md (non_vec_l): New mode attribute.
7310 (vec_extract<mode>): Renamed to ...
7311 (vec_extract<mode><non_vec_l>): ... this.
7312 (vec_init<mode>): Renamed to ...
7313 (vec_init<mode><non_vec_l>): ... this.
7314 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
7315 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
7317 * config/arm/iterators.md (V_elem_l): New mode attribute.
7318 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
7319 (vec_extract<mode><V_elem_l>): ... this.
7320 (vec_extractv2di): Renamed to ...
7321 (vec_extractv2didi): ... this.
7322 (vec_init<mode>): Renamed to ...
7323 (vec_init<mode><V_elem_l>): ... this.
7324 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
7325 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
7326 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
7327 Add element mode after gen_vec_extract* calls.
7328 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
7329 (vec_init<mode><unitmode>): ... this.
7330 (vec_extract<mode>): Renamed to ...
7331 (vec_extract<mode><unitmode>): ... this.
7332 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
7333 (vec_init<mode><unitmode>): ... this.
7334 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
7335 (vec_initv2sfsf): ... this.
7336 (vec_extractv2sf): Renamed to ...
7337 (vec_extractv2sfsf): ... this.
7338 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
7339 Add element mode after gen_vec_extract* calls.
7340 * config/mips/mips.md (unitmode): New mode iterator.
7341 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
7342 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
7343 * config/spu/spu.md (inner_l): New mode attribute.
7344 (vec_init<mode>): Renamed to ...
7345 (vec_init<mode><inner_l>): ... this.
7346 (vec_extract<mode>): Renamed to ...
7347 (vec_extract<mode><inner_l>): ... this.
7348 * config/sparc/sparc.md (veltmode): New mode iterator.
7349 (vec_init<VMALL:mode>): Renamed to ...
7350 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
7351 * config/ia64/vect.md (vec_initv2si): Renamed to ...
7352 (vec_initv2sisi): ... this.
7353 (vec_initv2sf): Renamed to ...
7354 (vec_initv2sfsf): ... this.
7355 (vec_extractv2sf): Renamed to ...
7356 (vec_extractv2sfsf): ... this.
7357 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
7358 (vec_init<mode>): Renamed to ...
7359 (vec_init<mode><VEC_base_l>): ... this.
7360 (vec_extract<mode>): Renamed to ...
7361 (vec_extract<mode><VEC_base_l>): ... this.
7362 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
7363 (vec_initv2sfsf): ... this.
7364 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
7365 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7366 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
7367 gen_vec_init* calls.
7369 2017-08-01 Richard Biener <rguenther@suse.de>
7371 PR tree-optimization/81297
7372 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
7373 TREE_OVERFLOW from INTEGER_CSTs.
7375 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
7377 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
7379 2017-07-31 Carl Love <cel@us.ibm.com>
7381 * config/rs6000/rs6000-c: Add support for built-in functions
7382 vector signed char vec_xl_be (signed long long, signed char *);
7383 vector unsigned char vec_xl_be (signed long long, unsigned char *);
7384 vector signed int vec_xl_be (signed long long, signed int *);
7385 vector unsigned int vec_xl_be (signed long long, unsigned int *);
7386 vector signed long long vec_xl_be (signed long long, signed long long *);
7387 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
7388 vector signed short vec_xl_be (signed long long, signed short *);
7389 vector unsigned short vec_xl_be (signed long long, unsigned short *);
7390 vector double vec_xl_be (signed long long, double *);
7391 vector float vec_xl_be (signed long long, float *);
7392 * config/rs6000/altivec.h (vec_xl_be): Add #define.
7393 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
7394 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
7396 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
7397 (altivec_expand_builtin): Add switch statement to call
7398 altivec_expand_xl_be for each builtin.
7399 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
7400 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
7401 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
7402 __builtin_vsx_le_be_v16qi.
7403 * doc/extend.texi: Update the built-in documentation file for the
7404 new built-in functions.
7406 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7409 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
7411 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7413 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7415 * config.gcc: Add z14.
7416 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7417 CPU model numbers for z13s and z14.
7418 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
7420 * config/s390/s390-opts.h (enum processor_type): Rename
7421 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7422 * config/s390/s390.c (processor_table): Add field for CPU name to
7423 be passed to Binutils.
7424 (s390_asm_output_machine_for_arch): Use the new field in
7425 processor_table for Binutils.
7426 (s390_expand_builtin): Replace arch12 with z14.
7427 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7428 (s390_get_sched_attrmask): Likewise.
7429 (s390_get_unit_mask): Likewise.
7430 * config/s390/s390.opt: Add z14 to processor_type enum.
7432 2017-07-31 Martin Jambor <mjambor@suse.cz>
7435 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7436 regardless of optimization level.
7438 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7439 Martin Liska <mliska@suse.cz>
7441 * predict.def: Remove old comment and adjust probability.
7442 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7445 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7448 * config/i386/i386.c (ix86_function_naked): New function.
7449 (ix86_can_use_return_insn_p): Return false for naked functions.
7450 (ix86_expand_prologue): Skip prologue for naked functions.
7451 (ix86_expand_epilogue): Skip epilogue for naked functions
7452 and emit trap instruction.
7453 (ix86_warn_func_return): New function.
7454 (ix86_attribute_table): Add "naked" attribute specification.
7455 (TARGET_WARN_FUNC_RETURN): Define.
7456 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7458 2017-07-31 Martin Liska <mliska@suse.cz>
7460 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7461 (dump_gimple_bb_header): Always dump BB info.
7462 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7464 2017-07-31 Martin Liska <mliska@suse.cz>
7467 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7468 also with current_function_decl non-null equality.
7470 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7473 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7474 change type to the element type, instead add eltype variable and
7475 use it where we are interested in the element type.
7477 PR tree-optimization/81603
7478 * ipa-polymorphic-call.c
7479 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7480 offset arithmetic in offset_int, bail out if the resulting bit offset
7481 doesn't fit into shwi.
7483 2017-07-31 Martin Liska <mliska@suse.cz>
7485 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7486 (gimplify_save_expr): Fix comment.
7488 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7491 * config/i386/i386.c (ix86_function_arg): Update arguments for
7493 (ix86_compute_frame_layout): Set the initial stack offset to
7494 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7495 INCOMING_FRAME_SP_OFFSET.
7496 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7497 stack before exception handler returns.
7498 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7499 the 'ERROR_CODE' for exception handler.
7501 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7503 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7504 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7505 (ASM_OUTPUT_REG_POP): Ditto.
7506 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7507 instead of asm_fprintf to output pure string.
7509 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7511 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7512 to imported_module_or_decl hook.
7513 (debug_nothing_tree_tree_tree_bool): Remove.
7514 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7515 * debug.c (do_nothing_debug_hooks): Use
7516 debug_nothing_tree_tree_tree_bool_bool instead of
7517 debug_nothing_tree_tree_tree_bool.
7518 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7519 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7520 * sdbout.c (sdb_debug_hooks): Likewise.
7521 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7522 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7524 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7525 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7526 attribute, don't add anything.
7528 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7530 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7531 (fold_build2_stat_loc): Likewise.
7532 (fold_build3_stat_loc): Likewise.
7533 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7534 (fold_build1_loc): Remove macro.
7535 (fold_build2_loc): Likewise.
7536 (fold_build3_loc): Likewise.
7538 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7540 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7541 (gimple_build_debug_bind_source_stat): Likewise.
7542 * gimple.h (gimple_build_debug_bind): Remove macro.
7543 (gimple_build_debug_bind_source): Likewise.
7545 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7547 * bitmap.c (bitmap_alloc): Adjust.
7548 (bitmap_gc_alloc): Likewise.
7549 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7551 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7553 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7554 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7555 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7556 (bitmap_gc_alloc_stat): Likewise.
7557 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7559 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7561 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7562 * rtl.h (shallow_copy_rtx): Remove macro.
7564 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7566 * emit-rtl.c (gen_raw_REG): Adjust.
7567 * gengenrtl.c (gendef): Likewise.
7568 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7569 * rtl.h (rtx_alloc): Remove macro.
7571 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7573 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7574 (build_tree_list_stat): Likewise.
7575 * tree.h (build_tree_list): Remove macro.
7576 (build_tree_list_vec): Likewise.
7578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7580 * tree.c (make_vector_stat): Remove _stat from name.
7581 (build_vector_stat): Likewise.
7582 * tree.h (make_vector_stat): Remove macro.
7583 (build_vector_stat): Likewise.
7585 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7587 * tree.h (build_var_debug_value): Remove prototype.
7589 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7591 * tree.c (tree_cons_stat): Remove _stat from name.
7592 * tree.h (tree_cons): Remove macro.
7594 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7596 * tree.c (build_vl_exp_stat): Remove _stat from name.
7597 * tree.h (build_vl_exp): Remove macro.
7599 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7601 * tree.c (build_decl_stat): Remove _stat from name.
7602 * tree.h (build_decl): Remove macro.
7604 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7606 * gimple.c (gimple_build_with_ops_stat): Adjust.
7607 (gimple_alloc_stat): Remove _stat from name.
7608 * gimple.h (gimple_alloc): Remove macro.
7610 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7612 * tree.c (make_tree_vec_stat): Remove _stat from name.
7613 (grow_tree_vec_stat): Likewise.
7614 * tree.h (make_tree_vec_stat): Adjust prototype.
7615 (grow_tree_vec_stat): Likewise.
7616 (make_tree_vec): Remove macro.
7617 (grow_tree_vec): Likewise.
7619 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7621 * fold-const.c (fold_build1_stat_loc): Adjust.
7622 (fold_build2_stat_loc): Likewise.
7623 (fold_build3_stat_loc): Likewise.
7624 * tree.c (build0_stat): Remove _stat from name.
7625 (build1_stat): Likewise.
7626 (build2_stat): Likewise.
7627 (build3_stat): Likewise.
7628 (build4_stat): Likewise.
7629 (build5_stat): Likewise.
7630 * tree.h (build1_loc): Remove macro, and rename _stat function
7632 (build2_loc): Likewise.
7633 (build3_loc): Likewise.
7634 (build4_loc): Likewise.
7635 (build5_loc): Likewise.
7637 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7639 * tree.c (make_int_cst_stat): Remove _stat from name.
7640 * tree.h (make_int_cst_stat): Adjust prototype.
7641 (make_int_cst): Remove macro.
7643 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7645 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7646 * tree.h (make_tree_binfo_stat): Adjust prototype.
7647 (make_tree_binfo): Remove.
7649 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7651 * tree.c (copy_node_stat): Rename to copy_node.
7652 (build_distinct_type_copy): Adjust.
7653 * tree.h (copy_node_stat): Adjust prototype.
7654 (copy_node): Remove macro.
7656 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7658 * tree.c (make_node_stat): rename to make_node.
7659 (build_tree_list_stat): Adjust.
7660 (build0_stat): Likewise.
7661 (build2_stat): Likewise.
7662 (build3_stat): Likewise.
7663 (build4_stat): Likewise.
7664 (build5_stat): Likewise.
7665 (build_decl_stat): Likewise.
7666 * tree.h (make_node_stat): Adjust prototype.
7667 (make_node): remove macro.
7669 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
7671 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
7672 (PPC_FEATURE2_SCV): Likewise.
7673 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
7675 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7677 * config/aarch64/aarch64.c
7678 (aarch64_internal_mov_immediate): Add new special pattern.
7679 * config/aarch64/aarch64.md (*movdi_aarch64):
7680 Add reg/32bit const mov case.
7682 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7683 Richard Sandiford <richard.sandiford@linaro.org>
7685 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7686 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7687 Add integer and movi cases.
7688 (movi-split-hf-df-sf split, fp16): New.
7689 (enabled): Added TARGET_FP_F16INST.
7690 * config/aarch64/iterators.md (GPF_HF): New.
7691 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7693 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7695 * config/aarch64/aarch64.c
7696 (aarch64_simd_container_mode): Add prototype.
7697 (aarch64_expand_mov_immediate): Add HI support.
7698 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7699 (aarch64_can_const_movi_rtx_p): New.
7700 (aarch64_preferred_reload_class):
7701 Remove restrictions of using FP registers for certain SIMD operations.
7702 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7703 (aarch64_valid_floating_const): Add integer move validation.
7704 (aarch64_simd_imm_scalar_p): Remove.
7705 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7706 (aarch64_legitimate_constant_p): Expand list of supported cases.
7707 * config/aarch64/aarch64-protos.h
7708 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7709 (aarch64_reinterpret_float_as_int): New.
7710 (aarch64_simd_imm_scalar_p): Remove.
7711 * config/aarch64/constraints.md (Uvi): New.
7712 (Dd): Split into Ds and new Dd.
7713 * config/aarch64/aarch64.md (*movsi_aarch64):
7715 (*movdi_aarch64): Add SIMD mov case.
7717 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7719 * tree-predcom.c: (struct chain): Handle store-store chain in which
7720 stores for elimination only store loop invariant values.
7721 (execute_pred_commoning_chain): Ditto.
7722 (prepare_initializers_chain_store_elim): Ditto.
7723 (prepare_finalizers): Ditto.
7724 (is_inv_store_elimination_chain): New function.
7725 (initialize_root_vars_store_elim_1): New function.
7727 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7729 * tree-predcom.c: Revise general description of the pass.
7730 (enum chain_type): New enum type for store elimination.
7731 (struct chain): New field supporting store elimination.
7732 (struct component): Ditto.
7733 (dump_chain): Dump store-stores chain.
7734 (release_chain): Release resources.
7735 (split_data_refs_to_components): Compute and create component
7736 contains only stores for elimination.
7737 (get_chain_last_ref_at): New function.
7738 (make_invariant_chain): Initialization.
7739 (make_rooted_chain): Specify chain type in parameter and record it.
7740 (add_looparound_copies): Skip for store-stores chain.
7741 (determine_roots_comp): Compute type of chain and pass it to
7743 (initialize_root_vars_store_elim_2): New function.
7744 (finalize_eliminated_stores): New function.
7745 (remove_stmt): Handle store for elimination.
7746 (execute_pred_commoning_chain): Execute predictive commoning on
7748 (determine_unroll_factor): Skip unroll for store-stores chain.
7749 (prepare_initializers_chain_store_elim): New function.
7750 (prepare_initializers_chain): Hanlde store-store chain.
7751 (prepare_finalizers_chain, prepare_finalizers): New function.
7752 (tree_predictive_commoning_loop): Return integer value indicating
7753 if loop is unrolled or lcssa form is corrupted.
7754 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7756 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7758 * tree-predcom.c (initialize_root): Delete.
7759 (execute_pred_commoning_chain): Initialize root vars and replace
7760 reference of non-combined chain directly, rather than call above
7763 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7765 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7766 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7768 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7770 * tree-predcom.c (struct chain): New field init_seq.
7771 (release_chain): Release init_seq.
7772 (prepare_initializers_chain): Record intialization stmts in above
7774 (insert_init_seqs): New function.
7775 (tree_predictive_commoning_loop): Call insert_init_seqs.
7777 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7779 * tree-predcom.c (determine_roots_comp): Skip trivial components.
7781 2017-07-28 Richard Biener <rguenther@suse.de>
7783 * match.pd: Remove superfluous :c.
7784 * genmatch.c (simplify::id): Add member.
7785 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
7787 (current_id): New global.
7788 (dt_node::parent): Move from ...
7789 (dt_operand::parent): ... here. Add for_id member.
7790 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
7791 (decision_tree::find_node): Relax order requirement when
7792 merging DT_TRUE nodes to ones inbetween the current simplify
7793 and the one we try to merge with. Add diagnostic whenever
7794 we need to enforce pattern order by not merging.
7795 (decision_tree::insert): Set current_id.
7796 (decision_tree::print_node): Dump parent node and for_id.
7797 (parser::last_id): Add member.
7798 (parser::push_simplify): Assign unique id.
7799 (parser::parser): Initialize last_id.
7801 2017-07-28 Martin Liska <mliska@suse.cz>
7804 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
7805 gimple_build_debug_bind.
7807 2017-07-28 Richard Biener <rguenther@suse.de>
7809 PR tree-optimization/81502
7810 * match.pd: Add pattern combining BIT_INSERT_EXPR with
7812 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
7814 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
7815 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
7816 for BIT_FIELD_REF args.
7817 * fold-const.c (make_bit_field_ref): Likewise.
7818 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7820 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7823 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
7824 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
7825 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
7826 Or it into SANITIZER_UNDEFINED.
7827 * ubsan.c: Include gimple-fold.h and varasm.h.
7828 (ubsan_expand_ptr_ifn): New function.
7829 (instrument_pointer_overflow): New function.
7830 (maybe_instrument_pointer_overflow): New function.
7831 (instrument_object_size): Formatting fix.
7832 (pass_ubsan::execute): Call instrument_pointer_overflow
7833 and maybe_instrument_pointer_overflow.
7834 * internal-fn.c (expand_UBSAN_PTR): New function.
7835 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
7836 * sanitizer.def (__ubsan_handle_pointer_overflow,
7837 __ubsan_handle_pointer_overflow_abort): New builtins.
7838 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
7839 * internal-fn.def (UBSAN_PTR): New internal function.
7840 * opts.c (sanitizer_opts): Add pointer-overflow.
7841 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
7842 * fold-const.c (build_range_check): Compute pointer range check in
7843 integral type if pointer arithmetics would be needed. Formatting
7846 2017-07-28 Martin Liska <mliska@suse.cz>
7849 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
7850 parameters that are of a variable-length.
7852 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
7854 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
7856 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
7857 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
7858 (CRT_CALL_STATIC_FUNCTION): Likewise.
7859 (ASM_DEFAULT_SPEC): New define.
7860 (ASM_SPEC32): Likewise.
7861 (ASM_SPEC64): Likewise.
7862 (ASM_SPEC_COMMON): Likewise.
7863 (ASM_SPEC): Likewise.
7864 (INVALID_64BIT): Likewise.
7865 (LINK_OS_DEFAULT_SPEC): Likewise.
7866 (LINK_OS_SPEC32): Likewise.
7867 (LINK_OS_SPEC64): Likewise.
7868 (POWERPC_LINUX): Likewise.
7869 (PTRDIFF_TYPE): Likewise.
7870 (RESTORE_FP_PREFIX): Likewise.
7871 (RESTORE_FP_SUFFIX): Likewise.
7872 (SAVE_FP_PREFIX): Likewise.
7873 (SAVE_FP_SUFFIX): Likewise.
7874 (SIZE_TYPE): Likewise.
7875 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
7876 (TARGET_64BIT): Likewise.
7877 (TARGET_64BIT): Likewise.
7878 (TARGET_AIX): Likewise.
7879 (WCHAR_TYPE_SIZE): Likewise.
7880 (WCHAR_TYPE): Undefine.
7881 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
7882 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
7883 (CPP_OS_RTEMS_SPEC): Delete.
7884 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
7885 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
7887 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
7889 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7891 PR tree-optimization/81578
7892 * tree-parloops.c (build_new_reduction): Bail out if
7893 reduction_code isn't one of the standard OpenMP reductions.
7894 Move the details printing after that decision.
7896 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7898 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
7899 related to reload_in_progress.
7900 (splat_input_operand): Likewise.
7901 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
7903 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
7905 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
7906 (TARGET_INSTANTIATE_DECLS): Likewise.
7907 (legitimate_indexed_address_p): Delete reload_in_progress code.
7908 (rs6000_debug_legitimate_address_p): Likewise.
7909 (rs6000_eliminate_indexed_memrefs): Likewise.
7910 (rs6000_emit_le_vsx_store): Likewise.
7911 (rs6000_emit_move_si_sf_subreg): Likewise.
7912 (rs6000_emit_move): Likewise.
7913 (register_to_reg_type): Likewise.
7914 (rs6000_pre_atomic_barrier): Likewise.
7915 (rs6000_machopic_legitimize_pic_address): Likewise.
7916 (rs6000_allocate_stack_temp): Likewise.
7917 (rs6000_address_for_fpconvert): Likewise.
7918 (rs6000_address_for_altivec): Likewise.
7919 (rs6000_secondary_memory_needed_rtx): Delete function.
7920 (rs6000_check_sdmode): Likewise.
7921 (rs6000_alloc_sdmode_stack_slot): Likewise.
7922 (rs6000_instantiate_decls): Likewise.
7923 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
7924 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
7925 Delete reload_in_progress.
7926 (*vec_reload_and_plus_<mptrsize>): Likewise.
7927 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
7928 (vsx_div_v2di): Likewise.
7929 (vsx_udiv_v2di): Likewise.
7931 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7933 * config/rs6000/rs6000.opt (mlra): Replace with stub.
7934 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
7935 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
7936 (rs6000_debug_reg_global): Delete print of LRA status.
7937 (rs6000_option_override_internal): Delete dead LRA related code.
7938 (rs6000_lra_p): Delete function.
7939 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
7941 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
7943 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
7944 * config/riscv/rtems.h: New file.
7946 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7947 Sudakshina Das <sudi.das@arm.com>
7949 * config/aarch64/aarch64.md
7950 (define_split for and<mode>3nr_compare): Move
7951 non aarch64_logical_operand to a register.
7952 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
7953 register immediate operand to a register.
7954 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
7956 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7959 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
7961 2017-07-27 Richard Biener <rguenther@suse.de>
7963 PR tree-optimization/81573
7964 PR tree-optimization/81494
7965 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
7966 multi defuse cycle case.
7968 2017-07-27 Richard Biener <rguenther@suse.de>
7970 PR tree-optimization/81571
7971 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
7974 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
7976 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
7977 earlier and only if MASK_FPU is set. Adjust formatting.
7979 2017-07-27 Martin Liska <mliska@suse.cz>
7981 * opt-functions.awk: Add validation of value of Init.
7982 * optc-gen.awk: Pass new argument.
7984 2017-07-27 Martin Liska <mliska@suse.cz>
7986 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
7987 Fix wrong condition.
7989 2017-07-27 Martin Liska <mliska@suse.cz>
7991 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
7992 BBs and edges seen by autoFDO.
7994 2017-07-27 Richard Biener <rguenther@suse.de>
7996 PR tree-optimization/81502
7997 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
7998 with incompatible but same sized type.
7999 (execute_update_addresses_taken): Likewise.
8001 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
8003 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
8004 flag_tree_loop_vectorize rather than flag_tree_vectorize.
8006 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8009 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
8010 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
8011 Change s_operand to memory_operand.
8013 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
8015 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
8016 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
8017 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
8018 Emit instructions rather than returning an expression. Handle TFmode
8019 and KFmode by casting to TImode.
8020 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
8021 (rs6000_emit_le_vsx_store): Likewise.
8022 * config/rs6000/vsx.md (VSX_TI): New iterator.
8023 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
8024 (*vsx_le_undo_permute_<mode>): Likewise.
8025 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
8026 emit the split sequence.
8027 (*vsx_le_perm_store_<mode>): Likewise.
8029 2017-07-27 Jakub Jelinek <jakub@redhat.com>
8031 PR tree-optimization/81555
8032 PR tree-optimization/81556
8033 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
8034 if true, force CHANGED for the recursive invocation.
8035 (reassociate_bb): Remember original length of ops array, pass
8036 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
8038 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
8039 attributes for noipa attribute. For naked attribute use
8040 lookup_attribute first before lookup_attribute_spec.
8041 * final.c (rest_of_handle_final): Disable IPA RA for functions with
8043 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
8044 for functions with noipa attribute.
8045 (cgraph_externally_visible_p): Return true for functions with noipa
8047 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
8048 for functions with noipa attribute.
8049 * doc/extend.texi: Document noipa function attribute.
8050 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
8051 also for functions with noipa attribute.
8052 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
8054 2017-07-26 Andrew Pinski <apinski@cavium.com>
8056 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
8057 vec_unalign_load_cost and vec_unalign_store_cost.
8059 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8061 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
8062 -mvsx-small-integer option.
8063 (ISA_3_0_MASKS_IEEE): Likewise.
8064 (OTHER_VSX_VECTOR_MASKS): Likewise.
8065 (POWERPC_MASKS): Likewise.
8066 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
8067 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
8068 code, only testing for DImode being allowed in non-VSX floating
8070 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
8071 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
8073 (rs6000_option_override_internal): Delete -mvsx-small-integer.
8074 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
8075 TARGET_P8_VECTOR test.
8076 (rs6000_secondary_reload_simple_move): Likewise.
8077 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
8078 since TARGET_P9_VECTOR was already tested.
8079 (rs6000_opt_masks): Remove -mvsx-small-integer.
8080 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
8081 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8083 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
8084 test for TARGET_VEXTRACTUB was used, and that uses
8086 (p9 extract splitter): Likewise.
8087 (vsx_extract_<mode>_di_p9): Likewise.
8088 (vsx_extract_<mode>_store_p9): Likewise.
8089 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
8090 for TARGET_P9_VECTOR was used. Delete code that is now dead with
8091 the elimination of TARGET_VSX_SMALL_INTEGER.
8092 (vsx_extract_<mode>_p8): Likewise.
8093 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
8094 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
8095 (vsx_set_<mode>_p9): Likewise.
8096 (vsx_set_v4sf_p9): Likewise.
8097 (vsx_set_v4sf_p9_zero): Likewise.
8098 (vsx_insert_extract_v4sf_p9): Likewise.
8099 (vsx_insert_extract_v4sf_p9_2): Likewise.
8100 * config/rs6000/rs6000.md (sign extend splitter): Change
8101 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
8102 (floatsi<mode>2_lfiwax_mem): Likewise.
8103 (floatunssi<mode>2_lfiwzx_mem): Likewise.
8104 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
8105 since a test for TARGET_P9_VECTOR was used.
8106 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8107 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
8108 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8109 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8110 TARGET_P8_VECTOR test.
8111 (fix_trunc<mode>si2_stfiwx): Likewise.
8112 (fix_trunc<mode>si2_internal): Likewise.
8113 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
8114 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8116 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8117 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8118 TARGET_P8_VECTOR test.
8119 (fixuns_trunc<mode>si2_stfiwx): Likewise.
8120 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
8121 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8123 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8124 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
8125 since a test for TARGET_P9_VECTOR was used.
8126 (splitter for loading small constants): Likewise.
8128 2017-07-26 Andrew Pinski <apinski@cavium.com>
8130 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
8133 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
8136 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
8137 (fp_valid_at): Likewise.
8139 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8141 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
8142 (qdf24xx_addrcost_table): Likewise.
8143 (cortexa57_tunings): Update to use generic_branch_cost.
8144 (cortexa72_tunings): Likewise.
8145 (cortexa73_tunings): Likewise.
8146 (qdf24xx_tunings): Likewise.
8148 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8150 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
8151 (thunderx2t99_branch_cost): Likewise.
8152 (cortexa35_tunings): Update to use generic_branch_cost.
8153 (cortexa53_tunings): Likewise.
8154 (cortexa57_tunings): Likewise.
8155 (cortexa72_tunings): Likewise.
8156 (cortexa73_tunings): Likewise.
8157 (thunderx2t99_tunings): Likewise.
8159 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8161 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
8162 (sparc_option_override): Honour MASK_FSMULD.
8163 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
8164 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
8165 * config/sparc/sparc.opt (mfsmuld): New option.
8166 * doc/invoke.texi (mfsmuld): Document option.
8168 2017-07-26 Marek Polacek <polacek@redhat.com>
8171 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
8173 2017-07-26 Richard Biener <rguenther@suse.de>
8175 * gimple-match-head.c (do_valueize): Return OP if valueize
8177 (get_def): New helper to get at the def stmt of a SSA name
8179 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
8180 do_valueize to get at the def stmt.
8181 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
8183 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
8186 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
8188 2017-07-26 Martin Liska <mliska@suse.cz>
8191 * function.c (expand_function_start): Make expansion of
8192 nonlocal_goto_save_area after parm_birth_insn.
8194 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8196 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
8197 from all CPU target flags enable members.
8199 2017-07-26 Richard Biener <rguenther@suse.de>
8201 * genmatch.c (dt_simplify::gen): Make iterator vars const.
8202 (decision_tree::gen): Make 'type' const.
8203 (write_predicate): Likewise.
8205 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8207 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
8208 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
8209 (rs6000_option_override_internal): Likewise.
8210 (rs6000_expand_vector_set): Likewise.
8211 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
8212 (TARGET_UPPER_REGS_SF): Likewise.
8213 (TARGET_UPPER_REGS_DI): Likewise.
8214 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
8215 (TARGET_DIRECT_MOVE_64BIT): Likewise.
8216 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
8217 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8218 (Splitters for DI constants in Altivec registers): Likewise.
8219 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
8220 (vsx_set_v4sf_p9): Likewise.
8221 (vsx_set_v4sf_p9_zero): Likewise.
8222 (vsx_insert_extract_v4sf_p9): Likewise.
8223 (vsx_insert_extract_v4sf_p9_2): Likewise.
8225 2017-07-25 Carl Love <cel@us.ibm.com>
8227 * doc/extend.texi: Update the built-in documentation file for the
8228 existing built-in functions
8229 vector signed char vec_cnttz (vector signed char);
8230 vector unsigned char vec_cnttz (vector unsigned char);
8231 vector signed short vec_cnttz (vector signed short);
8232 vector unsigned short vec_cnttz (vector unsigned short);
8233 vector signed int vec_cnttz (vector signed int);
8234 vector unsigned int vec_cnttz (vector unsigned int);
8235 vector signed long long vec_cnttz (vector signed long long);
8236 vector unsigned long long vec_cnttz (vector unsigned long long);
8238 2017-07-25 Andrew Pinski <apinski@cavium.com>
8240 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
8241 accesses where the use is for the first operand of a BIT_INSERT.
8243 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8246 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8247 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8249 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8251 * config/i386/gstabs.h: Delete.
8252 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
8254 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
8256 * config/i386/i386.c (ix86_decompose_address): Do not check for
8257 register RTX when looking at index_reg or base_reg.
8258 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
8260 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
8262 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
8263 to update EH info here.
8265 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8267 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
8269 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8271 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
8273 2017-07-25 Torsten Duwe <duwe@suse.de>
8275 * common.opt: Introduce -fpatchable-function-entry
8276 command line option, and its variables function_entry_patch_area_size
8277 and function_entry_patch_area_start.
8278 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
8279 including a two-value parser.
8280 * target.def (print_patchable_function_entry): New target hook.
8281 * targhooks.h (default_print_patchable_function_entry): New function.
8282 * targhooks.c (default_print_patchable_function_entry): Likewise.
8283 * toplev.c (process_options): Switch off IPA-RA if
8284 patchable function entries are being generated.
8285 * varasm.c (assemble_start_function): Look at the
8286 patchable-function-entry command line switch and current
8287 function attributes and maybe generate NOP instructions by
8288 calling the print_patchable_function_entry hook.
8289 * doc/extend.texi: Document patchable_function_entry attribute.
8290 * doc/invoke.texi: Document -fpatchable_function_entry
8291 command line option.
8292 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
8294 * doc/tm.texi: Re-generate.
8296 2017-07-25 Jakub Jelinek <jakub@redhat.com>
8299 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
8300 TARGET_AVX512DQ rather than TARGET_AVX512BW.
8302 2017-07-25 Tamar Christina <tamar.christina@arm.com>
8304 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
8306 2017-07-25 Richard Biener <rguenther@suse.de>
8308 PR tree-optimization/81455
8309 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
8310 not walk in cycles when looking for guards.
8312 2017-07-25 Richard Biener <rguenther@suse.de>
8314 PR tree-optimization/81529
8315 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
8316 when optimizing backedge uses.
8318 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
8320 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
8322 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
8323 to dl_section_ref. On AIX, append an expression to subtract
8324 the size of the section length to dl_section_ref.
8326 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
8328 * configure.ac: If any of the config.* scripts fail, exit 1.
8329 * configure: Regenerate.
8331 2017-07-25 Richard Biener <rguenther@suse.de>
8334 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
8335 of immediate uses, be more verbose on errors.
8337 2017-07-25 Richard Biener <rguenther@suse.de>
8339 PR tree-optimization/81510
8340 * tree-vect-loop.c (vect_is_simple_reduction): When the
8341 reduction stmt is not inside the loop bail out.
8343 2017-07-25 Richard Biener <rguenther@suse.de>
8345 PR tree-optimization/81303
8346 * tree-vect-loop-manip.c (vect_loop_versioning): Build
8347 profitability check against LOOP_VINFO_NITERSM1.
8349 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8351 * domwalk.c (cmp_bb_postorder): Simplify.
8352 (sort_bbs_postorder): New function. Use it...
8353 (dom_walker::walk): ...here to optimize common cases.
8355 2017-07-25 Martin Liska <mliska@suse.cz>
8358 * ipa-visibility.c (function_and_variable_visibility): Make the
8359 redirection just on target that supports aliasing.
8360 Fix GNU coding style.
8362 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
8365 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
8367 * config/arm/rtems.h: Likewise.
8368 * config/bfin/rtems.h: Likewise.
8369 * config/i386/rtemself.h: Likewise.
8370 * config/lm32/rtems.h: Likewise.
8371 * config/m32c/rtems.h: Likewise.
8372 * config/m68k/rtemself.h: Likewise.
8373 * config/microblaze/rtems.h: Likewise.
8374 * config/mips/rtems.h: Likewise.
8375 * config/moxie/rtems.h: Likewise.
8376 * config/nios2/rtems.h: Likewise.
8377 * config/powerpcspe/rtems.h: Likewise.
8378 * config/rs6000/rtems.h: Likewise.
8379 * config/rtems.h: Likewise.
8380 * config/sh/rtems.h: Likewise.
8381 * config/sh/rtemself.h: Likewise.
8382 * config/sparc/rtemself.h: Likewise.
8384 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8387 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
8388 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
8389 * tree-ssa-structalias.c (alias_get_name): Same.
8391 2017-07-25 Bin Cheng <bin.cheng@arm.com>
8394 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
8395 instructions if no du chain is found.
8397 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8399 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
8401 2017-07-25 Richard Biener <rguenther@suse.de>
8404 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
8407 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8409 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
8411 (ISA_2_7_MASKS_SERVER): Likewise.
8412 (ISA_3_0_MASKS_IEEE): Likewise.
8413 (OTHER_P8_VECTOR_MASKS): Likewise.
8414 (OTHER_VSX_VECTOR_MASKS): Likewise.
8415 (POWERPC_MASKS): Likewise.
8416 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
8417 duplicate list of options.
8418 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
8419 explicit -mupper-regs options.
8420 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
8421 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
8422 alias for -mupper-regs-df.
8423 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8424 (rs6000_init_hard_regno_mode_ok): Likewise.
8425 (rs6000_option_override_internal): Likewise.
8426 (rs6000_opt_masks): Likewise.
8427 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8428 options in terms of whether -mvsx or -mpower8-vector was used.
8429 (TARGET_UPPER_REGS_DI): Likewise.
8430 (TARGET_UPPER_REGS_SF): Likewise.
8431 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8432 -mupper-regs-* options.
8434 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8436 * passes.c (emergency_dump_function): Print some empty lines and a
8437 header before the RTL dump.
8439 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8441 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8443 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8446 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8447 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8449 2017-07-24 Carl Love <cel@us.ibm.com>
8451 * config/rs6000/rs6000-c.c: Add support for built-in functions
8452 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8453 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8454 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8455 vec_extract_fp_from_shortl): Add defines for the two builtins.
8456 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8457 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8459 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8460 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8461 * doc/extend.texi: Update the built-in documentation file for the
8462 new built-in function.
8464 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8467 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8469 * doc/generic.texi: Likewise.
8470 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8471 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8473 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8475 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8476 (aarch64_mls_elt_merge<mode>): Likewise.
8478 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8480 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8481 having __cxa_atexit.
8483 2017-07-23 Michael Collison <michael.collison@arm.com>
8485 * config/arm/arm.c (arm_option_override): Deprecate
8486 use of -mstructure-size-boundary.
8487 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8488 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8490 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8493 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8494 Reduce cost estimate for direct moves.
8496 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8499 * config/i386/i386.c (ix86_option_override_internal): Disable
8500 BMI, BMI2 and TBM instructions for -m16.
8502 2017-07-21 Carl Love <cel@us.ibm.com>
8504 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8505 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8506 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8507 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8508 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8509 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8510 VMULOSW): New enum "unspec" values.
8511 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8512 altivec_vmulosw): New patterns.
8513 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8514 VMULOSW): Add definitions.
8516 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8518 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8519 (qdf24xx): Likewise.
8520 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8521 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8522 (AARCH64_FL_V8_1): Renumber.
8523 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8524 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8525 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8526 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8527 rdma to feature modifiers list.
8529 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8532 * ipa-visibility (function_and_variable_visibility): Convert
8533 recursive PLT call to direct call if appropriate.
8535 2017-07-21 Andrew Pinski <apinski@cavium.com>
8537 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8538 operand 1 to see if the types precision matches.
8539 * fold-const.c (operand_equal_p): Likewise.
8541 2017-07-21 Richard Biener <rguenther@suse.de>
8543 PR tree-optimization/81303
8544 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8545 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8546 (vect_peeling_hash_get_lowest_cost): Adjust.
8547 (vect_enhance_data_refs_alignment): Likewise. Use
8548 vect_get_peeling_costs_all_drs to compute the penalty for no
8549 peeling to match up costs.
8551 2017-07-21 Richard Biener <rguenther@suse.de>
8553 PR tree-optimization/81500
8554 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8555 we didn't identify a reduction path.
8557 2017-07-21 Tom de Vries <tom@codesourcery.com>
8558 Cesar Philippidis <cesar@codesourcery.com>
8560 PR gcov-profile/81442
8561 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8564 2017-07-21 Tom de Vries <tom@codesourcery.com>
8567 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8569 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8570 nvptx_override_options_after_change.
8572 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8574 * dwarf2out.c (output_file_names): Avoid double testing for
8577 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8579 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8581 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8583 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8585 (try_crossjump_to_edge): Do not punt on partitioned functions.
8587 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8589 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8590 Put all BBs reachable only via paths crossing cold region to cold
8592 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8594 2016-07-21 Richard Biener <rguenther@suse.de>
8596 PR tree-optimization/81303
8597 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8598 into account prologue and epilogue iterations when raising
8599 min_profitable_iters to sth at least covering one vector iteration.
8601 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8603 * config/arm/arm.c (arm_test_cpu_arch_dat):
8606 2017-07-20 Nathan Sidwell <nathan@acm.org>
8608 Remove TYPE_METHODS.
8609 * tree.h (TYPE_METHODS): Delete.
8610 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8611 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8612 (dbxout_type_methods): Scan TYPE_FIELDS.
8613 (dbxout_type): Don't check TYPE_METHODS here.
8614 * function.c (use_register_for_decl): Always ignore register for
8615 class types when not optimizing.
8616 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8617 * tree.c (free_lang_data_in_type): Stitch out member functions and
8618 templates from TYPE_FIELDS.
8619 (build_distinct_type_copy, verify_type_variant,
8620 verify_type): Member fns are on TYPE_FIELDS.
8621 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8622 * tree-pretty-print.c (dump_generic_node): Likewise.
8624 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8627 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8628 V2TImode and V4TImode.
8629 (ix86_expand_vector_extract): Likewise.
8630 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8631 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8632 (ssescalarmode): Handle V4TImode and V2TImode.
8633 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8634 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8635 (VEXTRACTI128_MODE): New mode iterator.
8636 (splitter for *vec_extractv?ti first element): New.
8637 (VEC_INIT_MODE): New mode iterator.
8638 (vec_init<mode>): Consolidate 3 expanders into one using
8639 VEC_INIT_MODE mode iterator.
8641 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8643 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8644 non_spilled_static_chain_regno_p.
8646 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8648 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8650 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8652 * bb-reorder.c (connect_traces): Allow copying of blocks within
8655 2017-07-20 Richard Biener <rguenther@suse.de>
8657 * gimple.h (gimple_phi_result): Add gphi * overload.
8658 (gimple_phi_result_ptr): Likewise.
8659 (gimple_phi_arg): Likewise. Adjust index assert to only
8660 allow actual argument accesses rather than all slots available
8662 (gimple_phi_arg_def): Add gphi * overload.
8663 * tree-phinodes.c (make_phi_node): Initialize only actual
8665 (resize_phi_node): Clear memory not covered by old node,
8666 do not initialize excess argument slots.
8667 (reserve_phi_args_for_new_edge): Initialize new argument slot
8670 2017-07-20 Bin Cheng <bin.cheng@arm.com>
8672 PR tree-optimization/81388
8674 2016-07-21 Bin Cheng <bin.cheng@arm.com>
8676 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
8677 by removing computation of may_be_zero.
8679 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8680 Tom de Vries <tom@codesourcery.com>
8683 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8684 when gimple level profile disagrees with what RTL expander did.
8686 2017-07-20 Richard Biener <rguenther@suse.de>
8688 PR tree-optimization/61171
8689 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8690 (vect_analyze_stmt): Add slp instance parameter.
8691 (vectorizable_reduction): Likewise.
8692 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8693 (vect_is_simple_reduction): Deal with chains not detected
8694 as SLP reduction chain, specifically not properly associated
8695 chains containing a mix of plus/minus.
8696 (get_reduction_op): Remove.
8697 (get_initial_defs_for_reduction): Simplify, pass in whether
8698 this is a reduction chain, pass in the SLP node for the PHIs.
8699 (vect_create_epilog_for_reduction): Get the SLP instance as
8701 (vectorizable_reduction): Get the SLP instance as arg.
8702 During analysis remember the SLP node with the PHIs in the
8703 instance. Simplify getting at the vectorized reduction PHIs.
8704 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8705 through SLP instance.
8706 (vect_slp_analyze_operations): Likewise.
8707 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8708 (vect_transform_stmt): Likewise.
8710 2017-07-20 Tom de Vries <tom@codesourcery.com>
8712 PR tree-optimization/81489
8713 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8714 read of phi arg location to before loop that modifies phi.
8716 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8718 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8721 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8724 * except.c (execute): Fix ordering issue.
8726 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8728 PR rtl-optimization/81423
8729 * combine.c (make_compound_operation_int): Don't try to optimize
8730 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8732 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8734 PR rtl-optimization/81423
8735 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8736 with a constant that is -1 in the truncated to mode.
8738 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8740 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8741 (determine_unlikely_bbs): ... here.
8742 * predict.h (propagate_unlikely_bbs_forward): Declare.
8743 * cfgexpand.c (pass_expand::execute): Use it.
8744 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8746 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8747 propagate_unlikely_bbs_forward.
8749 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8752 * except.c (maybe_add_nop_after_section_switch): New function.
8755 2017-07-19 Tom de Vries <tom@codesourcery.com>
8757 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8759 2017-07-19 Tom de Vries <tom@codesourcery.com>
8761 * gimple.h (gimple_phi_arg): Make assert more strict.
8763 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8765 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8766 * config/rs6000/mmintrin.h: New file.
8767 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8769 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8771 PR tree-optimization/81346
8772 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
8774 2017-07-19 Tom de Vries <tom@codesourcery.com>
8776 * config/nvptx/nvptx.md (VECIM): Add V2DI.
8778 2017-07-19 Tom de Vries <tom@codesourcery.com>
8780 * config/nvptx/nvptx-modes.def: Add V2DImode.
8781 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
8782 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
8783 (nvptx_output_mov_insn): Handle lack of mov.b128.
8784 (nvptx_print_operand): Handle 'H' and 'L' codes.
8785 (nvptx_vector_mode_supported): Allow V2DImode.
8786 (nvptx_preferred_simd_mode): New function.
8787 (nvptx_data_alignment): New function.
8788 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
8789 nvptx_preferred_simd_mode.
8790 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
8792 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
8794 2017-07-19 Tom de Vries <tom@codesourcery.com>
8796 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
8797 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
8798 (nvptx_vector_mode_supported): New function. Allow V2SImode.
8799 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
8800 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
8801 (mov<VECIM>_insn): New define_insn.
8802 (define_expand "mov<VECIM>): New define_expand.
8804 2017-07-19 Tom de Vries <tom@codesourcery.com>
8806 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
8808 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8810 PR tree-optimization/81346
8811 * fold-const.h (fold_div_compare, range_check_type): Declare.
8812 * fold-const.c (range_check_type): New function.
8813 (build_range_check): Use range_check_type.
8814 (fold_div_compare): No longer static, rewritten into
8815 a match.pd helper function.
8816 (fold_comparison): Don't call fold_div_compare here.
8817 * match.pd (X / C1 op C2): New optimization using fold_div_compare
8820 2017-07-19 Nathan Sidwell <nathan@acm.org>
8822 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
8823 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
8824 * tree.c (find_decls_types_r, verify_type): Use
8825 TYPE_{MIN,MAX}_VALUE_RAW.
8826 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
8827 (hash_tree): Likewise.
8828 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8830 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8833 2017-07-18 Tom de Vries <tom@codesourcery.com>
8836 * omp-expand.c (expand_omp_for_static_chunk): Handle
8837 equal-argument loop exit phi.
8839 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
8842 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
8843 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
8844 operand 2 predicate.
8845 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
8846 operand 2 predicate.
8847 (ror,rol -> rorx splitters): Use const_int_operand as
8848 operand 2 predicate.
8850 2017-06-18 Richard Biener <rguenther@suse.de>
8852 PR tree-optimization/81410
8853 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
8854 the gap in the ! slp_perm SLP case after each group.
8856 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8859 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
8862 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8865 * predict.c (set_even_probabilities): Cleanup; do not affect
8866 probabilities that are already known.
8867 (combine_predictions_for_bb): Call even when count is set.
8869 2017-07-18 Nathan Sidwell <nathan@acm.org>
8871 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
8874 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8877 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
8878 optimization for loop niter analysis.
8880 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
8883 * config/avr/avr.c (avr_optimize_casesi): Don't use
8884 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
8886 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8888 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
8889 body_cost_vec from _vect_peel_extended_info.
8890 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
8891 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
8894 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8896 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
8898 2017-07-18 Richard Biener <rguenther@suse.de>
8900 PR tree-optimization/80620
8901 PR tree-optimization/81403
8902 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
8903 info when re-using a VN table entry.
8905 2017-07-18 Richard Biener <rguenther@suse.de>
8907 PR tree-optimization/81418
8908 * tree-vect-loop.c (vectorizable_reduction): Properly compute
8909 vectype_in. Verify that with lane-reducing reduction operations
8910 we have a single def-use cycle.
8912 2017-07-17 Carl Love <cel@us.ibm.com>
8914 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
8916 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8917 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8918 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8919 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8920 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8921 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8922 VMULOSW): New enum "unspec" values.
8923 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8924 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8925 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8926 altivec_vmulosw): New patterns.
8927 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8928 VMULOSW): Add definitions.
8929 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
8931 * config/alpha/alpha.c: Include predict.h.
8933 2017-07-17 Yury Gribov <tetra2005@gmail.com>
8935 * tree-vrp.c (compare_assert_loc): Fix comparison function
8936 to return predictable results.
8938 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8940 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
8943 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
8944 * doc/invoke.texi (mexpand-adddi): Update text.
8946 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8948 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
8949 that also clobbers the CC register. The old expand code is moved
8951 (*arc_clzsi2): ... here.
8952 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
8953 the CC register. The old expand code is moved to ...
8954 (arc_ctzsi2): ... here.
8956 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8958 * config/arc/arc.opt (mindexed-loads): Use initial value
8959 TARGET_INDEXED_LOADS_DEFAULT.
8960 (mauto-modify-reg): Use initial value
8961 TARGET_AUTO_MODIFY_REG_DEFAULT.
8962 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8963 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8964 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8965 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8967 2017-07-17 Martin Liska <mliska@suse.cz>
8970 * opts.c (finish_options): Do not allow -fgnu-tm
8971 w/ -fsanitize={kernel-,}address. Say sorry.
8973 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8976 * tree-loop-distribution.c (classify_partition): Only assert on
8977 numer of iterations.
8978 (merge_dep_scc_partitions): Delete prameter. Update function call.
8979 (distribute_loop): Remove code handling loop with unknown niters.
8980 (pass_loop_distribution::execute): Skip loop with unknown niters.
8982 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8985 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
8986 function sort_partitions_by_post_order.
8988 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8990 PR tree-optimization/81374
8991 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
8992 the max index of basic blocks, rather than number of basic blocks.
8994 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8996 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
8998 (arc_legitimate_pic_operand_p): Likewise.
8999 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
9001 (arc_needs_pcl_p): Likewise.
9002 (arc_legitimate_pc_offset_p): Likewise.
9003 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
9004 function is also used in constrains.md.
9005 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
9006 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
9007 PLUS. Only return true/false in known cases, otherwise assert.
9008 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
9009 is already called in arc_legitimate_constant_p.
9010 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
9012 (LEGITIMATE_PIC_OPERAND_P): Use
9013 arc_raw_symbolic_reference_mentioned_p function.
9014 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
9019 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9020 Andrew Burgess <andrew.burgess@embecosm.com>
9022 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
9023 (arc_return_address_register): New function.
9024 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
9025 (arc_handle_fndecl_attribute): Add naked attribute.
9026 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9027 (TARGET_WARN_FUNC_RETURN): Likewise.
9028 (arc_allocate_stack_slots_for_args): New function.
9029 (arc_warn_func_return): Likewise.
9030 (machine_function): Change type fn_type.
9031 (arc_compute_function_type): Consider new naked function type,
9032 change function return type.
9033 (arc_must_save_register): Adapt to handle new
9034 arc_compute_function_type's return type.
9035 (arc_expand_prologue): Likewise.
9036 (arc_expand_epilogue): Likewise.
9037 (arc_return_address_regs): Delete.
9038 (arc_return_address_register): New function.
9039 (arc_epilogue_uses): Use above function.
9040 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
9041 (arc_function_type): Change encoding, add naked type.
9042 (ARC_INTERRUPT_P): Change to handle the new encoding.
9043 (ARC_FAST_INTERRUPT_P): Likewise.
9044 (ARC_NORMAL_P): Define.
9045 (ARC_NAKED_P): Likewise.
9046 (arc_compute_function_type): Delete prototype.
9047 * config/arc/arc.md (in_ret_delay_slot): Use
9048 arc_return_address_register function.
9049 (simple_return): Likewise.
9050 (p_return_i): Likewise.
9052 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9054 PR tree-optimization/81428
9055 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
9056 can't be built for those types.
9058 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9060 Remove stuff dead since r239246.
9062 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
9063 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
9064 (avr_inform_devices): Remove dead stuff.
9066 2017-07-17 Tamar Christina <tamar.christina@arm.com>
9068 * config/arm/arm_neon.h: Fix softp typo.
9070 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9072 PR tree-optimization/81365
9073 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
9074 aggregate moves onto bb predecessor edges, make sure there are no
9075 loads that could alias the lhs in between the start of bb and the
9078 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9081 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
9082 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
9083 [LSHIFTRT, outer_code = TRUNCATE]: Same.
9085 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9087 PR tree-optimization/81396
9088 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
9089 (init_symbolic_number): Initialize it to 1.
9090 (perform_symbolic_merge): Add n_ops from both operands into the new
9092 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
9093 without base_addr as useless if they need more than one operation.
9094 (bswap_replace): Handle !bswap case for NULL base_addr.
9096 2017-07-17 Tom de Vries <tom@codesourcery.com>
9099 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
9102 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
9104 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
9105 conditional builtin define __FIX_LEON3FT_B2BST.
9107 2017-07-17 Daniel Cederman <cederman@gaisler.com>
9109 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
9110 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
9113 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9115 PR rtl-optimization/81424
9116 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
9117 to remove potential trapping from operands if -fnon-call-exceptions.
9119 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9121 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
9122 profile_proability for scalling.
9123 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
9125 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9127 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
9129 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9131 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
9132 fixpoint arithmetics.
9134 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9136 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
9137 fixpoint arithmetics.
9139 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9141 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
9142 fixpoint arithmetics.
9144 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9146 * profile-count.h (profile_probability::from_reg_br_prob_note,
9147 profile_probability::to_reg_br_prob_note): New functions.
9148 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
9149 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
9150 * predict.c (probability_reliable_p): Update.
9151 (edge_probability_reliable_p): Update.
9152 (br_prob_note_reliable_p): Update.
9153 (invert_br_probabilities): Update.
9154 (add_reg_br_prob_note): New function.
9155 (combine_predictions_for_insn): Update.
9156 * asan.c (asan_clear_shadow): Update.
9157 * cfgbuild.c (compute_outgoing_frequencies): Update.
9158 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
9159 (update_br_prob_note): Update.
9160 (rtl_verify_edges): Update.
9161 (purge_dead_edges): Update.
9162 (fixup_reorder_chain): Update.
9163 * emit-rtl.c (try_split): Update.
9164 * ifcvt.c (cond_exec_process_insns): Update.
9165 (cond_exec_process_if_block): Update.
9166 (dead_or_predicable): Update.
9167 * internal-fn.c (expand_addsub_overflow): Update.
9168 (expand_neg_overflow): Update.
9169 (expand_mul_overflow): Update.
9170 * loop-doloop.c (doloop_modify): Update.
9171 * loop-unroll.c (compare_and_jump_seq): Update.
9172 * optabs.c (emit_cmp_and_jump_insn_1): Update.
9173 * predict.h: Update.
9174 * reorg.c (mostly_true_jump): Update.
9176 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
9177 * config/alpha/alpha.c (emit_unlikely_jump): Update.
9178 * config/arc/arc.c: (emit_unlikely_jump): Update.
9179 * config/arm/arm.c: (emit_unlikely_jump): Update.
9180 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
9181 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
9182 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
9183 (ix86_print_operand): Update.
9184 (ix86_split_fp_branch): Update.
9185 (predict_jump): Update.
9186 * config/ia64/ia64.c (ia64_print_operand): Update.
9187 * config/mmix/mmix.c (mmix_print_operand): Update.
9188 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
9189 (rs6000_expand_split_stack_prologue): Update.
9190 * config/rs6000/rs6000.c: Update.
9191 * config/s390/s390.c (s390_expand_vec_strlen): Update.
9192 (s390_expand_vec_movstr): Update.
9193 (s390_expand_cs_tdsi): Update.
9194 (s390_expand_split_stack_prologue): Update.
9195 * config/sh/sh.c (sh_print_operand): Update.
9196 (expand_cbranchsi4): Update.
9197 (expand_cbranchdi4): Update.
9198 * config/sparc/sparc.c (output_v9branch): Update.
9199 * config/spu/spu.c (get_branch_target): Update.
9200 (ea_load_store_inline): Update.
9201 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
9202 * config/tilepro/tilepro.c: Update.
9204 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9206 * gimplify.c (mostly_copy_tree_r): Revert latest change.
9207 (gimplify_save_expr): Likewise.
9209 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9211 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
9213 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9215 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
9217 * timevar.def (TV_IPA_FNSUMMARY): Define.
9219 2017-07-16 Daniel Cederman <cederman@gaisler.com>
9221 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
9222 to back store errata sensitive sequence from being generated.
9223 (sqrtdf2_fix): Likewise.
9225 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9227 * tree-ssa-threadupdate.c (compute_path_counts,
9228 update_joiner_offpath_counts): Use profile_probability.
9230 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
9233 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9235 * config/arm/arm-c.c (arm_cpu_builtins): Define
9236 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9238 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9240 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9241 array entries to represent __ieee128 versions of the
9242 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
9243 scalar_extract_sig, and scalar_insert_exp built-in functions.
9244 (altivec_resolve_overloaded_builtin): Add special case handling
9245 for the __builtin_scalar_insert_exp function, as represented by
9246 the P9V_BUILTIN_VEC_VSIEDP constant.
9247 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
9248 exponent support for __ieee128 argument.
9249 (VSESQP): Add scalar extract signature support for __ieee128
9251 (VSTDCNQP): Add scalar test negative support for __ieee128
9253 (VSIEQP): Add scalar insert exponent support for __int128 argument
9254 with __ieee128 result.
9255 (VSIEQPF): Add scalar insert exponent support for __ieee128
9256 argument with __ieee128 result.
9257 (VSTDCQP): Add scalar test data class support for __ieee128
9259 (VSTDCNQP): Add overload support for scalar test negative with
9261 (VSTDCQP): Add overload support for scalar test data class
9263 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
9265 (UNSPEC_VSX_SIEXPQP): New constant.
9266 (xsxexpqp): New insn for VSX scalar extract exponent quad
9268 (xsxsigqp): New insn for VSX scalar extract significand quad
9270 (xsiexpqpf): New insn for VSX scalar insert exponent quad
9271 precision with floating point argument.
9272 (xststdcqp): New expand for VSX scalar test data class quad
9274 (xststdcnegqp): New expand for VSX scalar test negative quad
9276 (xststdcqp): New insn to match expansions for VSX scalar test data
9277 class quad precision and VSX scalar test negative quad precision.
9278 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
9279 special case operand checking to enforce that second operand of
9280 VSX scalar test data class with quad precision argument is a 7-bit
9282 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9283 prototypes and descriptions of __ieee128 versions of
9284 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
9285 scalar_test_data_class, and scalar_test_neg built-in functions.
9287 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9289 PR tree-optimization/81162
9290 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
9291 replace a negate with an add.
9293 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
9295 * doc/invoke.texi (arm/-mcpu): Document +crypto.
9297 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9299 * config/arm/arm-c.c (arm_cpu_builtins): Define
9300 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9302 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9304 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
9305 (armv8-r): Set ARM Cortex-R52 as default CPU.
9306 * config/arm/arm-tables.opt: Regenerate.
9307 * config/arm/arm-tune.md: Regenerate.
9308 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9310 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
9311 extension for -mcpu=cortex-r52.
9313 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9315 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
9316 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
9317 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
9318 (fp-armv8): Define it as FP_ARMv8 only.
9319 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
9320 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
9322 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
9323 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
9324 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
9325 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
9326 than TARGET_FPU_ARMV8.
9327 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
9328 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
9329 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
9331 * config/arm/neon.md (neon_vrint): Likewise.
9332 (neon_vcvt): Likewise.
9333 (neon_<fmaxmin_op><mode>): Likewise.
9334 (<fmaxmin><mode>3): Likewise.
9335 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
9336 * config/arm/predicates.md (arm_cond_move_operator): Check against
9337 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
9339 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
9341 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
9344 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9346 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
9347 loop in comment with memset.
9349 2017-07-14 Martin Liska <mliska@suse.cz>
9351 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
9352 * dwarf2out.c (is_java): Remove the function.
9353 (output_pubname): Remove usage of the function.
9354 (lower_bound_default): Remove usage of DW_LANG_Java.
9355 (gen_compile_unit_die): Likewise.
9356 * gcc.c: Remove compiler defaults for .java and .zip files.
9357 * gimple-expr.c (remove_suffix): Change as there's no longer
9358 extension than 4-letter one.
9359 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
9360 (gimplify_save_expr): Likewise.
9361 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
9362 as it's possible even for other languages than Java.
9363 * langhooks.h (struct lang_hooks): Remove Java from a comment.
9364 * lto-opts.c (lto_write_options): Remove reference to Java.
9365 * opts.c (strip_off_ending): Update file extension handling.
9366 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
9367 * tree-eh.c (lower_resx): Likewise.
9368 * tree.c (free_lang_data_in_type): Remove dead code.
9369 (find_decls_types_r): Likewise.
9370 (build_common_builtin_nodes): Remove Java from a comment.
9371 (verify_type): Remove dead code.
9372 * varasm.c (assemble_external): Remove Java from a comment.
9374 2017-07-14 Martin Liska <mliska@suse.cz>
9376 * opts.c (finish_options): Add quotes.
9377 (common_handle_option): Likewise.
9379 2017-07-14 Martin Liska <mliska@suse.cz>
9381 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
9382 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
9384 * dwarf2out.c (lower_bound_default): Do not handle
9386 (gen_compile_unit_die): Likewise.
9387 * gcc.c: Remove default extension binding for GNU Pascal.
9388 * stmt.c: Remove Pascal language from a comment.
9389 * xcoffout.c: Likewise.
9391 2017-07-13 David Malcolm <dmalcolm@redhat.com>
9394 * diagnostic-show-locus.c (fixit_cmp): New function.
9395 (layout::layout): Sort m_fixit_hints.
9396 (column_range::column_range): Assert that the values are valid.
9397 (struct char_span): New struct.
9398 (correction::overwrite): New method.
9399 (struct source_line): New struct.
9400 (line_corrections::add_hint): Add assertions. Reimplement memcpy
9401 calls in terms of classes source_line and char_span, and
9402 correction::overwrite.
9403 (selftest::test_overlapped_fixit_printing_2): New function.
9404 (selftest::diagnostic_show_locus_c_tests): Call it.
9406 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9408 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
9409 early if there is no lhs.
9411 2017-07-13 Martin Liska <mliska@suse.cz>
9413 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
9414 (gen_reference_type_die): Likewise.
9415 * stor-layout.c: Remove Pascal-related comment.
9417 2017-07-13 Martin Liska <mliska@suse.cz>
9419 * opts.c (finish_options): Add quotes to error messages.
9420 (parse_sanitizer_options): Likewise.
9422 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
9424 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
9426 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9428 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9430 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9432 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9434 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9436 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9439 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9440 provides the hardware capability bits, define the macro
9441 __BUILTIN_CPU_SUPPORTS__.
9442 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9443 if GLIBC does not provide the hardware capability bits. Add a
9444 gcc_unreachable call if the built-in cpu function is neither
9445 __builtin_cpu_is nor __builtin_cpu_supports.
9446 (rs6000_get_function_versions_dispatcher): Change the warning
9447 that an old GLIBC is used which does not export the capability
9448 bits to be an error.
9449 * doc/extend.texi (target_clones attribute): Document the
9450 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9451 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9452 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9453 the macros defined by GCC if the newer GLIBC is available.
9455 2017-07-12 Jeff Law <law@redhat.com>
9457 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9458 remaining includes slightly.
9459 * config/riscv/riscv-builtins.c: Include profile-count.h.
9461 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9464 * config/avr/avr.c (avr_set_current_function): In diagnostic
9465 messages: Quote keywords and (parts of) identifiers.
9466 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9469 2017-07-12 Carl Love <cel@us.ibm.com>
9471 * config/rs6000/rs6000-c.c: Add support for built-in functions
9472 vector bool char vec_revb (vector bool char);
9473 vector bool short vec_revb (vector short char);
9474 vector bool int vec_revb (vector bool int);
9475 vector bool long long vec_revb (vector bool long long);
9476 * doc/extend.texi: Update the built-in documentation file for the
9477 new built-in functions.
9479 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9481 * config/s390/s390.md: Remove movcc splitter.
9483 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9485 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9486 load/store on condition.
9488 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9491 * config/avr/avr.c (avr_encode_section_info)
9492 [progmem && !TREE_READONLY]: Error if progmem object needs
9495 2017-07-11 Michael Collison <michael.collison@arm.com>
9497 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9500 2017-07-11 Carl Love <cel@us.ibm.com>
9502 * config/rs6000/rs6000-c.c: Add support for builtins
9503 vector unsigned int vec_parity_lsbb (vector signed int);
9504 vector unsigned int vec_parity_lsbb (vector unsigned int);
9505 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9506 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9507 vector unsigned long long vec_parity_lsbb (vector signed long long);
9508 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9509 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9510 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9511 * doc/extend.texi: Update the built-in documentation file for the
9512 new built-in functions.
9514 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9516 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9517 (layout::m_primary_loc): New field.
9518 (layout::layout): Initialize new field. Move location filtering
9519 logic from here to...
9520 (layout::maybe_add_location_range): ...this new method. Add
9521 support for filtering to just the lines already specified by other
9523 (layout::will_show_line_p): New method.
9524 (gcc_rich_location::add_location_if_nearby): New method.
9525 (selftest::test_add_location_if_nearby): New test function.
9526 (selftest::diagnostic_show_locus_c_tests): Call it.
9527 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9530 2017-07-11 Tom de Vries <tom@codesourcery.com>
9532 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9533 (bb_first_real_insn): New function.
9534 (nvptx_single): Add extra initialization of broadcasted condition
9537 2017-07-11 Nathan Sidwell <nathan@acm.org>
9539 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9541 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9543 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9544 Binutils doc as TEXI will mess them up.
9545 * doc/invoke.texi (AVR Options): Same here.
9547 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9549 * config/sparc/sparc.opt (mfix-ut700): New option.
9550 (mfix-gr712rc): Likewise.
9551 (sparc_fix_b2bst): New variable.
9552 * doc/invoke.texi (SPARC options): Document them.
9553 (ARM options): Fix warnings.
9554 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9555 instructions to prevent sequences that can trigger the store-store
9556 errata for certain LEON3FT processors.
9557 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9558 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9559 * config/sparc/sparc.md (fix_b2bst): New attribute.
9560 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9562 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9565 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9567 (*rsqrtsf2_sse): Ditto.
9569 (div<mode>3): Macroize insn from divdf3 and divsf3
9570 using MODEF mode iterator.
9572 2017-07-10 Martin Sebor <msebor@redhat.com>
9574 PR tree-optimization/80397
9575 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9576 instead of testing for equality to INTEGER_TYPE.
9578 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9580 * config.gcc: Remove uclibc from arc target spec.
9582 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9584 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9586 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9589 * lto-wrapper.c (remove_option): New function.
9590 (merge_and_complain): Merge PIC/PIE options more realistically.
9592 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9594 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9598 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9599 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9600 * config.in: Regenerate.
9601 * configure: Regenerate.
9602 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9603 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9604 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9605 (TARGET_GASISR_PROLOGUES): ...target mask.
9606 * common/config/avr/avr-common.c
9607 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9608 Set -mgas-isr-prologues.
9609 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9610 INSERT_PASS_BEFORE for it.
9611 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9612 * config/avr/avr.c (avr_option_override)
9613 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9614 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9615 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9616 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9617 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9619 (make_avr_pass_pre_proep): New function.
9620 (emit_push_sfr) <treg>: Add argument to function and use it
9622 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9623 and set machine->gasisr.yes.
9624 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9625 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9626 __gcc_isr.n_pushed to .L__stack_usage.
9627 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9628 (avr_asm_final_postscan_insn): ...this new static function.
9629 * config/avr/avr.h (machine_function)
9630 <is_no_gccisr, use_L__stack_usage>: New fields.
9631 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9632 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9633 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9634 (gasisr, *gasisr): New expander and insn.
9635 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9636 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9637 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9639 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9641 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9644 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9646 Move jump-tables out of .text again.
9649 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9650 (ASM_OUTPUT_ADDR_VEC): New function.
9651 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9652 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9653 INSN_ADDRESSes as asm comment.
9654 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9655 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9656 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9657 * config/avr/avr.md (*tablejump): Adjust comment.
9658 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9659 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9661 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9662 (avr_output_addr_vec): New proto.
9663 (avr_log_t) <insn_addresses>: New field.
9665 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9668 * config/i386/i386.c (ix86_function_arg_advance): Set
9669 outgoing_args_on_stack to true if there are outgoing arguments
9671 (ix86_function_arg): Likewise.
9672 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
9673 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
9674 * config/i386/i386.h (machine_function): Add
9675 outgoing_args_on_stack.
9677 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
9679 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9681 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
9683 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9685 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9686 (REAL_H): Remove $(MACHMODE_H).
9687 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9689 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9690 $(MACHMODE_H) and double-int.h.
9691 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9693 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9696 2017-07-07 Andrew Pinski <apinski@cavium.com>
9698 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9699 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9701 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9703 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9704 Add warning if GCC was not configured to link against a GLIBC that
9705 exports the hardware capability bits.
9706 (make_resolver_func): Make resolver function private and not a
9707 COMDAT function. Create the name with clone_function_name instead
9708 of make_unique_name.
9711 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9712 correct operand in doing the split.
9714 2017-07-07 Carl Love <cel@us.ibm.com>
9716 * config/rs6000/rs6000-c: Add support for built-in function
9717 vector unsigned short vec_pack_to_short_fp32 (vector float,
9719 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9720 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9721 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9722 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9723 (convert_4f32_8i16): Add define_expand.
9724 * doc/extend.texi: Update the built-in documentation file for the
9725 new built-in function.
9727 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9729 * config/sparc/m8.md: New file.
9730 * config/sparc/sparc.md: Include m8.md.
9732 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9734 * config/sparc/sparc.opt: New option -mvis4b.
9735 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9736 (sparc_option_override): Handle VIS4B.
9737 (enum sparc_builtins): Define
9738 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9739 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9740 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9741 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9742 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9743 (check_constant_argument): New function.
9744 (sparc_vis_init_builtins): Define builtins
9745 __builtin_vis_dictunpack{8,16,32},
9746 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9747 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9748 __builtin_vis_fpcmpde{8,16,32}shl and
9749 __builtin_vis_fpcmpur{8,16,32}shl.
9750 (sparc_expand_builtin): Check that the constant operands to
9751 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9752 constant and in range.
9753 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9755 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9756 (SPARC_IMM5_P): Likewise.
9757 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9758 (enabled): Handle vis4b.
9759 (UNSPEC_DICTUNPACK): New unspec.
9760 (UNSPEC_FPCMPSHL): Likewise.
9761 (UNSPEC_FPUCMPSHL): Likewise.
9762 (UNSPEC_FPCMPDESHL): Likewise.
9763 (UNSPEC_FPCMPURSHL): Likewise.
9764 (cpu_feature): New CPU feature `vis4b'.
9765 (dictunpack{8,16,32}): New insns.
9766 (FPCSMODE): New mode iterator.
9767 (fpcscond): New code iterator.
9768 (fpcsucond): Likewise.
9769 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9770 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
9771 (fpcmpde{8,16,32}{si,di}shl): Likewise.
9772 (fpcmpur{8,16,32}{si,di}shl): Likewise.
9773 * config/sparc/constraints.md: Define constraints `q' for unsigned
9774 2-bit integer constants and `t' for unsigned 5-bit integer
9776 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
9778 (imm5_operand_dictunpack16): Likewise.
9779 (imm5_operand_dictunpack32): Likewise.
9780 (imm2_operand): Likewise.
9781 * doc/invoke.texi (SPARC Options): Document -mvis4b.
9782 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
9783 ditunpack* and fpcmp*shl builtins.
9785 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9787 * config.gcc: Handle m8 in --with-{cpu,tune} options.
9788 * config.in: Add HAVE_AS_SPARC6 define.
9789 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
9791 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
9793 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
9794 (CPP_CPU_SPEC): Handle m8.
9795 (ASM_CPU_SPEC): Likewise.
9796 * config/sparc/sparc-opts.h (enum processor_type): Add
9798 * config/sparc/sparc.c (m8_costs): New struct.
9799 (sparc_option_override): Handle TARGET_CPU_m8.
9800 (sparc32_initialize_trampoline): Likewise.
9801 (sparc64_initialize_trampoline): Likewise.
9802 (sparc_issue_rate): Likewise.
9803 (sparc_register_move_cost): Likewise.
9804 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
9805 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
9806 (ASM_CPU64_DEFAULT_SPEC): Likewise.
9807 (CPP_CPU_SPEC): Handle M8.
9808 (ASM_CPU_SPEC): Likewise.
9809 (AS_M8_FLAG): Define.
9810 * config/sparc/sparc.md: Add m8 to the cpu attribute.
9811 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
9812 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
9814 * configure: Regenerate.
9815 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
9818 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9820 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
9822 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
9823 ("*movdi_insn_sp32"): Do not set v3pipe.
9824 ("*movsi_insn"): Likewise.
9825 ("*movdi_insn_sp64"): Likewise.
9826 ("*movsf_insn"): Likewise.
9827 ("*movdf_insn_sp32"): Likewise.
9828 ("*movdf_insn_sp64"): Likewise.
9829 ("*zero_extendsidi2_insn_sp64"): Likewise.
9830 ("*sign_extendsidi2_insn"): Likewise.
9831 ("*mov<VM32:mode>_insn"): Likewise.
9832 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9833 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9834 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9835 ("<vlop:code><VL:mode>3"): Likewise.
9836 ("*not_<vlop:code><VL:mode>3"): Likewise.
9837 ("*nand<VL:mode>_vis"): Likewise.
9838 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
9839 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
9840 ("one_cmpl<VL:mode>2"): Likewise.
9841 ("faligndata<VM64:mode>_vis"): Likewise.
9842 ("alignaddrsi_vis"): Likewise.
9843 ("alignaddrdi_vis"): Likweise.
9844 ("alignaddrlsi_vis"): Likewise.
9845 ("alignaddrldi_vis"): Likewise.
9846 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9847 ("bmaskdi_vis"): Likewise.
9848 ("bmasksi_vis"): Likewise.
9849 ("bshuffle<VM64:mode>_vis"): Likewise.
9850 ("cmask8<P:mode>_vis"): Likewise.
9851 ("cmask16<P:mode>_vis"): Likewise.
9852 ("cmask32<P:mode>_vis"): Likewise.
9853 ("pdistn<P:mode>_vis"): Likewise.
9854 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9856 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9858 * config/sparc/sparc.md ("subtype"): New insn attribute.
9859 ("*wrgsr_sp64"): Set insn subtype.
9860 ("*rdgsr_sp64"): Likewise.
9861 ("alignaddrsi_vis"): Likewise.
9862 ("alignaddrdi_vis"): Likewise.
9863 ("alignaddrlsi_vis"): Likewise.
9864 ("alignaddrldi_vis"): Likewise.
9865 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9866 ("fexpand_vis"): Likewise.
9867 ("fpmerge_vis"): Likewise.
9868 ("faligndata<VM64:mode>_vis"): Likewise.
9869 ("bshuffle<VM64:mode>_vis"): Likewise.
9870 ("cmask8<P:mode>_vis"): Likewise.
9871 ("cmask16<P:mode>_vis"): Likewise.
9872 ("cmask32<P:mode>_vis"): Likewise.
9873 ("fchksm16_vis"): Likewise.
9874 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
9875 ("fmean16_vis"): Likewise.
9876 ("fp<plusminus_insn>64_vis"): Likewise.
9877 ("<plusminus_insn>v8qi3"): Likewise.
9878 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9879 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
9880 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
9881 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
9882 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
9883 ("*movqi_insn"): Likewise.
9884 ("*movhi_insn"): Likewise.
9885 ("*movsi_insn"): Likewise.
9886 ("movsi_pic_gotdata_op"): Likewise.
9887 ("*movdi_insn_sp32"): Likewise.
9888 ("*movdi_insn_sp64"): Likewise.
9889 ("movdi_pic_gotdata_op"): Likewise.
9890 ("*movsf_insn"): Likewise.
9891 ("*movdf_insn_sp32"): Likewise.
9892 ("*movdf_insn_sp64"): Likewise.
9893 ("*zero_extendhisi2_insn"): Likewise.
9894 ("*zero_extendqihi2_insn"): Likewise.
9895 ("*zero_extendqisi2_insn"): Likewise.
9896 ("*zero_extendqidi2_insn"): Likewise.
9897 ("*zero_extendhidi2_insn"): Likewise.
9898 ("*zero_extendsidi2_insn_sp64"): Likewise.
9899 ("ldfsr"): Likewise.
9900 ("prefetch_64"): Likewise.
9901 ("prefetch_32"): Likewise.
9902 ("tie_ld32"): Likewise.
9903 ("tie_ld64"): Likewise.
9904 ("*tldo_ldub_sp32"): Likewise.
9905 ("*tldo_ldub1_sp32"): Likewise.
9906 ("*tldo_ldub2_sp32"): Likewise.
9907 ("*tldo_ldub_sp64"): Likewise.
9908 ("*tldo_ldub1_sp64"): Likewise.
9909 ("*tldo_ldub2_sp64"): Likewise.
9910 ("*tldo_ldub3_sp64"): Likewise.
9911 ("*tldo_lduh_sp32"): Likewise.
9912 ("*tldo_lduh1_sp32"): Likewise.
9913 ("*tldo_lduh_sp64"): Likewise.
9914 ("*tldo_lduh1_sp64"): Likewise.
9915 ("*tldo_lduh2_sp64"): Likewise.
9916 ("*tldo_lduw_sp32"): Likewise.
9917 ("*tldo_lduw_sp64"): Likewise.
9918 ("*tldo_lduw1_sp64"): Likewise.
9919 ("*tldo_ldx_sp64"): Likewise.
9920 ("*mov<VM32:mode>_insn"): Likewise.
9921 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9922 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9924 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9926 * config/sparc/sparc.md ("type"): New insn type viscmp.
9927 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
9929 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
9930 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
9931 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9932 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
9934 ("n7_vis_logical_11cycle"): Likewise.
9935 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
9936 * config/sparc/niagara2.md ("niag3_vis": Likewise.
9937 * config/sparc/niagara.md ("niag_vis"): Likewise.
9938 * config/sparc/ultra3.md ("us3_fga"): Likewise.
9939 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
9941 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9943 * config/sparc/sparc.md: New instruction type `bmask'.
9944 (bmaskdi_vis): Use the `bmask' type.
9945 (bmasksi_vis): Likewise.
9946 * config/sparc/ultra3.md (us3_array): Likewise.
9947 * config/sparc/niagara7.md (n7_array): Likewise.
9948 * config/sparc/niagara4.md (n4_array): Likewise.
9949 * config/sparc/niagara2.md (niag2_vis): Likewise.
9950 (niag3_vis): Likewise.
9951 * config/sparc/niagara.md (niag_vis): Likewise.
9953 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9955 * ipa-comdats.c: Remove optimize check from gate.
9956 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
9957 for functions not optimized.
9958 (ipa_fn_summary_read): Skip optimize check.
9959 (ipa_fn_summary_write): Likewise.
9960 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
9962 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
9964 (can_inline_edge_p): Check flag_pcc_struct_return for match.
9965 (check_callers): Give up on caller which is not optimized.
9966 (inline_small_functions): Likewise.
9967 (ipa_inline): Do not give up when not optimizing.
9968 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
9969 away unoptimizes cdtors.
9970 (whole_program_function_and_variable_visibility): Do
9971 ipa_discover_readonly_nonaddressable_vars in LTO mode.
9972 * ipa.c (process_references): Do not check optimize.
9973 (symbol_table::remove_unreachable_nodes): Update optimize check.
9974 (set_writeonly_bit): Update optimize check.
9975 (pass_ipa_cdtor_merge::gate): Do not check optimize.
9976 (pass_ipa_single_use::gate): Remove.
9978 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9980 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
9981 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
9982 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
9983 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
9984 permute_load, permute_store, adjust_extract, adjust_splat,
9985 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
9986 replace_swap_with_copy, dump_swap_insn_table,
9987 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
9988 recombine_lvx_pattern, recombine_stvx_pattern,
9989 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
9990 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
9991 to file rs6000-p8swap.c.
9992 * config/rs6000/rs6000-p8swap.c: New file.
9993 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
9994 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
9995 and rs6000*-*-* targets.
9997 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9999 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
10001 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10003 * lto-wrapper.c (merge_and_complain): Do not merge
10004 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
10005 fsigned_zeros, ftrapping_math, fwrapv.
10006 (append_compiler_options): Do not track these options.
10007 (append_linker_options): Likewie
10009 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10011 * cgraphunit.c (cgraph_node::finalize_function): When
10012 !flag_toplevel_reorde set no_reorder flag.
10013 (varpool_node::finalize_decl): Likewise.
10014 (symbol_table::compile): Drop no toplevel reorder path.
10016 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10018 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
10019 edges; zero probability is not better than uninitialized.
10021 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10023 * asan.h (asan_sanitize_allocas_p): Declare.
10024 * asan.c (asan_sanitize_allocas_p): New function.
10025 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
10026 (handle_builtin_alloca): Likewise.
10027 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
10028 if !asan_sanitize_allocas_p.
10029 * params.def (asan-instrument-allocas): Add new option.
10030 * params.h (ASAN_PROTECT_ALLOCAS): Define.
10031 * opts.c (common_handle_option): Disable allocas sanitization for
10034 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10036 * asan.c: Include gimple-fold.h.
10037 (get_last_alloca_addr): New function.
10038 (handle_builtin_stackrestore): Likewise.
10039 (handle_builtin_alloca): Likewise.
10040 (asan_emit_allocas_unpoison): Likewise.
10041 (get_mem_refs_of_builtin_call): Add new parameter, remove const
10042 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
10043 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
10044 (instrument_builtin_call): Pass gimple iterator to
10045 get_mem_refs_of_builtin_call.
10046 (last_alloca_addr): New global.
10047 * asan.h (asan_emit_allocas_unpoison): Declare.
10048 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
10049 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
10050 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
10051 if function calls alloca.
10052 * gimple-fold.c (replace_call_with_value): Remove static keyword.
10053 * gimple-fold.h (replace_call_with_value): Declare.
10054 * internal-fn.c: Include asan.h.
10055 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
10056 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
10058 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10060 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
10061 (C_SELFTEST_FLAGS): New.
10062 (CPP_SELFTEST_FLAGS): New.
10063 (SELFTEST_DEPS): New, from deps of s-selftest.
10064 (C_SELFTEST_DEPS): New, from deps of s-selftest.
10065 (CPP_SELFTEST_DEPS): New.
10066 (selftest): Add dependency on s-selftest-c++.
10067 (s-selftest): Rename to...
10068 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
10069 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
10070 than SELFTEST_FLAGS.
10071 (selftest-gdb): Rename to...
10072 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
10074 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
10075 (selftest-valgrind): Rename to...
10076 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
10078 (selftest-valgrind): Reintroduce as an alias for
10079 selftest-c-valgrind.
10080 (s-selftest-c++): New.
10081 (selftest-c++-gdb): New.
10082 (selftest-c++-valgrind): New.
10084 2017-07-06 Olivier Hainque <hainque@adacore.com>
10086 * gcc.c (process_command): When deciding if undefined variables
10087 should be ignored when processing specs, accept "gcc -v" as well.
10089 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10091 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
10092 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
10094 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10096 * config/arm/arm-cpus.in (armv8-r): Add new entry.
10097 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
10098 * config/arm/arm-tables.opt: Regenerate.
10099 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
10101 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
10103 2017-07-06 Carl Love <cel@us.ibm.com>
10105 * ChangeLog: Clean up from mid air collision
10107 2017-07-06 Carl Love <cel@us.ibm.com>
10109 * config/rs6000/rs6000-c.c: Add support for built-in functions
10110 vector signed int vec_subc (vector signed int, vector signed int);
10111 vector signed __int128 vec_subc (vector signed __int128,
10112 vector signed __int128);
10113 vector unsigned __int128 vec_subc (vector unsigned __int128,
10114 vector unsigned __int128);
10115 vector signed int vec_sube (vector signed int, vector signed int,
10116 vector signed int);
10117 vector unsigned int vec_sube (vector unsigned int,
10118 vector unsigned int,
10119 vector unsigned int);
10120 vector signed __int128 vec_sube (vector signed __int128,
10121 vector signed __int128,
10122 vector signed__int128);
10123 vector unsigned __int128 vec_sube (vector unsigned __int128,
10124 vector unsigned __int128,
10125 vector unsigned __int128);
10126 vector signed int vec_subec (vector signed int, vector signed int,
10127 vector signed int);
10128 vector unsigned int vec_subec (vector unsigned int,
10129 vector unsigned int,
10130 vector unsigned int);
10131 vector signed __int128 vec_subec (vector signed __int128,
10132 vector signed __int128,
10133 vector signed__int128);
10134 vector unsigned __int128 vec_subec (vector unsigned __int128,
10135 vector unsigned __int128,
10136 vector unsigned __int128);
10137 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
10138 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
10139 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
10140 BU_ALTIVEC_OVERLOAD_X definitions.
10141 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
10142 * doc/extend.texi: Update the built-in documentation file for the new
10143 built-in functions.
10145 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10148 * diagnostic-show-locus.c (layout::layout): Use start and finish
10149 spelling location for the start and finish of each range.
10150 * genmatch.c (linemap_client_expand_location_to_spelling_point):
10151 Add unused aspect param.
10152 * input.c (expand_location_1): Add "aspect" param, and use it
10153 to access the correct part of the location.
10154 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
10156 (expand_location_to_spelling_point): Likewise.
10157 (linemap_client_expand_location_to_spelling_point): Add "aspect"
10158 param, and pass it to expand_location_1.
10160 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
10162 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
10163 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
10164 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
10165 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
10166 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
10167 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
10168 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
10169 _mm_maskz_getmant_ss): New intrinsics.
10170 (__builtin_ia32_getexpss128_mask): Changed to ...
10171 __builtin_ia32_getexpss128_round ... this.
10172 (__builtin_ia32_getexpsd128_mask): Changed to ...
10173 __builtin_ia32_getexpsd128_round ... this.
10174 * config/i386/i386-builtin-types.def
10175 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
10176 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
10177 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
10178 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
10179 __builtin_ia32_getmantss_mask_round): New builtins.
10180 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
10181 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
10182 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
10183 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
10184 * config/i386/sse.md
10185 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
10186 avx512f_sgetexp<mode><mask_scalar_name>
10187 <round_saeonly_scalar_name> ... this.
10188 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
10189 %0, %1, %2<round_saeonly_op3>}): Changed to ...
10190 vgetexp<ssescalarmodesuffix>
10191 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10192 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
10193 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
10194 avx512f_vgetmant<mode><mask_scalar_name>
10195 <round_saeonly_scalar_name> ... this.
10196 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10197 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
10198 vgetmant<ssescalarmodesuffix>
10199 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
10200 %0<mask_scalar_operand4>, %1, %2
10201 <round_saeonly_scalar_mask_op4>, %3} ... this.
10202 * config/i386/subst.md (mask_scalar_operand4,
10203 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
10204 round_saeonly_scalar_nimm_predicate): New subst attributes.
10206 2017-07-06 Julia Koval <julia.koval@intel.com>
10208 * config/i386/i386.c (ix86_erase_embedded_rounding):
10209 Remove code for old rounding pattern.
10211 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
10213 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
10215 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
10217 * doc/sourcebuild.texi (Test Directives, Variants of
10218 dg-require-support): Add documentation for dg-require-stack-check.
10220 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
10222 * config/i386/subst.md (mask_scalar, round_scalar,
10223 round_saeonly_scalar): New meta-templates.
10224 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
10225 round_scalar_mask_operand3, round_scalar_mask_op3,
10226 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
10227 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
10228 round_saeonly_scalar_constraint,
10229 round_saeonly_scalar_prefix): New subst attribute.
10230 * config/i386/sse.md
10231 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
10232 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
10233 <round_scalar_name> ... this.
10234 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
10235 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
10236 <round_scalar_name> ... this.
10237 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
10238 <sse>_vm<code><mode>3<mask_scalar_name>
10239 <round_saeonly_scalar_name> ... this.
10240 (v<plusminus_mnemonic><ssescalarmodesuffix>
10241 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10242 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10243 v<plusminus_mnemonic><ssescalarmodesuffix>
10244 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10245 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10246 (v<multdiv_mnemonic><ssescalarmodesuffix>
10247 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10248 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10249 v<multdiv_mnemonic><ssescalarmodesuffix>
10250 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10251 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10252 (v<maxmin_float><ssescalarmodesuffix>
10253 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
10254 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
10255 v<maxmin_float><ssescalarmodesuffix>
10256 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10257 %0<mask_scalar_operand3>, %1, %<iptr>2
10258 <round_saeonly_scalar_mask_op3>} ... this.
10260 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
10262 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
10263 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
10265 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
10266 Alan Hayward <alan.hayward@arm.com>
10267 David Sherwood <david.sherwood@arm.com>
10269 * combine.c (simplify_if_then_else): Remove "enum" before
10271 * compare-elim.c (can_eliminate_compare): Likewise.
10272 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10274 (aarch64_lookup_simd_builtin_type): Likewise.
10275 (aarch64_simd_builtin_type): Likewise.
10276 (aarch64_init_simd_builtin_types): Likewise.
10277 (aarch64_simd_expand_args): Likewise.
10278 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
10280 (aarch64_reverse_mask): Likewise.
10281 (aarch64_simd_emit_reg_reg_move): Likewise.
10282 (aarch64_gen_adjusted_ldpstp): Likewise.
10283 (aarch64_ccmp_mode_to_code): Likewise.
10284 (aarch64_operands_ok_for_ldpstp): Likewise.
10285 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10286 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10288 (aarch64_min_divisions_for_recip_mul): Likewise.
10289 (aarch64_reassociation_width): Likewise.
10290 (aarch64_get_condition_code_1): Likewise.
10291 (aarch64_simd_emit_reg_reg_move): Likewise.
10292 (aarch64_simd_attr_length_rglist): Likewise.
10293 (aarch64_reverse_mask): Likewise.
10294 (aarch64_operands_ok_for_ldpstp): Likewise.
10295 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10296 (aarch64_gen_adjusted_ldpstp): Likewise.
10297 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
10299 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
10300 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10301 (arm_lookup_simd_builtin_type): Likewise.
10302 (arm_simd_builtin_type): Likewise.
10303 (arm_init_simd_builtin_types): Likewise.
10304 (arm_expand_builtin_args): Likewise.
10305 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
10306 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
10307 (ft32_setup_incoming_varargs): Likewise.
10308 (ft32_function_arg): Likewise.
10309 (ft32_function_arg_advance): Likewise.
10310 (ft32_pass_by_reference): Likewise.
10311 (ft32_arg_partial_bytes): Likewise.
10312 (ft32_valid_pointer_mode): Likewise.
10313 (ft32_addr_space_pointer_mode): Likewise.
10314 (ft32_addr_space_legitimate_address_p): Likewise.
10315 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
10317 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
10318 (ix86_emit_outlined_ms2sysv_restore): Likewise.
10319 (iamcu_alignment): Likewise.
10320 (canonicalize_vector_int_perm): Likewise.
10321 (ix86_noce_conversion_profitable_p): Likewise.
10322 (ix86_mpx_bound_mode): Likewise.
10323 (ix86_operands_ok_for_move_multiple): Likewise.
10324 * config/microblaze/microblaze-protos.h
10325 (microblaze_expand_conditional_branch_reg): Likewise.
10326 * config/microblaze/microblaze.c
10327 (microblaze_expand_conditional_branch_reg): Likewise.
10328 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10330 (rs6000_reassociation_width): Likewise.
10331 (rs6000_invalid_binary_op): Likewise.
10332 (fusion_p9_p): Likewise.
10333 (emit_fusion_p9_load): Likewise.
10334 (emit_fusion_p9_store): Likewise.
10335 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
10337 (riscv_hard_regno_mode_ok_p): Likewise.
10338 (riscv_address_insns): Likewise.
10339 (riscv_split_symbol): Likewise.
10340 (riscv_legitimize_move): Likewise.
10341 (riscv_function_value): Likewise.
10342 (riscv_hard_regno_nregs): Likewise.
10343 (riscv_expand_builtin): Likewise.
10344 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
10345 (riscv_build_integer): Likewise.
10346 (riscv_split_integer): Likewise.
10347 (riscv_legitimate_constant_p): Likewise.
10348 (riscv_cannot_force_const_mem): Likewise.
10349 (riscv_regno_mode_ok_for_base_p): Likewise.
10350 (riscv_valid_base_register_p): Likewise.
10351 (riscv_valid_offset_p): Likewise.
10352 (riscv_valid_lo_sum_p): Likewise.
10353 (riscv_classify_address): Likewise.
10354 (riscv_legitimate_address_p): Likewise.
10355 (riscv_address_insns): Likewise.
10356 (riscv_load_store_insns): Likewise.
10357 (riscv_force_binary): Likewise.
10358 (riscv_split_symbol): Likewise.
10359 (riscv_force_address): Likewise.
10360 (riscv_legitimize_address): Likewise.
10361 (riscv_move_integer): Likewise.
10362 (riscv_legitimize_const_move): Likewise.
10363 (riscv_legitimize_move): Likewise.
10364 (riscv_address_cost): Likewise.
10365 (riscv_subword): Likewise.
10366 (riscv_output_move): Likewise.
10367 (riscv_canonicalize_int_order_test): Likewise.
10368 (riscv_emit_int_order_test): Likewise.
10369 (riscv_function_arg_boundary): Likewise.
10370 (riscv_pass_mode_in_fpr_p): Likewise.
10371 (riscv_pass_fpr_single): Likewise.
10372 (riscv_pass_fpr_pair): Likewise.
10373 (riscv_get_arg_info): Likewise.
10374 (riscv_function_arg): Likewise.
10375 (riscv_function_arg_advance): Likewise.
10376 (riscv_arg_partial_bytes): Likewise.
10377 (riscv_function_value): Likewise.
10378 (riscv_pass_by_reference): Likewise.
10379 (riscv_setup_incoming_varargs): Likewise.
10380 (riscv_print_operand): Likewise.
10381 (riscv_elf_select_rtx_section): Likewise.
10382 (riscv_save_restore_reg): Likewise.
10383 (riscv_for_each_saved_reg): Likewise.
10384 (riscv_register_move_cost): Likewise.
10385 (riscv_hard_regno_mode_ok_p): Likewise.
10386 (riscv_hard_regno_nregs): Likewise.
10387 (riscv_class_max_nregs): Likewise.
10388 (riscv_memory_move_cost): Likewise.
10389 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
10390 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
10391 (rl78_addr_space_address_mode): Likewise.
10392 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10394 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10395 (rs6000_reassociation_width): Likewise.
10396 (rs6000_invalid_binary_op): Likewise.
10397 (fusion_p9_p): Likewise.
10398 (emit_fusion_p9_load): Likewise.
10399 (emit_fusion_p9_store): Likewise.
10400 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
10401 (ok_for_simple_move_operands): Likewise.
10402 (ok_for_simple_move_strict_operands): Likewise.
10403 (ok_for_simple_arith_logic_operands): Likewise.
10404 (visium_legitimize_reload_address): Likewise.
10405 (visium_select_cc_mode): Likewise.
10406 (output_cbranch): Likewise.
10407 (visium_split_double_move): Likewise.
10408 (visium_expand_copysign): Likewise.
10409 (visium_expand_int_cstore): Likewise.
10410 (visium_expand_fp_cstore): Likewise.
10411 * config/visium/visium.c (visium_pass_by_reference): Likewise.
10412 (visium_function_arg): Likewise.
10413 (visium_function_arg_advance): Likewise.
10414 (visium_libcall_value): Likewise.
10415 (visium_setup_incoming_varargs): Likewise.
10416 (visium_legitimate_constant_p): Likewise.
10417 (visium_legitimate_address_p): Likewise.
10418 (visium_legitimize_address): Likewise.
10419 (visium_secondary_reload): Likewise.
10420 (visium_register_move_cost): Likewise.
10421 (visium_memory_move_cost): Likewise.
10422 (prepare_move_operands): Likewise.
10423 (ok_for_simple_move_operands): Likewise.
10424 (ok_for_simple_move_strict_operands): Likewise.
10425 (ok_for_simple_arith_logic_operands): Likewise.
10426 (visium_function_value_1): Likewise.
10427 (rtx_ok_for_offset_p): Likewise.
10428 (visium_legitimize_reload_address): Likewise.
10429 (visium_split_double_move): Likewise.
10430 (visium_expand_copysign): Likewise.
10431 (visium_expand_int_cstore): Likewise.
10432 (visium_expand_fp_cstore): Likewise.
10433 (visium_split_cstore): Likewise.
10434 (visium_select_cc_mode): Likewise.
10435 (visium_split_cbranch): Likewise.
10436 (output_cbranch): Likewise.
10437 (visium_print_operand_address): Likewise.
10438 * expmed.c (flip_storage_order): Likewise.
10439 * expmed.h (emit_cstore): Likewise.
10440 (flip_storage_order): Likewise.
10441 * genrecog.c (validate_pattern): Likewise.
10442 * hsa-gen.c (gen_hsa_addr): Likewise.
10443 * internal-fn.c (expand_arith_overflow): Likewise.
10444 * ira-color.c (allocno_copy_cost_saving): Likewise.
10445 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10446 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10447 (process_invariant_for_inheritance): Likewise.
10448 * lra-eliminations.c (move_plus_up): Likewise.
10449 * omp-low.c (lower_oacc_reductions): Likewise.
10450 * simplify-rtx.c (simplify_subreg): Likewise.
10451 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10452 (TARGET_CHKP_BOUND_MODE): Likewise..
10453 * targhooks.c (default_chkp_bound_mode): Likewise.
10454 (default_setup_incoming_vararg_bounds): Likewise.
10455 * targhooks.h (default_chkp_bound_mode): Likewise.
10456 (default_setup_incoming_vararg_bounds): Likewise.
10457 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10458 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10459 (have_whole_vector_shift): Likewise.
10460 * tree-vect-stmts.c (vectorizable_load): Likewise.
10461 * doc/tm.texi: Regenerate.
10463 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10465 Graceful degrade if Binutils PR21472 is not available.
10468 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10469 .rodata in flash test fails.
10470 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10471 * confgure: Regenerate.
10472 * config.in: Regenerate.
10473 * config/avr/avr.c (avr_asm_named_section)
10474 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10475 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10476 (avr_asm_init_sections): Same.
10478 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10480 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10481 (fma<VH:mode>4_intrinsic): Likewise.
10482 (*fmsub<VCVTF:mode>4): Likewise.
10483 (*fmsub<VH:mode>4_intrinsic): Likewise.
10485 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10488 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10489 Don't depend on "optimize > 0".
10490 (out_movhi_r_mr, out_movqi_mr_r): Same.
10491 (out_movhi_mr_r, out_movqi_r_mr): Same.
10492 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10493 io_address_operand on "optimize > 0".
10495 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10497 * tree-loop-distribution.c: Add general explanantion on the pass.
10498 (generate_loops_for_partition): Mark distributed loop.
10499 (pg_add_dependence_edges): New parameter. Handle alias data
10500 dependence specially and record it in the parameter if asked.
10501 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10502 (init_partition_graph_vertices, add_partition_graph_edge): New.
10503 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10504 (free_partition_graph_vdata, build_partition_graph): New.
10505 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10506 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10507 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10508 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10509 (version_for_distribution_p, finalize_partitions): New.
10510 (distribute_loop): Handle alias data dependence specially. Factor
10511 out loop fusion code as functions and call these functions.
10513 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10515 * tree-loop-distribution.c (classify_partition): New parameter and
10516 better handle reduction statement.
10517 (rdg_build_partitions): Revise comment.
10518 (distribute_loop): Compute statements in all partitions and pass it
10519 to classify_partition.
10521 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10523 * tree-loop-distribution.c (enum partition_type): New.
10524 (struct partition): New field type.
10525 (partition_merge_into): Add parameter. Update partition type.
10526 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10527 (build_rdg_partition_for_vertex): Compute partition type.
10528 (rdg_build_partitions): Dump partition type.
10529 (distribute_loop): Update calls to partition_merge_into.
10531 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10533 * tree-loop-distribution.c (struct ddr_hasher): New.
10534 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10536 (classify_partition): Call get_data_dependence.
10537 (pg_add_dependence_edges): Ditto.
10538 (distribute_loop): Release data dependence hash table.
10540 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10542 * tree-loop-distribution.c (ref_base_address): Delete.
10543 (similar_memory_accesses): Rename ...
10544 (share_memory_accesses): ... to this. Check if partitions access
10545 the same memory reference.
10546 (distribute_loop): Call share_memory_accesses.
10548 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10550 * tree-loop-distribution.c (struct partition): New field recording
10551 its data reference.
10552 (partition_alloc, partition_free): Init and release data refs.
10553 (partition_merge_into): Merge data refs.
10554 (build_rdg_partition_for_vertex): Collect data refs for partition.
10555 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10557 (distribute_loop): Remve data refs from vertice data of partition
10560 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10562 * tree-loop-distribution.c (params.h): Include header file.
10563 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10564 (datarefs_vec): New global var.
10565 (create_rdg_vertices): Use datarefs_vec directly.
10566 (free_rdg): Don't free data references.
10567 (build_rdg): Update use. Don't free data references.
10568 (distribute_loop): Compute global variable for data references.
10569 Bail out if there are too many data references.
10571 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10573 * tree-loop-distribution.c (loop_nest): New global var.
10574 (build_rdg): Use loop directly, rather than loop nest.
10575 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10577 (distribute_loop): Compute global variable loop nest. Update use.
10579 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10581 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10582 (partition_merge_into): New parameter. Dump reason for fusion.
10583 (distribute_loop): Update use of partition_merge_into.
10585 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10587 * tree-loop-distribution.c (bb_top_order_index): New.
10588 (bb_top_order_index_size, bb_top_order_cmp): New.
10589 (stmts_from_loop): Use topological order.
10590 (pass_loop_distribution::execute): Compute and release topological
10591 order for basic blocks.
10593 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10595 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10598 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10600 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10601 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10602 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10603 * internal-fn.def (LOOP_DIST_ALIAS): New.
10604 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10605 (fold_loop_internal_call): ... this.
10606 (vect_loop_dist_alias_call): New function.
10607 (set_uid_loop_bbs): Call fold_loop_internal_call.
10608 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10611 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10614 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10615 Require dead FLAGS_REG at the beginning of a peephole.
10617 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10620 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10621 arguments in the call to __builtin_ia32_sbb_u32.
10622 (_subborrow_u64): Swap _X and _Y arguments in the call to
10623 __builtin_ia32_sbb_u64.
10625 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10628 * tree-vrp.c (compare_assert_loc): Turn into a function template
10629 with stable template parameter. Only test if a->e is NULL,
10630 !a->e == !b->e has been verified already. Use e == NULL or
10631 e != NULL instead of e or ! e tests. If stable is true, don't use
10632 iterative_hash_expr, on the other side allow a or b or both NULL
10633 and sort the NULLs last.
10634 (process_assert_insertions): Sort using compare_assert_loc<false>
10635 instead of compare_assert_loc, later sort using
10636 compare_assert_loc<true> before calling process_assert_insertions_for
10637 in a loop. Use break instead of continue once seen NULL pointer.
10639 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10641 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10642 Cortex-R7 and Cortex-R8 processors.
10644 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10646 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10647 uninitialized while src is not.
10649 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10651 * common/config/arm/arm-common.c: Adjust include path for
10653 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10654 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10655 (arm-cpu-data.h): Likewise.
10656 (arm-cpu-cdata.h): Likewise.
10657 * config/arm/arm-cpu.h: Delete.
10658 * config/arm/arm-cpu-cdata.h: Delete.
10659 * config/arm/arm-cpu-data.h: Delete.
10661 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10663 * config/arm/arm-cpus.in (cortex-a55): New.
10664 (cortex-a75): Likewise.
10665 (cortex-a75.cortex-a55): Likewise.
10666 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10668 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
10669 * config/arm/arm-cpu-cdata.h: Regenerate.
10670 * config/arm/arm-cpu-data.h: Regenerate.
10671 * config/arm/arm-cpu.h: Regenerate.
10672 * config/arm/arm-tables.opt: Regenerate.
10673 * config/arm/arm-tune.md: Regenerate.
10675 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10677 * haifa-sched.c (sched_create_recovery_edges): Update profile.
10679 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10681 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10684 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10686 PR tree-optimization/81292
10687 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10688 full_string_p, also call adjust_related_strinfos if the adjustment
10689 is simple, otherwise invalidate related strinfos.
10691 2017-07-04 Martin Liska <mliska@suse.cz>
10694 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10695 newly created variable as DECL_IGNORED_P.
10697 2017-07-04 Martin Liska <mliska@suse.cz>
10700 * ipa-inline.c (inline_small_functions):
10701 Use xstrdup_for_dump.
10703 2017-07-04 Tom de Vries <tom@codesourcery.com>
10705 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10707 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10710 * config/darwin.c (darwin_function_switched_text_sections):
10713 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10715 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10717 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10719 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10721 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10723 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10724 min_profitable_iters, and th as inclusive lower bounds.
10725 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10726 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10727 for min_profitable_iters and min_profitable_estimate.
10728 (vect_transform_loop): Treat th as an inclusive lower bound.
10729 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10731 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10734 * config/darwin.c (darwin_function_switched_text_sections):
10735 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10736 in two pieces, and suppress the use of buf.
10738 2017-07-03 Nathan Sidwell <nathan@acm.org>
10740 * hash-table.h (hash_table_mod1): Fix indentation.
10742 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10744 PR middle-end/81290
10745 * predict.c (force_edge_cold): Be more careful about propagation
10747 * profile-count.h (profile_probability::guessed,
10748 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10750 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10752 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10754 * doc/invoke.texi (rcpc architecture extension): Document it.
10756 2017-07-03 Richard Biener <rguenther@suse.de>
10758 PR tree-optimization/60510
10759 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10760 the scalar reduction PHI and use it.
10761 (vectorizable_reduction): Properly guard the single_defuse_cycle
10762 path for non-SLP reduction chains where we cannot use it.
10763 Rework reduc_def/index and vector type deduction. Rework
10764 vector operand gathering during reduction op code-gen.
10765 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10766 chains dissolve the chain and leave it to non-SLP reduction
10769 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10771 * tree-data-ref.h (dr_alignment): Declare.
10772 * tree-data-ref.c (dr_alignment): New function.
10773 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
10774 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
10776 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
10778 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10780 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
10781 and base_misalignment fields.
10782 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
10783 * tree-data-ref.c: Include builtins.h.
10784 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
10785 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
10786 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
10787 * tree-vect-data-refs.c: Include tree-cfg.h.
10788 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
10789 fields instead of calculating an alignment here.
10790 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
10791 innermost_loop_behavior fields.
10793 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10795 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
10797 (DR_STEP_ALIGNMENT): New macro.
10798 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
10799 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
10800 (create_data_ref): Print it.
10801 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
10802 to tell whether the step preserves vector (mis)alignment.
10803 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10804 Move the check for an integer step and generalise to all INTEGER_CST.
10805 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
10806 Print the outer step alignment.
10808 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10810 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
10811 with offset_alignment.
10812 (DR_ALIGNED_TO): Delete.
10813 (DR_OFFSET_ALIGNMENT): New macro.
10814 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
10815 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
10816 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
10817 (create_data_ref): Likewise.
10818 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10819 (vect_analyze_data_refs): Likewise.
10820 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
10821 creating dummy innermost behavior.
10823 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10825 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
10826 with a "innermost_loop_behavior *" and refeence tree.
10827 * tree-data-ref.c (dr_analyze_innermost): Likewise.
10828 (create_data_ref): Update call accordingly.
10829 * tree-predcom.c (find_looparound_phi): Likewise.
10831 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10833 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
10834 fields with dr_wrt_vec_loop.
10835 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
10836 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
10837 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
10838 (vect_dr_behavior): New function.
10839 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10840 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
10841 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
10842 track whether the step preserves the misalignment.
10843 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10844 Use vect_dr_behavior.
10845 (vect_setup_realignment): Update call accordingly.
10846 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
10847 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
10848 call to vect_create_addr_base_for_vector_ref.
10849 (vect_create_cond_for_align_checks): Likewise.
10850 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
10851 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
10852 (vect_recog_mask_conversion_pattern): Likewise.
10853 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
10854 (new_stmt_vec_info): Remove redundant zeroing.
10856 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10858 * common/config/arm/arm-common.c (arm_be8_option): New function.
10859 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
10860 (ISA_ARMv6): Add isa_bit_be8.
10861 * config/arm/arm.h (arm_be8_option): Add prototype.
10862 (BE8_SPEC_FUNCTION): New define.
10863 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
10864 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
10865 (mlittle-endian): Similarly.
10866 (mbe8, mbe32): New options.
10867 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
10868 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
10870 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10872 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
10874 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10876 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
10877 (cleanup_tree_cfg_bb): Use it.
10878 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
10880 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
10882 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10885 * loop-doloop.c (add_test): Update profile.
10887 2017-07-03 Martin Liska <mliska@suse.cz>
10890 * sanopt.c (rewrite_usage_of_param): New function.
10891 (sanitize_rewrite_addressable_params): Likewise.
10892 (pass_sanopt::execute): Call rewrite_usage_of_param.
10894 2017-07-03 Richard Biener <rguenther@suse.de>
10896 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
10897 back to using VIEW_CONVERT_EXPR.
10899 2017-07-03 Martin Liska <mliska@suse.cz>
10902 * doc/extend.texi: Document when a resolver function is
10903 generated for target_clones.
10905 2017-07-03 Martin Liska <mliska@suse.cz>
10907 * asan.c (asan_emit_stack_protection): Unpoison just red zones
10908 and shadow memory of auto variables which are subject of
10909 use-after-scope sanitization.
10910 (asan_expand_mark_ifn): Add do set only when is_poison.
10912 2016-07-03 Richard Biener <rguenther@suse.de>
10914 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
10916 (vect_force_simple_reduction): Record reduction def -> phi mapping.
10917 (vectorizable_reduction): Perform reduction PHI creation when
10918 visiting a reduction PHI and adjust and simplify code generation
10919 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
10920 (vect_transform_loop): Visit reduction PHIs.
10921 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
10922 defs into the SLP tree.
10923 (vect_build_slp_tree): Reduction defs terminate the recursion.
10924 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
10926 (vect_get_vec_defs_for_stmt_copy): Export.
10927 (vect_get_vec_defs): Likewise.
10928 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
10930 (vect_get_vec_defs_for_stmt_copy): Declare.
10931 (vect_get_vec_defs): Likewise.
10933 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10935 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
10936 parameter with a "loop" parameter and use it instead of the
10937 loop containing DR_STMT. Don't check simple_iv when doing
10938 BB analysis. Describe the two analysis modes in the comment.
10940 2017-07-03 Tom de Vries <tom@codesourcery.com>
10942 PR tree-optimization/69468
10943 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
10944 (find_same_succ_bb): Handle ignore_edge_flags.
10946 2017-07-03 Tom de Vries <tom@codesourcery.com>
10948 PR tree-optimization/81192
10949 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
10951 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
10953 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
10955 2017-07-03 Tom de Vries <tom@codesourcery.com>
10957 PR tree-optimization/81192
10958 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
10959 BB_SAME_SUCC (bb) == NULL.
10961 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10963 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
10966 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10968 * dumpfile.c: Include profile-count.h
10969 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
10971 (insert_cond_bb): Update profile.
10972 * tree-cfg.h (insert_cond_bb): Update prototype.
10973 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
10974 * tree-dump.c: Do not include tree-cfg.
10976 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10978 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
10980 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10982 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
10985 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10987 * tree-complex.c (expand_complex_div_wide): update profile.
10989 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10990 Alan Hayward <alan.hayward@arm.com>
10991 David Sherwood <david.sherwood@arm.com>
10993 * Makefile.in (MACHMODE_H): Remove insn-modes.h
10994 (CORETYPES_H): New define.
10995 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
10996 (insn-modes-inline.h, s-modes-inline-h): New rules.
10997 (generated_files): Add insn-modes-inline.h.
10998 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
10999 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
11000 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
11001 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
11002 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
11003 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
11004 (build/gencodes.o, build/genconditions.o): Likewise.
11005 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
11006 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
11007 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
11008 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
11009 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
11010 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
11011 * coretypes.h: Include everything up to real.h for generators.
11012 Include insn-modes.h first. Include wide-int-print.h after
11013 wide-int.h. Include insn-modes-inline.h and then machmode.h.
11014 * machmode.h: Don't include insn-modes.h here.
11015 * function-tests.c: Remove includes of signop.h, machmode.h,
11016 double-int.h and wide-int.h.
11018 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
11020 * optc-save-gen.awk: Likewise.
11021 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
11022 * godump.c: Remove include of wide-int-print.h.
11023 * pretty-print.h: Likewise.
11024 * wide-int-print.cc: Likewise.
11025 * wide-int.cc: Likewise.
11026 * hash-map-tests.c: Remove include of signop.h.
11027 * hash-set-tests.c: Likewise.
11028 * rtl-tests.c: Likewise.
11029 * mkconfig.sh: Remove include of machmode.h.
11030 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
11032 (emit_insn_modes_inline_h): ...this new function. Emit the code
11033 into an insn-modes-inline.h header file, adding appropriate
11034 include guards and end comments.
11035 (emit_insn_modes_c_header): Remove include of machmode.h.
11036 (emit_min_insn_modes_c_header): Include coretypes.h rather than
11038 (main): Handle -i flag and call emit_insn_modes_inline_h when
11041 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11043 * tree-ssa-strlen.c (strinfo): Rename the length field to
11044 nonzero_chars. Add a full_string_p field.
11045 (compare_nonzero_chars, zero_length_string_p): New functions.
11046 (get_addr_stridx): Add an offset_out parameter.
11047 Use compare_nonzero_chars.
11048 (get_stridx): Update accordingly. Use compare_nonzero_chars.
11049 (new_strinfo): Update after above changes to strinfo.
11050 (set_endptr_and_length): Set full_string_p.
11051 (get_string_length): Update after above changes to strinfo.
11052 (unshare_strinfo): Update call to new_strinfo.
11053 (maybe_invalidate): Likewise.
11054 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
11055 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
11056 as a uhwi instead of an shwi. Update after above changes to
11057 strinfo and new_strinfo.
11058 (zero_length_string): Assert that chainsi contains full strings.
11059 Use zero_length_string_p. Update call to new_strinfo.
11060 (adjust_related_strinfos): Update after above changes to strinfo.
11061 Copy full_string_p from origsi.
11062 (adjust_last_stmt): Use zero_length_string_p.
11063 (handle_builtin_strlen): Update after above changes to strinfo and
11064 new_strinfo. Install the lhs as the string length if the previous
11065 entry didn't describe a full string.
11066 (handle_builtin_strchr): Update after above changes to strinfo
11068 (handle_builtin_strcpy): Likewise.
11069 (handle_builtin_strcat): Likewise.
11070 (handle_builtin_malloc): Likewise.
11071 (handle_pointer_plus): Likewise.
11072 (handle_builtin_memcpy): Likewise. Track nonzero characters
11073 that aren't necessarily followed by a nul terminator.
11074 (handle_char_store): Likewise.
11076 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11078 PR tree-optimization/80769
11079 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
11080 for malloc and calloc. Document the new invariant that all related
11081 strinfos have delayed lengths or none do.
11082 (verify_related_strinfos): Move earlier in file.
11083 (set_endptr_and_length): New function, split out from...
11084 (get_string_length): ...here. Also set the lengths of related
11086 (zero_length_string): Assert that chainsi has known (rather than
11088 (adjust_related_strinfos): Likewise.
11090 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11092 PR tree-optimization/81136
11093 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
11094 assert that two references with the same misalignment have the same
11095 compile-time misalignment if those compile-time misalignments
11098 2017-07-01 Andi Kleen <ak@linux.intel.com>
11100 * print-tree.c (print_node): Print all attributes.
11102 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11104 * cfg.c (scale_bbs_frequencies): New function.
11105 * cfg.h (scale_bbs_frequencies): Declare it.
11106 * cfgloopanal.c (single_likely_exit): Cleanup.
11107 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
11109 (scale_loop_profile): Likewise.
11110 (loop_version): Likewise.
11111 (create_empty_loop_on_edge): Update.
11112 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
11113 scale_loop_frequencies, scale_loop_profile, loopify,
11114 loop_version): Update prototypes.
11115 * modulo-sched.c (sms_schedule): Update.
11116 * predict.c (unlikely_executed_edge_p): Also check probability.
11117 (probably_never_executed_edge_p): Fix typo.
11118 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11119 * tree-parloops.c (gen_parallel_loop): Update.
11120 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
11121 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11122 * tree-ssa-loop-split.c (split_loop): Update.
11123 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11124 * tree-vect-loop-manip.c (vect_do_peeling): Update.
11125 (vect_loop_versioning): Update.
11126 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11128 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11130 * trans-mem.c (split_bb_make_tm_edge): Update profile.
11132 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11134 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
11135 to keep profile consistent.
11137 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11139 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
11140 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
11141 * profile-count.h (max_safe_multiplier): Make unsigned.
11142 (profile_count::guessed_zero): New.
11144 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11146 * bb-reorder.c (fix_up_crossing_landing_pad,
11147 fix_crossing_conditional_branches): Use make_single_succ_edge
11148 to keep profile consistent.
11150 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11152 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
11155 2017-07-01 Jakub Jelinek <jakub@redhat.com>
11158 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
11159 the right scopes, make sure cond_jump isn't preserved between multiple
11160 iterations. Search for fallthru edge whenever there are 3+ edges and
11161 use find_fallthru_edge for it.
11163 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11165 Patch by Alexander Monakov <amonakov@ispras.ru>
11166 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
11167 probabilities consistently.
11169 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11171 * pa.c (pa_expand_compare_and_swap_loop): Update call of
11172 emit_cmp_and_jump_insns.
11174 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11177 * tree-inline.c (expand_call_inline): Combine profile statuses.
11179 2017-06-30 Andrew Pinski <apinski@cavium.com>
11181 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
11182 fold_stmt returned true.
11184 2017-06-30 Nathan Sidwell <nathan@acm.org>
11186 * ggc.h (empty_string): Delete.
11187 * cfgexpand.c (expand_asm_stmt): Use plain "".
11188 * optabs.c (expand_asm_memory_barrier): Likewise.
11189 * stringpool.c (empty_string): Delete.
11190 (digit_vector, digit_string): Delete.
11191 (ggc_alloc_string): Use plain "", don't optimize single digit
11192 strings. Use ggc_alloc_atomic.
11194 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
11196 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
11197 comparison set and one other set, use the cost of the non-comparison
11200 2017-06-30 Nathan Sidwell <nathan@acm.org>
11202 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
11205 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
11207 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
11208 loops. Remove now unneeded calls to gimple_switch_set_label() that
11209 just set removed labels to NULL_TREE.
11211 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
11213 * tree-ssanames.c (set_range_info_raw): Abstract from ...
11214 (set_range_info): ...here. Only call set_range_info_raw if domain
11216 (set_nonzero_bits): Call set_range_info_raw.
11217 * tree-ssanames.h (set_range_info_raw): New.
11219 2017-06-30 Jakub Jelinek <jakub@redhat.com>
11222 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
11223 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
11224 of nonimmediate_operand and <store_mask_constraint> instead of m
11225 for the input operand. For V8FI iterator, always split if input
11226 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
11227 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
11228 <store_mask_predicate> instead of register_operand and
11229 <store_mask_constraint> instead of v for the input operand. Make
11230 sure both operands aren't MEMs for if not <mask_applied>.
11232 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
11234 * lto-wrapper.c (copy_file) Close both file descriptors before
11237 2017-06-30 Martin Liska <mliska@suse.cz>
11240 * multiple_target.c (create_dispatcher_calls): Make ifunc
11241 also for function that don't have calls or are not referenced.
11243 2017-06-30 Richard Biener <rguenther@suse.de>
11245 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
11246 analyze the first scalar stmt. Move vector type computation
11247 for the BB case here from ...
11248 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
11249 live operation processing in the SLP case properly.
11251 2017-06-30 Richard Biener <rguenther@suse.de>
11253 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
11255 2017-06-30 Martin Liska <mliska@suse.cz>
11258 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
11259 before BUILT_IN_UNWIND_RESUME when ASAN is used.
11261 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
11263 * doc/invoke.texi (AArch64): Add missing options and remove redundant
11266 2017-06-30 Richard Biener <rguenther@suse.de>
11268 PR tree-optimization/81249
11269 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
11270 condition reduction result to original scalar type.
11272 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11274 * profile-count.h (enum profile_quality): Fix typos and whitespace
11277 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11279 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
11280 type for branch probabilities.
11282 2017-06-29 Julian Brown <julian@codesourcery.com>
11283 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11285 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
11286 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
11287 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
11288 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
11290 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11292 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
11293 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
11294 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
11295 CC usage from generic code to here.
11296 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
11297 CC usage into the target macros.
11299 2017-06-29 Maya Rashish <coypu@sdf.org>
11301 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
11304 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11306 * arm/arm-builtins.c: Include profile-count.h
11307 * except.c (sjlj_emit_function_enter): Use
11308 profile_probability::unlikely.
11310 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11312 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
11313 and tocrel_offset be pointer args rather than implicitly using
11315 (legitimate_constant_pool_address_p, rs6000_emit_move,
11316 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
11317 tocrel_offset and use in toc_relative_expr_p call.
11318 (print_operand, print_operand_address): Use static tocrel_base_oac
11319 and tocrel_offset_oac.
11320 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
11323 2017-06-29 Maya Rashish <coypu@sdf.org>
11325 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
11327 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
11329 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
11330 objects, take into account only the alignment of 'op0' and 'mode1' if
11333 2017-06-29 Steve Ellcey <sellcey@cavium.com>
11335 * ccmp.c (ccmp_tree_comparison_p): New function.
11336 (ccmp_candidate_p): Update to use above function.
11337 (get_compare_parts): New function.
11338 (expand_ccmp_next): Update to use new functions.
11339 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
11341 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
11342 take mode as argument.
11343 * ccmp.h (expand_ccmp_expr): Add mode as argument.
11344 * expr.c (expand_expr_real_1): Pass mode as argument.
11346 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
11348 * combine.c (combine_instructions): Print insns to dump_file, together
11351 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11353 * asan.c (asan_emit_stack_protection): Update.
11354 (create_cond_insert_point): Update.
11355 * auto-profile.c (afdo_propagate_circuit): Update.
11356 * basic-block.h (struct edge_def): Turn probability to
11357 profile_probability.
11358 (EDGE_FREQUENCY): Update.
11359 * bb-reorder.c (find_traces_1_round): Update.
11360 (better_edge_p): Update.
11361 (sanitize_hot_paths): Update.
11362 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
11363 (make_single_succ_edge): Update.
11364 (check_bb_profile): Update.
11365 (dump_edge_info): Update.
11366 (update_bb_profile_for_threading): Update.
11367 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
11368 probabilitycount to 0.
11369 * cfgbuild.c (compute_outgoing_frequencies): Update.
11370 * cfgcleanup.c (try_forward_edges): Update.
11371 (outgoing_edges_match): Update.
11372 (try_crossjump_to_edge): Update.
11373 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
11374 (expand_gimple_tailcall): Update.
11375 (construct_init_block): Use make_single_succ_edge.
11376 (construct_exit_block): Use make_single_succ_edge.
11377 * cfghooks.c (verify_flow_info): Update.
11378 (redirect_edge_succ_nodup): Update.
11379 (split_edge): Update.
11380 (account_profile_record): Update.
11381 * cfgloopanal.c (single_likely_exit): Update.
11382 * cfgloopmanip.c (scale_loop_profile): Update.
11383 (set_zero_probability): Remove.
11384 (duplicate_loop_to_header_edge): Update.
11385 * cfgloopmanip.h (loop_version): Update prototype.
11386 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11387 (force_nonfallthru_and_redirect): Update.
11388 (update_br_prob_note): Update.
11389 (rtl_verify_edges): Update.
11390 (purge_dead_edges): Update.
11391 (rtl_lv_add_condition_to_bb): Update.
11392 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
11393 * cgraphunit.c (init_lowered_empty_function): Update.
11394 (cgraph_node::expand_thunk): Update.
11395 * cilk-common.c: Include profile-count.h
11396 * dojump.c (inv): Remove.
11397 (jumpifnot): Update.
11398 (jumpifnot_1): Update.
11399 (do_jump_1): Update.
11401 (do_jump_by_parts_greater_rtx): Update.
11402 (do_compare_rtx_and_jump): Update.
11403 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
11404 do_jump_1. do_compare_rtx_and_jump): Update prototype.
11405 * dwarf2cfi.c: Include profile-count.h
11406 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
11407 (sjlj_emit_dispatch_table): Likewise.
11408 * explow.c: Include profile-count.h
11409 * expmed.c (emit_store_flag_force): Update.
11410 (do_cmp_and_jump): Update.
11411 * expr.c (compare_by_pieces_d::generate): Update.
11412 (compare_by_pieces_d::finish_mode): Update.
11413 (emit_block_move_via_loop): Update.
11414 (store_expr_with_bounds): Update.
11415 (store_constructor): Update.
11416 (expand_expr_real_2): Update.
11417 (expand_expr_real_1): Update.
11418 * expr.h (try_casesi, try_tablejump): Update prototypes.
11419 * gimple-pretty-print.c (dump_probability): Update.
11420 (dump_profile): New.
11421 (dump_gimple_label): Update.
11422 (dump_gimple_bb_header): Update.
11423 * graph.c (draw_cfg_node_succ_edges): Update.
11424 * hsa-gen.c (convert_switch_statements): Update.
11425 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
11426 (find_if_case_1): Update.
11427 (find_if_case_2): Update.
11428 * internal-fn.c (expand_arith_overflow_result_store): Update.
11429 (expand_addsub_overflow): Update.
11430 (expand_neg_overflow): Update.
11431 (expand_mul_overflow): Update.
11432 (expand_vector_ubsan_overflow): Update.
11433 * ipa-cp.c (good_cloning_opportunity_p): Update.
11434 * ipa-split.c (split_function): Use make_single_succ_edge.
11435 * ipa-utils.c (ipa_merge_profiles): Update.
11436 * loop-doloop.c (add_test): Update.
11437 (doloop_modify): Update.
11438 * loop-unroll.c (compare_and_jump_seq): Update.
11439 (unroll_loop_runtime_iterations): Update.
11440 * lra-constraints.c (lra_inheritance): Update.
11441 * lto-streamer-in.c (input_cfg): Update.
11442 * lto-streamer-out.c (output_cfg): Update.
11443 * mcf.c (adjust_cfg_counts): Update.
11444 * modulo-sched.c (sms_schedule): Update.
11445 * omp-expand.c (expand_omp_for_init_counts): Update.
11446 (extract_omp_for_update_vars): Update.
11447 (expand_omp_ordered_sink): Update.
11448 (expand_omp_for_ordered_loops): Update.
11449 (expand_omp_for_generic): Update.
11450 (expand_omp_for_static_nochunk): Update.
11451 (expand_omp_for_static_chunk): Update.
11452 (expand_cilk_for): Update.
11453 (expand_omp_simd): Update.
11454 (expand_omp_taskloop_for_outer): Update.
11455 (expand_omp_taskloop_for_inner): Update.
11456 * omp-simd-clone.c (simd_clone_adjust): Update.
11457 * optabs.c (expand_doubleword_shift): Update.
11458 (expand_abs): Update.
11459 (emit_cmp_and_jump_insn_1): Update.
11460 (expand_compare_and_swap_loop): Update.
11461 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11462 * predict.c (predictable_edge_p): Update.
11463 (edge_probability_reliable_p): Update.
11464 (set_even_probabilities): Update.
11465 (combine_predictions_for_insn): Update.
11466 (combine_predictions_for_bb): Update.
11467 (propagate_freq): Update.
11468 (estimate_bb_frequencies): Update.
11469 (force_edge_cold): Update.
11470 * profile-count.c (profile_count::dump): Add missing space into dump.
11471 (profile_count::debug): Add newline.
11472 (profile_count::differs_from_p): Explicitly convert to unsigned.
11473 (profile_count::stream_in): Update.
11474 (profile_probability::dump): New member function.
11475 (profile_probability::debug): New member function.
11476 (profile_probability::differs_from_p): New member function.
11477 (profile_probability::differs_lot_from_p): New member function.
11478 (profile_probability::stream_in): New member function.
11479 (profile_probability::stream_out): New member function.
11480 * profile-count.h (profile_count_quality): Rename to ...
11481 (profile_quality): ... this one.
11482 (profile_probability): New.
11483 (profile_count): Update.
11484 * profile.c (compute_branch_probabilities): Update.
11485 * recog.c (peep2_attempt): Update.
11486 * sched-ebb.c (schedule_ebbs): Update.
11487 * sched-rgn.c (find_single_block_region): Update.
11488 (compute_dom_prob_ps): Update.
11489 (schedule_region): Update.
11490 * sel-sched-ir.c (compute_succs_info): Update.
11491 * stmt.c (struct case_node): Update.
11492 (do_jump_if_equal): Update.
11493 (get_outgoing_edge_probs): Update.
11494 (conditional_probability): Update.
11495 (emit_case_dispatch_table): Update.
11496 (expand_case): Update.
11497 (expand_sjlj_dispatch_table): Update.
11498 (emit_case_nodes): Update.
11499 * targhooks.c: Update.
11500 * tracer.c (better_p): Update.
11501 (find_best_successor): Update.
11502 * trans-mem.c (expand_transaction): Update.
11503 * tree-call-cdce.c: Update.
11504 * tree-cfg.c (gimple_split_edge): Upate.
11505 (move_sese_region_to_fn): Upate.
11506 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11507 * tree-eh.c (lower_resx): Upate.
11508 (cleanup_empty_eh_move_lp): Upate.
11509 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11510 * tree-inline.c (copy_edges_for_bb): Update.
11511 (copy_cfg_body): Update.
11512 * tree-parloops.c (gen_parallel_loop): Update.
11513 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11514 (gimple_gen_time_profiler): Update.
11515 * tree-ssa-dce.c (remove_dead_stmt): Update.
11516 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11517 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11518 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11519 (unloop_loops): Update.
11520 (try_peel_loop): Update.
11521 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11522 * tree-ssa-loop-split.c (connect_loops): Update.
11523 (split_loop): Update.
11524 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11525 (hoist_guard): Update.
11526 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11527 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11528 (value_replacement): Update.
11529 * tree-ssa-reassoc.c (branch_fixup): Update.
11530 * tree-ssa-tail-merge.c (replace_block_by): Update.
11531 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11532 (create_edge_and_update_destination_phis): Update.
11533 (compute_path_counts): Update.
11534 (recompute_probabilities): Update.
11535 (update_joiner_offpath_counts): Update.
11536 (freqs_to_counts_path): Update.
11537 (duplicate_thread_path): Update.
11538 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11539 (struct switch_conv_info): Update.
11540 (gen_inbound_check): Update.
11541 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11542 (vect_do_peeling): Update.
11543 (vect_loop_versioning): Update.
11544 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11545 (optimize_mask_stores): Update.
11546 * ubsan.c (ubsan_expand_null_ifn): Update.
11547 * value-prof.c (gimple_divmod_fixed_value): Update.
11548 (gimple_divmod_fixed_value_transform): Update.
11549 (gimple_mod_pow2): Update.
11550 (gimple_mod_pow2_value_transform): Update.
11551 (gimple_mod_subtract): Update.
11552 (gimple_mod_subtract_transform): Update.
11553 (gimple_ic): Update.
11554 (gimple_stringop_fixed_value): Update.
11555 (gimple_stringops_transform): Update.
11556 * value-prof.h: Update.
11558 2017-06-29 Carl Love <cel@us.ibm.com>
11560 * config/rs6000/rs6000-c.c: Add support for built-in functions
11561 vector signed int vec_signed (vector float);
11562 vector signed long long vec_signed (vector double);
11563 vector signed int vec_signed2 (vector double, vector double);
11564 vector signed int vec_signede (vector double);
11565 vector signed int vec_signedo (vector double);
11566 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11567 instruction generator.
11568 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11569 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11570 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11572 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11573 vunsignede_v2df): Add define_expands.
11574 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11575 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11576 VEC_UNSIGNEDO): Add definitions.
11577 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11578 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11579 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11580 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11581 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11582 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11583 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11584 vec_unsignede, vec_unsignedo): Add builtin defines.
11585 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11587 * doc/extend.texi: Update the built-in documentation file for the
11588 new built-in functions.
11590 2017-06-29 Richard Biener <rguenther@suse.de>
11592 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11593 reduction chains to LOOP_VINFO_REDUCTIONS.
11594 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11595 SLP reductions after processing reduction chains.
11597 2017-06-29 Nathan Sidwell <nathan@acm.org>
11599 * builtins.c (fold_builtin_FUNCTION): Use
11600 lang_hooks.decl_printable_name.
11602 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11604 PR middle-end/81194
11605 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11606 with only one label.
11607 * stmt.c (expand_case): Assert NCASES is greater than one.
11609 2017-06-29 Richard Biener <rguenther@suse.de>
11611 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11613 (group_case_labels): Likewise.
11614 (find_taken_edge): Push sanity checking on val to workers...
11615 (find_taken_edge_cond_expr): ... here
11616 (find_taken_edge_switch_expr): ... and here, handle cases
11617 with just a default label.
11618 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11619 (group_case_labels): Likewise.
11620 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11621 group_case_labels does anything cleanup the CFG again.
11623 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11625 PR tree-optimization/81196
11626 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11627 exit condition comparing two IVs.
11629 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11631 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11632 profile to the dummy entry at the end of the list of architectures.
11633 * config/arm/arm-cpu-cdata.h: Regenerated.
11635 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11636 Michael Collison <michael.collison@arm.com>
11639 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11641 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11642 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11643 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11644 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11645 of shift when the shift amount is masked with constant equal to
11646 the size of the mode.
11647 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11650 2017-06-29 Martin Liska <mliska@suse.cz>
11652 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11654 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11657 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11660 (edge_growth_cache_entry::edge_growth_cache_entry): New
11662 (reset_edge_growth_cache): Update to use constructor.
11664 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11666 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11667 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11668 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11670 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
11672 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
11673 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
11675 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
11677 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
11678 (*-linux-uclibc*): Add t-uclibc tmake_file.
11679 * config/t-musl: New.
11680 * config/t-uclibc: New.
11682 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11684 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11686 (gen_comm_data): Emit architectural setting of arch_prof.
11687 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11689 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11690 (armv8-m.base, armv8-m.main): Likewise.
11691 * arm-protos.h (arm_build_target): Add profile field.
11692 (arch_option): Likewise.
11693 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11695 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11696 arm_active_target.profile.
11698 2017-06-28 Richard Biener <rguenther@suse.de>
11700 PR middle-end/81227
11701 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11702 TYPE_OVERFLOW_WRAPS.
11703 * match.pd (negate_expr_p): Likewise.
11704 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11705 fold_build2, not fold_binary.
11707 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11709 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11710 Convert memory address to Pmode.
11711 (aarch64_print_operand): Assert MEM operands are always Pmode.
11713 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11716 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11717 Remove redundant if.
11718 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11719 * config/arm/aarch-common-protos.h
11720 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11721 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11723 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11726 * multiple_target.c (create_dispatcher_calls): Set the default
11727 clone to be static, not public.
11729 2017-06-28 Richard Biener <rguenther@suse.de>
11731 * tree-vect-loop.c (vectorizable_reduction): Move special
11732 cond reduction IV var creation ...
11733 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11734 parameter. Use STMT_VINFO_VECTYPE.
11735 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11738 2017-06-28 Martin Liska <mliska@suse.cz>
11741 * ipa-visibility.c (non_local_p): Handle visibility.
11743 2017-06-28 Martin Liska <mliska@suse.cz>
11746 * common.opt: Add IntegerRange to various options.
11747 * opt-functions.awk (integer_range_info): New function.
11748 * optc-gen.awk: Add integer_range_info to cl_options struct.
11749 * opts-common.c (decode_cmdline_option): Handle
11750 CL_ERR_INT_RANGE_ARG.
11751 (cmdline_handle_error): Likewise.
11752 * opts.c (print_filtered_help): Show valid interval in
11753 when --help is provided.
11754 * opts.h (struct cl_option): Add range_min and range_max fields.
11755 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11757 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11759 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11760 (x * C EQ/NE y * C): New transformation.
11762 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11764 * genmultilib (combination_space): Accept '+' in option names.
11766 2017-06-28 Martin Liska <mliska@suse.cz>
11769 * asan.c (instrument_derefs): Bail out inner references
11770 that are hard register variables.
11772 2017-06-28 Jakub Jelinek <jakub@redhat.com>
11775 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
11776 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
11778 2017-06-28 Richard Biener <rguenther@suse.de>
11780 * tree-vectorizer.h (vect_get_vec_defs): Remove.
11781 (vect_get_slp_defs): Adjust.
11782 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
11784 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
11786 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11787 get_initial_defs_for_reduction instead of vect_get_vec_defs.
11788 (vectorizable_reduction): Adjust.
11789 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
11791 (vect_get_slp_defs): Likewise.
11792 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
11793 (vectorizable_bswap): Adjust.
11794 (vectorizable_call): Likewise.
11795 (vectorizable_conversion): Likewise.
11796 (vectorizable_assignment): Likewise.
11797 (vectorizable_shift): Likewise.
11798 (vectorizable_operation): Likewise.
11799 (vectorizable_store): Likewise.
11800 (vectorizable_condition): Likewise.
11801 (vectorizable_comparison): Likewise.
11803 2017-06-28 Michael Collison <michael.collison@arm.com>
11806 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
11808 (arm_gen_movmemqi): Removed unused variable 'i'.
11809 Convert 'for' loop into 'while' loop.
11810 (arm_expand_prologue): Remove last unnecessary set of insn.
11811 (thumb_pop): Remove unused variable 'pushed_words'.
11812 (thumb_exit): Remove last unnecessary set of regs_to_pop.
11814 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11816 * config/s390/predicates.md: Use s390_rel_address_ok_p.
11817 * config/s390/s390-protos.h: Add prototype of
11818 s390_rel_address_ok_p.
11819 * config/s390/s390.c (s390_got_symbol): New function.
11820 (s390_rel_address_ok_p): New function.
11821 (legitimize_pic_address): Use s390_rel_address_ok_p.
11822 (s390_load_got): Use s390_got_symbol.
11823 (s390_option_override): Issue error if
11824 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
11825 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
11827 * config/s390/s390.opt: New option mpic-data-is-text-relative.
11829 2017-06-27 Andrew Pinski <apinski@cavium.com>
11831 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
11832 (X * copysign (1.0, X)): New pattern.
11833 (X * copysign (1.0, -X)): New pattern.
11834 (copysign (-1.0, CST)): New pattern.
11836 2017-06-27 Joseph Myers <joseph@codesourcery.com>
11838 * genmultilib (combination_space): Remove variable.
11839 Validate reuse rules against regular expression for any sequence
11840 of multilib options in any order.
11842 2017-06-27 Michael Collison <michael.collison@arm.com>
11844 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
11845 call aarch64_split_simd_combine.
11846 * (aarch64_combine_internal<mode>): Delete pattern.
11847 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
11848 Allow register and subreg operands.
11850 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11852 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
11853 specific need, just fallback on defaults.
11854 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
11856 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11857 Olivier Hainque <hainque@adacore.com>
11859 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
11861 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
11862 targets. Pick a default if no particular attempt applied.
11863 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
11866 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11868 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
11869 (x86_64-wrs-vxworks7): Likewise.
11871 2017-06-27 Marek Polacek <polacek@redhat.com>
11874 * ubsan.c (instrument_null): Check get_base_address's result for null.
11876 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11878 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
11880 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11882 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
11883 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
11884 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
11885 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
11886 New function types.
11887 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
11888 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
11889 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
11890 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
11891 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
11892 BUILT_IN_FEUPDATEENV): New builtins.
11893 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
11894 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
11895 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
11896 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
11898 (builtin_structptr_types): Adjust size.
11899 * tree.c (builtin_structptr_types): Add four entries.
11901 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11902 Olivier Hainque <hainque@adacore.com>
11904 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
11905 (TLS_SYM): New local macro, forcing reference to __tls__ on
11906 link command lines for VxWorks 7 RTPs, triggering initialization
11908 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
11909 OS features TLS support, true for RTPs on VxWorks 7.
11910 * config/vxworks.c (vxworks_override_options): Setup emutls
11913 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11915 * predict.c (test_prediction_value_range): Use -1U instead of -1
11916 to avoid narrowing conversion warning.
11917 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
11918 to avoid narrowing conversion warning.
11919 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
11921 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
11923 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11925 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
11926 64bit configurations.
11927 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
11928 (SIZE_TYPE): Likewise.
11929 * config/vxworks.c (vxworks_emutls_var_fields): Use
11930 long_unsigned_type_node instead of unsigned_type_node as the offset
11931 field type, which is "pointer" mode in emutls.c.
11933 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11936 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
11938 PR middle-end/81207
11939 * gimple-fold.c (replace_call_with_call_and_fold): Handle
11940 gimple_vuse copying separately from gimple_vdef copying.
11942 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11944 * value-prof.c (free_hist): Remove call to memset and the enclosing if
11947 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11948 Olivier Hainque <hainque@adacore.com>
11950 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
11951 for all vxworks7 targets.
11952 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
11953 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
11954 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
11955 variations for VX6/VX7 and 32/64bits later on in ...
11956 (VXWORKS_LIB_SPEC): Leverage new macros.
11957 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
11958 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
11960 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11962 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
11963 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
11965 2017-06-26 Carl Love <cel@us.ibm.com>
11967 * config/rs6000/rs6000-c.c: Add support for built-in functions
11968 vector bool char vec_reve (vector bool char);
11969 vector signed char vec_reve (vector signed char);
11970 vector unsigned char vec_reve (vector unsigned char);
11971 vector bool int vec_reve (vector bool int);
11972 vector signed int vec_reve (vector signed int);
11973 vector unsigned int vec_reve (vector unsigned int);
11974 vector bool long long vec_reve (vector bool long long);
11975 vector signed long long vec_reve (vector signed long long);
11976 vector unsigned long long vec_reve (vector unsigned long long);
11977 vector bool short vec_reve (vector bool short);
11978 vector signed short vec_reve (vector signed short);
11979 vector double vec_reve (vector double);
11980 vector float vec_reve (vector float);
11981 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
11982 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
11983 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
11984 (altivec_vreve): New pattern.
11985 * config/rs6000/altivec.h (vec_reve): New define.
11986 * doc/extend.texi (vec_rev): Update the built-in documentation file
11987 for the new built-in functions.
11989 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11991 PR tree-optimization/71815
11992 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
11994 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
11996 (slsr_process_phi): Likewise.
11997 (replace_uncond_cands_and_profitable_phis): Don't replace a
11998 multiply candidate with a stride of 1 (copy or cast).
11999 (phi_incr_cost): Call uses_consumed_by_stmt rather than
12001 (lowest_cost_path): Likewise.
12002 (total_savings): Likewise.
12004 2017-06-26 Richard Biener <rguenther@suse.de>
12007 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
12008 Use def_builtin_pure for all gather builtins.
12010 2017-06-26 Richard Biener <rguenther@suse.de>
12012 PR tree-optimization/81203
12013 * tree-tailcall.c (find_tail_calls): Do not move stmts into
12014 non-dominating BBs.
12016 2017-06-26 Marek Polacek <polacek@redhat.com>
12019 * doc/invoke.texi: Document -Wmultistatement-macros.
12021 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
12023 * doc/sourcebuild.texi (ARM-specific attributes): Document new
12024 arm_neon_ok_no_float_abi effective target.
12026 2017-06-26 Richard Biener <rguenther@suse.de>
12028 PR tree-optimization/80928
12029 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
12030 (copy_bbs): Set BB_DUPLICATED flag early.
12031 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
12033 (execute_on_shrinking_pred): Likewise.
12034 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
12035 BB_DUPLICATED blocks.
12036 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
12037 iterate over all PHIs considering removal of *gsi.
12039 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
12041 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
12044 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12046 * config/rs6000/rs6000-string.c: (expand_block_clear,
12047 do_load_for_compare, select_block_compare_mode,
12048 compute_current_alignment, expand_block_compare,
12049 expand_strncmp_align_check, expand_strn_compare,
12050 expand_block_move, rs6000_output_load_multiple)
12051 Move functions related to string/block move/compare
12052 to a separate file.
12053 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
12054 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
12055 for this function which is now used in two files.
12056 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
12057 * config.gcc: Add rs6000-string.o to extra_objs for
12058 targets powerpc*-*-* and rs6000*-*-*.
12060 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
12063 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
12064 32-bit, since indexed is not valid for DImode.
12065 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
12066 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
12067 (define_peephole2 for Altivec d-form load): Add 32-bit support.
12068 (define_peephole2 for Altivec d-form store): Likewise.
12071 * multiple_target.c (create_dispatcher_calls): Only create the
12072 dispatcher call if the function is the default clone of a
12073 versioned function.
12075 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
12077 PR middle-end/80902
12078 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
12079 a call, force the call to not be a tail call.
12081 2017-06-23 Jeff Law <law@redhat.com>
12083 * doc/contrib.texi: Add entry for Steven Pemberton's work on
12086 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
12088 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
12089 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
12090 handling for early expansion of vector shifts (sl,sr,sra,rl).
12091 (builtin_function_type): Add vector shift right instructions
12092 to the unsigned argument list.
12094 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12096 rtl-optimizatoin/79286
12097 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
12098 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
12099 trap. PIC register plus a const unspec without offset can never trap.
12101 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
12103 * tree.h (builtin_structptr_type): New type.
12104 (builtin_structptr_types): Declare new array.
12105 * tree.c (builtin_structptr_types): New array.
12106 (free_lang_data, build_common_tree_nodes): Use it.
12108 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
12111 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
12114 2017-06-22 Matt Turner <mattst88@gmail.com>
12116 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
12117 Lake models to skylake case. Assume skylake for unknown
12118 models with clflushopt.
12120 2017-06-22 Jeff Law <law@redhat.com>
12122 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
12123 frame sizes that do not satisfy aarch64_uimm12_shift.
12125 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
12127 * profile-count.h (apply_probability,
12128 apply_scale, probability_in): Fix checks for zero.
12130 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12132 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
12133 * doc/cppdiropts.texi (-I @var{dir}): Document it.
12135 2016-06-22 Richard Biener <rguenther@suse.de>
12137 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12138 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
12139 REDUC_MAX_EXPR support.
12140 (vectorizable_reduction): Likewise.
12141 (vect_create_epilog_for_reduction): Likewise.
12143 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
12145 * match.pd (A / (1 << B) -> A >> B): New.
12146 * generic-match-head.c: Include optabs-tree.h.
12147 * gimple-match-head.c: Likewise.
12148 * optabs-tree.h (target_supports_op_p): New.
12149 * optabs-tree.c (target_supports_op_p): New.
12151 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12153 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
12154 $gcc_cv_ld --help output.
12155 (gcc_cv_ld_demangle): Likewise.
12156 (gcc_cv_ld_eh_frame_hdr): Likewise.
12157 (gcc_cv_ld_pie): Likewise.
12158 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
12159 (gcc_cv_ld_buildid): Likewise.
12160 (gcc_cv_ld_sysroot): Likewise.
12161 (ld_bndplt_support): Likewise.
12162 (ld_pushpopstate_support): Likewise.
12163 * configure: Regenerate.
12164 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
12166 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12169 * config/i386/sse.md (round<mode>2): Renumber match_dup and
12170 operands indexes to avoid gap between operands and match_dups.
12172 2017-06-21 Andrew Pinski <apinski@cavium.com>
12174 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
12175 Increment Arith_shift and Arith_shift_reg by 1.
12176 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
12178 * config/aarch64/aarch64.c (thunderx_tunings): Enable
12179 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
12180 (aarch64_strip_extend): Add new argument and test for it.
12181 (aarch64_cheap_mult_shift_p): New function.
12182 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
12183 add a cost if it is true.
12184 Update calls to aarch64_strip_extend.
12185 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
12187 2017-06-21 Andrew Pinski <apinski@cavium.com>
12189 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
12191 (thunderxt88): Likewise.
12192 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
12193 (thunderx_prefetch_tune): New variable.
12194 (thunderx2t99_prefetch_tune): Update for the correct values.
12195 (thunderxt88_tunings): New variable.
12196 (thunderx_tunings): Use thunderx_prefetch_tune instead of
12197 generic_prefetch_tune.
12198 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
12200 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12202 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
12203 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
12204 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
12205 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
12206 (aarch64_atomic_cas<mode>, GPI): Likewise.
12208 2017-06-21 Martin Liska <mliska@suse.cz>
12210 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
12211 statements on cold and hot labels.
12212 * predict.c (tree_estimate_probability_bb): Remove the
12213 prediction from this place.
12215 2017-06-21 Martin Liska <mliska@suse.cz>
12217 PR tree-optimization/79489
12218 * gimplify.c (maybe_add_early_return_predict_stmt): New
12220 (gimplify_return_expr): Call the function.
12221 * predict.c (tree_estimate_probability_bb): Remove handling
12223 * predict.def: Update comment about early return predictor.
12224 * gimple-predict.h (is_gimple_predict): New function.
12225 * predict.def: Change default value of early return to 66.
12226 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
12228 * passes.def: Put pass_strip_predict_hints to the beginning of
12231 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
12233 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
12234 FUNCTION_DECL declarations.
12235 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
12237 (dwaf2out_decl): Likewise.
12238 * godump.c (go_early_global_decl): Skip call to the real debug hook
12239 for FUNCTION_DECL declarations.
12240 * passes.c (rest_of_decl_compilation): Skip call to the
12241 early_global_decl debug hook for FUNCTION_DECL declarations, unless
12242 -fdump-go-spec is passed.
12244 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12246 * config/i386/i386.c (struct builtin_isa): New field pure_p.
12247 Reorder for compactness.
12248 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
12249 (def_builtin_pure, def_builtin_pure2): New functions.
12250 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
12252 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12254 * match.pd (nop_convert): New predicate.
12255 ((A +- CST1) +- CST2): Allow some NOP conversions.
12257 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12260 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
12261 with ctors/dtors if GOVD_SHARED is set.
12263 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12265 * config/aarch64/aarch64.md (movti_aarch64):
12266 Emit mov rather than orr.
12267 (movtf_aarch64): Likewise.
12268 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
12269 Emit mov rather than orr.
12271 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12273 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
12274 Swap alternatives, make integer dup more expensive.
12276 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12278 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
12279 Return true for non-tls symbols.
12281 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
12283 * config/aarch64/aarch64-cores.def (cortex-a55): New.
12284 (cortex-a75): Likewise.
12285 (cortex-a75.cortex-a55): Likewise.
12286 * config/aarch64/aarch64-tune.md: Regenerate.
12287 * doc/invoke.texi (-mtune): Document new values for -mtune.
12289 2017-06-21 Tom de Vries <tom@codesourcery.com>
12291 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
12292 stack_size feature.
12293 (Effective-Target Keywords, Other attributes): Suggest using
12294 dg-add-options stack_size feature to get stack limit in stack_size
12295 effective target documentation.
12297 2017-06-21 Julian Brown <julian@codesourcery.com>
12298 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12300 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
12301 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
12302 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
12304 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
12305 attribute type list for neon_multiply.
12306 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
12307 attribute type list for neon_multiply.
12308 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
12309 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
12310 attribute type list for neon_multiply.
12311 * config/arm/types.md (crypto_pmull): Add.
12312 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
12313 attribute type list.
12315 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
12317 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
12320 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12322 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
12323 to make sure not to dereference a NULL cost_classes_ptr pointer.
12325 2017-06-20 Carl Love <cel@us.ibm.com>
12327 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12328 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12329 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12330 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12331 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12332 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12333 VMULOSW): New enum "unspec" values.
12334 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12335 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12336 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12337 altivec_vmulosw): New patterns.
12338 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12339 VMULOSW): Add definitions.
12341 2017-06-20 Julia Koval <julia.koval@intel.com>
12343 * config/i386/i386.c: Fix rounding expand for new pattern.
12344 * config/i386/subst.md: Fix pattern (parallel -> unspec).
12346 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12348 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
12349 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
12351 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12353 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
12356 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12358 * config/aarch64/aarch64-cores.def: Rearrange to sort by
12359 architecture, then by implementer ID.
12360 * config/aarch64/aarch64-tune.md: Regenerate.
12362 2017-06-20 Richard Biener <rguenther@suse.de>
12364 PR middle-end/81097
12365 * fold-const.c (split_tree): Fold to type before negating.
12367 2017-06-20 David Malcolm <dmalcolm@redhat.com>
12369 * diagnostic-show-locus.c
12370 (selftest::test_fixit_deletion_affecting_newline): New function.
12371 (selftest::diagnostic_show_locus_c_tests): Call it.
12373 2017-06-20 Andreas Schwab <schwab@suse.de>
12376 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
12379 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12381 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
12382 __ARM_FEATURE_COPROC according to support.
12384 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12386 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12387 Rewritten to avoid overflow for > 32-bit pointers.
12390 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
12391 by removing enum keyword.
12392 (ubsan_type_descriptor): Likewise. Formatting fix.
12395 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
12396 splitter): Require TARGET_SSE2 in the condition.
12398 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12401 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
12402 for doing vector set of SFmode on ISA 3.0.
12403 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
12404 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
12406 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
12407 SFmode value into a V4SF variable that was extracted from another
12408 V4SF variable without converting the element to double precision
12409 and back to single precision vector format.
12410 (vsx_insert_extract_v4sf_p9_2): Likewise.
12412 2017-06-19 Jakub Jelinek <jakub@redhat.com>
12414 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
12415 in UWHI to avoid undefined overflow.
12418 * ubsan.h (enum ubsan_encode_value_phase): New.
12419 (ubsan_encode_value): Change second argument to
12420 enum ubsan_encode_value_phase with default value of
12421 UBSAN_ENCODE_VALUE_GENERIC.
12422 * ubsan.c (ubsan_encode_value): Change second argument to
12423 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
12424 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
12425 create_tmp_var_raw instead of create_tmp_var and use a
12427 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12428 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12429 ubsan_encode_value callers.
12432 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12433 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12434 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12436 2017-06-19 Richard Biener <rguenther@suse.de>
12438 PR middle-end/81118
12439 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12440 estimates if we changed anything.
12442 2017-06-19 Richard Biener <rguenther@suse.de>
12444 PR tree-optimization/80887
12445 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12446 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12447 simplified lookups, then reset mprts_hook.
12448 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12450 (try_to_simplify): Likewise.
12452 2017-06-19 Martin Liska <mliska@suse.cz>
12455 * gimplify.c (gimplify_switch_expr):
12456 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12458 2017-06-19 Martin Liska <mliska@suse.cz>
12460 * doc/install.texi: Document that PGO runs in 4 stages.
12462 2017-06-19 Martin Liska <mliska@suse.cz>
12465 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12466 to dispatcher function name.
12467 * multiple_target.c (replace_function_decl): New function.
12468 (create_dispatcher_calls): Redirect both edges and references.
12470 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12472 * profile-count.c (profile_count::dump): Dump quality.
12473 (profile_count::differs_from_p): Update for unsigned val.
12474 * profile-count.h (profile_count_quality): New enum.
12475 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12477 2017-06-19 Richard Biener <rguenther@suse.de>
12479 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12480 struct function as arg.
12481 (estimate_numbers_of_iterations): Export overload with loop arg.
12482 (free_numbers_of_iterations_estimates_loop): Use an overload of
12483 free_numbers_of_iterations_estimates instead.
12484 * tree-cfg.c (remove_bb): Adjust.
12485 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12486 * tree-parloops.c (gen_parallel_loop): Likewise.
12487 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12489 (tree_unroll_loops_completely): Likewise.
12490 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12491 Use an overload instead and export.
12492 (estimated_loop_iterations): Adjust.
12493 (max_loop_iterations): Likewise.
12494 (likely_max_loop_iterations): Likewise.
12495 (estimate_numbers_of_iterations): Take struct function as arg
12497 (loop_exits_before_overflow): Adjust.
12498 (free_numbers_of_iterations_estimates_loop): Use an overload.
12499 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12500 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12502 2017-06-19 Richard Biener <rguenther@suse.de>
12505 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12506 RANGE_EXPR conservatively.
12508 2017-06-16 Carl Love <cel@us.ibm.com>
12510 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12511 definitions for vec_float, vec_float2, vec_floato,
12512 vec_floate built-ins.
12513 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12514 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12516 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12517 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12518 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12519 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12520 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12521 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12522 vec_floato): Add builtin defines.
12523 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12524 Update the built-in documentation file for the new built-in
12527 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12529 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12530 (mthumb): Mark as the negative of -marm.
12532 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12534 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12536 (ARM Options, -mtune): Document that this accepts the same
12537 extension options as -mcpu.
12538 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12540 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12542 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12543 permitted extensions.
12545 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12547 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12548 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12549 (armv8-m.main): Add option +nodsp.
12550 * config/arm/arm-cpu-cdata.h: Regenerated.
12552 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12554 * config/arm/t-fuchsia: New file.
12555 * config.gcc (arm*-*-fuchsia*): Use it.
12557 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12559 * config/arm/t-symbian: Rewrite for new option infrastructure.
12561 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12563 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12564 (MULTILIB_REQUIRED): Likewise.
12566 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12568 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12569 (MULTILIB_RESUE): Likewise.
12570 (MULTILIB_MATCHES): Likewise.
12571 (MULTLIB_REQUIRED): Likewise.
12573 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12575 * config/arm/t-rtems: Rewrite for new option framework.
12577 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12579 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12580 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12581 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12582 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12583 * config/arm/t-multilib: ... here.
12584 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12585 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12586 armv7-a and armv8*-a when A-profile libraries have not been built.
12587 * config/arm/t-rmprofile: Rewrite.
12589 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12591 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12592 with a backslash. Remove the backslash after substituting unescaped
12594 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12596 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12598 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12599 the driver to pass through the default setting of -mfloat-abi.
12600 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12602 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12603 (all_feat_combs): New rule.
12604 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12606 * config/arm/t-aprofile: Rewrite.
12608 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12610 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12611 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12612 fpu is not specified by the user/command-line.
12613 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12614 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12615 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12616 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12617 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12618 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12620 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12622 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12623 * config/arm/t-arm-elf: Rewritten.
12625 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12627 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12628 have some floating-point instructions.
12629 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12630 (TARGET_MAYBE_HARD_FLOAT): New macro.
12631 * config/arm/arm-builtins.c (arm_init_builtins): Use
12632 TARGET_MAYBE_HARD_FLOAT.
12633 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12635 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12637 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12638 (configargs.h): Include it.
12639 (arm_print_hint_for_fpu_option): New function.
12640 (arm_parse_fpu_option): New function.
12641 (candidate_extension): New class.
12642 (arm_canon_for_multilib): New function.
12643 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12644 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12645 (ARCH_CANONICAL_SPECS): New macro.
12646 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12648 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12650 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12651 are set after handling multilib fragments. Set target_cpu_default2
12654 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12656 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12658 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12659 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12661 (check_arch): Likewise.
12662 * config/arm/arm.c (arm_configure_build_target): Handle
12663 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12664 options in the default.
12666 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12668 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
12669 when an option is an alias of another.
12670 * config/arm/parsecpu.awk (optalias): New parser token.
12671 (gen_comm_data): Mark non-alias options as such. Emit entries
12672 for extension aliases.
12673 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
12674 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
12675 (armv6kz, armv6zk, armv6t2): Likewise.
12676 (armv7): Make vfpv3-d16 an alias.
12677 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
12679 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
12680 Sort in canonical order.
12681 (armv8-a): Sort in canonical order.
12682 (armv8.1-a, armv8.2-a): Likewise.
12683 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12685 (cortex-a9): Sort in canonical order.
12686 * config/arm/arm.c (selftests.h): Include it.
12687 (arm_test_cpu_arch_data): New function.
12688 (arm_run_self_tests): New function.
12689 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12690 (targetm): Move declaration to the end of the file.
12691 * arm-cpu-cdata.h: Regenerated.
12693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12695 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12696 call to target_mode_check describing the type of option passed.
12697 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12698 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12699 arm_parse_cpu_option_name to match parameters against list of
12701 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12702 arm_arch_core_flags data structure.
12703 * config/arm/arm-cpu_cdata.h: Regenerated.
12705 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12707 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12709 (arm_print_hint_for_cpu_option): Likewise.
12710 (arm_print_hint_for_arch_option): Likewise.
12711 (arm_parse_cpu_option_name): Likewise.
12712 (arm_parse_arch_option_name): Likewise.
12713 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12714 of entries in the all_fpus list.
12715 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12716 (arm_parse_cpu_option_name): Declare.
12717 (arm_parse_arch_option_name): Declare.
12718 (arm_parse_option_features): Declare.
12719 (arm_intialize_isa): Declare.
12720 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12722 (gen_comm_data): ... here. Make definitions non-static.
12723 * config/arm/arm-cpu-data.h: Regenerated.
12724 * config/arm/arm-cpu-cdata.h: Regenerated.
12726 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12728 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12729 (cpu_arch_extension): New structure.
12730 (cpu_arch_option, arch_option, cpu_option): New structures.
12731 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12732 architecture types.
12733 (gen_data): Generate new format data tables.
12734 * config/arm/arm.c (cpu_tune): New structure.
12735 (cpu_option, processors): Delete.
12736 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12737 (arm_print_hint_for_cpu_option): ... this and ...
12738 (arm_print_hint_for_arch_option): ... this.
12739 (arm_parse_arch_cpu_name): Delete. Replace with ...
12740 (arm_parse_cpu_option_name): ... this and ...
12741 (arm_parse_arch_option_name): ... this.
12742 (arm_unrecognized_feature): Change type of target parameter to
12744 (arm_parse_arch_cpu_features): Delete. Replace with ...
12745 (arm_parse_option_features): ... this.
12746 (arm_configure_build_target): Rework to use new configuration data
12748 (arm_print_tune_info): Rework for new configuration data tables.
12749 * config/arm/arm-cpu-data.h: Regenerated.
12750 * config/arm/arm-cpu.h: Regenerated.
12752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12754 * Makefile.in (OBJS): Move sbitmap.o from here ...
12755 (OBJS-libcommon): ... to here.
12757 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12759 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12760 (ISA_ALL_CRYPTO): New macro.
12761 (ISA_ALL_SIMD): New macro
12762 (ISA_ALL_FP): New macro.
12763 * config/arm/arm.c (fpu_bitlist): Update initializer.
12764 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12766 (arm9e): Add fpu. Add option for nofp
12767 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12768 (arm926ej-s, arm1026ej-s): Likewise.
12769 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12770 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
12771 neon-fp16, neon-vfpv4, nofp and nosimd.
12772 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
12773 (cortex-a8): Add fpu. Add option for nofp.
12774 (cortex-a9): Add fpu. Add options for nosimd and nofp.
12775 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
12776 (cortex-r4f): Add fpu.
12777 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
12778 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
12780 (cortex-r8): Likewise.
12781 (cortex-m4): Add fpu. Add option for nofp.
12782 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
12783 (cortex-a17.cortex-a7): Likewise.
12784 (cortex-a32): Add fpu. Add options for crypto and nofp.
12785 (cortex-a35, cortex-a53): Likewise.
12786 (cortex-a57): Add fpu. Add option for crypto.
12787 (cortex-a72, cortex-a73): Likewise.
12788 (exynos-m1): Likewise.
12789 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
12790 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
12791 (cortex-m33): Add fpu. Add option for nofp.
12792 * config/arm/arm-cpu-cdata.h: Regenerated
12793 * config/arm/arm-cpu-data.h: Regenerated.
12795 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12797 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
12798 (armv5te, armv5tej): Likewise.
12799 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
12800 (armv7): Add options fp and vfpv3-d16.
12801 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
12802 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
12804 (armv7ve): Likewise.
12805 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
12806 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
12807 (armv8-a): Add nocrypto option.
12808 (armv8.1-a, armv8.2-a): Likewise.
12809 (armv8-m.main): add options fp, fp.dp and nofp.
12811 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12813 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
12815 (armv8-a+crc): Delete.
12816 (armv8.1-a): Add options simd, crypto and nofp.
12817 (armv8.2-a): Add options fp16, simd, crypto and nofp.
12818 (armv8.2-a+fp16): Delete.
12819 (armv8-m.main): Add option dsp.
12820 (armv8-m.main+dsp): Delete.
12821 (cortex-a8): Add fpu. Add nofp option.
12822 (cortex-a9): Add fpu. Add nofp and nosimd options.
12823 * config/arm/parsecpu.awk (gen_data): Generate option tables and
12824 link to main cpu and architecture data structures.
12825 (gen_comm_data): Only put isa attributes from the main architecture
12827 (option): New statement for architecture and CPU entries.
12828 * arm.c (struct cpu_option): New structure.
12829 (struct processors): Add entry for options.
12830 (arm_unrecognized_feature): New function.
12831 (arm_parse_arch_cpu_name): Ignore any characters after the first
12833 (arm_parse_arch_cpu_feature): New function.
12834 (arm_configure_build_target): Separate out any CPU and architecture
12835 features and parse separately. Don't error out if -mfpu=auto is
12836 used with only an architecture string.
12837 (arm_print_asm_arch_directives): New function.
12838 (arm_file_start): Call it.
12839 * config/arm/arm-cpu-cdata.h: Regenerated.
12840 * config/arm/arm-cpu-data.h: Likewise.
12841 * config/arm/arm-tables.opt: Likewise.
12843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12845 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
12846 assembler when it is not -mfpu=auto.
12848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12850 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
12851 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
12852 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
12853 (ASM_CPU_SPEC): Rewrite.
12854 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
12855 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
12856 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
12857 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
12858 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
12859 copied string is NUL-terminated. Also strip any characters prefixed
12861 (arm_rewrite_selected_arch): New function.
12862 (arm_rewrite_march): New function.
12864 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12866 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
12867 (x_arm_cpu_string, x_arm_tune_string): Likewise.
12868 (march, mcpu, mtune): Convert to string-based options.
12869 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
12870 (arm_parse_arch_cpu_name): New function.
12871 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
12872 identify selected architecture or CPU.
12873 (arm_option_save): New function.
12874 (TARGET_OPTION_SAVE): Redefine.
12875 (arm_option_restore): Restore string options.
12876 (arm_option_print): Print string options.
12878 2017-06-16 Martin Sebor <msebor@redhat.com>
12880 PR tree-optimization/80933
12881 PR tree-optimization/80934
12882 * builtins.c (fold_builtin_3): Do not handle bcmp here.
12883 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
12884 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
12885 (gimple_fold_builtin): Call them.
12887 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12889 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
12890 as unlikely; update profile.
12892 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12894 * predict.c (force_edge_cold): Handle declaring edges impossible
12897 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12899 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
12901 (try_unroll_loop_completely): Fix reporting.
12903 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12905 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
12907 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
12910 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
12911 if given a non-constant argument for an intrinsic which requires a
12914 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12916 * profile.c (compare_freqs): New function.
12917 (branch_prob): Sort edge list.
12918 (find_spanning_tree): Assume that the list is priority sorted.
12920 2017-06-16 Richard Biener <rguenther@suse.de>
12922 PR tree-optimization/81090
12923 * passes.def (pass_record_bounds): Remove.
12924 * tree-pass.h (make_pass_record_bounds): Likewise.
12925 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
12926 make_pass_record_bounds): Likewise.
12927 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
12928 not free niter estimates at the beginning but at the end.
12929 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
12931 2017-06-16 Richard Biener <rguenther@suse.de>
12933 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
12934 initializer to workaround ICE in host GCC 4.8.
12936 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12938 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
12940 (clone_inlined_nodes): Update.
12942 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12944 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
12945 prefetch settings, and enable prefetching by default at -O3.
12947 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12949 * config/aarch64/aarch64.c (aarch64_override_options_internal):
12950 Set flag_prefetch_loop_arrays according to tuning data.
12952 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12954 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
12955 New tune structure.
12956 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
12957 [Unrelated to main purpose of the patch] Place the pointer field last
12958 to enable type checking errors when tune structure are wrongly merged.
12959 * config/aarch64/aarch64.c (generic_prefetch_tune,)
12960 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
12961 (thunderx2t99_prefetch_tune): New tune constants.
12962 (tune_params *_tunings): Update all tunings (no functional change).
12963 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
12964 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
12965 from tunings structures.
12967 2017-06-16 Jakub Jelinek <jakub@redhat.com>
12970 * ubsan.c (instrument_null): Add T argument, use it instead
12971 of computing it based on IS_LHS.
12972 (instrument_object_size): Likewise.
12973 (pass_ubsan::execute): Adjust instrument_null and
12974 instrument_object_size callers to pass gimple_get_lhs or
12975 gimple_assign_rhs1 result to it. Use instrument_null instead of
12976 calling get_base_address and instrument_mem_ref. Handle
12977 aggregate call arguments for object-size sanitization.
12979 2017-06-16 Yury Gribov <tetra2005@gmail.com>
12981 PR tree-optimization/81089
12982 * tree-vrp.c (is_masked_range_test): Validate operands of
12985 2017-06-15 Martin Sebor <msebor@redhat.com>
12988 * dumpfile.c (dump_register): Avoid calling memset to initialize
12989 a class with a default ctor.
12990 * gcc.c (struct compiler): Remove const qualification.
12991 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
12992 * hash-table.h: Ditto.
12993 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
12995 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
12996 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
12998 * params.h (struct param_info): Make struct members non-const.
12999 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
13000 with default initialization.
13001 * vec.h (vec_copy_construct, vec_default_construct): New helper
13003 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
13004 with vec_copy_construct.
13005 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
13006 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
13007 * doc/invoke.texi (-Wclass-memaccess): Document.
13009 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13011 * emit-rtl.h (is_leaf): Update comment about local
13012 register allocator.
13014 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13017 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
13018 for a variable to have a section before checking if the section has a
13020 Set section to.persistent if persistent attribute is set.
13021 Warn if .persistent attribute is used on an automatic variable.
13023 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
13025 PR rtl-optimization/80474
13026 * reorg.c (update_block): Do not ignore instructions in a delay slot.
13028 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
13030 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
13033 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
13035 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
13036 (casesi): Emit bounds checking as RTL.
13037 (casesi_internal_mips16_<mode>): Remove bounds checking.
13039 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
13041 * config/xtensa/xtensa.c (xtensa_option_override): Append
13042 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
13043 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
13044 xtensa_doloop_hooks): Define unconditionally.
13045 (xtensa_reorg_loops): Only call reorg_loops in the presence of
13047 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
13048 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
13049 for it in xtensa_option_override.
13050 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
13051 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
13053 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
13055 * doc/cppopts.texi: Document '-' special value to -MF.
13057 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
13059 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
13060 (cortex_a53_fconst): Likewise.
13061 (cortex_a53_fpmul): Likewise.
13062 (cortex_a53_f_load_64): Likewise.
13063 (cortex_a53_f_load_many): Likewise.
13064 (cortex_a53_advsimd_alu): Likewise.
13065 (cortex_a53_advsimd_alu_q): Likewise.
13066 (cortex_a53_advsimd_mul): Likewise.
13067 (cortex_a53_advsimd_mul_q): Likewise.
13068 (fpmac bypass): Add new bypass for fpmac-fpmac case.
13069 Add missing fmul, r2f_cvt and fconst cases.
13071 2017-06-14 Richard Biener <rguenther@suse.de>
13073 PR middle-end/81088
13074 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
13076 (fold_binary_loc): When associating do not treat pre-existing
13077 TREE_OVERFLOW on literal constants as a reason to allow
13078 TREE_OVERFLOW on associated literal constants.
13080 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13082 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
13083 (MASK_FEATURES): New macro.
13084 * config/sparc/sparc.c (sparc_option_override): Remove the special
13085 handling of -mfpu and generalize it to all MASK_FEATURES switches.
13087 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13089 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
13090 a division of 0 if non-call exceptions are enabled.
13092 2017-06-14 Andrew Pinski <apinski@cavium.com>
13093 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13096 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
13097 Improve vector initialization code gen for only variable case.
13099 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13101 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
13103 2017-06-14 Richard Biener <rguenther@suse.de>
13105 PR tree-optimization/81083
13106 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
13109 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13111 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
13112 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
13113 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
13114 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
13115 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
13116 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
13118 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13120 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
13121 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
13123 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13125 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
13127 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13129 * config/rs6000/t-rtems: Don't handle SPE.
13131 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13133 * config/rs6000/t-linux: Don't handle SPE.
13135 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13137 * config/rs6000/eabispe.h: Delete file.
13139 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13141 * config/rs6000/t-spe: Delete file.
13143 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13145 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
13146 (rs6000_legitimate_offset_address_p): Return false for anything in
13147 V2SImode or V2SFmode.
13149 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13151 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
13152 except V2SF and V2SI. Rearrange the vector modes, and add comments.
13153 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
13155 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
13156 (rs6000_legitimate_offset_address_p): Ditto.
13157 (rs6000_emit_move): Ditto.
13158 (rs6000_init_builtins): Remove V4HI_type_node.
13160 2017-06-13 Martin Liska <mliska@suse.cz>
13163 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
13164 (gate_asan): Likewise.
13165 * asan.h (asan_no_sanitize_address_p): Remove the function.
13166 (sanitize_flags_p): New function.
13167 * builtins.def: Fix coding style.
13168 * common.opt: Use renamed enum value.
13169 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
13170 * doc/extend.texi: Document no_sanitize attribute.
13171 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
13172 to SANITIZE_UNDEFINED_NONDEFAULT.
13173 * gcc.c (sanitize_spec_function): Use the renamed enum value.
13174 * gimple-fold.c (optimize_atomic_compare_exchange_p):
13175 Use sanitize_flags_p.
13176 * gimplify.c (gimplify_function_tree): Likewise.
13177 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
13178 * opts.c (parse_no_sanitize_attribute): New function.
13179 (common_handle_option): Use renamed enum value.
13180 * opts.h (parse_no_sanitize_attribute): Declare.
13181 * tree.c (sanitize_flags_p): New function.
13182 * tree.h: Declared here.
13183 * tsan.c: Use sanitize_flags_p.
13184 * ubsan.c (ubsan_expand_null_ifn): Likewise.
13185 (instrument_mem_ref): Likewise.
13186 (instrument_bool_enum_load): Likewise.
13187 (do_ubsan_in_current_function): Remove the function.
13188 (pass_ubsan::execute): Use sanitize_flags_p.
13189 * ubsan.h: Remove do_ubsan_in_current_function
13190 * tree-cfg.c (print_no_sanitize_attr_value): New function.
13191 (dump_function_to_file): Use it here.
13193 2017-06-13 Martin Jambor <mjambor@suse.cz>
13195 PR tree-optimization/80803
13196 PR tree-optimization/81063
13197 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
13198 (propagate_subaccesses_across_link): Enqueue subtree whenever
13199 necessary instead of relying on the caller.
13201 2017-06-13 Martin Jambor <mjambor@suse.cz>
13203 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
13204 that have a first_link.
13205 (sort_and_splice_var_accesses): Do not check first_link before
13207 (subtree_mark_written_and_enqueue): Likewise.
13208 (propagate_all_subaccesses): Likewise and do not stop at first
13209 parent with a first_link.
13211 2017-06-13 Martin Jambor <mjambor@suse.cz>
13213 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
13216 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13218 * match.pd: New pattern.
13220 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13222 * tree-vrp.c (is_masked_range_test): New function.
13223 (register_edge_assert_for): Determine ranges for
13226 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13228 PR tree-optimization/67328
13229 * fold-const.c (maskable_range_p): New function.
13230 (build_range_check): Generate bittests if possible.
13232 2017-06-13 Martin Liska <mliska@suse.cz>
13234 * gimple-pretty-print.c (dump_probability): Add new argument.
13235 (dump_edge_probability): Dump both probability and count.
13236 (dump_gimple_label): Likewise.
13237 (dump_gimple_bb_header): Likewise.
13239 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
13242 * config/avr/avr-devices.c: Fix indentation.
13243 * config/avr/gen-avr-mmcu-specs.c: Dito.
13245 2017-06-13 Richard Biener <rguenther@suse.de>
13247 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
13248 instead get vector type from stmt_info.
13249 (vectorizable_reduction): Adjust. Remove dead code.
13251 2017-06-13 Richard Biener <rguenther@suse.de>
13253 PR middle-end/81065
13254 * fold-const.c (extract_muldiv_1): Remove bogus distribution
13255 case of C * (x * C2 + C3).
13256 (fold_addr_of_array_ref_difference): Properly fold index difference.
13258 2017-06-12 David S. Miller <davem@davemloft.net>
13261 * config/sparc/sparc.md (return expander): Emit frame blockage if
13262 function uses alloca.
13264 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13266 * combine.c (make_field_assignment): Check len rather than the mode
13267 precision when calling force_to_mode.
13269 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
13271 Support multilibs and devices that see flash in RAM address range.
13274 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
13275 (avr_mcu_t) <flash_pm_offset>: New field.
13276 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
13277 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
13278 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
13279 (AVR_TINY_PM_OFFSET): Remove macro.
13280 * config/avr/avr.opt (-mshort-calls): New option.
13281 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13282 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
13283 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
13284 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
13285 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
13286 instead of avr_arch->have_jmp_call.
13287 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
13288 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
13289 avr_arch->flash_pm_offset to define.
13290 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
13291 new field flash_pm_offset. Add entry for avrxmega3.
13292 (avr_texinfo): Add entry for avrxmega3.
13293 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
13294 attiny212, attiny214,
13295 attiny412, attiny414, attiny416, attiny417,
13296 attiny814, attiny816, attiny817,
13297 attiny1614, attiny1616, attiny1617,
13298 attiny3214, attiny3216, attiny3217.
13299 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
13300 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
13301 (avr_print_operand_address) [AVR_TINY]: Same.
13302 (avr_asm_init_sections) <readonly_data_section>: Only patch
13303 callback if avr_arch->flash_pm_offset = 0.
13304 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
13305 for rodata if avr_arch->flash_pm_offset != 0.
13306 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
13307 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
13308 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
13309 (m_options): Append opt_rcall.
13310 (m_dirnames): Append dir_rcall.
13311 * config/avr/t-multilib: Regenerate.
13313 * configure.ac [target=avr]: Check whether avrxmega3 default
13314 linker description file works as needed.
13315 * configure: Regenerate.
13316 * doc/avr-mmcu.texi: Regenerate.
13317 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
13318 <__AVR_ARCH__>: Document avrxmega3 and 103.
13319 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
13320 <__AVR_SHORT_CALLS__>: Document it.
13321 <__AVR_PM_BASE_ADDRESS__>: Document it.
13322 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
13323 (AVR Variable Attributes) <progmem>: Document this is
13324 not needed for avrxmega3.
13325 (AVR Named Address Spaces) <__flash>: Dito.
13327 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
13329 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
13331 2017-06-12 Doug Rupp <rupp@adacore.com>
13333 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
13334 Append vxworks-stdint.h to the tm_file list.
13335 * config/vxworks-stdint.h: New file.
13337 2017-06-12 Martin Liska <mliska@suse.cz>
13339 PR tree-optimization/81041
13340 * tree-profile.c (gimple_gen_ic_func_profiler):
13341 Create an extra BB in profile-generate
13342 (gimple_gen_time_profiler): Likewise.
13344 2017-06-12 Jakub Jelinek <jakub@redhat.com>
13346 PR tree-optimization/81003
13347 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
13348 (update_range_test): Use it instead of force_gimple_operand_gsi.
13350 2017-06-12 Richard Biener <rguenther@suse.de>
13352 PR tree-optimization/81053
13353 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
13354 with backedge value not defined in loop. Simplify def stmt
13357 2017-06-11 Tom de Vries <tom@codesourcery.com>
13360 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
13362 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
13363 nvptx_cannot_force_const_mem.
13365 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13367 * opts.c (finish_options): Move test for flag_split_stack after
13368 it has been initialized.
13370 2017-06-11 Jason Merrill <jason@redhat.com>
13372 * tree.h (id_equal): New.
13373 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
13374 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
13375 instead of strcmp of IDENTIFIER_POINTER.
13377 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13379 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
13380 (mark_all_inlined_calls_cdtor): Fix formating.
13381 (inline_transform): Rescale profile before inlining.
13383 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13385 * cgraph.h (cgraph_edge::clone): Update prototype.
13386 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
13387 (cgraph_node::create_clone): Update.
13388 (cgraph_node::create_version_clone): Update.
13389 * tree-inline.c (copy_bb): Update.
13390 (expand_call_inline): Update.
13392 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13394 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
13395 factored out from ...
13396 (rs6000_emit_prologue): ... here.
13398 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13400 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
13401 factored out from ...
13402 (rs6000_emit_prologue): ... here.
13404 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13406 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
13408 (handle_missing_profiles): Fix computation of tp_first_run.
13409 (counts_to_freqs): Do not touch freqs when count is 0.
13411 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13413 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
13416 2017-06-10 Tom de Vries <tom@codesourcery.com>
13418 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13419 attributes): Document signal effective target.
13421 2017-06-10 Tom de Vries <tom@codesourcery.com>
13423 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13424 Document effective target stack_size.
13426 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13428 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13429 to the edit_context if they can be auto-applied.
13431 2017-06-9 Ian Lance Taylor <iant@golang.org>
13433 * opts.c (finish_options): If -fsplit-stack, disable implicit
13434 -forder-blocks-and-partition.
13435 * doc/invoke.texi (Optimize Options): Document that when using
13436 -fsplit-stack -forder-blocks-and-partition is not implicitly
13439 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13441 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13442 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13443 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13444 * builtins.def (abort, trap, unreachable): Declare cold.
13445 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13446 * tree-core.h (ECF_COLD): New.
13447 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13448 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13450 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13452 * predict.c (unlikely_executed_stmt_p): Cleanup.
13454 2017-06-09 Richard Biener <rguenther@suse.de>
13456 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13457 model if the ref is always written to.
13459 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13461 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13463 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13465 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13468 2017-06-09 Tom de Vries <tom@codesourcery.com>
13471 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13472 "target cannot support label values" when encountering LABEL_REF.
13474 2017-06-09 Martin Liska <mliska@suse.cz>
13476 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13477 (gimple_gen_ic_func_profiler): Emit direct comparison
13478 of __gcov_indirect_call_callee with NULL.
13479 (gimple_gen_time_profiler): Change probability from
13480 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13482 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13484 * profile.c (edge_gcov_counts): Turn to pointer.
13485 (compute_branch_probabilities, compute_branch_probabilities): Update.
13486 (branch_prob): Do not clear edge_gcov_count.
13487 * profile.h (edge_gcov_counts): Turn to pointer.
13488 (edge_gcov_count): Update.
13490 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13492 * gimple.h (gimple_check_failed): Mark cold.
13494 2017-06-09 Richard Biener <rguenther@suse.de>
13496 PR tree-optimization/66623
13497 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13498 refactor check_reduction into two parts, properly computing
13499 whether we have to check reduction validity for outer loop
13502 2017-06-09 Richard Biener <rguenther@suse.de>
13504 PR tree-optimization/79483
13505 * graphite-scop-detection.c (order): New global.
13506 (get_order): Compute bb to order mapping that satisfies code
13507 generation constraints.
13508 (cmp_pbbs): New helper.
13509 (build_scops): Start domwalk at entry block, sort generated
13512 2017-06-09 Richard Biener <rguenther@suse.de>
13514 PR middle-end/81007
13515 * ipa-polymorphic-call.c
13516 (ipa_polymorphic_call_context::restrict_to_inner_class):
13517 Skip FIELD_DECLs with error_mark_node type.
13518 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13521 2017-06-09 Martin Liska <mliska@suse.cz>
13523 * predict.c (struct branch_predictor): New struct.
13524 (test_prediction_value_range): New test.
13525 (predict_c_tests): New function.
13526 * selftest-run-tests.c (selftest::run_tests): Run the function.
13527 * selftest.h: Declare new tests.
13529 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13532 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13533 gen_add3_insn did not fail.
13534 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13535 r0, construct that number in a temporary reg and add that reg to r0.
13536 If asked to put the result in r0 as well, fail.
13538 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13540 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13541 for early expansion of vec_eqv.
13543 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13545 PR middle-end/81005
13546 * ubsan.c (instrument_null): Avoid pointless code temporary.
13547 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13549 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13553 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13556 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13559 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13561 * predict.c (unlikely_executed_edge_p): Move ahead.
13562 (probably_never_executed_edge_p): Use it.
13564 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13566 PR middle-end/79988
13567 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13568 gimple_call_builtin_p call.
13570 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13572 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13573 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13574 rtl_check_failed_type2, rtl_check_failed_code1,
13575 rtl_check_failed_code2, rtl_check_failed_code_mode,
13576 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13577 rtvec_check_failed_bounds, rtl_check_failed_flag,
13578 _fatal_insn_not_found, _fatal_insn): Likewise.
13579 * tree.h (tree_contains_struct_check_failed,
13580 tree_check_failed, tree_not_check_failed,
13581 tree_class_check_failed, tree_range_check_failed,
13582 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13583 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13584 tree_operand_check_failed, omp_clause_check_failed,
13585 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13588 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13590 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13591 flag_branch_probabilities.
13592 * ipa-inline.c (edge_badness): Likewise.
13593 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13594 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13595 * predict.c (maybe_hot_frequency_p): Likewise.
13596 (probably_never_executed): Likewise.
13597 * sched-ebb.c (schedule_ebbs): Likewise.
13598 * sched-rgn.c (find_single_block_region): Likewise.
13599 * tracer.c (tail_duplicate): Likewise.
13601 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13603 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13604 longer requires x_flag_profile_use.
13606 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13608 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13609 instead of flag_reorder_blocks_and_partition.
13610 * dbxout.c (dbxout_function_end): Likewise.
13611 * dwarf2out.c (gen_subprogram_die): Likewise.
13612 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13613 * hw-doloop.c (reorg_loops): Likewise.
13614 * varasm.c (assemble_start_function,
13615 assemble_end_function): Likewise.
13616 (decide_function_section): Do not check for
13617 flag_reorder_blocks_and_partition.
13619 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13621 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13623 (chkp_get_hard_register_fake_addr_expr): Ditto.
13624 (chkp_build_addr_expr): Add check for hard reg case.
13625 (chkp_parse_array_and_component_ref): Ditto.
13626 (chkp_find_bounds_1): Ditto.
13627 (chkp_process_stmt): Don't generate bounds store for
13630 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13632 * predict.c (maybe_hot_bb_p): Do not check profile status.
13633 (maybe_hot_edge_p): Likewise.
13634 (probably_never_executed): Check for zero counts even if profile
13636 (unlikely_executed_edge_p): New function.
13637 (unlikely_executed_stmt_p): New function.
13638 (unlikely_executed_bb_p): New function.
13639 (set_even_probabilities): Use unlikely predicates.
13640 (combine_predictions_for_bb): Likewise.
13641 (predict_paths_for_bb): Likewise.
13642 (predict_paths_leading_to_edge): Likewise.
13643 (determine_unlikely_bbs): New function.
13644 (estimate_bb_frequencies): Use it.
13645 (compute_function_frequency): Use zero counts even if profile is
13647 * profile-count.h: Fix typo.
13649 2017-08-08 Julia Koval <julia.koval@intel.com>
13651 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13652 _mm512_mask_cvtsepi16_storeu_epi8,
13653 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13654 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13655 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13656 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13657 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13658 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13659 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13660 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13661 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13662 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13663 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13664 __builtin_ia32_pmovuswb256mem_mask,
13665 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13666 __builtin_ia32_pmovwb512mem_mask): New builtins.
13668 2017-08-08 Julia Koval <julia.koval@intel.com>
13670 PR target/73350,80862
13671 * config/i386/subst.md (round): Fix round pattern.
13672 * config/i386/i386.c (ix86_erase_embedded_rounding):
13673 Fix erasing rounding for the fixed pattern.
13675 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13677 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
13679 2017-06-08 Martin Liska <mliska@suse.cz>
13681 PR gcov-profile/80911
13682 * gcov.c (block_info::block_info): New constructor.
13684 2017-06-07 Carl Love <cel@us.ibm.com>
13686 * config/rs6000/rs6000-c: The return type of the following
13687 built-in functions was implemented as int not long long. Fix sign
13688 of return value for the unsigned version of vec_mulo and vec_mule.
13689 vector unsigned long long vec_bperm (vector unsigned long long,
13690 vector unsigned char)
13691 vector signed long long vec_mule (vector signed int,
13693 vector unsigned long long vec_mule (vector unsigned int,
13694 vector unsigned int)
13695 vector signed long long vec_mulo (vector signed int,
13697 vector unsigned long long vec_mulo (vector unsigned int,
13698 vector unsigned int)
13699 * doc/extend.texi: Fix the documentation for the built-in
13702 2017-06-07 Carl Love <cel@us.ibm.com>
13705 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13708 2017-06-07 Carl Love <cel@us.ibm.com>
13710 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13711 support, Generate doublehv for signed int/float for BE case only.
13713 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13715 * doc/invoke.texi (mcx16): Rewrite.
13717 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13719 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13720 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13721 *mov<mode>_softfloat, and an anonymous splitter): Use
13722 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13724 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13726 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13728 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13729 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13730 (rs6000_debug_reg_global): Adjust.
13731 (rs6000_init_hard_regno_mode_ok): Adjust.
13732 (rs6000_dbx_register_number): Adjust.
13733 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13734 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13735 Remove SPE_ACC and SPEFSCR.
13736 (REG_ALLOC_ORDER): Ditto.
13737 (FRAME_POINTER_REGNUM): Change to 111.
13738 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13739 (REG_CLASS_NAMES): Ditto.
13740 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13741 (REGISTER_NAMES): Ditto.
13742 (ADDITIONAL_REG_NAMES): Ditto.
13743 (rs6000_reg_names): Ditto.
13744 * config/rs6000/rs6000.md: Renumber some register number
13747 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13749 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13751 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13752 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13754 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13755 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13756 Delete the SPE high registers.
13757 (REG_ALLOC_ORDER): Ditto.
13758 (enum reg_class): Remove SPE_HIGH_REGS.
13759 (REG_CLASS_NAMES): Ditto.
13760 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13761 (REGISTER_NAMES): Ditto.
13762 (rs6000_reg_names): Ditto.
13763 * doc/tm.texi.in: Remove SPE as example.
13764 * doc/tm.texi: Regenerate.
13766 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13768 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13769 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13770 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
13771 * config/rs6000/rs6000.md (type): Remove "brinc".
13773 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13775 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
13776 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
13777 * config/rs6000/linuxspe.h: Delete file.
13778 * config/rs6000/rs6000.md: Don't include spe.md.
13779 * config/rs6000/spe.h: Delete file.
13780 * config/rs6000/spe.md: Delete file.
13781 * config/rs6000/t-rs6000: Remove spe.md.
13783 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13785 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
13786 (reg_or_none500mem_operand): Delete.
13787 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
13788 instead of reg_or_none500mem_operand.
13790 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13792 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13793 handling of SPE flags.
13794 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
13796 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13798 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13800 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
13801 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
13802 paired_divv2sf3): Similar.
13803 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
13804 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
13805 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
13807 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
13808 Rename the paired_* instruction patterns.
13809 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
13811 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
13812 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
13813 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
13814 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
13815 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
13816 PAIRED_VECTOR_MODE.
13817 (struct machine_function): Delete field spe_insn_chain_scanned_p.
13818 (spe_func_has_64bit_regs_p): Delete.
13819 (spe_expand_predicate_builtin): Delete.
13820 (spe_expand_evsel_builtin): Delete.
13821 (TARGET_DWARF_REGISTER_SPAN): Do not define.
13822 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
13823 (invalid_e500_subreg): Delete.
13824 (rs6000_legitimize_address): Always force_reg op2 as well, for
13825 paired single memory accesses.
13826 (rs6000_member_type_forces_blk): Delete.
13827 (rs6000_spe_function_arg): Delete.
13828 (rs6000_expand_unop_builtin): Delete SPE handling.
13829 (rs6000_expand_binop_builtin): Ditto.
13830 (spe_expand_stv_builtin): Delete.
13831 (bdesc_2arg_spe): Delete.
13832 (spe_expand_builtin): Delete.
13833 (spe_expand_predicate_builtin): Delete.
13834 (spe_expand_evsel_builtin): Delete.
13835 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
13836 (spe_init_builtins): Delete.
13837 (spe_func_has_64bit_regs_p): Delete.
13838 (savres_routine_name): Delete "info" parameter. Adjust callers.
13839 (rs6000_emit_stack_reset): Ditto.
13840 (rs6000_dwarf_register_span): Delete.
13841 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
13842 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
13843 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
13845 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
13847 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
13848 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
13849 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
13850 mulv2sf3, divv2sf3): Delete expanders.
13852 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13854 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
13856 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13858 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
13859 * config/rs6000/rs6000.c: Ditto.
13861 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13863 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
13864 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
13865 comparison_operator.
13867 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13869 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
13870 * config/rs6000/rs6000.opt: Ditto.
13871 * config/rs6000/t-rtems: Ditto.
13873 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13875 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
13876 TARGET_E500_SINGLE by 0, simplify.
13877 * config/rs6000/rs6000.c: Ditto.
13878 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
13879 (spe_build_register_parallel): Delete.
13880 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
13881 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
13882 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
13883 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
13884 (E500_CONVERT): Delete.
13885 * config/rs6000/spe.md: Remove many patterns and all define_constants.
13887 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13889 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
13890 * config/rs6000/dfp.md: Ditto.
13891 (negdd2, *negdd2_fpr): Merge.
13892 (absdd2, *absdd2_fpr): Merge.
13893 (negtd2, *negtd2_fpr): Merge.
13894 (abstd2, *abstd2_fpr): Merge.
13895 * config/rs6000/e500.h: Delete file.
13896 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
13897 TARGET_FPRS by 1 and simplify.
13898 * config/rs6000/rs6000-c.c: Ditto.
13899 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
13900 TARGET_DF_SPE by 0.
13901 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
13903 * config/rs6000/rs6000.md: Ditto.
13904 (floatdidf2, *floatdidf2_fpr): Merge.
13905 (move_from_CR_gt_bit): Delete.
13906 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
13907 (E500_CR_IOR_COMPARE): Delete.
13908 (All patterns that require !TARGET_FPRS): Delete.
13909 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
13911 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13913 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
13915 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13917 * graphds.c (add_edge): Intitialize edge's attached data.
13918 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
13919 pointer parameter. Call pointed function on each edge during
13920 graph traversing. Skip traversing the edge when the function
13922 (graphds_dfs, graphds_scc): Ditto.
13923 (for_each_edge): New parameter. Pass the new parameter to callback
13925 * graphds.h (skip_edge_callback): New function pointer type.
13926 (graphds_dfs, graphds_scc): New function pointer parameter.
13927 (graphds_edge_callback, for_each_edge): New parameter.
13929 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13931 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
13932 out code checking if runtime alias check is possible to below ...
13933 Call the new function.
13934 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
13935 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
13937 2017-06-07 Marek Polacek <polacek@redhat.com>
13940 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
13941 TYPE_OVERFLOW_WRAPS check.
13943 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13945 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
13946 if versioning is required.
13947 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
13948 peeling with the check for versioning.
13950 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13952 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
13953 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
13954 Set true to new parameter if new ssa variable is defined.
13955 (vect_gen_vector_loop_niters): Refactor. Set range information
13956 for the new vector loop bound variable.
13957 (vect_do_peeling): Ditto.
13959 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13961 * tree-affine.c (ssa.h): Include header file.
13962 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
13963 has wrapping overflow behavior.
13965 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13967 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
13969 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13971 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
13972 (tree_to_aff_combination): ... here.
13974 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13976 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
13977 reg_pressure model function.
13978 (ivopts_global_cost_for_size): Delete.
13979 (determine_set_costs, iv_ca_recount_cost): Call new model function
13980 ivopts_estimate_reg_pressure.
13982 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13984 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
13985 expensive than udiv. Remove floating point cases from mod.
13987 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13989 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
13990 Increase idiv cost.
13992 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13994 * config/aarch64/aarch64.md
13995 (copysignsf3): Fix mask generation.
13997 2017-06-07 Jakub Jelinek <jakub@redhat.com>
13999 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
14001 (class dump_manager): Add register_dumps method.
14002 * dumpfile.c: Include langhooks.h.
14003 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
14004 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
14005 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
14006 (dump_manager::dump_register): Start with 512 entries instead of 32.
14007 (dump_manager::register_dumps): New method.
14008 * toplev.c (general_init): Instead of invoking register_dumps
14009 langhook, invoke register_dumps method on the dump manager.
14010 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
14013 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
14015 * doc/md.texi: Clarify the restrictions on a define_insn condition.
14016 Say that # requires an associated define_split to exist, and that
14017 the define_split must be suitable for use after register allocation.
14019 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14021 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
14022 (compute_outgoing_frequencies): Also initialize zero counts.
14023 (find_many_sub_basic_blocks): Do not produce uninitialized profile
14024 around loops; preserve more of profile when nothing changes.
14026 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
14028 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
14030 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
14031 * config/arm/arm-cpu-cdata.h: Regenerate.
14032 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
14033 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
14034 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
14035 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
14036 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
14038 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
14040 * config/arm/t-rmprofile: Likewise.
14041 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
14043 2017-06-06 David S. Miller <davem@davemloft.net>
14046 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
14047 blockage if function uses alloca.
14049 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14051 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
14052 New "uid" fields to hold pretty-print IDs of group and ref.
14053 Memory references are now identified as <group_id>:<ref_id>
14054 instead of using [random] addresses.
14055 (dump_mem_details): Simplify, no functional change.
14056 (dump_mem_ref): Simplify and make output more concise.
14057 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
14058 (find_or_create_group): Initialize group uid.
14059 (record_ref): Initialize ref uid. Improve debug output.
14060 (prune_group_by_reuse, should_issue_prefetch_p,)
14061 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
14062 (mark_nontemporal_store, determine_loop_nest_reuse):
14063 Improve debug output.
14065 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14067 * dbgcnt.def (prefetch): New debug counter.
14068 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
14069 (schedule_prefetches): Stop issueing prefetches if debug counter
14072 2017-06-06 Tom de Vries <tom@codesourcery.com>
14074 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
14075 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
14077 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14079 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
14080 Use aarch64_reg_or_zero predicate for operand 4.
14081 (aarch64_compare_and_swap<mode> define_insn_and_split):
14082 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
14083 (aarch64_store_exclusive<mode>): Likewise for operand 2.
14085 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14087 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
14088 (arm_compute_save_core_reg_mask): This.
14089 (thumb1_compute_save_reg_mask): Rename into ...
14090 (thumb1_compute_save_core_reg_mask): This.
14091 (arm_compute_save_reg0_reg12_mask): Adapt comment.
14092 (arm_compute_frame_layout): Likewise.
14094 2017-06-06 Richard Biener <rguenther@suse.de>
14096 PR tree-optimization/80974
14097 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
14098 keep or clear leaders SSA info.
14100 2017-06-06 Tom de Vries <tom@codesourcery.com>
14102 * config/nvptx/nvptx.c (split_mode_p): New function.
14103 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
14105 2017-06-06 Tom de Vries <tom@codesourcery.com>
14107 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
14109 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14112 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
14115 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14117 * shrink-wrap.c (handle_simple_exit): Update profile.
14118 (try_shrink_wrapping): Upate profile.
14120 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14122 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
14123 (tree_guess_outgoing_edge_probabilities): New.
14124 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
14125 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
14127 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14129 * ipa-split.c (split_function): Initialize return bb profile.
14131 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14133 * profile.c (compute_branch_probabilities): Also initialize
14134 EXIT_BLOCK profile.
14136 2017-06-06 Richard Biener <rguenther@suse.de>
14138 PR tree-optimization/80928
14139 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
14140 (vect_analyze_loop_operations): Properly guard analysis for
14142 (vect_transform_loop): Likewise.
14143 (vect_analyze_loop_2): Also reset SLP type on PHIs.
14144 (vect_model_induction_cost): Do not cost for pure SLP.
14145 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
14146 of induction in inner loop vectorization.
14147 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
14148 (vect_get_and_check_slp_defs): Handle vect_induction_def.
14149 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
14151 (vect_analyze_slp_cost_1): Cost induction.
14152 (vect_detect_hybrid_slp_stmts): Handle PHIs.
14153 (vect_get_slp_vect_defs): Likewise.
14154 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
14155 (vect_transform_stmt): Handle SLP reductions.
14156 * tree-vectorizer.h (vectorizable_induction): Adjust.
14158 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14160 * config/rs6000/rs6000.c (make_resolver_func): Update
14161 init_lowered_empty_function call.
14163 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
14165 * doc/invoke.texi: Document the -fprofile-abs-path option.
14166 * common.opt (fprofile-abs-path): New option.
14167 * gcov-io.h (gcov_write_filename): Declare.
14168 * gcov-io.c (gcov_write_filename): New function.
14169 * coverage.c (coverage_begin_function): Use gcov_write_filename.
14170 * profile.c (output_location): Likewise.
14172 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14174 * shring-wrap.c: Revert accidental commit.
14176 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
14178 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
14180 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14182 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
14184 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
14185 profile in callgraph edge.
14186 * profile-count.h (apply_probability): If THIS is 0, then result is 0
14187 (apply_scale): Likewise.
14188 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
14189 Also scale profile when inlining function with zero profile.
14190 (initialize_cfun): Update exit block profile even when it is zero.
14191 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
14192 when profile is read.
14194 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14196 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
14197 (CLONE_*): New constants to define the processors we can generate
14198 code for with the target_clone attribute.
14199 (rs6000_clone_map): New array to identify which clone processors
14200 the current program is running on.
14201 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
14202 target_clone attribute.
14203 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
14204 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
14205 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
14206 (cpu_expand_builtin): Add support for target_clone attribute.
14207 (rs6000_valid_attribute_p): Allow "default" attribute.
14208 (get_decl_name): New debug function to simplify printing the
14209 current function name in debugging statements.
14210 (rs6000_clone_priority): New functions to support the target_clone
14211 attribute, and be able to generate code to switch between ISA 2.05
14212 through ISA 3.0 (power6 through power9).
14213 (rs6000_compare_version_priority): Likewise.
14214 (rs6000_get_function_versions_dispatcher): Likewise.
14215 (make_resolver_func): Likewise.
14216 (add_condition_to_bb): Likewise.
14217 (dispatch_function_versions): Likewise.
14218 (rs6000_generate_version_dispatcher_body): Likewise.
14219 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
14220 (fusion_gpr_load_p): Fix a spacing issue.
14221 * doc/extend.texi (Common Function Attributes): Document that the
14222 PowerPC supports the target_clone attribute.
14224 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14226 * config/arm/arm.h: explain F symbol found in description of ARM
14227 register allocation in its legend.
14229 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14231 * config/mips/frame-header-opt.c: Include profile-count.h.
14232 * config/riscv/riscv.c: Include profile-count.h
14234 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14236 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
14238 (sm_set_flag_if_changed): Add bbs field.
14239 (execute_sm_if_changed_flag_set): Pass BBS.
14240 (execute_sm): Update.
14242 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14244 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
14247 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14249 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
14250 (peephole2): New peephole2 to emit the above.
14251 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
14253 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14255 * config/aarch64/aarch64.c (define_peephole2 above
14256 *sub_<shift>_<mode>): New peephole.
14258 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14260 * config/i386/i386.c (make_resolver_func): Update.
14261 * Makefile.in: Add profile-count.h and profile-count.o
14262 * auto-profile.c (afdo_indirect_call): Update to new API.
14263 (afdo_set_bb_count): Update.
14264 (afdo_propagate_edge): Update.
14265 (afdo_propagate_circuit): Update.
14266 (afdo_calculate_branch_prob): Update.
14267 (afdo_annotate_cfg): Update.
14268 * basic-block.h: Include profile-count.h
14269 (struct edge_def): Turn count to profile_count.
14270 (struct basic_block_def): Likewie.
14271 (REG_BR_PROB_BASE): Move to profile-count.h
14272 (RDIV): Move to profile-count.h
14273 * bb-reorder.c (max_entry_count): Turn to profile_count.
14274 (find_traces): Update.
14275 (rotate_loop):Update.
14276 (connect_traces):Update.
14277 (sanitize_hot_paths):Update.
14278 * bt-load.c (migrate_btr_defs): Update.
14279 * cfg.c (RDIV): Remove.
14280 (init_flow): Use alloc_block.
14281 (alloc_block): Uninitialize count.
14282 (unchecked_make_edge): Uninitialize count.
14283 (check_bb_profile): Update.
14284 (dump_edge_info): Update.
14285 (dump_bb_info): Update.
14286 (update_bb_profile_for_threading): Update.
14287 (scale_bbs_frequencies_int): Update.
14288 (scale_bbs_frequencies_gcov_type): Update.
14289 (scale_bbs_frequencies_profile_count): New.
14290 * cfg.h (update_bb_profile_for_threading): Update.
14291 (scale_bbs_frequencies_profile_count): Declare.
14292 * cfgbuild.c (compute_outgoing_frequencies): Update.
14293 (find_many_sub_basic_blocks): Update.
14294 * cfgcleanup.c (try_forward_edges): Update.
14295 (try_crossjump_to_edge): Update.
14296 * cfgexpand.c (expand_gimple_tailcall): Update.
14297 (construct_exit_block): Update.
14298 * cfghooks.c (verify_flow_info): Update.
14299 (dump_bb_for_graph): Update.
14300 (split_edge): Update.
14301 (make_forwarder_block): Update.
14302 (duplicate_block): Update.
14303 (account_profile_record): Update.
14304 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
14305 (get_estimated_loop_iterations): Update.
14306 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
14307 (single_likely_exit): Update.
14308 * cfgloopmanip.c (scale_loop_profile): Update.
14310 (set_zero_probability): Update.
14311 (lv_adjust_loop_entry_edge): Update.
14312 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14313 (purge_dead_edges): Update.
14314 (rtl_account_profile_record): Update.
14315 * cgraph.c (cgraph_node::create): Uninitialize count.
14316 (symbol_table::create_edge): Uninitialize count.
14317 (cgraph_update_edges_for_call_stmt_node): Update.
14318 (cgraph_edge::dump_edge_flags): Update.
14319 (cgraph_node::dump): Update.
14320 (cgraph_edge::maybe_hot_p): Update.
14321 * cgraph.h: Include profile-count.h
14322 (create_clone), create_edge, create_indirect_edge): Update.
14323 (cgraph_node): Turn count to profile_count.
14324 (cgraph_edge0: Likewise.
14325 (make_speculative, clone): Update.
14326 (create_edge): Update.
14327 (init_lowered_empty_function): Update.
14328 * cgraphclones.c (cgraph_edge::clone): Update.
14329 (duplicate_thunk_for_node): Update.
14330 (cgraph_node::create_clone): Update.
14331 * cgraphunit.c (cgraph_node::analyze): Update.
14332 (cgraph_node::expand_thunk): Update.
14333 * final.c (dump_basic_block_info): Update.
14334 * gimple-streamer-in.c (input_bb): Update.
14335 * gimple-streamer-out.c (output_bb): Update.
14336 * graphite.c (print_global_statistics): Update.
14337 (print_graphite_scop_statistics): Update.
14338 * hsa-brig.c: Include basic-block.h.
14339 * hsa-dump.c: Include basic-block.h.
14340 * hsa-gen.c (T sum_slice): Update.
14341 (convert_switch_statements):Update.
14342 * hsa-regalloc.c: Include basic-block.h.
14343 * ipa-chkp.c (chkp_produce_thunks): Update.
14344 * ipa-cp.c (struct caller_statistics): Update.
14345 (init_caller_stats): Update.
14346 (gather_caller_stats): Update.
14347 (ipcp_cloning_candidate_p): Update.
14348 (good_cloning_opportunity_p): Update.
14349 (get_info_about_necessary_edges): Update.
14350 (dump_profile_updates): Update.
14351 (update_profiling_info): Update.
14352 (update_specialized_profile): Update.
14353 (perhaps_add_new_callers): Update.
14354 (decide_about_value): Update.
14355 (ipa_cp_c_finalize): Update.
14356 * ipa-devirt.c (struct odr_type_warn_count): Update.
14357 (struct decl_warn_count): Update.
14358 (struct final_warning_record): Update.
14359 (possible_polymorphic_call_targets): Update.
14360 (ipa_devirt): Update.
14361 * ipa-fnsummary.c (redirect_to_unreachable): Update.
14362 * ipa-icf.c (sem_function::merge): Update.
14363 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
14364 * ipa-inline.c (compute_uninlined_call_time): Update.
14365 (compute_inlined_call_time): Update.
14366 (want_inline_small_function_p): Update.
14367 (want_inline_self_recursive_call_p): Update.
14368 (edge_badness): Update.
14369 (lookup_recursive_calls): Update.
14370 (recursive_inlining): Update.
14371 (inline_small_functions): Update.
14372 (dump_overall_stats): Update.
14373 (dump_inline_stats): Update.
14374 * ipa-profile.c (ipa_profile_generate_summary): Update.
14375 (ipa_propagate_frequency): Update.
14376 (ipa_profile): Update.
14377 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14378 * ipa-utils.c (ipa_merge_profiles): Update.
14379 * loop-doloop.c (doloop_modify): Update.
14380 * loop-unroll.c (report_unroll): Update.
14381 (unroll_loop_runtime_iterations): Update.
14382 * lto-cgraph.c (lto_output_edge): Update.
14383 (lto_output_node): Update.
14384 (input_node): Update.
14385 (input_edge): Update.
14386 (merge_profile_summaries): Update.
14387 * lto-streamer-in.c (input_cfg): Update.
14388 * lto-streamer-out.c (output_cfg): Update.
14389 * mcf.c (create_fixup_graph): Update.
14390 (adjust_cfg_counts): Update.
14391 (sum_edge_counts): Update.
14392 * modulo-sched.c (sms_schedule): Update.
14393 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
14394 * predict.c (maybe_hot_count_p): Update.
14395 (probably_never_executed): Update.
14396 (dump_prediction): Update.
14397 (combine_predictions_for_bb): Update.
14398 (propagate_freq): Update.
14399 (handle_missing_profiles): Update.
14400 (counts_to_freqs): Update.
14401 (rebuild_frequencies): Update.
14402 (force_edge_cold): Update.
14403 * predict.h: Include profile-count.h
14404 (maybe_hot_count_p, counts_to_freqs): UPdate.
14405 * print-rtl-function.c: Do not include cfg.h
14406 * print-rtl.c: Include basic-block.h
14407 * profile-count.c: New file.
14408 * profile-count.h: New file.
14409 * profile.c (is_edge_inconsistent): Update.
14410 (correct_negative_edge_counts): Update.
14411 (is_inconsistent): Update.
14412 (set_bb_counts): Update.
14413 (read_profile_edge_counts): Update.
14414 (compute_frequency_overlap): Update.
14415 (compute_branch_probabilities): Update; Initialize and deinitialize
14417 (branch_prob): Update.
14418 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
14419 (edge_gcov_count): New.
14420 (bb_gcov_count): New.
14421 * shrink-wrap.c (try_shrink_wrapping): Update.
14422 * tracer.c (better_p): Update.
14423 * trans-mem.c (expand_transaction): Update.
14424 (ipa_tm_insert_irr_call): Update.
14425 (ipa_tm_insert_gettmclone_call): Update.
14426 * tree-call-cdce.c: Update.
14427 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14428 (gimple_duplicate_sese_tail): Update.
14429 (gimple_account_profile_record): Update.
14430 (execute_fixup_cfg): Update.
14431 * tree-inline.c (copy_bb): Update.
14432 (copy_edges_for_bb): Update.
14433 (initialize_cfun): Update.
14434 (freqs_to_counts): Update.
14435 (copy_cfg_body): Update.
14436 (expand_call_inline): Update.
14437 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14438 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14439 (try_unroll_loop_completely): Update.
14440 (try_peel_loop): Update.
14441 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14442 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14443 * tree-ssa-loop-split.c (connect_loops): Update.
14444 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14445 * tree-ssa-reassoc.c (branch_fixup): Update.
14446 * tree-ssa-tail-merge.c (replace_block_by): Update.
14447 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14448 (compute_path_counts): Update.
14449 (update_profile): Update.
14450 (recompute_probabilities): Update.
14451 (update_joiner_offpath_counts): Update.
14452 (estimated_freqs_path): Update.
14453 (freqs_to_counts_path): Update.
14454 (clear_counts_path): Update.
14455 (ssa_fix_duplicate_block_edges): Update.
14456 (duplicate_thread_path): Update.
14457 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14458 (struct switch_conv_info): Update.
14459 * tree-tailcall.c (decrease_profile): Update.
14460 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14461 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14462 * value-prof.c (check_counter): Update.
14463 (gimple_divmod_fixed_value): Update.
14464 (gimple_mod_pow2): Update.
14465 (gimple_mod_subtract): Update.
14466 (gimple_ic_transform): Update.
14467 (gimple_stringop_fixed_value): Update.
14468 * value-prof.h (gimple_ic): Update.
14470 2017-06-02 Carl Love <cel@us.ibm.com>
14472 * config/rs6000/rs6000-c: Add support for built-in functions
14473 vector double vec_doublee (vector signed int);
14474 vector double vec_doublee (vector unsigned int);
14475 vector double vec_doublee (vector float);
14476 vector double vec_doubleh (vector signed int);
14477 vector double vec_doubleh (vector unsigned int);
14478 vector double vec_doubleh (vector float);
14479 vector double vec_doublel (vector signed int);
14480 vector double vec_doublel (vector unsigned int);
14481 vector double vec_doublel (vector float);
14482 vector double vec_doubleo (vector signed int);
14483 vector double vec_doubleo (vector unsigned int);
14484 vector double vec_doubleo (vector float);.
14485 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14486 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14488 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14489 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14490 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14492 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14493 vec_doublel, vec_doubleh.
14494 * doc/extend.texi: Update the built-in documentation file for the
14495 new built-in functions.
14497 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14500 * ipa-inline-analysis.c (free_growth_caches): Set
14501 edge_removal_hook_holder to NULL after removing it.
14503 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14505 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14506 comparision with zero.
14508 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14509 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14510 for early expansion of vec_min and vec_max builtins.
14511 (builtin_function_type): Add min/max unsigned variants to those
14512 identified as having unsigned arguments.
14514 2017-06-02 Olivier Hainque <hainque@adacore.com>
14516 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14518 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14520 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14521 Use VALL_F16 iterator rather than VALL.
14523 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14525 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14526 Emit CBNZ inside loop when doing a strong exchange and comparing
14527 against zero. Generate the CC flags after the loop.
14529 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14531 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14532 (dl_section_ref): New.
14533 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14534 On AIX, append an expression to subtract the size of the
14535 section length to dl_section_ref.
14537 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14539 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14540 for early expansion of vector absolute builtins.
14542 2017-06-02 Richard Biener <rguenther@suse.de>
14544 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14545 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14547 2017-06-02 Richard Biener <rguenther@suse.de>
14549 PR tree-optimization/80948
14550 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14551 stmt order as well.
14553 2017-06-02 Richard Biener <rguenther@suse.de>
14555 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14557 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14558 for inductions as relevant.
14560 2017-06-02 Richard Biener <rguenther@suse.de>
14562 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14563 (vectorizable_induction): ... this. Remove dead code.
14565 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14567 * builtins. (expand_builtin_alloca): Remove second parameter and
14568 infer its value from the first parameter instead.
14569 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14571 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14573 PR rtl-optimization/80903
14574 * loop-doloop.c (add_test): Unshare sequence.
14576 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14578 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14580 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14582 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14584 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14585 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14586 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14587 (xlogue_layout::get_stub_rtx): Make static.
14588 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14589 (xlogue_layout::compute_stub_managed_regs): Rename to...
14590 (xlogue_layout::count_stub_managed_regs): ...this.
14591 (xlogue_layout::is_stub_managed_reg): New function.
14592 (xlogue_layout::m_stub_names): Rename to...
14593 (xlogue_layout::s_stub_names): ...this, make static.
14594 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14595 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14596 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14597 xlogue_layout::s_stub_names): Instantiate statics.
14598 (stub_managed_regs): Remove.
14599 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14600 (disable_call_ms2sysv_xlogues): Rename to...
14601 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14602 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14604 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14605 change after reload_completed.
14606 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14608 (ix86_expand_prologue): Likewise.
14609 (ix86_expand_epilogue): Likewise.
14610 (ix86_expand_split_stack_prologue): Likewise.
14611 (ix86_compute_frame_layout): Remove frame parameter ...
14612 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14613 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14615 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14616 (ix86_frame): Move from here ...
14617 * config/i386/i386.h (ix86_frame): ... to here.
14618 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14619 complete ix86_frame data structure instead. Remove some_ld_name.
14621 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14623 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14624 symbols that hold a DECL_VALUE_EXPR.
14626 2017-06-01 Martin Jambor <mjambor@suse.cz>
14628 PR tree-optimization/80898
14629 * tree-sra.c (process_subtree_disqualification): Removed.
14630 (disqualify_candidate): Do not acll
14631 process_subtree_disqualification.
14632 (subtree_mark_written_and_enqueue): New function.
14633 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14634 RHS has been disqualified and re-queue LHS if necessary. Apart
14635 from that, ignore disqualified RHS.
14637 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14639 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14640 address fetch for z10 or later.
14642 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14644 * config/arc/arc.md (tst_movb): Add guard when splitting.
14646 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14648 * config/arc/arc.c (arc_can_eliminate): Test against
14649 arc_frame_pointer_needed.
14651 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14653 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14654 to prevent store reordering.
14655 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14656 (type): Add block type.
14657 (stack_tie): Define special instruction to be used in
14660 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14662 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14663 constraint. It is not valid for the pattern.
14664 (noncommutative_binary_comparison): Likewise.
14666 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14668 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
14671 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14673 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
14674 be used by the reg-alloc.
14676 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14678 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
14679 reg-alloc when having mul64 or mul32x16 instructions.
14680 (mulsidi3): Likewise.
14681 (umulsidi3): Likewise.
14682 (mulsi32x16): New pattern.
14683 (mulsi64): Likewise.
14684 (mulsidi64): Likewise.
14685 (umulsidi64): Likewise.
14686 (MUL32x16_REG): Define.
14687 (mul64_600): Use MUL32x16_REG.
14688 (mac64_600): Likewise.
14689 (umul64_600): Likewise.
14690 (umac64_600): Likewise.
14692 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14694 * config/arc/arc.md (mulsi3_700): Make it commutative.
14696 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14698 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14700 (*sign_extendsidi2_insn): Likewise for movstosw.
14702 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14704 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14705 the type of the input discriminant value. Convert the
14706 discriminant value of signedness vary.
14708 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14710 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14711 Add to -Wall section.
14713 2017-06-01 Richard Biener <rguenther@suse.de>
14715 PR middle-end/66313
14716 * fold-const.c (fold_plusminus_mult_expr): If the factored
14717 factor may be zero use a wrapping type for the inner operation.
14718 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14719 and handle moved defs.
14720 (process_assignment): Properly guard the unary op case. Return a
14721 tri-state indicating that moving the stmt before the call may allow
14722 to continue. Pass through to_move.
14723 (find_tail_calls): Handle moving unrelated defs before
14726 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14729 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14730 splitter result in the canonical way.
14732 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14734 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14735 also for 32bit target. Update insn attributes.
14736 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14738 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14740 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14741 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14742 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14743 (_mm_maskz_min_ss): New intrinsics.
14745 2017-05-31 Martin Liska <mliska@suse.cz>
14747 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14748 Change comment style to one we normally use.
14749 (vectorizable_reduction): Likewise.
14750 (vectorizable_induction): Likewise.
14751 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14752 (vectorizable_call): Likewise.
14753 (vectorizable_simd_clone_call): Likewise.
14754 (vectorizable_conversion): Likewise.
14755 (vectorizable_assignment): Likewise.
14756 (vectorizable_shift): Likewise.
14757 (vectorizable_operation): Likewise.
14758 (vectorizable_store): Likewise.
14759 (vectorizable_load): Likewise.
14760 * tree-vectorizer.h: Likewise.
14762 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14764 * passes.c (emergency_dump_function): New.
14765 * tree-pass.h (emergency_dump_function): Declare.
14766 * plugin.c (plugins_internal_error_function): Remove.
14767 * plugin.h (plugins_internal_error_function): Remove declaration.
14768 * toplev.c (internal_error_function): New static function. Use it...
14769 (general_init): ...here.
14771 2017-05-31 Graham Markall <graham.markall@embecosm.com>
14773 * config/arc/arc.c (arc_print_operand): Handle constant operands.
14774 (arc_rtx_costs): Add costs for new patterns.
14775 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
14776 * config/arc/predicates.md: Add _1_2_3_operand predicate.
14778 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14780 * tree-ssa-strlen.c (get_next_strinfo): New function.
14781 (get_stridx_plus_constant): Use it.
14782 (zero_length_string): Likewise.
14783 (adjust_related_strinfos): Likewise.
14784 (adjust_last_stmt): Likewise.
14786 2017-05-31 Richard Biener <rguenther@suse.de>
14789 * config/i386/i386.c (ix86_expand_builtin): Remove assert
14790 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
14792 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14794 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
14795 loop_vinfo argument and use of dependence distance vectors.
14796 Check instead whether the two references differ only in their
14797 initial value and assume that they have the same alignment if the
14798 difference is a multiple of the vector alignment.
14799 (vect_analyze_data_refs_alignment): Update call accordingly.
14801 2017-05-31 Martin Liska <mliska@suse.cz>
14804 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
14806 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14808 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
14809 (create_intersect_range_checks): Move from ...
14810 * tree-data-ref.c (create_intersect_range_checks_index)
14811 (create_intersect_range_checks): ... to here.
14812 (create_runtime_alias_checks): New function factored from ...
14813 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
14814 here. Call above function.
14815 * tree-data-ref.h (create_runtime_alias_checks): New function.
14817 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14819 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
14820 segment length for dr_b and compute it in wide_int.
14822 2017-05-31 Richard Biener <rguenther@suse.de>
14824 PR tree-optimization/80906
14825 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
14826 and pass through iv_map.
14827 (copy_bb_and_scalar_dependences): Adjust.
14828 (translate_pending_phi_nodes): Likewise.
14829 (copy_loop_close_phi_args): Handle code-generating IVs instead
14832 2017-05-30 David Malcolm <dmalcolm@redhat.com>
14834 * diagnostic-color.c (color_dict): Add "type-diff".
14835 (parse_gcc_colors): Update comment.
14836 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
14837 -fdiagnostics-show-template-tree and -fno-elide-type.
14838 (GCC_COLORS): Add type-diff to example.
14840 (-fdiagnostics-show-template-tree): New.
14841 (-fno-elide-type): New.
14842 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
14843 the pp_format_decoder callback. Call any m_format_postprocessor's
14845 (pretty_printer::pretty_printer): Initialize
14846 m_format_postprocessor.
14847 (pretty_printer::~pretty_printer): Delete any
14848 m_format_postprocessor.
14849 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
14850 (class format_postprocessor): New class.
14851 (struct pretty_printer::format_decoder): Document the new parameters.
14852 (struct pretty_printer::m_format_postprocessor): New field.
14853 * tree-diagnostic.c (default_tree_printer): Update for new
14854 bool and const char ** params.
14855 * tree-diagnostic.h (default_tree_printer): Likewise.
14857 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
14859 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
14860 (lwa_operand): Delete rs6000_gen_cell_microcode test.
14861 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14862 rs6000_gen_cell_microcode code.
14863 (rs6000_final_prescan_insn): Delete.
14864 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
14865 "warn-cell-microcode" entries.
14866 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
14867 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
14868 throughout. Change cc_reg_not_micro_cr0_operand to
14869 cc_reg_not_cr0_operand throughout.
14870 (*extendhi<mode>2_noload): Delete.
14871 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
14872 (mwarn-cell-microcode): Delete.
14873 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
14874 -mgen-cell-microcode and -mwarn-cell-microcode.
14876 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
14879 * config/i386/constraints.md (Yd): New constraint.
14881 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
14882 and (?Yd, r) alternatives. Update insn attributes.
14883 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
14884 and (?*Yd, r) alternatives. Update insn attributes.
14885 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
14887 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
14889 * gimplify.c (gimplify_modify_expr): Don't create a
14890 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
14893 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14895 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
14897 2017-05-30 Richard Biener <rguenther@suse.de>
14899 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
14900 and reduc_def fields.
14901 (STMT_VINFO_REDUC_TYPE): New define.
14902 (STMT_VINFO_REDUC_DEF): Likewise.
14903 (vect_force_simple_reduction): Adjust prototype.
14904 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
14905 (vect_is_simple_reduction): Remove check_reduction argument.
14906 (vect_force_simple_reduction): Adjust and set
14907 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14908 (vectorizable_reduction): Do not re-do reduction analysis
14909 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14910 * tree-parloops.c (gather_scalar_reductions): Adjust.
14912 2017-05-30 Richard Biener <rguenther@suse.de>
14914 PR middle-end/80901
14915 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
14918 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
14920 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
14921 Introduce unknown_misalignment parameter and remove vf.
14922 (vect_peeling_hash_get_lowest_cost):
14923 Pass unknown_misalignment parameter.
14924 (vect_enhance_data_refs_alignment):
14925 Fix unsupportable data ref treatment.
14927 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14929 * tree-vect-data-refs.c (vect_get_data_access_cost):
14930 Workaround for SLP handling.
14931 (vect_enhance_data_refs_alignment):
14932 Compute costs for doing no peeling at all, compare to the best
14933 peeling costs so far and avoid peeling if cheaper.
14935 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14937 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
14938 Return peeling info and set costs to zero for unlimited cost
14940 (vect_enhance_data_refs_alignment): Also inspect all datarefs
14941 with unknown misalignment. Compute and costs for unknown
14942 misalignment, compare them to the costs for known misalignment
14943 and choose the cheapest for peeling.
14945 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14947 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
14948 (vect_get_peeling_costs_all_drs): Create function.
14949 (vect_peeling_hash_get_lowest_cost):
14950 Use vect_get_peeling_costs_all_drs.
14951 (vect_peeling_supportable): Create function.
14952 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
14954 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14956 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
14957 DR_HAS_NEGATIVE_STEP.
14958 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
14959 (vect_enhance_data_refs_alignment): Use.
14960 (vect_duplicate_ssa_name_ptr_info): Use.
14961 * tree-vectorizer.h (dr_misalignment): Use.
14962 (known_alignment_for_access_p): Use.
14964 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14967 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
14969 (has_section_name): New function.
14971 2017-05-30 Martin Liska <mliska@suse.cz>
14974 * auto-profile.c (get_function_decl_from_block): Fix
14977 2017-05-30 Richard Biener <rguenther@suse.de>
14979 PR middle-end/80876
14980 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
14982 2017-05-30 Martin Liska <mliska@suse.cz>
14984 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
14985 * dumpfile.h (struct dump_file_info): Remove ctors.
14987 2017-05-30 Martin Liska <mliska@suse.cz>
14989 * predict.def: Fix GNU coding style.
14991 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14993 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14994 Mark 'to' argument with ATTRIBUTE_UNUSED.
14996 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14998 * config/xtensa/xtensa.c (xtensa_emit_call): Use
14999 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
15000 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
15003 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
15005 * doc/install.texi (Options specification): Restore entry of
15006 --enable-sjlj-exceptions.
15008 2017-05-27 Michael Eager <eager@eagercon.com>
15011 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
15013 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
15015 * config/microblaze/microblaze.h
15016 (FIXED_REGISTERS): Update in macro.
15017 (CALL_USED_REGISTERS): Update in macro.
15019 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
15021 * doc/install.texi: Add links to macOS binary distributions.
15023 2017-05-27 Jakub Jelinek <jakub@redhat.com>
15027 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15029 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15031 2017-05-26 Martin Liska <mliska@suse.cz>
15033 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
15035 2017-05-26 Martin Liska <mliska@suse.cz>
15037 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
15038 always leading ';; '.
15039 (dump_bb_info): Likewise.
15040 (brief_dump_cfg): Likewise.
15041 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
15042 * dumpfile.c: Remove usage of TDF_VERBOSE.
15043 * dumpfile.h (enum dump_kind): Likewise.
15044 (dump_gimple_bb_header): Do not use TDF_COMMENT.
15045 * print-tree.c (debug_verbose): Remove.
15046 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
15047 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
15048 * tree-diagnostic.c (default_tree_printer): Replace
15049 TDF_DIAGNOSTIC with TDF_SLIM.
15051 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15053 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
15054 in parameter loop, rather than loop_vinfo.
15055 (create_intersect_range_checks): Ditto.
15056 (vect_create_cond_for_alias_checks): Update call to above functions.
15058 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15060 PR tree-optimization/80815
15061 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
15062 for merging runtime alias checks. Handle negative DR_STEPs.
15064 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15066 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
15068 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
15069 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
15070 out code pruning runtime alias checks.
15071 * tree-data-ref.c (prune_runtime_alias_test_list): New function
15072 factored out from above.
15073 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15075 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15077 (prune_runtime_alias_test_list): New decalaration.
15079 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15081 * tree-vect-data-refs.c (compare_tree): Rename and move ...
15082 * tree-data-ref.c (data_ref_compare_tree): ... to here.
15083 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
15084 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
15085 (operator==, comp_dr_with_seg_len_pair): Ditto.
15086 (vect_prune_runtime_alias_test_list): Ditto.
15088 2017-05-26 Martin Liska <mliska@suse.cz>
15091 * params.def: Bound partial-inlining-entry-probability param.
15093 2017-05-26 Marek Polacek <polacek@redhat.com>
15096 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
15099 2017-05-26 Richard Biener <rguenther@suse.de>
15101 PR tree-optimization/80842
15102 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
15105 2017-05-26 Richard Biener <rguenther@suse.de>
15107 PR tree-optimization/80844
15108 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
15110 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
15112 * doc/md.texi (Machine Constraints): Update x86 family
15113 machine constraints section to match 'config/i386/constraints.md'.
15115 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
15117 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
15119 2017-05-25 Nathan Sidwell <nathan@acm.org>
15121 * doc/invoke.texi (--enable-languages): Update documentation.
15123 2017-05-25 Martin Liska <mliska@suse.cz>
15125 * dumpfile.c: Add TDF_FOLDING.
15126 * dumpfile.h (enum dump_kind): Likewise.
15127 * genmatch.c (dt_simplify::gen_1): Use it.
15129 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15131 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
15133 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15135 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15136 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
15138 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15140 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
15141 * match.pd (X == C): Rewrite it here.
15142 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
15143 with_certain_nonzero_bits2): New predicates.
15144 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
15146 2017-05-24 Nathan Sidwell <nathan@acm.org>
15148 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
15151 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
15154 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15156 * config/powerpcspe: New port. Files are copied from the rs6000
15157 port, with "rs6000" in filenames replaced by "powerpcspe".
15159 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
15161 PR rtl-optimization/80754
15162 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
15164 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
15166 * config/sparc/sparc.md (length): Return the correct value for -mflat
15167 sibcalls to match output_sibcall.
15169 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15173 * config/rs6000/rs6000.c (struct machine_function): Add new field
15175 (rs6000_get_separate_components): Init that field, use it.
15176 (rs6000_components_for_bb): Use the field.
15178 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
15180 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
15182 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
15184 PR middle-end/80823
15185 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
15187 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15190 * config/s390/s390.c (s390_check_qrst_address): Check incoming
15191 address against address_operand predicate.
15192 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
15194 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
15196 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
15197 parameters passed indirectly.
15199 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
15201 * config/i386/i386.md (*movdi_internal): Remove SSE4
15202 alternative 18 (?r, *v). Update insn attributes.
15203 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
15204 Update insn attributes.
15205 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
15206 Update insn attributes.
15207 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
15208 alternative 1 (r, v). Remove isa attribute.
15209 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
15210 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
15211 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
15213 2017-05-23 Tom de Vries <tom@codesourcery.com>
15215 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
15218 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15220 * cgraphunit.c (symbol_table::process_new_functions): Update.
15221 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
15222 (inline_generate_summary): Rename to ...
15223 (ipa_fn_summary_generate): ... this one.
15224 (inline_read_summary): Rename to ...
15225 (ipa_fn_summary_read): ... this one.
15226 (inline_write_summary): Rename to ...
15227 (ipa_fn_summary_write): ... this one.
15228 (inline_free_summary): Rename to ...
15229 (ipa_free_fn_summary): ... this one.
15230 (pass_data_local_fn_summary, pass_local_fn_summary,
15231 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
15232 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
15233 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
15234 make_pass_ipa_fn_summary): New.
15235 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
15236 inline_write_summary, inline_free_summary): Remove.
15237 (ipa_free_fn_summary) : New.
15238 * ipa-inline.c (ipa_inline): Update.
15239 (pass_ipa_inline): Do not generate summaries.
15240 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
15242 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
15243 and add pass_ipa_fn_summary.
15244 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
15246 (make_pass_inline_parameters): Remove.
15248 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
15250 * omp-low.c (struct omp_context): Remove "default_kind" member.
15253 * omp-offload.c (execute_oacc_device_lower): Remove the
15254 parallelism dimensions function attributes for unparallelized
15255 OpenACC kernels constructs.
15257 2017-05-23 Martin Liska <mliska@suse.cz>
15259 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
15261 (cgraph_edge::make_speculative): Likewise.
15262 (cgraph_edge::resolve_speculation): Likewise.
15263 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15264 (cgraph_node::dump): Likewise.
15265 * cgraph.h: Likewise.
15266 * cgraphunit.c (analyze_functions): Likewise.
15267 (symbol_table::compile): Likewise.
15268 * ipa-cp.c (print_all_lattices): Likewise.
15269 (determine_versionability): Likewise.
15270 (initialize_node_lattices): Likewise.
15271 (ipcp_verify_propagated_values): Likewise.
15272 (estimate_local_effects): Likewise.
15273 (update_profiling_info): Likewise.
15274 (create_specialized_node): Likewise.
15275 (perhaps_add_new_callers): Likewise.
15276 (decide_about_value): Likewise.
15277 (decide_whether_version_node): Likewise.
15278 (identify_dead_nodes): Likewise.
15279 (ipcp_store_bits_results): Likewise.
15280 * ipa-devirt.c (dump_targets): Likewise.
15281 (ipa_devirt): Likewise.
15282 * ipa-icf.c (sem_item::dump): Likewise.
15283 (sem_function::equals): Likewise.
15284 (sem_variable::equals): Likewise.
15285 (sem_item_optimizer::read_section): Likewise.
15286 (sem_item_optimizer::execute): Likewise.
15287 (congruence_class::dump): Likewise.
15288 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
15289 (dump_inline_summary): Likewise.
15290 (estimate_node_size_and_time): Likewise.
15291 (inline_analyze_function): Likewise.
15292 * ipa-inline-transform.c (inline_call): Likewise.
15293 * ipa-inline.c (report_inline_failed_reason): Likewise.
15294 (want_early_inline_function_p): Likewise.
15295 (edge_badness): Likewise.
15296 (update_edge_key): Likewise.
15297 (inline_small_functions): Likewise.
15298 * ipa-profile.c (ipa_profile): Likewise.
15299 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
15300 (ipa_make_edge_direct_to_target): Likewise.
15301 (remove_described_reference): Likewise.
15302 (ipa_impossible_devirt_target): Likewise.
15303 (propagate_controlled_uses): Likewise.
15304 (ipa_print_node_params): Likewise.
15305 (ipcp_transform_function): Likewise.
15306 * ipa-pure-const.c (pure_const_read_summary): Likewise.
15307 (propagate_pure_const): Likewise.
15308 * ipa-reference.c (generate_summary): Likewise.
15309 (read_write_all_from_decl): Likewise.
15310 (propagate): Likewise.
15311 (ipa_reference_read_optimization_summary): Likewise.
15312 * ipa-utils.c (ipa_merge_profiles): Likewise.
15313 * ipa.c (walk_polymorphic_call_targets): Likewise.
15314 (symbol_table::remove_unreachable_nodes): Likewise.
15315 (ipa_single_use): Likewise.
15316 * passes.c (execute_todo): Likewise.
15317 * predict.c (drop_profile): Likewise.
15318 * symtab.c (symtab_node::get_dump_name): New function.
15319 (symtab_node::dump_name): Likewise.
15320 (symtab_node::dump_asm_name): Likewise.
15321 (symtab_node::dump_references): Likewise.
15322 (symtab_node::dump_referring): Likewise.
15323 (symtab_node::dump_base): Likewise.
15324 (symtab_node::debug_symtab): Likewise.
15325 * tree-sra.c (convert_callers_for_node): Likewise.
15326 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
15327 * value-prof.c (init_node_map): Likewise.
15329 2017-05-23 Martin Liska <mliska@suse.cz>
15331 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
15332 and symtab_node::debug_symtab to symbol_table::debug.
15333 * cgraphunit.c (analyze_functions): Use the renamed function.
15334 (symbol_table::compile): Likewise.
15335 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
15336 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
15337 * passes.c (execute_todo): Likewise.
15338 * symtab.c (symbol_table::dump): New function.
15339 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
15341 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15343 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
15344 that nonconst implies exec.
15346 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15348 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15349 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15350 (inline_edge_summary_vec): Turn into ...
15351 (ipa_call_summaries): ... this one.
15352 (redirect_to_unreachable, edge_set_predicate,
15353 evaluate_properties_for_edge, inline_summary_alloc,
15354 reset_ipa_call_summary, reset_inline_summary,
15355 inline_summary_t::duplicate): Update.
15356 (inline_edge_duplication_hook): Turn to ...
15357 (ipa_call_summary_t::duplicate): ... this one.
15358 (inline_edge_removal_hook): Turn to ...
15359 (ipa_call_summary_t::remove): ... this one.
15360 (dump_inline_edge_summary): Turn to ...
15361 (dump_ipa_call_summary): ... this one.
15362 (estimate_function_body_sizes): Update.
15363 (inline_update_callee_summaries): Update.
15364 (remap_edge_change_prob): Update.
15365 (remap_edge_summaries): Update.
15366 (inline_merge_summary): Update.
15367 (do_estimate_edge_time): Update.
15368 (inline_generate_summary): Update.
15369 (inline_read_section): Update.
15370 (inline_read_summary): Update.
15371 (inline_free_summary): Update.
15372 * ipa-inline.c (can_inline_edge_p): Update.
15373 (compute_inlined_call_time): Update.
15374 (want_inline_small_function_p): Update.
15375 (edge_badness): Update.
15376 (early_inliner): Update.
15377 * ipa-inline.h (inline_edge_summary): Turn to ...
15378 (ipa_call_summary): ... this one.
15379 (ipa_call_summary_t): New class.
15380 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15381 (ipa_call_summaries): New.
15382 (inline_edge_summary): Remove.
15383 (estimate_edge_growth): Update.
15384 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15385 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15386 * ipa-split.c (execute_split_functions): Update.
15387 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15389 2017-05-23 Tom de Vries <tom@codesourcery.com>
15391 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15392 attributes): Document rdrand effective target.
15394 2017-05-23 Tom de Vries <tom@codesourcery.com>
15396 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15397 attributes): Sort alphabetically.
15399 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
15401 * config/avr/genmultilib.awk: Use gsub instead of gensub.
15403 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15406 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
15407 V2DF/V2DI splat into two separate patterns, one that handles
15408 registers, and the other that only handles memory. Drop support
15409 for splatting from a GPR on ISA 2.07 and then splitting the
15410 splat into direct move and splat.
15411 (vsx_splat_<mode>_reg): Likewise.
15412 (vsx_splat_<mode>_mem): Likewise.
15414 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
15416 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
15418 2017-05-22 Jakub Jelinek <jakub@redhat.com>
15420 PR middle-end/80809
15421 * omp-low.c (finish_taskreg_remap): New function.
15422 (finish_taskreg_scan): If unit size of ctx->record_type
15423 is non-constant, unshare the size expression and replace
15424 decls in it with possible outer var refs.
15426 PR middle-end/80809
15427 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15428 GOVD_SHARED rather than GOVD_PRIVATE with it.
15429 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15430 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15432 PR middle-end/80853
15433 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15434 as last argument to build_outer_var_ref for pointer bases of array
15435 section reductions.
15437 2017-05-19 Martin Sebor <msebor@redhat.com>
15439 * print-tree.c (print_node): Print DECL_READ_P flag.
15441 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15443 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15444 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15445 * cgraph.c: Likewise.
15446 * cgraphunit.c: Likewise.
15447 * gengtype.c: Likewise.
15448 * ipa-cp.c: Likewise.
15449 * ipa-devirt.c: Likewise.
15450 * ipa-icf.c: Likewise.
15451 * ipa-predicate.c: Likewise.
15452 * ipa-profile.c: Likewise.
15453 * ipa-prop.c: Likewise.
15454 * ipa-split.c: Likewise.
15456 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15457 edge_predicate_pool, dump_inline_hints,
15458 inline_summary::account_size_time, redirect_to_unreachable,
15459 edge_set_predicate, set_hint_predicate,
15460 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15461 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15462 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15463 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15464 ipa_call_summary_t::remove, initialize_growth_caches,
15465 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15466 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15467 mark_modified, unmodified_parm_1, unmodified_parm,
15468 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15469 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15470 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15471 will_be_nonconstant_predicate, record_modified_bb_info,
15472 get_minimal_bb, record_modified, param_change_prob,
15473 phi_result_unknown_predicate, predicate_for_phi_result,
15474 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15475 estimate_function_body_sizes, compute_inline_parameters,
15476 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15477 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15478 inline_update_callee_summaries, remap_edge_change_prob,
15479 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15480 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15481 inline_analyze_function, inline_summary_t::insert,
15482 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15483 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15484 inline_free_summary): Move to ipa-fnsummary.h
15485 (predicate_t): Remove.
15486 * ipa-fnsummary.c: New file.
15487 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15488 (enum inline_hints_vals, inline_hints, agg_position_info,
15489 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15490 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15491 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15492 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15493 inline_read_summary, inline_write_summary, inline_free_summary,
15494 inline_analyze_function, initialize_inline_failed,
15495 inline_merge_summary, inline_update_overall_summary,
15496 compute_inline_parameters): Move to ipa-fnsummary.h
15497 * ipa-fnsummary.h: New file.
15498 * ipa-inline-transform.h: Include ipa-inline.h.
15499 * ipa-inline.c: LIkewise.
15501 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15503 * ipa-inline.c (edge_badness): Use inlined_time instead of
15504 inline_summaries->get.
15506 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15508 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15510 2017-05-22 Nathan Sidwell <nathan@acm.org>
15512 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15513 (fdump-lang): Document 'raw' option.
15514 * dumpfile.h (TDI_tu): Delete.
15515 * dumpfile.c (dump_files): Remove translation-unit.
15516 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15518 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15520 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15521 command option from $(AWK) call.
15522 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15523 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15524 [FORMAT]: Remove handling of variable.
15525 * config/avr/t-multilib: Regenerate.
15527 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15529 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15531 (dump_inline_summary): Do not print self_time.
15532 (estimate_function_body_sizes): Do not set self_time.
15533 (compute_inline_parameters): Likewise.
15534 (inline_read_section, inline_write_summary): Do not stream self_time.
15535 * ipa-inline.h (inline_summary): Drop self_time.
15537 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15539 * ipa-inline-analysis.c (account_size_time): Rename to ...
15540 (inline_summary::account_size_time): ... this one.
15541 (reset_ipa_call_summary): Turn to ...
15542 (ipa_call_summary::reset): ... this one.
15543 (reset_inline_summary): Turn to ...
15544 (inline_summary::reset): ... this one.
15545 (inline_summary_t::remove): Update.
15546 (inline_summary_t::duplicate): Update.
15547 (ipa_call_summary_t::remove): Update.
15548 (dump_inline_summary): Update.
15549 (estimate_function_body_sizes): Update.
15550 (compute_inline_parameters): Update.
15551 (estimate_node_size_and_time): Update.
15552 (inline_merge_summary): Update.
15553 (inline_update_overall_summary): Update.
15554 (inline_read_section): Update.
15555 (inline_write_summary): Update.
15556 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15557 add account_size_time and reset member functions.
15558 (ipa_call_summary): Add reset function.
15559 * ipa-predicate.h (predicate::operator &): Constify.
15561 2017-05-22 Richard Biener <rguenther@suse.de>
15563 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15565 2017-05-19 Jason Merrill <jason@redhat.com>
15567 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15569 2017-05-19 Marek Polacek <polacek@redhat.com>
15572 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15573 TYPE_OVERFLOW_WRAPS checks.
15575 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15577 * tree-core.h (enum omp_clause_default_kind): Add
15578 "OMP_CLAUSE_DEFAULT_PRESENT".
15579 * tree-pretty-print.c (dump_omp_clause): Handle it.
15580 * gimplify.c (enum gimplify_omp_var_data): Add
15581 "GOVD_MAP_FORCE_PRESENT".
15582 (gimplify_adjust_omp_clauses_1): Map it to
15583 "GOMP_MAP_FORCE_PRESENT".
15584 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15586 * gimplify.c (oacc_default_clause): Clarify.
15588 2017-05-19 Nathan Sidwell <nathan@acm.org>
15590 LANG_HOOK_REGISTER_DUMPS
15591 * toplev.c (general_init): Call register dump lang hook.
15592 * doc/invoke.texi: Document -fdump-lang option family.
15593 * dumpfile.c (dump_files): Remove class dump here.
15594 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15595 * dumpfile.h (tree_dump_index): Remove TDI_class.
15596 * langhooks-def.h (lhd_register_dumps): Declare.
15597 (LANG_HOOKS_REGISTER_DUMPS): Define.
15598 (LANG_HOOKS_INITIALIZER): Add it.
15599 * langhooks.c (lhd_register_dumps): Define.
15600 * langhooks.h (struct lang_hooks): Add register_dumps.
15602 2017-05-19 Nathan Sidwell <nathan@acm.org>
15604 * context.h (context::set_passes): New.
15605 * context.c (context::context): Do not create pass manager.
15606 * toplev.c (general_init): Create pass manager here.
15608 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15610 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15611 use this splitter if two add or or instructions would also work for
15612 the constant we want to generate.
15614 2017-05-19 Richard Biener <rguenther@suse.de>
15617 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15618 predicate evaluation.
15620 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15622 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15624 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15627 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15629 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15630 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15631 (inline_edge_summary_vec): Turn into ...
15632 (ipa_call_summaries): ... this one.
15633 (redirect_to_unreachable, edge_set_predicate,
15634 evaluate_properties_for_edge, inline_summary_alloc,
15635 reset_ipa_call_summary, reset_inline_summary,
15636 inline_summary_t::duplicate): Update.
15637 (inline_edge_duplication_hook): Turn to ...
15638 (ipa_call_summary_t::duplicate): ... this one.
15639 (inline_edge_removal_hook): Turn to ...
15640 (ipa_call_summary_t::remove): ... this one.
15641 (dump_inline_edge_summary): Turn to ...
15642 (dump_ipa_call_summary): ... this one.
15643 (estimate_function_body_sizes): Update.
15644 (inline_update_callee_summaries): Update.
15645 (remap_edge_change_prob): Update.
15646 (remap_edge_summaries): Update.
15647 (inline_merge_summary): Update.
15648 (do_estimate_edge_time): Update.
15649 (inline_generate_summary): Update.
15650 (inline_read_section): Update.
15651 (inline_read_summary): Update.
15652 (inline_free_summary): Update.
15653 * ipa-inline.c (can_inline_edge_p): Update.
15654 (compute_inlined_call_time): Update.
15655 (want_inline_small_function_p): Update.
15656 (edge_badness): Update.
15657 (early_inliner): Update.
15658 * ipa-inline.h (inline_edge_summary): Turn to ...
15659 (ipa_call_summary): ... this one.
15660 (ipa_call_summary_t): New class.
15661 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15662 (ipa_call_summaries): New.
15663 (inline_edge_summary): Remove.
15664 (estimate_edge_growth): Update.
15665 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15666 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15667 * ipa-split.c (execute_split_functions): Update.
15668 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15670 2017-05-19 Richard Biener <rguenther@suse.de>
15672 PR middle-end/80764
15673 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
15675 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
15677 * config/rs6000/rs6000.c (struct machine_function): Add field
15678 fpr_is_wrapped_separately.
15679 (rs6000_get_separate_components): Use 64 components. Handle the
15680 new FPR components.
15681 (rs6000_components_for_bb): Handle the FPR components.
15682 (rs6000_emit_prologue_components): Handle the FPR components.
15683 (rs6000_emit_epilogue_components): Handle the FPR components.
15684 (rs6000_set_handled_components): Handle the FPR components.
15685 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15686 that are already separately shrink-wrapped.
15687 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15688 that are already separately shrink-wrapped.
15690 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15693 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15696 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15697 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15698 cases where the register allocator uses a move and an offsettable
15699 memory operation to/from a FPR register on ISA 2.06/2.07.
15700 (define_peephole2 for Altivec d-form store): Likewise.
15702 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15705 * config/i386/mmx.md (*mov<mode>_internal): Enable
15706 alternatives 11, 12, 13 and 14 also for 32bit targets.
15707 Remove alternatives 15, 16, 17 and 18.
15708 * config/i386/sse.md (vec_concatv2di): Change
15709 alternative (!x, *y) to (x, ?!*Yn).
15711 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15713 * dumpfile.h (enum dump_kind): Remove stray comma.
15715 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15717 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15718 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15719 predicate::num_conditions
15720 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15721 (CHANGED): turn into predicate::changed.
15722 (agg_position_info): Move to ipa-predicate.h
15723 (add_condition, predicate::add_clause, predicate::operator &=,
15724 predicate::or_with, predicate::evaluate, predicate::probability,
15725 dump_condition, dump_clause, predicate::dump,
15726 predicate::remap_after_duplication, predicate::remap_after_inlining,
15727 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15728 (evaluate_conditions_for_known_args): Update.
15729 (set_cond_stmt_execution_predicate): Update.
15730 * ipa-inline.h: Include ipa-predicate.h
15731 (condition, inline_param_summary, conditions, agg_position_info,
15732 predicate): Move to ipa-predicate.h
15733 * ipa-predicate.c: New file.
15734 * ipa-predicate.h: New file.
15736 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15738 * final.c (leaf_function_p): Check we are not in a sequence.
15740 2017-05-18 Martin Liska <mliska@suse.cz>
15742 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15743 * dumpfile.c (dump_register): Use new enum dump_kind.
15744 (get_dump_file_name): Likewise.
15745 (dump_enable_all): Likewise.
15746 (dump_switch_p_1): Likewise.
15747 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15748 * dumpfile.h (enum dump_kind): New enum type.
15749 (struct dump_file_info): Create constructor and
15750 format fields and comments.
15751 * passes.c (pass_manager::register_one_dump_file):
15753 * statistics.c (statistics_early_init): Likewise.
15754 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15755 TDF_TREE with TDF_SLIM.
15756 (gather_memory_references_ref): Likewise.
15758 2017-05-18 Martin Liska <mliska@suse.cz>
15760 * vec.h (struct vnull): Use it.
15762 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15764 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15765 (true_predicate, false_predicate, true_predicate_p,
15766 false_predicate_p): Remove.
15767 (single_cond_predicate, not_inlined_predicate): Turn to member function
15769 (add_condition): Update.
15770 (add_clause): Turn to...
15771 (predicate::add_clause): ... this one; update; allow passing NULL
15773 (and_predicates): Turn to ...
15774 (predicate::operator &=): ... this one.
15775 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
15776 (or_predicates): Turn to ...
15777 (predicate::or_with): ... this one.
15778 (evaluate_predicate): Turn to ...
15779 (predicate::evaluate): ... this one.
15780 (predicate_probability): Turn to ...
15781 (predicate::probability): ... this one.
15782 (dump_condition): Update.
15783 (dump_predicate): Turn to ...
15784 (predicate::dump): ... this one.
15785 (account_size_time): Update.
15786 (edge_set_predicate): Update.
15787 (set_hint_predicate): UPdate.
15788 (evaluate_conditions_for_known_args): Update.
15789 (evaluate_properties_for_edge): Update.
15790 (remap_predicate_after_duplication): Turn to...
15791 (predicate::remap_after_duplication): ... this one.
15792 (remap_hint_predicate_after_duplication): Update.
15793 (inline_summary_t::duplicate): UPdate.
15794 (dump_inline_edge_summary): Update.
15795 (dump_inline_summary): Update.
15796 (set_cond_stmt_execution_predicate): Update.
15797 (set_switch_stmt_execution_predicate): Update.
15798 (compute_bb_predicates): Update.
15799 (will_be_nonconstant_expr_predicate): Update.
15800 (will_be_nonconstant_predicate): Update.
15801 (phi_result_unknown_predicate): Update.
15802 (predicate_for_phi_result): Update.
15803 (array_index_predicate): Update.
15804 (estimate_function_body_sizes): Update.
15805 (estimate_node_size_and_time): Update.
15806 (estimate_ipcp_clone_size_and_time): Update.
15807 (remap_predicate): Rename to ...
15808 (predicate::remap_after_inlining): ... this one.
15809 (remap_hint_predicate): Update.
15810 (inline_merge_summary): Update.
15811 (inline_update_overall_summary): Update.
15812 (estimate_size_after_inlining): Update.
15813 (read_predicate): Rename to ...
15814 (predicate::stream_in): ... this one.
15815 (read_inline_edge_summary): Update.
15816 (write_predicate): Rename to ...
15817 (predicate::stream_out): ... this one.
15818 (write_inline_edge_summary): Update.
15819 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
15820 (clause_t): Turn to uint32_t
15821 (predicate): Turn to class; implement constructor and operators
15823 (size_time_entry): Update.
15824 (inline_summary): Update.
15825 (inline_edge_summary): Update.
15827 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
15829 * fold-const.c (fold_binary_loc): Move transformation...
15830 * match.pd (C - X CMP X): ... here.
15832 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
15834 * config/sparc/sparc.c (sparc_option_override): Set function
15835 alignment for -mcpu=niagara7 to 64 to match the I$ line.
15836 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
15838 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
15840 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
15842 2017-05-18 Marek Polacek <polacek@redhat.com>
15845 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
15846 (pass_ubsan::execute): Call gimple_assign_single_p instead of
15847 gimple_assign_load_p.
15849 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
15851 PR middle-end/80692
15852 * real.c (do_compare): Give decimal_do_compare preference over
15853 comparing just the signs.
15855 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
15857 * doc/md.texi (Canonicalization of Instructions): Describe the
15858 canonical form of instructions that inherently set a condition
15861 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
15863 PR middle-end/80775
15864 * tree-cfg.c: Move deletion of unreachable case statements to after
15865 the merging of consecutive case labels.
15867 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
15869 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
15870 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
15871 restoring of callee-saved registers.
15873 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
15875 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
15876 * config/visium/visium.c (single_set_and_flags): Likewise.
15877 * config/visium/visium.md (Substitutions): Likewise.
15879 2017-05-17 Martin Liska <mliska@suse.cz>
15881 * cfg.c: Introduce dump_flags_t type and
15882 use it instead of int type.
15884 * cfghooks.c: Likewise.
15885 * cfghooks.h (struct cfg_hooks): Likewise.
15886 * cfgrtl.c: Likewise.
15887 * cfgrtl.h: Likewise.
15888 * cgraph.c (cgraph_node::get_body): Likewise.
15889 * coretypes.h: Likewise.
15890 * domwalk.c: Likewise.
15891 * domwalk.h: Likewise.
15892 * dumpfile.c (struct dump_option_value_info): Likewise.
15893 (dump_enable_all): Likewise.
15894 (dump_switch_p_1): Likewise.
15895 (opt_info_switch_p): Likewise.
15896 * dumpfile.h (enum tree_dump_index): Likewise.
15897 (struct dump_file_info): Likewise.
15898 * genemit.c: Likewise.
15899 * generic-match-head.c: Likewise.
15900 * gengtype.c (open_base_files): Likewise.
15901 * gimple-pretty-print.c: Likewise.
15902 * gimple-pretty-print.h: Likewise.
15903 * graph.c (print_graph_cfg): Likewise.
15904 * graphite-scop-detection.c (dot_all_sese): Likewise.
15905 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
15906 * loop-unroll.c (report_unroll): Likewise.
15907 * passes.c (pass_manager::register_one_dump_file): Likewise.
15908 * print-tree.c: Likewise.
15909 * statistics.c: Likewise.
15910 * tree-cfg.c: Likewise.
15911 * tree-cfg.h: Likewise.
15912 * tree-dfa.c: Likewise.
15913 * tree-dfa.h: Likewise.
15914 * tree-dump.c (dump_function): Likewise.
15915 * tree-dump.h (struct dump_info): Likewise.
15916 * tree-pretty-print.c: Likewise.
15917 * tree-pretty-print.h: Likewise.
15918 * tree-ssa-live.c: Likewise.
15919 * tree-ssa-live.h: Likewise.
15920 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
15921 * tree-vect-loop.c: Likewise.
15922 * tree-vect-slp.c: Likewise.
15924 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
15925 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15927 PR tree-optimization/80457
15928 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
15929 of all arguments to a statement as scalar_to_vec operations.
15930 (vectorizable_call): Adjust call to vect_model_simple_cost for
15932 (vectorizable_conversion): Likewise.
15933 (vectorizable_assignment): Likewise.
15934 (vectorizable_shift): Likewise.
15935 (vectorizable_operation): Likewise.
15936 (vectorizable_comparison): Likewise.
15937 (vect_is_simple_cond): Record the def types for operands.
15938 (vectorizable_condition): Likewise, call vect_model_simple_cost.
15939 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
15940 for statement argument count.
15942 2017-05-16 Carl Love <cel@us.ibm.com>
15944 * config/rs6000/rs6000-c: Add support for built-in functions
15945 vector unsigned long long vec_bperm (vector unsigned long long,
15946 vector unsigned char)
15947 vector signed long long vec_mule (vector signed int,
15949 vector unsigned long long vec_mule (vector unsigned int,
15950 vector unsigned int)
15951 vector signed long long vec_mulo (vector signed int,
15953 vector unsigned long long vec_mulo (vector unsigned int,
15954 vector unsigned int)
15955 vector signed char vec_sldw (vector signed char,
15956 vector signed char,
15958 vector unsigned char vec_sldw (vector unsigned char,
15959 vector unsigned char,
15961 vector signed short vec_sldw (vector signed short,
15962 vector signed short,
15964 vector unsigned short vec_sldw (vector unsigned short,
15965 vector unsigned short,
15967 vector signed int vec_sldw (vector signed int,
15970 vector unsigned int vec_sldw (vector unsigned int,
15971 vector unsigned int,
15973 vector signed long long vec_sldw (vector signed long long,
15974 vector signed long long,
15976 vector unsigned long long vec_sldw (vector unsigned long long,
15977 vector unsigned long long,
15979 * config/rs6000/rs6000-c: Add support for built-in functions
15980 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
15981 * config/rs6000/altivec.h: Add defintion for vec_sldw.
15982 * doc/extend.texi: Update the built-in documentation for the
15983 new built-in functions.
15985 2017-05-16 Marek Polacek <polacek@redhat.com>
15989 * tree.c (save_expr): Don't fold the expression.
15991 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
15993 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
15994 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
15995 and (?*y,m). Update insn attributes.
15997 2017-05-16 Martin Liska <mliska@suse.cz>
15999 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
16000 flags argument of print_gimple_stmt, print_gimple_expr,
16001 print_generic_stmt and print_generic_expr.
16002 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
16003 * coretypes.h: Likewise.
16004 * except.c (dump_eh_tree): Likewise.
16005 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
16006 * gimple-pretty-print.h: Likewise.
16007 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
16008 (backprop::push_to_worklist): Likewise.
16009 (backprop::pop_from_worklist): Likewise.
16010 (backprop::process_use): Likewise.
16011 (backprop::intersect_uses): Likewise.
16012 (note_replacement): Likewise.
16013 * gimple-ssa-store-merging.c
16014 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
16015 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
16016 (pass_store_merging::execute): Likewise.
16017 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
16018 (ssa_base_cand_dump_callback): Likewise.
16019 (dump_incr_vec): Likewise.
16020 (replace_refs): Likewise.
16021 (replace_mult_candidate): Likewise.
16022 (create_add_on_incoming_edge): Likewise.
16023 (create_phi_basis): Likewise.
16024 (insert_initializers): Likewise.
16025 (all_phi_incrs_profitable): Likewise.
16026 (introduce_cast_before_cand): Likewise.
16027 (replace_one_candidate): Likewise.
16028 * gimplify.c (gimplify_expr): Likewise.
16029 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
16030 (set_rename): Likewise.
16031 (rename_uses): Likewise.
16032 (copy_loop_phi_nodes): Likewise.
16033 (add_close_phis_to_merge_points): Likewise.
16034 (copy_loop_close_phi_args): Likewise.
16035 (copy_cond_phi_args): Likewise.
16036 (graphite_copy_stmts_from_block): Likewise.
16037 (translate_pending_phi_nodes): Likewise.
16038 * graphite-poly.c (print_pdr): Likewise.
16039 (dump_gbb_cases): Likewise.
16040 (dump_gbb_conditions): Likewise.
16041 (print_scop_params): Likewise.
16042 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
16043 (build_cross_bb_scalars_use): Likewise.
16044 (gather_bbs::before_dom_children): Likewise.
16045 * hsa-dump.c (dump_hsa_immed): Likewise.
16046 * ipa-cp.c (print_ipcp_constant_value): Likewise.
16047 (get_replacement_map): Likewise.
16048 * ipa-inline-analysis.c (dump_condition): Likewise.
16049 (estimate_function_body_sizes): Likewise.
16050 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
16051 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
16052 * ipa-prop.c (ipa_dump_param): Likewise.
16053 (ipa_print_node_jump_functions_for_edge): Likewise.
16054 (ipa_modify_call_arguments): Likewise.
16055 (ipa_modify_expr): Likewise.
16056 (ipa_dump_param_adjustments): Likewise.
16057 (ipa_dump_agg_replacement_values): Likewise.
16058 (ipcp_modif_dom_walker::before_dom_children): Likewise.
16059 * ipa-pure-const.c (check_stmt): Likewise.
16060 (pass_nothrow::execute): Likewise.
16061 * ipa-split.c (execute_split_functions): Likewise.
16062 * omp-offload.c (dump_oacc_loop_part): Likewise.
16063 (dump_oacc_loop): Likewise.
16064 * trans-mem.c (tm_log_emit): Likewise.
16065 (tm_memopt_accumulate_memops): Likewise.
16066 (dump_tm_memopt_set): Likewise.
16067 (dump_tm_memopt_transform): Likewise.
16068 * tree-cfg.c (gimple_verify_flow_info): Likewise.
16069 (print_loop): Likewise.
16070 * tree-chkp-opt.c (chkp_print_addr): Likewise.
16071 (chkp_gather_checks_info): Likewise.
16072 (chkp_get_check_result): Likewise.
16073 (chkp_remove_check_if_pass): Likewise.
16074 (chkp_use_outer_bounds_if_possible): Likewise.
16075 (chkp_reduce_bounds_lifetime): Likewise.
16076 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
16077 (chkp_mark_completed_bounds): Likewise.
16078 (chkp_register_incomplete_bounds): Likewise.
16079 (chkp_mark_invalid_bounds): Likewise.
16080 (chkp_maybe_copy_and_register_bounds): Likewise.
16081 (chkp_build_returned_bound): Likewise.
16082 (chkp_get_bound_for_parm): Likewise.
16083 (chkp_build_bndldx): Likewise.
16084 (chkp_get_bounds_by_definition): Likewise.
16085 (chkp_generate_extern_var_bounds): Likewise.
16086 (chkp_get_bounds_for_decl_addr): Likewise.
16087 * tree-chrec.c (chrec_apply): Likewise.
16088 * tree-data-ref.c (dump_data_reference): Likewise.
16089 (dump_subscript): Likewise.
16090 (dump_data_dependence_relation): Likewise.
16091 (analyze_overlapping_iterations): Likewise.
16092 * tree-inline.c (expand_call_inline): Likewise.
16093 (tree_function_versioning): Likewise.
16094 * tree-into-ssa.c (dump_defs_stack): Likewise.
16095 (dump_currdefs): Likewise.
16096 (dump_names_replaced_by): Likewise.
16097 (dump_update_ssa): Likewise.
16098 (update_ssa): Likewise.
16099 * tree-object-size.c (pass_object_sizes::execute): Likewise.
16100 * tree-parloops.c (build_new_reduction): Likewise.
16101 (try_create_reduction_list): Likewise.
16102 (ref_conflicts_with_region): Likewise.
16103 (oacc_entry_exit_ok_1): Likewise.
16104 (oacc_entry_exit_single_gang): Likewise.
16105 * tree-pretty-print.h: Likewise.
16106 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
16107 (get_scalar_evolution): Likewise.
16108 (add_to_evolution): Likewise.
16109 (get_loop_exit_condition): Likewise.
16110 (analyze_evolution_in_loop): Likewise.
16111 (analyze_initial_condition): Likewise.
16112 (analyze_scalar_evolution): Likewise.
16113 (instantiate_scev): Likewise.
16114 (number_of_latch_executions): Likewise.
16115 (gather_chrec_stats): Likewise.
16116 (final_value_replacement_loop): Likewise.
16117 (scev_const_prop): Likewise.
16118 * tree-sra.c (dump_access): Likewise.
16119 (disqualify_candidate): Likewise.
16120 (create_access): Likewise.
16121 (reject): Likewise.
16122 (maybe_add_sra_candidate): Likewise.
16123 (create_access_replacement): Likewise.
16124 (analyze_access_subtree): Likewise.
16125 (analyze_all_variable_accesses): Likewise.
16126 (sra_modify_assign): Likewise.
16127 (initialize_constant_pool_replacements): Likewise.
16128 (find_param_candidates): Likewise.
16129 (decide_one_param_reduction): Likewise.
16130 (replace_removed_params_ssa_names): Likewise.
16131 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
16132 * tree-ssa-copy.c (dump_copy_of): Likewise.
16133 (copy_prop_visit_cond_stmt): Likewise.
16134 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
16135 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
16136 (record_equivalences_from_stmt): Likewise.
16137 * tree-ssa-dse.c (compute_trims): Likewise.
16138 (delete_dead_call): Likewise.
16139 (delete_dead_assignment): Likewise.
16140 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
16141 (forward_propagate_into_cond): Likewise.
16142 (pass_forwprop::execute): Likewise.
16143 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16144 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
16146 (move_computations_worker): Likewise.
16147 (execute_sm): Likewise.
16148 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
16149 (remove_exits_and_undefined_stmts): Likewise.
16150 (remove_redundant_iv_tests): Likewise.
16151 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
16152 (adjust_iv_update_pos): Likewise.
16153 * tree-ssa-math-opts.c (bswap_replace): Likewise.
16154 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
16155 (value_replacement): Likewise.
16156 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
16157 * tree-ssa-pre.c (print_pre_expr): Likewise.
16158 (get_representative_for): Likewise.
16159 (create_expression_by_pieces): Likewise.
16160 (insert_into_preds_of_block): Likewise.
16161 (eliminate_insert): Likewise.
16162 (eliminate_dom_walker::before_dom_children): Likewise.
16163 (eliminate): Likewise.
16164 (remove_dead_inserted_code): Likewise.
16165 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
16166 * tree-ssa-reassoc.c (get_rank): Likewise.
16167 (eliminate_duplicate_pair): Likewise.
16168 (eliminate_plus_minus_pair): Likewise.
16169 (eliminate_not_pairs): Likewise.
16170 (undistribute_ops_list): Likewise.
16171 (eliminate_redundant_comparison): Likewise.
16172 (update_range_test): Likewise.
16173 (optimize_range_tests_var_bound): Likewise.
16174 (optimize_vec_cond_expr): Likewise.
16175 (rewrite_expr_tree): Likewise.
16176 (rewrite_expr_tree_parallel): Likewise.
16177 (linearize_expr): Likewise.
16178 (break_up_subtract): Likewise.
16179 (linearize_expr_tree): Likewise.
16180 (attempt_builtin_powi): Likewise.
16181 (attempt_builtin_copysign): Likewise.
16182 (transform_stmt_to_copy): Likewise.
16183 (transform_stmt_to_multiply): Likewise.
16184 (dump_ops_vector): Likewise.
16185 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
16186 (print_scc): Likewise.
16187 (set_ssa_val_to): Likewise.
16188 (visit_reference_op_store): Likewise.
16189 (visit_use): Likewise.
16190 (sccvn_dom_walker::before_dom_children): Likewise.
16191 (run_scc_vn): Likewise.
16192 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16194 (expr_hash_elt::print): Likewise.
16195 (const_and_copies::pop_to_marker): Likewise.
16196 (const_and_copies::record_const_or_copy_raw): Likewise.
16197 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
16198 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
16199 (dump_predicates): Likewise.
16200 (find_uninit_use): Likewise.
16201 (warn_uninitialized_phi): Likewise.
16202 (pass_late_warn_uninitialized::execute): Likewise.
16203 * tree-ssa.c (verify_vssa): Likewise.
16204 (verify_ssa): Likewise.
16205 (maybe_optimize_var): Likewise.
16206 * tree-vrp.c (dump_value_range): Likewise.
16207 (dump_all_value_ranges): Likewise.
16208 (dump_asserts_for): Likewise.
16209 (register_edge_assert_for_2): Likewise.
16210 (vrp_visit_cond_stmt): Likewise.
16211 (vrp_visit_switch_stmt): Likewise.
16212 (vrp_visit_stmt): Likewise.
16213 (vrp_visit_phi_node): Likewise.
16214 (simplify_cond_using_ranges_1): Likewise.
16215 (fold_predicate_in): Likewise.
16216 (evrp_dom_walker::before_dom_children): Likewise.
16217 (evrp_dom_walker::push_value_range): Likewise.
16218 (evrp_dom_walker::pop_value_range): Likewise.
16219 (execute_early_vrp): Likewise.
16221 2017-05-16 Richard Biener <rguenther@suse.de>
16223 * dwarf2out.c (loc_list_from_tree_1): Do not create
16224 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
16226 2017-05-16 Richard Biener <rguenther@suse.de>
16228 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
16230 (note_variable_value_in_expr): If we resolved the decl ref
16231 do not push to the stack.
16233 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
16235 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
16236 operations in fast-math mode.
16237 (vaddq_f16): Likewise.
16238 (vmul_f16): Likewise.
16239 (vmulq_f16): Likewise.
16240 (vsub_f16): Likewise.
16241 (vsubq_f16): Likewise.
16242 * config/arm/neon.md (add<mode>3): New.
16244 (fma:<VH:mode>3): New. Also remove outdated comment.
16247 2017-05-16 Martin Liska <mliska@suse.cz>
16251 * ipa-devirt.c (warn_types_mismatch): Fix typo.
16252 (odr_types_equivalent_p): Likewise.
16254 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
16256 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
16258 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
16261 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
16262 non-interunit SSE move alternatives with '?'.
16263 (zero-extendsidi peephole2): New peephole to skip intermediate
16264 general register in SSE zero-extend sequence.
16266 2017-05-15 Jeff Law <law@redhat.com>
16268 * reorg.c (relax_delay_slots): Create a new variable to hold
16269 the temporary target rather than clobbering TARGET_LABEL.
16271 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
16272 missing argument to extract_bit_field call.
16273 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
16275 2017-05-15 Martin Liska <mliska@suse.cz>
16278 * gcc.c (process_command): Do not allow empty argument of -o option.
16280 2017-05-15 Renlin Li <renlin.li@arm.com>
16282 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
16283 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
16284 * config/aarch64/constraints.md (Usf): Add long call check.
16285 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
16286 (call_value): Likewise.
16287 (sibcall): Likewise.
16288 (sibcall_value): Likewise.
16290 (call_value_insn): New.
16291 (sibcall_insn): Update rtx pattern.
16292 (sibcall_value_insn): Likewise.
16293 (call_internal): Remove.
16294 (call_value_internal): Likewise.
16295 (sibcall_internal): Likewise.
16296 (sibcall_value_internal): Likewise.
16297 (call_reg): Likewise.
16298 (call_symbol): Likewise.
16299 (call_value_reg): Likewise.
16300 (call_value_symbol): Likewise.
16302 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
16305 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
16307 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
16309 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
16310 compatible with CCGOCmode and with CCZmode.
16312 2017-05-14 Martin Sebor <msebor@redhat.com>
16314 PR middle-end/77671
16315 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
16316 (gimple_fold_builtin_snprintf): Same.
16317 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
16318 (gimple_fold_builtin_snprintf): Same.
16319 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
16320 of character types.
16321 (is_call_safe): New function.
16322 (try_substitute_return_value): Call it.
16323 (try_simplify_call): New function.
16324 (pass_sprintf_length::handle_gimple_call): Call it.
16326 2017-05-14 Martin Sebor <msebor@redhat.com>
16328 PR middle-end/80669
16329 * builtins.c (expand_builtin_stpncpy): Simplify.
16331 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
16333 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
16334 * config/i386/i386.h
16335 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
16336 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
16337 (struct machine_function): Add new members call_ms2sysv,
16338 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
16339 (struct machine_frame_state): New fields sp_realigned and
16340 sp_realigned_offset.
16341 * config/i386/i386.c
16342 (enum xlogue_stub): New enum.
16343 (enum xlogue_stub_sets): New enum.
16344 (class xlogue_layout): New class.
16345 (struct ix86_frame): New fields stack_realign_allocate_offset,
16346 stack_realign_offset and outlined_save_offset. Modify comments to
16347 detail stack layout when using out-of-line stubs.
16348 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
16349 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
16350 -mcall-ms2sysv-xlogues.
16351 (stub_managed_regs): New static variable.
16352 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
16353 registers managed by out-of-line stub.
16354 (disable_call_ms2sysv_xlogues): New function.
16355 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
16356 m->call_ms2sysv when appropriate and compute frame layout for
16358 (sp_valid_at, fp_valid_at): New inline functions.
16359 (choose_basereg): New function.
16360 (choose_baseaddr): Add align parameter, use choose_basereg and modify
16362 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
16363 Use align parameter of choose_baseaddr to generated aligned SSE movs
16365 (pro_epilogue_adjust_stack): Modify to track
16366 machine_frame_state::sp_realigned.
16367 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
16368 (ix86_nsaved_sseregs): Likewise.
16369 (ix86_emit_save_regs): Likewise.
16370 (ix86_emit_save_regs_using_mov): Likewise.
16371 (ix86_emit_save_sse_regs_using_mov): Likewise.
16372 (get_scratch_register_on_entry): Likewise.
16373 (gen_frame_set): New function.
16374 (gen_frame_load): Likewise.
16375 (gen_frame_store): Likewise.
16376 (emit_outlined_ms2sysv_save): Likewise.
16377 (emit_outlined_ms2sysv_restore): Likewise.
16378 (ix86_expand_prologue): Modify stack re-alignment code and call
16379 emit_outlined_ms2sysv_save when appropriate.
16380 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
16381 parameter rtx_insn *insn, which allows the function to be used to only
16382 generate the notes.
16383 (ix86_expand_epilogue): Modify validity checks of frame and stack
16384 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
16385 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
16386 * config/i386/predicates.md
16387 (save_multiple): New predicate.
16388 (restore_multiple): Likewise.
16389 * config/i386/sse.md
16390 (save_multiple<mode>): New pattern.
16391 (save_multiple_realign<mode>): Likewise.
16392 (restore_multiple<mode>): Likewise.
16393 (restore_multiple_and_return<mode>): Likewise.
16394 (restore_multiple_leave_return<mode>): Likewise.
16395 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
16397 2017-05-14 Julia Koval <julia.koval@intel.com>
16399 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
16400 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
16401 (__builtin_ia32_xsetbv): New builtins.
16402 * config/i386/i386.c (ix86_expand_special_args_builtin):
16404 (ix86_expand_builtin): Special expand for new intrinsics.
16405 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
16406 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
16407 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
16409 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16411 * cfganal.c (inverted_post_order_compute): Change argument type
16413 * cfganal.h (inverted_post_order_compute): Adjust prototype.
16414 * df-core.c (rest_of_handle_df_initialize): Adjust.
16415 (rest_of_handle_df_finish): Likewise.
16416 (df_analyze_1): Likewise.
16417 (df_analyze): Likewise.
16418 (loop_inverted_post_order_compute): Change argument to be a vec *.
16419 (df_analyze_loop): Adjust.
16420 (df_get_n_blocks): Likewise.
16421 (df_get_postorder): Likewise.
16422 * df.h (struct df_d): Change field to be a vec.
16423 * lcm.c (compute_laterin): Adjust.
16424 (compute_available): Likewise.
16425 * lra-lives.c (lra_create_live_ranges_1): Likewise.
16426 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
16427 * tree-ssa-pre.c (compute_antic): Likewise.
16429 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16431 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16432 (depth_first_search::depth_first_search): Change structure init
16433 function to this constructor.
16434 (depth_first_search::add_bb): Rename function to this member.
16435 (depth_first_search::execute): Likewise.
16436 (flow_dfs_compute_reverse_finish): Adjust.
16438 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16440 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16441 (longest_simple_path): Likewise.
16442 * shrink-wrap.c (spread_components): Likewise.
16443 (disqualify_problematic_components): Likewise.
16444 (emit_common_heads_for_components): Likewise.
16445 (emit_common_tails_for_components): Likewise.
16446 (insert_prologue_epilogue_for_components): Likewise.
16448 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16450 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16453 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16455 * df-core.c (df_set_blocks): Start using auto_bitmap.
16456 (df_compact_blocks): Likewise.
16457 * df-problems.c (df_rd_confluence_n): Likewise.
16458 * df-scan.c (df_insn_rescan_all): Likewise.
16459 (df_process_deferred_rescans): Likewise.
16460 (df_update_entry_block_defs): Likewise.
16461 (df_update_exit_block_uses): Likewise.
16462 (df_entry_block_bitmap_verify): Likewise.
16463 (df_exit_block_bitmap_verify): Likewise.
16464 (df_scan_verify): Likewise.
16465 * lra-constraints.c (lra_constraints): Likewise.
16466 (undo_optional_reloads): Likewise.
16467 (lra_undo_inheritance): Likewise.
16468 * lra-remat.c (calculate_gen_cands): Likewise.
16469 (do_remat): Likewise.
16470 * lra-spills.c (assign_spill_hard_regs): Likewise.
16471 (spill_pseudos): Likewise.
16472 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16473 (bitmap_set_subtract_values): Likewise.
16475 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16477 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16478 management with auto_bitmap.
16479 (fix_inter_tick): Likewise.
16480 (fix_recovery_deps): Likewise.
16481 * ira.c (add_store_equivs): Likewise.
16482 (find_moveable_pseudos): Likewise.
16483 (split_live_ranges_for_shrink_wrap): Likewise.
16484 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16485 (rtx_reuse_manager::seen_def_p): Likewise.
16486 (rtx_reuse_manager::set_seen_def): Likewise.
16487 * print-rtl.h (class rtx_reuse_manager): Likewise.
16489 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16491 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16493 (migrate_btr_def): Likewise.
16494 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16495 * df-core.c (loop_post_order_compute): Likewise.
16496 (loop_inverted_post_order_compute): Likewise.
16497 * hsa-common.h: Likewise.
16498 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16499 * init-regs.c (initialize_uninitialized_regs): Likewise.
16500 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16501 (inline_small_functions): Likewise.
16502 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16503 * ira.c (combine_and_move_insns): Likewise.
16504 (build_insn_chain): Likewise.
16505 * loop-invariant.c (find_invariants): Likewise.
16506 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16507 * predict.c (tree_predict_by_opcode): Likewise.
16508 (predict_paths_leading_to): Likewise.
16509 (predict_paths_leading_to_edge): Likewise.
16510 (estimate_loops_at_level): Likewise.
16511 (estimate_loops): Likewise.
16512 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16513 (spread_components): Likewise.
16514 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16515 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16516 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16517 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16518 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16519 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16520 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16521 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16522 (mark_threaded_blocks): Likewise.
16523 (thread_through_all_blocks): Likewise.
16524 * tree-ssa.c (verify_ssa): Likewise.
16525 (execute_update_addresses_taken): Likewise.
16526 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16528 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16530 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16532 (post_order_compute): Likewise.
16533 (inverted_post_order_compute): Likewise.
16534 (pre_and_rev_post_order_compute_fn): Likewise.
16536 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16538 * genrecog.c (int_set::int_set): Explicitly construct our
16539 auto_vec base class.
16540 * vec.h (auto_vec::auto_vec): New constructor.
16542 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16544 * bitmap.h (class auto_bitmap): New constructor taking
16545 bitmap_obstack * argument.
16547 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16549 * bitmap.h (class auto_bitmap): Change type of m_bits to
16550 bitmap_head, and adjust ctor / dtor and member operators.
16552 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16554 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16555 when returned register mode doesn't match original mode.
16557 2017-05-12 Jeff Law <law@redhat.com>
16558 Jakub Jelinek <jakub@redhat.com>
16560 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16561 we look for cc setter after the compare-elim changes.
16562 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16563 within the vector to match what compare-elim now expects.
16564 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16565 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16567 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16568 after the compare-elim changes.
16569 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16570 the vector to match what compare-elim now expects.
16571 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16572 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16573 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16574 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16575 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16577 * config/visium/visium.c (single_set_and_flags): Fix where
16578 we look for cc setter after the compare-elim changes.
16579 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16580 with the vector to match what compare-elim now expects.
16581 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16582 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16583 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16584 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16585 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16586 (neg<mode>2_insn_set_overflow): Likewise.
16588 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16590 PR middle-end/79794
16591 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16592 maybe_expand_insn call, set ops[0].target. If still set after call,
16593 set alt_rtl. Add extra arg to recursive calls.
16594 (extract_bit_field): Add alt_rtl argument. Pass to
16596 * expmed.h (extract_bit_field): Fix prototype.
16597 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16598 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16599 to extract_bit_field_calls.
16600 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16601 Pass alt_rtl to extract_bit_field calls.
16602 * calls.c (store_unaligned_arguments_into_psuedos)
16603 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16604 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16606 * optabs.h (struct expand_operand): Add target bitfield.
16608 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16610 * compare-elim.c (try_eliminate_compare): Canonicalize
16611 operation with embedded compare to
16612 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16613 (set (reg) (operation)].
16615 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16617 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16620 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16621 cost of adding a carry flag for ADC instruction.
16622 [case MINUS]: Ignore the cost of subtracting a carry flag
16623 for SBB instruction.
16625 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16627 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16629 * config/rs6000/bmiintrin.h: New file.
16630 * config/rs6000/bmi2intrin.h: New file.
16631 * config/rs6000/x86intrin.h: New file.
16633 2017-05-12 Jeff Law <law@redhat.com>
16635 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16638 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16640 PR middle-end/80707
16641 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16643 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16645 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16646 early expansion of vector divide builtins.
16647 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16648 builtins identified as having unsigned arguments.
16650 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16652 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16653 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16654 expansion of vector logical operations (and, andc, or, xor,
16657 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16659 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16660 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16662 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16664 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16665 early GIMPLE expansion of vector multiplies.
16667 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16669 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
16670 TARGET_HAVE_MOVT conditional.
16671 (movt splitter): Likewise.
16673 2017-05-12 Richard Biener <rguenther@suse.de>
16675 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
16676 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16677 Fold all stmts not inplace.
16679 2017-05-12 Richard Biener <rguenther@suse.de>
16681 PR tree-optimization/80713
16682 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
16683 inserted_exprs bit for not removed stmts.
16685 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16687 PR middle-end/69921
16688 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16689 parallelized" attribute for parallelized OpenACC kernels.
16690 * omp-offload.c (execute_oacc_device_lower): Use it.
16692 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16693 Set "oacc kernels" attribute.
16694 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16695 parameter. Adjust all users.
16696 (oacc_fn_attrib_kernels_p): Remove function.
16697 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16698 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16699 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16700 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16701 assert "oacc kernels" attribute is set.
16703 2017-05-11 Carl Love <cel@us.ibm.com>
16705 * config/rs6000/rs6000-c: Add support for built-in functions
16706 vector unsigned char vec_popcnt (vector signed char)
16707 vector unsigned char vec_popcnt (vector unsigned char)
16708 vector unsigned short vec_popcnt (vector signed short)
16709 vector unsigned short vec_popcnt (vector unsigned short)
16710 vector unsigned int vec_popcnt (vector signed int)
16711 vector unsigned int vec_popcnt (vector unsigned int)
16712 vector unsigned long long vec_popcnt (vector signed long long)
16713 vector unsigned long long vec_popcnt (vector unsigned long long)
16714 vector signed long long vec_slo (vector signed long long,
16715 vector signed char)
16716 vector signed long long vec_slo (vector signed long long,
16717 vector unsigned char)
16718 vector unsigned long long vec_slo (vector unsigned long long,
16719 vector signed char)
16720 vector unsigned long long vec_slo (vector unsigned long long,
16721 vector unsigned char)
16722 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16723 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16724 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16725 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16726 * doc/extend.texi: Update the built-in documentation file for the
16727 new built-in functions.
16729 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16731 * attribs.h (sorted_attr_string): Move machine independent
16732 functions for target clone support from the i386 port to common
16733 code. Rename ix86_function_versions to common_function_versions.
16734 Rename make_name to make_unique_name.
16735 (common_function_versions): Likewise.
16736 (make_unique_name): Likewise.
16737 (make_dispatcher_decl): Likewise.
16738 (is_function_default_version): Likewise.
16739 * attribs.c (attr_strcmp): Likewise.
16740 (sorted_attr_string): Likewise.
16741 (common_function_versions): Likewise.
16742 (make_unique_name): Likewise.
16743 (make_dispatcher_decl): Likewise.
16744 (is_function_default_version): Likewise.
16745 * config/i386/i386.c (attr_strcmp): Likewise.
16746 (sorted_attr_string): Likewise.
16747 (ix86_function_versions): Likewise.
16748 (make_name): Likewise.
16749 (make_dispatcher_decl): Likewise.
16750 (is_function_default_version): Likewise.
16751 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16753 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16756 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16757 Account for direct move costs for vec_construct of integer
16760 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16763 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16764 (UNSPEC_STX_ATOMIC): Ditto.
16765 (loaddi_via_sse): New insn.
16766 (storedi_via_sse): Ditto.
16767 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16768 Update corresponding peephole2 patterns.
16769 (atomic_storedi_fpu): Ditto.
16771 2017-05-11 Julia Koval <julia.koval@intel.com>
16773 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
16774 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
16776 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
16777 (__builtin_ia32_rsqrt14ss_mask): New builtins.
16778 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
16780 2017-05-11 Nathan Sidwell <nathan@acm.org>
16782 * graphite-poly.c: Include dumpfile.h.
16784 * dumpfle.h (dump_function): Declare here ...
16785 * tree-dump.h (dump_function): ... not here.
16786 * dumpfile.c: #include tree-cfg.h.
16787 (dump_function): Move here from ...
16788 * tree-dump.c (dump_function): ... here.
16789 * gimplify.c: #include splay-tree.h, not tree-dump.h.
16790 * graphite-poly.c: Don't include tree-dump.h.
16791 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
16792 * print-tree.c: Likewise.
16793 * stor-layout.c: Likewise.
16794 * tree-nested.c: Likewise.
16796 * dumpfile.c (dump_start): Use TDF_FLAGS.
16797 (dump_enable_all): Fix TDF_KIND check thinko.
16799 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16801 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16802 array entries to represent two legal parameterizations of the
16803 overloaded __builtin_cmpb function, as represented by the
16804 P6_OV_BUILTIN_CMPB constant.
16805 (altivec_resolve_overloaded_builtin): Add special case handling
16806 for the __builtin_cmpb function, as represented by the
16807 P6_OV_BUILTIN_CMPB constant.
16808 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
16809 (BU_P6_64BIT_2): New macro.
16810 (BU_P6_OVERLOAD_2): New macro
16811 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
16812 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
16813 (CMPB): Add overload support to represent both 32-bit and 64-bit
16814 compare-bytes function.
16815 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
16816 support for TARGET_CMPB.
16817 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
16818 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16819 documentation of the __builtin_cmpb overloaded built-in function.
16821 2017-05-11 Richard Biener <rguenther@suse.de>
16823 PR tree-optimization/80705
16824 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
16825 bases are not vectorizable.
16827 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16829 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
16830 when counting register pressure.
16832 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16834 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
16835 (iv_ca_more_deps): Renamed to ...
16836 (iv_ca_compare_deps): ... this.
16837 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
16839 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16841 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
16843 (determine_group_iv_costs): ... here.
16844 (find_inv_vars_cb): Record inv var if it's not recorded before.
16846 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16848 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
16849 (get_shiftadd_cost): Ditto.
16851 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16853 * tree-ssa-address.c: Include header file.
16854 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
16856 (add_to_parts): Refactor.
16857 (addr_to_parts): New parameter. Update use of move_hint_to_base.
16858 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
16861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16863 PR tree-optimization/53090
16864 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
16866 (extract_cond_operands): Detect condition with IV on both sides
16867 and return COMP_IV_EXPR_2.
16868 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
16869 (rewrite_use_compare): Simplify by removing call to function
16870 extract_cond_operands.
16872 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16874 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
16875 (extract_cond_operands): Detect condition comparing against non-
16876 invariant bound and return appropriate enum value.
16877 (find_interesting_uses_cond): Update use of extract_cond_operands.
16878 Handle its return value accordingly.
16879 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
16881 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16883 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
16884 nonlinear iv_use computation in loop invariant sensitive way.
16886 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16888 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
16889 (find_iv_candidates): Call relate_compare_use_with_all_cands.
16891 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16893 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
16894 (dump_cand): Support iv_cand.inv_exprs.
16895 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
16897 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
16900 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16902 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
16904 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
16905 as local function. Include necessary header files.
16906 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
16908 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16910 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
16912 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16914 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
16915 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
16916 RSHIFT_EXPR and BIT_NOT_EXPR.
16918 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16920 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
16921 (adjust_setup_cost): New parameter supporting round up adjustment.
16922 (struct address_cost_data): Delete.
16923 (force_expr_to_var_cost): Don't bound cost with spill_cost.
16924 (split_address_cost, ptr_difference_cost): Delete.
16925 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
16926 (struct ainc_cost_data): New struct.
16927 (get_address_cost_ainc): New function.
16928 (get_address_cost, get_computation_cost): Reimplement.
16929 (determine_group_iv_cost_address): Record inv_expr for all uses of
16931 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
16932 (iv_ca_has_deps): Reimplemented to ...
16933 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
16935 (iv_ca_extend): Call iv_ca_more_deps.
16937 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16939 * tree-ssa-address.c (struct mem_address): Move to header file.
16940 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
16941 * tree-ssa-address.h (struct mem_address): Move from C file.
16942 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
16944 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16946 * tree-affine.h (aff_combination_type): New interface.
16947 (aff_combination_zero_p): Remove static.
16948 (aff_combination_const_p): New interface.
16949 (aff_combination_singleton_var_p): New interfaces.
16951 2017-05-11 Richard Biener <rguenther@suse.de>
16953 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16954 Skip unreachable blocks and destinations.
16955 (eliminate): Move stmt removal and fixup ...
16956 (fini_eliminate): ... here. Skip inserted exprs.
16957 (pass_pre::execute): Move fini_pre after fini_eliminate.
16958 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
16959 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
16960 PRE to get rid of dead code that has invalid SSA form and
16961 split critical edges again.
16963 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16965 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16967 2017-05-11 Richard Biener <rguenther@suse.de>
16969 * passes.c (execute_function_todo): Verify loops if they are
16970 said to be up-to-date.
16971 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
16972 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
16974 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
16977 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
16978 handle calling assemble_external ourself.
16981 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
16982 modes with zero size. Enhance comment.
16984 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16986 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
16987 built-ins for vec_xl and vec_xst with short and char pointer
16990 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16992 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
16993 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
16994 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
16995 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
16996 (_mm_maskz_min_round_ss): New intrinsics.
16997 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
16998 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
16999 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
17000 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
17001 (__builtin_ia32_minss_mask_round): New builtins.
17002 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17003 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17004 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
17006 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
17007 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
17009 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
17012 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
17014 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
17015 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
17016 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
17017 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
17018 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
17019 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
17020 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
17021 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17022 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
17023 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
17024 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
17025 (__builtin_ia32_mulss_mask_round): New builtins.
17026 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17027 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17028 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
17030 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
17031 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17033 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17036 2017-05-10 Julia Koval <julia.koval@intel.com>
17038 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
17039 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
17040 (_mm256_setr_m128i): New intrinsics.
17042 2017-05-10 Julia Koval <julia.koval@intel.com>
17044 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
17045 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
17046 (_mm_maskz_rcp14_ss): New intrinsics.
17047 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
17048 (__builtin_ia32_rcp14ss_mask): New builtins.
17049 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
17051 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
17053 PR tree-optimization/51513
17054 * tree-cfg.c (gimple_seq_unreachable_p): New function.
17055 (assert_unreachable_fallthru_edge_p): Use it.
17056 (group_case_labels_stmt): Likewise.
17057 * tree-cfg.h: Prototype it.
17058 * stmt.c: Include cfghooks.h and tree-cfg.h.
17059 (emit_case_dispatch_table) <gap_label>: New local variable.
17060 Use it to fill dispatch table gaps.
17061 Test for default_label before updating probabilities.
17062 (expand_case) <default_label>: Remove unneeded initialization.
17063 Test for unreachable default case statement and remove its edge.
17064 Set default_label accordingly.
17065 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
17067 2017-05-10 Carl Love <cel@us.ibm.com>
17069 * config/rs6000/rs6000-c: Add support for built-in functions
17070 vector signed char vec_neg (vector signed char)
17071 vector signed short int vec_neg (vector short int)
17072 vector signed int vec_neg (vector signed int)
17073 vector signed long long vec_neg (vector signed long long)
17074 vector float vec_neg (vector float)
17075 vector double vec_neg (vector double)
17076 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
17078 * config/rs6000/altivec.h: Add define for vec_neg
17079 * doc/extend.texi: Update the built-in documentation for the
17080 new built-in functions.
17082 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17084 PR tree-optimization/77644
17085 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
17087 2017-05-10 Nathan Sidwell <nathan@acm.org>
17089 * dumpfile.h (TDI_lang_all): New.
17090 (TDF_KIND): New. Renumber others
17091 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
17093 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
17095 (get_dump_file_name): Adjust suffix generation.
17096 (dump_enable_all): Use TDF_KIND.
17097 * doc/invoke.texi (-fdump-lang-all): Document.
17099 * dumpfile.h: Tabify.
17101 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
17104 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
17105 Move member access before delete.
17107 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
17109 * tree-inline.c (expand_call_inline): Split block at stmt
17112 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
17115 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
17116 are now unused after splitting mov{sf,sd}_hardfloat.
17117 (f32_lr2): Likewise.
17118 (f32_lm): Likewise.
17119 (f32_lm2): Likewise.
17120 (f32_li): Likewise.
17121 (f32_li2): Likewise.
17122 (f32_lv): Likewise.
17123 (f32_sr): Likewise.
17124 (f32_sr2): Likewise.
17125 (f32_sm): Likewise.
17126 (f32_sm2): Likewise.
17127 (f32_si): Likewise.
17128 (f32_si2): Likewise.
17129 (f32_sv): Likewise.
17130 (f32_dm): Likewise.
17131 (f32_vsx): Likewise.
17132 (f32_av): Likewise.
17133 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
17134 For movsf, order stores so the VSX stores occur before the GPR
17135 store which encourages the register allocator to use a traditional
17136 FPR instead of a GPR. For movsd, order the stores so that the GPR
17137 store comes before the VSX stores to allow the power6 to work.
17138 This is due to the power6 not having a 32-bit integer store
17139 instruction from a FPR.
17140 (movsf_hardfloat): Likewise.
17141 (movsd_hardfloat): Likewise.
17143 2017-05-09 Martin Sebor <msebor@redhat.com>
17145 PR translation/80280
17146 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
17149 PR translation/80280
17150 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
17151 data member added in r247778.
17152 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
17154 2017-05-09 Nathan Sidwell <nathan@acm.org>
17156 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
17158 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
17161 2017-05-09 Marek Polacek <polacek@redhat.com>
17163 * doc/invoke.texi: Fix typo.
17165 2017-05-09 Richard Biener <rguenther@suse.de>
17167 * tree-vrp.c (vrp_val_is_max): Adjust comment.
17168 (vrp_val_is_min): Likewise.
17169 (set_value_range_to_value): Likewise.
17170 (set_value_range_to_nonnegative): Likewise.
17171 (gimple_assign_nonzero_p): Likewise.
17172 (gimple_stmt_nonzero_p): Likewise.
17173 (vrp_int_const_binop): Likewise. Remove unreachable case.
17174 (adjust_range_with_scev): Adjust comments.
17175 (compare_range_with_value): Likewise.
17176 (extract_range_from_phi_node): Likewise.
17177 (test_for_singularity): Likewise.
17179 2017-05-09 Richard Biener <rguenther@suse.de>
17181 * tree-vrp.c (get_single_symbol): Add assert that we don't
17182 get overflowed constants as invariant part.
17183 (compare_values_warnv): Add comment before the TREE_NO_WARNING
17184 checks. Use wi::cmp instead of recursing for integer constants.
17185 (compare_values): Just ignore whether we assumed undefined
17186 overflow instead of failing the compare.
17187 (extract_range_for_var_from_comparison_expr): Add comment before the
17188 TREE_NO_WARNING sets.
17189 (test_for_singularity): Likewise.
17190 (extract_range_from_comparison): Do not disable optimization
17191 when we assumed undefined overflow.
17192 (extract_range_basic): Remove init of unused var.
17194 2017-05-09 Richard Biener <rguenther@suse.de>
17196 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
17197 (extract_range_from_multiplicative_op_1): Adjust.
17198 (extract_range_from_binary_expr_1): Use int_const_binop.
17200 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17203 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
17204 rs6000_store_data_bypass_p in seven define_bypass directives and
17205 in several comments.
17206 * config/rs6000/rs6000-protos.h: Add prototype for
17207 rs6000_store_data_bypass_p function.
17208 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
17209 function implements slightly different (rs6000-specific) semantics
17210 than store_data_bypass_p, returning false rather than aborting
17211 with assertion error when arguments do not satisfy the
17212 requirements of store data bypass.
17213 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
17214 rs6000_store_data_bypass_p.
17216 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
17218 * config/xtensa/xtensa-protos.h
17219 (xtensa_initial_elimination_offset): New declaration.
17220 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17221 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
17222 macro definition, add case for FRAME_POINTER_REGNUM when
17223 FRAME_GROWS_DOWNWARD.
17224 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
17225 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
17226 xtensa_initial_elimination_offset.
17228 2017-05-08 Nathan Sidwell <nathan@acm.org>
17230 * doc/invoke.texi: Alphabetize -fdump options.
17232 2017-05-08 Martin Sebor <msebor@redhat.com>
17234 PR translation/80280
17235 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
17237 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
17239 * target.def (compute_frame_layout): New optional target hook.
17240 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
17241 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
17242 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
17244 * reload1.c (verify_initial_elim_offsets): Likewise.
17245 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
17246 (use_simple_return_p): Call arm_compute_frame_layout if needed.
17247 (arm_get_frame_offsets): Split up into this ...
17248 (arm_compute_frame_layout): ... and this function.
17250 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
17252 * config/aarch64/constraints.md (Usa): New constraint.
17253 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
17255 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
17257 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
17258 with_multilib_list after it has been checked.
17260 2017-05-08 Richard Biener <rguenther@suse.de>
17262 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
17263 (bitmap_set_subtract_values): Likewise.
17265 2017-05-08 Richard Biener <rguenther@suse.de>
17267 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
17268 (gimple_assign_nonzero): ... this and remove strict_overflow_p
17270 (gimple_stmt_nonzero_warnv_p): Rename to ...
17271 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
17273 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
17274 (extract_range_basic): Adjust, do not disable propagation on
17275 strict overflow sensitive simplification.
17276 (vrp_visit_cond_stmt): Likewise.
17278 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
17280 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
17281 body size unconditionally.
17283 2017-05-07 Jeff Law <law@redhat.com>
17286 2017-05-06 Jeff Law <law@redhat.com>
17287 PR tree-optimization/78496
17288 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17291 PR tree-optimization/78496
17292 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17293 (simplify_stmt_using_ranges): Call it.
17294 (vrp_dom_walker::before_dom_children): Extract equivalences
17295 from an ASSERT_EXPR with an equality comparison against a
17298 2017-05-06 Jeff Law <law@redhat.com>
17300 PR tree-optimization/78496
17301 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17304 PR tree-optimization/78496
17305 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17306 (simplify_stmt_using_ranges): Call it.
17307 (vrp_dom_walker::before_dom_children): Extract equivalences
17308 from an ASSERT_EXPR with an equality comparison against a
17311 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17313 * lra-constraints.c (lra_copy_reg_equiv): New function.
17314 (split_reg): Use it to copy equivalence information from the
17315 original register to the spill register.
17317 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17319 PR rtl-optimization/75964
17320 * simplify-rtx.c (simplify_const_relational_operation): Remove
17321 invalid handling of comparisons of integer ABS.
17323 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
17325 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
17326 initialize to zero.
17327 (init_regs): Remove declaration.
17328 (function_arg_advance_32): Initialize error_p as boolean variable.
17330 2017-05-05 Nathan Sidwell <nathan@acm.org>
17332 * store-motion.c (remove_reachable_equiv_notes): Reformat long
17333 lines. Use for (;;).
17335 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17337 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
17338 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
17339 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
17340 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
17341 VF=2 that require versioning.
17343 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17345 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
17348 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17350 * diagnostic.h (diagnostic_override_option_index): Convert from
17351 macro to inline function.
17353 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17355 * diagnostic.c (last_module_changed_p): New function.
17356 (set_last_module): New function.
17357 (diagnostic_report_current_module): Convert macro usage to
17358 the above functions.
17359 * diagnostic.h (diagnostic_context::last_module): Strengthen
17360 from const line_map * to const line_map_ordinary *.
17361 (diagnostic_last_module_changed): Delete macro.
17362 (diagnostic_set_last_module): Delete macro.
17364 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17366 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
17367 with diagnostic_report_diagnostic.
17368 (diagnostic_n_impl_richloc): Likewise.
17369 * diagnostic.h (report_diagnostic): Delete macro.
17370 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
17371 with diagnostic_report_diagnostic.
17372 * substring-locations.c (format_warning_va): Likewise.
17374 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17376 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
17377 save/restor of format_spec. Move option-printing code to...
17378 (print_option_information): ...this new function, and
17379 reimplement by simply printing to the pretty_printer,
17380 rather than appending to the format string.
17382 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17384 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
17385 handling logic into...
17386 (update_effective_level_from_pragmas): ...this new function.
17388 2017-05-04 Andrew Waterman <andrew@sifive.com>
17390 * config/riscv/riscv.opt (mstrict-align): New option.
17391 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
17392 (SLOW_UNALIGNED_ACCESS): Define.
17393 (riscv_slow_unaligned_access): Declare.
17394 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
17396 (riscv_slow_unaligned_access): New variable.
17397 (rocket_tune_info): Set slow_unaligned_access to true.
17398 (optimize_size_tune_info): Set slow_unaligned_access to false.
17399 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
17400 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
17401 (riscv_option_override): Set riscv_slow_unaligned_access.
17402 * doc/invoke.texi: Add -mstrict-align to RISC-V.
17404 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
17406 * config/riscv/riscv.md: Unify indentation.
17408 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17413 * config/rs6000/rs6000.md (u code attribute): Add FIX and
17415 (extendsi<mode>2): Add support for doing sign extension via
17416 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
17417 don't have ISA 3.0 instructions.
17418 (extendsi<mode>2 splitter): Likewise.
17419 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
17420 generate the normal insns since SImode can now go in vector
17421 registers. Disallow the special UNSPECs needed for previous
17422 machines to hide SImode being used. Add new insns
17423 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
17424 (fix_trunc<mode>si2_stfiwx): Likewise.
17425 (fix_trunc<mode>si2_internal): Likewise.
17426 (fixuns_trunc<mode>si2): Likewise.
17427 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17428 (fctiw<u>z_<mode>_smallint): Likewise.
17429 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17430 of floating point to 32-bit integer from doing a direct move to
17431 the GPR registers to do a store.
17432 (fctiwz_<mode>): Break long line.
17434 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17436 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17437 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17438 (addr_list, addr_offset_valid_p): New.
17439 (split_address_groups): Check offset validity with above function.
17440 (gt-tree-ssa-loop-ivopts.h): Include header file.
17442 2017-05-05 Nathan Sidwell <nathan@acm.org>
17444 * config.gcc (arm*-*-*): Add missing 'fi'.
17446 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17448 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17449 included in -fopt-info does not matter.
17450 * doc/optinfo.texi (-fopt-info): Fix description of default
17451 behavour. Explicitly say order of options included in -fopt-info
17454 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17456 * config.gcc: Allow combinations of aprofile and rmprofile values for
17457 --with-multilib-list.
17458 * config/arm/t-multilib: New file.
17459 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17460 variables. Remove setting of ISA and floating-point ABI in
17461 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17462 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17463 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17464 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17466 * config/arm/t-rmprofile: Likewise except for the matches changes.
17467 * doc/install.texi (--with-multilib-list): Document the combination of
17468 aprofile and rmprofile values and warn about pitfalls in doing that.
17470 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17472 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17473 (movdi_aarch64): Likewise.
17475 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17477 PR tree-optimization/80632
17478 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17480 (build_arrays): Initialize it for virtual phis.
17481 (fix_phi_nodes): Use it for virtual phis.
17483 PR tree-optimization/80558
17484 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17485 [x, y] op z into [x op, y op z] for op & or | if conditions
17488 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17489 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17492 * config/arm/arm.md (use_literal_pool): Remove.
17493 (64-bit immediate split): No longer takes cost into consideration
17494 if arm_disable_literal_pool is enabled.
17495 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17496 used when arm_disable_literal_pool is enabled.
17497 (arm_max_const_double_inline_cost): Remove use of
17498 arm_disable_literal_pool.
17499 (push_minipool_fix): Add assert.
17500 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17501 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17502 (no_literal_pool_sf_immediate): New.
17504 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17506 PR tree-optimization/80613
17507 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17508 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17510 2017-05-05 Richard Biener <rguenther@suse.de>
17512 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17514 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17516 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17517 of this flag from insn conditions due to removal from r247495.
17519 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17521 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17523 (arm_early_store_addr_dep_ptr): Likewise.
17524 * config/arm/aarch-common-protos.h
17525 (arm_early_load_addr_dep_ptr): Add prototype.
17526 (arm_early_store_addr_dep_ptr): Likewise.
17527 * config/arm/cortex-a53.md: Add new bypasses.
17529 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17531 * tree.c (next_type_uid): Change type to unsigned.
17532 (type_hash_canon): Decrement back next_type_uid if
17533 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17534 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17537 2017-05-04 Martin Sebor <msebor@redhat.com>
17539 * builtins.c: Fix a trivial typo in a comment.
17541 PR middle-end/79234
17542 * builtins.c (check_sizes): Adjust to handle reading past the end.
17543 Avoid printing excessive upper bound of ranges. Use %E to print
17544 tree nodes instead of converting them to %wu.
17545 (expand_builtin_memchr): New function.
17546 (compute_dest_size): Rename...
17547 (compute_objsize): ...to this.
17548 (expand_builtin_memcpy): Adjust.
17549 (expand_builtin_mempcpy): Adjust.
17550 (expand_builtin_strcat): Adjust.
17551 (expand_builtin_strcpy): Adjust.
17552 (check_strncat_sizes): Adjust.
17553 (expand_builtin_strncat): Adjust.
17554 (expand_builtin_strncpy): Adjust and simplify.
17555 (expand_builtin_memset): Adjust.
17556 (expand_builtin_bzero): Adjust.
17557 (expand_builtin_memcmp): Adjust.
17558 (expand_builtin): Handle memcmp.
17559 (maybe_emit_chk_warning): Check strncat just once.
17561 2017-05-04 Martin Sebor <msebor@redhat.com>
17563 PR preprocessor/79214
17564 PR middle-end/79222
17565 PR middle-end/79223
17566 * builtins.c (check_sizes): Add inlining context and issue
17567 warnings even when -Wno-system-headers is set.
17568 (check_strncat_sizes): Same.
17569 (expand_builtin_strncat): Same.
17570 (expand_builtin_memmove): New function.
17571 (expand_builtin_stpncpy): Same.
17572 (expand_builtin): Handle memmove and stpncpy.
17574 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17576 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17577 which is not used any more.
17579 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17581 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17583 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17585 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17586 (cortexa53_tunings): Likewise.
17587 (cortexa57_tunings): Likewise.
17588 (cortexa72_tunings): Likewise.
17589 (cortexa73_tunings): Likewise.
17591 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17593 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17594 Set loop alignment to 8.
17596 2017-05-04 Martin Sebor <msebor@redhat.com>
17598 PR translation/80280
17599 * builtins.c (expand_builtin_object_size): Add missing quoting to
17600 %D and like directives.
17601 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17602 (hsa_type_for_tree_type): Same.
17603 (verify_function_arguments): Same.
17604 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17605 * varasm.c (get_section): Same.
17608 2017-05-04 Martin Sebor <msebor@redhat.com>
17610 PR translation/80280
17611 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17613 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17615 * config/aarch64/aarch64.c (generic_addrcost_table):
17616 Change HI/TI mode setting.
17618 2017-05-04 Martin Jambor <mjambor@suse.cz>
17620 PR tree-optimization/80622
17621 * tree-sra.c (comes_initialized_p): New function.
17622 (build_accesses_from_assign): Only set write lazily when
17623 comes_initialized_p is false.
17624 (analyze_access_subtree): Use comes_initialized_p.
17625 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17626 instead of testing for PARM_DECL.
17628 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17630 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17631 constraint on operand 0 to allow more general addressing modes.
17632 Adjust output template.
17633 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17635 * config/aarch64/aarch64-protos.h
17636 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17637 * config/aarch64/constraints.md (Dp): New address constraint.
17638 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17641 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17643 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17644 update use of estimate_ipcp_clone_size_and_time.
17645 (estimate_local_effects): Update use of
17646 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17647 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17648 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17649 Return nonspecialized time.
17651 2017-05-04 Richard Biener <rguenther@suse.de>
17653 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17654 for the last VUSE which def dominates the PHI. Directly call
17656 (get_continuation_for_phi_1): Remove.
17658 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17660 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17661 to explain the use of truncating division. Cap the number of
17662 iterations to the maximum given by nb_iterations_upper_bound,
17665 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17667 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17668 * configure: Regenerate.
17669 * config.in: Regenerate.
17670 * config/i386/driver-mingw32.c: new file.
17671 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
17672 * config.host: Link driver-mingw32.o on MinGW host.
17673 * doc/install.texi: Document new --enable-mingw-wildcard configure
17676 2017-05-04 Marek Polacek <polacek@redhat.com>
17678 PR tree-optimization/80612
17679 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
17681 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17682 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17684 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17685 (movt splitter): Likewise.
17686 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17687 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17688 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17689 block for Thumb-1 with MOVT.
17690 (thumb2_legitimate_address_p): Move code block ...
17691 (can_avoid_literal_pool_for_label_p): ... into this new function.
17692 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17694 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17695 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17696 "M-profile targets with the MOVT instruction".
17698 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17700 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17701 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17702 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17704 2017-05-04 Martin Liska <mliska@suse.cz>
17706 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17707 variable cond_code.
17709 2017-05-04 Richard Biener <rguenther@suse.de>
17711 * tree.c (array_at_struct_end_p): Handle arrays at struct
17712 end with flexarrays more conservatively. Refactor and treat
17713 arrays of arrays or aggregates more strict. Fix
17714 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17715 * tree.c (array_at_struct_end_p): Adjust prototype.
17716 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17717 * gimple-fold.c (get_range_strlen): Likewise.
17718 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17720 2017-05-04 Richard Biener <rguenther@suse.de>
17722 PR tree-optimization/31130
17723 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17725 (supports_overflow_infinity): Likewise.
17726 (is_negative_overflow_infinity): Likewise.
17727 (is_positive_overflow_infinity): Likewise.
17728 (is_overflow_infinity): Likewise.
17729 (stmt_overflow_infinity): Likewise.
17730 (overflow_infinity_range_p): Likewise.
17731 (usable_range_p): Remove as always returning true.
17732 (make_overflow_infinity): Remove.
17733 (negative_overflow_infinity): Likewise.
17734 (positive_overflow_infinity): Likewise.
17735 (avoid_overflow_infinity): Likewise.
17736 (set_value_range): Adjust accordingly.
17737 (set_value_range_to_nonnegative): Likewise, remove now unused
17738 overflow_infinity arg.
17739 (vrp_operand_equal_p): Adjust.
17740 (update_value_range): Likewise.
17741 (range_int_cst_singleton_p): Likewise.
17742 (operand_less_p): Likewise.
17743 (compare_values_warnv): Likewise.
17744 (extract_range_for_var_from_comparison_expr): Likewise.
17745 (vrp_int_const_binop): Likewise.
17746 (zero_nonzero_bits_from_vr): Likewise.
17747 (extract_range_from_multiplicative_op_1): Likewise.
17748 (extract_range_from_binary_expr_1): Likewise.
17749 (extract_range_from_unary_expr): Likewise.
17750 (extract_range_from_comparison): Likewise.
17751 (extract_range_basic): Likewise.
17752 (adjust_range_with_scev): Likewise.
17753 (compare_ranges): Likewise.
17754 (compare_range_with_value): Likewise.
17755 (dump_value_range): Likewise.
17756 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17758 (simplify_cond_using_ranges): Adjust.
17760 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17762 * brig-builtins.def: Added a builtin for class_f64.
17763 * builtin-types.def: Added a builtin type needed by class_f64.
17765 2017-05-03 Jason Merrill <jason@redhat.com>
17767 * timevar.def: Add TV_CONSTEXPR.
17769 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17771 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
17773 2017-05-03 Martin Jambor <mjambor@suse.cz>
17775 * ipa-prop.c (ipa_update_after_lto_read): Removed.
17776 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
17777 * ipa-cp.c (ipcp_propagate_stage): Do not call
17778 ipa_update_after_lto_read.
17779 * ipa-inline.c (ipa_inline): Likewise.
17781 2017-05-03 Martin Jambor <mjambor@suse.cz>
17783 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
17784 tag. Added a default constructor and a destructor.
17785 (ipa_edge_args_sum_t): New class;
17786 (ipa_edge_args_sum): Declare.
17787 (ipa_edge_args_vector): Remove declaration.
17788 (IPA_EDGE_REF): Use ipa_edge_args_sum.
17789 (ipa_free_edge_args_substructures): Remove declaration.
17790 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
17791 (ipa_edge_args_info_available_for_edge_p): Likewise.
17792 * ipa-prop.c (ipa_edge_args_vector): Removed.
17793 (edge_removal_hook_holder): Likewise.
17794 (edge_duplication_hook_holder): Likewise.
17795 (ipa_edge_args_sum): New variable.
17796 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
17797 ipa_edge_args_vector.
17798 (ipa_free_edge_args_substructures): Likewise.
17799 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
17800 ipa_edge_args_vector.
17801 (ipa_edge_removal_hook): Turned into method
17802 ipa_edge_args_sum_t::remove.
17803 (ipa_edge_duplication_hook): Turned into method
17804 ipa_edge_args_sum_t::duplicate.
17805 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
17806 registering edge hooks.
17807 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
17808 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
17809 ipa_edge_args_sum instead of ipa_edge_args_vector.
17810 * ipa-profile.c (ipa_profile): Likewise.
17812 2017-05-03 Martin Jambor <mjambor@suse.cz>
17814 * symbol-summary.h (function_summary): New method exists.
17815 (function_summary::symtab_removal): Deallocate through release.
17816 (call_summary): New class.
17817 (gt_ggc_mx): New overload.
17818 (gt_pch_nx): Likewise.
17819 (gt_pch_nx): Likewise.
17821 2017-05-03 Jeff Law <law@redhat.com>
17823 PR tree-optimization/78496
17824 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
17825 from simplify_cond_using_ranges. Split off code to walk
17826 backwards through casts into ...
17827 (simplify_cond_using_ranges_2): New function.
17828 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
17829 (execute_vrp): After identifying jump threads, call
17830 simplify_cond_using_ranges_2.
17832 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
17835 * ipa-inline.h (inline_summary): Add ctor.
17836 (create_ggc): Do not use ggc_cleared_alloc.
17838 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
17839 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17841 * gcc.c (handle_braces): Support escaping in switch matching
17843 * doc/invoke.texi (Spec Files): Document it.
17844 Remove superfluous @code markup in items.
17846 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17848 * diagnostic-show-locus.c (struct column_range): New struct.
17849 (get_affected_columns): New function.
17850 (get_printed_columns): New function.
17851 (struct correction): New struct.
17852 (correction::ensure_capacity): New function.
17853 (correction::ensure_terminated): New function.
17854 (struct line_corrections): New struct.
17855 (line_corrections::~line_corrections): New dtor.
17856 (line_corrections::add_hint): New function.
17857 (layout::print_trailing_fixits): Reimplement in terms of the new
17859 (selftest::test_overlapped_fixit_printing): New function.
17860 (selftest::diagnostic_show_locus_c_tests): Call it.
17862 2017-05-03 Nathan Sidwell <nathan@acm.org>
17864 Canonicalize canonical type hashing
17865 * tree.h (type_hash_canon_hash): Declare.
17866 * tree.c (type_hash_list, attribute_hash_list): Move into
17867 type_hash_canon_hash.
17868 (build_type_attribute_qual_variant): Break out hash code calc into
17869 type_hash_canon_hash.
17870 (type_hash_canon_hash): New. Generic type hash computation.
17871 (build_range_type_1, build_array_type_1, build_function_type,
17872 build_method_type_directly, build_offset_type, build_complex_type,
17873 make_vector_type): Call it.
17875 2017-05-03 Richard Biener <rguenther@suse.de>
17877 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
17878 When all DRs have unknown misaligned do not always peel
17879 when there is a store but apply the same costing model as if
17880 there were only loads.
17882 2017-05-03 Richard Biener <rguenther@suse.de>
17885 PR tree-optimization/80492
17886 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17887 compare_base_decls returning dont-know properly.
17889 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
17891 * config/arm/iterators.md (CCSI): New mode iterator.
17892 (arch): New mode attribute.
17893 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
17894 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
17895 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
17896 code iterator for success result mode.
17897 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
17898 the corresponding new insn generators.
17900 2017-05-03 Bin Cheng <bin.cheng@arm.com>
17903 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17904 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17906 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
17908 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
17909 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
17910 (DDR_A): Wrap DDR argument in brackets.
17911 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
17912 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
17913 (DDR_REVERSED_P): Likewise.
17915 2017-05-03 Jakub Jelinek <jakub@redhat.com>
17917 PR tree-optimization/79472
17918 * tree-switch-conversion.c (struct switch_conv_info): Add
17919 contiguous_range and default_case_nonstandard fields.
17920 (collect_switch_conv_info): Compute contiguous_range and
17921 default_case_nonstandard fields, don't clear final_bb if
17922 contiguous_range and only the default case doesn't have the required
17924 (check_all_empty_except_final): Set default_case_nonstandard instead
17925 of failing if contiguous_range and the default case doesn't have empty
17927 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
17928 and only the default case doesn't have the required constants. Skip
17930 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
17931 if default_case_nonstandard.
17932 (build_constructors): Build constant 1 just once. Assert that default
17933 values aren't inserted in between cases if contiguous_range. Skip
17935 (build_arrays): Skip virtual phis.
17936 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
17937 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
17938 Handle virtual phis.
17939 (gen_inbound_check): Handle default_case_nonstandard case.
17940 (process_switch): Adjust check_final_bb caller. Call
17941 gather_default_values with the first non-default case instead of
17942 default case if default_case_nonstandard.
17944 2017-05-02 Nathan Sidwell <nathan@acm.org>
17946 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
17947 check. Fix formatting.
17949 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
17951 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
17952 errors when comparing specialized and unspecialized times.
17954 2017-05-02 David Malcolm <dmalcolm@redhat.com>
17956 * diagnostic-show-locus.c
17957 (layout::should_print_annotation_line_p): Make private.
17958 (layout::print_annotation_line): Make private.
17959 (layout::annotation_line_showed_range_p): Make private.
17960 (layout::show_ruler): Make private.
17961 (layout::print_source_line): Make private. Pass in line and
17962 line_width, rather than calling location_get_source_line. Drop
17964 (layout::print_leading_fixits): New method.
17965 (layout::print_any_fixits): Rename to...
17966 (layout::print_trailing_fixits): ...this, and make private.
17967 Don't print newline fixits.
17968 (diagnostic_show_locus): Move logic for printing one row into...
17969 (layout::print_line): ...this new function. Move the
17970 location_get_source_line call and error-handling from
17971 print_source_line to here. Call print_leading_fixits, and rename
17972 print_any_fixits to print_trailing_fixits.
17973 (selftest::test_fixit_insert_containing_newline): Update now that
17974 newlines are partially supported.
17975 (selftest::test_fixit_insert_containing_newline_2): New test.
17976 (selftest::test_fixit_replace_containing_newline): Update comments.
17977 (selftest::diagnostic_show_locus_c_tests): Call the new test.
17978 * edit-context.c (class added_line): New class.
17979 (class edited_line): Describe newline handling in comment.
17980 (edited_line::actually_edited_p): New method.
17981 (edited_line::print_content): Delete redundant decl.
17982 (edited_line::m_predecessors): New field.
17983 (edited_file::print_content): Call edited_line::print_content.
17984 (edited_file::print_diff): Update to support newlines.
17985 (edited_file::print_diff_hunk): Likewise.
17986 (edited_file::print_run_of_changed_lines): New function.
17987 (edited_file::print_diff_line): Convert to...
17988 (print_diff_line): ...this.
17989 (edited_file::get_effective_line_count): New function.
17990 (edited_line::edited_line): Initialize new field m_predecessors.
17991 (edited_line::~edited_line): Clean up m_predecessors.
17992 (edited_line::apply_fixit): Handle newlines.
17993 (edited_line::get_effective_line_count): New function.
17994 (edited_line::print_content): New function.
17995 (edited_line::print_diff_lines): New function.
17996 (selftest::test_applying_fixits_insert_containing_newline): New
17998 (selftest::test_applying_fixits_replace_containing_newline): New
18000 (selftest::insert_line): New function.
18001 (selftest::test_applying_fixits_multiple_lines): Add example of
18003 (selftest::edit_context_c_tests): Call the new tests.
18005 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18007 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
18008 parameter cand. Update dump information.
18009 (get_computation_cost): Update uses.
18011 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18013 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
18014 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
18015 (get_computation_at, rewrite_use_address): Update use of
18016 get_computation_aff.
18018 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18020 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
18021 (get_computation): Delete.
18022 (get_computation_cost): Implement like get_computation_cost_at.
18023 Use get_computation_at.
18024 (get_computation_cost_at): Delete.
18025 (rewrite_use_nonlinear_expr): Use get_computation_at.
18026 (rewrite_use_compare, remove_unused_ivs): Ditto.
18028 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18030 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
18032 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18034 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
18035 (ivopts_global_cost_for_size): Rename parameter and update uses.
18036 (iv_ca_recount_cost): Update uses.
18037 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
18038 candidates seperately in n_invs and n_cands.
18039 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
18041 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18043 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
18044 (find_inv_vars_cb): New.
18045 (find_depends): Renamed to ...
18046 (find_inv_vars): ... this.
18047 (add_candidate_1, force_var_cost): Call find_inv_vars.
18048 (split_address_cost, determine_group_iv_cost_cond): Ditto.
18050 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18052 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
18053 inv_vars. Add inv_exprs.
18054 (struct iv_cand): Rename depends_on to inv_vars.
18055 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
18056 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
18057 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
18058 (dump_cand): Dump inv_vars.
18059 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
18060 (record_invariant, find_depends, add_candidate_1): Ditto.
18061 (set_group_iv_cost, force_var_cost): Ditto.
18062 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
18063 (get_computation_cost_at, get_computation_cost): Ditto.
18064 (determine_group_iv_cost_generic): Ditto.
18065 (determine_group_iv_cost_address): Ditto.
18066 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
18067 (determine_group_iv_costs): Ditto.
18068 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
18069 (iv_ca_set_remove_invariants): Renamed to ...
18070 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
18071 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
18072 (iv_ca_set_add_invariants): Renamed to ...
18073 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
18074 (iv_ca_set_cp): Use iv_ca_set_add_invs.
18075 (iv_ca_has_deps): Support inv_vars and inv_exprs.
18076 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
18077 (create_new_ivs): Remove useless dump.
18079 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18081 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
18083 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
18084 (iv_ca_set_no_cp, create_new_iv): Ditto.
18086 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18088 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18090 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18092 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
18093 function tree_check2.
18095 2017-05-02 Martin Liska <mliska@suse.cz>
18097 * doc/gcov.texi: Add missing preposition.
18098 * gcov.c (function_info::function_info): Properly fill up
18099 all member variables.
18101 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18103 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
18105 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18107 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
18109 2017-05-02 Martin Liska <mliska@suse.cz>
18112 * lto-streamer-in.c (lto_read_tree_1): Remove
18113 LTO_STREAMER_DEBUG.
18114 * lto-streamer.c (struct tree_hash_entry): Likewise.
18115 (struct tree_entry_hasher): Likewise.
18116 (tree_entry_hasher::hash): Likewise.
18117 (tree_entry_hasher::equal): Likewise.
18118 (lto_streamer_init): Likewise.
18119 (lto_orig_address_map): Likewise.
18120 (lto_orig_address_get): Likewise.
18121 (lto_orig_address_remove): Likewise.
18122 * lto-streamer.h: Likewise.
18123 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
18124 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
18126 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
18128 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
18129 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
18130 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
18131 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
18132 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
18133 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
18134 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
18135 (mm_maskz_sub_ss): New intrinsics.
18136 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18137 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18138 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
18139 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
18140 (__builtin_ia32_subss_mask_round): New builtins.
18141 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18142 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18143 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
18145 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
18146 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18148 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18151 2017-05-02 Martin Jambor <mjambor@suse.cz>
18153 PR tree-optimization/78687
18154 * tree-sra.c (access): New field parent.
18155 (process_subtree_disqualification): New function.
18156 (disqualify_candidate): Call it.
18157 (build_accesses_from_assign): Reset write flag if creating an
18159 (build_access_subtree): Fill in parent field and also prpagate
18160 down grp_write flag.
18161 (create_artificial_child_access): New parameter set_grp_write, set
18162 grp_write to its value.
18163 (propagate_subaccesses_across_link): Also propagate grp_write flag
18165 (propagate_all_subaccesses): Push the closest parent back to work
18166 queue if add_access_to_work_queue returned true.
18168 2017-05-02 Richard Biener <rguenther@suse.de>
18170 * common.opt (fstrict-overflow): Alias negative to fwrapv.
18171 * doc/invoke.texi (fstrict-overflow): Remove all traces of
18172 -fstrict-overflow documentation.
18173 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
18174 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
18175 flag_strict_overflow.
18176 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
18177 * lto-opts.c (lto_write_options): Do not stream it.
18178 * lto-wrapper.c (merge_and_complain): Do not handle it.
18179 * opts.c (default_options_table): Do not set -fstrict-overflow.
18180 (finish_options): Likewise do not clear it when sanitizing.
18181 * simplify-rtx.c (simplify_const_relational_operation): Do not
18182 test flag_strict_overflow.
18184 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18186 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
18187 using enabled attribute.
18188 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
18189 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
18190 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
18191 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
18192 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
18193 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
18194 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
18195 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
18196 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
18197 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
18199 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18201 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
18203 2017-05-02 Richard Biener <rguenther@suse.de>
18205 PR tree-optimization/80591
18207 2017-04-10 Richard Biener <rguenther@suse.de>
18209 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18212 2017-05-02 Richard Biener <rguenther@suse.de>
18214 PR tree-optimization/80549
18215 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
18216 (cleanup_tree_cfg_noloop): Create forwarders to known loop
18217 headers if they do not have a preheader.
18219 2017-05-02 Martin Liska <mliska@suse.cz>
18222 * common.opt: Fix typo.
18223 * doc/invoke.texi: Likewise.
18225 2017-05-01 Jan Beulich <jbeulich@suse.com>
18227 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
18228 swapping, add (x,x,m,x,n) alternative.
18230 2017-05-01 Nathan Sidwell <nathan@acm.org>
18232 * calls.c (combine_pending_stack_adjustment_and_call): Remove
18233 unnecessary unadjusted_alignment check.
18235 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18238 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
18240 * gimplify.c (gimplify_cilk_detach): New function.
18241 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
18242 * tree-core.h: Document EXPR_CILK_SPAWN.
18243 * tree.h (EXPR_CILK_SPAWN): Define.
18245 2017-05-01 David Malcolm <dmalcolm@redhat.com>
18247 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
18248 to use new fixit_hint representation, using the "replace" logic.
18249 (get_line_span_for_fixit_hint): Likewise.
18250 (layout::print_any_fixits): Likewise.
18251 (selftest::test_one_liner_many_fixits): Rename to...
18252 (selftest::test_one_liner_many_fixits_1): ...this, and update
18253 comment and expected output to reflect that the multiple fix-it
18254 hints are now consolidated into one insertion.
18255 (selftest::test_one_liner_many_fixits_2): New test.
18256 (selftest::test_diagnostic_show_locus_one_liner): Update for
18258 (selftest::test_fixit_consolidation): Update for fix-it API
18260 * diagnostic.c (print_parseable_fixits): Likewise.
18261 * edit-context.c (edited_line::m_line_events): Convert from
18262 auto_vec <line_event *> to auto_vec <line_event>.
18263 (class line_event): Convert from abstract base class to a concrete
18264 class, taking over the role of replace_event.
18265 (class insert_event): Delete.
18266 (class replace_event): Rename to class line_event. Convert to
18268 (edit_context::add_fixits): Reimplement.
18269 (edit_context::apply_insert): Delete.
18270 (edit_context::apply_replace): Rename to...
18271 (edit_context::apply_fixit): ...this. Convert to half-open range.
18272 (edited_file::apply_insert): Delete.
18273 (edited_file::apply_replace): Rename to...
18274 (edited_file::apply_fixit): ...this.
18275 (edited_line::~edited_line): Drop deletion of events.
18276 (edited_line::apply_insert): Delete.
18277 (edited_line::apply_replace): Rename to...
18278 (edited_line::apply_fixit): ...this. Convert to half-open range.
18279 Update for change to type of m_line_events.
18280 * edit-context.h (edit_context::apply_insert): Delete.
18281 (edit_context::apply_replace): Rename to...
18282 (edit_context::apply_fixit): ...this.
18284 2017-05-01 Martin Sebor <msebor@redhat.com>
18286 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
18289 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
18292 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
18293 __get_cpuid_max returns 0.
18294 (__get_cpuid_count): Ditto.
18296 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
18298 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
18299 replacement expression is another instance of one of its arguments.
18301 2017-05-01 Jakub Jelinek <jakub@redhat.com>
18304 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
18305 check for stack push/pop autoinc.
18306 * config/i386/i386.c (ix86_agi_dependent): Return false
18307 if the only reason why modified_in_p returned true is that
18308 addr is SP based and set_insn is a push or pop.
18310 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18312 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
18315 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18318 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
18319 (account_size_time): Use two predicates - exec_pred and
18321 (evaluate_conditions_for_known_args): Compute both clause and
18323 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
18324 (inline_summary_t::duplicate): Update.
18325 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
18327 (compute_inline_parameters): Likewise.
18328 (estimate_edge_size_and_time): Update caluclation of time.
18329 (estimate_node_size_and_time): Compute both time and nonspecialized
18331 (estimate_ipcp_clone_size_and_time): Update.
18332 (inline_merge_summary): Update.
18333 (do_estimate_edge_time): Update.
18334 (do_estimate_edge_size): Update.
18335 (do_estimate_edge_hints): Update.
18336 (inline_read_section, inline_write_summary): Stream both new predicates.
18337 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
18339 (compute_inlined_call_time): Cleanup.
18340 (big_speedup_p): Update.
18341 (edge_badness): Update.
18342 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
18343 (size_time_entry): Replace predicate by exec_predicate and
18344 nonconst_predicate.
18345 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
18346 (estimate_edge_time): Return also nonspec_time.
18347 (reset_edge_growth_cache): Update.
18349 2017-04-29 Jakub Jelinek <jakub@redhat.com>
18351 PR rtl-optimization/80491
18352 * ifcvt.c (noce_process_if_block): When looking for x setter
18353 with missing else_bb, don't check only the insn right before
18354 cond_earliest, but look for the last insn that x is modified in
18355 within the same bb.
18357 PR rtl-optimization/80491
18358 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
18360 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
18362 PR tree-optimization/80487
18363 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
18365 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18367 PR tree-optimization/79697
18368 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
18369 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
18370 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18372 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
18373 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
18375 2017-04-28 Martin Sebor <msebor@redhat.com>
18377 PR tree-optimization/80523
18378 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
18379 (init_target_to_host_charmap, target_to_host, target_strtol10): New
18381 (maybe_warn, format_directive, parse_directive): Use new functions.
18382 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
18384 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
18386 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
18388 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18390 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
18391 target_header_dir): Set correctly.
18392 * configure: Regenerated.
18393 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18394 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18395 instead of SYSTEM_HEADER_DIR.
18397 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
18399 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
18400 (estimate_local_effects): Likewise.
18401 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
18402 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
18403 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
18404 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
18405 do_estimate_edge_time, estimate_edge_time): Likewise.
18406 * ipa-inline-analysis.c (estimate_node_size_and_time,
18407 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
18408 (estimate_time_after_inlining): Remove.
18410 2017-04-28 Martin Liska <mliska@suse.cz>
18412 * doc/gcov.texi: Enhance documentation of gcov.
18414 2017-04-28 Martin Liska <mliska@suse.cz>
18416 * doc/gcov.texi: Sort options in alphabetic order.
18417 * doc/gcov-dump.texi: Likewise.
18418 * doc/gcov-tool.texi: Likewise.
18419 * gcov.c (print_usage): Likewise.
18420 * gcov-dump.c (print_usage): Likewise.
18421 * gcov-tool.c (print_merge_usage_message): Likewise.
18422 (print_rewrite_usage_message): Likewise.
18423 (print_overlap_usage_message): Likewise.
18425 2017-04-28 Martin Liska <mliska@suse.cz>
18427 PR gcov-profile/53915
18428 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18430 2017-04-28 Martin Liska <mliska@suse.cz>
18432 PR gcov-profile/79891
18433 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18434 is marked by compiler as living on a line.
18435 (get_cycles_count): Remove usage of the union.
18436 (output_intermediate_file): Likewise.
18437 (find_source): Fix GNU coding style.
18438 (accumulate_line_counts): Remove old non-all block mode.
18439 (output_lines): Remove usage of the union.
18440 * profile.c (output_location): Include all BBs, even if
18441 belonging to a same line (and file) as a previous BB.
18443 2017-04-28 Martin Liska <mliska@suse.cz>
18445 * gcov.c (process_args): Handle new argument 'w'.
18446 (read_graph_file): Assign ID to BBs.
18447 (output_branch_count): Display BB # if verbose flag is set.
18448 (output_lines): Likewise for arcs.
18449 (print_usage): Add '--verbose' option help.
18450 * doc/gcov.texi: Document --verbose (-w) option.
18452 2017-04-28 Martin Liska <mliska@suse.cz>
18454 * gcov.c (struct block_location_info): New struct.
18455 (process_file): Fill up the new structure.
18456 (read_graph_file): Replace usage of encoding by the newly added
18458 (add_line_counts): Likewise.
18459 (accumulate_line_counts): Remove usage of the union.
18460 (function_info::function_info): New function.
18461 (function_info::~function_info): Likewise.
18462 (process_file): Call delete instead of release_function.
18463 (release_function): Release the function.
18464 (release_structures): Call delete instead of release_function.
18465 (solve_flow_graph): Replace usage of num_blocks.
18466 (find_exception_blocks): Likewise.
18467 (output_lines): Fix GNU coding style.
18469 2017-04-28 Martin Liska <mliska@suse.cz>
18472 * coverage.c (coverage_remove_note_file): New function.
18473 * coverage.h: Declare the function.
18474 * toplev.c (finalize): Clean if an error has been seen.
18476 2017-04-28 Martin Liska <mliska@suse.cz>
18478 PR gcov-profile/80031
18479 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18480 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18481 * gcov.c (read_graph_file): Read just number of blocks.
18482 * profile.c (branch_prob): Do not stream 0 flags per a basic
18485 2017-04-28 Martin Liska <mliska@suse.cz>
18487 * gcov-dump.c (tag_*): Add new argument to declarations.
18488 (dump_gcov_file): Likewise.
18489 (tag_blocks): Add and use new argument depth.
18490 (tag_arcs): Likewise.
18491 (tag_lines): Likewise.
18492 (tag_counters): Likewise.
18493 (tag_summary): Likewise.
18494 (dump_working_sets): Use depth to do a proper indentation.
18496 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18499 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18500 * symtab.c (symtab_node::debug_symtab): Move definition here.
18502 2017-04-28 Richard Biener <rguenther@suse.de>
18504 * lto-streamer.h (LTO_major_version): Bump to 7.
18506 2017-04-28 Richard Biener <rguenther@suse.de>
18508 * tree-vrp.c (assert_info): New struct.
18509 (add_assert_info): New helper.
18510 (register_edge_assert_for_2): Refactor to add asserts to a vector
18512 (register_edge_assert_for_1): Likewise.
18513 (register_edge_assert_for): Likewise.
18514 (finish_register_edge_assert_for): New helper actually registering
18515 asserts where live on edge.
18516 (find_conditional_asserts): Adjust.
18517 (find_switch_asserts): Likewise.
18518 (evrp_dom_walker::try_find_new_range): Generalize.
18519 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18521 2017-04-27 Marek Polacek <polacek@redhat.com>
18524 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18525 arg10 and arg11 to itype.
18527 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18529 * doc/extend.texi (Object Size Checking): Improve grammar.
18531 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18534 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18535 that the logic for permitting reciprocal estimates matches that
18538 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18541 * tree.c (type_cache_hasher::equal): Only compare
18542 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18543 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18544 non-aggregate element types.
18545 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18546 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18548 2017-04-27 Richard Biener <rguenther@suse.de>
18550 PR middle-end/80533
18551 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18552 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18553 keeping a reference to a trailing array.
18555 2017-04-27 Richard Biener <rguenther@suse.de>
18557 PR middle-end/80539
18558 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18559 being in loop-closed SSA form conservatively.
18560 (chrec_fold_multiply_poly_poly): Likewise.
18562 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18564 PR middle-end/79665
18565 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18566 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18568 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18571 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18572 (aarch64_function_arg_alignment): Return unsigned int again, but still
18573 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18574 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18575 Don't emit -Wpsabi note.
18576 (aarch64_function_arg_boundary): Likewise.
18577 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18580 2017-04-26 Nathan Sidwell <nathan@acm.org>
18582 * tree.h (crc32_unsigned_n): Declare.
18583 (crc32_unsigned, crc32_unsigned): Make inline.
18584 * tree.c (crc32_unsigned_bits): Replace with ...
18585 (crc32_unsigned_n): ... this.
18586 (crc32_unsigned, crc32_byte): Remove.
18587 (crc32_string): Remove unnecessary braces.
18589 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18591 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18592 * ipa-inline-analysis.c (MAX_TIME): Remove.
18593 (account_size_time): Use sreal for time.
18594 (dump_inline_summary): Update.
18595 (estimate_function_body_sizes): Update.
18596 (estimate_edge_size_and_time): Update.
18597 (estimate_calls_size_and_time): Update.
18598 (estimate_node_size_and_time): Update.
18599 (inline_merge_summary): Update.
18600 (inline_update_overall_summary): Update.
18601 (estimate_time_after_inlining): Update.
18602 (inline_read_section): Update.
18603 (inline_write_summary): Update.
18604 * ipa-inline.c (compute_uninlined_call_time): Update.
18605 (compute_inlined_call_time): Update.
18606 (recursive_inlining): Update.
18607 (inline_small_functions): Update.
18608 (dump_overall_stats): Update.
18609 * ipa-inline.h: Include sreal.h.
18610 (size_time_entry): Turn time to sreal.
18611 (inline_summary): Turn self_time nad time to sreal.
18613 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18615 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18617 (sreal::stream_out, sreal::stream_in): New.
18618 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18620 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18622 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18625 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18628 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18630 (dimode_scalar_chain::compute_convert_gain): Ditto.
18631 (dimode_scalar_chain::make_vector_copies): Ditto.
18632 (dimode_scalar_chain::convert_reg): Ditto.
18633 (dimode_scalar_chain::convert_insn): Ditto.
18634 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18635 (VI248_AVX512BW_1): New mode iterator.
18636 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18637 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18640 2017-04-25 Martin Sebor <msebor@redhat.com>
18642 PR tree-optimization/80497
18643 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18644 constants are representable in HOST_WIDE_INT.
18645 (parse_directive): Ditto.
18647 2017-04-25 Martin Sebor <msebor@redhat.com>
18650 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18651 (new_zero_array): Adjust signature.
18652 (dom_info::dom_init): Used unsigned rather that size_t.
18653 (dom_info::dom_info): Same.
18655 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18656 Jakub Jelinek <jakub@redhat.com>
18659 * config/arm/arm.c: Include gimple.h.
18660 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18661 returns negative, increment ncrn only if it returned positive.
18662 (arm_needs_doubleword_align): Return int instead of bool,
18663 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18664 members, but if there is any such non-FIELD_DECL
18665 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18666 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18667 returns negative, increment nregs only if it returned positive.
18668 (arm_setup_incoming_varargs): Likewise.
18669 (arm_function_arg_boundary): Emit -Wpsabi note if
18670 arm_needs_doubleword_align returns negative, return
18671 DOUBLEWORD_ALIGNMENT only if it returned positive.
18673 2017-04-25 Marek Polacek <polacek@redhat.com>
18676 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18677 first argument to type.
18679 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
18682 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
18683 type checks to test for compatibility instead of equality.
18685 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18686 Jakub Jelinek <jakub@redhat.com>
18689 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18691 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18692 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18693 the alignment computation, but return their maximum in warn_alignment.
18694 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18695 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18697 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18698 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18701 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18703 * config/arc/simdext.md (dmpyh): Fix typo.
18705 2017-04-25 Richard Biener <rguenther@suse.de>
18707 PR tree-optimization/80492
18708 * alias.c (compare_base_decls): Handle registers with asm
18709 specification conservatively.
18710 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18711 compare_base_decls returning dont-know properly.
18713 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18715 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18716 (legitimate_offset_address_p): New function.
18717 (arc_legitimate_address_p): Use above function.
18719 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18721 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18723 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18725 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18726 ACCH registers whenever they are available.
18728 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18730 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18731 double regs fix when not used.
18733 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18735 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18737 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18738 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18740 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18742 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18743 when emitting short ADD instructions.
18745 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18747 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18749 (cmpsi_cc_c_insn): Likewise.
18750 (cbranchsi4_scratch): Compute proper instruction length using
18751 compact_hreg_operand.
18752 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18754 2017-04-25 Richard Biener <rguenther@suse.de>
18756 PR middle-end/80509
18757 * passes.c (pass_manager::pass_manager): Initialize
18758 m_name_to_pass_map.
18760 2017-04-25 Richard Biener <rguenther@suse.de>
18762 PR tree-optimization/79201
18763 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18765 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18768 * config/s390/vector.md: Split MEM->GPR vector moves for
18769 non-s_operand addresses.
18771 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18774 * config/s390/predicates.md (reload_const_wide_int_operand): New
18776 * config/s390/s390.md ("movti"): Remove d/P alternative.
18777 ("movti_bigconst"): New pattern definition.
18779 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18782 * s390-protos.h (s390_expand_cs_hqi): Removed.
18783 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
18784 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
18785 modes as well as CCZ1mode and CCZmode.
18786 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
18787 signature of s390_emit_compare_and_swap.
18788 (s390_expand_cs_hqi): Likewise, make static.
18789 (s390_expand_cs_tdsi): Generate an explicit compare before trying
18790 compare-and-swap, in some cases.
18791 (s390_expand_cs): Wrapper function.
18792 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
18794 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
18795 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
18796 patterns for small and large integers. Forbid symref memory operands.
18797 Move expander to s390.c. Require cc register.
18798 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
18799 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
18800 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
18801 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
18802 symref memory operands. Remove CC mode and call s390_match_ccmode
18804 ("atomic_exchange<mode>"): Allow and implement all integer modes.
18806 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18808 * config/s390/s390.md (define_peephole2): New peephole to help
18809 combining the load-and-test pattern with volatile memory.
18811 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18813 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
18814 with CCZmode for TARGET_Z196.
18816 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18818 PR rtl-optimization/80501
18819 * combine.c (make_compound_operation_int): Set subreg_code to SET
18820 even for AND with mask of the sign bit of mode.
18822 PR rtl-optimization/80500
18823 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
18824 sum's initial value.
18826 2017-04-25 Julian Brown <julian@codesourcery.com>
18827 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18829 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
18831 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
18833 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
18835 2017-04-25 Julian Brown <julian@codesourcery.com>
18836 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18838 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
18839 (thunderx2t99_sha): New Reservation.
18841 2017-04-25 Julian Brown <julian@codesourcery.com>
18842 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18844 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
18845 type for 1-element load.
18847 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
18849 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
18851 2017-04-24 Martin Jambor <mjambor@suse.cz>
18853 PR tree-optimization/80293
18854 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
18855 char arrays not totally scalarizable if it is false.
18856 (analyze_all_variable_accesses): Pass correct value in the new
18857 parameter. Add a statistics counter.
18859 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
18861 PR middle-end/79931
18862 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
18864 2017-04-24 Richard Biener <rguenther@suse.de>
18866 PR tree-optimization/80494
18867 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
18868 out for complex types.
18870 2017-04-24 Richard Biener <rguenther@suse.de>
18872 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
18873 * tree-ssa-sccvn.c (print_scc): Print SCC size.
18874 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
18875 (DFS): Adjust and never fail.
18876 (sccvn_dom_walker::fail): Remove.
18877 (sccvn_dom_walker::before_dom_children): Adjust.
18878 (run_scc_vn): Likewise and never fail.
18879 * tree-ssa-pre.c (pass_pre::execute): Adjust.
18880 (pass_fre::execute): Likewise.
18882 2017-04-24 Richard Biener <rguenther@suse.de>
18884 PR tree-optimization/79725
18885 * tree-ssa-sink.c (statement_sink_location): Return whether
18886 failure reason was zero uses. Move that check later.
18887 (sink_code_in_bb): Deal with zero uses by removing the stmt
18890 2017-04-24 Richard Biener <rguenther@suse.de>
18893 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
18894 pointer-based references.
18896 2017-04-24 Richard Biener <rguenther@suse.de>
18899 * pass_manager.h (pass_manager::operator new): Remove.
18900 (pass_manager::operator delete): Likewise.
18901 * passes.c (pass_manager::operator new): Remove.
18902 (pass_manager::operator delete): Likewise.
18903 (pass_manager::pass_manager): Zero individual pass members.
18905 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
18908 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
18909 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
18910 Check "XEXP (src, 1)" operand here.
18911 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
18912 Check "XEXP (src, 1)" operand here.
18913 (dimode_scalar_chain::make_vector_copies): Detect count register
18914 of a shift instruction. Zero extend count register from QImode
18915 to DImode to satisfy vector shift pattern count operand predicate.
18916 Substitute vector shift count operand with a DImode copy.
18917 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
18920 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
18922 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
18923 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18924 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
18925 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18926 (UNSPEC_NOREX_MEM): Remove definition.
18928 2017-04-21 Richard Biener <rguenther@suse.de>
18930 PR tree-optimization/79547
18931 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18932 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
18933 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
18934 without any constraints.
18936 2017-04-21 Richard Biener <rguenther@suse.de>
18938 PR tree-optimization/78847
18939 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
18941 2017-04-21 Richard Biener <rguenther@suse.de>
18943 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
18944 (build_distinct_type_copy): Likewise.
18945 (build_variant_type_copy): Likewise.
18946 * tree.c (build_qualified_type): Pass down mem-stat info.
18947 (build_distinct_type_copy): Likewise.
18948 (build_variant_type_copy): Likewise.
18950 2017-04-21 Richard Biener <rguenther@suse.de>
18952 PR tree-optimization/80237
18953 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
18955 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
18956 for a simplified result.
18958 2016-04-21 Richard Biener <rguenther@suse.de>
18960 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
18961 sth as strict as a simple_iv but a chrec without symbols and an
18962 operand defined in the loop we are peeling (and not some subloop).
18963 (propagate_constants_for_unrolling): Propagate all constants.
18965 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18968 * config/i386/i386.c (print_reg): Remove assert for disalowed
18969 regno values, call output_operand_lossage instead.
18971 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18974 * config/i386/constraints.md (Yc): New register constraint.
18975 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
18976 Use Yc constraint for alternative 2 of operand 0. Remove
18977 preferred_for_speed attribute.
18979 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18981 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
18982 lastprivate clauses in SIMT case.
18984 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
18986 * doc/invoke.texi (-Wextra-semi): Document new warning option.
18988 2017-04-20 Richard Biener <rguenther@suse.de>
18990 PR tree-optimization/57796
18991 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
18992 as N scalar stores.
18993 (vect_model_load_cost): Cost gathers as N scalar loads.
18995 2017-04-20 Richard Biener <rguenther@suse.de>
18997 * ggc-page.c (ggc_allocated_p): Rename to ...
18998 (safe_lookup_page_table_entry): ... this and return the lookup
19000 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
19002 2017-04-20 Richard Biener <rguenther@suse.de>
19004 PR tree-optimization/80453
19005 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
19006 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
19007 from the conditions.
19008 (vn_phi_eq): Pass them down.
19009 (vn_phi_lookup): Record them.
19010 (vn_phi_insert): Likewise.
19012 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
19014 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
19015 uninitialized variable warning to avoid buffer overrun.
19017 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
19020 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
19021 is suppressed for '{ 0 }' in C.
19023 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19025 * BASE-VER: Set to 8.0.0.
19027 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
19029 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
19030 priority .init_array and .fini_array section with SECTION_NOTYPE
19033 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19035 PR middle-end/80423
19036 * tree.h (build_array_type): Add typeless_storage default argument.
19037 * tree.c (type_cache_hasher::equal): Also compare
19038 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
19039 (build_array_type): Add typeless_storage argument, set
19040 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
19042 (build_nonshared_array_type): Adjust build_array_type_1 caller.
19043 (build_array_type): Likewise. Add typeless_storage argument.
19045 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19046 Jakub Jelinek <jakub@redhat.com>
19048 PR tree-optimization/80426
19049 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
19050 operation on symbolic operands, also compute the overflow for the
19051 invariant part when the operation degenerates into a negation.
19053 2017-04-19 Jakub Jelinek <jakub@redhat.com>
19056 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
19057 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
19060 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
19062 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
19065 * config/avr/avr.c (tree.h): Include it.
19066 (cgraph.h): Include it.
19067 (avr_encode_section_info): Don't warn for uninitialized progmem
19068 variable if it's just an alias.
19070 2017-04-19 Richard Biener <rguenther@suse.de>
19073 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
19074 when needed by AutoPGO.
19076 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
19079 * doc/lto.texi: Remove an extra 'that'.
19081 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
19083 PR rtl-optimization/80429
19084 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
19085 are only used in debug insns.
19087 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19088 Vladimir Makarov <vmakarov@redhat.com>
19090 * config/sparc/predicates.md (input_operand): Add comment. Return
19091 true for any memory operand when LRA is in progress.
19092 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
19094 2017-04-18 Jeff Law <law@redhat.com>
19097 * mips.md ({return,simple_return}_internal): Do not overwrite
19100 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19102 PR tree-optimization/80443
19103 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
19104 instead of adding 1, subtract -1 and similarly instead of subtracting
19107 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
19109 PR rtl-optimization/80357
19110 * haifa-sched.c (tmp_bitmap): New variable.
19111 (model_recompute): Handle duplicate use records.
19112 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
19113 (free_global_sched_pressure_data): Free it.
19115 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
19118 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19119 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19120 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19121 instead of SYSTEM_HEADER_DIR.
19123 2017-04-18 Jeff Law <law@redhat.com>
19125 PR middle-end/80422
19126 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
19127 predecessors after walking up the insn chain.
19129 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19132 * dwarf2out.c (modified_type_die): Try harder not to emit internal
19133 sizetype type into debug info.
19135 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19138 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
19139 unneeded test for TARGET_UPPER_REGS_SF.
19140 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
19142 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19145 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
19146 instead of gsi_after_labels.
19148 2017-04-18 Jeff Law <law@redhat.com>
19150 * regcprop.c (maybe_mode_change): Avoid creating copies of the
19154 2017-04-13 Jeff Law <law@redhat.com>
19155 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19156 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19158 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
19161 * config/avr/avr.c (intl.h): Include it.
19162 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
19164 2017-04-18 Martin Liska <mliska@suse.cz>
19166 PR gcov-profile/78783
19167 * gcov-tool.c (gcov_output_files): Validate that destination
19168 file is either removed by the tool or by a user.
19170 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19171 Guy Benyei <guybe@mellanox.com>
19173 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
19174 block, and do not negate it, the stored id is already negative.
19176 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19178 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
19180 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19183 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
19184 masks of options that should be turned off if the VSX vector
19185 options are turned off.
19186 (OTHER_P8_VECTOR_MASKS): Likewise.
19187 (OTHER_VSX_VECTOR_MASKS): Likewise.
19188 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
19189 rs6000_disable_incompatible_switches to validate no type switches
19191 (rs6000_incompatible_switch): New function to disallow turning on
19192 other vector options if -mno-vsx, -mno-power8-vector, or
19193 -mno-power9-vector are specified.
19195 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19197 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
19199 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19201 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
19202 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
19203 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
19204 (ARG_POINTER_CFA_OFFSET): Likewise.
19206 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19208 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
19209 conditions to take advantage of various optimizations.
19211 2017-04-13 Jeff Law <law@redhat.com>
19213 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19214 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19215 (zero_extendsidi2_dext): Likewise.
19217 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19220 * fold-const.c (fold_ternary_loc): Revert
19221 use op0 instead of fold_convert_loc (loc, type, arg0) part of
19224 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
19226 PR rtl-optimization/80343
19227 * lra-remat.c (update_scratch_ops): Assign original hard reg to
19228 new scratch pseudo.
19230 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
19233 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
19234 to ubsan_encode_value.
19236 2017-04-13 Jeff Law <law@redhat.com>
19238 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
19239 appearing in DEBUG_INSNs.
19241 2017-04-13 Martin Liska <mliska@suse.cz>
19243 PR gcov-profile/80413
19244 * gcov-io.c (gcov_write_string): Copy to buffer just when
19245 allocated size is greater than zero.
19247 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19250 * dwarf2out.c (decls_for_scope): Ignore declarations of
19251 current_function_decl in BLOCK_NONLOCALIZED_VARS.
19253 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
19256 * ipa-visibility.c (non_local_p): Fix typos.
19257 (localize_node): When localizing symbol in same comdat group,
19258 dissolve the group only when we know external symbols are going
19260 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
19262 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19264 PR tree-optimization/79390
19265 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
19266 order does not result in usable sequence, retry with reversed operand
19272 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
19273 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
19274 op0 instead of fold_convert_loc (loc, type, arg0).
19276 2017-04-12 Jeff Law <law@redhat.com>
19278 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
19279 has a delay slot in the generated code.
19281 * config/cris/cris.md (cris_preferred_reload_class): Return
19282 GENNONACR_REGS rather than GENERAL_REGS.
19284 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19287 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
19288 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
19289 signedness of the result type.
19291 2017-04-12 Richard Biener <rguenther@suse.de>
19292 Jeff Law <law@redhat.com>
19294 PR tree-optimization/80359
19295 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
19296 trim stores to TARGET_MEM_REFs.
19298 2017-04-12 Richard Biener <rguenther@suse.de>
19300 PR tree-optimization/79390
19301 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
19302 threading case even more.
19304 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19307 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
19308 for quad_address_p for TImode, instead of just not indexed_address.
19310 2017-04-12 Richard Biener <rguenther@suse.de>
19311 Bernd Edlinger <bernd.edlinger@hotmail.de>
19313 PR middle-end/79671
19314 * alias.c (component_uses_parent_alias_set_from): Handle
19315 TYPE_TYPELESS_STORAGE.
19316 (get_alias_set): Likewise.
19317 * tree-core.h (tree_type_common): Add typeless_storage flag.
19318 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
19319 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
19320 for types containing members with TYPE_TYPELESS_STORAGE.
19321 (place_field): Likewise.
19322 (layout_type): Likewise for ARRAY_TYPE.
19323 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
19324 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
19325 TYPE_TYPELESS_STORAGE.
19326 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
19328 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19331 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19332 first argument to type.
19334 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19338 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
19339 CONST0_RTX (mode) rather than const0_rtx where appropriate.
19340 (rs6000_expand_binop_builtin): Likewise.
19341 (rs6000_expand_ternop_builtin): Likewise; also add missing
19342 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
19343 vshasigma built-ins.
19344 * doc/extend.texi: Document that vec_xxpermdi's third argument
19345 must be a constant.
19347 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
19349 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
19350 Use shift_const cost parameter when calculating gain of STV shifts.
19352 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
19354 PR rtl-optimization/70478
19355 * lra-constraints.c (process_alt_operands): Check memory for
19356 disfavoring memory insn operand.
19358 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19360 PR middle-end/80100
19361 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
19362 left shift in unsigned HOST_WIDE_INT type.
19364 PR rtl-optimization/80385
19365 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
19366 (not (neg X)) into (plus X -1) for complex or non-integral modes.
19369 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
19370 if they have any depend clauses.
19372 2017-04-11 Martin Liska <mliska@suse.cz>
19375 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
19376 * ipa-split.c (split_function): Create a local comdat symbol
19377 if caller is in a comdat group.
19379 2017-04-11 Martin Liska <mliska@suse.cz>
19382 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
19385 2017-04-11 Martin Sebor <msebor@redhat.com>
19387 PR middle-end/80364
19388 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
19389 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
19391 (directive::set_width, directive::set_precision, format_character):
19393 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
19396 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
19399 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
19400 conflict, set target->arch_name instead of target->cpu_name.
19402 2017-04-11 Richard Biener <rguenther@suse.de>
19404 PR tree-optimization/80374
19405 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
19406 build_zero_cst, remove fold_convertible_p check again.
19408 2017-04-11 Martin Liska <mliska@suse.cz>
19411 * ubsan.c (instrument_object_size): Do not instrument register
19414 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19417 * config/i386/i386-builtin-types.def
19418 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
19419 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
19420 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
19421 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
19422 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
19423 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
19424 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
19425 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
19426 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
19427 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19428 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19429 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19430 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19431 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19432 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19433 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19434 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19435 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19436 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19437 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19438 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19439 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19440 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19441 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19442 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19443 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19444 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19445 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19446 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19447 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19448 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19449 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19450 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19451 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19452 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19453 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19454 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19455 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19456 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19457 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19458 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19459 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19460 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19461 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19462 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19463 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19465 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19466 flag to second_arg_count, handle 4 argument function type _COUNT
19467 aliases, handle second_arg_count on second argument rather than last.
19469 2017-04-10 Jeff Law <law@redhat.com>
19471 PR tree-optimization/80374
19472 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19473 record anything if we can not convert integer_zero_node to the
19476 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19479 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19480 Enhance special handling given to the TARGET_P9_MINMAX option in
19481 relation to certain other options.
19483 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19485 PR tree-optimization/80153
19486 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19487 remove POINTER_PLUS_EXPR's base part directly, rather than through
19490 2017-04-10 Richard Biener <rguenther@suse.de>
19491 Bin Cheng <bin.cheng@arm.com>
19493 PR tree-optimization/80153
19494 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19495 the first element of pointer type aff_tree. Build result expr in
19497 (add_elt_to_tree): Convert to type unconditionally. Remove other
19498 fold_convert calls.
19499 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19500 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19502 2017-04-10 Richard Biener <rguenther@suse.de>
19504 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19507 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19509 PR rtl-optimization/70478
19510 * lra-constraints.c (curr_small_class_check): New.
19511 (update_and_check_small_class_inputs): New.
19512 (process_alt_operands): Update curr_small_class_check. Disfavor
19513 alternative insn memory operands. Check available regs for small
19516 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19519 * config/mips/mips.opt (-mvirt): Update description.
19520 * doc/invoke.texi (-mvirt): Likewise.
19522 2017-04-10 Richard Biener <rguenther@suse.de>
19524 PR middle-end/80362
19525 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19526 looking for NEGATE_EXPR in -A / -B to A / B folding.
19528 2017-04-10 Martin Liska <mliska@suse.cz>
19530 PR gcov-profile/80224
19531 * gcov.c (print_usage): Fix usage string.
19532 (get_gcov_intermediate_filename): Remove.
19533 (output_gcov_file): Use both for normal and intermediate format.
19534 (generate_results): Do not initialize special file for
19535 intermediate format.
19537 2017-04-10 Richard Biener <rguenther@suse.de>
19539 PR tree-optimization/80304
19540 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19543 2017-04-10 Nathan Sidwell <nathan@acm.org>
19546 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19547 (rs6000_init_builtins): Use it.
19549 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19551 * config/arm/arm.md (<mrc>): Add mode to SET source.
19552 (<mrrc>): Likewise.
19554 2017-04-10 Richard Biener <rguenther@suse.de>
19556 PR middle-end/80344
19557 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19559 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19562 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19563 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19564 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19565 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19566 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19567 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19568 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19569 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19570 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19571 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19572 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19573 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19574 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19575 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19576 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19577 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19578 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19579 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19580 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19581 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19582 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19583 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19584 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19586 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19588 PR rtl-optimization/70478
19589 * lra-constraints.c: Reverse the last patch.
19591 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19593 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19594 Add comment for WCHAR_T.
19596 2017-04-08 Martin Liska <mliska@suse.cz>
19599 2017-04-07 Martin Liska <mliska@suse.cz>
19602 * ipa-split.c (split_function): Add function part to a same comdat
19605 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19608 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19610 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19612 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19613 (vec_pairing): ...to this.
19614 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19615 (rs6000_sched_init): Adjust for name change.
19616 (struct rs6000_sched_context): Likewise.
19617 (rs6000_init_sched_context): Likewise.
19618 (rs6000_set_sched_context): Likewise.
19620 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19626 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19628 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19629 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19630 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19632 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19634 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19636 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19638 PR rtl-optimization/70703
19639 * ira-color.c (update_conflict_hard_regno_costs): Use
19640 int64_t instead of HOST_WIDE_INT.
19642 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19644 PR rtl-optimization/70478
19645 * lra-constraints.c (process_alt_operands): Disfavor alternative
19646 insn memory operands.
19648 2017-04-07 Jeff Law <law@redhat.com>
19650 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19651 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19653 2017-04-07 Martin Liska <mliska@suse.cz>
19656 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19657 Show error message instead of an ICE.
19659 2017-04-07 Martin Liska <mliska@suse.cz>
19662 * ipa-split.c (split_function): Add function part to a same comdat
19665 2017-04-07 Richard Biener <rguenther@suse.de>
19667 PR middle-end/80341
19668 * tree.c (get_unwidened): Also handle ! for_type case for
19670 * convert.c (do_narrow): Split out from ...
19671 (convert_to_integer_1): ... here. Do not pass final truncation
19672 type to get_unwidened for TRUNC_DIV_EXPR.
19674 2017-04-07 Richard Biener <rguenther@suse.de>
19676 * tree-affine.c (wide_int_ext_for_comb): Take type rather
19678 (aff_combination_const): Adjust.
19679 (aff_combination_scale): Likewise.
19680 (aff_combination_add_elt): Likewise.
19681 (aff_combination_add_cst): Likewise.
19682 (aff_combination_convert): Likewise.
19683 (add_elt_to_tree): Likewise. Remove unused argument.
19684 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19686 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19688 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19690 * config/arm/arm.c (arm_default_short_enums): Use
19691 ARM_DEFAULT_SHORT_ENUMS.
19692 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19694 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19697 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19698 members with redundant out-of-class redeclaration.
19700 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19703 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19704 * config/i386/i386.md (*zero_extendsidi2):
19705 Add (?*x,*x) and (?*v,*v) alternatives.
19707 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19710 * config/i386/i386.c (ix86_expand_builtin)
19711 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19712 mode from insn data. Convert operands to insn operand mode.
19713 Copy operands that don't satisfy insn predicate to a register.
19715 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19717 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19720 2017-04-06 Richard Biener <rguenther@suse.de>
19722 PR tree-optimization/80334
19723 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19724 preserve alignment of accesses.
19726 2017-04-06 Richard Biener <rguenther@suse.de>
19728 PR tree-optimization/80262
19729 * tree-sra.c (build_ref_for_offset): Preserve address-space
19731 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19732 Drop useless address-space information on MEM_REF offsets.
19734 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19736 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19738 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19740 PR rtl-optimization/70703
19741 * ira-color.c (update_conflict_hard_regno_costs): Use
19742 HOST_WIDE_INT instead of long.
19744 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19747 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19748 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19750 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19751 for x86_64 target. Handle -m3dnowa option.
19753 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19755 PR rtl-optimization/70703
19756 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19757 (update_conflict_hard_regno_costs): Use long instead of unsigned
19758 arithmetic for cost calculation.
19760 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19761 Bernd Edlinger <bernd.edlinger@hotmail.de>
19764 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19767 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19770 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
19771 ptr_mode with Pmode throughout.
19772 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
19773 into probe_stack_range and use DImode.
19775 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
19778 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
19779 call_eh_return is true.
19781 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19783 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
19784 Initialize last_match_fntype_index.
19786 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19789 * tree-nvr.c: Include internal-fn.h.
19790 (pass_return_slot::execute): Ignore internal calls without
19793 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19794 Richard Biener <rguenther@suse.de>
19797 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
19798 captures used multiple times, except for the last use.
19799 * generic-match-head.c: Include gimplify.h.
19801 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19803 PR tree-optimization/79390
19804 * target.h (struct noce_if_info): Declare.
19805 * targhooks.h (default_noce_conversion_profitable_p): Declare.
19806 * target.def (noce_conversion_profitable_p): New target hook.
19807 * ifcvt.h (struct noce_if_info): New type, moved from ...
19808 * ifcvt.c (struct noce_if_info): ... here.
19809 (noce_conversion_profitable_p): Renamed to ...
19810 (default_noce_conversion_profitable_p): ... this. No longer
19812 (noce_try_store_flag_constants, noce_try_addcc,
19813 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
19814 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
19815 instead of noce_conversion_profitable_p.
19816 * config/i386/i386.c: Include ifcvt.h.
19817 (ix86_option_override_internal): Don't override
19818 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
19819 (ix86_noce_conversion_profitable_p): New function.
19820 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
19821 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
19822 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
19823 * doc/tm.texi: Regenerated.
19825 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19827 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
19830 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19833 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
19834 instructions for small multiply cores.
19836 2017-04-04 Jeff Law <law@redhat.com>
19838 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
19840 (mips_expand_vec_perm_const): Initialize elements in orig_perm
19841 that are not set by the loop over the elements.
19843 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19846 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
19847 int mode, convert_modes it to mode as unsigned, otherwise use
19848 lowpart_subreg to mode rather than SImode.
19849 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
19850 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
19851 Use DImode instead of SImode for the shift count operand.
19852 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
19855 2017-04-04 Richard Biener <rguenther@suse.de>
19857 PR middle-end/80281
19858 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
19859 arithmetic done for the negate or the plus. Simplify.
19860 (A - (-B) -> A + B): Likewise.
19861 * fold-const.c (split_tree): Make sure to not negate pointers.
19863 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
19865 PR rtl-optimization/60818
19866 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
19867 a compare of comparisons with the thing compared if this results
19868 in a different machine mode.
19870 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
19872 * alias.c (base_alias_check): Fix typo in comment.
19873 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
19874 * cgraphunit.c (symbol_table::compile): Likewise.
19875 * collect2.c (maybe_run_lto_and_relink): Likewise.
19876 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
19877 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
19878 * config/avr/avr.c (avr_map_op_t): Likewise.
19879 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
19880 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
19881 * config/epiphany/epiphany.md (movcc): Likewise.
19882 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
19883 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
19885 * config/mips/mips.c (mips_save_restore_reg): Likewise.
19886 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
19887 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
19888 * config/sh/sh.c (sh_rtx_costs): Likewise.
19889 * fold-const.c (fold_truth_andor): Likewise.
19890 * genautomata.c (collapse_flag): Likewise.
19891 * gengtype.h (struct type::u::s): Likewise.
19892 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
19893 * input.c (FORMAT_AMOUNT): Likewise.
19894 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
19895 (known_aggs_to_agg_replacement_list): Likewise.
19896 * ipa-inline-analysis.c: Likewise.
19897 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
19898 * ipa-polymorphic-call.c
19899 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
19900 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
19901 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
19903 * modulo-sched.c (apply_reg_moves): Likewise.
19904 * omp-expand.c (build_omp_regions_1): Likewise.
19905 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
19906 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
19907 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
19908 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19909 * value-prof.c: Likewise.
19910 * var-tracking.c (val_reset): Likewise.
19912 2017-04-03 Richard Biener <rguenther@suse.de>
19914 PR tree-optimization/80275
19915 * fold-const.c (split_address_to_core_and_offset): Handle
19918 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
19920 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
19921 descriptors is at least equal to that of functions.
19923 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19925 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
19927 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19930 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
19931 (mov<IMOD4:mode>): New expander.
19932 (*mov<IMOD4:mode>_internal): New insn and split pattern.
19934 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19936 PR rtl-optimization/79405
19937 * fwprop.c (propagations_left): New variable.
19938 (forward_propagate_into): Decrement it.
19939 (fwprop_init): Initialize it.
19940 (fw_prop): If the variable has reached zero, stop propagating.
19941 (fwprop_addr): Ditto.
19943 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19946 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
19947 a FUNCTION_DECL, pass it as decl instead of origin to
19950 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
19952 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
19955 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
19958 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
19959 TARGET_VSX_SMALL_INTEGER.
19961 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19963 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19964 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
19966 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19968 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
19969 extraction from odd-numbered MSA register.
19971 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19973 PR middle-end/80173
19974 * expmed.c (store_bit_field_1): Don't attempt to create
19975 a word subreg out of hard registers wider than word if they
19976 have HARD_REGNO_NREGS of 1 for their mode.
19978 PR middle-end/80163
19979 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
19980 conversions to integer types wider than word and pointer.
19983 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
19984 (rtx_equal_for_cselib_p): Pass 0 to it.
19985 * cselib.c (cselib_hasher::equal): Likewise.
19986 (rtx_equal_for_cselib_1): Add depth argument. If depth
19987 is 128, don't look up VALUE locs and punt. Increment
19988 depth in recursive calls when walking VALUE locs.
19990 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
19992 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
19993 (make_gcov_file_name): Use the canonical path name for generating
19995 (read_line): Fix handling of files with ascii null bytes.
19997 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
19999 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
20000 to initialise a vector register instead
20001 of using a const_int.
20003 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20005 PR translation/80189
20006 * gimplify.c (omp_default_clause): Use %qs instead of %s in
20007 diagnostic messages.
20009 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
20012 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
20013 (dfp_diex_<mode>): Update mode of operand 1.
20014 * doc/extend.texi (dxex, dxexq): Document change to return type.
20015 (diex, diexq): Document change to argument type.
20017 2017-03-30 Martin Jambor <mjambor@suse.cz>
20020 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
20021 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
20022 it reflects the signature changes performed at the callee side.
20023 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
20024 to cgraph_build_function_type_skip_args.
20025 (build_function_decl_skip_args): Adjust call to the above function.
20027 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20030 * config/i386/sse.md
20031 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
20032 register as dest whenever it is a MEM not rtx_equal_p to the
20033 corresponding dup operand, and when forcing into reg move the
20034 reg into the memory afterwards.
20035 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
20036 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
20037 for the force_reg mode.
20038 (avx512vl_vextractf128<mode>): Use register as dest either
20039 always when a MEM, or when it is a MEM not rtx_equal_p to the
20040 corresponding dup operand, or even not when it is a CONST_VECTOR
20041 depending on the mode and lo vs. hi.
20042 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
20044 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
20045 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
20046 Likewise. Require that operands[2] is even.
20047 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
20048 Remove extraneous parens. Require that operands[2] is a multiple
20050 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
20051 operands[0] is a MEM if <mask_applied>, the predicates/constraints
20052 disallow memory then.
20054 2017-03-30 Richard Biener <rguenther@suse.de>
20056 PR tree-optimization/77498
20057 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
20058 to non-constants over backedges.
20060 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
20062 PR rtl-optimization/80233
20063 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
20064 as last_combined_insn. Do not test for BARRIER_P separately.
20066 2017-03-29 Andreas Schwab <schwab@suse.de>
20069 * calls.c (prepare_call_address): Convert funexp to Pmode before
20070 copying to temp reg.
20072 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20074 PR tree-optimization/80158
20075 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
20076 Handle possible future case of more than one alternate
20078 (replace_rhs_if_not_dup): Likewise.
20079 (replace_one_candidate): Likewise.
20081 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
20083 PR rtl-optimization/80193
20084 * ira.c (ira): Do not check allocation for LRA.
20086 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
20088 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
20089 (nvptx_output_simt_exit): Declare.
20090 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
20091 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
20092 (init_softstack_frame): Move initialization of crtl->is_leaf to...
20093 (nvptx_declare_function_name): ...here. Emit declaration of local
20094 memory space buffer for omp_simt_enter insn.
20095 (nvptx_output_unisimt_switch): New.
20096 (nvptx_output_softstack_switch): New.
20097 (nvptx_output_simt_enter): New.
20098 (nvptx_output_simt_exit): New.
20099 * config/nvptx/nvptx.h (struct machine_function): New fields
20100 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
20101 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
20102 (UNSPECV_SIMT_EXIT): Ditto.
20103 (omp_simt_enter_insn): New insn.
20104 (omp_simt_enter): New expansion.
20105 (omp_simt_exit): New insn.
20106 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
20108 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
20109 (expand_GOMP_SIMT_ENTER_ALLOC): New.
20110 (expand_GOMP_SIMT_EXIT): New.
20111 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
20112 (GOMP_SIMT_ENTER_ALLOC): Ditto.
20113 (GOMP_SIMT_EXIT): Ditto.
20114 * target-insns.def (omp_simt_enter): New insn.
20115 (omp_simt_exit): Ditto.
20116 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
20118 (lower_rec_simd_input_clauses): Implement SIMT privatization.
20119 (lower_rec_input_clauses): Likewise.
20120 (lower_lastprivate_clauses): Handle SIMT privatization.
20122 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
20123 (ompdevlow_adjust_simt_enter): New.
20124 (find_simtpriv_var_op): New.
20125 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
20126 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
20128 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
20129 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
20130 (copy_decl_for_dup_finish): Ditto.
20132 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
20134 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
20137 * config/i386/i386.c (ix86_option_override_internal): Always
20138 allow -mpreferred-stack-boundary=3 for 64-bit targets.
20140 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20142 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
20144 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20146 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
20147 mark new edge's irreducible flag accordign to it.
20148 (vect_do_peeling): Check loop preheader edge's irreducible flag
20149 and pass it to function slpeel_add_loop_guard.
20151 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
20153 PR tree-optimization/80218
20154 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
20155 Update block frequencies and counts.
20157 2017-03-28 Richard Biener <rguenther@suse.de>
20159 PR tree-optimization/78644
20160 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
20161 of a simplification result we may not use it at all.
20163 2017-03-28 Richard Biener <rguenther@suse.de>
20166 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
20167 without arguments, generate default definition of a SSA name.
20169 2017-03-28 Richard Biener <rguenther@suse.de>
20171 PR middle-end/80222
20172 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
20173 TYPE_REF_CAN_ALIAS_ALL references.
20174 * fold-const.c (fold_indirect_ref_1): Likewise.
20176 2017-03-28 Martin Liska <mliska@suse.cz>
20179 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
20180 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
20182 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20183 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
20185 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
20186 (EXTRA_SPECS): Define.
20187 (SUBTARGET_EXTRA_SPECS): Likewise.
20188 (SUBTARGET_CPP_SPEC): Likewise.
20189 * config/arc/elf.h (EXTRA_SPECS): Renamed to
20190 SUBTARGET_EXTRA_SPECS.
20191 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
20193 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20195 * config/arc/simdext.md (vst64_insn): Update pattern.
20196 (vld32wh_insn): Likewise.
20197 (vld32wl_insn): Likewise.
20198 (vld64_insn): Likewise.
20199 (vld32_insn): Likewise.
20201 2017-03-28 Marek Polacek <polacek@redhat.com>
20204 * fold-const.c (fold_comparison): Use protected_set_expr_location
20205 instead of SET_EXPR_LOCATION.
20207 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
20209 * tree.c (add_expr): Avoid name lookup warning.
20211 2017-03-27 Jeff Law <law@redhat.com>
20213 PR tree-optimization/80216
20214 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
20215 function name. Limit recursion depth.
20216 (record_temporary_equivalences): Corresponding changes.
20218 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
20220 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
20223 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20226 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
20228 * cfgcleanup.c (reg_note_cfa_p): New array.
20229 (insns_have_identical_cfa_notes): New function.
20230 (old_insns_match_p): Don't cross-jump in between /f
20231 and non-/f instructions. If both i1 and i2 are frame related,
20232 verify all CFA notes, their order and content.
20234 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20237 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
20238 HImode and SImode with zero extend to DImode to one insn.
20239 (bswap<mode>2_extenddi): Likewise.
20240 (bswapsi2_extenddi): Likewise.
20241 (bswaphi2_extendsi): Likewise.
20242 (bswaphi2): Combine bswap HImode and SImode into one insn.
20243 Separate memory insns from swapping register.
20244 (bswapsi2): Likewise.
20245 (bswap<mode>2): Likewise.
20246 (bswaphi2_internal): Delete, no longer used.
20247 (bswapsi2_internal): Likewise.
20248 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
20249 store, and gpr<-gpr swap insns.
20250 (bswap<mode>2_store): Likewise.
20251 (bswaphi2_reg): Register only splitter, combine with the splitter.
20252 (bswaphi2 splitter): Likewise.
20253 (bswapsi2_reg): Likewise.
20254 (bswapsi2 splitter): Likewise.
20255 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
20256 the insns into load, store, and register/register insns.
20257 (bswapdi2_ldbrx): Likewise.
20258 (bswapdi2_load): Likewise.
20259 (bswapdi2_store): Likewise.
20260 (bswapdi2_reg): Likewise.
20262 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
20264 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
20265 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
20267 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
20270 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
20272 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20273 special handling for target option conflicts between dform
20274 options (-mpower9-dform, -mpower9-dform-vector,
20275 -mpower9-dform-scalar) and -mno-direct-move.
20277 2017-03-27 Richard Biener <rguenther@suse.de>
20279 PR tree-optimization/80181
20280 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
20282 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20284 * config/arc/predicates.md (move_double_src_operand): Replace the
20285 call to move_double_src_operand with a call to address_operand.
20287 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20289 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
20290 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
20291 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
20293 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20295 * config/arc/predicates.md (long_immediate_loadstore_operand):
20296 Consider scaled addresses cases.
20298 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20300 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
20301 restored when in interrupt.
20302 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
20303 doesn't have delay slot.
20305 2017-03-27 Richard Biener <rguenther@suse.de>
20308 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
20309 inlined thunk clones.
20311 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20314 * asan.c (instrument_derefs): Copy over last operand from
20315 original COMPONENT_REF to the new COMPONENT_REF with
20316 DECL_BIT_FIELD_REPRESENTATIVE.
20317 * ubsan.c (instrument_object_size): Likewise.
20319 2017-03-27 Richard Biener <rguenther@suse.de>
20321 PR tree-optimization/80170
20322 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
20323 sure DR/SCEV didnt fold in constants we do not see when looking
20324 at the reference base alignment.
20326 2017-03-27 Richard Biener <rguenther@suse.de>
20328 PR middle-end/80171
20329 * gimple-fold.c (fold_ctor_reference): Properly guard against
20330 NULL return value from canonicalize_constructor_val.
20332 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
20335 * config/i386/i386.c (ix86_expand_builtin)
20336 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
20337 flags reg setting and flags reg using instructions.
20338 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
20339 clobbering instructions to zero extend op2.
20341 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
20343 * doc/install.texi (Configuration) <--with-aix-soname>:
20344 Update link to AIX ld.
20346 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
20348 PR rtl-optimization/80160
20349 PR rtl-optimization/80159
20350 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
20351 reg_alternate_class into account.
20353 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
20356 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
20357 to consider in curr_insn_transform.
20359 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20361 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
20362 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
20363 and emit_mode_inner.
20365 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20367 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
20368 argument to the overloaded builtin variants. Use the new flag to
20369 deprecate certain builtin variants.
20370 * config/s390/s390-builtin-types.def: Add new builtin types.
20371 * config/s390/s390-builtins.h: Support new flags field for
20372 overloaded builtins.
20373 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
20374 (s390_macro_to_expand): Enable vector float data type.
20375 (s390_cpu_cpp_builtins_internal): Indicate support of the new
20376 builtins by incrementing the __VEC__ version number.
20377 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
20379 (s390_resolve_overloaded_builtin): Emit error messages depending
20380 on the builtin flags.
20381 * config/s390/s390.c (s390_expand_builtin): Support additional
20382 flags argument. Change error message to match the messages
20383 emitted in s390-c.c.
20384 * config/s390/s390.md: New UNSPEC_* constants.
20385 (op_type): Add new instruction types.
20386 * config/s390/vecintrin.h: Add new builtins and test data class
20388 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
20389 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
20390 (VEC_INEXACT, VEC_NOINEXACT): New constants.
20391 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
20392 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
20393 ("vec_mergel<mode>"): V_HW -> VEC_HW.
20395 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
20396 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
20397 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
20398 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
20400 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
20401 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
20402 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
20403 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
20405 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
20406 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
20407 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
20408 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
20409 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
20410 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
20411 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
20413 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
20414 ("vec_scatter_element<V_HW_4:mode>_DI")
20415 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
20416 ("vec_fpint<mode>", "vflls")
20417 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
20418 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
20419 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
20420 ("*vec_cmphe<mode>_cc"): ... these.
20422 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
20423 mode constant instead of magic value.
20425 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20427 * config/s390/s390.c (s390_expand_vec_compare): Support other
20428 vector floating point modes than just V2DF.
20429 (s390_expand_vcond): Likewise.
20430 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20431 (s390_cannot_change_mode_class): Prevent mode changes between TF
20432 and V1TF in vector registers.
20433 * config/s390/s390.md (DF, SF): New mode attributes.
20434 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20435 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20436 SFmode support for VRs.
20437 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20439 (VFT, VF_HW): New mode iterators.
20440 (vw, sdx): New mode attributes.
20441 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20442 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20443 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20444 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20445 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20446 also the new vector floating point modes. Renaming to ...
20448 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20449 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20450 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20451 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20452 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20453 ("vec_unordered<mode>"): ... these.
20455 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20456 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20457 ("*vec_extendv2df"): New insn definitions.
20459 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20461 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20462 ("mulditi3_2", "*muldi3_sign"): New patterns.
20463 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20464 rename the pattern definition.
20466 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20468 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20470 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20472 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20474 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20475 instruction if possible.
20476 * config/s390/vector.md (vec_halfnumelts): New mode
20478 ("*vec_vllezlf<mode>"): New pattern.
20480 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20482 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20483 ("popcountv4si2", "popcountv2di2"): Rename to ...
20484 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20485 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20487 ("popcount<mode>2_vxe"): New pattern.
20489 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20491 * common/config/s390/s390-common.c (processor_flags_table): Add
20493 * config.gcc: Add arch12.
20494 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20495 Default to arch12 for unknown CPU model numbers.
20496 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20497 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20498 PROCESSOR_max sanity check.
20499 * config/s390/s390-opts.h (enum processor_type): Add
20501 * config/s390/s390.c (processor_table): Add arch12.
20502 (s390_expand_builtin): Add check for B_VXE flag.
20503 (s390_issue_rate): Add PROCESSOR_ARCH12.
20504 (s390_get_sched_attrmask): Likewise.
20505 (s390_get_unit_mask): Likewise.
20506 (s390_sched_score): Enable z13 scheduling for arch12.
20507 (s390_sched_reorder): Likewise.
20508 (s390_sched_variable_issue): Likewise.
20509 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20511 (s390_tune_attr): Use z13 scheduling also for arch12.
20512 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20513 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20514 (TARGET_VXE_P): New macros.
20515 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20516 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20517 * config/s390/s390.opt: Add arch12 as processor_type.
20519 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20521 * config/s390/s390.md
20522 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20523 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20524 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20526 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20527 Rename expanders to ...
20529 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20530 ("fixuns_truncdddi2_emu"): ... these.
20532 ("fixuns_trunc<mode>si2_emu"): New expander.
20534 ("*fixuns_truncdfdi2_z13"): Rename to ...
20535 ("*fixuns_truncdfdi2_vx"): ... this.
20537 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20539 * config/s390/2964.md: Remove the single element vector compare
20540 instructions which are no longer used.
20541 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20543 (s390_canonicalize_comparison): Remove handling of DFmode
20545 (s390_expand_vec_compare_scalar): Remove function.
20546 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20547 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20549 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20551 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20553 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20555 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20556 will anyway by matched by mov<mode>_64dfp.
20558 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20560 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20561 vlef/vstef. Add missing operand to vleif.
20563 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20565 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20566 pair for all vector types with 64 bit elements.
20567 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20568 * config/s390/vector.md (V_HW_64): ... here.
20569 (V_128_NOSINGLE): New mode iterator.
20570 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20571 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20572 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20573 ("*vec_load_pairv2di"): Change to ...
20574 ("*vec_load_pair<mode>"): ... this one.
20576 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20578 * config/s390/constraints.md: Add comments.
20579 (jKK): Reject element sizes > 8 bytes.
20580 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20582 * config/s390/s390.md: Add the s_operand checks formerly in
20583 s390_split_ok_p to various splitters where they are still
20585 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20586 for 128 bit vectors. Plus two splitters.
20588 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20590 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20593 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20596 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20597 error if the boundary argument is not constant.
20599 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20601 PR rtl-optimization/80112
20602 * loop-doloop.c (doloop_condition_get): Don't check condition
20603 if cmp isn't SET with IF_THEN_ELSE src.
20605 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20607 PR tree-optimization/80158
20608 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20609 replacing a candidate statement, also replace it for the
20610 candidate's alternate interpretation.
20611 (replace_rhs_if_not_dup): Likewise.
20612 (replace_one_candidate): Likewise.
20614 2017-03-24 Richard Biener <rguenther@suse.de>
20616 PR tree-optimization/80167
20617 * graphite-isl-ast-to-gimple.c
20618 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20620 (translate_isl_ast_to_gimple::get_rename): Likewise.
20622 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20624 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20625 handling of certain combinations of target options, including the
20626 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20627 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20629 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20632 * config/arm/arm.md (*load_multiple): Add reload_completed to
20633 matching condition.
20635 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20636 Richard Biener <rguenth@suse.de>
20638 PR tree-optimization/79908
20639 PR tree-optimization/80136
20640 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20641 been cast away, gimplify_and_add suffices.
20643 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20645 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20647 2017-03-23 Richard Biener <rguenther@suse.de>
20649 PR tree-optimization/80032
20650 * gimplify.c (gimple_push_cleanup): Forced unconditional
20651 cleanups still have to go to the conditional_cleanups
20654 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20656 PR tree-optimization/80072
20657 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20659 (next_operand_entry_id): Change type to unsigned int.
20660 (sort_by_operand_rank): Make sure to return the right return value
20661 even if unsigned fields are bigger than INT_MAX.
20662 (struct oecount): Change cnt and id type to unsigned int.
20663 (oecount_hasher::equal): Formatting fix.
20664 (oecount_cmp): Make sure to return the right return value
20665 even if unsigned fields are bigger than INT_MAX.
20666 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
20669 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
20670 TREE_READONLY on result if writing it more than once.
20673 * doc/invoke.texi (-fsanitize=thread): Document that with
20674 -fnon-call-exceptions atomics are not able to throw
20678 * tsan.c: Include tree-eh.h.
20679 (instrument_builtin_call): Call maybe_clean_eh_stmt or
20680 maybe_clean_or_replace_eh_stmt where needed.
20681 (instrument_memory_accesses): Add cfg_changed argument.
20682 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
20683 if it returned true.
20684 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20686 PR rtl-optimization/63191
20687 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20688 wrapper function, moved the whole old content into ...
20689 (ix86_delegitimize_address_1): ... this. New inline function.
20690 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20691 true as last argument instead of ix86_delegitimize_address.
20693 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20695 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20696 cortexa57_branch_cost.
20698 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20700 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20702 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20705 * doc/md.texi (Constraints): Document wA constraint.
20706 * config/rs6000/constraints.md (wA): New.
20707 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20708 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20709 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20710 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20712 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20715 * gimplify.c (is_oacc_declared): New function.
20716 (oacc_default_clause): Use it to set default flags for acc declared
20717 variables inside parallel regions.
20718 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20719 declared variables.
20720 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20721 declare attribute to any decl as necessary.
20723 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20726 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20727 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20728 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20729 (arm_arch_lpae): This.
20730 * config/arm/arm.c (arm_arch7ve): Rename into ...
20731 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20732 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20735 2017-03-22 Martin Liska <mliska@suse.cz>
20738 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20739 error message instead of an ICE.
20741 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20743 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20745 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20747 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20749 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20752 2017-03-21 Martin Sebor <msebor@redhat.com>
20754 * doc/extend.texi: Use "cannot" instead of "can't."
20755 * doc/hostconfig.texi: Same.
20756 * doc/install.texi: Same.
20757 * doc/invoke.texi: Same.
20758 * doc/loop.texi: Same.
20759 * doc/md.texi: Same.
20760 * doc/objc.texi: Same.
20761 * doc/rtl.texi: Same.
20762 * doc/tm.texi: Same.
20763 * doc/tm.texi.in: Same.
20764 * doc/trouble.texi: Same.
20766 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20769 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20770 (collect_checksum_attributes): Set it.
20771 (die_checksum_ordered): Use it.
20773 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20775 PR tree-optimization/79908
20776 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
20777 change: For a VA_ARG whose LHS has been cast away, use
20778 force_gimple_operand to construct the side effects.
20780 2017-03-21 David Malcolm <dmalcolm@redhat.com>
20782 PR translation/80001
20783 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
20784 more amenable to translation.
20785 (oacc_loop_auto_partitions): Likewise.
20787 2017-03-21 Marek Polacek <polacek@redhat.com>
20788 Martin Sebor <msebor@redhat.com>
20790 PR tree-optimization/80109
20791 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
20792 on INTEGRAL_TYPE_P.
20794 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20795 Segher Boessenkool <segher@kernel.crashing.org>
20798 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
20799 check reg_used_between_p between insn and one of succ or succ2
20800 depending on if succ is artificial insn not inserted into insn
20803 2017-03-21 Martin Liska <mliska@suse.cz>
20805 PR gcov-profile/80081
20806 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
20807 * doc/gcc.texi: Include gcov-dump stuff.
20808 * doc/gcov-dump.texi: New file.
20810 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
20812 PR rtl-optimization/79150
20813 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
20814 conditional jump, if the jump is the last insn of the loop.
20816 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20817 Richard Biener <rguenth@suse.de>
20819 PR tree-optimization/79908
20820 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20821 been cast away, use force_gimple_operand to construct the side
20824 2017-03-21 Martin Liska <mliska@suse.cz>
20826 PR libfortran/79956
20827 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
20830 2017-03-21 Brad Spengler <spender@grsecurity.net>
20833 * plugin.c (htab_hash_plugin): New function.
20834 (add_new_plugin): Use it and adjust.
20835 (parse_plugin_arg_opt): Adjust.
20836 (init_one_plugin): Likewise.
20838 2017-03-21 Richard Biener <rguenther@suse.de>
20840 PR tree-optimization/80032
20841 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
20842 if set force the cleanup to happen unconditionally.
20843 (gimplify_target_expr): Push inserted clobbers with force_uncond
20844 to avoid them being removed by control-dependent DCE.
20846 2017-03-21 Richard Biener <rguenther@suse.de>
20848 PR tree-optimization/80122
20849 * tree-inline.c (copy_bb): Do not expans va-arg packs or
20850 va_arg_pack_len when the inlined call stmt requires pack
20852 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
20854 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20857 * tsan.c (instrument_builtin_call): If the memory model argument
20858 is not a constant, assume it is valid.
20861 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
20864 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20866 PR rtl-optimization/79910
20867 * combine.c (can_combine_p): Do not allow combining an I0 or I1
20868 if its dest is used by an insn before I2 (other than the combined
20869 insns themselves, which are properly handled already).
20871 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20874 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20876 * combine.c (record_used_regs): New static function.
20877 (try_combine): Handle situations where there is an additional
20878 instruction between I2 and I3 which needs to have a LOG_LINK
20882 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20884 * combine.c (try_combine): Delete redundant i1 test. Call
20885 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20887 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20890 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
20891 alternatives 13/14.
20893 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20895 PR tree-optimization/80054
20896 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
20897 the optimization if a PHI or any of its arguments is not dominated
20898 by the candidate's basis. Use gphi* rather than gimple* as
20900 (replace_profitable_candidates): Clean up a gimple* variable that
20901 should be a gphi* variable.
20903 2017-03-20 Martin Sebor <msebor@redhat.com>
20906 * doc/extend.texi (attribute constructor): Document present limitation.
20908 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
20911 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
20912 __POWER9_VECTOR__ #ifdef control, change template definition to
20913 use Power9-specific built-in function.
20914 (vec_any_eq): Likewise.
20915 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
20916 to control outcomes from this test.
20917 (vector_ae_<mode>p): For VEC_F modes, likewise.
20919 2017-03-20 Ian Lance Taylor <iant@google.com>
20921 * config/i386/i386.c (ix86_function_regparm): Save an extra
20922 register for -fsplit-stack with DECL_STATIC_CHAIN.
20924 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20927 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
20928 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
20930 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20932 * config/riscv/riscv.c (riscv_print_operand): Use "fence
20934 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
20937 2017-03-20 Marek Polacek <polacek@redhat.com>
20940 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
20942 2017-03-20 Richard Biener <rguenther@suse.de>
20944 PR tree-optimization/80113
20945 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
20946 allocate extra SSA name for PHI def.
20947 (add_close_phis_to_outer_loops): Likewise.
20948 (add_close_phis_to_merge_points): Likewise.
20949 (copy_loop_close_phi_args): Likewise.
20950 (copy_cond_phi_nodes): Likewise.
20952 2017-03-20 Martin Liska <mliska@suse.cz>
20954 PR middle-end/79753
20955 * tree-chkp.c (chkp_build_returned_bound): Do not build
20956 returned bounds for a LHS that's not a BOUNDED_P type.
20958 2017-03-20 Martin Liska <mliska@suse.cz>
20962 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
20963 COMPLEX_CST and VECTOR_CST.
20965 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20968 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
20969 target operand. A new splitter adds the clobber statement in case
20970 the target operand is dead anyway.
20972 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
20974 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
20975 to age-old versions of binutils and glibc.
20977 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
20979 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
20981 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20983 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
20985 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20987 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
20988 requirement for binutils 2.13.
20990 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20992 * combine.c (try_combine): Delete redundant i1 test. Call
20993 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20995 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
20997 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
20998 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
21000 <riscv64-*-elf>: Re-arrange section
21001 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
21002 <riscv32-*-linux>: Likewise.
21003 <riscv64-*-elf>: Likewise
21004 <riscv64-*-linux>: Likewise.
21006 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
21009 * aarch64.opt(verbose-cost-dump): Fix typo.
21011 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
21014 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
21015 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
21017 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
21019 * reload.c (find_reloads): When reloading a nonoffsettable address,
21020 use RELOAD_OTHER for it and its address reloads.
21022 PR rtl-optimization/79910
21023 * combine.c (record_used_regs): New static function.
21024 (try_combine): Handle situations where there is an additional
21025 instruction between I2 and I3 which needs to have a LOG_LINK
21028 2017-03-17 Jeff Law <law@redhat.com>
21030 PR tree-optimization/71437
21031 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
21032 conditional in the hash table first.
21033 (vrp_dom_walker::before_dom_children): Extract condition from
21034 ASSERT_EXPR. Record condition, its inverion and any implied
21035 conditions as well.
21037 2017-03-17 Marek Polacek <polacek@redhat.com>
21038 Markus Trippelsdorf <markus@trippelsdorf.de>
21040 PR tree-optimization/80079
21041 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
21044 2017-03-17 Richard Biener <rguenther@suse.de>
21046 PR middle-end/80075
21047 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
21048 Properly verify the LHS before the RHS possibly claims to be
21050 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
21053 2017-03-17 Martin Jambor <mjambor@suse.cz>
21055 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
21056 (List of -O2 options): Likewise.
21057 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
21060 2017-03-17 Tom de Vries <tom@codesourcery.com>
21062 * gcov-dump.c (print_usage): Print bug_report_url.
21064 2017-03-17 Richard Biener <rguenther@suse.de>
21066 PR middle-end/80050
21067 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
21068 (parser::peek): Likewise.
21070 2017-03-17 Richard Biener <rguenther@suse.de>
21072 PR tree-optimization/80048
21073 * sese.c (free_sese_info): Properly release rename_map and
21074 copied_bb_map elements.
21076 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
21078 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
21079 Add linked-list forward and backlinks. Insert on
21080 construction, remove on destruction.
21081 (class pass_store_merging): Add m_stores_head field.
21082 (pass_store_merging::terminate_and_process_all_chains):
21083 Iterate over m_stores_head list.
21084 (pass_store_merging::terminate_all_aliasing_chains):
21086 (pass_store_merging::execute): Check for debug stmts first.
21087 Push new chains onto the m_stores_head stack.
21089 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21092 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
21093 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
21094 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
21096 2017-03-16 Jeff Law <law@redhat.com>
21098 PR tree-optimization/71437
21099 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
21100 member function. Implementation moved into after_dom_children
21101 member function and into the threader's thread_outgoing_edges
21103 (dom_opt_dom_walker::after_dom_children): Simplify by moving
21104 some code into new thread_outgoing_edges.
21105 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
21106 definition. Simplify marker handling (do it here). Assume we always
21107 have the available expression and the const/copies tables.
21108 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
21110 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
21111 * tree-vrp.c (equiv_stack): No longer file scoped.
21112 (vrp_dom_walker): New class.
21113 (vrp_dom_walker::before_dom_children): New member function.
21114 (vrp_dom_walker::after_dom_children): Likewise.
21115 (identify_jump_threads): Setup domwalker. Use it rather than
21116 walking edges in a random order by hand. Simplify setup/finalization.
21117 (finalize_jump_threads): Remove.
21118 (vrp_finalize): Do not call identify_jump_threads here.
21119 (execute_vrp): Do it here instead and call thread_through_all_blocks
21122 PR tree-optimization/71437
21123 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
21125 (simplify_stmt_for_jump_threading): Add basic_block argument. All
21127 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
21128 (dom_opt_dom_walker::thread_across_edge): Remove
21129 handle_dominating_asserts argument. All callers changed.
21130 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
21131 changes. Remove calls to lhs_of_dominating_assert. Other
21132 uses of handle_dominating_asserts turn into unconditional code
21133 (simplify_control_stmt_condition_1): Likewise.
21134 (simplify_control_stmt_condition): Likewise.
21135 (thread_through_normal_block, thread_across_edge): Likewise.
21136 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
21137 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
21138 object if it is not an SSA_NAME.
21139 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
21140 before calling into the VRP specific simplifiers.
21141 (identify_jump_threads): Remove handle_dominating_asserts
21144 2017-03-16 Jakub Jelinek <jakub@redhat.com>
21147 * tree-diagnostic.c (default_tree_printer): No longer static.
21148 * tree-diagnostic.h (default_tree_printer): New prototype.
21150 2017-03-16 Tamar Christina <tamar.christina@arm.com>
21152 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
21153 Change ins into fmov.
21155 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21157 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
21158 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
21159 Use h_con constraint for operand 1.
21160 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
21161 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
21163 2017-03-15 Jeff Law <law@redhat.com>
21165 PR tree-optimization/71437
21166 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
21167 (record_temporary_equivalences): Use it.
21169 PR tree-optimization/71437
21170 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
21171 tree-ssa-scopedtables.
21172 (lookup_avail_expr, build_and_record_new_cond): Likewise.
21173 (record_conditions, record_cond, vuse_eq): Likewise.
21174 (record_edge_info): Adjust to API tweak of record_conditions.
21175 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
21176 (record_temporary_equivalences, optimize_stmt): Likewise.
21177 (eliminate_redundant_computations): Likewise.
21178 (record_equivalences_from_stmt): Likewise.
21179 * tree-ssa-scopedtables.c: Include options.h and params.h.
21180 (vuse_eq): New function, moved from tree-ssa-dom.c
21181 (build_and_record_new_cond): Likewise.
21182 (record_conditions): Likewise. Accept vector of conditions rather
21183 than edge_equivalence structure for first argument.
21184 for the first argument.
21185 (avail_exprs_stack::lookup_avail_expr): New member function, moved
21186 from tree-ssa-dom.c.
21187 (avail_exprs_stack::record_cond): Likewise.
21188 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
21189 from tree-ssa-dom.c.
21190 (avail_exprs_stack): Add new member functions lookup_avail_expr
21192 (record_conditions): Declare.
21194 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
21197 * lra-constraints.c (process_alt_operands): Increase reject for
21198 reloading an input/output operand.
21200 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
21203 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
21204 insns to convert from signed/unsigned char/short to IEEE 128-bit
21206 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
21208 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
21211 * config/i386/i386.c (ix86_vector_duplicate_value): Create
21212 subreg of inner mode for values already in registers.
21214 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
21216 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
21217 iteration reg is used after the loop.
21219 2017-03-14 Martin Sebor <msebor@redhat.com>
21221 PR tree-optimization/79800
21222 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
21223 precision in negative-positive range.
21224 (format_floating): Call non-const overload with adjusted precision.
21226 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21229 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
21232 2017-03-14 Martin Sebor <msebor@redhat.com>
21234 PR middle-end/80020
21235 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
21236 * builtins.def (aligned_alloc): Use it.
21239 * Makefile.in (GTFILES): Add calls.c.
21240 * calls.c: Include "gt-calls.h".
21242 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
21244 PR rtl-optimization/79728
21245 * regs.h (struct target_regs): New field
21246 x_contains_allocatable_regs_of_mode.
21247 (contains_allocatable_regs_of_mode): New macro.
21248 * reginfo.c (init_reg_sets_1): Initialize it, and change
21249 contains_reg_of_mode so it includes global regs as well.
21250 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
21251 rather than contains_regs_of_mode.
21253 2017-03-14 Martin Liska <mliska@suse.cz>
21255 * doc/invoke.texi: Document options that can't be combined with
21256 -fcheck-pointer-bounds.
21258 2017-03-14 Martin Liska <mliska@suse.cz>
21260 PR middle-end/79831
21261 * doc/invoke.texi (-Wchkp): Document the option.
21263 2017-03-14 Martin Liska <mliska@suse.cz>
21265 * Makefile.in: Install gcov-dump.
21267 2017-03-14 Martin Liska <mliska@suse.cz>
21269 * multiple_target.c (expand_target_clones): Bail out for
21270 an invalid attribute.
21272 2017-03-14 Richard Biener <rguenther@suse.de>
21274 * alias.c (struct alias_set_entry): Pack properly.
21275 * cfgloop.h (struct loop): Likewise.
21276 * cse.c (struct set): Likewise.
21277 * ipa-utils.c (struct searchc_env): Likewise.
21278 * loop-invariant.c (struct invariant): Likewise.
21279 * lra-remat.c (struct cand): Likewise.
21280 * recog.c (struct change_t): Likewise.
21281 * rtl.h (struct address_info): Likewise.
21282 * symbol-summary.h (function_summary): Likewise.
21283 * tree-loop-distribution.c (struct partition): Likewise.
21284 * tree-object-size.c (struct object_size_info): Likewise.
21285 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
21286 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
21287 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
21288 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
21289 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
21290 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
21291 (struct _stmt_vec_info): Likewise.
21293 2017-03-14 Martin Liska <mliska@suse.cz>
21296 * multiple_target.c (create_dispatcher_calls): Check that
21297 a target can create a function dispatcher.
21299 2017-03-14 Martin Liska <mliska@suse.cz>
21302 * multiple_target.c (expand_target_clones): Drop local.local
21303 flag for default implementation.
21305 2017-03-14 Richard Biener <rguenther@suse.de>
21307 PR tree-optimization/80030
21308 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
21310 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
21312 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
21313 gcc_fallthrough() instead of __attribute__((fallthrough));
21315 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21317 * doc/gcc.texi: Remove "up" link to (DIR).
21318 * doc/gccint.texi: Ditto.
21320 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21322 * doc/install.texi (Specific) <avr>: Remove reference to
21325 2017-03-13 Jeff Law <law@redhat.com>
21327 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
21328 attribute rather than comments.
21330 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
21331 match_scratch operand is highest.
21333 2017-03-13 Martin Liska <mliska@suse.cz>
21335 PR middle-end/78339
21336 * ipa-pure-const.c (warn_function_noreturn): If the declarations
21337 is a CHKP clone, use original declaration.
21339 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21341 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
21342 (arc_conditional_register_usage): Use a different allocation order
21343 when optimizing for size.
21344 * common/config/arc/arc-common.c (arc_option_optimization_table):
21345 Section anchors default on when optimizing for size.
21347 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21349 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
21351 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21353 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
21354 * config/arc/arc.md (cpu_facility): Add cd variant.
21355 (*movqi_insn): Add code density variant.
21356 (*movhi_insn): Likewise.
21357 (*movqi_insn): Likewise.
21358 (*addsi3_mixed): Likewise.
21359 (subsi3_insn): Likewise.
21361 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21363 * config/arc/arc.md (movsi_cond_exec): Update constraint.
21365 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21367 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
21368 expressions with MINUS and UNARY ops.
21370 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21373 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
21375 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
21376 between vec_select and vector argument.
21377 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
21378 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
21379 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
21380 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
21381 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
21382 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
21384 2017-03-13 Richard Biener <rguenther@suse.de>
21387 * params.def (vect-max-peeling-for-alignment): Fix typo.
21389 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21391 * doc/install.texi (Specific) <mips-*-*>: Remove description of
21392 issue that only occurred with binutils below 2.18.
21394 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21396 * doc/install.texi (Specific) <cris-axis-elf>: No longer
21397 refer to binutils 2.11/2.12 minimum.
21399 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21401 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
21402 ftp.kernel.org and simplify binutils requirement.
21404 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
21406 * doc/invoke.texi (Warning Options): Fix spelling of link-time
21408 (Optimize Options): Ditto. Also remove redundancy.
21410 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21412 PR translation/79848
21413 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
21415 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
21416 to G_ to avoid double translation.
21418 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21420 PR translation/79923
21421 * auto-profile.c (get_combined_location): Convert leading
21422 character of diagnostics to lower case and remove trailing period.
21423 (read_profile): Likewise for various diagnostics.
21424 * config/arm/arm.c (arm_option_override): Remove trailing period
21425 from various diagnostics.
21426 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
21427 (msp430_expand_delay_cycles): Likewise.
21429 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21432 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21433 full command-line argument, rather than just "str".
21434 (aarch64_validate_march): Likewise.
21435 (aarch64_validate_mtune): Likewise.
21437 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21439 PR rtl-optimization/78911
21440 * lra-assigns.c (must_not_spill_p): New function.
21441 (spill_for): Use it.
21443 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21445 PR tree-optimization/79981
21446 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21447 ATOMIC_COMPARE_EXCHANGE ifn result.
21448 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21449 IFN_ATOMIC_COMPARE_EXCHANGE.
21451 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21454 * opts.c (parse_sanitizer_options): Add missing question mark to
21455 "did you mean" message.
21457 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21459 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21461 (VMULEUH_UNS): Likewise.
21462 (VMULOUB_UNS): Likewise.
21463 (VMULOUH_UNS): Likewise.
21464 * config/rs6000/rs6000.c (builtin_function_type): Remove
21465 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21467 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21470 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21471 x with result of extra_parsing_for_operand_code_0.
21472 (function_reader::extra_parsing_for_operand_code_0): Convert
21473 return type from void to rtx, returning x. When reading
21474 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21475 larger size containing struct block_symbol.
21477 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21479 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21480 -mfloat128-hardware without -m64.
21482 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21485 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21486 entries to the case statement that marks unsigned arguments to
21487 overloaded functions.
21489 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21491 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21492 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21494 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21497 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21498 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21500 2017-03-10 Martin Liska <mliska@suse.cz>
21504 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21505 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21508 2017-03-10 Olivier Hainque <hainque@adacore.com>
21510 * tree-switch-conversion (array_value_type): Start by resetting
21511 candidate type to it's main variant.
21513 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21515 PR rtl-optimization/79909
21516 * combine.c (try_combine): Use simplify_replace_rtx on individual
21517 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21518 of the whole CALL_INSN_FUNCTION_USAGE.
21520 PR tree-optimization/79972
21521 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21522 get_range_info on SSA_NAMEs. Formatting fixes.
21524 2017-03-10 Richard Biener <rguenther@suse.de>
21525 Jakub Jelinek <jakub@redhat.com>
21527 PR tree-optimization/77975
21528 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21529 edge to be constant.
21530 (get_val_for): For constant x return it. Formatting fix.
21531 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21532 would use the same bases as the current one.
21534 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21536 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21537 instead of vec_select for V1TImode.
21538 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21540 (VSX_LE_128): Add V1TI to this mode iterator.
21541 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21542 (*vsx_le_perm_store_<mode>): Likewise.
21543 (pre-reload splitter for VSX stores): Likewise.
21544 (post-reload splitter for VSX stores): Likewise.
21545 (*vsx_xxpermdi2_le_<mode>): Likewise.
21546 (*vsx_lxvd2x2_le_<mode>): Likewise.
21547 (*vsx_stxvd2x2_le_<mode>): Likewise.
21549 2017-03-09 Michael Eager <eager@eagercon.com>
21551 Correct failures with --enable-checking=yes,rtl.
21553 * config/microblaze/microblaze.c (microblaze_expand_shift):
21554 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21555 test for const0_rtx.
21556 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21557 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21559 2017-03-09 Richard Biener <rguenther@suse.de>
21561 PR tree-optimization/79977
21562 * graphite-scop-detection.c (scop_detection::merge_sese):
21563 Handle the case of extra exits to blocks dominating the entry.
21565 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21567 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21570 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21572 PR rtl-optimization/79949
21573 * lra-constraints.c (process_alt_operands): Check memory when
21574 trying to predict a cycle. Print about the overall increase.
21576 2017-03-09 Richard Biener <rguenther@suse.de>
21578 PR middle-end/79971
21579 * gimple-expr.c (useless_type_conversion_p): Preserve
21580 TYPE_SATURATING for fixed-point types.
21582 2017-03-09 Richard Biener <rguenther@suse.de>
21585 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21586 alignment of BLKmode params.
21588 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21591 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21592 (VALL_NO_V2Q): Likewise.
21594 * config/aarch64/aarch64-simd.md
21595 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21597 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21598 VALL_NO_V2Q mode iterator.
21599 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21601 2017-03-09 Martin Liska <mliska@suse.cz>
21603 PR tree-optimization/79631
21604 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21605 tree_int_cst_sign_bit just for INTEGER constants.
21607 2017-03-09 Martin Liska <mliska@suse.cz>
21611 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21614 2017-03-09 Marek Polacek <polacek@redhat.com>
21617 * tree.c (inchash::add_expr): Handle TREE_VEC.
21619 2017-03-09 Martin Liska <mliska@suse.cz>
21622 (chkp_narrow_size_and_offset): New function.
21623 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21624 (void chkp_parse_bit_field_ref): New function.
21625 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21626 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21628 2017-03-09 Martin Liska <mliska@suse.cz>
21631 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21632 (chkp_find_bounds_1): Remove gcc_unreachable.
21634 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21637 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21638 BUILT_IN_SYNC*, determine the access type from the size suffix and
21639 always build a MEM_REF with that type. Handle forgotten
21640 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21643 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21644 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21645 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21646 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21647 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21648 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21649 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21650 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21651 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21652 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21653 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21654 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21655 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21656 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21657 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21658 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21659 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21660 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21661 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21662 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21663 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21664 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21665 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21666 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21667 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21668 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
21669 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
21670 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
21671 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
21672 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
21673 definitions outside of __OPTIMIZE__ guarded section.
21676 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
21677 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
21678 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
21679 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
21682 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21684 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21685 ("vfenez<mode>"): Add missing constraints.
21687 2017-03-08 Martin Sebor <msebor@redhat.com>
21690 * config/nds32/nds32.c (nds32_option_override):
21691 Fix misspelled diagnostic.
21693 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21696 * gimplify.c (gimplify_omp_for): Replace index var in outer
21697 taskloop statement with an artificial variable and add
21698 OMP_CLAUSE_PRIVATE clause for it.
21700 2017-03-08 Richard Biener <rguenther@suse.de>
21702 PR tree-optimization/79955
21703 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21704 for accesses that are completely outside of the variable.
21706 2017-03-08 Andrew Haley <aph@redhat.com>
21708 PR tree-optimization/79943
21709 * tree-ssa-loop-split.c (compute_new_first_bound): When
21710 calculating the new upper bound, (END-BEG) should be added, not
21713 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21715 * config/avr/avr.md (setmemhi): Make sure match_dup
21716 operand number comes before match_scratch.
21718 2017-03-08 Richard Biener <rguenther@suse.de>
21720 PR tree-optimization/79920
21721 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21722 with ncopies == 1 to ...
21723 (vect_transform_slp_perm_load): ... here. Properly compute
21724 all element loads by iterating VF times over the group. Do
21725 not handle ncopies (computed in a broken way) in
21726 vect_create_mask_and_perm.
21728 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21731 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21732 is a uniform vector, use uniform_vector_p return value instead of
21733 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21735 2017-03-07 Marek Polacek <polacek@redhat.com>
21737 PR middle-end/79809
21738 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21739 (alloca_call_type): Likewise.
21741 2017-03-07 Martin Liska <mliska@suse.cz>
21743 * gcov.c (process_args): Put comment to correct location.
21745 2017-03-07 Martin Liska <mliska@suse.cz>
21747 PR middle-end/68270
21748 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21749 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21750 (chkp_narrow_bounds_for_field): Likewise.
21751 (chkp_parse_array_and_component_ref): Pass one more argument to
21754 2017-03-07 Richard Biener <rguenther@suse.de>
21756 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21759 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21761 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21762 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21764 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21767 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21768 to end of description.
21769 (PARAM_MAX_STORES_TO_MERGE): Likewise.
21771 2017-03-07 Jakub Jelinek <jakub@redhat.com>
21773 PR rtl-optimization/79901
21774 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
21776 (*avx512f_<code><mode>3<mask_name>): ... this.
21777 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
21778 iterator instead of VI8_AVX2_AVX512BW.
21780 PR rtl-optimization/79901
21781 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
21782 min/max expander, expand it using expand_vec_cond_expr.
21785 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
21788 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21791 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
21792 to void * for PCH reasons.
21793 * dwarf2out.c (output_loc_operands, output_die): Cast
21794 v.val_vec.array to unsigned char *.
21796 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
21799 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
21802 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
21804 PR rtl-optimization/79571
21805 * lra-constraints.c (process_alt_operands): Calculate static
21806 reject and subtract it from overall when only addresses will be
21809 2017-03-06 Julia Koval <julia.koval@intel.com>
21812 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
21813 incoming stack boundary to 128 for 64-bit targets.
21815 2017-03-06 Richard Biener <rguenther@suse.de>
21817 PR tree-optimization/79894
21818 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
21819 to NULL after folding it.
21821 2017-03-06 Richard Biener <rguenther@suse.de>
21823 PR tree-optimization/79824
21824 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
21825 check disabling peeling for gaps.
21827 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
21829 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21830 attributes): Document gettimeofday.
21832 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21834 * config/s390/s390.c (s390_option_override_internal): Set
21835 PARAM_MIN_VECT_LOOP_BOUND
21837 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21839 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
21840 * config/s390/s390.md: Likewise.
21842 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21845 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
21846 (<avx2_avx512>_perm<mode>): Rename to ...
21847 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
21849 (<avx512>_perm<mode>_mask): Rename to ...
21850 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
21852 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
21853 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
21854 instead of VI8F_256_512.
21855 (avx512f_perm<mode>): New define_expand.
21856 (avx512f_perm<mode>_mask): Likewise.
21857 (avx512f_perm<mode>_1<mask_name>): New define_insn.
21858 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
21860 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21862 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
21863 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
21865 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
21867 2017-03-06 Martin Liska <mliska@suse.cz>
21870 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
21871 when having a SSA NAME w/o VAR_DECL assigned to it.
21873 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21875 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
21876 msa_dpsub_<su>_d): Fix MODE for vec_select.
21878 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21880 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
21882 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
21884 2017-03-06 Richard Biener <rguenther@suse.de>
21886 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
21887 * plugin.c (register_plugin_info): Likewise.
21888 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
21890 2017-03-05 Jakub Jelinek <jakub@redhat.com>
21892 * config/i386/sse.md (sse_storehps, sse_storelps,
21893 avx_<castmode><avxsizesuffix>_<castmode>,
21894 avx512f_<castmode><avxsizesuffix>_<castmode>,
21895 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
21896 in condition that at least one operand is not a MEM.
21898 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21900 PR middle-end/79805
21901 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
21902 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
21904 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
21905 gimple_call_nothrow_p flag based on whether original builtin can throw.
21906 If it can, emit following stmts on the fallthrough edge.
21907 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
21908 don't create new bb if inserting just debug stmts on the edge, try to
21909 insert them on the fallthru bb or just reset debug stmts.
21911 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
21914 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
21915 restore recog_data (including the operand rtxes inside it) around
21916 the call to get_insn_template.
21918 2017-03-03 Martin Sebor <msebor@redhat.com>
21920 PR tree-optimization/79699
21921 * context.c (context::~context): Free MPFR caches to avoid
21922 a memory leak on program exit.
21924 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21926 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
21927 Use wide_int::ulow () instead of .elt (0).
21929 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21931 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
21932 (*pushxf): Limit oF constraint to 32bit targets and add oC
21933 constraint for 64bit targets.
21934 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
21935 (*pushdf): Change rmF constraint to rmC.
21937 2017-03-03 Martin Liska <mliska@suse.cz>
21939 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
21940 Remove unused variable.
21942 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21945 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
21946 is a memory operand, increase num_memory.
21947 (ix86_expand_args_builtin): Likewise.
21949 2017-03-03 Jan Hubicka <jh@suse.cz>
21952 * ipa-devirt.c (maybe_record_node): Properly handle
21953 __cxa_pure_virtual visibility.
21955 2017-03-03 Martin Liska <mliska@suse.cz>
21957 PR tree-optimization/79803
21958 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
21960 (pass_loop_prefetch::execute): Disabled optimization if an
21961 assumption about L1 cache size is not met.
21963 2017-03-03 Martin Liska <mliska@suse.cz>
21965 PR rtl-optimization/79574
21966 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
21967 (hash_scan_set): Likewise.
21968 (dump_hash_table): Likewise.
21969 (hoist_code): Likewise.
21971 2017-03-03 Richard Biener <rguenther@suse.de>
21973 * fixed-value.c (fixed_from_string): Restore use of elt (1)
21974 in place of uhigh ().
21975 (fixed_convert_from_real): Likewise.
21977 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21980 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
21982 2017-03-03 Richard Biener <rguenther@suse.de>
21984 PR middle-end/79818
21985 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
21986 TYPE_OVERFLOW_UNDEFINED check.
21988 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21990 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
21992 (vector_ae_<mode>_p): Likewise.
21993 (vector_nez_<mode>_p): Likewise.
21994 (vector_ne_v2di_p): Likewise.
21995 (vector_ae_v2di_p): Likewise.
21996 (vector_ne_<mode>_p): Likewise.
21997 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
21999 (vsx_tsqrt<mode>2_fe): Likewise.
22001 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
22004 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
22006 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22008 PR rtl-optimization/79780
22009 * cprop.c (one_cprop_pass): When second and further conditional trap
22010 in a single basic block is turned into an unconditional trap, turn it
22011 into a deleted note to avoid RTL verification failures.
22013 2017-03-02 Richard Biener <rguenther@suse.de>
22015 * fold-const.c (const_binop): Use ulow () instead of elt (0).
22017 2017-03-02 Richard Biener <rguenther@suse.de>
22019 PR tree-optimization/79345
22021 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
22022 param and abort the walk, returning -1 if it is hit.
22023 (walk_aliased_vdefs): Take a limit param and pass it on.
22024 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
22025 defaulting to 0 and return a signed int.
22026 * tree-ssa-uninit.c (struct check_defs_data): New struct.
22027 (check_defs): New helper.
22028 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
22029 about uninitialized memory.
22030 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
22031 bogus uninitialized warning.
22032 (fixed_convert_from_real): Likewise.
22034 2017-03-02 Bin Cheng <bin.cheng@arm.com>
22036 PR tree-optimization/66768
22037 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
22038 iv_use if base object can't be determined.
22040 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22042 PR tree-optimization/79345
22043 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
22044 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
22045 (get_pattern_stats): Initialize it.
22046 * genemit.c (gen_expand): Verify match_scratch numbers come after
22047 match_operand/match_dup numbers.
22048 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
22049 match_scratch numbers.
22050 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
22052 * config/s390/s390.md (trunctdsd2): Likewise.
22054 2017-03-02 Richard Biener <rguenther@suse.de>
22056 * wide-int.h (wide_int_storage::operator=): Implement in terms
22059 2017-03-02 Richard Biener <rguenther@suse.de>
22061 PR tree-optimization/79777
22062 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
22063 the to insert expression to sth existing.
22065 2017-03-01 Martin Sebor <msebor@redhat.com>
22067 PR middle-end/79692
22068 * gimple-ssa-sprintf.c
22069 (directive::known_width_and_precision): New function.
22070 (format_integer): Use it.
22071 (get_mpfr_format_length): Consider the full range of precision
22072 when computing %g output with the # flag. Set the likely byte
22073 count to 3 rather than 1 when precision is indeterminate.
22074 (format_floating): Correct the lower bound of precision.
22076 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22078 * doc/invoke.texi: Document default code model for 64-bit Linux.
22080 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22083 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
22084 udiv rather than div since input pattern is unsigned.
22086 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
22088 * config/i386/i386.c (print_reg): Warn for values of
22089 unsupported size in integer register.
22091 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
22094 * config/rs6000/predicates.md (current_file_function_operand): Do
22095 not allow self calls to be local if the function is replaceable.
22097 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
22100 * config/rs6000/altivec.h (vec_ctz and others): Change the
22101 preprocessor macro that controls conditional compilation from
22102 _ARCH_PWR9 to __POWER9_VECTOR__.
22103 (vec_all_ne): Change parameterization of __altivec_scalar_pred
22104 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
22105 control (instead of _ARCH_PWR9 control) so that template
22106 definition uses power9-specific function.
22107 (vec_any_eq): Likewise.
22108 (vec_all_ne): Change macro definition to use a power9-specific
22109 expansion under #ifdef __POWER9_VECTOR__ control (instead of
22110 _ARCH_PWR9 control).
22111 (vec_any_eq) Likewise.
22112 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
22113 expansion for CMPNEF to remove support for xvcmpnesp instruction.
22114 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
22115 support for xvcmpnedp instruction.
22116 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
22117 macro expansion so that Power9 implementation of vec_all_ne does
22118 not use the AltiVec predicate framework.
22119 (VCMPNEH_P): Likewise.
22120 (VCMPNEW_P): Likewise.
22121 (VCMPNED_P): Likewise.
22122 (VCMPNEFP_P): Likewise.
22123 (VCMPNEDP_P): Likewise.
22124 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
22125 implementation of vec_any_eq to not use AltiVec predicate
22127 (VCMPAEH_P): Likewise.
22128 (VCMPAEW_P): Likewise.
22129 (VCMPAED_P): Likewise.
22130 (VCMPAEFP_P): Likewise.
22131 (VCMPAEDP_P): Likewise.
22132 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
22133 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
22134 not use the AltiVec predicate framework.
22135 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
22136 of vec_any_eq to not use AltiVec predicate framework.
22137 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22138 support for predefined __POWER9_VECTOR__ macro to indicate that
22139 Power9 instruction selection is enabled.
22140 (altivec_overloaded_builtins): Remove extraneous
22141 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
22142 function argument types RS6000_BTI_bool_V16QI and
22143 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
22144 entry for overloaded function argument types RS6000_BTI_bool_V4SI
22145 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
22146 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
22147 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
22148 Power9 for implementations of vec_cmpne. Change the signature for
22149 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
22150 (representing vec_all_ne) to remove the previously described first
22151 argument of type RS6000_BTI_INTSI, as this was an artifact of
22152 reliance on the AltiVec predicate framework, which is no longer
22153 used in the implementation of these functions. Add
22154 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
22155 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
22156 since, unlike the AltiVec predicate framework implementation, we
22157 do not share function descriptors between vec_alle and vec_anyeq.
22158 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
22159 set of modes that receive special treatment even when
22160 TARGET_P9_VECTOR is true. The special treatment emits code that
22161 does not depend on Power9 instructions.
22162 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
22163 define_expand to not rely on AltiVec predicate framework.
22164 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22166 (vector_ne_v2di_p): Change this define_expand to not rely on
22167 AltiVec predicate framework.
22168 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
22170 (vector_ne_<mode>_p): Change this define_expand to not rely on
22171 AltiVec predicate framework.
22172 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22174 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
22175 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
22176 define_insn pattern.
22177 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
22178 define_insn pattern because the xvcmpne<VSs>. instruction is not
22180 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
22181 instruction is not supported.
22183 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22185 * config/nvptx/nvptx.c: Include intl.h.
22187 2017-03-01 Martin Jambor <mjambor@suse.cz>
22190 * ipa-prop.h (ipa_bits): Removed field known.
22191 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
22192 to pointers. Adjusted their comments to warn about their sharing.
22193 (ipcp_transformation_summary): Change bits to a vector of pointers.
22194 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
22195 (ipa_get_ipa_bits_for_value): Declare.
22196 * tree-vrp.h (value_range): Mark as GTY((for_user)).
22197 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
22198 (ipa_bits_hash_table): Likewise.
22199 (ipa_vr_ggc_hash_traits): Likewise.
22200 (ipa_vr_hash_table): Likewise.
22201 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
22202 being pointers and vr_known being removed.
22203 (ipa_set_jf_unknown): Likewise.
22204 (ipa_get_ipa_bits_for_value): New function.
22205 (ipa_set_jfunc_bits): Likewise.
22206 (ipa_get_value_range): New overloaded functions.
22207 (ipa_set_jfunc_vr): Likewise.
22208 (ipa_compute_jump_functions_for_edge): Use the above functions to
22209 construct bits and vr parts of jump functions.
22210 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
22211 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22213 (ipcp_grow_transformations_if_necessary): Also allocate
22214 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22216 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
22217 them. Fix too long lines.
22218 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
22219 vr_known being removed.
22220 (ipa_read_jump_function): Use new setter functions to construct bits
22221 and vr parts of jump functions or set them to NULL.
22222 (write_ipcp_transformation_info): Adjust for bits being pointers.
22223 (read_ipcp_transformation_info): Likewise.
22224 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
22226 Include gt-ipa-prop.h.
22227 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
22229 (ipcp_store_bits_results): Likewise.
22230 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
22231 Do not write to existing jump functions but use a temporary instead.
22233 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22236 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
22237 attempt to use its first operand as BIT_FIELD_REF base.
22239 2017-03-01 Richard Biener <rguenther@suse.de>
22241 PR middle-end/79721
22242 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
22243 interpolating formula in wrapping arithmetic.
22244 (chrec_apply): Convert chrec_evaluate return value to wanted type.
22246 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22248 PR tree-optimization/79734
22249 * tree-vect-generic.c (expand_vector_condition): Optimize
22250 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
22251 Handle VEC_COND_EXPR where comparison has different inner width from
22252 type's inner width.
22254 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
22256 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
22257 markup, and similar issues. Remove @opindex entries for things
22258 that aren't options. Add missing -mmpy-option entries.
22260 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22262 PR tree-optimization/79737
22263 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
22264 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
22265 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
22266 instead of byte_size. Formatting fix.
22267 (shift_bytes_in_array_right): Formatting fix.
22269 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
22272 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
22273 condition on optimize for the leaf function test.
22275 2017-02-28 Martin Liska <mliska@suse.cz>
22278 * read-rtl-function.c (function_reader::handle_unknown_directive):
22279 Bail out when one uses -flto.
22281 2017-02-28 Martin Liska <mliska@suse.cz>
22283 * common.opt: Replace space with tabular for options of <number>
22285 * config/i386/i386.opt: Show <number> value for
22286 -mlarge-data-threshold.
22287 * opts.c (print_filtered_help): Do not display number in hexadecimal
22290 2017-02-28 Martin Liska <mliska@suse.cz>
22292 * common.opt: Fix --help=option -Q for options which are of
22295 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
22297 * config/i386/i386.c (print_reg): Error out for values
22298 of 8-bit size in invalid integer register.
22300 2017-02-28 Martin Sebor <msebor@redhat.com>
22302 PR tree-optimization/79691
22303 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
22305 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22308 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
22309 gcc_unreachable with output_operand_lossage.
22311 2017-02-28 Richard Biener <rguenther@suse.de>
22313 PR tree-optimization/79740
22314 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
22316 (visit_nary_op): Insert the nary into the hashtable if we
22317 pattern-matched sth.
22318 * tree-ssa-pre.c (eliminate_insert): Robustify.
22320 2017-02-28 Richard Biener <rguenther@suse.de>
22322 PR middle-end/79731
22323 * fold-const.c (decode_field_reference): Reject out-of-bound
22326 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22328 * config/i386/i386.c: Include intl.h.
22329 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
22330 instead of just cond ? "..." : "...".
22331 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
22332 * coverage.c (read_counts_file): Likewise.
22333 * omp-offload.c: Include intl.h.
22334 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
22335 of just cond ? "..." : "...".
22336 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
22337 of just cond ? "..." : "...".
22339 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
22342 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
22344 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
22345 'tune for' CPU name.
22346 * config/arm/arm-cpu-data.h: Regenerated.
22348 2017-02-28 Richard Biener <rguenther@suse.de>
22350 PR tree-optimization/79732
22351 * tree-inline.c (expand_call_inline): Do not shadow var.
22353 2017-02-28 Richard Biener <rguenther@suse.de>
22355 PR tree-optimization/79723
22356 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
22357 address-space properly.
22359 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
22361 * doc/optinfo.texi (Optimization groups): Fix option used for
22363 * doc/invoke.texi (-fopt-info): Document "omp".
22364 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
22365 (OPTGROUP_ALL): Add OPTGROUP_OMP.
22366 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
22367 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
22368 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
22370 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
22372 * dumpfile.c (optgroup_options): Instead of "openmp", associate
22373 OPTGROUP_OMP with "omp".
22375 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
22378 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
22379 for arithmetic shift of unsigned V2DI.
22381 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
22383 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
22384 arc/linux.h headers.
22385 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
22386 (LINK_SPEC): Likewise.
22387 (ARC_TLS_EXTRA_START_SPEC): Likewise.
22388 (EXTRA_SPECS): Likewise.
22389 (STARTFILE_SPEC): Likewise.
22390 (ENDFILE_SPEC): Likewise.
22391 (LIB_SPEC): Likewise.
22392 (TARGET_SDATA_DEFAULT): Likewise.
22393 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
22394 (MULTILIB_DEFAULTS): Likewise.
22395 (DWARF2_UNWIND_INFO): Likewise.
22396 * config/arc/big.h: New file.
22397 * config/arc/elf.h: Likewise.
22398 * config/arc/linux.h: Likewise.
22399 * config/arc/t-uClibc: Remove.
22401 2017-02-27 Bin Cheng <bin.cheng@arm.com>
22403 PR tree-optimization/77536
22404 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
22405 (tree_transform_and_unroll_loop): Use above function to compute the
22406 estimated niter of unrolled loop and use it when scaling profile.
22407 Also use count info rather than frequency if it's non-zero.
22408 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
22409 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
22410 (vect_transform_loop): Call above function.
22412 2017-02-27 Richard Biener <rguenther@suse.de>
22414 PR tree-optimization/45397
22415 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
22416 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
22417 (visit_nary_op): Add pattern matching for CSEing sign-changed
22418 or truncated operations with wider ones.
22420 2017-02-27 Richard Biener <rguenther@suse.de>
22422 PR tree-optimization/79690
22423 * tree-vect-stmts.c (vectorizable_store): Use vector type
22424 built from the DR with address-space.
22426 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22428 * doc/invoke.texi (Optimize Options): Refine the description
22429 of asan-use-after-return.
22431 2017-02-25 Alan Modra <amodra@gmail.com>
22433 PR rtl-optimization/79584
22434 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22435 base, not ad->base_term, the reg within base. Remove assertion
22436 that ad->base == ad->base_term. Replace gen_int_mode using
22437 bogus mode with const0_rtx.
22439 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22441 PR middle-end/79396
22442 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22443 FMA_EXPR like tcc_binary or tcc_unary.
22445 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22448 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22450 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22451 (output_loc_operands): Handle DW_OP_call_ref and
22452 DW_OP_GNU_variable_value.
22453 (struct variable_value_struct): New type.
22454 (struct variable_value_hasher): Likewise.
22455 (variable_value_hash): New variable.
22456 (string_types): Remove.
22457 (copy_loc_descr): New function.
22458 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22459 (prepend_loc_descr_to_each): New function.
22460 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22461 instead of add_loc_descr_to_each if the first argument is single
22462 location list and the second has multiple.
22463 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22464 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22465 when looking for variable value which doesn't have other location info.
22466 (loc_list_from_tree): Formatting fix.
22467 (gen_array_type_die): Simplify DW_AT_string_length handling.
22468 (adjust_string_types): Remove.
22469 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22470 string_types. Call resolve_variable_values.
22471 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22472 (resolve_addr_in_expr): Likewise. Add A argument.
22473 (copy_deref_exprloc): Remove deref argument. Adjust for the
22474 original expression being DW_OP_GNU_variable_value with optionally
22475 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22476 optionally after it.
22477 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22478 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22479 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22480 (variable_value_hasher::hash, variable_value_hasher::equal): New
22482 (resolve_variable_value_in_expr, resolve_variable_value,
22483 resolve_variable_values, note_variable_value_in_expr,
22484 note_variable_value): New functions.
22485 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22488 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22491 * opts.h (handle_generated_option): Add GENERATED_P argument.
22492 * opts-common.c (handle_option): Adjust function comment.
22493 (handle_generated_option): Add GENERATED_P argument, pass it to
22495 (control_warning_option): Pass false to handle_generated_option
22497 * opts.c (maybe_default_option): Pass true to handle_generated_option
22499 * optc-gen.awk: Likewise.
22501 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22503 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22504 a REG, look at the REG it is a SUBREG of.
22505 (splitter for cmpeqsi_t): Ditto.
22507 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22509 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22510 the special USEs with the pattern of the insn, not the insn itself.
22512 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22515 * doc/invoke.texi: Document -mload-store-pairs.
22517 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22518 Sandra Loosemore <sandra@codesourcery.com>
22520 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22521 argument isn't a CONST_INT.
22522 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22523 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22524 (nios2_validate_compare): Bypass alternate compare logic if *op2
22525 is not a CONST_INT.
22526 (ldstwm_operation_p): Return false if first_base is not a REG or
22527 if first_offset is not a CONST_INT.
22529 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22531 * config/cris/cris.md: Use correct operand in a define_peephole2.
22533 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22535 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22537 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22539 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22540 this_insn if it is an INSN or JUMP_INSN.
22541 (force_offsettable): Look at base, not at addr.
22542 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22543 on things that aren't necessarily CONST_INTs.
22545 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22547 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22548 -mfpmath=sse is the default also for x86-32 targets with SSE2
22549 instruction set when @option{-ffast-math} is enabled
22551 2017-02-24 Jeff Law <law@redhat.com>
22553 PR rtl-optimizatoin/79286
22554 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22557 2017-02-24 Richard Biener <rguenther@suse.de>
22559 PR tree-optimization/79389
22560 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22563 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22565 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22566 function comment to reflect reality.
22567 (loop_exits_before_overflow): Fix typo in function description.
22569 2017-02-24 Richard Biener <rguenther@suse.de>
22571 PR tree-optimization/79389
22572 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22573 properly that a threading opportunity exists. Detect conditional
22574 copy/constant propagation opportunities.
22576 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22578 * config/visium/visium.md (type): Add trap.
22579 (b): New mode attribute.
22580 (*btst): Rename into...
22581 (*btst<mode>): ...this and adjust.
22582 (*cbranchsi4_btst_insn): Rename into...
22583 (*cbranch<mode>4_btst_insn): ...this and adjust.
22584 (trap): New define_insn.
22586 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22588 PR tree-optimization/79389
22589 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22590 (noce_reversed_cond_code): New function.
22591 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22592 reversed_comparison_code. Formatting fix.
22593 (noce_try_store_flag): Test rev_cond != NULL in addition to
22594 reversed_comparison_code.
22595 (noce_try_store_flag_constants): Likewise.
22596 (noce_try_store_flag_mask): Likewise.
22597 (noce_try_addcc): Use rev_cond if non-NULL instead of
22598 reversed_comparison_code.
22599 (noce_try_cmove_arith): Likewise. Formatting fixes.
22600 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22601 (noce_find_if_block): Initialize rev_cond.
22602 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22603 instead of false as last argument never attempt to reverse it
22606 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22608 PR tree-optimization/79663
22609 * tree-predcom.c (combine_chains): Process refs in reverse order
22610 only for ZERO length chains, and add explaining comment.
22612 2017-02-23 Jeff Law <law@redhat.com>
22614 PR tree-optimization/79578
22615 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22616 in call to operand_equal_p.
22618 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22621 * config/i386/cpuid.h: Fix another undefined behavior.
22623 2017-02-23 Richard Biener <rguenther@suse.de>
22625 PR tree-optimization/79683
22626 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22627 vector types for data-refs.
22629 2017-02-23 Martin Liska <mliska@suse.cz>
22631 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22633 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22635 PR middle-end/79665
22636 * internal-fn.c (get_range_pos_neg): Moved to ...
22637 * tree.c (get_range_pos_neg): ... here. No longer static.
22638 * tree.h (get_range_pos_neg): New prototype.
22639 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22640 are known to be in between 0 and signed maximum inclusive, try to
22641 expand both unsigned and signed divmod and use the cheaper one from
22644 2017-02-22 Jeff Law <law@redhat.com>
22646 PR tree-optimization/79578
22647 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22648 to compare base operands.
22650 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22653 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22654 gpc_reg_operand instead of fpr_reg_operand.
22656 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22658 * config/mips/mips.c (mips_return_in_memory): Force FP
22659 vector types to be returned in memory for o32 ABI.
22661 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22663 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22664 instead of DW_TAG_member for static data member declarations and don't
22665 set no_linkage_name for static inline data members.
22666 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
22669 2017-02-22 Martin Liska <mliska@suse.cz>
22671 * doc/invoke.texi: Replace inequality signs with square brackets
22674 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22676 PR tree-optimization/68644
22677 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
22679 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
22682 * lra-constraints.c (simplify_operand_subreg): Handle
22683 WORD_REGISTER_OPERATIONS targets.
22685 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22688 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22689 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22690 elimination by swapping fld*.
22692 2017-02-22 Richard Biener <rguenther@suse.de>
22694 PR tree-optimization/79673
22695 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22696 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22697 irrelevant address-space qualifiers and avoiding a
22698 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22700 2017-02-22 Richard Biener <rguenther@suse.de>
22702 PR tree-optimization/79666
22703 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22704 to not symbolically negate if that may introduce undefined
22707 2017-02-22 Martin Liska <mliska@suse.cz>
22710 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22711 * data-streamer-out.c (streamer_write_gcov_count_stream):
22713 * value-prof.c (stream_out_histogram_value): Make assert more
22714 precise based on type of counter.
22716 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22719 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22720 Use nonimmediate_operand instead of memory_operand for operand 1.
22721 (float-extend standard_x87sse_constant_load splitter): Ditto.
22723 2017-02-21 Jeff Law <law@redhat.com>
22725 PR tree-optimization/79621
22726 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22727 blocks with edges to themselves.
22729 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22732 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22733 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22734 Use gimple_call_builtin_p.
22737 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22738 on temporarily removed DEBUG_INSNs.
22740 PR tree-optimization/79649
22741 * tree-loop-distribution.c (classify_partition): Give up on
22742 non-generic address space loads/stores.
22744 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22746 * doc/loop.texi (Loop manipulation): Remove nonexistent
22747 tree_ssa_loop_version from the documentation.
22748 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22750 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22753 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22754 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22755 * config/rs6000/rs6000.c: Include except.h.
22756 (rs6000_expand_split_stack_prologue): Call
22757 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22759 2017-02-21 Martin Jambor <mjambor@suse.cz>
22762 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22763 have been analyzed.
22765 2017-02-21 Martin Jambor <mjambor@suse.cz>
22767 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22768 for backward compatibility only.
22769 * doc/invoke.texi (Option Summary): Remove all references to
22770 -fipa-cp-alignment.
22772 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
22776 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22778 * lra-constraints.c (curr_insn_transform): Handle
22779 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22781 2017-02-21 Martin Liska <mliska@suse.cz>
22783 * config/i386/i386.opt: Replace -masm-dialect with -masm.
22785 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
22787 PR translation/79638
22788 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
22790 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
22793 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
22794 (arm_function_ok_for_sibcall): Return false for an indirect call by
22795 descriptor if all the argument registers are used.
22796 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
22797 alignment of the function.
22799 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22801 PR tree-optimization/61441
22802 * simplify-rtx.c (simplify_const_unary_operation): For
22803 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
22804 the sNaN unmodified.
22806 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22808 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22809 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22810 instead of SYSTEM_HEADER_DIR.
22812 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
22813 Martin Liška <mliska@suse.cz>
22815 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
22816 Fix typos and grammar, use active voice, and clarify.
22818 2017-02-20 Marek Polacek <polacek@redhat.com>
22820 PR middle-end/79537
22821 * gimplify.c (gimplify_expr): Handle unused *&&L;.
22824 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
22826 2017-02-20 Jakub Jelinek <jakub@redhat.com>
22829 * config/i386/i386.c (ix86_expand_builtin): Handle
22830 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
22831 ix86_builtins_isa[fcode].isa as a requirement of those
22832 flags and any other flag in the bitmask.
22833 (ix86_init_mmx_sse_builtins): Use 0 instead of
22834 ~OPTION_MASK_ISA_64BIT as mask.
22835 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
22836 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
22837 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
22838 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
22840 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22843 * lra-constraints.c (split_reg): Check requested split mode
22844 is supported by the register.
22846 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22848 * lra-constraints.c (simplify_operand_subreg): Remove early
22851 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22854 * lra-constraints.c (curr_insn_transform): Tighten condition
22855 for converting SUBREG reloads from OP_OUT to OP_INOUT.
22857 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22860 * lra-constraints.c (curr_insn_transform): Handle
22861 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22863 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
22866 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
22868 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
22870 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
22873 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
22876 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22878 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
22881 2017-02-18 Jakub Jelinek <jakub@redhat.com>
22883 * final.c (last_columnnum, override_columnnum): New variables.
22884 (final_start_function): Set last_columnnum, pass it to begin_prologue
22885 hook and pass 0 to dwarf2out_begin_prologue.
22886 (final_scan_insn): Update override_columnnum. Pass last_columnnum
22887 to source_line debug hook.
22888 (notice_source_line): Compute last_columnnum and for debug_column_info
22889 return true on column changes.
22890 * debug.h (struct gcc_debug_hooks): Add column argument to
22891 source_line and begin_prologue hooks.
22892 (debug_nothing_int_charstar_int_bool): Remove prototype.
22893 (debug_nothing_int_int_charstar,
22894 debug_nothing_int_int_charstar_int_bool): New prototypes.
22895 (dwarf2out_begin_prologue): Add column argument.
22896 * debug.c (do_nothing_debug_hooks): Adjust source_line and
22897 begin_prologue hooks.
22898 (debug_nothing_int_charstar_int_bool): Remove.
22899 (debug_nothing_int_int_charstar,
22900 debug_nothing_int_int_charstar_int_bool): New functions.
22901 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
22902 through to dwarf2out_source_line.
22903 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
22904 (dwarf2out_source_line): Add column argument, emit it if requested.
22905 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
22907 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22908 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22909 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
22910 through to dwarf2out_begin_prologue.
22911 (vmsdbgout_source_line): Add column argument, pass it through to
22912 dwarf2out_source_line.
22913 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
22914 dbxout_source_line caller.
22915 (dbxout_source_line): Add column argument.
22917 * common.opt (gno-column-info, gcolumn-info): New options.
22918 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
22919 (check_die): Also test for multiple DW_AT_decl_column attributes.
22920 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
22921 DW_AT_decl_column if requested.
22922 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
22924 (gen_variable_die): Likewise.
22925 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
22926 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
22929 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
22930 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
22931 (ix86_handle_option): Handle OPT_m3dnowa.
22932 * doc/invoke.texi (-m3dnowa): Document.
22933 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
22934 -m3dnowa instead of -m3dnow -march=athlon.
22937 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
22938 instead of gcc_assert for K, r and R code checks. Formatting fixes.
22940 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22943 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
22944 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
22945 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
22946 generator for vsx_xxpermdi_<mode>_be.
22947 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
22948 force big-endian semantics.
22949 (vsx_xxpermdi_<mode>_be): New define_expand with same
22950 implementation as previous version of vsx_xxpermdi_<mode>.
22952 2017-02-17 Jakub Jelinek <jakub@redhat.com>
22954 PR tree-optimization/79327
22955 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
22956 variable, its initialization and use.
22958 2017-02-17 Julia Koval <julia.koval@intel.com>
22960 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
22961 (OPTION_MASK_ISA_PKU_UNSET): New.
22962 (ix86_handle_option): Handle -mrdpid.
22963 * config/i386/cpuid.h (bit_RDPID): New.
22964 * config/i386/driver-i386.c (host_detect_local_cpu):
22965 Detect RDPID feature.
22966 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
22967 * config/i386/i386-c.c (ix86_target_macros_internal):
22969 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
22970 (ix86_valid_target_attribute_inner_p): Add "rdpid".
22971 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22972 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
22973 * config/i386/i386.md (define_insn "rdpid"): New.
22974 * config/i386/i386.opt Add -mrdpid.
22975 * config/i386/immintrin.h (_rdpid_u32): New.
22977 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
22979 PR rtl-optimization/79541
22980 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
22981 instead of transforming it into USE.
22983 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
22985 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
22986 If HONOR_SNANS (SFmode) force the input to a register.
22987 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
22988 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
22989 an frsp or similar insn.
22991 2017-02-17 Martin Liska <mliska@suse.cz>
22993 PR rtl-optimization/79577
22994 * params.def (selsched-max-sched-times): Increase minimum to 1.
22996 2017-02-17 Martin Liska <mliska@suse.cz>
22998 PR rtl-optimization/79574
22999 * gcse.c (want_to_gcse_p): Prevent integer overflow.
23001 2017-02-17 Martin Liska <mliska@suse.cz>
23003 PR tree-optimization/79529
23004 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
23005 ssa_defined_default_def_p to handle cases which are implicitly
23007 * tree-ssa.c (ssa_defined_default_def_p): New function.
23008 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
23009 which are implicitly defined.
23010 * tree-ssa.h (ssa_defined_default_def_p): Declare.
23012 2017-02-17 Richard Biener <rguenther@suse.de>
23014 PR middle-end/79576
23015 * params.def (max-ssa-name-query-depth): Limit to 10.
23017 2017-02-17 Richard Biener <rguenther@suse.de>
23019 PR tree-optimization/79552
23020 * tree-ssa-structalias.c (visit_loadstore): Properly verify
23023 2017-02-17 Richard Biener <rguenther@suse.de>
23026 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
23028 2017-02-17 Marek Polacek <polacek@redhat.com>
23030 PR middle-end/79536
23031 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
23032 (fold_negate_expr): New wrapper.
23034 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
23036 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
23037 Correct terminology and de-emphasize pre-standard behavior.
23039 2017-02-16 Alan Modra <amodra@gmail.com>
23041 PR rtl-optimization/79286
23042 * ira.c (def_dominates_uses): New function.
23043 (update_equiv_regs): Don't create an equivalence for insns that
23044 may trap where the register def does not dominate the use.
23046 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
23048 PR rtl-optimization/78127
23049 * lra.c (lra): Call lra_eliminate before finish the loop after
23052 2017-02-16 Richard Biener <rguenther@suse.de>
23054 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
23056 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
23057 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
23058 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
23059 (isl_val_int_from_wi): New function.
23060 (extract_affine_gmp): Rename to ...
23061 (extract_affine_wi): ... this, take a widest_int.
23062 (extract_affine_int): Just wrap extract_affine_wi.
23063 (add_param_constraints): Use isl_val_int_from_wi.
23064 (add_loop_constraints): Likewise, and extract_affine_wi.
23066 2017-02-15 Jeff Law <law@redhat.com>
23068 PR middle-end/79521
23069 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
23070 ira_init_register_move_cost_if_necessary.
23072 2017-02-15 Martin Sebor <msebor@redhat.com>
23074 PR middle-end/32003
23075 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
23076 removed in a prior commit.
23078 2017-02-15 Bin Cheng <bin.cheng@arm.com>
23080 PR tree-optimization/79347
23081 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
23082 counters during peeling.
23084 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
23086 * Makefile.in (site.exp): Remove "set ISLVER".
23088 2017-02-15 Jakub Jelinek <jakub@redhat.com>
23091 * real.c (real_from_integer): Call real_convert even for decimal.
23093 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
23096 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
23098 2017-02-14 Andrew Pinski <apinski@cavium.com>
23100 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
23101 cores and change the partno/implementer to be correct.
23102 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
23103 the 'B" as the implementer.
23104 * config/aarch64/aarch64-tune.md: Regenerate.
23106 2017-02-14 Carl Love <cel@us.ibm.com>
23108 * config/rs6000/rs6000.c: Add case statement entry to make the
23109 xvcvuxdsp built-in argument unsigned.
23110 * config/rs6000/vsx.md: Fix the source and return operand types so they
23111 match the instruction definitions from the ISA document. Fix typo
23112 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
23115 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
23118 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
23119 member early_clobber_alts.
23120 * lra-lives.c (reg_early_clobber_p): New.
23121 (process_bb_lives): Use it.
23122 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
23123 (debug_operand_data): Initialize early_clobber_alts.
23124 (setup_operand_alternative): Set up early_clobber_alts.
23125 (collect_non_operand_hard_regs): Ditto. Pass early clobber
23126 alternatives to new_insn_reg.
23127 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
23129 (lra_update_insn_regno_info): Pass the new arg.
23131 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23133 PR middle-end/79505
23134 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
23135 (new_oacc_loop_raw): Don't clear already cleared fields.
23138 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
23139 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
23140 _mm512_prefetch_i64gather_ps): New inline functions and macros.
23142 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
23145 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
23147 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
23150 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
23151 the extra instruction to the right place to store 128-bit constant
23154 2017-02-14 Martin Sebor <msebor@redhat.com>
23156 PR middle-end/79448
23157 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
23158 warning for strings of unknown length.
23160 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23162 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
23164 2017-02-14 Jeff Law <law@redhat.com>
23167 * ira-costs.c (scan_one_insn): Initialize register move costs
23168 for pseudos seen in USE/CLOBBER insns.
23170 PR tree-optimization/79095
23171 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
23172 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
23173 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
23174 if the operands are known to be not equal, then the resulting range
23176 (intersect_ranges): If the new range is ~[0,0] and the old range is
23177 wide, then prefer ~[0,0].
23178 * tree-vrp.c (overflow_comparison_p_1): New function.
23179 (overflow_comparison_p): New function.
23180 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
23181 if NAME is used in an overflow test.
23182 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
23183 overflow check that can be expressed as an equality test, then adjust
23184 ops to be that equality test.
23186 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23188 * config/s390/s390-builtin-types.def: Remove flags argument.
23189 * config/s390/s390.c (s390_init_builtins): Likewise.
23191 2017-02-14 Martin Liska <mliska@suse.cz>
23193 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
23194 vector. Fix trailing white spaces.
23196 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
23198 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
23201 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23203 PR rtl-optimization/68664
23204 * config/arm/arm.c (arm_sched_can_speculate_insn):
23205 New function. Declare prototype.
23206 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23208 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23210 PR rtl-optimization/68664
23211 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
23213 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23215 2017-02-14 Amit Pawar <amit.pawar@amd.com>
23217 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
23218 max skip bytes for function, loop and jump.
23220 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23222 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
23223 ABS_EXPR for gimple dump.
23225 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23228 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
23230 PR tree-optimization/79408
23231 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
23232 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
23233 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
23234 also if rhs1 is INTEGER_CST.
23236 2017-02-14 Richard Biener <rguenther@suse.de>
23238 PR middle-end/79432
23239 * tree-into-ssa.c (insert_phi_nodes): When the function can
23240 have abnormal edges rewrite SSA names with broken use-def
23241 dominance out of SSA and register them for PHI insertion.
23243 2017-02-13 Martin Sebor <msebor@redhat.com>
23245 PR middle-end/79496
23246 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
23247 clearing info.nowrite flag when snprintf size argument is a range.
23249 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23251 * cprop.c (cprop_jump): Add missing space in string literal.
23252 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
23253 (get_constraint_for_component_ref): Likewise.
23254 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
23255 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
23256 * lra-constraints.c (process_alt_operands): Likewise.
23257 * ipa-inline.c (inline_small_functions): Likewise.
23258 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
23259 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23260 * trans-mem.c (diagnose_tm_1_op): Likewise.
23261 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
23262 (grid_parallel_clauses_gridifiable): Likewise.
23264 * config/nvptx/mkoffload.c (process): Add space in between
23267 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
23268 "MOD4_SSE_REGS" and "ALL_REGS".
23270 * spellcheck.c (test_data): Add , in between "foo" and "food".
23272 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23275 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
23276 boundary crossing check and subsequent code generation agree.
23278 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23280 * config/aarch64/aarch64.c (has_memory_op): Delete.
23281 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
23284 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23286 PR rtl-optimization/79388
23287 PR rtl-optimization/79450
23288 * combine.c (distribute_notes): When removing TEM_INSN for which
23289 corresponding dest has last value recorded, invalidate that last
23292 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23294 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
23295 of explicit '@'. Add missing assembly comment marker on branch costs
23298 2017-02-13 Nathan Sidwell <nathan@acm.org>
23300 * gengtype-lex.l (<in_struct>): Add '/'.
23302 2017-02-13 Martin Liska <mliska@suse.cz>
23305 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
23307 2017-02-13 Richard Biener <rguenther@suse.de>
23309 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
23311 * configure: Re-generate.
23312 * config.in: Likewise.
23313 * graphite-dependences.c: Simplify as if
23314 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
23315 * graphite-isl-ast-to-gimple.c: Likewise.
23316 * graphite-optimize-isl.c: Likewise.
23317 * graphite-poly.c: Likewise.
23318 * graphite-sese-to-poly.c: Likewise.
23319 * graphite.h: Likewise.
23320 * toplev.c: Include isl/version.h and use isl_version () for
23321 printing the ISL version.
23322 * doc/install.texi: Update ISL requirement.
23324 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23326 * doc/standards.texi (Standards): Update reference to
23329 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23331 * doc/extend.texi (Named Address Spaces): sourceware.org now
23333 * doc/install.texi (Binaries): Ditto.
23336 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23338 * doc/cpp.texi: Replace "stringify"/"stringification" with C
23339 standard terminology "stringize"/"stringizing" throughout.
23340 * doc/cppinternals.texi: Likewise.
23342 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23344 * doc/extend.texi: Fix some spelling mistakes and typos.
23345 * doc/invoke.texi: Likewise.
23347 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23350 * params.def (inline-min-speedup) Change from 10 to 8.
23352 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23354 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
23357 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23360 * ipa-inline-analysis.c (get_minimal_bb): New function.
23361 (record_modified): Use it.
23362 (remap_edge_change_prob): Handle also ancestor functions.
23364 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
23366 * doc/contrib.texi (Contributors): Remove broken link into
23367 the Mauve CVS repository.
23369 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23371 PR middle-end/79454
23372 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
23373 result computation whenever lhs doesn't have vector mode, not
23374 just when it has BLKmode.
23376 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23378 * doc/makefile.texi (profiledbootstrap): Refer to the
23379 installation instructions only in textual form.
23381 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23384 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
23386 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23388 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
23389 (Specific): Update mingw-w64 reference.
23391 (Specific): Remove broken link to Renesas RX processor.
23393 2017-02-10 Richard Biener <rguenther@suse.de>
23395 * toplev.c (process_options): Do not mention obsolete graphite
23396 options when printing sorry message about missing graphite support.
23397 Mention -floop-nest-optimize.
23399 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
23401 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
23402 (vtst_p16): Likewise.
23403 (vtstq_p8): Likewise.
23404 (vtstq_p16): Likewise.
23406 (vtstq_p64): Likewise.
23407 * config/arm/arm_neon.h (vgetq_lane_p64): New.
23408 (vset_lane_p64): New.
23409 (vsetq_lane_p64): New.
23411 2017-02-10 Jakub Jelinek <jakub@redhat.com>
23413 PR tree-optimization/79411
23414 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
23415 stmt operands are SSA_NAMEs used in abnormal phis.
23416 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
23419 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23422 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
23425 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23427 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23429 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23431 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23435 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23436 not arbitrary TREE_CONSTANT.
23439 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23440 "omp declare target link" attribute unless is_global_var.
23441 * omp-offload.c (find_link_var_op): Likewise.
23443 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23444 Chung-Lin Tang <cltang@codesourcery.com>
23446 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23448 (gimplify_adjust_omp_clauses): Don't delete TILE.
23449 (gimplify_omp_for): Deal with TILE.
23450 * internal-fn.c (expand_GOACC_TILE): New function.
23451 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23453 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23454 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23456 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23457 avoid DIV for outermost collapse var.
23458 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23459 Remove out of date comments, fix whitespace.
23460 * omp-general.c (omp_extract_for_data): Deal with tiling.
23461 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23462 adjust OLF_DIM_BASE value.
23463 (struct omp_for_data): Add tiling field.
23464 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23465 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23466 for auto loops. Remove default auto determining, moved to
23467 oacc_loop_fixed_partitions.
23468 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23469 stmts, add e_mask field.
23470 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23471 (oacc_thread_numbers): Use oacc_dim_call.
23472 (oacc_xform_tile): New.
23473 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23474 (finish_oacc_loop): Adjust for ifns vector.
23475 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23476 add case for GOACC_TILE fns.
23477 (oacc_loop_xform_loop): Delete.
23478 (oacc_loop_process): Iterate over call list directly, and add
23479 handling for GOACC_TILE fns.
23480 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23482 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23483 vector partitioning to outer loops. Assign 2 partitions to loops
23484 when available. Add TILE handling.
23485 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23486 (execite_oacc_device_lower): Process GOACC_TILE fns,
23487 ignore unknown specs.
23488 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23489 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23490 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23492 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23494 * configure.ac (ACX_BUGURL): Update.
23495 * configure: Regenerate.
23497 2017-02-09 Richard Biener <rguenther@suse.de>
23499 PR tree-optimization/69823
23500 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23501 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23503 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23505 * config/arc/arc-c.def: Add __NPS400__ definition.
23506 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23507 (TARGET_NPS400): Define.
23509 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23511 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23513 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23514 pointer, arch_info.
23515 (arc_cpu_types): Fill the arch_info field with a pointer into the
23516 arc_arch_types table.
23517 (arc_selected_cpu): Declare.
23518 * config/arc/arc.c (arc_selected_cpu): Make global.
23519 (arc_selected_arch): Delete.
23520 (arc_base_cpu): Delete.
23521 (arc_override_options): Remove references to deleted variables,
23522 update access to arch information.
23523 (ARC_OPT): Update access to arch information.
23524 (ARC_OPTX): Likewise.
23525 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23526 (TARGET_ARC600): Update access to arch information.
23527 (TARGET_ARC601): Likewise.
23528 (TARGET_ARC700): Likewise.
23529 (TARGET_EM): Likewise.
23530 (TARGET_HS): Likewise.
23531 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23534 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23537 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23538 condition/operands for integer GE/LE/GEU/LEU operations.
23540 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23542 PR translation/79397
23543 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23546 2017-02-08 Martin Jambor <mjambor@suse.cz>
23549 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23550 whether allocation happened.
23551 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23552 nothing was allocated.
23554 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23556 PR tree-optimization/79408
23557 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23558 constant, but SSA_NAME with a known integer range, use the minimum
23559 of that range instead of op1 to determine if modulo can be replaced
23560 with its first operand.
23562 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23564 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23566 2017-02-08 Richard Biener <rguenther@suse.de>
23568 PR tree-optimization/71824
23569 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23570 Check all loops contained in the merged region.
23572 2017-02-07 Andrew Pinski <apinski@cavium.com>
23574 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23576 2017-02-07 Andrew Pinski <apinski@cavium.com>
23578 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23579 (thunderxt88): Likewise.
23580 (thunderxt81): Disable LSE and change v8.1 to v8.
23581 (thunderxt83): Likewise.
23583 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23584 Richard Biener <rguenther@suse.de>
23586 PR middle-end/79399
23587 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23588 type from int to size_t.
23589 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23591 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23593 PR rtl-optimization/79386
23594 * cprop.c (bypass_conditional_jumps): Initialize
23595 bypass_last_basic_block already before splitting bbs after
23596 unconditional traps...
23597 (bypass_conditional_jumps): ... rather than here.
23600 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23601 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23602 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23603 fix -masm=intel patterns.
23605 2017-02-07 Richard Biener <rguenther@suse.de>
23607 PR tree-optimization/79256
23608 PR middle-end/79278
23609 * builtins.c (get_object_alignment_2): Use min_align_of_type
23610 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23611 and ADJUST_FIELD_ALIGN.
23613 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23615 * doc/tm.texi: Regenerate.
23616 * stor-layout.c (layout_decl): Adjust.
23617 (update_alignment_for_field): Likewise.
23618 (place_field): Likewise.
23619 (min_align_of_type): Likewise.
23620 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23621 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23622 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23623 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23624 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23625 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23626 * config/i386/i386.c (x86_field_alignment): Likewise.
23627 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23628 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23629 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23630 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23631 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23632 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23636 2017-01-30 Richard Biener <rguenther@suse.de>
23638 PR tree-optimization/79256
23639 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23640 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23643 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23645 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23646 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23647 builtins to SImode and emit a zero-extend, if necessary.
23649 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23651 * docs/invoke.texi (RISC-V Options): Alphabetize.
23653 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23655 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23658 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23660 * config/riscv/riscv.c: New file.
23661 * common/config/riscv/riscv-common.c: Likewise.
23662 * config.gcc: Likewise.
23663 * config/riscv/constraints.md: Likewise.
23664 * config/riscv/elf.h: Likewise.
23665 * config/riscv/generic.md: Likewise.
23666 * config/riscv/linux.h: Likewise.
23667 * config/riscv/multilib-generator: Likewise.
23668 * config/riscv/peephole.md: Likewise.
23669 * config/riscv/pic.md: Likewise.
23670 * config/riscv/predicates.md: Likewise.
23671 * config/riscv/riscv-builtins.c: Likewise.
23672 * config/riscv/riscv-c.c: Likewise.
23673 * config/riscv/riscv-ftypes.def: Likewise.
23674 * config/riscv/riscv-modes.def: Likewise.
23675 * config/riscv/riscv-opts.h: Likewise.
23676 * config/riscv/riscv-protos.h: Likewise.
23677 * config/riscv/riscv.h: Likewise.
23678 * config/riscv/riscv.md: Likewise.
23679 * config/riscv/riscv.opt: Likewise.
23680 * config/riscv/sync.md: Likewise.
23681 * config/riscv/t-elf-multilib: Likewise.
23682 * config/riscv/t-linux: Likewise.
23683 * config/riscv/t-linux-multilib: Likewise.
23684 * config/riscv/t-riscv: Likewise.
23685 * configure.ac: Likewise.
23686 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23687 Waterman as RISC-V maintainers.
23688 * doc/install.texi: Add RISC-V entries.
23689 * doc/invoke.texi: Add RISC-V options section.
23690 * doc/md.texi: Add RISC-V constraints section.
23691 * configure: Regenerated.
23693 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23696 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23697 false values to be constant vectors with all 0 or all 1 bits set.
23698 (vcondu<mode><mode>): Likewise.
23699 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23701 (fpmask_comparison_operator): Update comment.
23702 (vecint_comparison_operator): New predicate.
23703 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23704 vector conditionals when the true and false values are constant
23705 vectors with all 0 bits or all 1 bits set.
23707 2017-02-06 Martin Sebor <msebor@redhat.com>
23709 PR tree-optimization/79376
23710 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23712 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23714 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23715 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23716 to simplify split condition.
23718 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23720 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23721 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23724 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23726 PR rtl-optimization/68664
23727 * target.def (can_speculate_insn): New hook.
23728 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23729 * doc/tm.texi: Regenerate.
23730 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23731 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23732 (rs6000_sched_can_speculate_insn): New function.
23734 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23736 PR tree-optimization/79284
23737 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23738 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23739 vectorizable_mask_load_store, vectorizable_operation,
23740 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23741 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23742 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23743 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23744 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23745 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23746 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23747 is_gimple_assign (stmt). Replace another such test with
23748 is_gimple_assign (stmt).
23750 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23753 * config/avr/avr.c (rtl-iter.h): Include it.
23754 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23755 (avr_legitimate_combined_insn): ...and implementation.
23757 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23759 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23760 * config/s390/s390.c (s390_const_operand_ok)
23761 (s390_canonicalize_comparison, s390_extract_part)
23762 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23763 (s390_contiguous_bitmask_p, s390_rtx_costs)
23764 (legitimize_pic_address): Likewise.
23765 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23766 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23767 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23768 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23769 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
23771 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23773 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
23774 REGNO($0) == REGNO($1).
23776 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23778 * config/s390/linux.h(SIZE_TYPE): Add comment.
23780 2017-02-06 Julian Brown <julian@codesourcery.com>
23781 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23782 Virendra Pathak <virendra.pathak@broadcom.com>
23784 * config/aarch64/aarch64-cores.def: Change the scheduler
23786 * config/aarch64/aarch64.md: Include thunderx2t99.md.
23787 * config/aarch64/thunderx2t99.md: New file.
23789 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23791 * doc/standards.texi (Go Language): Update link to language
23794 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23796 * tree-eh.c (lower_resx): Sanitize profile.
23797 (cleanup_empty_eh_move_lp): Likewise.
23799 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23802 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
23804 * cfgloopmanip.h (loop_version): Update prototype.
23805 * modulo-sched.c (sms_schedule): Update call of loop_version.
23806 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
23807 * tree-parloops.c (gen_parallel_loop): Likewise.
23808 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
23809 * tree-ssa-loop-split.c (split_loop): Likewise.
23810 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
23811 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23813 2017-02-05 Martin Liska <mliska@suse.cz>
23816 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
23818 (print_operand_address): Initialize a struct to zero.
23820 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23822 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
23823 garbage collector only in textual form.
23825 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23827 * doc/extend.texi (x86 specific memory model extensions for
23828 transactional memory): Simplify a phrase.
23830 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
23833 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
23834 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
23835 (atomic_storedi_1): Likewise.
23837 2017-02-04 Jakub Jelinek <jakub@redhat.com>
23839 PR tree-optimization/79338
23840 * tree-parloops.c (gather_scalar_reductions): Don't call
23841 vect_analyze_loop_form for loop->inner before destroying loop's
23844 2017-02-03 Martin Sebor <msebor@redhat.com>
23846 PR tree-optimization/79327
23847 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
23848 when precision has resulted in leading zeros.
23849 (format_integer): Adjust the likely counter to assume an unknown
23850 argument that may be zero is non-zero.
23852 2017-02-03 Jason Merrill <jason@redhat.com>
23855 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
23856 avoid copying non-taken branch.
23858 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23860 PR tree-optimization/79340
23861 * tree-vect-loop.c (vectorizable_reduction): Release
23862 vec_defs elements after safe_splicing them into other vectors.
23865 PR tree-optimization/79327
23866 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
23867 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
23869 (format_integer): Use wide_int_to_tree instead of build_int_cst
23870 + to_?hwi. If argmin is NULL, just set argmin and argmax to
23871 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
23872 of shortest and longest sequence.
23874 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
23876 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
23877 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
23879 2017-02-03 Walter Lee <walt@tilera.com>
23882 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
23883 after initial stackframe link reg save.
23884 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
23886 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23889 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
23890 wu for stxssp alternative.
23892 2017-02-03 Martin Sebor <msebor@redhat.com>
23894 PR tree-optimization/79352
23895 * gimple-fold.c (get_range_strlen): Add argument.
23896 (get_range_strlen): Change return type to bool.
23897 (get_maxval_strlen): Pass in a dummy argument.
23898 * gimple-fold.h (get_range_strlen): Change return type to bool.
23899 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
23900 * tree.h (array_at_struct_end_p): Add argument.
23901 * tree.c (array_at_struct_end_p): Handle it.
23903 2017-02-03 Martin Liska <mliska@suse.cz>
23906 * multiple_target.c (create_dispatcher_calls): Redirect edge
23907 from a caller of a dispatcher.
23908 (expand_target_clones): Make the clones local.
23909 (ipa_target_clone): Do both target clones and resolvers.
23910 (ipa_dispatcher_calls): Remove the pass.
23911 (pass_dispatcher_calls::gate): Likewise.
23912 (make_pass_dispatcher_calls): Likewise.
23913 * passes.def (pass_target_clone): Put as very first IPA early
23916 2017-02-03 Martin Liska <mliska@suse.cz>
23918 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
23919 in case of a function with ifunc attribute.
23921 2017-02-03 Martin Liska <mliska@suse.cz>
23923 * cgraph.c (cgraph_node::dump): Dump function version info.
23924 * symtab.c (symtab_node::dump_base): Add missing new line.
23926 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23928 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
23929 (ifcombine_ifandif): Use it.
23931 2017-02-03 Martin Liska <mliska@suse.cz>
23933 * doc/invoke.texi: Document default value for
23934 use-after-scope-direct-emission-threshold.
23936 2017-02-03 Martin Liska <mliska@suse.cz>
23938 PR tree-optimization/79339
23939 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
23940 (format_floating): Likewise.
23942 2017-02-03 Martin Liska <mliska@suse.cz>
23945 * ipa-prop.c (ipa_node_params_t::insert): Remove current
23947 (ipa_node_params_t::remove): Likewise.
23948 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
23949 initialization from removed ipa_node_params_t::insert.
23950 (ipa_node_params::~ipa_node_params): Move from removed
23951 ipa_node_params_t::release.
23952 * symbol-summary.h (symbol_summary::m_released): New member.
23953 Do not release a summary twice. Do not allow to call finalizer
23954 for types of a summary that live in GGC memory.
23956 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23958 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
23961 2017-02-02 Martin Sebor <msebor@redhat.com>
23963 PR middle-end/79275
23964 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
23965 (format_string): Tighten up the range of output for non-constant
23966 strings and correct the expected range for wide non-constant strings.
23968 2017-02-02 Martin Sebor <msebor@redhat.com>
23970 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
23972 PR middle-end/32003
23973 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
23975 (-fdump-tree-@var): Add to index and document how to come up
23976 with pass-specific option and dump file names.
23977 (-fdump-passes): Clarify where to look for output.
23979 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23981 PR middle-end/77445
23982 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
23983 statistics of the analyzed path; allow threading for speed when
23984 any of BBs along the path are optimized for speed.
23986 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
23988 PR middle-end/78468
23989 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
23990 settings of the virtual registers.
23993 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
23995 * explow.c (get_dynamic_stack_size): Take known alignment of stack
23996 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
23999 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24001 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
24002 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
24004 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24006 * config/s390/s390.md: Add missing comments with the expanded
24008 * config/s390/vector.md: Likewise.
24009 * config/s390/vx-builtins.md: Likewise.
24011 2017-02-02 Jakub Jelinek <jakub@redhat.com>
24014 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
24015 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
24016 conditions on a single line.
24018 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24020 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24021 __S390_VX__ to __VX__.
24023 2017-02-01 Andrew Pinski <apinski@cavium.com>
24025 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
24026 stmt_info to record_stmt_cost.
24027 (vect_get_known_peeling_cost): Pass stmt_info if known to
24029 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
24030 cpu_vector_cost field into
24031 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
24032 field into vec_int_stmt_cost and vec_fp_stmt_cost.
24033 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
24034 splitting of scalar_stmt_cost and vec_stmt_cost.
24035 (thunderx_vector_cost): Likewise.
24036 (cortexa57_vector_cost): LIkewise.
24037 (exynosm1_vector_cost): Likewise.
24038 (xgene1_vector_cost): Likewise.
24039 (thunderx2t99_vector_cost): Improve after the splitting of the two
24041 (aarch64_builtin_vectorization_cost): Update for the splitting of
24042 scalar_stmt_cost and vec_stmt_cost.
24044 2017-02-01 Torvald Riegel <triegel@redhat.com>
24045 Richard Henderson <rth@redhat.com>
24047 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
24048 conditional on existance of a fast atomic load.
24049 * optabs-query.c (can_atomic_load_p): New function.
24050 * optabs-query.h (can_atomic_load_p): Declare it.
24051 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
24052 no fast atomic load is available for the particular size of access.
24053 (expand_atomic_compare_and_swap): Likewise.
24054 (expand_atomic_load): Likewise.
24055 (expand_atomic_store): Likewise.
24056 (expand_atomic_fetch_op): Likewise.
24057 * testsuite/lib/target-supports.exp
24058 (check_effective_target_sync_int_128): Remove x86 because it provides
24059 no fast atomic load.
24060 (check_effective_target_sync_int_128_runtime): Likewise.
24062 2017-02-01 Richard Biener <rguenther@suse.de>
24064 * graphite.c: Include tree-vectorizer.h for find_loop_location.
24065 (graphite_transform_loops): Provide opt-info for optimized nests.
24066 * tree-parloop.c (parallelize_loops): Provide opt-info for
24067 parallelized loops.
24069 2017-02-01 Richard Biener <rguenther@suse.de>
24071 PR middle-end/79315
24072 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
24073 was not set before.
24075 2017-02-01 Richard Biener <rguenther@suse.de>
24077 PR tree-optimization/71824
24078 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24079 Verify the loops are valid in the merged SESE region.
24080 (scop_detection::can_represent_loop_1): Check analyzing the
24081 evolution of the number of iterations in the region succeeds.
24083 2017-01-31 Ian Lance Taylor <iant@golang.org>
24085 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
24086 REG_ARGS_SIZE note to 32-bit push insns and call insn.
24088 2017-01-31 David Malcolm <dmalcolm@redhat.com>
24090 PR preprocessor/79210
24091 * input.c (get_substring_ranges_for_loc): Replace line_width
24092 assertion with error-handling.
24094 2017-01-31 Richard Biener <rguenther@suse.de>
24096 PR tree-optimization/77318
24097 * graphite-sese-to-poly.c (extract_affine): Fix assert.
24098 (create_pw_aff_from_tree): Take loop parameter.
24099 (add_condition_to_pbb): Pass loop of the condition to
24100 create_pw_aff_from_tree.
24102 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24104 * config/s390/s390.c (s390_asan_shadow_offset): New function.
24105 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
24107 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
24111 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
24113 (convert_int_to_float128): Likewise.
24114 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
24115 (convert_int_to_float128): Likewise.
24116 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
24117 (UNSPEC_IEEE128_CONVERT): Likewise.
24118 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
24119 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
24120 Use local variables for IBM extended format.
24121 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
24122 (fix_trunc<mode>si2_fprs): Likewise.
24123 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
24124 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
24125 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
24126 to know that we can now have integers of all sizes in vector
24128 (fix<uns>_<mode>di2_hw): Likewise.
24129 (float<uns>_<mode>si2_hw): Likewise.
24130 (fix_<mode>si2_hw): Likewise.
24131 (fixuns_<mode>si2_hw): Likewise.
24132 (float<uns>_<mode>di2_hw): Likewise.
24133 (float_<mode>di2_hw): Likewise.
24134 (float_<mode>si2_hw): Likewise.
24135 (floatuns_<mode>di2_hw): Likewise.
24136 (floatuns_<mode>si2_hw): Likewise.
24137 (xscvqp<su>wz_<mode>): Delete, no longer used.
24138 (xscvqp<su>dz_<mode>): Likewise.
24139 (xscv<su>dqp_<mode>): Likewise.
24140 (ieee128_mfvsrd_64bit): Likewise.
24141 (ieee128_mfvsrd_32bit): Likewise.
24142 (ieee128_mfvsrwz): Likewise.
24143 (ieee128_mtvsrw): Likewise.
24144 (ieee128_mtvsrd_64bit): Likewise.
24145 (ieee128_mtvsrd_32bit): Likewise.
24147 2017-01-31 Martin Liska <mliska@suse.cz>
24150 * ipa-prop.c (ipa_free_all_node_params): Call release method
24151 instead of ~sumbol_summary to not to trigger double times
24154 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
24156 PR tree-optimization/71691
24157 * bitmap.h (class auto_bitmap): New.
24158 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
24159 is_maybe_undefined instead of ssa_undefined_value_p.
24161 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24163 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24164 __S390_ARCH_LEVEL__ to __ARCH__.
24166 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24168 PR tree-optimization/79267
24169 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
24170 if should_remove_lhs_p is true.
24172 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
24175 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
24176 (add_alignment_attribute): New.
24177 (base_type_die): Add alignment attribute.
24178 (subrange_type_die): Likewise.
24179 (modified_type_die): Likewise.
24180 (gen_array_type_die): Likewise.
24181 (gen_descr_array_type_die: Likewise.
24182 (gen_enumeration_type_die): Likewise.
24183 (gen_subprogram_die): Likewise.
24184 (gen_variable_die): Likewise.
24185 (gen_field_die): Likewise.
24186 (gen_ptr_to_mbr_type_die): Likewise.
24187 (gen_struct_or_union_type_die): Likewise.
24188 (gen_subroutine_type_die): Likewise.
24189 (gen_typedef_die): Likewise.
24190 (base_type_cmp): Compare alignment attribute.
24192 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24195 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
24196 (setb_unsigned) New pattern for setb with CCUNS.
24197 * config/rs6000/rs6000.c (expand_block_compare): Use a different
24198 subfc./subfe sequence to avoid overflow problems. Generate a
24199 shorter sequence with cmpld/setb for power9.
24200 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
24201 for generating subfc. instruction.
24202 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
24203 now uses this instruction.
24205 2017-01-30 Ian Lance Taylor <iant@google.com>
24208 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
24209 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
24211 2017-01-30 Martin Sebor <msebor@redhat.com>
24213 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
24214 Move constant to the right of a relational operator.
24215 (get_mpfr_format_length, format_character, format_string): Ditto.
24216 (should_warn_p, maybe_warn): Same.
24218 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
24220 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
24223 * asan.c (get_translation_unit_decl): Remove function.
24224 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
24226 2017-01-30 Martin Liska <mliska@suse.cz>
24228 PR gcov-profile/79259
24229 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
24230 -fprofile-generate.
24232 2017-01-30 Martin Liska <mliska@suse.cz>
24235 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
24236 Initialize variables with NULL value.
24238 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
24241 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
24243 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
24245 2017-01-30 Richard Biener <rguenther@suse.de>
24247 PR tree-optimization/79276
24248 * tree-vrp.c (process_assert_insertions): Properly adjust common
24249 when removing a duplicate.
24251 * gcc.dg/torture/pr79276.c: New testcase.
24253 2017-01-30 Richard Biener <rguenther@suse.de>
24255 PR tree-optimization/79256
24256 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24257 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24259 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
24261 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
24264 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
24265 ("*r<noxa>sbg_<mode>_sll_bitmask")
24266 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
24267 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
24268 Use contiguous_bitmask_nowrap_operand.
24270 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24273 * config/rs6000/altivec.h (vec_xl): Revise #define.
24274 (vec_xst): Likewise.
24276 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
24278 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
24280 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
24282 PR rtl-optimization/79194
24283 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
24284 traps before call to bypass_conditional_jumps.
24286 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24288 PR tree-optimization/71374
24289 * lra-constraints.c (check_conflict_input_operands): New.
24290 (match_reload): Use it.
24292 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24295 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
24296 account to calculate conflict_set.
24298 2017-01-27 Bin Cheng <bin.cheng@arm.com>
24300 PR rtl-optimization/78559
24301 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
24302 other_insn in combine.
24304 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24306 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
24307 uint16_type_node for BT_UINT16.
24309 2017-01-27 David Malcolm <dmalcolm@redhat.com>
24311 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
24312 "RTL Tests" to menu.
24313 (GIMPLE Tests): New node.
24314 (RTL Tests): New node.
24316 2017-01-27 Richard Biener <rguenther@suse.de>
24318 PR tree-optimization/79245
24319 * tree-loop-distribution.c (distribute_loop): Apply cost
24320 modeling also to detected patterns.
24322 2017-01-27 Richard Biener <rguenther@suse.de>
24324 PR tree-optimization/71433
24325 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
24326 (compare_assert_loc): New function.
24327 (process_assert_insertions): Sort and optimize assert locations
24328 to remove duplicates and push down identical assertions on
24329 edges to their destination block.
24331 2017-01-27 Richard Biener <rguenther@suse.de>
24333 PR tree-optimization/79244
24334 * tree-vrp.c (remove_range_assertions): Forcefully propagate
24335 out SSA names even if abnormal.
24337 2017-01-27 Jakub Jelinek <jakub@redhat.com>
24339 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
24340 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
24341 instead of MPFR_RNDN.
24343 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
24346 * arm.c (arm_option_override): Don't call build_target_option_node
24347 until after doing all option overrides.
24348 (arm_valid_target_attribute_tree): Likewise.
24350 2017-01-27 Martin Liska <mliska@suse.cz>
24352 * doc/invoke.texi (-fprofile-arcs): Document profiling support
24353 for {cd}tors and C++ {cd}tors.
24355 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
24357 * config/s390/s390.md ("*setmem_long_and")
24358 ("*setmem_long_and_31z"): Use zero_extend instead of and.
24360 2017-01-26 Martin Sebor <msebor@redhat.com>
24362 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
24365 2017-01-26 Martin Sebor <msebor@redhat.com>
24367 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
24368 HAVE_DFmode before using XFmode or DFmode.
24369 (parse_directive): Avoid using the z length modifier to avoid
24370 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
24372 PR middle-end/78703
24373 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
24374 to accept adjustment as an array.
24375 (get_int_range): New function.
24376 (struct directive): Make width and prec arrays.
24377 (directive::set_width, directive::set_precision): Call get_int_range.
24378 (format_integer, format_floating): Handle width and precision ranges.
24379 (format_string, parse_directive): Same.
24381 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24384 * dwarf2out.c (generate_skeleton_bottom_up): For children with
24385 comdat_type_p set, just clone them, but keep the children in the
24389 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
24390 which have direct callers with -fvar-tracking-assignments enabled
24392 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
24393 inside of type units.
24395 2017-01-26 Martin Sebor <msebor@redhat.com>
24397 PR middle-end/78703
24398 * gimple-ssa-sprintf.c (struct result_range): Add likely and
24400 (struct format_result): Replace number_chars, number_chars_min,
24401 and number_chars_max with a single member of struct result_range.
24403 (format_result::operator+=): Adjust.
24404 (struct fmtresult): Remove bounded. Handle likely and unlikely
24406 (fmtresult::adjust_for_width_or_precision): New function.
24407 (fmtresult:type_max_digits): New function.
24408 (bytes_remaining): Handle likely and unlikely counters.
24409 (min_bytes_remaining): Remove.
24410 (format_percent): Simplify.
24411 (format_integer, format_floating): Set likely and unlikely counters.
24412 (get_string_length, format_character, format_string): Same.
24413 (format_plain, should_warn_p): New function.
24414 (maybe_warn): Call should_warn_p. Update diagnostic messages
24415 and handle those for all directives, including plain strings.
24416 (format_directive): Handle likely and unlikely counters.
24417 Remove unnecessary quoting from diagnostics. Add an informational
24419 (add_bytes): Remove.
24420 (pass_sprintf_length::compute_format_length): Simplify.
24421 (try_substitute_return_value): Handle likely and unlikely counters.
24423 2017-01-26 Carl Love <cel@us.ibm.com>
24425 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
24426 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24428 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24431 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24432 endianess for subregs into account.
24433 * lra-constraints.c (lra_constraints): Do risky transformations
24434 always on the first iteration.
24435 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24436 last_call_used_reg_set.
24437 (process_bb_lives): Define and use last_call_used_reg_set.
24438 * lra.c (lra): Always continue after lra_constraints on the first
24441 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24443 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24445 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24447 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24449 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24450 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24451 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24452 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24453 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24454 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24455 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24456 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24457 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24459 2017-01-26 Marek Polacek <polacek@redhat.com>
24462 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24463 for the third operand.
24465 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24467 PR middle-end/79236
24468 * omp-low.c (struct omp_context): Add simt_stmt field.
24469 (scan_omp_for): Return omp_context *.
24470 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24471 context to the _simt_ SIMD stmt.
24472 (lower_omp_for): For combined SIMD with sibling _simt_
24473 SIMD, make sure to use the same decls in _looptemp_
24474 clauses as in the sibling.
24476 2017-01-26 David Sherwood <david.sherwood@arm.com>
24478 PR middle-end/79212
24479 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24482 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24485 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24486 emit fld b; fld a; if possible.
24488 * brig-builtins.def: Update copyright years.
24489 * config/arm/arm_acle_builtins.def: Update copyright years.
24491 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24494 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24495 constraint instead of o for the stxsd instruction.
24497 2017-01-25 Carl Love <cel@us.ibm.com>
24499 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24500 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24502 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24504 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24506 2017-01-25 Richard Biener <rguenther@suse.de>
24508 PR tree-optimization/69264
24509 * target.def (vector_alignment_reachable): Improve documentation.
24510 * doc/tm.texi: Regenerate.
24511 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24513 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24514 earlier changes with respect to TYPE_USER_ALIGN.
24515 (vector_alignment_reachable_p): Likewise. Improve dumping.
24517 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24520 * config/arm/arm.md (xordi3): Force constant operand into a register
24523 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24525 * doc/invoke.texi (-fstore-merging): Correct default optimization
24526 levels at which it is enabled.
24527 (-O): Move -fstore-merging from list to...
24530 2017-01-25 Richard Biener <rguenther@suse.de>
24533 * omp-expand.c: Include debug.h.
24534 (expand_omp_taskreg): Make sure to generate early debug before
24535 outlining anything from a function.
24536 (expand_omp_target): Likewise.
24537 (grid_expand_target_grid_body): Likewise.
24539 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24542 * asan.c (get_translation_unit_decl): New function.
24543 (asan_add_global): Extract modules file name from globals
24544 TRANSLATION_UNIT_DECL name.
24546 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24549 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24550 for long calls with APCS frame and VFP.
24552 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24554 * cfg.c (original_copy_tables_initialized_p): New function.
24555 * cfg.h (original_copy_tables_initialized_p): New decl.
24556 * cfgrtl.c (relink_block_chain): Guard the call to
24557 free_original_copy_tables with a call to
24558 original_copy_tables_initialized_p.
24559 * cgraph.h (symtab_node::native_rtl_p): New decl.
24560 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24561 (symtab_node::needed_p): Don't assert for early assembly output
24562 for __RTL functions.
24563 (cgraph_node::finalize_function): Set "force_output" for __RTL
24565 (cgraph_node::analyze): Bail out early for __RTL functions.
24566 (analyze_functions): Update assertion to support __RTL functions.
24567 (cgraph_node::expand): Bail out early for __RTL functions.
24568 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24570 * function.h (struct function): Update comment for field
24572 * gimple-expr.c: Include "tree-pass.h".
24573 (gimple_has_body_p): Return false for __RTL functions.
24574 * Makefile.in (OBJS): Add run-rtl-passes.o.
24575 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24577 (gcc::pass_manager::get_clean_slate): New accessor.
24578 * passes.c: Include "insn-addr.h".
24579 (should_skip_pass_p): Add logging. Update logic for running
24580 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24581 property-provider override so it is only done for gimple passes.
24583 (skip_pass): New function.
24584 (execute_one_pass): Call skip_pass when skipping passes.
24585 * read-md.c (md_reader::read_char): Support filtering
24586 the input to a subset of line numbers.
24587 (md_reader::md_reader): Initialize fields
24588 m_first_line and m_last_line.
24589 (md_reader::read_file_fragment): New function.
24590 * read-md.h (md_reader::read_file_fragment): New decl.
24591 (md_reader::m_first_line): New field.
24592 (md_reader::m_last_line): New field.
24593 * read-rtl-function.c (function_reader::create_function): Only
24594 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24595 curr_properties. Set DECL_INITIAL to a dummy block.
24596 (read_rtl_function_body_from_file_range): New function.
24597 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24599 * run-rtl-passes.c: New file.
24600 * run-rtl-passes.h: New file.
24602 2017-01-24 Jeff Law <law@redhat.com>
24604 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24607 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24609 PR tree-optimization/79159
24610 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24611 (record_nonwrapping_iv): Improve boundary using above function if no
24612 value range information.
24614 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24615 Martin Jambor <mjambor@suse.cz>
24617 * brig-builtins.def: New file.
24618 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24619 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24620 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24621 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24622 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24623 * builtin-types.def (BT_INT8): New.
24624 (BT_INT16): Likewise.
24625 (BT_UINT8): Likewise.
24626 (BT_UINT16): Likewise.
24627 (BT_FN_ULONG): Likewise.
24628 (BT_FN_UINT_INT): Likewise.
24629 (BT_FN_UINT_ULONG): Likewise.
24630 (BT_FN_UINT_LONG): Likewise.
24631 (BT_FN_UINT_PTR): Likewise.
24632 (BT_FN_ULONG_PTR): Likewise.
24633 (BT_FN_INT8_FLOAT): Likewise.
24634 (BT_FN_INT16_FLOAT): Likewise.
24635 (BT_FN_UINT32_FLOAT): Likewise.
24636 (BT_FN_UINT16_FLOAT): Likewise.
24637 (BT_FN_UINT8_FLOAT): Likewise.
24638 (BT_FN_UINT64_FLOAT): Likewise.
24639 (BT_FN_UINT16_UINT32): Likewise.
24640 (BT_FN_UINT32_UINT16): Likewise.
24641 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24642 (BT_FN_INT_PTR_INT): Likewise.
24643 (BT_FN_UINT_PTR_UINT): Likewise.
24644 (BT_FN_LONG_PTR_LONG): Likewise.
24645 (BT_FN_ULONG_PTR_ULONG): Likewise.
24646 (BT_FN_VOID_UINT64_UINT64): Likewise.
24647 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24648 (BT_FN_INT8_INT8_INT8): Likewise.
24649 (BT_FN_INT16_INT16_INT16): Likewise.
24650 (BT_FN_INT_INT_INT): Likewise.
24651 (BT_FN_UINT_FLOAT_UINT): Likewise.
24652 (BT_FN_FLOAT_UINT_UINT): Likewise.
24653 (BT_FN_ULONG_UINT_UINT): Likewise.
24654 (BT_FN_ULONG_UINT_PTR): Likewise.
24655 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24656 (BT_FN_UINT_UINT_UINT): Likewise.
24657 (BT_FN_VOID_UINT_PTR): Likewise.
24658 (BT_FN_UINT_UINT_PTR: Likewise.
24659 (BT_FN_UINT32_UINT64_PTR): Likewise.
24660 (BT_FN_INT_INT_UINT_UINT): Likewise.
24661 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24662 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24663 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24664 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24665 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24666 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24667 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24668 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
24669 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
24670 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
24671 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
24672 * doc/frontends.texi: List BRIG FE.
24673 * doc/install.texi (Testing): Add BRIG tesring requirements.
24674 * doc/invoke.texi (Overall Options): Mention BRIG.
24675 * doc/standards.texi (Standards): Doucment BRIG HSA version.
24677 2017-01-24 Richard Biener <rguenther@suse.de>
24679 PR translation/79208
24680 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
24682 2017-01-24 Martin Jambor <mjambor@suse.cz>
24685 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24686 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24687 and known_contexts.
24689 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24691 PR middle-end/79123
24692 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24693 casts from signed to unsigned really don't have a range.
24695 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24697 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24698 GMP_RNDx for compatiblity.
24700 2017-01-24 Martin Liska <mliska@suse.cz>
24703 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24704 that would prevent us to call alloca with -1 as argument.
24706 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24708 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24709 Avoid -Wformat-security warning.
24711 2017-01-23 Andrew Pinski <apinski@cavium.com>
24713 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24716 2017-01-23 Martin Sebor <msebor@redhat.com>
24718 PR middle-end/78703
24719 * gimple-ssa-sprintf.c (warn_level): New global.
24720 (format_integer): Use it here and throughout the rest of the file.
24721 Use the same switch to compute sign as base.
24722 (maybe_warn): New function.
24723 (format_directive): Factor out warnings into maybe_warn.
24724 Add debugging output. Use warn_level.
24725 (add_bytes): Use warn_level.
24726 (pass_sprintf_length::compute_format_length): Add debugging output.
24727 (try_substitute_return_value): Same.
24728 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24730 PR middle-end/78703
24731 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24732 (struct fmtresult, format_integer, format_floating): Adjust.
24733 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24734 (get_string_length, format_string,format_directive): Same.
24735 (pass_sprintf_length::compute_format_length): Same.
24736 (try_substitute_return_value): Simplify slightly.
24738 PR middle-end/78703
24739 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24740 (fmtresult::operator+=): Outlined.
24741 (struct fmtresult): Add ctors.
24742 (struct conversion_spec): Rename...
24743 (struct directive): ...to this. Add and remove data members.
24744 (directive::set_width, directive::set_precision): New functions.
24745 (format_percent): Use fmtresult ctor.
24746 (get_width_and_precision): Remove.
24747 (format_integer): Make naming changes. Avoid computing width and
24749 (format_floating): Same. Adjust indentation.
24750 (format_character, format_none): New functions.
24751 (format_string): Moved character handling to format_character.
24752 (format_directive): Remove arguments, change return type.
24753 (parse_directive): New function.
24754 (pass_sprintf_length::compute_format_length): Move directive
24755 parsing to parse_directive.
24757 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24759 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24760 (assign_assembler_name_if_needed): ... this.
24761 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24762 (assign_assembler_name_if_needed): ... this.
24763 (free_lang_data_in_cgraph): Adjust callers.
24764 * cgraphunit.c (cgraph_node::analyze): Likewise.
24765 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24768 2017-01-23 Richard Biener <rguenther@suse.de>
24770 PR tree-optimization/79088
24771 PR tree-optimization/79188
24772 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
24773 resetting loop bounds after last path deletion. Reset loop
24774 bounds of the target loop, make code match the comments.
24775 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
24776 Make sure loops need no fixups.
24778 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24780 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
24781 exponent support with double type for first argument.
24782 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
24783 type returned by __builtin_vec_extract_sig,
24784 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
24785 functions from "vector int" to "vector unsigned int" or from
24786 "vector long long int" to "vector unsigned long long int".
24787 Changed type returned by __builtin_vec_extract_exp,
24788 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
24789 functions from "vector int" to "vector unsigned int" or from
24790 "vector long long int" to "vector unsigned long long int".
24791 Changed return type of __builtin_vec_test_data_class,
24792 __builtin_vec_test_data_class_sp, and
24793 __builtin_vec_test_data_class_dp from "vector int" to
24794 "vector bool int" or from "vector long long int" to "vector bool
24795 long long int" and changed second argument type from "unsigned
24796 int" to "int". Added new overloaded function forms "vector float
24797 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
24798 "vector float __builtin_vec_insert_exp_sp (vector float, vector
24799 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
24800 double, vector unsigned long long int)" and "vector double
24801 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
24802 long int)". Changed return type of
24803 __builtin_scalar_test_data_class and
24804 __builtin_scalar_test_data_class_sp and
24805 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
24806 int" and changed second argument from "unsigned int" to "int".
24807 Changed type returned by __builtin_scalar_test_neg,
24808 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
24809 from "int" to "bool int". Added new overloaded function form
24810 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
24811 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
24812 exponent double-precision with floating point first argument.
24813 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
24814 documentation of scalar_test_data_class, scalar_test_neg,
24815 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
24816 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
24817 vec_test_data_class built-in functions to reflect refinements in
24818 their type signatures.
24820 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
24822 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
24824 (aarch64_elf_asm_destructor): Likewise.
24826 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
24828 PR rtl-optimization/78634
24829 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
24830 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
24831 * ifcvt.c (noce_try_cmove): Add missing cost check.
24833 PR rtl-optimization/71724
24834 * combine.c (if_then_else_cond): Look for situations where it is
24835 beneficial to undo the work of one of the recursive calls.
24837 2017-01-23 Bin Cheng <bin.cheng@arm.com>
24839 PR tree-optimization/70754
24840 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
24841 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
24842 combined stmt before it if not NULL.
24843 (combine_chains): Process refs reversely and compute dominance point
24846 2017-01-23 Martin Liska <mliska@suse.cz>
24848 PR tree-optimization/79196
24849 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
24850 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
24852 (strlen_optimize_stmt): Call the renamed function.
24854 2017-01-23 Michael Matz <matz@suse.de>
24856 PR tree-optimization/78384
24857 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
24859 2017-01-23 Richard Biener <rguenther@suse.de>
24861 PR tree-optimization/79186
24862 * tree-vrp.c (register_new_assert_for): Make sure we've seen
24863 both incoming edges before moving an assert.
24865 2017-01-23 Martin Jambor <mjambor@suse.cz>
24867 * ipa-prop.c (load_from_param_1): Removed.
24868 (load_from_unmodified_param): Bits from load_from_param_1 put back
24870 (load_from_param): Removed.
24871 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
24872 with stmt. Reverted back to use of load_from_unmodified_param.
24874 2017-01-23 Martin Jambor <mjambor@suse.cz>
24877 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
24878 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
24879 field a pointer to garbage collected vector, mark lattices and
24880 ipcp_orig_node with GTY((skip)).
24881 (ipa_get_param_count): Adjust to descriptors being a pointer.
24882 (ipa_get_param): Likewise.
24883 (ipa_get_type): Likewise.
24884 (ipa_get_param_move_cost): Likewise.
24885 (ipa_set_param_used): Likewise.
24886 (ipa_get_controlled_uses): Likewise.
24887 (ipa_set_controlled_uses): Likewise.
24888 (ipa_is_param_used): Likewise.
24889 (ipa_node_params_t): Move into garbage collector. New methods insert
24891 (ipa_node_params_sum): Annotate wth GTY(()).
24892 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
24894 (ipa_load_from_parm_agg): Adjust declaration.
24895 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
24896 * ipa-profile.c (ipa_profile): Likewise.
24897 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
24898 (ipa_populate_param_decls): Make descriptors parameter garbage
24900 (ipa_dump_param): Adjust to descriptors being a pointer.
24901 (ipa_alloc_node_params): Likewise.
24902 (ipa_initialize_node_params): Likewise.
24903 (load_from_param_1): Make descriptors parameter garbage collected.
24904 (load_from_unmodified_param): Likewise.
24905 (load_from_param): Likewise.
24906 (ipa_load_from_parm_agg): Likewise.
24907 (ipa_node_params::~ipa_node_params): Removed.
24908 (ipa_free_all_node_params): Remove call to delete operator.
24909 (ipa_node_params_t::insert): New.
24910 (ipa_node_params_t::remove): Likewise.
24911 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
24912 copy known_csts and known_contexts vectors.
24913 (ipa_read_node_info): Adjust to descriptors being a pointer.
24914 (ipcp_modif_dom_walker): Make m_descriptors field garbage
24916 (ipcp_transform_function): Make descriptors variable garbage
24919 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
24921 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
24922 * config/i386/avx512dqintrin.h: Ditto.
24923 * config/i386/avx512fintrin.h: Ditto.
24924 * config/i386/i386.c: Handle new builtins.
24925 * config/i386/i386-builtin.def: Add new builtins.
24926 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
24927 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
24929 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24930 Martin Liska <mliska@suse.cz>
24932 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
24933 * asan.c (asan_expand_poison_ifn): Support stores and use
24934 appropriate ASAN report function.
24935 * internal-fn.c (expand_ASAN_POISON_USE): New function.
24936 * internal-fn.def (ASAN_POISON_USE): Declare.
24937 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
24938 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
24939 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
24940 ASAN_POISON calls w/o LHS.
24941 * tree-ssa.c (execute_update_addresses_taken): Create clobber
24942 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
24943 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
24944 * gimplify.c (asan_poison_variables): Add attribute
24945 use_after_scope_memory to variables that really needs to live
24947 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
24948 having the attribute.
24950 2017-01-23 Martin Liska <mliska@suse.cz>
24952 * asan.c (create_asan_shadow_var): New function.
24953 (asan_expand_poison_ifn): Likewise.
24954 * asan.h (asan_expand_poison_ifn): New declaration.
24955 * internal-fn.c (expand_ASAN_POISON): Likewise.
24956 * internal-fn.def (ASAN_POISON): New builtin.
24957 * sanopt.c (pass_sanopt::execute): Expand
24958 asan_expand_poison_ifn.
24959 * tree-inline.c (copy_decl_for_dup_finish): Make function
24961 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
24962 * tree-ssa.c (is_asan_mark_p): New function.
24963 (execute_update_addresses_taken): Rewrite local variables
24964 (identified just by use-after-scope as addressable) into SSA.
24966 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24968 * doc/install.texi (Specific): opensource.apple.com uses https
24969 now. Remove trailing slash.
24971 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24973 * README.Portability: Remove note on an Irix compatibility issue.
24975 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
24977 * gcov.c (INCLUDE_ALGORITHM): Define.
24978 (INCLUDE_VECTOR): Define.
24979 No longer include <vector> and <algorithm> directly.
24981 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
24983 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
24985 * doc/invoke.texi (Code Gen Options): Ditto.
24987 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
24990 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
24992 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
24994 rtl-optimization/79125
24995 * cprop.c (local_cprop_pass): Handle cases where we make an
24996 unconditional trap.
24998 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
25002 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
25003 read from, for big endian.
25005 2017-01-20 Jiong Wang <jiong.wang@arm.com>
25007 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
25008 register pauth builtins for LP64 only.
25010 2017-01-20 Marek Polacek <polacek@redhat.com>
25013 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
25016 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25018 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
25020 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
25021 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
25022 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
25024 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25027 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
25028 in big-endian mode when they are not a single duplicated value.
25030 2017-01-20 Richard Biener <rguenther@suse.de>
25032 * BASE-VER: Bump to 7.0.1.
25034 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25036 * omp-low.c (omplow_simd_context): New struct. Use it...
25037 (lower_rec_simd_input_clauses): ...here and...
25038 (lower_rec_input_clauses): ...here to hold common data. Adjust all
25039 references to idx, lane, max_vf, is_simt.
25041 2017-01-20 Graham Markall <graham.markall@embecosm.com>
25043 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
25046 2017-01-20 Martin Jambor <mjambor@suse.cz>
25048 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
25049 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
25051 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
25052 (GTFILES): Rename hsa.c to hsa-common.c.
25053 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
25054 * hsa-dump.c: Likewise.
25055 * hsa-gen.c: Likewise.
25056 * hsa-regalloc.c: Likewise.
25057 * ipa-hsa.c: Likewise.
25058 * omp-expand.c: Likewise.
25059 * omp-low.c: Likewise.
25060 * toplev.c: Likewise.
25062 2017-01-20 Marek Polacek <polacek@redhat.com>
25065 * doc/invoke.texi: Document -Wduplicated-branches.
25066 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
25067 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
25068 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
25069 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
25070 return 0 only when not OEP_LEXICOGRAPHIC.
25071 (fold_build_cleanup_point_expr): Use the expression
25072 location when building CLEANUP_POINT_EXPR.
25073 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
25074 * tree.c (add_expr): Handle error_mark_node.
25076 2017-01-20 Martin Liska <mliska@suse.cz>
25079 * tree-profile.c (init_ic_make_global_vars): Do not call
25081 (gimple_init_gcov_profiler): Likewise.
25083 2017-01-20 Martin Liska <mliska@suse.cz>
25086 * cgraph.h (maybe_create_reference): Remove argument and
25088 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
25090 * ipa-cp.c (create_specialized_node): Likewise.
25091 * symtab.c (symtab_node::maybe_create_reference): Handle
25092 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
25094 2017-01-20 Martin Liska <mliska@suse.cz>
25096 * read-rtl-function.c (function_reader::create_function): Use
25097 build_decl instread of build_decl_stat.
25099 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
25101 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
25102 * config/i386/avx512dqintrin.h: Ditto.
25103 * config/i386/avx512fintrin.h: Ditto.
25104 * config/i386/i386-builtin-types.def: Add new types.
25105 * config/i386/i386.c: Handle new types.
25106 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
25107 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
25108 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
25109 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
25110 (__builtin_ia32_kshiftridi): New.
25111 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
25113 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
25117 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
25118 define to rs6000_init_stack_protect_guard.
25119 (rs6000_init_stack_protect_guard): New function.
25121 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25122 Yunqiang Su <yunqiang.su@imgtec.com>
25124 * config.gcc (supported_defaults): Add madd4.
25125 (with_madd4): Add validation.
25126 (all_defaults): Add madd4.
25127 * config/mips/mips.opt (mmadd4): New option.
25128 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25130 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25132 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
25133 (ISA_HAS_FUSED_MADD4): Likewise.
25134 * doc/invoke.texi (-mmadd4): Document the new option.
25135 * doc/install.texi (--with-madd4): Document the new option.
25137 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25139 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
25140 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
25141 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
25142 (aarch64_init_pauth_hint_builtins): New.
25143 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
25144 (aarch64_expand_builtin): Expand new builtins.
25146 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25148 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
25149 * combine-stack-adj.c (no_unhandled_cfa): Handle
25150 REG_CFA_TOGGLE_RA_MANGLE.
25151 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
25152 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
25153 info for return address signing.
25154 (aarch64_expand_epilogue): Likewise.
25156 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25158 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
25159 * config/aarch64/aarch64-protos.h
25160 (aarch64_return_address_signing_enabled): New declaration.
25161 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
25163 (aarch64_expand_prologue): Sign return address before it's pushed onto
25165 (aarch64_expand_epilogue): Authenticate return address fetched from
25167 (aarch64_override_options): Sanity check for ILP32 and ISA level.
25168 (aarch64_attributes): New function attributes for "sign-return-address".
25169 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
25170 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
25171 ("*do_return"): Generate combined instructions according to key index.
25172 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
25173 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
25175 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
25176 * config/aarch64/aarch64.opt (msign-return-address=): New.
25177 * doc/extend.texi (AArch64 Function Attributes): Documents
25178 "sign-return-address=".
25179 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
25181 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25183 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
25184 overall option summary.
25186 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25188 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
25189 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
25190 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
25191 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
25193 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
25195 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
25196 -mpower9-minmax by default for -mcpu=power9.
25197 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
25198 128-bit floating point.
25200 2017-01-20 Alan Modra <amodra@gmail.com>
25202 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
25203 optimizing for size.
25205 2017-01-20 Alan Modra <amodra@gmail.com>
25208 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
25209 for strcmp and strncmp from corresponding builtin decl.
25211 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25213 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
25214 instead of i386/rtems-64.h.
25215 * config/i386/rtems-64.h: Remove.
25217 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25221 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
25223 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
25225 2017-01-19 Tamar Christina <tamar.christina@arm.com>
25227 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
25228 Change int to HOST_WIDE_INT.
25229 * config/aarch64/aarch64-protos.h
25230 (aarch64_simd_gen_const_vector_dup): Likewise.
25231 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
25233 2017-01-19 David Malcolm <dmalcolm@redhat.com>
25235 * langhooks-def.h (lhd_type_for_size): New decl.
25236 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
25237 * langhooks.c (lhd_type_for_size): New function, taken from
25240 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
25242 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
25243 define_bypass for CR latency.
25244 (power9-cracked-alu): Update bypass latency and remove power9-branch.
25245 (power9-alu2): Add define_bypass for CR latency.
25247 (power9-mul): Update insn latency.
25248 (power9-mul-compare): Update insn latency, bypass latency and remove
25251 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25253 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
25255 * config/aarch64/aarch64.md
25256 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
25257 aarch64_nopcrelative_literal_loads.
25258 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
25260 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
25262 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
25263 TARGET_LOONGSON_3A.
25264 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
25266 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
25269 * config.gcc (supported_defaults): Add lxc1-sxc1.
25270 (with_lxc1_sxc1): Add validation.
25271 (all_defaults): Add lxc1-sxc1.
25272 * config/mips/mips.opt (mlxc1-sxc1): New option.
25273 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25275 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25276 __mips_no_lxc1_sxc1.
25277 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
25278 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
25279 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
25281 2017-01-19 Richard Biener <rguenther@suse.de>
25283 PR tree-optimization/72488
25284 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
25285 sure to restore SSA info.
25286 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
25288 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
25290 PR rtl-optimization/79121
25291 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
25292 of the inner type when shifting an extended value.
25294 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25297 * symtab.c (symtab_node::equal_address_to): Fix comparing of
25298 interposable aliases.
25300 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
25303 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
25304 Use the evmergelohi instruction.
25305 (mov_si<mode>_e500_subreg4_2_le): Likewise.
25306 (mov_sitf_e500_subreg8_2_be): Likewise.
25307 (mov_sitf_e500_subreg12_2_le): Likewise.
25308 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
25309 (mov_si<mode>_e500_subreg4_2_be): Likewise.
25310 (mov_sitf_e500_subreg8_2_le): Likewise.
25311 (mov_sitf_e500_subreg12_2_be): Likewise.
25313 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25315 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
25316 attribute from vecsimple to vecperm.
25317 (altivec_vbpermq2): Likewise.
25319 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25322 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
25324 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25325 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
25326 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
25327 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
25328 case where N arg is SIZE_MAX.
25329 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
25330 (cmpstrsi): Add pattern.
25332 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25334 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25335 __builtin_vec_revb builtins.
25336 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
25337 built-in functions to support generation of the ISA 3.0 XXBR<x>
25338 vector byte reverse instructions.
25339 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
25340 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
25341 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
25342 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
25343 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
25344 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
25345 (P9V_BUILTIN_VEC_REVB): Likewise.
25346 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
25347 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
25348 (p9_xxbrq_v16qi): Likewise.
25349 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
25350 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
25351 (p9_xxbrh_v8hi): Likewise.
25352 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
25353 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
25354 vec_revb built-in functions.
25356 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
25358 PR rtl-optimization/78952
25359 * config/i386/i386.md (any_extract): New code iterator.
25360 (*insvqi_2): Use any_extract for source operand.
25361 (*insvqi_3): Use any_shiftrt for source operand.
25363 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25365 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
25367 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
25369 2017-01-18 Matthias Klose <doko@ubuntu.com>
25371 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
25373 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25375 * config/rs6000/altivec.h (vec_bperm): Change #define.
25376 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
25377 (altivec_vbpermq2): New define_insn.
25378 (altivec_vbpermd): Likewise.
25379 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
25380 function interface.
25381 (VBPERMD): Likewise.
25382 (VBPERM): New polymorphic function interface.
25383 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
25384 Add entries for P9V_BUILTIN_VEC_VBPERM.
25385 * doc/extend.texi: Add interfaces for vec_bperm.
25387 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25389 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
25390 first letter of error messages.
25391 (s390_resolve_overloaded_builtin): Likewise.
25392 * config/s390/s390.c (s390_expand_builtin): Likewise.
25393 (s390_invalid_arg_for_unprototyped_fn): Likewise.
25394 (s390_valid_target_attribute_inner_p): Likewise.
25395 * config/s390/s390.md ("tabort"): Likewise.
25397 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
25399 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
25400 (ISA_AVOID_DIV_HILO): New macro.
25401 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
25402 (ISA_HAS_DDIV): Likewise.
25404 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25406 * doc/invoke.texi (fabi-version): Correct number of occurrences.
25408 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25410 * doc/invoke.texi (fabi-version): Spelling fix.
25412 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25415 * doc/invoke.texi (fabi-version): Mention mangling fix for
25418 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25421 * doc/invoke.texi (fabi-version): Document discriminator mangling.
25423 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
25426 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
25427 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25429 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25431 (stack_protect_test): Ditto.
25432 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25433 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25435 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25436 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25437 -mstack-protector-guard-offset=.
25438 (RS/6000 and PowerPC Options): Ditto.
25440 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25442 * config/i386/i386.h (MASK_CLASS_P): New define.
25443 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25444 there are no registers from different register sets also when
25445 mask registers are used. Update function comment.
25446 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25447 to (*k/*r) and (*k/*km) alternatives.
25449 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25451 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25452 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25453 (EH_RETURN_HANDLER_RTX): New define.
25454 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25455 Force frame pointer in EH return functions.
25456 (aarch64_expand_epilogue): Add barrier for eh_return.
25457 (aarch64_final_eh_return_addr): Remove.
25458 (aarch64_eh_return_handler_rtx): New function.
25459 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25461 (aarch64_eh_return_handler_rtx): New prototype.
25463 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25465 * config/rs6000/altivec.h (vec_rlmi): New #define.
25466 (vec_vrlnm): Likewise.
25467 (vec_rlnm): Likewise.
25468 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25469 (UNSPEC_VRLNM): Likewise.
25470 (VIlong): New mode iterator.
25471 (altivec_vrl<VI_char>mi): New define_insn.
25472 (altivec_vrl<VI_char>nm): Likewise.
25473 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25475 (VRLDNM): Likewise.
25476 (RLNM): New polymorphic function entry.
25477 (VRLWMI): New monomorphic function entry.
25478 (VRLDMI): Likewise.
25479 (RLMI): New polymorphic function entry.
25480 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25481 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25482 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25485 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25488 * dwarf2out.c (field_byte_offset): Restore the
25489 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25490 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25491 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25494 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25497 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25499 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25502 * dwarf2out.c (add_data_member_location_attribute): For constant
25503 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25504 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25505 DW_AT_byte_size attributes.
25507 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25509 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25510 after forcing to constant memory when the code model is medium.
25512 2017-01-17 Julia Koval <julia.koval@intel.com>
25515 * config/i386/avx512fintrin.h
25516 (_mm512_i32gather_ps): Change __addr type to void const*.
25517 (_mm512_mask_i32gather_ps): Ditto.
25518 (_mm512_i32gather_pd): Ditto.
25519 (_mm512_mask_i32gather_pd): Ditto.
25520 (_mm512_i64gather_ps): Ditto.
25521 (_mm512_mask_i64gather_ps): Ditto.
25522 (_mm512_i64gather_pd): Ditto.
25523 (_mm512_mask_i64gather_pd): Ditto.
25524 (_mm512_i32gather_epi32): Ditto.
25525 (_mm512_mask_i32gather_epi32): Ditto.
25526 (_mm512_i32gather_epi64): Ditto.
25527 (_mm512_mask_i32gather_epi64): Ditto.
25528 (_mm512_i64gather_epi32): Ditto.
25529 (_mm512_mask_i64gather_epi32): Ditto.
25530 (_mm512_i64gather_epi64): Ditto.
25531 (_mm512_mask_i64gather_epi64): Ditto.
25532 (_mm512_i32scatter_ps): Change __addr type to void*.
25533 (_mm512_mask_i32scatter_ps): Ditto.
25534 (_mm512_i32scatter_pd): Ditto.
25535 (_mm512_mask_i32scatter_pd): Ditto.
25536 (_mm512_i64scatter_ps): Ditto.
25537 (_mm512_mask_i64scatter_ps): Ditto.
25538 (_mm512_i64scatter_pd): Ditto.
25539 (_mm512_mask_i64scatter_pd): Ditto.
25540 (_mm512_i32scatter_epi32): Ditto.
25541 (_mm512_mask_i32scatter_epi32): Ditto.
25542 (_mm512_i32scatter_epi64): Ditto.
25543 (_mm512_mask_i32scatter_epi64): Ditto.
25544 (_mm512_i64scatter_epi32): Ditto.
25545 (_mm512_mask_i64scatter_epi32): Ditto.
25546 (_mm512_i64scatter_epi64): Ditto.
25547 (_mm512_mask_i64scatter_epi64): Ditto.
25548 * config/i386/avx512pfintrin.h
25549 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25550 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25551 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25552 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25553 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25554 (_mm512_prefetch_i32scatter_ps): Ditto.
25555 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25556 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25557 (_mm512_prefetch_i64scatter_pd): Ditto.
25558 (_mm512_prefetch_i64scatter_ps): Ditto.
25559 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25560 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25561 * config/i386/avx512vlintrin.h
25562 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25563 (_mm_mmask_i32gather_ps): Ditto.
25564 (_mm256_mmask_i32gather_pd): Ditto.
25565 (_mm_mmask_i32gather_pd): Ditto.
25566 (_mm256_mmask_i64gather_ps): Ditto.
25567 (_mm_mmask_i64gather_ps): Ditto.
25568 (_mm256_mmask_i64gather_pd): Ditto.
25569 (_mm_mmask_i64gather_pd): Ditto.
25570 (_mm256_mmask_i32gather_epi32): Ditto.
25571 (_mm_mmask_i32gather_epi32): Ditto.
25572 (_mm256_mmask_i32gather_epi64): Ditto.
25573 (_mm_mmask_i32gather_epi64): Ditto.
25574 (_mm256_mmask_i64gather_epi32): Ditto.
25575 (_mm_mmask_i64gather_epi32): Ditto.
25576 (_mm256_mmask_i64gather_epi64): Ditto.
25577 (_mm_mmask_i64gather_epi64): Ditto.
25578 (_mm256_i32scatter_ps): Change __addr type to void*.
25579 (_mm256_mask_i32scatter_ps): Ditto.
25580 (_mm_i32scatter_ps): Ditto.
25581 (_mm_mask_i32scatter_ps): Ditto.
25582 (_mm256_i32scatter_pd): Ditto.
25583 (_mm256_mask_i32scatter_pd): Ditto.
25584 (_mm_i32scatter_pd): Ditto.
25585 (_mm_mask_i32scatter_pd): Ditto.
25586 (_mm256_i64scatter_ps): Ditto.
25587 (_mm256_mask_i64scatter_ps): Ditto.
25588 (_mm_i64scatter_ps): Ditto.
25589 (_mm_mask_i64scatter_ps): Ditto.
25590 (_mm256_i64scatter_pd): Ditto.
25591 (_mm256_mask_i64scatter_pd): Ditto.
25592 (_mm_i64scatter_pd): Ditto.
25593 (_mm_mask_i64scatter_pd): Ditto.
25594 (_mm256_i32scatter_epi32): Ditto.
25595 (_mm256_mask_i32scatter_epi32): Ditto.
25596 (_mm_i32scatter_epi32): Ditto.
25597 (_mm_mask_i32scatter_epi32): Ditto.
25598 (_mm256_i32scatter_epi64): Ditto.
25599 (_mm256_mask_i32scatter_epi64): Ditto.
25600 (_mm_i32scatter_epi64): Ditto.
25601 (_mm_mask_i32scatter_epi64): Ditto.
25602 (_mm256_i64scatter_epi32): Ditto.
25603 (_mm256_mask_i64scatter_epi32): Ditto.
25604 (_mm_i64scatter_epi32): Ditto.
25605 (_mm_mask_i64scatter_epi32): Ditto.
25606 (_mm256_i64scatter_epi64): Ditto.
25607 (_mm256_mask_i64scatter_epi64): Ditto.
25608 (_mm_i64scatter_epi64): Ditto.
25609 (_mm_mask_i64scatter_epi64): Ditto.
25610 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25611 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25612 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25613 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25614 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25615 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25616 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25617 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25618 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25619 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25620 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25621 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25622 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25623 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25624 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25625 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25626 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25627 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25628 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25629 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25630 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25631 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25632 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25633 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25634 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25635 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25636 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25637 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25638 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25639 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25640 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25641 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25642 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25643 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25644 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25645 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25646 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25647 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25648 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25649 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25650 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25651 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25652 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25653 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25654 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25655 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25656 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25657 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25658 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25659 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25660 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25661 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25662 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25663 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25664 definitions accordingly.
25666 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25667 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25670 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
25673 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
25676 * ira-conflicts.c (ira_build_conflicts): Update total conflict
25677 hard regs for inner regno.
25679 2017-01-17 Martin Liska <mliska@suse.cz>
25682 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
25683 assumption and add comment.
25685 2017-01-17 Nathan Sidwell <nathan@acm.org>
25687 * ipa-visibility.c (localize_node): New function, broken out of ...
25688 (function_and_variable_visibility): ... here. Call it.
25690 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25692 PR middle-end/77445
25693 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25694 correctly set frequency of oudgoing edge.
25695 (duplicate_thread_path): Fix profile updating.
25697 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25700 * configure.ac: Add GCC_BASE_VER.
25701 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25702 version from BASE-VER file.
25703 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25704 (gcc.o): Depend on $(BASEVER).
25705 * common.opt (dumpfullversion): New option.
25706 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25707 * doc/invoke.texi: Document -dumpfullversion.
25708 * doc/install.texi: Document --with-gcc-major-version-only.
25709 * configure: Regenerated.
25711 2017-01-17 Richard Biener <rguenther@suse.de>
25713 PR tree-optimization/71433
25714 * tree-vrp.c (register_new_assert_for): Merge same asserts
25715 on all incoming edges.
25716 (process_assert_insertions_for): Handle insertions at the
25719 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25721 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25722 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25724 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25727 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25730 2017-01-17 Alan Modra <amodra@gmail.com>
25733 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25734 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25735 symbolic stack limit when pic.
25737 2017-01-16 Martin Sebor <msebor@redhat.com>
25739 PR tree-optimization/78608
25740 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25742 2017-01-16 Jeff Law <law@redhat.com>
25745 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25746 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25747 for several include directories that may be relative to sysroot.
25748 * config/i386/x-mingw32 (gplus_includedir): Define.
25749 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25750 (native_system_includedir): Likewise.
25751 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25752 override if TARGET_SYSTEM_ROOT is defined.
25753 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25755 PR tree-optimization/79090
25756 PR tree-optimization/33562
25757 PR tree-optimization/61912
25758 PR tree-optimization/77485
25759 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25760 and computed trims into the dump file.
25762 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25764 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25766 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25769 * gimplify.c (gimplify_init_constructor): If want_value and
25770 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
25774 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
25775 sequence. Formatting fixes.
25776 (doloop_optimize): Formatting fixes.
25779 * gcc.c (debug_level_greater_than_spec_func): New function.
25780 (static_spec_functions): Add debug-level-gt spec function.
25781 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
25783 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25784 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25785 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
25786 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
25787 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
25788 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
25790 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
25792 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
25793 QImode fixups to general and mask registers only.
25795 2017-01-16 Carl Love <cel@us.ibm.com>
25797 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
25798 for built-in functions
25799 vector signed char vec_nabs (vector signed char)
25800 vector signed short vec_nabs (vector signed short)
25801 vector signed int vec_nabs (vector signed int)
25802 vector signed long long vec_nabs (vector signed long long)
25803 vector float vec_nabs (vector float)
25804 vector double vec_nabs (vector double)
25805 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
25807 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
25808 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
25809 * doc/extend.texi: Update the documentation file for the new built-in
25812 2017-01-16 Martin Sebor <msebor@redhat.com>
25814 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
25817 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25819 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
25820 UNSPEC_VSX__XXSPLTD to require special splat handling.
25822 2017-01-16 David Malcolm <dmalcolm@redhat.com>
25825 * system.h: Poison strndup.
25827 2017-01-16 Alan Modra <amodra@gmail.com>
25830 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
25833 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
25835 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
25837 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
25839 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
25840 call recog here. Assert that INSN_CODE (insn) is non-negative.
25842 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25845 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
25847 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
25848 in the currently scheduled RTL fragment.
25850 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25852 PR rtl-optimization/78751
25853 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
25856 2017-01-14 Jeff Law <law@redhat.com>
25858 PR tree-optimization/79090
25859 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
25860 variable length stores.
25861 (compute_trims): Delete dead assignment to *trim_tail.
25862 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
25865 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
25867 PR rtl-optimization/78626
25868 PR rtl-optimization/78727
25869 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
25870 of a block, and split such blocks after everything else is finished.
25872 2017-01-14 Alan Modra <amodra@gmail.com>
25875 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
25876 target legitimate_combined_insn.
25877 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
25878 (rs6000_legitimate_combined_insn): New function.
25879 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
25881 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
25882 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
25883 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
25885 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
25887 * doc/frontends.texi (G++ and GCC): Remove references to Java.
25889 2017-01-13 Jeff Law <law@redhat.com>
25891 PR tree-optimization/33562
25892 PR tree-optimization/61912
25893 PR tree-optimization/77485
25894 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
25896 (delete_dead_assignment): Likewise.
25897 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
25898 statement to delete_dead_call and delete_dead_assignment.
25900 2017-01-13 David Malcolm <dmalcolm@redhat.com>
25903 * substring-locations.c (format_warning_va): Strengthen case 1 so
25904 that both endpoints of the substring must be within the format
25905 range for just the substring to be printed.
25907 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
25909 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
25910 * config/i386/i386.c (ix86_target_string): Add missing options
25911 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
25912 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
25913 flags_other and ix86_target_other to flags2_other. Display unknown
25915 (ix86_valid_target_attribute_inner_p): Add missing options and
25916 reorder options by implied ISAs, as in ix86_target_string.
25918 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25920 * hash-table.h (hash_table::too_empty_p): New function.
25921 (hash_table::expand): Use it.
25922 (hash_table::traverse): Likewise.
25923 (hash_table::empty_slot): Use sizeof (value_type) instead of
25924 sizeof (PTR) to convert bytes to elements. Shrink the table
25925 if the current size is excessive for the current number of
25928 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25930 * ira-costs.c (record_reg_classes): Break from the inner loop
25931 early once alt_fail is known to be true. Update outer loop
25932 handling accordingly.
25934 2017-01-13 Jeff Law <law@redhat.com>
25936 * tree-ssa-dse.c (decrement_count): New function.
25937 (increment_start_addr, maybe_trim_memstar_call): Likewise.
25938 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
25939 when we know the partially dead statement is a mem* function.
25941 PR tree-optimization/61912
25942 PR tree-optimization/77485
25943 * tree-ssa-dse.c: Include expr.h.
25944 (maybe_trim_constructor_store): New function.
25945 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
25947 PR tree-optimization/33562
25948 PR tree-optimization/61912
25949 PR tree-optimization/77485
25950 * doc/invoke.texi: Document new dse-max-object-size param.
25951 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
25952 * tree-ssa-dse.c: Include params.h.
25953 (dse_store_status): New enum.
25954 (initialize_ao_ref_for_dse): New, partially extracted from
25956 (valid_ao_ref_for_dse, normalize_ref): New.
25957 (setup_live_bytes_from_ref, compute_trims): Likewise.
25958 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
25959 (maybe_trim_partially_dead_store): Likewise.
25960 (maybe_trim_complex_store): Likewise.
25961 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
25962 Track what bytes live from the original store. Return tri-state
25963 for dead, partially dead or live.
25964 (dse_dom_walker): Add constructor, destructor and new private members.
25965 (delete_dead_call, delete_dead_assignment): New extracted from
25967 (dse_optimize_stmt): Make a member of dse_dom_walker.
25968 Use initialize_ao_ref_for_dse.
25970 PR tree-optimization/33562
25971 PR tree-optimization/61912
25972 PR tree-optimization/77485
25973 * sbitmap.h (bitmap_count_bits): Prototype.
25974 (bitmap_clear_range, bitmap_set_range): Likewise.
25975 * sbitmap.c (bitmap_clear_range): New function.
25976 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
25978 2017-01-13 Martin Liska <mliska@suse.cz>
25981 * function.c (set_cfun): Add new argument force.
25982 * function.h (set_cfun): Likewise.
25983 * ipa-inline-transform.c (inline_call): Use the function when
25984 strict alising from is dropped for function we inline to.
25986 2017-01-13 Richard Biener <rguenther@suse.de>
25988 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
25989 for dumping GIMPLE INTEGER_CSTs.
25991 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25993 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
25994 to 201112L since C++17.
25996 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
25999 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
26000 if -fsanitize=kernel-address is present.
26002 2017-01-13 Richard Biener <rguenther@suse.de>
26004 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
26005 as _Literal ( type ) number in case usual suffixes do not
26006 preserve all information.
26008 2017-01-13 Richard Biener <rguenther@suse.de>
26010 PR tree-optimization/77283
26011 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
26012 and ssa-iterators.h.
26013 (is_feasible_trace): Implement a cost model based on joiner
26016 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
26019 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
26020 char or short to __float128/_Float128 directly.
26022 2017-01-12 Martin Sebor <msebor@redhat.com>
26024 to -Wformat-overflow.
26025 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
26026 (min_bytes_remaining): Same.
26027 (get_string_length): Same.
26028 (format_string): Same.
26029 (format_directive): Same.
26031 (pass_sprintf_length::handle_gimple_call): Same.
26033 2017-01-12 Jakub Jelinek <jakub@redhat.com>
26035 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
26036 info.nowrite calls with no lhs that can't throw. Return bool
26037 whether gsi_remove has been called or not.
26038 (pass_sprintf_length::handle_gimple_call): Return bool whether
26039 try_substitute_return_value called gsi_remove. Formatting fix.
26040 (pass_sprintf_length::execute): Don't use gsi_remove if
26041 handle_gimple_call returned true.
26044 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
26045 be removed due to side-effects, don't remove following barrier nor
26046 turn the successor edge into fallthru edge.
26048 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26051 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
26052 element-reversing loads and stores as not swappable.
26054 2017-01-12 Nathan Sidwell <nathan@acm.org>
26055 Nicolai Stange <nicstange@gmail.com>
26057 * combine.c (try_combine): Don't ignore result of overlap checking
26058 loop. Combine overlap & asm check into single loop.
26060 2017-01-12 Richard Biener <rguenther@suse.de>
26062 * tree-pretty-print.c (dump_generic_node): Provide -gimple
26063 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
26065 2017-01-12 Richard Biener <rguenther@suse.de>
26067 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
26068 and TS_TARGET_OPTION directly derive from TS_BASE.
26069 * tree-core.h (tree_optimization_option): Derive from tree_base.
26070 (tree_target_option): Likewise.
26072 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
26074 * config/i386/i386.c (memory_address_length): Increase len
26075 only when rip_relative_addr_p returns false.
26077 2017-01-11 Julia Koval <julia.koval@intel.com>
26079 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
26080 (OPTION_MASK_ISA_SGX_SET): New.
26081 (ix86_handle_option): Handle OPT_msgx.
26082 * config.gcc: Added sgxintrin.h.
26083 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
26084 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
26085 * config/i386/i386.c (ix86_target_string): Add -msgx.
26087 (ix86_option_override_internal): Handle new options.
26088 (ix86_valid_target_attribute_inner_p): Add sgx.
26089 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
26090 * config/i386/i386.opt: Add msgx.
26091 * config/i386/sgxintrin.h: New file.
26092 * config/i386/x86intrin.h: Add sgxintrin.h.
26094 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26097 * fold-const.c (maybe_nonzero_address): Return 1 for function
26099 (tree_single_nonzero_warnv_p): Don't handle function local objects
26103 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
26106 2017-01-11 David Malcolm <dmalcolm@redhat.com>
26109 * opts.c: Include "spellcheck.h"
26110 (struct string_fragment): New struct.
26111 (struct edit_distance_traits<const string_fragment &>): New
26113 (get_closest_sanitizer_option): New function.
26114 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
26116 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26118 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
26120 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
26121 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
26122 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
26123 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
26124 for initial die_offset if dwarf_split_debug_info.
26125 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
26126 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
26128 (output_skeleton_debug_sections): Formatting fix. Use
26129 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
26130 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
26132 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
26134 * config/arm/cortex-a53.md: Add bypasses for
26135 cortex_a53_r2f_cvt.
26136 (cortex_a53_r2f): Only use for transfers.
26137 (cortex_a53_f2r): Likewise.
26138 (cortex_a53_r2f_cvt): Add reservation for conversions.
26139 (cortex_a53_f2r_cvt): Likewise.
26141 2017-01-11 Tamar Christina <tamar.christina@arm.com>
26143 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
26144 to all inlined functions, change static to extern.
26146 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
26149 * config/arm/arm.c (legitimize_pic_address): Handle reference to
26151 (arm_assemble_integer): Likewise.
26153 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
26155 * config.gcc: Use new awk script to check CPU, FPU and architecture
26156 parameters for --with-... options.
26157 * config/arm/parsecpu.awk: New file
26158 * config/arm/arm-cpus.in: New file.
26159 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
26161 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
26163 * config/arm/t-arm: Update dependency rules.
26164 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
26165 of processing .def files.
26166 * config/arm/genopt.sh: Deleted.
26167 * config/arm/gentune.sh: Deleted.
26168 * config/arm/arm-cores.def: Deleted.
26169 * config/arm/arm-arches.def: Deleted.
26170 * config/arm/arm-fpus.def: Deleted.
26171 * config/arm/arm-tune.md: Regenerated.
26172 * config/arm/arm-tables.opt: Regenerated.
26173 * config/arm/arm-cpu.h: New generated file.
26174 * config/arm/arm-cpu-data.h: New generated file.
26175 * config/arm/arm-cpu-cdata.h: New generated file.
26177 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
26180 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
26182 (input_varpool_node): Unpack dynamically_initialized bit.
26184 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
26186 PR rtl-optimization/79032
26187 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
26188 the alignment of the adjusted memory reference against that of MODE,
26189 instead of the alignment of the original memory reference.
26191 2017-01-11 Martin Jambor <mjambor@suse.cz>
26193 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
26195 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
26196 decorated functions.
26198 2017-01-11 Richard Biener <rguenther@suse.de>
26200 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
26201 set range/nonnull info for PHI results. Do not set it on
26202 stmts marked for removal.
26204 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
26206 * expr.c (store_field): In the bitfield case, fetch the return value
26207 from the registers before applying a single big-endian adjustment.
26208 Always do a final load for a BLKmode value not larger than a word.
26210 2017-01-10 David Malcolm <dmalcolm@redhat.com>
26213 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26214 that we correctly handle column numbers greater than
26215 LINE_MAP_MAX_COLUMN_NUMBER.
26217 2017-01-10 Martin Sebor <msebor@redhat.com>
26219 PR middle-end/78245
26220 * gimple-ssa-sprintf.c (get_destination_size): Call
26221 {init,fini}object_sizes.
26222 * tree-object-size.c (addr_object_size): Adjust.
26223 (pass_through_call): Adjust.
26224 (pass_object_sizes::execute): Adjust.
26225 * tree-object-size.h (fini_object_sizes): Declare.
26227 2017-01-10 Martin Sebor <msebor@redhat.com>
26229 PR tree-optimization/78775
26230 * builtins.c (get_size_range): Move...
26231 * calls.c: ...to here.
26232 (alloc_max_size): Accept zero argument.
26233 (operand_signed_p): Remove.
26234 (maybe_warn_alloc_args_overflow): Call get_size_range.
26235 * calls.h (get_size_range): Declare.
26237 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
26239 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
26240 from TI's devices.csv file as of September 2016.
26241 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
26243 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
26245 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
26246 * doc/invoke.texi: Likewise.
26247 * doc/md.texi: Likewise.
26248 * doc/objc.texi: Likewise.
26250 2017-01-10 Joshua Conner <joshconner@google.com>
26252 * config/arm/fuchsia-elf.h: New file.
26253 * config/fuchsia.h: New file.
26254 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
26255 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
26257 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
26259 2016-01-10 Richard Biener <rguenther@suse.de>
26261 PR tree-optimization/79034
26262 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26263 Propagate out degenerate PHIs in the joiner.
26265 2017-01-10 Martin Liska <mliska@suse.cz>
26267 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
26268 (sort_congruence_classes_by_decl_uid): Likewise.
26269 (sort_congruence_class_groups_by_decl_uid): Likewise.
26270 (sem_item_optimizer::merge_classes): Sort class, groups in these
26271 classes and members in the groups by DECL_UID of declarations.
26272 This would make merge operations stable.
26274 2017-01-10 Martin Liska <mliska@suse.cz>
26276 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
26277 usage of m_classes_vec.
26278 (sem_item_optimizer::~sem_item_optimizer): Likewise.
26279 (sem_item_optimizer::get_group_by_hash): Likewise.
26280 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
26281 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
26282 (sem_item_optimizer::verify_classes): Likewise.
26283 (sem_item_optimizer::process_cong_reduction): Likewise.
26284 (sem_item_optimizer::dump_cong_classes): Likewise.
26285 (sem_item_optimizer::merge_classes): Likewise.
26286 * ipa-icf.h (congruence_class_hash): Rename from
26287 congruence_class_group_hash. Remove declaration of m_classes_vec.
26289 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
26291 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
26292 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
26293 * config.gcc: Add avx512vpopcntdqintrin.h.
26294 * config/i386/avx512vpopcntdqintrin.h: New.
26295 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
26296 * config/i386/i386-builtin-types.def: Add new types.
26297 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
26298 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
26299 __builtin_ia32_vpopcountq_v8di_mask): New.
26300 * config/i386/i386-c.c (ix86_target_macros_internal): Define
26301 __AVX512VPOPCNTDQ__.
26302 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
26303 (PTA_AVX512VPOPCNTDQ): Define.
26304 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
26305 TARGET_AVX512VPOPCNTDQ_P): Define.
26306 * config/i386/i386.opt: Add mavx512vpopcntdq.
26307 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
26308 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
26310 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26312 PR middle-end/77484
26313 * predict.def (PRED_CALL): Set to 67.
26315 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
26317 * expr.c (store_field): In the bitfield case, if the value comes from
26318 a function call and is of an aggregate type returned in registers, do
26319 not modify the field mode; extract the value in all cases if the mode
26320 is BLKmode and the size is not larger than a word.
26322 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
26325 * config/i386/cpuid.h: Fix undefined behavior.
26327 2017-01-04 Jeff Law <law@redhat.com>
26329 PR tree-optimization/79007
26330 PR tree-optimization/67955
26331 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
26332 conservative for pt.null when flag_non_call_exceptions is on.
26334 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26336 PR translation/79019
26337 PR translation/79020
26338 * params.def (PARAM_INLINE_MIN_SPEEDUP,
26339 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
26340 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
26342 * config/avr/avr.opt (maccumulate-args): Likewise.
26343 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
26344 * common.opt (freport-bug): Likewise.
26345 * cif-code.def (CIF_FINAL_ERROR): Likewise.
26346 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
26347 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
26348 translatable string.
26349 * config/i386/i386.c (function_value_32): Likewise.
26350 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
26351 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
26353 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
26354 * common/config/msp430/msp430-common.c (msp430_handle_option):
26356 * symtab.c (symtab_node::verify_base): Likewise.
26357 * opts.c (set_debug_level): Likewise.
26358 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
26359 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
26360 missing whitespace to translatable strings.
26361 * config/avr/avr.md (bswapsi2): Fix typo in comment.
26362 * config/sh/superh.h: Likewise.
26363 * config/i386/xopintrin.h: Likewise.
26364 * config/i386/znver1.md: Likewise.
26365 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
26366 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
26367 * double-int.h (struct double_int): Likewise.
26368 * double-int.c (div_and_round_double): Likewise.
26369 * wide-int.cc: Likewise.
26370 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
26371 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
26372 * cfgcleanup.c (crossjumps_occured): Renamed to ...
26373 (crossjumps_occurred): ... this.
26374 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
26377 PR tree-optimization/78899
26378 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
26379 returning bool return struct loop *, NULL for failure and the new
26381 (versionable_outer_loop_p): Don't version outer loop if it has
26382 dont_vectorized bit set.
26383 (tree_if_conversion): When versioning outer loop, ensure
26384 tree_if_conversion is performed also on the inner loop of the
26385 non-vectorizable outer loop copy.
26386 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
26387 LOOP_VECTORIZED in inner loop of the scalar outer loop and
26388 prevent vectorization of it.
26389 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
26390 the outer loop vectorization of the non-scalar version is attempted
26391 before vectorization of the inner loop in scalar version. If
26392 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
26393 vectorization of its inner loop.
26394 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
26395 has 2 inner loops, rename also on edges from bb whose single pred
26396 is outer_loop->header. Fix typo in function comment.
26398 2017-01-09 Martin Sebor <msebor@redhat.com>
26401 * asan.c (asan_emit_stack_protection): Increase local buffer size
26402 to avoid snprintf truncation warning.
26404 2017-01-09 Andrew Pinski <apinski@cavium.com>
26406 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
26407 to reference thunderx2t99 for the tuning structure
26408 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
26410 (thunderx2t99_extra_costs): This.
26411 * config/aarch64/aarch64-tune.md: Regenerate.
26412 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
26413 (vulcan_addrcost_table): This.
26414 (vulcan_regmove_cost): Rename to ...
26415 (thunderx2t99_regmove_cost): This.
26416 (vulcan_vector_cost): Rename to ...
26417 (thunderx2t99_vector_cost): this.
26418 (vulcan_branch_cost): Rename to ...
26419 (thunderx2t99_branch_cost): This.
26420 (vulcan_tunings): Rename to ...
26421 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
26422 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
26424 2017-01-09 Martin Jambor <mjambor@suse.cz>
26428 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26429 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26430 (propagate_vr_accross_jump_function): Use the above function for all
26431 value range computations for pass-through jump functions and type
26432 converasion from explicit value range values.
26433 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26434 parameters from TYPE_ARG_TYPES.
26435 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26436 (ipa_write_node_info): Stream type of the actual argument.
26437 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26439 2017-01-09 Martin Liska <mliska@suse.cz>
26442 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26443 (lookup_compiler): Do not show error message with have_E.
26445 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26447 PR tree-optimization/78938
26448 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26449 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26450 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26451 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26454 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26456 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26459 2017-01-09 Richard Biener <rguenther@suse.de>
26461 PR tree-optimization/78997
26462 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26463 name condition properly.
26465 2017-01-09 Richard Biener <rguenther@suse.de>
26468 * dwarf2out.c (is_cxx): New overload with context.
26469 (is_naming_typedef_decl): Use it.
26471 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26473 * invoke.texi (Option Summary): Correct spacing in option lists
26474 and add line breaks to fix over-long lines.
26476 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26478 PR middle-end/17660
26480 * extend.texi (Common Variable Attributes): Add xref to GCC
26481 Internals manual to explain mode attribute keywords.
26483 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26486 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26487 and Preprocessor Options.
26488 (Options for Linking): Document -pthread here....
26489 (RS/6000 and PowerPC Options): ...not here.
26490 (Solaris 2 Options): ...or here.
26491 * doc/cppopts.texi: Document -pthread.
26493 2017-01-08 Martin Sebor <msebor@redhat.com>
26495 PR middle-end/77708
26496 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26497 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26498 New member functions.
26499 (format_directive): Used them.
26501 (pass_sprintf_length::handle_gimple_call): Same.
26502 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26503 to avoid truncation for any argument.
26504 (extract_affine_mul): Same.
26505 * tree.c (get_file_function_name): Same.
26507 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26509 PR middle-end/77484
26510 * predict.def (PRED_INDIR_CALL): Set to 86.
26512 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26514 PR preprocessor/54124
26515 * doc/cppopts.texi: Reformat -d subtable to list the full name
26516 of the options. Add cross-reference to the docs for the general
26517 compiler -d options.
26518 * doc/invoke.texi (Developer Options): Add cross-reference to the
26519 preprocessor-specific -d option documentation.
26521 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26523 PR preprocessor/13498
26524 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26525 redudant material, and reflect new command-line options.
26526 (System Headers): Likewise.
26528 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26530 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26531 -isystem, and -idirafter. Copy-edit.
26532 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26533 default for -ftrack-macro-expansion. Delete obsolete and
26534 badly-formatted implementation details about -fdebug-cpp output.
26535 * doc/cppwarnopts.texi: Copy-edit.
26537 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26540 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26541 that the transition from a max line width >= 1<<10 to narrower
26542 lines works correctly.
26544 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26546 * doc/options.texi (PerFunction): New.
26547 * opt-functions.awk (switch_flags): Map both Optimization and
26548 PerFunction to CL_OPTIMIZATION.
26549 * opth-gen.awk: Test for PerFunction flag along with
26551 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26552 it only when the latter is present. Skip those that don't in
26553 the hash function generator.
26554 * common.opt (fvar-tracking): Mark as PerFunction instead of
26556 (fvar-tracking-assignments): Likewise.
26557 (fvar-tracking-assignments-toggle): Likewise.
26558 (fvar-tracking-uninit): Likewise.
26560 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26562 PR translation/79018
26563 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26566 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26569 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26570 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26571 TARGET_LONG_JUMP_TABLE_OFFSETS.
26572 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26573 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26574 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26575 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26576 * config/m68k/m68k.md (tablejump expander): Likewise.
26577 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26578 TARGET_LONG_JUMP_TABLE_OFFSETS.
26579 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26580 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26582 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26583 David Holsgrove <david.holsgrove@xilinx.com>
26585 * common/config/microblaze/microblaze-common.c
26586 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26587 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26589 * config/microblaze/microblaze.c (microblaze_must_save_register)
26590 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26592 (microblaze_eh_return): New function.
26593 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26594 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26595 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26596 * config/microblaze/microblaze.md (eh_return): New pattern.
26598 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26600 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26601 GCC_DIAGNOSTIC_STRINGIFY): Define.
26603 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26605 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26607 * config/arm/arm.md (<mcrr>): New.
26609 * config/arm/arm.c (arm_arch5te): New.
26610 (arm_option_override): Set arm_arch5te.
26611 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26613 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26614 (arm_mcrr_qualifiers): ... this. New.
26615 (MRRC_QUALIFIERS): Define to...
26616 (arm_mrrc_qualifiers): ... this. New.
26617 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26619 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26620 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26621 (MRRCI, mrrc, MRRC): New.
26622 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26623 VUNSPEC_MRRC2): New.
26625 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26627 * config/arm/arm.md (<mcr>): New.
26629 * config/arm/arm.c (arm_coproc_builtin_available): Add
26630 support for mcr, mrc, mcr2 and mrc2.
26631 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26632 (arm_mcr_qualifiers): ... this. New.
26633 (MRC_QUALIFIERS): Define to ...
26634 (arm_mrc_qualifiers): ... this. New.
26635 (MCR_QUALIFIERS): Define to ...
26636 (arm_mcr_qualifiers): ... this. New.
26637 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26639 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26640 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26641 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26642 VUNSPEC_MRC2): New.
26644 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26646 * config/arm/arm.md (*ldc): New.
26650 * config/arm/arm.c (arm_coproc_builtin_available): Add
26651 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26652 (arm_coproc_ldc_stc_legitimate_address): New.
26653 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26654 'qualifier_const_pointer'.
26655 (LDC_QUALIFIERS): Define to...
26656 (arm_ldc_qualifiers): ... this. New.
26657 (STC_QUALIFIERS): Define to...
26658 (arm_stc_qualifiers): ... this. New.
26659 * config/arm/arm-protos.h
26660 (arm_coproc_ldc_stc_legitimate_address): New.
26661 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26662 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26663 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26664 stc2, stcl, stc2l): New.
26665 * config/arm/constraints.md (Uz): New.
26666 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26667 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26668 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
26669 VUNSPEC_STC2L): New.
26671 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26673 * config/arm/arm.md (<cdp>): New.
26674 * config/arm/arm.c (neon_const_bounds): Rename this ...
26675 (arm_const_bounds): ... this.
26676 (arm_coproc_builtin_available): New.
26677 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
26678 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
26679 (CDP_QUALIFIERS): Define to...
26680 (arm_cdp_qualifiers): ... this. New.
26682 (arm_expand_builtin_args): Add case for 6 arguments.
26683 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26684 (arm_const_bounds): ... this.
26685 (arm_coproc_builtin_available): New.
26686 * config/arm/arm_acle.h (__arm_cdp): New.
26688 * config/arm/arm_acle_builtins.def (cdp): New.
26690 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26691 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26692 'arm_const_bounds'.
26693 * config/arm/types.md (coproc): New.
26694 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26695 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26696 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26697 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26699 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26701 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26702 (UBINOP_QUALIFIERS): New.
26704 (acle_builtin_data): New. Change comment.
26705 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26706 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26707 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26708 arm_acle_builtins.def.
26709 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26710 (arm_init_acle_builtins): New.
26711 (CRC32_BUILTIN): Remove.
26712 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26713 crc32cb, crc32ch and crc32cw.
26714 (arm_init_crc32_builtins): Remove.
26715 (arm_init_builtins): Use arm_init_acle_builtins rather
26716 than arm_init_crc32_builtins.
26717 (arm_expand_acle_builtin): New.
26718 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26719 * config/arm/arm_acle_builtins.def: New.
26721 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26723 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26724 (arm_builtin_datum): ... this.
26725 (arm_init_neon_builtin): Rename to ...
26726 (arm_init_builtin): ... this. Add a new parameters PREFIX
26727 and USE_SIG_IN_NAME.
26728 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26729 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26730 'arm_builtin_datum'.
26731 (arm_init_vfp_builtins): Likewise.
26732 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26733 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26734 (arm_expand_neon_args): Rename to ...
26735 (arm_expand_builtin_args): ... this. Rename builtin_arg
26736 enum values and differentiate between ARG_BUILTIN_MEMORY
26737 and ARG_BUILTIN_NEON_MEMORY.
26738 (arm_expand_neon_builtin_1): Rename to ...
26739 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26740 values, arm_expand_builtin_args and add bool parameter NEON.
26741 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26742 (arm_expand_vfp_builtin): Likewise.
26743 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26745 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26747 PR middle-end/77484
26748 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26749 * predict.c (tree_estimate_probability_bb): Reverse direction of
26750 polymorphic call predictor.
26752 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26754 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26755 (determine_pass_name_match): ...this new function and...
26756 (should_skip_pass_p): ...this new function.
26758 2017-01-06 Nathan Sidwell <nathan@acm.org>
26760 * ipa-visibility.c (function_and_variable_visibility): Reformat
26761 comments and long lines. Remove extrneous if.
26762 * symtab.c (symtab_node::make_decl_local): Fix code format.
26763 (symtab_node::set_section_for_node): Fix comment typo.
26765 2017-01-06 Martin Liska <mliska@suse.cz>
26768 * lra-constraints.c: Rename invariant to lra_invariant.
26769 * predict.c (set_even_probabilities): Initialize e to NULL.
26771 2017-01-05 Martin Sebor <msebor@redhat.com>
26773 PR tree-optimization/78910
26774 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
26775 (format_integer): Correct off-by-one error in the handling
26776 of precision with negative numbers in signed conversions..
26778 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
26780 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
26782 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26784 PR tree-optimization/71016
26785 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
26786 factor_out_conditional_conversion. Formatting fix.
26787 (factor_out_conditional_conversion): Add cond_stmt argument.
26788 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
26789 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
26792 2017-01-05 David Malcolm <dmalcolm@redhat.com>
26794 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
26795 read-rtl-function.o, and selftest-rtl.o.
26796 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
26797 (selftest::aarch64_test_loading_full_dump): New function.
26798 (selftest::aarch64_run_selftests): New function.
26799 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
26800 selftest::aarch64_run_selftests.
26801 * config/i386/i386.c
26802 (selftest::ix86_test_loading_dump_fragment_1): New function.
26803 (selftest::ix86_test_loading_call_insn): New function.
26804 (selftest::ix86_test_loading_full_dump): New function.
26805 (selftest::ix86_test_loading_unspec): New function.
26806 (selftest::ix86_run_selftests): Call the new functions.
26807 * emit-rtl.c (maybe_set_max_label_num): New function.
26808 * emit-rtl.h (maybe_set_max_label_num): New decl.
26809 * function.c (instantiate_decls): Guard call to
26810 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
26811 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
26813 * gensupport.c (gen_reader::gen_reader): Pass "false"
26814 for new "compact" param of rtx_reader.
26815 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
26816 rather than an empty string for NULL strings.
26817 * read-md.c: Potentially include config.h rather than bconfig.h.
26818 Wrap include of errors.h with #ifdef GENERATOR_FILE.
26819 (have_error): New global, copied from errors.c.
26820 (md_reader::read_name): Rename to...
26821 (md_reader::read_name_1): ...this, adding "out_loc" param,
26822 and converting "missing name or number" to returning false, rather
26824 (md_reader::read_name): Reimplement in terms of read_name_1.
26825 (md_reader::read_name_or_nil): New function.
26826 (md_reader::read_string): Handle "(nil)" by returning NULL.
26827 (md_reader::md_reader): Add new param "compact".
26828 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
26829 (md_reader::read_file): New method.
26830 * read-md.h (md_reader::md_reader): Add new param "compact".
26831 (md_reader::read_file): New method.
26832 (md_reader::is_compact): New accessor.
26833 (md_reader::read_name): Convert return type from void to file_location.
26834 (md_reader::read_name_or_nil): New decl.
26835 (md_reader::read_name_1): New decl.
26836 (md_reader::m_compact): New field.
26837 (noop_reader::noop_reader): Pass "false" for new "compact" param
26839 (rtx_reader::rtx_reader): Add new "compact" param.
26840 (rtx_reader::read_rtx_operand): Make virtual and convert return
26841 type from void to rtx.
26842 (rtx_reader::read_until): New decl.
26843 (rtx_reader::handle_any_trailing_information): New virtual function.
26844 (rtx_reader::postprocess): New virtual function.
26845 (rtx_reader::finalize_string): New virtual function.
26846 (rtx_reader::m_in_call_function_usage): New field.
26847 (rtx_reader::m_reuse_rtx_by_id): New field.
26848 * read-rtl-function.c: New file.
26849 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
26850 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
26851 (selftest::verify_three_block_rtl_cfg): New decl.
26852 * read-rtl-function.h: New file.
26853 * read-rtl.c: Potentially include config.h rather than bconfig.h.
26854 For host, include function.h, memmodel.h, and emit-rtl.h.
26855 (one_time_initialization): New function.
26856 (struct compact_insn_name): New struct.
26857 (compact_insn_names): New array.
26858 (find_code): Handle insn codes in compact dumps.
26859 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
26860 (bind_subst_iter_and_attr): Likewise.
26861 (add_condition_to_string): Likewise.
26862 (add_condition_to_rtx): Likewise.
26863 (apply_attribute_uses): Likewise.
26864 (add_current_iterators): Likewise.
26865 (apply_iterators): Likewise.
26866 (initialize_iterators): Guard usage of apply_subst_iterator with
26867 #ifdef GENERATOR_FILE.
26868 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
26869 (md_reader::read_mapping): Likewise.
26870 (add_define_attr_for_define_subst): Likewise.
26871 (add_define_subst_attr): Likewise.
26872 (read_subst_mapping): Likewise.
26873 (check_code_iterator): Likewise.
26874 (rtx_reader::read_rtx): Likewise. Move one-time initialization
26876 (one_time_initialization): New function.
26877 (rtx_reader::read_until): New method.
26878 (read_flags): New function.
26879 (parse_reg_note_name): New function.
26880 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
26881 Handle reuse_rtx ids.
26882 Wrap iterator lookup within #ifdef GENERATOR_FILE.
26883 Add parsing support for RTL dumps, mirroring the special-cases in
26884 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
26885 values, and calling handle_any_trailing_information.
26886 (rtx_reader::read_rtx_operand): Convert return type from void
26887 to rtx, returning return_rtx. Handle case 'e'. Call
26888 finalize_string on XSTR and XTMPL fields.
26889 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
26890 "(nil)" values were omitted. Call the postprocess vfunc on the
26892 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
26893 class ctor. Initialize m_in_call_function_usage. Call
26894 one_time_initialization.
26895 * rtl-tests.c (selftest::test_uncond_jump): Call
26896 set_new_first_and_last_insn.
26897 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
26898 * selftest-rtl.c: New file.
26899 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
26900 (selftest::get_insn_by_uid): New decl.
26901 * selftest-run-tests.c (selftest::run_tests): Call
26902 read_rtl_function_c_tests.
26903 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
26904 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
26907 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
26909 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
26910 operands in a special way. Assert that pos+len <= mode precision.
26912 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26914 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
26915 3 argument Alias with unlimited for the negative form.
26916 (fno-vect-cost-model): Removed.
26918 2017-01-05 Martin Liska <mliska@suse.cz>
26920 * hsa-gen.c (gen_hsa_divmod): New function.
26921 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
26923 2017-01-05 Martin Liska <mliska@suse.cz>
26926 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
26929 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26931 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
26932 small constant length operands.
26934 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26936 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
26937 between loop iterations.
26939 2017-01-05 Martin Liska <mliska@suse.cz>
26942 * gimplify.c (gimplify_decl_expr): Compare to
26943 asan_poisoned_variables instread of checking flags.
26944 (gimplify_target_expr): Likewise.
26945 (gimplify_expr): Likewise.
26946 (gimplify_function_tree): Conditionally initialize
26947 asan_poisoned_variables.
26949 2017-01-04 Jeff Law <law@redhat.com>
26951 PR tree-optimizatin/78812
26952 * rtl.h (contains_mem_rtx_p): Prototype.
26953 * ifcvt.c (containts_mem_rtx_p): Move from here to...
26954 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
26955 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
26956 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
26957 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
26959 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26961 * input.c (assert_char_at_range): Default-initialize actual_range.
26963 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26965 * df-scan.c (df_ref_create_structure): Make regno unsigned,
26966 to match the caller.
26968 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26970 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
26971 insns after final jump in test to emit dummy move.
26973 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26975 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
26976 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
26978 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26980 * multiple_target.c (create_dispatcher_calls): Init e_next.
26981 * tree-ssa-loop-split.c (split_loop): Init border.
26982 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
26985 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
26990 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
26991 (altivec_register_operand): Do not return true if the operand
26992 contains a SUBREG mixing SImode and SFmode.
26993 (vsx_register_operand): Likewise.
26994 (vsx_reg_sfsubreg_ok): New predicate.
26995 (vfloat_operand): Do not return true if the operand contains a
26996 SUBREG mixing SImode and SFmode.
26997 (vint_operand): Likewise.
26998 (vlogical_operand): Likewise.
26999 (gpc_reg_operand): Likewise.
27000 (int_reg_operand): Likewise.
27001 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
27002 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
27003 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
27005 (rs6000_emit_move_si_sf_subreg): New helper function.
27006 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
27007 fixup SUBREGs involving SImode and SFmode.
27008 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
27009 numbers for the new peephole2 optimization.
27010 (peephole2 for SFmode unions): New peephole2 to optimize cases in
27011 the GLIBC math library that do AND/IOR/XOR operations on single
27012 precision floating point.
27013 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
27014 target macros to say whether we need to avoid SUBREGs mixing
27016 (TARGET_ALLOW_SF_SUBREG): Likewise.
27017 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
27018 (UNSPEC_SI_FROM_SF): Likewise.
27019 (iorxor): Change spacing.
27020 (and_ior_xor): New iterator for AND, IOR, and XOR.
27021 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
27022 (movdi_from_sf_zero_ext): Likewise.
27023 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
27024 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
27025 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
27026 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
27027 (fms<mode>4): Likewise.
27028 (fnma<mode>4): Likewise.
27029 (fnms<mode>4): Likewise.
27030 (nfma<mode>4): Likewise.
27031 (nfms<mode>4): Likewise.
27033 2017-01-04 Marek Polacek <polacek@redhat.com>
27036 * doc/invoke.texi: Document -Wpointer-compare.
27038 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27040 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
27043 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
27044 descriptions for -gdwarf-5 and emit them as uleb128 instead of
27047 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
27050 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
27051 documentation of the powerpc_popcntb_ok attribute.
27052 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27053 code to issue warning messages if a requested CPU configuration is
27054 not supported by the binary (assembler and loader) toolchain.
27055 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
27056 made to define a built-in function that has been disabled.
27057 (paired_init_builtins): Add assertion to prevent ICE if attempt is
27058 made to define a built-in function that has been disabled.
27059 (altivec_init_builtins): Add comment explaining why definition
27060 of the DST built-in functions is not preceded by an assertion
27061 check. Add assertions to prevent ICE if attempts are made to
27062 define an altivec predicate or an abs* built-in function that has
27064 (htm_init_builtins): Add comment explaining why definition of the
27065 htm built-in functions is not preceded by an assertion check.
27067 2017-01-04 Jeff Law <law@redhat.com>
27069 PR tree-optimizatin/67955
27070 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
27071 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
27072 the points-to solution does not include pt_null. Use DECL_PT_UID
27075 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
27077 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
27078 Use gen_int_mode instead of gen_lopwart for const_int operands.
27080 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27082 PR tree-optimization/71563
27083 * match.pd: Simplify X << Y into X if Y is known to be 0 or
27084 out of range value - has low bits known to be zero.
27086 2017-01-04 Alan Modra <amodra@gmail.com>
27088 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
27089 * configure: Regenerate.
27090 * config.in: Regenerate.
27092 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27095 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
27096 a substring of the message, but strcmp with the whole message. Ifdef
27097 ENABLE_NLS, translate the message first using dgettext.
27099 2017-01-03 Jeff Law <law@redhat.com>
27101 PR tree-optimizatin/78856
27102 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
27103 (mark_threaded_blocks): Remove code to truncate thread paths that
27104 cross multiple loop headers. Instead invalidate the cached loop
27105 iteration information and handle case of a thread path walking
27106 into an irreducible region.
27108 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
27111 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
27112 assertions. Add support for doing the signbit if the IEEE 128-bit
27113 floating point value is in a GPR.
27114 * config/rs6000/rs6000.md (Fsignbit): Delete.
27115 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
27116 Update the length attribute if the value is in a GPR.
27117 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
27118 the sign or zero extension instruction, since the value is always 0/1.
27119 (signbit<mode>2_dm2): Delete using <Fsignbit>.
27122 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
27123 extracting SImode to a GPR register so that we can generate a
27124 store, limit the vector to be in a traditional Altivec register
27125 for the vextuwrx instruction.
27127 2017-01-03 Ian Lance Taylor <iant@google.com>
27129 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
27131 2017-01-03 Martin Sebor <msebor@redhat.com>
27133 PR tree-optimization/78696
27134 * gimple-ssa-sprintf.c (format_floating): Correct handling of
27135 precision. Use MPFR for %f for greater fidelity. Correct handling
27137 (pass_sprintf_length::compute_format_length): Set width and precision
27138 specified by asrerisk to void_node for vararg functions.
27139 (try_substitute_return_value): Adjust dump output.
27141 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
27143 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
27145 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
27147 * doc/invoke.texi (SPARC options): Document -mlra as the default.
27148 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
27149 -mlra/-mno-lra was passed to the compiler.
27151 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
27153 PR rtl-optimization/65618
27154 * emit-rtl.c (try_split): Move initialization of "before" and
27155 "after" to just before the call to emit_insn_after_setloc.
27157 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
27159 * doc/md.texi (Standard Names): Remove reference to Java frontend.
27161 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
27163 * dwarf2out.c (gen_enumeration_type_die): When
27164 -gno-strict-dwarf, add a DW_AT_encoding attribute.
27166 2017-01-03 Jakub Jelinek <jakub@redhat.com>
27168 PR tree-optimization/78965
27169 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
27170 Change first argument from const call_info & to call_info &. For %n
27171 set info.nowrite to false.
27173 PR middle-end/78901
27174 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
27175 possibly throwing calls.
27177 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
27178 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
27179 and fns handling, rather than in a separate case SSA_NAME.
27181 2017-01-02 Jeff Law <law@redhat.com>
27183 * config/darwin-driver.c (darwin_driver_init): Const-correctness
27184 fixes for first_period and second_period variables.
27186 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
27189 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
27190 (*insvqi_1): New insn pattern.
27191 (*insvqi_1_mem_rex64): Ditto.
27192 (*insvqi_2): Ditto.
27193 (*insvqi_3): Rename from *insvqi.
27195 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
27197 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
27199 * doc/cfg.texi (Edges): Remove reference to Java.
27200 (Maintaining the CFG): Ditto.
27202 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27204 PR middle-end/77674
27205 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
27206 transparent aliases.
27208 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27210 PR middle-end/77484
27211 * predict.def (PRED_CALL): Update hitrate.
27212 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
27213 * predict.c (tree_estimate_probability_bb): Split CALL predictor
27214 into direct/indirect/polymorphic variants.
27216 2017-01-01 Jakub Jelinek <jakub@redhat.com>
27218 Update copyright years.
27220 * gcc.c (process_command): Update copyright notice dates.
27221 * gcov-dump.c (print_version): Ditto.
27222 * gcov.c (print_version): Ditto.
27223 * gcov-tool.c (print_version): Ditto.
27224 * gengtype.c (create_file): Ditto.
27225 * doc/cpp.texi: Bump @copying's copyright year.
27226 * doc/cppinternals.texi: Ditto.
27227 * doc/gcc.texi: Ditto.
27228 * doc/gccint.texi: Ditto.
27229 * doc/gcov.texi: Ditto.
27230 * doc/install.texi: Ditto.
27231 * doc/invoke.texi: Ditto.
27233 Copyright (C) 2017 Free Software Foundation, Inc.
27235 Copying and distribution of this file, with or without modification,
27236 are permitted in any medium without royalty provided the copyright
27237 notice and this notice are preserved.