1 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
5 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
12 switching to a new text section.
14 2017-09-18 Richard Biener <rguenther@suse.de>
16 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
18 (build_alias_set): Reject aliases with no access function.
20 2017-09-18 Richard Biener <rguenther@suse.de>
22 PR tree-optimization/79622
23 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
25 (build_cross_bb_scalars_use): Likewise.
27 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
29 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
31 2017-09-18 Alan Modra <amodra@gmail.com>
34 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
35 stack_pointer_rtx for count 0. Update comments. Break up
38 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
40 config/i386/i386.c: (xlogue_layout::STUB_NAME_MAX_LEN): Increase to 20
42 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
43 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
44 or avx version of the stub.
46 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
49 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
50 compute the minimum stack alignment. Also update preferred stack
51 boundary for leaf functions.
53 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
55 PR tree-optimization/82228
56 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
59 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
61 * common/config/nds32/nds32-common.c
62 (nds32_option_optimization_table): Refine formatting.
63 (nds32_option_optimization_table): Use -fsched-pressure and
64 -fomit-frame-pointer for specific optimization level.
66 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
68 * config/nds32/nds32.c: Refine formatting and comments.
69 * config/nds32/nds32.h: Likewise.
70 * config/nds32/nds32.md: Likewise.
71 * config/nds32/nds32-cost.c: Likewise.
72 * config/nds32/nds32-isr.c: Likewise.
73 * config/nds32/nds32-md-auxiliary.c: Likewise.
74 * config/nds32/nds32-multiple.md: Likewise.
75 * config/nds32/nds32-predicates.c: Likewise.
77 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
78 Jakub Jelinek <jakub@redhat.com>
80 Add support for -std=c++2a.
81 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
83 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
85 2017-09-15 Steve Ellcey <sellcey@cavium.com>
88 * doc/extend.texi (Common Function Attributes): Add
89 references to ARM, AArch64, and S/390 specific attributes.
90 (Function Specific Option Pragmas): Add AArch64 and S/390
91 to list of back ends that support the target pragma.
93 2017-09-15 Nathan Sidwell <nathan@acm.org>
95 * doc/standards.texi: Fix C++17 description. Update URLs for
98 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
100 * common.opt (Wcast-align=strict): New warning option.
101 * doc/invoke.texi: Document -Wcast-align=strict.
103 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
105 * cgraph.h (cgraph_thunk_info): Add comments.
106 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
107 assert for VIRTUAL_* arguments stricter.
109 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
111 PR tree-optimization/71026
112 * match.pd: Move RDIV patterns from fold-const.c
113 * fold-const.c (distribute_real_division): Removed.
114 (fold_binary_loc): Remove calls to distribute_real_divison.
116 2017-09-15 Jakub Jelinek <jakub@redhat.com>
118 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
119 c++1z and gnu++1z as deprecated. Change other references to
120 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
121 Change -Wc++1z-compat to -Wc++17-compat.
122 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
123 * dwarf2out.c (highest_c_language): Handle C++17.
124 (gen_compile_unit_die): Likewise.
126 2017-09-15 Jakub Jelinek <jakub@redhat.com>
128 PR rtl-optimization/82192
129 * combine.c (make_extraction): Don't look through non-paradoxical
130 SUBREGs or TRUNCATE if pos + len is or might be bigger than
133 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
134 Alan Hayward <alan.hayward@arm.com>
135 David Sherwood <david.sherwood@arm.com>
137 * target.def (function_arg_offset): New hook.
138 * targhooks.h (default_function_arg_offset): Declare.
139 * targhooks.c (default_function_arg_offset): New function.
140 * function.c (locate_and_pad_parm): Use
141 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
142 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
143 (TARGET_FUNCTION_ARG_OFFSET): ...this.
144 * doc/tm.texi: Regenerate.
145 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
146 * config/spu/spu.c (spu_function_arg_offset): New function.
147 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
148 * system.h (FUNCTION_ARG_OFFSET): Poison.
150 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
151 Alan Hayard <alan.hayward@arm.com>
152 David Sherwood <david.sherwood@arm.com>
154 * target.def (truly_noop_truncation): New hook.
155 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
156 than TRULY_NOOP_TRUNCATION.
157 * hooks.h (hook_bool_uint_uint_true): Declare.
158 * hooks.c (hook_bool_uint_uint_true): New function.
159 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
160 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
161 * doc/tm.texi: Regenerate.
162 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
163 rather than TRULY_NOOP_TRUNCATION in comments.
164 (simplify_comparison): Likewise.
165 (record_truncated_value): Likewise.
166 * expmed.c (extract_bit_field_1): Likewise.
167 (extract_split_bit_field): Likewise.
168 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
169 instead of TRULY_NOOP_TRUNCATION.
170 * function.c (assign_parm_setup_block): Likewise.
171 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
172 * rtlhooks.c: Include target.h.
173 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
174 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
175 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
176 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
177 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
178 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
179 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
180 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
181 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
182 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
183 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
184 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
185 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
186 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
187 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
188 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
189 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
190 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
191 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
192 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
193 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
194 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
195 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
196 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
197 * config/mips/mips.c (mips_truly_noop_truncation): New function.
198 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
199 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
200 rather than TRULY_NOOP_TRUNCATION in comments.
201 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
202 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
203 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
204 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
205 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
206 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
207 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
208 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
209 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
210 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
211 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
212 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
213 rather than TRULY_NOOP_TRUNCATION in comments.
214 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
215 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
216 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
217 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
218 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
219 TRULY_NOOP_TRUNCATION condition.
220 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
221 (TRULY_NOOP_TRUNCATION): Delete.
222 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
223 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
224 * config/spu/spu.c (spu_truly_noop_truncation): New function.
225 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
226 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
227 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
228 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
229 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
230 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
231 rather than TRULY_NOOP_TRUNCATION in comments.
232 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
233 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
234 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
235 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
236 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
237 * system.h (TRULY_NOOP_TRUNCATION): Poison.
239 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
242 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
243 (*cmp_ior): Likewise.
244 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
245 (*ior_scc_scc_cmp): Likewise.
246 (*and_scc_scc): Likewise.
247 (*and_scc_scc_cmp): Likewise.
249 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
250 Alan Hayard <alan.hayward@arm.com>
251 David Sherwood <david.sherwood@arm.com>
253 * target.def (can_change_mode_class): New hook.
254 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
255 (hard_regno_nregs): Likewise.
256 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
257 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
258 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
259 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
260 (LOAD_EXTEND_OP): Update accordingly.
261 * doc/tm.texi: Regenerate.
262 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
263 CANNOT_CHANGE_MODE_CLASS.
264 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
265 (REG_CAN_CHANGE_MODE_P): ...this new macro.
266 * combine.c (simplify_set): Update accordingly.
267 * emit-rtl.c (validate_subreg): Likewise.
268 * recog.c (general_operand): Likewise.
269 * regcprop.c (mode_change_ok): Likewise.
270 * reload1.c (choose_reload_regs): Likewise.
271 (inherit_piecemeal_p): Likewise.
272 * rtlanal.c (simplify_subreg_regno): Likewise.
273 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
274 instead of CANNOT_CHANGE_MODE_CLASS.
275 (reload_cse_simplify_operands): Likewise.
276 * reload.c (push_reload): Use targetm.can_change_mode_class
277 instead of CANNOT_CHANGE_MODE_CLASS.
278 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
279 REG_CANNOT_CHANGE_MODE_P.
280 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
281 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
282 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
283 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
284 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
285 (arm_can_change_mode_class): New function.
286 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
287 than CANNOT_CHANGE_MODE_CLASS in comments.
288 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
289 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
290 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
291 (ix86_can_change_mode_class): ...this new function, inverting the
292 sense of the return value.
293 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
294 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
295 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
296 (ia64_can_change_mode_class): New function.
297 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
298 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
299 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
300 (m32c_can_change_mode_class): ...this new function, inverting the
301 sense of the return value.
302 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
303 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
304 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
305 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
306 (mips_can_change_mode_class): ...this new function, inverting the
307 sense of the return value.
308 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
309 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
310 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
311 (msp430_can_change_mode_class): New function.
312 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
313 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
314 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
315 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
316 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
317 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
318 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
319 (pa_cannot_change_mode_class): Replace with...
320 (pa_can_change_mode_class): ...this new function, inverting the
321 sense of the return value.
322 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
323 than CANNOT_CHANGE_MODE_CLASS in comments.
324 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
325 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
326 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
327 (pdp11_cannot_change_mode_class): Replace with...
328 (pdp11_can_change_mode_class): ...this new function, inverting the
329 sense of the return value.
330 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
331 * config/powerpcspe/powerpcspe-protos.h
332 (rs6000_cannot_change_mode_class_ptr): Delete.
333 * config/powerpcspe/powerpcspe.c
334 (rs6000_cannot_change_mode_class_ptr): Delete.
335 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
336 (rs6000_option_override_internal): Assign to
337 targetm.can_change_mode_class instead of
338 rs6000_cannot_change_mode_class_ptr.
339 (rs6000_cannot_change_mode_class): Replace with...
340 (rs6000_can_change_mode_class): ...this new function, inverting the
341 sense of the return value.
342 (rs6000_debug_cannot_change_mode_class): Replace with...
343 (rs6000_debug_can_change_mode_class): ...this new function.
344 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
345 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
346 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
347 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
348 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
350 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
351 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
352 (rs6000_option_override_internal): Assign to
353 targetm.can_change_mode_class instead of
354 rs6000_cannot_change_mode_class_ptr.
355 (rs6000_cannot_change_mode_class): Replace with...
356 (rs6000_can_change_mode_class): ...this new function, inverting the
357 sense of the return value.
358 (rs6000_debug_cannot_change_mode_class): Replace with...
359 (rs6000_debug_can_change_mode_class): ...this new function.
360 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
361 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
362 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
363 (s390_can_change_mode_class): ...this new function, inverting the
364 sense of the return value.
365 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
366 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
367 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
368 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
369 (sh_cannot_change_mode_class): Replace with...
370 (sh_can_change_mode_class): ...this new function, inverting the
371 sense of the return value.
372 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
373 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
374 (sparc_can_change_mode_class): New function.
375 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
376 * config/spu/spu.c (spu_can_change_mode_class): New function.
377 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
378 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
379 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
380 (visium_can_change_mode_class): New function.
381 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
383 2017-09-15 Richard Biener <rguenther@suse.de>
385 PR tree-optimization/82217
386 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
387 but not undefined case.
389 2017-09-15 Jakub Jelinek <jakub@redhat.com>
392 * postreload.c (reload_cse_simplify_operands): Skip
393 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
395 2017-09-15 Richard Biener <rguenther@suse.de>
397 PR tree-optimization/68823
398 * graphite-scop-detection.c (build_alias_set): If we have a
399 possible dependence check whether we can handle them by just
400 looking at the DRs DR_ACCESS_FNs.
401 (build_scops): If build_alias_set fails, fail the SCOP.
403 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
405 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
406 to support float128 built-in functions that require the ISA 3.0
408 (BU_FLOAT128_3_HW): Likewise.
409 (SQRTF128): Add support for the IEEE 128-bit square root and fma
413 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
414 support for built-in functions that need the ISA 3.0 IEEE 128-bit
415 floating point instructions.
416 (rs6000_invalid_builtin): Likewise.
417 (rs6000_builtin_mask_names): Likewise.
418 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
419 (RS6000_BTM_FLOAT128_HW): Likewise.
420 (RS6000_BTM_COMMON): Likewise.
421 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
423 * doc/extend.texi (RS/6000 built-in functions): Document the
424 IEEE 128-bit floating point square root and fused multiply-add
427 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
429 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
430 reg (r2) isn't in the set of registers defined in the prologue.
432 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
433 Alan Hayward <alan.hayward@arm.com>
434 David Sherwood <david.sherwood@arm.com>
436 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
437 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
438 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
439 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
440 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
442 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
443 max_vectorization_factor.
444 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
446 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
447 Alan Hayward <alan.hayward@arm.com>
448 David Sherwood <david.sherwood@arm.com>
450 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
451 (vect_worthwhile_without_simd_p): Declare.
452 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
453 (vectorizable_reduction): Use it.
454 * tree-vect-stmts.c (vectorizable_shift): Likewise.
455 (vectorizable_operation): Likewise.
457 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
458 Alan Hayward <alan.hayward@arm.com>
459 David Sherwood <david.sherwood@arm.com>
461 * tree-vectorizer.h (vect_get_num_copies): New function.
462 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
463 * tree-vect-loop.c (vectorizable_reduction): Likewise.
464 (vectorizable_induction): Likewise.
465 (vectorizable_live_operation): Likewise.
466 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
467 (vectorizable_bswap): Likewise.
468 (vectorizable_call): Likewise.
469 (vectorizable_conversion): Likewise.
470 (vectorizable_assignment): Likewise.
471 (vectorizable_shift): Likewise.
472 (vectorizable_operation): Likewise.
473 (vectorizable_store): Likewise.
474 (vectorizable_load): Likewise.
475 (vectorizable_condition): Likewise.
476 (vectorizable_comparison): Likewise.
477 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
479 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
480 Alan Hayward <alan.hayward@arm.com>
481 David Sherwood <david.sherwood@arm.com>
483 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
486 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
487 Alan Hayward <alan.hayward@arm.com>
488 David Sherwood <david.sherwood@arm.com>
490 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
491 an inline wrapper that provides a location.
492 (gimple_build_vector): Likewise.
493 * gimple-fold.c (gimple_build_vector_from_val): New function.
494 (gimple_build_vector): Likewise.
495 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
496 functions to build the initial value. Always return a gimple value.
497 (get_initial_defs_for_reduction): Likewise. Only compute
499 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
500 vect_init_vector on the results from get_initial_def(s)_for_reduction.
501 (vectorizable_induction): Use gimple_build_vector rather than
504 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
505 Alan Hayward <alan.hayward@arm.com>
506 David Sherwood <david.sherwood@arm.com>
508 * target.h (vec_perm_indices): New typedef.
509 (auto_vec_perm_indices): Likewise.
510 * optabs-query.h: Include target.h
511 (can_vec_perm_p): Take a vec_perm_indices *.
512 * optabs-query.c (can_vec_perm_p): Likewise.
513 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
514 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
515 * tree-vect-generic.c (lower_vec_perm): Likewise.
516 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
517 (vect_grouped_load_supported): Likewise.
518 (vect_shift_permute_load_chain): Likewise.
519 (vect_permute_store_chain): Use auto_vec_perm_indices.
520 (vect_permute_load_chain): Likewise.
521 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
522 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
523 Update uses of can_vec_perm_p.
524 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
525 mode with a number of elements. Take a vec_perm_indices *.
526 (vect_create_epilog_for_reduction): Update accordingly.
527 Use auto_vec_perm_indices.
528 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
529 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
530 (vect_transform_slp_perm_load): Likewise.
531 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
532 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
533 (vect_gen_perm_mask_checked): Likewise.
534 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
535 (vect_gen_perm_mask_checked): Likewise.
536 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
537 (vectorizable_store): Likewise.
538 (vectorizable_load): Likewise.
539 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
540 (vectorizable_bswap): Likewise.
542 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
543 Alan Hayward <alan.hayward@arm.com>
544 David Sherwood <david.sherwood@arm.com>
546 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
547 * tree.c (build_vector): Likewise.
548 (build_vector_from_ctor): Update accordingly.
549 (build_vector_from_val): Likewise.
550 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
551 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
552 * tree-vect-generic.c (add_rshift): Likewise.
553 (expand_vector_divmod): Likewise.
554 (optimize_vector_constructor): Likewise.
555 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
556 (vect_transform_slp_perm_load): Likewise.
557 (vect_schedule_slp_instance): Likewise.
558 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
559 (vectorizable_call): Likewise.
560 (vect_gen_perm_mask_any): Likewise. Add elements in order.
561 * expmed.c (make_tree): Likewise.
562 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
563 a vector passed to build_vector.
564 (fold_convert_const): Likewise.
565 (exact_inverse): Likewise.
566 (fold_ternary_loc): Likewise.
567 (fold_relational_const): Likewise.
568 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
569 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
570 (const_unop): Likewise. Store the reduction accumulator in a
571 variable rather than an array.
572 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
573 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
574 the new vector, rather than constructing it after the input arrays.
575 (native_interpret_vector): Use auto_vec<tree> when building
576 a vector passed to build_vector. Add elements in order.
577 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
578 auto_vec<tree> when building a vector passed to build_vector.
579 (vect_create_epilog_for_reduction): Likewise.
580 (vectorizable_induction): Likewise.
581 (get_initial_def_for_reduction): Likewise. Fix indentation of
583 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
585 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
586 passed to build_vector.
588 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
589 Alan Hayward <alan.hayward@arm.com>
590 David Sherwood <david.sherwood@arm.com>
592 * tree-core.h (tree_base::u): Add an "nelts" field.
593 (tree_vector): Use VECTOR_CST_NELTS as the length.
594 * tree.c (tree_size): Likewise.
595 (make_vector): Initialize VECTOR_CST_NELTS.
596 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
597 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
598 TYPE_VECTOR_SUBPARTS.
599 * expr.c (const_vector_mask_from_tree): Consistently use "units"
600 as the number of units, setting it from VECTOR_CST_NELTS.
601 (const_vector_from_tree): Likewise.
602 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
603 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
604 (fold_negate_expr_1): Likewise.
605 (fold_convert_const): Likewise.
606 (const_binop): Likewise. Differentiate the number of output and
608 (const_unop): Likewise.
609 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
610 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
611 in cases that did the opposite.
613 2017-09-14 Richard Biener <rguenther@suse.de>
615 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
618 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
620 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
622 2017-09-14 Jakub Jelinek <jakub@redhat.com>
625 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
626 if and where to split a bb, except for splitting before debug insn
627 sequences followed by non-label real insn. Delete debug insns
628 in between basic blocks.
630 * combine.c (make_compound_operation_int): Formatting fixes.
632 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
633 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
634 * config/netbsd.h (LINK_EH_SPEC): Likewise.
635 * config/sol2.h (LINK_EH_SPEC): Likewise.
636 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
637 * config/s390/linux.h (LINK_SPEC): Likewise.
638 * config/freebsd.h (LINK_EH_SPEC): Likewise.
639 * config/openbsd.h (LINK_EH_SPEC): Likewise.
640 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
641 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
642 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
643 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
644 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
645 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
647 2017-09-13 Jakub Jelinek <jakub@redhat.com>
649 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
651 (ENDFILE_LINUX_SPEC): Likewise.
652 (LINK_EH_SPEC): Likewise.
653 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
654 (LINK_OS_LINUX_SPEC32): Likewise.
655 (LINK_OS_LINUX_SPEC64): Likewise.
656 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
657 (LINK_OS_LINUX_SPEC): Likewise.
659 2017-09-13 Martin Liska <mliska@suse.cz>
662 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
663 CASE_HIGH is NULL_TREE.
665 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
666 Alan Hayward <alan.hayward@arm.com>
667 David Sherwood <david.sherwood@arm.com>
669 * target.def (secondary_memory_needed): New hook.
670 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
671 instead of SECONDARY_MEMORY_NEEDED.
672 (secondary_memory_needed_mode): Likewise.
673 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
674 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
675 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
676 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
677 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
678 * doc/tm.texi: Regenerate.
679 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
680 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
681 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
682 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
683 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
684 * config/i386/i386.c (inline_secondary_memory_needed): Put the
685 mode argument first and change the reg_class arguments to reg_class_t.
686 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
687 Make static. Update the call to inline_secondary_memory_needed.
688 (ix86_register_move_cost): Update the call to
689 inline_secondary_memory_needed.
690 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
691 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
693 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
694 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
696 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
697 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
698 * config/mips/mips.c (mips_secondary_memory_needed): Make static
699 and match hook interface. Add comment from mips.h.
700 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
701 * config/mmix/mmix.md (truncdfsf2): Refer to
702 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
704 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
705 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
706 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
707 (pa_secondary_memory_needed): New function.
708 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
709 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
710 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
711 (pdp11_secondary_memory_needed): Make static and match hook interface.
712 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
713 * config/powerpcspe/powerpcspe-protos.h
714 (rs6000_secondary_memory_needed_ptr): Delete.
715 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
717 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
718 (rs6000_option_override_internal): Assign to
719 targetm.secondary_memory_needed rather than
720 rs6000_secondary_memory_needed_ptr.
721 (rs6000_secondary_memory_needed): Match hook interface.
722 (rs6000_debug_secondary_memory_needed): Likewise.
723 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
724 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
725 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
726 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
727 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
728 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
730 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
731 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
732 (rs6000_option_override_internal): Assign to
733 targetm.secondary_memory_needed rather than
734 rs6000_secondary_memory_needed_ptr.
735 (rs6000_secondary_memory_needed): Match hook interface.
736 (rs6000_debug_secondary_memory_needed): Likewise.
737 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
738 * config/s390/s390.c (s390_secondary_memory_needed): New function.
739 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
740 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
741 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
742 (sparc_secondary_memory_needed): New function.
743 * lra-constraints.c (check_and_process_move): Refer to
744 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
746 (curr_insn_transform): Likewise.
747 (process_alt_operands): Use targetm.secondary_memory_needed
748 instead of TARGET_SECONDARY_MEMORY_NEEDED.
749 (check_secondary_memory_needed_p): Likewise.
750 (choose_split_class): Likewise.
751 * reload.c: Unconditionally include code that was previously
752 conditional on SECONDARY_MEMORY_NEEDED.
753 (push_secondary_reload): Use targetm.secondary_memory_needed
754 instead of TARGET_SECONDARY_MEMORY_NEEDED.
755 (push_reload): Likewise.
756 * reload1.c: Unconditionally include code that was previously
757 conditional on SECONDARY_MEMORY_NEEDED.
758 (choose_reload_regs): Use targetm.secondary_memory_needed
759 instead of TARGET_SECONDARY_MEMORY_NEEDED.
760 (gen_reload): Likewise.
761 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
763 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
764 Alan Hayward <alan.hayward@arm.com>
765 David Sherwood <david.sherwood@arm.com>
767 * target.def (secondary_memory_needed_mode): New hook:
768 * targhooks.c (default_secondary_memory_needed_mode): Declare.
769 * targhooks.h (default_secondary_memory_needed_mode): New function.
770 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
771 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
772 * doc/tm.texi: Regenerate.
773 * lra-constraints.c (check_and_process_move): Use
774 targetm.secondary_memory_needed_mode instead of
775 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
776 (curr_insn_transform): Likewise.
777 * reload.c (get_secondary_mem): Likewise.
778 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
779 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
781 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
782 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
783 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
784 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
785 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
787 * config/powerpcspe/powerpcspe-protos.h
788 (rs6000_secondary_memory_needed_mode): Delete.
789 * config/powerpcspe/powerpcspe.c
790 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
791 (rs6000_secondary_memory_needed_mode): Make static.
792 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
793 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
795 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
797 (rs6000_secondary_memory_needed_mode): Make static.
798 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
799 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
800 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
801 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
802 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
804 (sparc_secondary_memory_needed_mode): New function.
805 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
807 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
809 * config/aarch64/constraints.md (Umq): New constraint.
810 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
812 (mov<mode>): Update condition.
814 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
816 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
817 when bitposition is the same.
819 2017-09-13 Richard Biener <rguenther@suse.de>
821 * dwarf2out.c (output_die_symbol): Remove.
822 (output_die): Do not output a DIEs symbol.
824 2017-09-13 Richard Biener <rguenther@suse.de>
827 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
828 default-def to avoid breaking iterator update with the weird
829 interaction with cgraph_update_edges_for_call_stmt_node.
831 2017-09-13 Richard Biener <rguenther@suse.de>
833 * tree-cfg.c (verify_gimple_assign_binary): Add verification
834 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
835 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
836 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
838 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
840 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
841 Disable pc relative literal load irrespective of
842 TARGET_FIX_ERR_A53_84341 for default.
844 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
846 * config/sparc/sparc.c (output_return): Output the source location of
847 the insn in the delay slot, if any.
848 (output_sibcall): Likewise.
850 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
853 * common.opt (-static-pie): New alias.
854 (shared): Negate static-pie.
855 (-no-pie): Update help text.
857 (static-pie): New option.
858 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
860 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
861 (LINK_EH_SPEC): Likewise.
862 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
863 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
864 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
865 * gcc.c (LINK_COMMAND_SPEC): Likewise.
866 (init_gcc_specs): Likewise.
867 (init_spec): Likewise.
868 (display_help): Update help message for -pie.
869 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
872 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
874 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
875 (movdi_aarch64): Likewise.
876 (movti_aarch64): Likewise.
878 2017-09-12 Simon Wright <simon@pushface.org>
881 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
882 calculation of the minor version, always output as 0.
884 2017-09-12 Jakub Jelinek <jakub@redhat.com>
887 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
888 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
889 on it early, rather than manual conversion late. For
890 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
891 instead of performing manual conversion.
893 2017-09-12 Carl Love <cel@us.ibm.com>
895 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
896 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
898 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
899 VMULOSW): Add definitions.
900 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
901 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
902 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
903 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
904 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
906 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
908 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
910 (movti_aarch64): Likewise.
911 (movdf_aarch64): Likewise.
912 (movtf_aarch64): Likewise.
913 (load_pairdi): Likewise.
914 (store_pairdi): Likewise.
915 (load_pairdf): Likewise.
916 (store_pairdf): Likewise.
917 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
918 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
919 (ldr_got_small_<mode>): Likewise.
920 (ldr_got_small_28k_<mode>): Likewise.
921 (ldr_got_tiny): Likewise.
922 * config/aarch64/iterators.md (ldst_sz): New.
923 (ldpstp_sz): Likewise.
924 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
926 (thunderx_load): Split load_8 to load_16.
927 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
929 (thunderx2t99_storepair_basic): Split store_8 to store_16.
930 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
931 (xgene1_store_pair): Split store_8 to store_16.
932 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
933 (falkor_st_0_st_sd): Split store_8 to store_16.
935 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
937 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
938 and store1/2/3/4 to store_4/8/12/16.
939 * config/aarch64/aarch64.md: Update for rename.
940 * config/arm/arm.md: Likewise.: Likewise.
941 * config/arm/arm.c: Likewise.
942 * config/arm/thumb1.md: Likewise.
943 * config/arm/thumb2.md: Likewise.
944 * config/arm/vfp.md: Likewise.
945 * config/arm/arm-generic.md: Likewise.
946 * config/arm/arm1020e.md: Likewise.
947 * config/arm/arm1026ejs.md: Likewise.
948 * config/arm/arm1136jfs.md: Likewise.
949 * config/arm/arm926ejs.md: Likewise.
950 * config/arm/cortex-a15.md: Likewise.
951 * config/arm/cortex-a17.md: Likewise.
952 * config/arm/cortex-a5.md: Likewise.
953 * config/arm/cortex-a53.md: Likewise.
954 * config/arm/cortex-a57.md: Likewise.
955 * config/arm/cortex-a7.md: Likewise.
956 * config/arm/cortex-a8.md: Likewise.
957 * config/arm/cortex-a9.md: Likewise.
958 * config/arm/cortex-m4.md: Likewise.
959 * config/arm/cortex-m7.md: Likewise.
960 * config/arm/cortex-r4.md: Likewise.
961 * config/arm/exynos-m1.md: Likewise.
962 * config/arm/fa526.md: Likewise.
963 * config/arm/fa606te.md: Likewise.
964 * config/arm/fa626te.md: Likewise.
965 * config/arm/fa726te.md: Likewise.
966 * config/arm/fmp626.md: Likewise.
967 * config/arm/iwmmxt.md: Likewise.
968 * config/arm/ldmstm.md: Likewise.
969 * config/arm/marvell-pj4.md: Likewise.
970 * config/arm/xgene1.md: Likewise.
971 * config/aarch64/thunderx.md: Likewise.
972 * config/aarch64/thunderx2t99.md: Likewise.
973 * config/aarch64/falkor.md: Likewise.
975 2017-09-12 Martin Liska <mliska@suse.cz>
977 * attribs.c (private_lookup_attribute): New function.
978 * attribs.h (private_lookup_attribute): Declared here.
979 (lookup_attribute): Called from this place.
981 2017-09-12 Richard Biener <rguenther@suse.de>
983 PR tree-optimization/82157
984 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
985 stmts with side-effects.
987 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
988 Alan Hayward <alan.hayward@arm.com>
989 David Sherwood <david.sherwood@arm.com>
991 * target.def (hard_regno_nregs): New hook.
992 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
993 * targhooks.h (default_hard_regno_nregs): Declare.
994 * targhooks.c (default_hard_regno_nregs): New function.
995 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
996 (TARGET_HARD_REGNO_NREGS): ...this hook.
997 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
998 (CLASS_MAX_NREGS): Likewise.
999 * doc/tm.texi: Regenerate.
1000 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
1001 instead of HARD_REGNO_NREGS.
1002 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
1003 HARD_REGNO_NREGS in the comment.
1004 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
1005 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
1006 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
1007 Return an unsigned int.
1008 (TARGET_HARD_REGNO_NREGS): Redefine.
1009 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
1010 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
1011 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1012 (arc_hard_regno_nregs): New function.
1013 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
1014 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
1015 (arm_hard_regno_nregs): New function.
1016 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
1017 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
1018 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
1019 (TARGET_HARD_REGNO_NREGS): Redefine.
1020 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
1021 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
1022 (HARD_REGNO_NREGS): Delete.
1023 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
1024 (cr16_hard_regno_nregs): New function.
1025 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
1026 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
1027 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
1028 (cris_hard_regno_nregs): New function.
1029 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
1030 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
1031 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1032 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
1033 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
1034 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
1035 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
1036 (frv_hard_regno_nregs): Make static. Take and return an
1038 (frv_class_max_nregs): Remove outdated copy of documentation.
1039 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
1040 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
1041 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
1042 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
1043 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
1044 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
1045 (TARGET_HARD_REGNO_NREGS): Redefine.
1046 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
1047 (CLASS_MAX_NREGS): Update comment.
1048 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
1049 (ia64_hard_regno_nregs): New function.
1050 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
1051 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
1052 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
1053 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
1054 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
1056 (m32c_hard_regno_nregs): Likewise. Make static.
1057 (TARGET_HARD_REGNO_NREGS): Redefine.
1058 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
1059 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
1060 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
1061 (m68k_hard_regno_nregs): New function.
1062 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
1063 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
1064 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
1065 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
1066 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
1067 Take and return an unsigned int.
1068 (TARGET_HARD_REGNO_NREGS): Redefine.
1069 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
1070 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1071 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
1072 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
1073 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
1074 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
1075 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
1076 (msp430_hard_regno_nregs): Make static. Take and return an
1078 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
1079 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
1080 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
1081 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
1082 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
1083 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
1084 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
1085 (TARGET_HARD_REGNO_NREGS): Redefine.
1086 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
1087 (PA_HARD_REGNO_NREGS): ...this.
1088 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
1089 (PA_HARD_REGNO_NREGS): ...this.
1090 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1091 (pa_hard_regno_nregs): New function.
1092 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
1093 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
1094 (pdp11_hard_regno_nregs): New function.
1095 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
1096 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
1097 (rs6000_hard_regno_nregs_hook): New function.
1098 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
1099 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1100 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
1101 Take and return an unsigned int. Move earlier in file.
1102 (TARGET_HARD_REGNO_NREGS): Redefine.
1103 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
1104 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
1105 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
1106 (rl78_hard_regno_nregs): Make static. Take and return an
1108 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
1109 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
1110 (rs6000_hard_regno_nregs_hook): New function.
1111 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
1112 * config/rx/rx.c (rx_hard_regno_nregs): New function.
1113 (TARGET_HARD_REGNO_NREGS): Redefine.
1114 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
1115 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
1116 instead of HARD_REGNO_NREGS.
1117 (s390_hard_regno_nregs): New function.
1118 (s390_hard_regno_mode_ok): Add comment from s390.h.
1119 (TARGET_HARD_REGNO_NREGS): Redefine.
1120 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
1121 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
1122 (sh_hard_regno_nregs): New function.
1123 (sh_pass_in_reg_p): Use it.
1124 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
1125 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1126 (sparc_hard_regno_nregs): New function.
1127 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
1128 * config/spu/spu.c (spu_hard_regno_nregs): New function.
1129 (spu_function_arg_advance): Use it, supplying a valid register number.
1130 (TARGET_HARD_REGNO_NREGS): Redefine.
1131 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
1132 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
1133 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
1134 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
1135 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
1136 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
1137 (CLASS_MAX_NREGS): Remove copy of old documentation.
1138 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
1139 (visium_hard_regno_nregs): New function.
1140 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
1141 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
1142 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1143 xtensa_hard_regno_nregs): New function.
1144 * system.h (HARD_REGNO_NREGS): Poison.
1146 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1148 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
1149 hard_regno_nregs instead of HARD_REGNO_NREGS.
1150 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
1151 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
1152 (c6x_expand_epilogue): Likewise.
1153 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
1154 (frv_read_iacc_argument): Likewise.
1155 * config/sh/sh.c: Include regs.h.
1156 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1157 (regs_used): Likewise.
1158 (output_stack_adjust): Likewise.
1159 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
1160 * expmed.c: Include regs.h.
1161 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1162 * ree.c: Include regs.h.
1163 (combine_reaching_defs): Use hard_regno_nregs instead of
1165 (add_removable_extension): Likewise.
1167 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1169 * regs.h (hard_regno_nregs): Turn into a function.
1170 (end_hard_regno): Update accordingly.
1171 * caller-save.c (setup_save_areas): Likewise.
1172 (save_call_clobbered_regs): Likewise.
1173 (replace_reg_with_saved_mem): Likewise.
1174 (insert_restore): Likewise.
1175 (insert_save): Likewise.
1176 * combine.c (can_change_dest_mode): Likewise.
1177 (move_deaths): Likewise.
1178 (distribute_notes): Likewise.
1179 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
1180 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
1181 (rs6000_split_multireg_move): Likewise.
1182 (rs6000_register_move_cost): Likewise.
1183 (rs6000_memory_move_cost): Likewise.
1184 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
1185 (rs6000_split_multireg_move): Likewise.
1186 (rs6000_register_move_cost): Likewise.
1187 (rs6000_memory_move_cost): Likewise.
1188 * cselib.c (cselib_reset_table): Likewise.
1189 (cselib_lookup_1): Likewise.
1190 * emit-rtl.c (set_mode_and_regno): Likewise.
1191 * function.c (aggregate_value_p): Likewise.
1192 * ira-color.c (setup_profitable_hard_regs): Likewise.
1193 (check_hard_reg_p): Likewise.
1194 (calculate_saved_nregs): Likewise.
1195 (assign_hard_reg): Likewise.
1196 (improve_allocation): Likewise.
1197 (calculate_spill_cost): Likewise.
1198 * ira-emit.c (modify_move_list): Likewise.
1199 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
1200 (ira_hard_reg_in_set_p): Likewise.
1201 * ira.c (setup_reg_mode_hard_regset): Likewise.
1202 (clarify_prohibited_class_mode_regs): Likewise.
1203 (check_allocation): Likewise.
1204 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1205 (lra_setup_reg_renumber): Likewise.
1206 (setup_try_hard_regno_pseudos): Likewise.
1207 (spill_for): Likewise.
1208 (assign_hard_regno): Likewise.
1209 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1210 * lra-constraints.c (in_class_p): Likewise.
1211 (lra_constraint_offset): Likewise.
1212 (simplify_operand_subreg): Likewise.
1213 (lra_constraints): Likewise.
1214 (split_reg): Likewise.
1215 (split_if_necessary): Likewise.
1216 (invariant_p): Likewise.
1217 (inherit_in_ebb): Likewise.
1218 * lra-lives.c (process_bb_lives): Likewise.
1219 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
1220 (get_hard_regs): Likewise.
1221 (do_remat): Likewise.
1222 * lra-spills.c (assign_spill_hard_regs): Likewise.
1223 * mode-switching.c (create_pre_exit): Likewise.
1224 * postreload.c (reload_combine_recognize_pattern): Likewise.
1225 * recog.c (peep2_find_free_register): Likewise.
1226 * regcprop.c (kill_value_regno): Likewise.
1227 (set_value_regno): Likewise.
1228 (copy_value): Likewise.
1229 (maybe_mode_change): Likewise.
1230 (find_oldest_value_reg): Likewise.
1231 (copyprop_hardreg_forward_1): Likewise.
1232 * regrename.c (check_new_reg_p): Likewise.
1233 (regrename_do_replace): Likewise.
1234 * reload.c (push_reload): Likewise.
1235 (combine_reloads): Likewise.
1236 (find_dummy_reload): Likewise.
1237 (operands_match_p): Likewise.
1238 (find_reloads): Likewise.
1239 (find_equiv_reg): Likewise.
1240 (reload_adjust_reg_for_mode): Likewise.
1241 * reload1.c (count_pseudo): Likewise.
1242 (count_spilled_pseudo): Likewise.
1243 (find_reg): Likewise.
1244 (clear_reload_reg_in_use): Likewise.
1245 (free_for_value_p): Likewise.
1246 (allocate_reload_reg): Likewise.
1247 (choose_reload_regs): Likewise.
1248 (reload_adjust_reg_for_temp): Likewise.
1249 (emit_reload_insns): Likewise.
1250 (delete_output_reload): Likewise.
1251 * rtlanal.c (subreg_get_info): Likewise.
1252 * sched-deps.c (sched_analyze_reg): Likewise.
1253 * sel-sched.c (init_regs_for_mode): Likewise.
1254 (mark_unavailable_hard_regs): Likewise.
1255 (choose_best_reg_1): Likewise.
1256 (verify_target_availability): Likewise.
1257 * valtrack.c (dead_debug_insert_temp): Likewise.
1258 * var-tracking.c (track_loc_p): Likewise.
1259 (emit_note_insn_var_location): Likewise.
1260 * varasm.c (make_decl_rtl): Likewise.
1261 * reginfo.c (choose_hard_reg_mode): Likewise.
1262 (init_reg_modes_target): Refer directly to
1263 this_target_regs->x_hard_regno_nregs.
1265 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1267 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
1268 instead of hard_regno_nregs.
1270 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1272 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
1273 end_hard_regno instead of hard_regno_nregs.
1274 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
1275 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
1276 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
1277 * ira-color.c (improve_allocation): Likewise.
1278 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1279 * lra-lives.c (mark_regno_live): Likewise.
1280 (mark_regno_dead): Likewise.
1281 * lra-remat.c (operand_to_remat): Likewise.
1282 * lra.c (collect_non_operand_hard_regs): Likewise.
1283 * postreload.c (reload_combine_note_store): Likewise.
1284 (move2add_valid_value_p): Likewise.
1285 * reload.c (regno_clobbered_p): Likewise.
1287 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1289 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
1291 * config/v850/v850.c (v850_reorg): Likewise.
1292 * reload.c (refers_to_regno_for_reload_p): Likewise.
1293 (find_equiv_reg): Likewise.
1294 * reload1.c (reload_reg_reaches_end_p): Likewise.
1296 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1298 * caller-save.c (add_used_regs): Use REG_NREGS instead of
1300 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
1301 * config/arm/arm.c (output_move_neon): Likewise.
1302 (arm_attr_length_move_neon): Likewise.
1303 (neon_split_vcombine): Likewise.
1304 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
1305 (c6x_mark_reg_written): Likewise.
1306 (c6x_dwarf_register_span): Likewise.
1307 * config/i386/i386.c (ix86_save_reg): Likewise.
1308 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
1309 (rws_access_reg): Likewise.
1310 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
1311 * mode-switching.c (create_pre_exit): Likewise.
1312 * ree.c (combine_reaching_defs): Likewise.
1313 (add_removable_extension): Likewise.
1314 * regcprop.c (find_oldest_value_reg): Likewise.
1315 (copyprop_hardreg_forward_1): Likewise.
1316 * reload.c (reload_inner_reg_of_subreg): Likewise.
1317 (push_reload): Likewise.
1318 (combine_reloads): Likewise.
1319 (find_dummy_reload): Likewise.
1320 (reload_adjust_reg_for_mode): Likewise.
1321 * reload1.c (find_reload_regs): Likewise.
1322 (forget_old_reloads_1): Likewise.
1323 (reload_reg_free_for_value_p): Likewise.
1324 (reload_adjust_reg_for_temp): Likewise.
1325 (emit_reload_insns): Likewise.
1326 (delete_output_reload): Likewise.
1327 * sel-sched.c (choose_best_reg_1): Likewise.
1328 (choose_best_pseudo_reg): Likewise.
1330 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1331 Alan Hayward <alan.hayward@arm.com>
1332 David Sherwood <david.sherwood@arm.com>
1334 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
1335 * target.def (slow_unaligned_access): New hook.
1336 * targhooks.h (default_slow_unaligned_access): Declare.
1337 * targhooks.c (default_slow_unaligned_access): New function.
1338 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
1339 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
1340 * doc/tm.texi: Regenerate.
1341 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
1342 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
1343 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
1345 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
1346 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
1348 (rs6000_slow_unaligned_access): New function.
1349 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1350 (expand_block_compare): Likewise.
1351 (expand_strn_compare): Likewise.
1352 (rs6000_rtx_costs): Likewise.
1353 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
1354 (riscv_slow_unaligned_access): Likewise.
1355 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
1356 (riscv_slow_unaligned_access_p): ...this and make static.
1357 (riscv_option_override): Update accordingly.
1358 (riscv_slow_unaligned_access): New function.
1359 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1360 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
1361 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1362 (rs6000_slow_unaligned_access): New function.
1363 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1364 (rs6000_rtx_costs): Likewise.
1365 * config/rs6000/rs6000-string.c (expand_block_compare)
1366 (expand_strn_compare): Use targetm.slow_unaligned_access instead
1367 of SLOW_UNALIGNED_ACCESS.
1368 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
1369 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
1370 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
1371 of SLOW_UNALIGNED_ACCESS.
1372 * expmed.c (simple_mem_bitfield_p): Likewise.
1373 * expr.c (alignment_for_piecewise_move): Likewise.
1374 (emit_group_load_1): Likewise.
1375 (emit_group_store): Likewise.
1376 (copy_blkmode_from_reg): Likewise.
1377 (emit_push_insn): Likewise.
1378 (expand_assignment): Likewise.
1379 (store_field): Likewise.
1380 (expand_expr_real_1): Likewise.
1381 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1382 * lra-constraints.c (simplify_operand_subreg): Likewise.
1383 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1384 * gimple-ssa-store-merging.c: Likewise in block comment at start
1386 * tree-ssa-strlen.c: Include target.h.
1387 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
1388 of SLOW_UNALIGNED_ACCESS.
1389 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
1391 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1393 PR rtl-optimization/82185
1394 * expmed.c (emit_store_flag_int): Only test tem if it has been
1397 2017-09-12 Richard Biener <rguenther@suse.de>
1400 * match.pd ((FTYPE) N CMP CST): Fix typo.
1402 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1404 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1406 (mips_near_type_p): Add 'short_call' attribute as a synonym
1408 * doc/extend.texi (short_call): Document new function attribute.
1410 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1413 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1414 assertion check that in the condition.
1415 (get_atomic_generic_size): Likewise. Before testing if parameter
1416 has pointer type, if it has array type, call for C++
1417 default_conversion to perform array-to-pointer conversion.
1419 2017-09-12 Richard Biener <rguenther@suse.de>
1421 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1422 for operations we cannot scalarize.
1424 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1426 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1427 vectors heap vectors. Clean up comments.
1428 Make visited_bbs a reference.
1429 (profitable_jump_thread_path): Make GC
1430 vectors heap vectors. Clean up comments.
1432 (convert_and_register_jump_thread_path): Make GC vectors heap
1434 (check_subpath_and_update_thread_path): Same. Clean up comments.
1435 Make visited_bbs a reference.
1436 (handle_phi): Abstract common code to to
1437 register_jump_thread_path_if_profitable.
1438 Rename VAR_BB to DEF_BB.
1440 Make GC vectors heap vectors.
1441 Make visited_bbs a reference.
1442 (handle_assignment): Same.
1443 (register_jump_thread_path_if_profitable): New.
1444 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1446 Make GC vectors heap vectors. Clean up comments.
1447 Make visited_bbs a reference.
1448 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1449 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1452 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1455 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1456 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1458 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1460 Revert r251800 and r251799.
1462 2017-09-11 Martin Jambor <mjambor@suse.cz>
1465 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1466 arguments in advance.
1467 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1468 use it to find predecessor edges.
1469 (naive_outof_ssa): Collect vector of predecessors.
1471 2017-09-08 Jason Merrill <jason@redhat.com>
1473 PR c++/70029 - ICE with ref-qualifier and -flto
1474 * langhooks.h (struct lang_hooks_for_types): Add
1475 copy_lang_qualifiers.
1476 * attribs.c (build_type_attribute_qual_variant): Use it.
1477 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1479 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1480 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1482 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1485 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1486 (*mulsi3_sp64): New instruction.
1487 (mulsi3): New expander.
1489 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1491 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1493 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1495 * sancov.c: Include memmodel.h.
1497 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1500 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1503 2017-09-07 Carl Love <cel@us.ibm.com>
1505 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1506 the sldi instruction.
1508 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1510 * sancov.c: Include tm_p.h.
1512 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1515 * output.h (switch_to_other_text_partition): New declaration.
1516 * varasm.c (switch_to_other_text_partition): New function.
1517 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1518 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1519 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1520 to the other text partition before emitting LCL label and switch back
1521 after emitting the word after it.
1523 2017-09-07 Richard Biener <rguenther@suse.de>
1525 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1526 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1527 critical edges here, after loop init.
1528 (pass_data_pre): Remove PROP_no_crit_edges flags.
1529 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1530 for valueization of call args to avoid leaking VN_TOP.
1531 (visit_use): Assert we do not visit default defs.
1532 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1533 Use error_mark_node to more easily detect leaking VN_TOP.
1534 All default-defs are varying, not VN_TOP. Mark them visited.
1535 (run_scc_vn): Make code match comment.
1537 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1539 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1540 OPTION_MASK_FLOAT128_KEYWORD.
1541 (POWERPC_MASKS): Likewise.
1542 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1543 support for the -mfloat128-type option, and make -mfloat128
1544 default on PowerPC Linux systems. Define or undefine
1545 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1546 Define __float128 to be __ieee128 if IEEE 128-bit support is
1547 enabled, or undefine it.
1548 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1549 Delete defining __FLOAT128_TYPE__.
1550 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1551 -mfloat128-type option and make -mfloat128 default on PowerPC
1553 (TARGET_FLOAT128_TYPE): Likewise.
1554 (-mfloat128-type): Likewise.
1555 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1556 Delete the -mfloat128-type option and make -mfloat128 default on
1557 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1558 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1559 128-bit floating point is enabled. Change tests from using
1560 -mfloat128-type to -mfloat128.
1561 (rs6000_mangle_type): Use the correct mangling for the __float128
1562 type even if normal long double is restricted to 64-bits.
1563 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1565 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1566 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1567 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1569 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1570 documentation for -mfloat128.
1572 2017-09-06 Olivier Hainque <hainque@adacore.com>
1574 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1576 2017-09-06 Wish Wu <wishwu007@gmail.com>
1577 Jakub Jelinek <jakub@redhat.com>
1579 * asan.c (initialize_sanitizer_builtins): Add
1580 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1581 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1582 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1583 BT_FN_VOID_UINT64_PTR variables.
1584 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1585 (BT_FN_VOID_UINT16_UINT16): Likewise.
1586 (BT_FN_VOID_UINT32_UINT32): Likewise.
1587 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1588 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1589 (BT_FN_VOID_UINT64_PTR): Likewise.
1590 * common.opt (flag_sanitize_coverage): New variable.
1591 (fsanitize-coverage=trace-pc): Remove.
1592 (fsanitize-coverage=): Add.
1593 * flag-types.h (enum sanitize_coverage_code): New enum.
1594 * fold-const.c (fold_range_test): Disable non-short-circuit
1595 optimization if flag_sanitize_coverage.
1596 (fold_truth_andor): Likewise.
1597 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1598 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1599 (coverage_sanitizer_opts): New array.
1600 (get_closest_sanitizer_option): Add OPTS argument, handle also
1601 OPT_fsanitize_coverage_.
1602 (parse_sanitizer_options): Adjusted to also handle
1603 OPT_fsanitize_coverage_.
1604 (common_handle_option): Add OPT_fsanitize_coverage_.
1605 * sancov.c (instrument_comparison, instrument_switch): New function.
1606 (sancov_pass): Add trace-cmp support.
1607 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1608 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1609 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1610 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1611 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1612 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1613 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1614 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1615 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1616 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1618 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1620 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1621 error. Only quit immediately if parsing is complete.
1622 (BEGIN): Initialize fatal_err and parse_done.
1623 (begin fpu, end fpu): Check number of arguments.
1624 (begin arch, end arch): Likewise.
1625 (begin cpu, end cpu): Likewise.
1626 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1627 (optalias): Likewise.
1629 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1631 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1632 * config/arm/arm-isa.h: Delete. Move definitions to ...
1633 * arm-cpus.in: ... here. Use new feature and fgroup values.
1634 * config/arm/arm.c (arm_option_override): Use lower case for feature
1636 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1637 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1638 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1640 (print_isa_bits_for): New function.
1641 (gen_isa): New function.
1642 (gen_comm_data): Use print_isa_bits_for.
1643 (define feature): New keyword.
1644 (define fgroup): New keyword.
1645 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
1646 (arm-isa.h): Add rule to generate file.
1647 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1648 case for feature bit names.
1650 2017-09-06 Richard Biener <rguenther@suse.de>
1652 * tree-ssa-pre.c (NECESSARY): Remove.
1653 (create_expression_by_pieces): Do not touch pass-local flags.
1654 (insert_into_preds_of_block): Likewise.
1655 (do_pre_regular_insertion): Likewise.
1656 (eliminate_insert): Likewise.
1657 (eliminate_dom_walker::before_dom_children): Likewise.
1658 (fini_eliminate): Do not look at inserted_exprs.
1659 (mark_operand_necessary): Remove.
1660 (remove_dead_inserted_code): Replace with simple work-list
1661 algorithm based on inserted_exprs and SSA uses.
1662 (pass_pre::execute): Re-order fini_eliminate and
1663 remove_dead_inserted_code.
1665 2017-09-06 Olivier Hainque <hainque@adacore.com>
1667 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1668 for VxWorks 7. Adjust surrounding comments.
1670 2017-09-06 Richard Biener <rguenther@suse.de>
1672 * gimple-ssa-strength-reduction.c
1673 (find_candidates_dom_walker::before_dom_children): Also allow
1676 2017-09-06 Richard Biener <rguenther@suse.de>
1678 PR tree-optimization/82108
1679 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1680 for gap in the non-permutation SLP case.
1682 2017-09-06 Martin Jambor <mjambor@suse.cz>
1684 PR tree-optimization/82078
1685 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1686 add_access_to_work_queue...
1687 (build_accesses_from_assign): ...here.
1688 (propagate_all_subaccesses): Make sure racc is the group
1689 representative, if there is one.
1691 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1694 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1697 2017-09-06 Richard Biener <rguenther@suse.de>
1699 * gimple-ssa-strength-reduction.c
1700 (find_candidates_dom_walker::before_doom_children): Use a
1701 type and not a mode check.
1703 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1706 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1708 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1710 2017-09-05 Jeff Law <law@redhat.com>
1712 PR tree-optimization/64910
1713 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1714 cases where we have 3 or more operands.
1716 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1719 * omp-low.c (lower_omp_for): Recompute tree invariant if
1720 gimple_omp_for_initial/final is ADDR_EXPR.
1723 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1724 into gimple val before gimplification fo the COND_EXPR.
1726 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1728 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1729 REGION_COPY argument.
1730 (thread_through_all_blocks): Remove unused argument to
1731 duplicate_thread_path.
1733 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1734 Alan Hayward <alan.hayward@arm.com>
1735 David Sherwood <david.sherwood@arm.com>
1737 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1738 Take a scalar_mode rather than a machine_mode.
1739 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1740 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1741 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1742 (aarch64_gen_adjusted_ldpstp): Likewise.
1743 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
1745 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1746 Alan Hayward <alan.hayward@arm.com>
1747 David Sherwood <david.sherwood@arm.com>
1749 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
1750 Take a scalar_int_mode instead of a machine_mode.
1751 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1752 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1753 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1754 (aarch64_simd_attr_length_rglist): Delete.
1755 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
1756 a scalar_int_mode instead of a machine_mode.
1757 (aarch64_add_offset): Likewise.
1758 (aarch64_internal_mov_immediate): Likewise
1759 (aarch64_add_constant_internal): Likewise.
1760 (aarch64_add_constant): Likewise.
1761 (aarch64_movw_imm): Likewise.
1762 (aarch64_rtx_arith_op_extract_p): Likewise.
1763 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1764 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1765 Remove assert that the mode isn't a vector.
1766 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1767 (aarch64_expand_mov_immediate): Update calls after above changes.
1768 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
1769 (aarch64_and_bitmask_imm): Check for scalar integer modes.
1770 (aarch64_move_imm): Likewise.
1771 (aarch64_can_const_movi_rtx_p): Likewise.
1772 (aarch64_strip_extend): Likewise.
1773 (aarch64_extr_rtx_p): Likewise.
1774 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
1775 a CONST_INT when the mode parameter is VOIDmode.
1776 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
1778 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1780 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
1781 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1782 (bitwise_type_for_mode): Update accordingly.
1784 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1786 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
1787 * stor-layout.c (mode_for_size_tree): Likewise.
1788 (mode_for_array): Update accordingly.
1789 (layout_decl): Likewise.
1790 (compute_record_mode): Likewise. Only set the mode once.
1792 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1794 * target.def (get_mask_mode): Change return type to opt_mode.
1796 * doc/tm.texi: Regenerate.
1797 * targhooks.h (default_get_mask_mode): Return an opt_mode.
1798 * targhooks.c (default_get_mask_mode): Likewise.
1799 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1800 * optabs-query.c (can_vec_mask_load_store_p): Update use of
1801 targetm.get_mask_mode.
1802 * tree.c (build_truth_vector_type): Likewise.
1804 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1806 * machmode.h (mode_for_vector): Return an opt_mode.
1807 * stor-layout.c (mode_for_vector): Likewise.
1808 (mode_for_int_vector): Update accordingly.
1809 (layout_type): Likewise.
1810 * config/i386/i386.c (emit_memmov): Likewise.
1811 (ix86_expand_set_or_movmem): Likewise.
1812 (ix86_expand_vector_init): Likewise.
1813 (ix86_get_mask_mode): Likewise.
1814 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
1816 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
1817 * expmed.c (extract_bit_field_1): Likewise.
1818 * expr.c (expand_expr_real_2): Likewise.
1819 * optabs-query.c (can_vec_perm_p): Likewise.
1820 (can_vec_mask_load_store_p): Likewise.
1821 * optabs.c (expand_vec_perm): Likewise.
1822 * targhooks.c (default_get_mask_mode): Likewise.
1823 * tree-vect-stmts.c (vectorizable_store): Likewise.
1824 (vectorizable_load): Likewise.
1825 (get_vectype_for_scalar_type_and_size): Likewise.
1827 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1829 * machmode.h (mode_for_int_vector): New function.
1830 * stor-layout.c (mode_for_int_vector): Likewise.
1831 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
1832 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
1833 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1834 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1835 (s390_expand_vcond): Likewise.
1837 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1839 * machmode.h (opt_machine_mode): New type.
1840 (opt_mode<T>): Allow construction from anything that can be
1842 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
1843 (mode_for_size): Return an opt_machine_mode.
1844 * stor-layout.c (mode_for_size): Likewise.
1845 (mode_for_size_tree): Update call accordingly.
1846 (bitwise_mode_for_mode): Likewise.
1847 (make_fract_type): Likewise.
1848 (make_accum_type): Likewise.
1849 * caller-save.c (replace_reg_with_saved_mem): Update call
1851 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1852 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1853 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1854 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1855 * expmed.c (extract_bit_field_1): Likewise.
1856 * reload.c (get_secondary_mem): Likewise.
1857 * varasm.c (assemble_integer): Likewise.
1858 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
1861 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1863 * machmode.h (decimal_float_mode_for_size): New function.
1864 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
1865 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
1866 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
1867 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
1868 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
1869 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
1871 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1873 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
1874 (get_builtin_sync_mode): Likewise.
1875 (expand_ifn_atomic_compare_exchange): Likewise.
1876 (expand_builtin_atomic_clear): Likewise.
1877 (expand_builtin_atomic_test_and_set): Likewise.
1878 (fold_builtin_atomic_always_lock_free): Likewise.
1879 * calls.c (compute_argument_addresses): Likewise.
1880 (emit_library_call_value_1): Likewise.
1881 (store_one_arg): Likewise.
1882 * combine.c (combine_instructions): Likewise.
1883 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
1884 * config/arm/arm.c (arm_function_value): Likewise.
1885 (aapcs_allocate_return_reg): Likewise.
1886 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
1887 * config/i386/i386.c (construct_container): Likewise.
1888 (ix86_gimplify_va_arg): Likewise.
1889 (ix86_expand_sse_cmp): Likewise.
1890 (emit_memmov): Likewise.
1891 (emit_memset): Likewise.
1892 (expand_small_movmem_or_setmem): Likewise.
1893 (ix86_expand_pextr): Likewise.
1894 (ix86_expand_pinsr): Likewise.
1895 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
1896 * config/microblaze/microblaze.c (microblaze_block_move_straight):
1898 * config/mips/mips.c (mips_function_value_1) Likewise.
1899 (mips_block_move_straight): Likewise.
1900 (mips_expand_ins_as_unaligned_store): Likewise.
1901 * config/powerpcspe/powerpcspe.c
1902 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1903 (rs6000_darwin64_record_arg_flush): Likewise.
1904 * config/rs6000/rs6000.c
1905 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1906 (rs6000_darwin64_record_arg_flush): Likewise.
1907 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
1908 (sparc_function_value_1): Likewise.
1909 * config/spu/spu.c (adjust_operand): Likewise.
1910 (spu_emit_branch_or_set): Likewise.
1911 (arith_immediate_p): Likewise.
1912 * emit-rtl.c (gen_lowpart_common): Likewise.
1913 * expr.c (expand_expr_real_1): Likewise.
1914 * function.c (assign_parm_setup_block): Likewise.
1915 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1916 * reload1.c (alter_reg): Likewise.
1917 * stor-layout.c (mode_for_vector): Likewise.
1918 (layout_type): Likewise.
1920 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1922 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
1923 (spu_convert_move): Likewise.
1924 * lower-subreg.c (resolve_simple_move): Likewise.
1926 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1929 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
1930 define_insn to a define_expand.
1931 (altivec_vsum2sws_direct): New define_insn.
1932 (altivec_vsumsws): Convert from a define_insn to a define_expand.
1934 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1936 * config/arm/arm.c (arm_option_params_internal): Improve setting of
1939 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
1944 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
1945 realign stack if stack alignment needed is less than incoming
1948 2017-09-05 Marek Polacek <polacek@redhat.com>
1951 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
1954 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1956 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
1958 2017-09-05 Richard Biener <rguenther@suse.de>
1960 PR tree-optimization/82084
1961 * fold-const.c (can_native_encode_string_p): Handle wide characters.
1963 2017-09-05 Richard Biener <rguenther@suse.de>
1965 PR tree-optimization/82102
1966 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
1968 2017-09-05 Martin Liska <mliska@suse.cz>
1970 PR tree-optimization/82032
1971 * tree-cfg.c (generate_range_test): New function.
1972 * tree-cfg.h (generate_range_test): Declared here.
1973 * tree-cfgcleanup.c (convert_single_case_switch): New function.
1974 (cleanup_control_expr_graph): Use it.
1975 * tree-switch-conversion.c (try_switch_expansion): Remove
1977 (emit_case_nodes): Use generate_range_test.
1979 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1982 * config/i386/i386.md (*<btsc><mode>_mask): Add
1983 TARGET_USE_BT to insn constraint.
1984 (*btr<mode>_mask): Ditto.
1986 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1988 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1989 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1991 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1994 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1995 TARGET_NEON and TARGET_IWMMXT.
1996 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1997 TARGET_NEON and TARGET_IWMMXT.
1998 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
2000 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
2002 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
2003 (ix86_rewrite_tls_address): Ditto.
2004 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
2005 (ix86_rewrite_tls_address_1): Ditto.
2006 (ix86_rewrite_tls_address): Ditto.
2007 * config/i386/predicates.md (tls_address_pattern): New predicate.
2008 * config/i386/i386.md (TLS address splitter): New splitter.
2010 2017-09-04 Richard Biener <rguenther@suse.de>
2012 PR tree-optimization/82084
2013 * fold-const.h (can_native_encode_string_p): Declare.
2014 * fold-const.c (can_native_encode_string_p): Factor out from ...
2015 (native_encode_string): ... here.
2016 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
2017 vectorizing stores from constants we later cannot handle.
2019 2017-09-04 Marek Polacek <polacek@redhat.com>
2022 * doc/invoke.texi: Update -Wtautological-compare documentation.
2024 2017-09-04 Jeff Law <law@redhat.com>
2026 PR tree-optimization/64910
2027 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
2028 swap the first and last operand if the last is a constant.
2030 2017-09-04 Marek Polacek <polacek@redhat.com>
2033 * convert.c (do_narrow): When sanitizing signed integer overflows,
2034 bail out for signed types.
2035 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
2037 2017-09-04 Richard Biener <rguenther@suse.de>
2039 PR tree-optimization/82060
2040 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
2041 Move devirtualization after stmt folding and before EH/AB/noreturn
2042 cleanup to get the stmt refs canonicalized. Use a bool instead
2043 of gimple_modified_p since that doesn't work for NOPs. Schedule
2044 NOPs generated by folding for removal.
2046 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2047 Alan Hayward <alan.hayward@arm.com>
2048 David Sherwood <david.sherwood@arm.com>
2050 * coretypes.h (pad_direction): New enum.
2051 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
2052 (FUNCTION_ARG_PADDING): Likewise.
2053 * target.def (function_arg_padding): New hook.
2054 * targhooks.h (default_function_arg_padding): Declare.
2055 * targhooks.c (default_function_arg_padding): New function.
2056 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
2057 (TARGET_FUNCTION_ARG_PADDING): ...this.
2058 * doc/tm.texi: Regenerate.
2059 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
2060 instead of direction.
2061 (compute_argument_addresses): Likewise.
2062 (load_register_parameters): Likewise.
2063 (emit_library_call_value_1): Likewise.
2064 (store_one_arg): Use targetm.calls.function_arg_padding instead
2065 of FUNCTION_ARG_PADDING.
2066 (must_pass_in_stack_var_size_or_pad): Likewise.
2067 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
2068 (emit_group_store): Likewise.
2069 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
2070 instead of FUNCTION_ARG_PADDING.
2071 (emit_push_insn): Likewise, and propagate enum change throughout
2073 * function.h (direction): Delete.
2074 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
2076 * function.c (assign_parm_find_stack_rtl): Likewise.
2077 (assign_parm_setup_block_p): Likewise.
2078 (assign_parm_setup_block): Likewise.
2079 (gimplify_parameters): Likewise.
2080 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
2081 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
2083 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
2084 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2085 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
2086 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
2087 (aarch64_function_arg_padding): ...this new function.
2088 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
2089 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2090 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
2091 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2092 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
2093 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2094 (arm_pad_arg_upward): Replace with...
2095 (arm_function_arg_padding): ...this new function.
2096 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
2098 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
2099 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
2100 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2101 (ia64_hpux_function_arg_padding): Replace with...
2102 (ia64_function_arg_padding): ...this new function. Use pad_direction
2103 instead of direction. Check for TARGET_HPUX.
2104 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
2105 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2106 (iq2000_function_arg_padding): New function.
2107 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
2108 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
2109 (mips_function_arg_padding): ...this new function.
2110 (mips_pad_reg_upward): Update accordingly.
2111 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2112 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
2113 targetm.calls.function_arg_padding.
2114 (FUNCTION_ARG_PADDING): Delete.
2115 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2116 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
2117 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2118 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
2119 (nios2_block_reg_padding): Return pad_direction instead of direction.
2120 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
2121 instead of direction.
2122 (nios2_function_arg_padding): Likewise. Make static.
2123 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2124 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
2125 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
2126 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
2127 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2128 (pa_function_arg_padding): Make static. Return pad_direction instead
2130 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
2131 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2132 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
2133 instead of direction. Use targetm.calls.function_arg_padding.
2134 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
2135 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2136 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
2137 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
2138 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
2140 (function_arg_padding): Rename to...
2141 (rs6000_function_arg_padding): ...this. Make static. Return
2142 pad_direction instead of direction.
2143 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2144 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
2145 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2146 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
2147 instead of direction. Use targetm.calls.function_arg_padding.
2148 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
2149 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2150 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
2151 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
2152 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2153 (function_arg_padding): Rename to...
2154 (rs6000_function_arg_padding): ...this. Make static. Return
2155 pad_direction instead of direction.
2156 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2157 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
2158 * config/s390/s390.c (s390_function_arg_padding): New function.
2159 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2160 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
2161 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
2162 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2163 (function_arg_padding): Rename to...
2164 (sparc_function_arg_padding): ...this. Make static. Return
2165 pad_direction instead of direction.
2166 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
2167 * config/spu/spu.c (spu_function_arg_padding): New function.
2168 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2169 * system.h (FUNCTION_ARG_PADDING): Poison.
2171 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2172 Alan Hayward <alan.hayward@arm.com>
2173 David Sherwood <david.sherwood@arm.com>
2175 * target.def (modes_tieable_p): New hook.
2176 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
2177 (TARGET_MODES_TIEABLE_P): ...this.
2178 * doc/tm.texi.in: Regenerate.
2179 * hooks.h (hook_bool_mode_mode_true): Declare.
2180 * hooks.c (hook_bool_mode_mode_true): New function.
2181 * combine.c (subst): Use targetm.modes_tieable_p instead of
2183 * dse.c (find_shift_sequence): Likewise.
2184 * expmed.c (extract_low_bits): Likewise.
2185 * lower-subreg.c: Include target.h.
2186 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
2188 * rtlanal.c (rtx_cost): Likewise.
2189 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
2190 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
2191 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
2192 (TARGET_MODES_TIEABLE_P): Redefine.
2193 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
2194 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
2195 (TARGET_MODES_TIEABLE_P): Redefine.
2196 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
2197 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
2198 (arc_modes_tieable_p): New function.
2199 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
2200 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
2201 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
2202 (arm_modes_tieable_p): Make static.
2203 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
2204 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
2205 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
2206 (TARGET_MODES_TIEABLE_P): Redefine.
2207 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
2208 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
2209 (TARGET_MODES_TIEABLE_P): Redefine.
2210 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
2211 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
2212 (cr16_modes_tieable_p): New function.
2213 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
2214 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
2215 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
2216 (TRULY_NOOP_TRUNCATION): Update comment.
2217 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
2218 (TRULY_NOOP_TRUNCATION): Update comment.
2219 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
2220 (frv_modes_tieable_p): New function.
2221 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
2222 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
2223 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
2224 (TARGET_MODES_TIEABLE_P): Redefine.
2225 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
2226 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
2227 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
2228 (TARGET_MODES_TIEABLE_P): Redefine.
2229 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
2230 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
2231 (ia64_modes_tieable_p): New function.
2232 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
2233 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
2234 (iq2000_modes_tieable_p): New function.
2235 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
2236 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
2237 (lm32_modes_tieable_p): New function.
2238 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
2239 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
2240 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
2241 (TARGET_MODES_TIEABLE_P): Redefine.
2242 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
2243 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
2244 (m32r_modes_tieable_p): New function.
2245 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
2246 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
2247 (m68k_modes_tieable_p): New function.
2248 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
2249 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
2250 (mcore_modes_tieable_p): New function.
2251 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
2252 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
2254 (TARGET_MODES_TIEABLE_P): Redefine.
2255 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
2256 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
2257 * config/mips/mips.c (mips_modes_tieable_p): Make static.
2258 (TARGET_MODES_TIEABLE_P): Redefine.
2259 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
2260 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
2261 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
2262 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
2263 (mn10300_modes_tieable_p): ...this and make static.
2264 (TARGET_MODES_TIEABLE_P): Redefine.
2265 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
2266 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
2267 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
2268 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
2269 (msp430_modes_tieable_p): Make static.
2270 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
2271 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
2272 (TARGET_MODES_TIEABLE_P): Redefine.
2273 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
2274 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
2275 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
2276 (TARGET_MODES_TIEABLE_P): Redefine.
2277 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
2278 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
2279 * config/pa/pa.c (pa_modes_tieable_p): Make static.
2280 (TARGET_MODES_TIEABLE_P): Redefine.
2281 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
2282 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
2283 (pdp11_modes_tieable_p): New function.
2284 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
2285 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
2286 (rs6000_modes_tieable_p): New function.
2287 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2288 * config/powerpcspe/powerpcspe.md: Update comment.
2289 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
2290 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
2291 (TARGET_MODES_TIEABLE_P): Redefine.
2292 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
2293 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
2294 (rl78_modes_tieable_p): New function.
2295 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
2296 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
2297 (rs6000_modes_tieable_p): New function.
2298 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2299 * config/rs6000/rs6000.md: Update comment.
2300 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
2301 * config/rx/rx.c (rx_modes_tieable_p): New function.
2302 (TARGET_MODES_TIEABLE_P): Redefine.
2303 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
2304 * config/s390/s390.c (s390_modes_tieable_p): New function.
2305 (TARGET_MODES_TIEABLE_P): Redefine.
2306 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
2307 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
2308 (sh_modes_tieable_p): New function.
2309 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
2310 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
2311 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
2312 (sparc_modes_tieable_p): Make static.
2313 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
2314 * config/spu/spu.c (spu_modes_tieable_p): New function.
2315 (TARGET_MODES_TIEABLE_P): Redefine.
2316 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
2317 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
2318 (TARGET_MODES_TIEABLE_P): Redefine.
2319 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
2320 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
2321 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
2322 * config/v850/v850.c (v850_modes_tieable_p): New function.
2323 (TARGET_MODES_TIEABLE_P): Redefine.
2324 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
2325 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
2326 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
2327 (visium_modes_tieable_p): New function.
2328 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
2329 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
2330 (xtensa_modes_tieable_p): New function.
2331 * system.h (MODES_TIEABLE_P): Poison.
2333 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2334 Alan Hayward <alan.hayward@arm.com>
2335 David Sherwood <david.sherwood@arm.com>
2337 * target.def (hard_regno_mode_ok): New hook.
2338 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
2339 (TARGET_HARD_REGNO_MODE_OK): ...this.
2340 * doc/tm.texi.in: Regenerate.
2341 * hooks.h (hook_bool_uint_mode_true): Declare.
2342 * hooks.c (hook_bool_uint_mode_true): New function.
2343 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
2345 * genpreds.c (write_insn_preds_c): Add an include of target.h.
2346 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
2347 instead of HARD_REGNO_MODE_OK.
2348 * caller-save.c: Include target.h.
2349 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
2351 * combine.c (can_combine_p): Likewise.
2352 (combinable_i3pat): Likewise.
2353 (can_change_dest_mode): Likewise.
2354 * expr.c (init_expr_target): Likewise.
2355 (convert_move): Likewise.
2356 (convert_modes): Likewise.
2357 * ira.c (setup_prohibited_class_mode_regs): Likewise.
2358 (setup_prohibited_mode_move_regs): Likewise.
2359 * ira.h (target_ira): Likewise.
2360 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2361 * lra-constraints.c (process_alt_operands): Likewise.
2362 (split_reg): Likewise.
2363 * recog.c (peep2_find_free_register): Likewise.
2364 * ree.c (combine_reaching_defs): Likewise.
2365 * regcprop.c (maybe_mode_change): Likewise.
2366 * reginfo.c (init_reg_sets_1): Likewise.
2367 (choose_hard_reg_mode): Likewise.
2368 (simplifiable_subregs): Likewise.
2369 * regrename.c (check_new_reg_p): Likewise.
2370 * reload.c (find_valid_class): Likewise.
2371 (find_valid_class_1): Likewise.
2372 (reload_inner_reg_of_subreg): Likewise.
2373 (push_reload): Likewise.
2374 (combine_reloads): Likewise.
2375 (find_dummy_reload): Likewise.
2376 (find_reloads): Likewise.
2377 * reload1.c (find_reg): Likewise.
2378 (set_reload_reg): Likewise.
2379 (allocate_reload_reg): Likewise.
2380 (choose_reload_regs): Likewise.
2381 (reload_adjust_reg_for_temp): Likewise.
2382 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
2383 (simplify_subreg_regno): Likewise.
2384 * sel-sched.c (init_regs_for_mode): Likewise.
2385 * varasm.c (make_decl_rtl): Likewise.
2386 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
2387 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
2389 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
2390 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
2391 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2392 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
2393 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
2394 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2395 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
2396 (arc_mode_class): Delete.
2397 (HARD_REGNO_MODE_OK): Delete.
2398 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2399 (arc_hard_regno_mode_ok): Rename old array to...
2400 (arc_hard_regno_mode_ok_modes): ...this.
2401 (arc_conditional_register_usage): Update accordingly.
2402 (arc_mode_class): Make static.
2403 (arc_hard_regno_mode_ok): New function.
2404 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2405 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2406 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2407 (arm_hard_regno_mode_ok): Make static.
2408 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2410 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2411 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2412 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2414 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2415 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2416 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2417 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2418 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2419 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2420 * config/bfin/predicates.md (valid_reg_operand): Use
2421 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2422 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2423 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2424 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2425 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2426 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2427 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2428 (cr16_hard_regno_mode_ok): Make static and return a bool.
2429 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2430 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2431 (cris_hard_regno_mode_ok): New function.
2432 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2433 (epiphany_mode_class): Delete.
2434 (HARD_REGNO_MODE_OK): Delete.
2435 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2436 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2437 (hard_regno_mode_ok): Rename to...
2438 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2439 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2440 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2442 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2443 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2444 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2445 (frv_hard_regno_mode_ok): Make static and return a bool.
2446 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2448 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2449 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2450 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2451 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2453 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2454 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2455 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2456 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2458 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2459 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2460 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2461 (ia64_hard_regno_mode_ok): New function.
2462 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2463 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2464 (iq2000_hard_regno_mode_ok): New function.
2465 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2466 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2467 (lm32_hard_regno_mode_ok): New function.
2468 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2469 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2470 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2471 instead of HARD_REGNO_MODE_OK.
2472 (m32c_hard_regno_ok): Rename to...
2473 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2474 (m32c_cannot_change_mode_class): Update accordingly.
2475 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2476 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2477 (m32r_mode_class): Delete.
2478 (HARD_REGNO_MODE_OK): Delete.
2479 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2480 (m32r_hard_regno_mode_ok): Rename to...
2481 (m32r_hard_regno_modes): ...this.
2482 (m32r_mode_class): Make static.
2483 (m32r_hard_regno_mode_ok): New function.
2484 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2485 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2486 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2487 (m68k_hard_regno_mode_ok): Make static.
2488 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2489 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2490 (mcore_hard_regno_mode_ok): New function.
2491 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2492 (HARD_REGNO_MODE_OK): Delete.
2493 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2495 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2496 (microblaze_hard_regno_mode_ok): New function.
2497 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2498 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2499 (mips_hard_regno_mode_ok): Delete.
2500 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2501 (mips_hard_regno_mode_ok_p): ...this and make static.
2502 (mips_hard_regno_mode_ok_p): Rename to...
2503 (mips_hard_regno_mode_ok_uncached): ...this.
2504 (mips_hard_regno_mode_ok): New function.
2505 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2506 of HARD_REGNO_MODE_OK.
2507 (mips_option_override): Update after above name changes.
2508 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2509 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2510 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2511 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2512 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2513 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2514 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2515 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2516 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2517 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2518 (msp430_hard_regno_mode_ok): Make static and return a bool.
2519 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2520 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2521 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2523 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2524 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2525 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2526 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2527 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2528 (PA_HARD_REGNO_MODE_OK): ...this
2529 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2530 (PA_HARD_REGNO_MODE_OK): ...this.
2531 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2532 (pa_hard_regno_mode_ok): New function.
2533 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2534 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2535 (pdp11_hard_regno_mode_ok): New function.
2536 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2537 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2539 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2541 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2542 (rs6000_hard_regno_mode_ok): Rename to...
2543 (rs6000_hard_regno_mode_ok_uncached): ...this.
2544 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2545 (rs6000_hard_regno_mode_ok): New function.
2546 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2547 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2548 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2549 (riscv_hard_regno_mode_ok): ...this and make static.
2550 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2551 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2552 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2553 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2554 (rl78_hard_regno_mode_ok): Make static and return bool.
2555 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2556 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2558 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2559 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2560 (rs6000_hard_regno_mode_ok): Rename to...
2561 (rs6000_hard_regno_mode_ok_uncached): ...this.
2562 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2563 (rs6000_hard_regno_mode_ok): New function.
2564 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2565 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2566 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2567 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2568 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2569 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2570 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2571 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2572 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2573 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2574 (sh_hard_regno_mode_ok): Make static.
2575 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2576 instead of HARD_REGNO_MODE_OK.
2577 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2578 (sparc_mode_class): Delete.
2579 (HARD_REGNO_MODE_OK): Delete.
2580 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2581 (hard_regno_mode_classes): Make static.
2582 (sparc_mode_class): Likewise.
2583 (sparc_hard_regno_mode_ok): New function.
2584 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2585 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2586 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2588 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2589 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2590 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2591 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2592 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2593 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2594 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2595 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2596 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2597 (visium_hard_regno_mode_ok): New function.
2598 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2599 instead of HARD_REGNO_MODE_OK.
2600 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2601 (HARD_REGNO_MODE_OK): Delete.
2602 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2603 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2604 (xtensa_option_override): Update accordingly.
2605 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2606 (xtensa_hard_regno_mode_ok): New function.
2607 * system.h (HARD_REGNO_MODE_OK): Poison.
2609 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2610 Alan Hayward <alan.hayward@arm.com>
2611 David Sherwood <david.sherwood@arm.com>
2613 * target.def (hard_regno_call_part_clobbered): New hook.
2614 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2615 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2616 * doc/tm.texi: Regenerate.
2617 * hooks.h (hook_bool_uint_mode_false): Declare.
2618 * hooks.c (hook_bool_uint_mode_false): New function.
2619 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2620 * cselib.c (cselib_process_insn): Use
2621 targetm.hard_regno_call_part_clobbered instead of
2622 HARD_REGNO_CALL_PART_CLOBBERED.
2623 * ira-conflicts.c (ira_build_conflicts): Likewise.
2624 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2625 * lra-constraints.c (need_for_call_save_p): Likewise.
2626 * lra-lives.c: Include target.h.
2627 (check_pseudos_live_through_calls): Use
2628 targetm.hard_regno_call_part_clobbered instead of
2629 HARD_REGNO_CALL_PART_CLOBBERED.
2630 * regcprop.c: Include target.h.
2631 (copyprop_hardreg_forward_1): Use
2632 targetm.hard_regno_call_part_clobbered instead of
2633 HARD_REGNO_CALL_PART_CLOBBERED.
2634 * reginfo.c (choose_hard_reg_mode): Likewise.
2635 * regrename.c (check_new_reg_p): Likewise.
2636 * reload.c (find_equiv_reg): Likewise.
2637 * reload1.c (emit_reload_insns): Likewise.
2638 * sched-deps.c (deps_analyze_insn): Likewise.
2639 * sel-sched.c (init_regs_for_mode): Likewise.
2640 (mark_unavailable_hard_regs): Likewise.
2641 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2642 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2643 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
2645 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2646 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2647 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
2649 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
2651 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2652 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2653 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
2655 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2656 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2657 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2659 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2660 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2662 * config/powerpcspe/powerpcspe.c
2663 (rs6000_hard_regno_call_part_clobbered): New function.
2664 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2665 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2666 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2668 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2669 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2670 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2672 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2673 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2674 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2676 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2677 Alan Hayward <alan.hayward@arm.com>
2678 David Sherwood <david.sherwood@arm.com>
2680 * rtl.h (subreg_memory_offset): Declare.
2681 * emit-rtl.c (subreg_memory_offset): New function.
2682 * expmed.c (store_bit_field_1): Use it.
2683 * expr.c (undefined_operand_subword_p): Likewise.
2684 * simplify-rtx.c (simplify_subreg): Likewise.
2686 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2688 PR rtl-optimization/57448
2691 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2692 using atomic_load pattern.
2693 (expand_atomic_store): Likewise.
2695 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2698 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2699 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2700 handling. Use replace_call_with_value with NULL instead of
2701 gsi_replace, unlink_stmt_vdef and release_defs.
2703 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2706 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2708 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2711 * rtl.h (emit_library_call_value_1): Declare.
2712 (emit_library_call): Replace declaration with a series of overloads.
2713 Remove the parameter count argument.
2714 (emit_library_call_value): Likewise.
2715 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2716 with an "rtx_mode_t *".
2717 (emit_library_call_value): Delete.
2718 (emit_library_call): Likewise.
2719 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2720 (asan_emit_allocas_unpoison): Likewise.
2721 * builtins.c (expand_builtin_powi): Likewise.
2722 (expand_asan_emit_allocas_unpoison): Likewise.
2723 * cfgexpand.c (expand_main_function): Likewise.
2724 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2725 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2726 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2727 * config/arm/arm.c (arm_trampoline_init): Likewise.
2728 (arm_call_tls_get_addr): Likewise.
2729 (arm_expand_divmod_libfunc): Likewise.
2730 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2731 (smulsi3_highpart): Likewise.
2732 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2733 (c6x_expand_compare): Likewise.
2734 (c6x_expand_movmem): Likewise.
2735 * config/frv/frv.c (frv_trampoline_init): Likewise.
2736 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2737 (ix86_expand_divmod_libfunc): Likewise.
2738 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2739 (ia64_expand_compare): Likewise.
2740 (ia64_profile_hook): Likewise.
2741 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2742 (nonlocal_goto): Likewise.
2743 (restore_stack_nonlocal): Likewise.
2744 * config/m32r/m32r.c (block_move_call): Likewise.
2745 (m32r_trampoline_init): Likewise.
2746 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
2747 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
2748 (m68k_call_m68k_read_tp): Likewise.
2749 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
2750 (microblaze_expand_divide): Likewise.
2751 * config/mips/mips.h (mips_args): Likewise.
2752 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
2753 (MIPS_ICACHE_SYNC): Likewise.
2754 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
2755 (nios2_trampoline_init): Likewise.
2756 * config/pa/pa.c (hppa_tls_call): Likewise.
2757 (pa_trampoline_init): Likewise.
2758 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
2759 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
2760 (expand_strn_compare): Likewise.
2761 (rs6000_generate_compare): Likewise.
2762 (rs6000_expand_float128_convert): Likewise.
2763 (output_profile_hook): Likewise.
2764 (rs6000_trampoline_init): Likewise.
2765 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
2766 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
2767 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
2768 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
2769 (rs6000_generate_compare): Likewise.
2770 (rs6000_expand_float128_convert): Likewise.
2771 (output_profile_hook): Likewise.
2772 (rs6000_trampoline_init): Likewise.
2773 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
2774 * config/sh/sh.c (sh_trampoline_init): Likewise.
2775 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
2776 (sparc_emit_float_lib_cmp): Likewise.
2777 (sparc32_initialize_trampoline): Likewise.
2778 (sparc64_initialize_trampoline): Likewise.
2779 (sparc_profile_hook): Likewise.
2780 * config/spu/spu.c (ea_load_store): Likewise.
2781 * config/spu/spu.md (floatunssidf2): Likewise.
2782 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
2783 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
2784 * config/visium/visium.c (expand_block_move_4): Likewise.
2785 (expand_block_move_2): Likewise.
2786 (expand_block_move_1): Likewise.
2787 (expand_block_set_4): Likewise.
2788 (expand_block_set_2): Likewise.
2789 (expand_block_set_1): Likewise.
2790 (visium_trampoline_init): Likewise.
2791 (visium_profile_hook): Likewise.
2792 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
2793 (xtensa_setup_frame_addresses): Likewise.
2794 (xtensa_trampoline_init): Likewise.
2795 * except.c (sjlj_emit_function_enter): Likewise.
2796 (sjlj_emit_function_exit): Likewise.
2797 * explow.c (allocate_dynamic_stack_space): Likewise.
2798 (probe_stack_range): Likewise.
2799 * expr.c (convert_mode_scalar): Likewise.
2800 * optabs.c (expand_binop): Likewise.
2801 (expand_twoval_binop_libfunc): Likewise.
2802 (expand_unop): Likewise.
2803 (prepare_cmp_insn): Likewise.
2804 (prepare_float_lib_cmp): Likewise.
2805 (expand_float): Likewise.
2806 (expand_fix): Likewise.
2807 (expand_fixed_convert): Likewise.
2808 (maybe_emit_sync_lock_test_and_set): Likewise.
2809 (expand_atomic_compare_and_swap): Likewise.
2810 (expand_mem_thread_fence): Likewise.
2811 (expand_atomic_fetch_op): Likewise.
2813 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
2815 * doc/generic.texi (OpenACC): Adjust URL.
2816 * doc/invoke.texi (C Dialect Options): Ditto.
2818 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
2820 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
2821 predicate for operand 1. Add (m,<S>) constraint.
2822 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
2823 Prevent memory operand 1 with register operand 2.
2825 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
2827 PR rtl-optimization/82024
2828 * combine.c (try_combine): If the combination result is a PARALLEL,
2829 and we only need to retain the SET in there that would be placed
2830 at I2, check that we can place that at I3 instead, before doing so.
2832 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2835 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
2837 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
2838 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
2841 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
2842 Jeff Law <law@redhat.com>
2844 * varasm.c (bss_initializer_p): Do not put constants into .bss
2845 (categorize_decl_for_section): Handle bss_initializer_p returning
2846 false when DECL_INITIAL is NULL.
2848 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2851 * config/s390/s390.c (s390_can_inline_p): New function.
2853 2017-09-01 Jeff Law <law@redhat.com>
2855 PR tree-optimization/82052
2856 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
2857 Always initialize the returned slot after a hash table miss
2858 when INSERT is true.
2860 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
2862 * config/s390/s390.md (mem_signal_fence): Remove.
2863 * doc/md.texi (mem_signal_fence): Remove.
2864 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
2866 * target-insns.def (mem_signal_fence): Remove.
2868 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2871 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
2874 * asan.c (create_odr_indicator): Strip name encoding from assembler
2875 name before appending it after __odr_asan_.
2877 2017-09-01 Martin Liska <mliska@suse.cz>
2879 PR tree-optimization/82059
2880 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
2881 frequency only when an edge is redirected.
2883 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2885 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
2886 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
2887 (arc_conditional_register_usage): Remove ARC600 lp_count
2889 (arc_file_start): Emit Tag_ARC_CPU_variation.
2890 (arc_can_use_doloop_p): New conditions to use ZOLs.
2891 (hwloop_fail): New function.
2892 (hwloop_optimize): Likewise.
2893 (hwloop_pattern_reg): Likewise.
2894 (arc_doloop_hooks): New struct, to be used with reorg_loops.
2895 (arc_reorg_loops): New function, calls reorg_loops.
2896 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
2897 (arc600_corereg_hazard): Remove ZOL checking, case handled by
2899 (arc_loop_hazard): Remove function, functionality moved into
2901 (arc_hazard): Remove arc_loop_hazard call.
2902 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
2903 into hwloop_optimize.
2904 (arc_label_align): Remove ZOL handling.
2905 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
2906 * config/arc/arc.md (doloop_begin): Remove pattern.
2907 (doloop_begin_i): Likewise.
2908 (doloop_end_i): Likewise.
2909 (doloop_fallback): Likewise.
2910 (doloop_fallback_m): Likewise.
2911 (doloop_end): Reimplement expand.
2912 (arc_lp): New pattern for LP instruction.
2913 (loop_end): New pattern.
2914 (loop_fail): Likewise.
2915 (decrement_and_branch_until_zero): Likewise.
2916 * config/arc/arc.opt (mlpc-width): New option.
2917 * doc/invoke.texi (mlpc-width): Document option.
2919 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2921 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
2922 (arc_ccfsm_advance): Fix checking for delay slots.
2923 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
2925 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2927 * config/arc/arc.md (movqi_insn): Add stores to save constant long
2929 (movhi_insn): Update store instruction constraint which are saving
2930 6-bit short immediates.
2931 (movsi_insn): Consider also short scaled load operations.
2932 (zero_extendhisi2_i): Use Usd constraint instead of T.
2933 (extendhisi2_i): Add q constraint.
2934 (arc_clzsi2): Add type and length attributes.
2935 (arc_ctzsi2): Likewise.
2936 * config/arc/constraints.md (Usc): Update constraint, the
2937 assembler can parse two relocations for a single instruction.
2939 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2941 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
2942 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
2944 2017-08-31 Olivier Hainque <hainque@adacore.com>
2946 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
2947 match as powerpc-wrs-vxworks*.
2949 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
2951 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
2952 register constraint for by-element operand.
2953 (aarch64_mls_elt_merge<mode>): Likewise.
2955 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2957 * config/arc/arc.c (arc_can_follow_jump): Check for short
2960 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2962 * config.gcc: Use g.opt for arc.
2963 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
2964 functionality moved to ...
2965 (legitimate_scaled_address_p): New function, ...here.
2966 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
2967 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
2968 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
2970 (arc_override_options): Handle G option.
2971 (arc_output_pic_addr_const): Correct function definition.
2972 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
2973 (arc_decl_anon_ns_mem_p): Delete.
2974 (arc_in_small_data_p): Overhaul this function to take into
2975 consideration the value given via G option.
2976 (arc_rewrite_small_data_1): Renamed and corrected old
2977 arc_rewrite_small_data function.
2978 (arc_rewrite_small_data): New function.
2979 (small_data_pattern): Don't use pic_offset_table_rtx.
2980 * config/arc/arc.h (CC1_SPEC): Recognize G option.
2981 * config/arc/simdext.md (movmisalignv2hi): Use
2982 prepare_move_operands function.
2984 (movmisalign*): Likewise.
2985 * doc/invoke.texi (ARC options): Document -G option.
2987 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2989 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2991 * config/arc/arc.c (arc_print_operand): Output scalled address for
2992 sdata whenever is possible.
2993 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2994 load/stores are available.
2995 (compact_sda_memory_operand): Check for the alignment required by
2996 code density instructions.
2997 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2999 * config/arc/constraints.md (Usd): Update constraint.
3000 (Us0): New constraint.
3001 (Usc): Update constraint.
3003 2017-08-31 Richard Biener <rguenther@suse.de>
3006 * dwarf2out.c (dwarf2out_early_global_decl): Process each
3009 2017-08-31 Tamar Christina <tamar.christina@arm.com>
3011 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
3012 Resize type_signature.
3014 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3015 Alan Hayward <alan.hayward@arm.com>
3016 David Sherwood <david.sherwood@arm.com>
3018 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
3019 subregs whose inner modes can be stored in GPRs.
3020 (aarch64_classify_index): Likewise.
3022 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3023 Alan Hayward <alan.hayward@arm.com>
3024 David Sherwood <david.sherwood@arm.com>
3026 * config/aarch64/iterators.md (V_cmp_result): Rename to...
3027 (V_INT_EQUIV): ...this.
3028 (v_cmp_result): Rename to...
3029 (v_int_equiv): ...this.
3030 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
3031 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
3032 (copysign<mode>3): Likewise.
3033 (aarch64_simd_bsl<mode>_internal): Likewise.
3034 (aarch64_simd_bsl<mode>): Likewise.
3035 (vec_cmp<mode><mode>): Likewise.
3036 (vcond<mode><mode>): Likewise.
3037 (vcond<v_cmp_mixed><mode>): Likewise.
3038 (vcondu<mode><v_cmp_mixed>): Likewise.
3039 (aarch64_cm<optab><mode>): Likewise.
3040 (aarch64_cmtst<mode>): Likewise.
3041 (aarch64_fac<optab><mode>): Likewise.
3042 (vec_perm_const<mode>): Likewise.
3043 (vcond_mask_<mode><v_cmp_result>): Rename to...
3044 (vcond_mask_<mode><v_int_equiv>): ...this.
3045 (vec_cmp<mode><v_cmp_result>): Rename to...
3046 (vec_cmp<mode><v_int_equiv>): ...this.
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/aarch64-modes.def: Remove 32-, 48- and 64-byte
3054 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
3055 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
3056 (UNSPEC_LD4_DREG): New unspecs.
3057 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
3058 (aarch64_ld2<mode>_dreg_be): Replace with...
3059 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
3061 (aarch64_ld3<mode>_dreg_le)
3062 (aarch64_ld3<mode>_dreg_be): Replace with...
3063 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
3065 (aarch64_ld4<mode>_dreg_le)
3066 (aarch64_ld4<mode>_dreg_be): Replace with...
3067 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
3070 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3072 PR tree-optimization/81987
3073 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
3074 insert an initializer in a location not dominated by the stride
3077 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3079 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
3080 on the entire header of the finally block in the fallthru case.
3082 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3084 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
3086 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
3088 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
3089 rs6000_emit_move_from_cr and call renamed function.
3090 (rs6000_emit_prologue): Call renamed functions.
3091 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
3092 movesi_from_cr, remove volatile CRs.
3094 2017-08-30 Jon Beniston <jon@beniston.com>
3095 Richard Biener <rguenther@suse.de>
3097 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
3098 of VECTOR_MODE_P check.
3099 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
3100 element vector types.
3102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3104 * df.h (df_read_modify_subreg_p): Remove in favor of...
3105 * rtl.h (read_modify_subreg_p): ...this new function. Take a
3106 const_rtx instead of an rtx.
3107 * cprop.c (local_cprop_find_used_regs): Update accordingly.
3108 * df-problems.c (df_word_lr_mark_ref): Likewise.
3109 * ira-lives.c (mark_pseudo_reg_live): Likewise.
3110 (mark_pseudo_reg_dead): Likewise.
3111 (mark_ref_dead): Likewise.
3112 * reginfo.c (init_subregs_of_mode): Likewise.
3113 * sched-deps.c (sched_analyze_1): Likewise.
3114 * df-scan.c (df_def_record_1): Likewise.
3115 (df_uses_record): Likewise.
3116 (df_read_modify_subreg_p): Remove in favor of...
3117 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
3118 const_rtx instead of an rtx.
3120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3121 Alan Hayward <alan.hayward@arm.com>
3122 David Sherwood <david.sherwood@arm.com>
3124 * rtl.h (partial_subreg_p): New function.
3125 * caller-save.c (save_call_clobbered_regs): Use it.
3126 * calls.c (expand_call): Likewise.
3127 * combine.c (combinable_i3pat): Likewise.
3128 (simplify_set): Likewise.
3129 (make_extraction): Likewise.
3130 (make_compound_operation_int): Likewise.
3131 (gen_lowpart_or_truncate): Likewise.
3132 (force_to_mode): Likewise.
3133 (make_field_assignment): Likewise.
3134 (reg_truncated_to_mode): Likewise.
3135 (record_truncated_value): Likewise.
3136 (move_deaths): Likewise.
3137 * cse.c (record_jump_cond): Likewise.
3138 (cse_insn): Likewise.
3139 * cselib.c (cselib_lookup_1): Likewise.
3140 * expmed.c (extract_bit_field_using_extv): Likewise.
3141 * function.c (assign_parm_setup_reg): Likewise.
3142 * ifcvt.c (noce_convert_multiple_sets): Likewise.
3143 * ira-build.c (create_insn_allocnos): Likewise.
3144 * lra-coalesce.c (merge_pseudos): Likewise.
3145 * lra-constraints.c (match_reload): Likewise.
3146 (simplify_operand_subreg): Likewise.
3147 (curr_insn_transform): Likewise.
3148 * lra-lives.c (process_bb_lives): Likewise.
3149 * lra.c (new_insn_reg): Likewise.
3150 (lra_substitute_pseudo): Likewise.
3151 * regcprop.c (mode_change_ok): Likewise.
3152 (maybe_mode_change): Likewise.
3153 (copyprop_hardreg_forward_1): Likewise.
3154 * reload.c (push_reload): Likewise.
3155 (find_reloads): Likewise.
3156 (find_reloads_subreg_address): Likewise.
3157 * reload1.c (alter_reg): Likewise.
3158 (eliminate_regs_1): Likewise.
3159 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3161 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
3163 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
3164 back to if statements, including unpack.
3166 2017-08-30 Martin Liska <mliska@suse.cz>
3169 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
3171 (func_checker::compare_asm_inputs_outputs): ... this function.
3172 (func_checker::compare_gimple_asm): Use the function to compare
3173 also ASM constrains.
3174 * ipa-icf-gimple.h: Rename the function.
3176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3177 Alan Hayward <alan.hayward@arm.com>
3178 David Sherwood <david.sherwood@arm.com>
3180 * coretypes.h (complex_mode): New type.
3181 * gdbhooks.py (build_pretty_printer): Handle it.
3182 * machmode.h (complex_mode): New class.
3183 (complex_mode::includes_p): New function.
3184 (is_complex_int_mode): Likewise.
3185 (is_complex_float_mode): Likewise.
3186 * genmodes.c (get_mode_class): Handle complex mode classes.
3187 * function.c (expand_function_end): Use is_complex_int_mode.
3189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3190 Alan Hayward <alan.hayward@arm.com>
3191 David Sherwood <david.sherwood@arm.com>
3193 * coretypes.h (scalar_mode_pod): New typedef.
3194 * gdbhooks.py (build_pretty_printer): Handle it.
3195 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
3196 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
3197 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
3198 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
3201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3202 Alan Hayward <alan.hayward@arm.com>
3203 David Sherwood <david.sherwood@arm.com>
3205 * machmode.h (mode_for_vector): Take a scalar_mode instead
3207 * stor-layout.c (mode_for_vector): Likewise.
3208 * explow.c (promote_mode): Use as_a <scalar_mode>.
3209 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
3211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3212 Alan Hayward <alan.hayward@arm.com>
3213 David Sherwood <david.sherwood@arm.com>
3215 * target.def (preferred_simd_mode): Take a scalar_mode
3216 instead of a machine_mode.
3217 * targhooks.h (default_preferred_simd_mode): Likewise.
3218 * targhooks.c (default_preferred_simd_mode): Likewise.
3219 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
3220 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
3221 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
3222 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
3223 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
3224 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
3225 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
3226 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
3228 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
3229 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
3230 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
3231 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
3232 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
3233 * doc/tm.texi: Regenerate.
3234 * optabs-query.c (can_vec_mask_load_store_p): Return false for
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 (scalar_mode_supported_p): Take a scalar_mode
3242 instead of a machine_mode.
3243 * targhooks.h (default_scalar_mode_supported_p): Likewise.
3244 * targhooks.c (default_scalar_mode_supported_p): Likewise.
3245 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
3246 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3247 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
3248 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
3249 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
3250 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
3251 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
3252 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
3253 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
3254 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
3255 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
3256 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
3258 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
3259 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
3260 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3261 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3262 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
3264 * doc/tm.texi: Regenerate.
3266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3267 Alan Hayward <alan.hayward@arm.com>
3268 David Sherwood <david.sherwood@arm.com>
3270 * coretypes.h (opt_scalar_mode): New typedef.
3271 * gdbhooks.py (build_pretty_printers): Handle it.
3272 * machmode.h (mode_iterator::get_2xwider): Add overload for
3274 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
3276 * expr.c (convert_mode_scalar): Likewise.
3277 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3278 * optabs.c (expand_float): Likewise.
3279 (expand_fix): Likewise.
3280 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3283 Alan Hayward <alan.hayward@arm.com>
3284 David Sherwood <david.sherwood@arm.com>
3286 * optabs.c (expand_float): Explicitly check for scalars before
3287 using a branching expansion.
3288 (expand_fix): Likewise.
3290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3291 Alan Hayward <alan.hayward@arm.com>
3292 David Sherwood <david.sherwood@arm.com>
3294 * expr.c (convert_mode): Split scalar handling out into...
3295 (convert_mode_scalar): ...this new function. Treat the modes
3298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3299 Alan Hayward <alan.hayward@arm.com>
3300 David Sherwood <david.sherwood@arm.com>
3302 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
3304 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
3306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3307 Alan Hayward <alan.hayward@arm.com>
3308 David Sherwood <david.sherwood@arm.com>
3310 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
3311 rather than a machine_mode.
3312 (fixed_from_string): Likewise.
3313 (fixed_convert): Likewise.
3314 (fixed_convert_from_int): Likewise.
3315 (fixed_convert_from_real): Likewise.
3316 (real_convert_from_fixed): Likewise.
3317 * fixed-value.c (fixed_from_double_int): Likewise.
3318 (fixed_from_string): Likewise.
3319 (fixed_convert): Likewise.
3320 (fixed_convert_from_int): Likewise.
3321 (fixed_convert_from_real): Likewise.
3322 (real_convert_from_fixed): Likewise.
3323 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
3325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3326 Alan Hayward <alan.hayward@arm.com>
3327 David Sherwood <david.sherwood@arm.com>
3329 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
3330 of separate mode class checks. Do not allow vector modes here.
3331 (immed_wide_int_const): Use as_a <scalar_mode>.
3332 * explow.c (trunc_int_for_mode): Likewise.
3333 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
3334 (wi::shwi): Likewise.
3335 (wi::min_value): Likewise.
3336 (wi::max_value): Likewise.
3337 * dwarf2out.c (loc_descriptor): Likewise.
3338 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
3341 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3342 Alan Hayward <alan.hayward@arm.com>
3343 David Sherwood <david.sherwood@arm.com>
3345 * tree.h (SCALAR_TYPE_MODE): New macro.
3346 * expr.c (expand_expr_addr_expr_1): Use it.
3347 (expand_expr_real_2): Likewise.
3348 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
3349 (fold_convert_const_fixed_from_int): Likewise.
3350 (fold_convert_const_fixed_from_real): Likewise.
3351 (native_encode_fixed): Likewise
3352 (native_encode_complex): Likewise
3353 (native_encode_vector): Likewise.
3354 (native_interpret_fixed): Likewise.
3355 (native_interpret_real): Likewise.
3356 (native_interpret_complex): Likewise.
3357 (native_interpret_vector): Likewise.
3358 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
3359 (simd_clone_adjust_argument_types): Likewise.
3360 (simd_clone_init_simd_arrays): Likewise.
3361 (simd_clone_adjust): Likewise.
3362 * stor-layout.c (layout_type): Likewise.
3363 * tree.c (build_minus_one_cst): Likewise.
3364 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3365 * tree-inline.c (estimate_move_cost): Likewise.
3366 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
3367 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
3368 (vectorizable_reduction): Likewise.
3369 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
3370 (vect_recog_mixed_size_cond_pattern): Likewise.
3371 (check_bool_pattern): Likewise.
3372 (adjust_bool_pattern): Likewise.
3373 (search_type_for_mask_1): Likewise.
3374 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3375 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3376 (vectorizable_load): Likewise.
3377 (vectorizable_store): Likewise.
3378 * ubsan.c (ubsan_encode_value): Likewise.
3379 * varasm.c (output_constant): Likewise.
3381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3382 Alan Hayward <alan.hayward@arm.com>
3383 David Sherwood <david.sherwood@arm.com>
3385 * coretypes.h (scalar_mode): New class.
3386 * machmode.h (scalar_mode): Likewise.
3387 (scalar_mode::includes_p): New function.
3388 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
3389 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
3390 * genmodes.c (get_mode_class): Handle remaining scalar modes.
3391 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
3392 * expmed.c (store_bit_field_1): Likewise.
3393 (extract_bit_field_1): Likewise.
3394 * expr.c (write_complex_part): Likewise.
3395 (read_complex_part): Likewise.
3396 (emit_move_complex_push): Likewise.
3397 (expand_expr_real_2): Likewise.
3398 * function.c (assign_parm_setup_reg): Likewise.
3399 (assign_parms_unsplit_complex): Likewise.
3400 * optabs.c (expand_binop): Likewise.
3401 * rtlanal.c (subreg_get_info): Likewise.
3402 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3403 * varasm.c (output_constant_pool_2): Likewise.
3405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3406 Alan Hayward <alan.hayward@arm.com>
3407 David Sherwood <david.sherwood@arm.com>
3409 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3411 (expmed_mult_highpart_optab): Likewise.
3412 (expmed_mult_highpart): Likewise.
3414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3415 Alan Hayward <alan.hayward@arm.com>
3416 David Sherwood <david.sherwood@arm.com>
3418 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3419 instead of a machine_mode.
3420 (builtin_memset_read_str): Likewise.
3421 * builtins.c (c_readstr): Likewise.
3422 (builtin_memcpy_read_str): Likewise.
3423 (builtin_strncpy_read_str): Likewise.
3424 (builtin_memset_read_str): Likewise.
3425 (builtin_memset_gen_str): Likewise.
3426 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3427 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3428 instead of a machine_mode.
3429 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3431 (make_extraction): Likewise.
3432 (try_widen_shift_mode): Take and return scalar_int_modes instead
3434 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3435 a scalar_int_mode instead of a machine_mode.
3436 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3437 (avr_addr_space_pointer_mode): Likewise.
3438 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3439 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3440 (msp430_unwind_word_mode): Likewise.
3441 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3442 (spu_addr_space_pointer_mode): Likewise.
3443 (spu_addr_space_address_mode): Likewise.
3444 (spu_libgcc_cmp_return_mode): Likewise.
3445 (spu_libgcc_shift_count_mode): Likewise.
3446 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3447 (rl78_addr_space_pointer_mode): Likewise.
3448 (fl78_unwind_word_mode): Likewise.
3449 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3451 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3452 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3453 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3454 (mips_valid_pointer_mode): Likewise.
3455 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3456 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3457 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3459 (ft32_addr_space_address_mode): Likewise.
3460 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3461 scalar_int_mode instead of a machine_mode.
3462 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3464 (m32c_addr_space_address_mode): Likewise.
3465 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3466 (rs6000_eh_return_filter_mode): Likewise.
3467 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3468 (rs6000_eh_return_filter_mode): Likewise.
3469 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3470 (s390_libgcc_shift_count_mode): Likewise.
3471 (s390_unwind_word_mode): Likewise.
3472 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3474 * target.def (mode_rep_extended): Likewise.
3475 (valid_pointer_mode): Likewise.
3476 (addr_space.valid_pointer_mode): Likewise.
3477 (eh_return_filter_mode): Return a scalar_int_mode rather than
3479 (libgcc_cmp_return_mode): Likewise.
3480 (libgcc_shift_count_mode): Likewise.
3481 (unwind_word_mode): Likewise.
3482 (addr_space.pointer_mode): Likewise.
3483 (addr_space.address_mode): Likewise.
3484 * doc/tm.texi: Regenerate.
3485 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3487 (do_jump): Use scalar_int_mode for local variables.
3488 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3489 rather than a machine_mode.
3490 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3491 (scompare_loc_descriptor_wide): Likewise.
3492 (scompare_loc_descriptor_narrow): Likewise.
3493 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3495 * except.c (sjlj_emit_dispatch_table): Likewise.
3496 (expand_builtin_eh_copy_values): Likewise.
3497 * explow.c (convert_memory_address_addr_space_1): Likewise.
3498 Take a scalar_int_mode rather than a machine_mode.
3499 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3500 than a machine_mode.
3501 (memory_address_addr_space): Use scalar_int_mode for local variables.
3502 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3503 rather than a machine_mode.
3504 * expmed.c (mask_rtx): Likewise.
3505 (init_expmed_one_conv): Likewise.
3506 (expand_mult_highpart_adjust): Likewise.
3507 (extract_high_half): Likewise.
3508 (expmed_mult_highpart_optab): Likewise.
3509 (expmed_mult_highpart): Likewise.
3510 (expand_smod_pow2): Likewise.
3511 (expand_sdiv_pow2): Likewise.
3512 (emit_store_flag_int): Likewise.
3513 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3515 (extract_low_bits): Likewise.
3516 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3518 * expr.c (pieces_addr::adjust): Likewise.
3519 (can_store_by_pieces): Likewise.
3520 (store_by_pieces): Likewise.
3521 (clear_by_pieces_1): Likewise.
3522 (expand_expr_addr_expr_1): Likewise.
3523 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3524 (expand_expr_real_1): Likewise.
3525 (try_casesi): Likewise.
3526 * final.c (shorten_branches): Likewise.
3527 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3528 type of "mode" to machine_mode.
3529 * internal-fn.c (expand_arith_overflow_result_store): Take a
3530 scalar_int_mode rather than a machine_mode.
3531 (expand_mul_overflow): Use scalar_int_mode for local variables.
3532 * loop-doloop.c (doloop_modify): Likewise.
3533 (doloop_optimize): Likewise.
3534 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3535 than a machine_mode.
3536 (expand_doubleword_shift_condmove): Likewise.
3537 (expand_doubleword_shift): Likewise.
3538 (expand_doubleword_clz): Likewise.
3539 (expand_doubleword_popcount): Likewise.
3540 (expand_doubleword_parity): Likewise.
3541 (expand_absneg_bit): Use scalar_int_mode for local variables.
3542 (prepare_float_lib_cmp): Likewise.
3543 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3544 rather than a machine_mode.
3545 (convert_memory_address_addr_space): Likewise.
3546 (get_mode_bounds): Likewise.
3547 (get_address_mode): Return a scalar_int_mode rather than a
3549 * rtlanal.c (get_address_mode): Likewise.
3550 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3551 than a machine_mode.
3552 * targhooks.c (default_mode_rep_extended): Likewise.
3553 (default_valid_pointer_mode): Likewise.
3554 (default_addr_space_valid_pointer_mode): Likewise.
3555 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3556 than a machine_mode.
3557 (default_libgcc_cmp_return_mode): Likewise.
3558 (default_libgcc_shift_count_mode): Likewise.
3559 (default_unwind_word_mode): Likewise.
3560 (default_addr_space_pointer_mode): Likewise.
3561 (default_addr_space_address_mode): Likewise.
3562 * targhooks.h (default_eh_return_filter_mode): Likewise.
3563 (default_libgcc_cmp_return_mode): Likewise.
3564 (default_libgcc_shift_count_mode): Likewise.
3565 (default_unwind_word_mode): Likewise.
3566 (default_addr_space_pointer_mode): Likewise.
3567 (default_addr_space_address_mode): Likewise.
3568 (default_mode_rep_extended): Take a scalar_int_mode rather than
3570 (default_valid_pointer_mode): Likewise.
3571 (default_addr_space_valid_pointer_mode): Likewise.
3572 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3574 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3575 rather than a machine_mode.
3576 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3577 for local variables.
3578 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3579 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3580 than a machine_mode.
3582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3583 Alan Hayward <alan.hayward@arm.com>
3584 David Sherwood <david.sherwood@arm.com>
3586 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3587 the mode argument to scalar_int_mode.
3588 (do_jump_by_parts_zero_rtx): Likewise.
3589 (do_jump_by_parts_equality_rtx): Likewise.
3590 (do_jump_by_parts_greater): Take a mode argument.
3591 (do_jump_by_parts_equality): Likewise.
3592 (do_jump_1): Update calls accordingly.
3594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3595 Alan Hayward <alan.hayward@arm.com>
3596 David Sherwood <david.sherwood@arm.com>
3598 * is-a.h (safe_dyn_cast): New function.
3599 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3600 (jump_table_for_label): Likewise.
3601 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3602 instead of an rtx_insn *.
3603 (shorten_branches): Use dyn_cast instead of LABEL_P and
3604 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3605 rtx_jump_table_data::get_data_mode.
3606 (final_scan_insn): Likewise.
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 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3613 trying to combine a full-register integer set with a subreg
3616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3617 Alan Hayward <alan.hayward@arm.com>
3618 David Sherwood <david.sherwood@arm.com>
3620 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3621 that is always either address_mode or pointer_mode.
3623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3624 Alan Hayward <alan.hayward@arm.com>
3625 David Sherwood <david.sherwood@arm.com>
3627 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
3628 when the two are known to be equal.
3630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3631 Alan Hayward <alan.hayward@arm.com>
3632 David Sherwood <david.sherwood@arm.com>
3634 * simplify-rtx.c (simplify_const_unary_operation): Use
3635 is_a <scalar_int_mode> instead of checking for a nonzero
3636 precision. Forcibly convert op_mode to a scalar_int_mode
3637 in that case. More clearly differentiate the operand and
3638 result modes and use the former when deciding what the value
3639 of a count-bits operation should be. Use is_int_mode instead
3640 of checking for a MODE_INT. Remove redundant check for whether
3641 this mode has a zero precision.
3643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3644 Alan Hayward <alan.hayward@arm.com>
3645 David Sherwood <david.sherwood@arm.com>
3647 * optabs.c (widen_leading): Change the type of the mode argument
3648 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
3649 (widen_bswap): Likewise.
3650 (expand_parity): Likewise.
3651 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
3652 (expand_ffs): Likewise.
3653 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3662 (expand_expr_real_1): Update call accordingly.
3664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3665 Alan Hayward <alan.hayward@arm.com>
3666 David Sherwood <david.sherwood@arm.com>
3668 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3669 value_mode arguments. Use scalar_int_mode internally.
3670 (store_bit_field_1): Rename the new integer mode from imode
3671 to op0_mode and use it instead of GET_MODE (op0). Update calls
3672 to store_split_bit_field, store_bit_field_using_insv and
3673 store_fixed_bit_field.
3674 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3675 Use scalar_int_mode internally. Use a bit count rather than a mode
3676 when calculating the largest bit size for get_best_mode.
3677 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3678 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3679 Remove assertion that OP0 has a scalar integer mode.
3680 (store_split_bit_field): Add op0_mode and value_mode arguments.
3681 Update calls to extract_fixed_bit_field.
3682 (extract_bit_field_using_extv): Add an op0_mode argument.
3683 Use scalar_int_mode internally.
3684 (extract_bit_field_1): Rename the new integer mode from imode to
3685 op0_mode and use it instead of GET_MODE (op0). Update calls to
3686 extract_split_bit_field, extract_bit_field_using_extv and
3687 extract_fixed_bit_field.
3688 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3689 to extract_split_bit_field and extract_fixed_bit_field_1.
3690 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3691 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3693 (extract_split_bit_field): Add an op0_mode argument. Update call
3694 to extract_fixed_bit_field.
3696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3697 Alan Hayward <alan.hayward@arm.com>
3698 David Sherwood <david.sherwood@arm.com>
3700 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3701 * explow.c (hard_function_value): Likewise.
3702 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3703 convert_to_mode call outside the loop.
3704 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3705 for the mode iterator. Require the mode specified by max_pieces
3707 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3709 (copy_blkmode_to_reg): Likewise.
3710 (set_storage_via_setmem): Likewise.
3711 * optabs.c (prepare_cmp_insn): Likewise.
3712 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3713 * stor-layout.c (finish_bitfield_representative): Likewise.
3715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3716 Alan Hayward <alan.hayward@arm.com>
3717 David Sherwood <david.sherwood@arm.com>
3719 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3720 * expr.c (convert_move): Use them.
3721 (convert_modes): Likewise.
3722 (store_expr_with_bounds): Likewise.
3724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3725 Alan Hayward <alan.hayward@arm.com>
3726 David Sherwood <david.sherwood@arm.com>
3728 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3729 parameter for the mode of "x". Remove the "known_x", "known_mode"
3730 and "known_ret" arguments. Change the type of the mode argument
3732 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3733 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3734 (reg_num_sign_bit_copies_for_combine): Likewise.
3735 * rtlanal.c (nonzero_bits1): Likewise.
3736 (num_sign_bit_copies1): Likewise.
3737 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3738 (reg_num_sign_bit_copies_general): Likewise.
3739 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3740 (reg_nonzero_bits_general): Likewise.
3742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3743 Alan Hayward <alan.hayward@arm.com>
3744 David Sherwood <david.sherwood@arm.com>
3746 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
3747 than in subroutines. Return 1 for non-integer modes.
3748 (cached_num_sign_bit_copies): Change the type of the mode parameter
3750 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
3751 classes. Handle CONST_INT_P first and then check whether X also
3752 has a scalar integer mode. Check the same thing for inner registers
3753 of a SUBREG and for values that are being extended or truncated.
3755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3756 Alan Hayward <alan.hayward@arm.com>
3757 David Sherwood <david.sherwood@arm.com>
3759 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
3760 in subroutines. Return the mode mask for non-integer modes.
3761 (cached_nonzero_bits): Change the type of the mode parameter
3763 (nonzero_bits1): Likewise. Remove early exit for other mode
3764 classes. Handle CONST_INT_P first and then check whether X
3765 also has a scalar integer mode.
3767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3768 Alan Hayward <alan.hayward@arm.com>
3769 David Sherwood <david.sherwood@arm.com>
3771 * expr.c (widest_int_mode_for_size): Make the comment match the code.
3772 Return a scalar_int_mode and assert that the size is greater than
3774 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
3775 (op_by_pieces_d::op_by_pieces_d): Likewise.
3776 (op_by_pieces_d::run): Likewise.
3777 (can_store_by_pieces): Likewise.
3779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3780 Alan Hayward <alan.hayward@arm.com>
3781 David Sherwood <david.sherwood@arm.com>
3783 * combine.c (extract_left_shift): Add a mode argument and update
3785 (make_compound_operation_int): Change the type of the mode parameter
3786 to scalar_int_mode and update the call to extract_left_shift.
3788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3789 Alan Hayward <alan.hayward@arm.com>
3790 David Sherwood <david.sherwood@arm.com>
3792 * combine.c (simplify_and_const_int): Change the type of the mode
3793 parameter to scalar_int_mode.
3794 (simplify_and_const_int_1): Likewise. Update recursive call.
3796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3797 Alan Hayward <alan.hayward@arm.com>
3798 David Sherwood <david.sherwood@arm.com>
3800 * combine.c (simplify_compare_const): Check that the mode is a
3801 scalar_int_mode (rather than VOIDmode) before testing its
3803 (simplify_comparison): Move COMPARISON_P handling out of the
3804 loop and restrict the latter part of the loop to scalar_int_modes.
3805 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
3806 and when considering SUBREG_REGs. Use is_int_mode instead of
3807 checking GET_MODE_CLASS against MODE_INT.
3809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3810 Alan Hayward <alan.hayward@arm.com>
3811 David Sherwood <david.sherwood@arm.com>
3813 * combine.c (try_widen_shift_mode): Move check for equal modes to...
3814 (simplify_shift_const_1): ...here. Use scalar_int_mode for
3815 shift_unit_mode and for modes involved in scalar shifts.
3817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3818 Alan Hayward <alan.hayward@arm.com>
3819 David Sherwood <david.sherwood@arm.com>
3821 * combine.c (force_int_to_mode): New function, split out from...
3822 (force_to_mode): ...here. Keep xmode up-to-date and use it
3823 instead of GET_MODE (x).
3825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3826 Alan Hayward <alan.hayward@arm.com>
3827 David Sherwood <david.sherwood@arm.com>
3829 * optabs-query.h (extraction_insn::struct_mode): Change type to
3830 opt_scalar_int_mode and update comment.
3831 (extraction_insn::field_mode): Change type to scalar_int_mode.
3832 (extraction_insn::pos_mode): Likewise.
3833 * combine.c (make_extraction): Update accordingly.
3834 * optabs-query.c (get_traditional_extraction_insn): Likewise.
3835 (get_optab_extraction_insn): Likewise.
3836 * recog.c (simplify_while_replacing): Likewise.
3837 * expmed.c (narrow_bit_field_mem): Change the type of the mode
3838 parameter to opt_scalar_int_mode.
3840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3841 Alan Hayward <alan.hayward@arm.com>
3842 David Sherwood <david.sherwood@arm.com>
3844 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
3845 to a scalar_int_mode instead of a machine_mode.
3846 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
3847 (get_best_mode): Return a boolean and use a pointer argument to store
3848 the selected mode. Replace the limit mode parameter with a bit limit.
3849 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
3850 for the values returned by bit_field_mode_iterator::next_mode.
3851 (store_bit_field): Update call to get_best_mode.
3852 (store_fixed_bit_field): Likewise.
3853 (extract_fixed_bit_field): Likewise.
3854 * expr.c (optimize_bitfield_assignment_op): Likewise.
3855 * fold-const.c (optimize_bit_field_compare): Likewise.
3856 (fold_truth_andor_1): Likewise.
3857 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
3858 Update for new type of m_mode.
3859 (get_best_mode): As above.
3861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3862 Alan Hayward <alan.hayward@arm.com>
3863 David Sherwood <david.sherwood@arm.com>
3865 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
3866 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
3867 (store_bit_field): Check is_a <scalar_int_mode> before calling
3868 strict_volatile_bitfield_p.
3869 (extract_bit_field): Likewise.
3871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3872 Alan Hayward <alan.hayward@arm.com>
3873 David Sherwood <david.sherwood@arm.com>
3875 * target.def (cstore_mode): Return a scalar_int_mode.
3876 * doc/tm.texi: Regenerate.
3877 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
3878 * targhooks.h (default_cstore_mode): Likewise.
3879 * targhooks.c (default_cstore_mode): Likewise, using a forced
3881 * expmed.c (emit_cstore): Expect the target of the cstore to be
3884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3885 Alan Hayward <alan.hayward@arm.com>
3886 David Sherwood <david.sherwood@arm.com>
3888 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
3890 (niter_desc): Likewise mode.
3891 (iv_analyze): Add a mode parameter.
3893 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
3894 and change its type to scalar_int_mode.
3895 * loop-iv.c: Update commentary at head of file.
3896 (iv_constant): Pass the mode paraeter before the rtx it describes
3897 and change its type to scalar_int_mode. Remove VOIDmode handling.
3898 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
3899 (iv_extend): Likewise.
3900 (shorten_into_mode): Likewise.
3901 (iv_add): Use scalar_int_mode.
3902 (iv_mult): Likewise.
3903 (iv_shift): Likewise.
3904 (canonicalize_iv_subregs): Likewise.
3905 (get_biv_step_1): Pass the outer_mode parameter before the rtx
3906 it describes and change its mode to scalar_int_mode. Also change
3907 the type of the returned inner_mode to scalar_int_mode.
3908 (get_biv_step): Likewise, turning outer_mode from a pointer
3909 into a direct parameter. Update call to get_biv_step_1.
3910 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
3911 iv_constant and get_biv_step.
3912 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
3913 and change its type to scalar_int_mode. Don't initialise iv->mode
3914 to VOIDmode and remove later checks for its still being VOIDmode.
3915 Update calls to iv_analyze_op and iv_analyze_expr. Check
3916 is_a <scalar_int_mode> when changing the mode under consideration.
3917 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
3918 Update call to iv_analyze_expr.
3919 (iv_analyze_op): Add a mode parameter. Reject subregs whose
3920 inner register is not also a scalar_int_mode. Update call to
3922 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
3923 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
3924 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
3925 separate mode class checks. Update calls to iv_analyze. Remove
3926 fix-up of VOIDmodes after iv_analyze_biv.
3927 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
3928 don't have a scalar_int_mode. Update call to biv_p.
3930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3931 Alan Hayward <alan.hayward@arm.com>
3932 David Sherwood <david.sherwood@arm.com>
3934 * cfgexpand.c (convert_debug_memory_address): Use
3935 as_a <scalar_int_mode>.
3936 * combine.c (expand_compound_operation): Likewise.
3937 (make_extraction): Likewise.
3938 (change_zero_ext): Likewise.
3939 (simplify_comparison): Likewise.
3940 * cse.c (cse_insn): Likewise.
3941 * dwarf2out.c (minmax_loc_descriptor): Likewise.
3942 (mem_loc_descriptor): Likewise.
3943 (loc_descriptor): Likewise.
3944 * expmed.c (init_expmed_one_mode): Likewise.
3945 (synth_mult): Likewise.
3946 (emit_store_flag_1): Likewise.
3947 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
3948 of a comparison with size.
3949 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
3950 (reduce_to_bit_field_precision): Likewise.
3951 * function.c (expand_function_end): Likewise.
3952 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
3953 * loop-doloop.c (doloop_modify): Likewise.
3954 * optabs.c (expand_binop): Likewise.
3955 (expand_unop): Likewise.
3956 (expand_copysign_absneg): Likewise.
3957 (prepare_cmp_insn): Likewise.
3958 (maybe_legitimize_operand): Likewise.
3959 * recog.c (const_scalar_int_operand): Likewise.
3960 * rtlanal.c (get_address_mode): Likewise.
3961 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3962 (simplify_cond_clz_ctz): Likewise.
3963 * tree-nested.c (get_nl_goto_field): Likewise.
3964 * tree.c (build_vector_type_for_mode): Likewise.
3965 * var-tracking.c (use_narrower_mode): Likewise.
3967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3968 Alan Hayward <alan.hayward@arm.com>
3969 David Sherwood <david.sherwood@arm.com>
3971 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
3972 * builtins.c (expand_builtin_signbit): Use it.
3973 * cfgexpand.c (expand_debug_expr): Likewise.
3974 * dojump.c (do_jump): Likewise.
3975 (do_compare_and_jump): Likewise.
3976 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
3977 * expmed.c (make_tree): Likewise.
3978 * expr.c (expand_expr_real_2): Likewise.
3979 (expand_expr_real_1): Likewise.
3980 (try_casesi): Likewise.
3981 * fold-const-call.c (fold_const_call_ss): Likewise.
3982 * fold-const.c (unextend): Likewise.
3983 (extract_muldiv_1): Likewise.
3984 (fold_single_bit_test): Likewise.
3985 (native_encode_int): Likewise.
3986 (native_encode_string): Likewise.
3987 (native_interpret_int): Likewise.
3988 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3989 * internal-fn.c (expand_addsub_overflow): Likewise.
3990 (expand_neg_overflow): Likewise.
3991 (expand_mul_overflow): Likewise.
3992 (expand_arith_overflow): Likewise.
3993 * match.pd: Likewise.
3994 * stor-layout.c (layout_type): Likewise.
3995 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3996 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3997 * tree-ssanames.c (get_range_info): Likewise.
3998 * tree-switch-conversion.c (array_value_type) Likewise.
3999 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
4000 (vect_recog_divmod_pattern): Likewise.
4001 (vect_recog_mixed_size_cond_pattern): Likewise.
4002 * tree-vrp.c (extract_range_basic): Likewise.
4003 (simplify_float_conversion_using_ranges): Likewise.
4004 * tree.c (int_fits_type_p): Likewise.
4005 * ubsan.c (instrument_bool_enum_load): Likewise.
4006 * varasm.c (mergeable_string_section): Likewise.
4007 (narrowing_initializer_constant_valid_p): Likewise.
4008 (output_constant): Likewise.
4010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4011 Alan Hayward <alan.hayward@arm.com>
4012 David Sherwood <david.sherwood@arm.com>
4014 * machmode.h (NARROWEST_INT_MODE): New macro.
4015 * expr.c (alignment_for_piecewise_move): Use it instead of
4016 GET_CLASS_NARROWEST_MODE (MODE_INT).
4017 (push_block): Likewise.
4018 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4020 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4023 Alan Hayward <alan.hayward@arm.com>
4024 David Sherwood <david.sherwood@arm.com>
4026 * postreload.c (move2add_valid_value_p): Change the type of the
4027 mode parameter to scalar_int_mode.
4028 (move2add_use_add2_insn): Add a mode parameter and use it instead
4030 (move2add_use_add3_insn): Likewise.
4031 (reload_cse_move2add): Update accordingly.
4033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4034 Alan Hayward <alan.hayward@arm.com>
4035 David Sherwood <david.sherwood@arm.com>
4037 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
4039 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
4040 * optabs.c (expand_unop): Likewise.
4042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4043 Alan Hayward <alan.hayward@arm.com>
4044 David Sherwood <david.sherwood@arm.com>
4046 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
4047 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
4048 (popcount_loc_descriptor): Likewise.
4049 (bswap_loc_descriptor): Likewise.
4050 (rotate_loc_descriptor): Likewise.
4051 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
4052 calling the functions above.
4054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4055 Alan Hayward <alan.hayward@arm.com>
4056 David Sherwood <david.sherwood@arm.com>
4058 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
4060 (try_combine): Likewise.
4061 (simplify_if_then_else): Likewise.
4062 * cse.c (cse_insn): Likewise.
4063 * dwarf2out.c (mem_loc_descriptor): Likewise.
4064 * emit-rtl.c (gen_lowpart_common): Likewise.
4065 * simplify-rtx.c (simplify_truncation): Likewise.
4066 (simplify_binary_operation_1): Likewise.
4067 (simplify_const_relational_operation): Likewise.
4068 (simplify_ternary_operation): Likewise.
4069 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
4071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4072 Alan Hayward <alan.hayward@arm.com>
4073 David Sherwood <david.sherwood@arm.com>
4075 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
4076 * reload.c (push_reload): Likewise.
4077 (find_reloads): Likewise.
4079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4080 Alan Hayward <alan.hayward@arm.com>
4081 David Sherwood <david.sherwood@arm.com>
4083 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
4084 (make_compound_operation_int): Likewise.
4085 (change_zero_ext): Likewise.
4086 * expr.c (convert_move): Likewise.
4087 (convert_modes): Likewise.
4088 * fwprop.c (forward_propagate_subreg): Likewise.
4089 * loop-iv.c (get_biv_step_1): Likewise.
4090 * optabs.c (widen_operand): Likewise.
4091 * postreload.c (move2add_valid_value_p): Likewise.
4092 * recog.c (simplify_while_replacing): Likewise.
4093 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4094 (simplify_binary_operation_1): Likewise. Remove redundant
4095 mode equality check.
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 * combine.c (combine_simplify_rtx): Add checks for
4102 is_a <scalar_int_mode>.
4103 (simplify_if_then_else): Likewise.
4104 (make_field_assignment): Likewise.
4105 (simplify_comparison): Likewise.
4106 * ifcvt.c (noce_try_bitop): Likewise.
4107 * loop-invariant.c (canonicalize_address_mult): Likewise.
4108 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4111 Alan Hayward <alan.hayward@arm.com>
4112 David Sherwood <david.sherwood@arm.com>
4114 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
4115 is_a <scalar_int_mode> instead of != BLKmode.
4117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4118 Alan Hayward <alan.hayward@arm.com>
4119 David Sherwood <david.sherwood@arm.com>
4121 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
4122 instead of != VOIDmode.
4123 * combine.c (if_then_else_cond): Likewise.
4124 (change_zero_ext): Likewise.
4125 * dwarf2out.c (mem_loc_descriptor): Likewise.
4126 (loc_descriptor): Likewise.
4127 * rtlanal.c (canonicalize_condition): Likewise.
4128 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
4130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4131 Alan Hayward <alan.hayward@arm.com>
4132 David Sherwood <david.sherwood@arm.com>
4134 * simplify-rtx.c (simplify_binary_operation_1): Use
4135 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
4137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4138 Alan Hayward <alan.hayward@arm.com>
4139 David Sherwood <david.sherwood@arm.com>
4141 * wide-int.h (int_traits<unsigned char>) New class.
4142 (int_traits<unsigned short>) Likewise.
4143 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
4144 Use GET_MODE_UNIT_PRECISION and remove redundant test for
4146 * combine.c (set_nonzero_bits_and_sign_copies): Use
4147 is_a <scalar_int_mode>.
4148 (find_split_point): Likewise.
4149 (combine_simplify_rtx): Likewise.
4150 (simplify_logical): Likewise.
4151 (expand_compound_operation): Likewise.
4152 (expand_field_assignment): Likewise.
4153 (make_compound_operation): Likewise.
4154 (extended_count): Likewise.
4155 (change_zero_ext): Likewise.
4156 (simplify_comparison): Likewise.
4157 * dwarf2out.c (scompare_loc_descriptor): Likewise.
4158 (ucompare_loc_descriptor): Likewise.
4159 (minmax_loc_descriptor): Likewise.
4160 (mem_loc_descriptor): Likewise.
4161 (loc_descriptor): Likewise.
4162 * expmed.c (init_expmed_one_mode): Likewise.
4163 * lra-constraints.c (lra_constraint_offset): Likewise.
4164 * optabs.c (prepare_libcall_arg): Likewise.
4165 * postreload.c (move2add_note_store): Likewise.
4166 * reload.c (operands_match_p): Likewise.
4167 * rtl.h (load_extend_op): Likewise.
4168 * rtlhooks.c (gen_lowpart_general): Likewise.
4169 * simplify-rtx.c (simplify_truncation): Likewise.
4170 (simplify_unary_operation_1): Likewise.
4171 (simplify_binary_operation_1): Likewise.
4172 (simplify_const_binary_operation): Likewise.
4173 (simplify_const_relational_operation): Likewise.
4174 (simplify_subreg): Likewise.
4175 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4176 * var-tracking.c (adjust_mems): Likewise.
4177 (prepare_call_arguments): Likewise.
4179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4180 Alan Hayward <alan.hayward@arm.com>
4181 David Sherwood <david.sherwood@arm.com>
4183 * machmode.h (is_int_mode): New fuction.
4184 * combine.c (find_split_point): Use it.
4185 (combine_simplify_rtx): Likewise.
4186 (simplify_if_then_else): Likewise.
4187 (simplify_set): Likewise.
4188 (simplify_shift_const_1): Likewise.
4189 (simplify_comparison): Likewise.
4190 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
4191 * cse.c (notreg_cost): Likewise.
4192 (cse_insn): Likewise.
4193 * cselib.c (cselib_lookup_1): Likewise.
4194 * dojump.c (do_jump_1): Likewise.
4195 (do_compare_rtx_and_jump): Likewise.
4196 * dse.c (get_call_args): Likewise.
4197 * dwarf2out.c (rtl_for_decl_init): Likewise.
4198 (native_encode_initializer): Likewise.
4199 * expmed.c (emit_store_flag_1): Likewise.
4200 (emit_store_flag): Likewise.
4201 * expr.c (convert_modes): Likewise.
4202 (store_field): Likewise.
4203 (expand_expr_real_1): Likewise.
4204 * fold-const.c (fold_read_from_constant_string): Likewise.
4205 * gimple-ssa-sprintf.c (get_format_string): Likewise.
4206 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
4207 * optabs.c (expand_binop): Likewise.
4208 (expand_unop): Likewise.
4209 (expand_abs_nojump): Likewise.
4210 (expand_one_cmpl_abs_nojump): Likewise.
4211 * simplify-rtx.c (mode_signbit_p): Likewise.
4212 (val_signbit_p): Likewise.
4213 (val_signbit_known_set_p): Likewise.
4214 (val_signbit_known_clear_p): Likewise.
4215 (simplify_relational_operation_1): Likewise.
4216 * tree.c (vector_type_mode): Likewise.
4218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4219 Alan Hayward <alan.hayward@arm.com>
4220 David Sherwood <david.sherwood@arm.com>
4222 * machmode.h (smallest_mode_for_size): Fix formatting.
4223 (smallest_int_mode_for_size): New function.
4224 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
4225 instead of smallest_mode_for_size.
4226 * combine.c (make_extraction): Likewise.
4227 * config/arc/arc.c (arc_expand_movmem): Likewise.
4228 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
4229 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4230 * config/s390/s390.c (s390_expand_insv): Likewise.
4231 * config/sparc/sparc.c (assign_int_registers): Likewise.
4232 * config/spu/spu.c (spu_function_value): Likewise.
4233 (spu_function_arg): Likewise.
4234 * coverage.c (get_gcov_type): Likewise.
4235 (get_gcov_unsigned_t): Likewise.
4236 * dse.c (find_shift_sequence): Likewise.
4237 * expmed.c (store_bit_field_1): Likewise.
4238 * expr.c (convert_move): Likewise.
4239 (store_field): Likewise.
4240 * internal-fn.c (expand_arith_overflow): Likewise.
4241 * optabs-query.c (get_best_extraction_insn): Likewise.
4242 * optabs.c (expand_twoval_binop_libfunc): Likewise.
4243 * stor-layout.c (layout_type): Likewise.
4244 (initialize_sizetypes): Likewise.
4245 * targhooks.c (default_get_mask_mode): Likewise.
4246 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
4248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4249 Alan Hayward <alan.hayward@arm.com>
4250 David Sherwood <david.sherwood@arm.com>
4252 * machmode.h (opt_mode::else_blk): New function.
4253 (int_mode_for_mode): Declare.
4254 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
4255 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
4257 * cfgexpand.c (expand_debug_expr): Likewise.
4258 * combine.c (gen_lowpart_or_truncate): Likewise.
4259 (gen_lowpart_for_combine): Likewise.
4260 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
4261 * config/avr/avr.c (avr_to_int_mode): Likewise.
4262 (avr_out_plus_1): Likewise.
4263 (avr_out_plus): Likewise.
4264 (avr_out_round): Likewise.
4265 * config/i386/i386.c (ix86_split_to_parts): Likewise.
4266 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4267 (s390_expand_vcond): Likewise.
4268 * config/spu/spu.c (spu_split_immediate): Likewise.
4269 (spu_expand_mov): Likewise.
4270 * dse.c (get_stored_val): Likewise.
4271 * expmed.c (store_bit_field_1): Likewise.
4272 (convert_extracted_bit_field): Use int_mode_for_mode instead of
4274 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
4275 (extract_low_bits): Likewise.
4276 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
4277 handling rather than repeating the check.
4278 (emit_group_store): Likewise.
4279 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
4280 * optabs.c (expand_absneg_bit): Likewise.
4281 (expand_copysign_absneg): Likewise.
4282 (expand_copysign_bit): Likewise.
4283 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4284 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
4285 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
4286 * var-tracking.c (prepare_call_arguments): Likewise.
4287 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
4288 int_mode_for_mode instead of mode_for_size.
4289 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4292 Alan Hayward <alan.hayward@arm.com>
4293 David Sherwood <david.sherwood@arm.com>
4295 * machmode.h (int_mode_for_size): New function.
4296 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
4297 instead of mode_for_size.
4298 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
4300 * combine.c (expand_field_assignment): Use int_mode_for_size
4301 instead of mode_for_size.
4302 (make_extraction): Likewise.
4303 (simplify_shift_const_1): Likewise.
4304 (simplify_comparison): Likewise.
4305 * dojump.c (do_jump): Likewise.
4306 * dwarf2out.c (mem_loc_descriptor): Likewise.
4307 * emit-rtl.c (init_derived_machine_modes): Likewise.
4308 * expmed.c (flip_storage_order): Likewise.
4309 (convert_extracted_bit_field): Likewise.
4310 * expr.c (copy_blkmode_from_reg): Likewise.
4311 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
4312 * internal-fn.c (expand_mul_overflow): Likewise.
4313 * lower-subreg.c (simple_move): Likewise.
4314 * optabs-libfuncs.c (init_optabs): Likewise.
4315 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4316 * tree.c (vector_type_mode): Likewise.
4317 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
4318 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
4319 * tree-vect-generic.c (expand_vector_parallel): Likewise.
4320 * tree-vect-stmts.c (vectorizable_load): Likewise.
4321 (vectorizable_store): Likewise.
4323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4324 Alan Hayward <alan.hayward@arm.com>
4325 David Sherwood <david.sherwood@arm.com>
4327 * coretypes.h (pod_mode): New type.
4328 (scalar_int_mode_pod): New typedef.
4329 * machmode.h (pod_mode): New class.
4330 (int_n_data_t::m): Change type to scalar_int_mode_pod.
4331 * genmodes.c (emit_mode_int_n): Update accordingly.
4332 * lower-subreg.h (target_lower_subreg): Change type to
4333 scalar_int_mode_pod.
4334 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
4335 scalar_int_mode_pod.
4337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4338 Alan Hayward <alan.hayward@arm.com>
4339 David Sherwood <david.sherwood@arm.com>
4341 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
4342 machine_mode to scalar_int_mode.
4343 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
4344 (rs6000_option_override_internal): Remove cast to int.
4345 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
4346 machine_mode to scalar_int_mode.
4347 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
4348 (rs6000_option_override_internal): Remove cast to int.
4349 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
4350 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
4352 * config/s390/s390.c (s390_expand_builtin): Likewise.
4353 * coretypes.h (scalar_int_mode): New type.
4354 (opt_scalar_int_mode): New typedef.
4355 * machmode.h (scalar_int_mode): New class.
4356 (scalar_int_mode::includes_p): New function.
4357 (byte_mode): Change type to scalar_int_mode.
4358 (word_mode): Likewise.
4359 (ptr_mode): Likewise.
4360 * emit-rtl.c (byte_mode): Likewise.
4361 (word_mode): Likewise.
4362 (ptr_mode): Likewise.
4363 (init_derived_machine_modes): Update accordingly.
4364 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
4365 and MODE_PARTIAL_INT.
4366 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
4367 opt_scalar_int_mode.
4369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4370 Alan Hayward <alan.hayward@arm.com>
4371 David Sherwood <david.sherwood@arm.com>
4373 * target.def (libgcc_floating_mode_supported_p): Take a
4375 * doc/tm.texi: Regenerate.
4376 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
4378 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4379 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
4382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4383 Alan Hayward <alan.hayward@arm.com>
4384 David Sherwood <david.sherwood@arm.com>
4386 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
4387 * doc/tm.texi: Regenerate.
4388 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
4389 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
4390 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
4391 * targhooks.h (default_floatn_mode): Likewise.
4392 * targhooks.c (default_floatn_mode): Likewise.
4393 * tree.c (build_common_tree_nodes): Update accordingly.
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 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
4400 (mode_iterator::iterate_p): Likewise.
4401 (mode_iterator::get_wider): Likewise.
4402 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4405 Alan Hayward <alan.hayward@arm.com>
4406 David Sherwood <david.sherwood@arm.com>
4408 * coretypes.h (opt_scalar_float_mode): New typedef.
4409 * machmode.h (float_mode_for_size): New function.
4410 * emit-rtl.c (double_mode): Delete.
4411 (init_emit_once): Use float_mode_for_size.
4412 * stor-layout.c (layout_type): Likewise.
4413 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4416 Alan Hayward <alan.hayward@arm.com>
4417 David Sherwood <david.sherwood@arm.com>
4419 * output.h (assemble_real): Take a scalar_float_mode.
4420 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4421 * config/arm/arm.md (consttable_4): Likewise.
4422 (consttable_8): Likewise.
4423 (consttable_16): Likewise.
4424 * config/mips/mips.md (consttable_float): Likewise.
4425 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4426 * varasm.c (assemble_real): Take a scalar_float_mode.
4427 (output_constant_pool_2): Update accordingly.
4428 (output_constant): Likewise.
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 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4435 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4436 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4437 (native_encode_real): Likewise.
4438 (native_interpret_real): Likewise.
4439 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4440 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4443 Alan Hayward <alan.hayward@arm.com>
4444 David Sherwood <david.sherwood@arm.com>
4446 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4447 <scalar_float_mode>. Simplify.
4448 (gen_extend_conv_libfunc): Likewise.
4450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4451 Alan Hayward <alan.hayward@arm.com>
4452 David Sherwood <david.sherwood@arm.com>
4454 * coretypes.h (scalar_float_mode): New type.
4455 * machmode.h (mode_traits::from_int): Use machine_mode if
4456 USE_ENUM_MODES is defined.
4457 (is_a): New function.
4459 (dyn_cast): Likewise.
4460 (scalar_float_mode): New class.
4461 (scalar_float_mode::includes_p): New function.
4462 (is_float_mode): Likewise.
4463 * gdbhooks.py (MachineModePrinter): New class.
4464 (build_pretty_printer): Use it for scalar_float_mode.
4465 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4466 (format_helper::format_helper): Turn into a template.
4467 * genmodes.c (get_mode_class): New function.
4468 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4469 or machine_mode if none.
4470 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4471 as_a <scalar_float_mode>.
4472 * dwarf2out.c (mem_loc_descriptor): Likewise.
4473 (insert_float): Likewise.
4474 (add_const_value_attribute): Likewise.
4475 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4476 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4477 (expand_unop): Update accordingly.
4478 (expand_abs_nojump): Likewise.
4479 (expand_copysign_absneg): Take a scalar_float_mode.
4480 (expand_copysign_bit): Likewise.
4481 (expand_copysign): Update accordingly.
4483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4484 Alan Hayward <alan.hayward@arm.com>
4485 David Sherwood <david.sherwood@arm.com>
4487 * coretypes.h (opt_mode): New class.
4488 * machmode.h (opt_mode): Likewise.
4489 (opt_mode::else_void): New function.
4490 (opt_mode::require): Likewise.
4491 (opt_mode::exists): Likewise.
4492 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4493 (GET_MODE_2XWIDER_MODE): Likewise.
4494 (mode_iterator::get_wider): Update accordingly.
4495 (mode_iterator::get_2xwider): Likewise.
4496 (mode_iterator::get_known_wider): Likewise, turning into a template.
4497 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4498 forcing a wider mode to exist.
4499 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4500 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4501 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4502 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4503 * lower-subreg.c (init_lower_subreg): Likewise.
4504 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4505 on the final iteration.
4506 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4507 a wider mode exists before asking for a move pattern.
4508 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4509 forcing a wider mode to exist.
4510 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4511 returning false if no such mode exists.
4512 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4513 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4514 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4515 Avoid checking for a MODE_INT if we already know the mode is not a
4517 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4518 forcing a wider mode to exist.
4519 (expmed_mult_highpart_optab): Likewise.
4520 (expmed_mult_highpart): Likewise.
4521 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4523 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4524 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4525 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4526 * internal-fn.c (expand_mul_overflow): Update use of
4527 GET_MODE_2XWIDER_MODE.
4528 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4529 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4530 GET_MODE_WIDER_MODE.
4531 (convert_plusminus_to_widen): Likewise.
4532 * tree-switch-conversion.c (array_value_type): Likewise.
4533 * var-tracking.c (emit_note_insn_var_location): Likewise.
4534 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4535 Return false inside rather than outside the loop if no wider mode
4537 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4538 and GET_MODE_2XWIDER_MODE
4539 (can_compare_p): Use else_void.
4540 * gdbhooks.py (OptMachineModePrinter): New class.
4541 (build_pretty_printer): Use it for opt_mode.
4543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4544 Alan Hayward <alan.hayward@arm.com>
4545 David Sherwood <david.sherwood@arm.com>
4547 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4548 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4551 Alan Hayward <alan.hayward@arm.com>
4552 David Sherwood <david.sherwood@arm.com>
4554 * machmode.h (mode_traits): New structure.
4555 (get_narrowest_mode): New function.
4556 (mode_iterator::start): Likewise.
4557 (mode_iterator::iterate_p): Likewise.
4558 (mode_iterator::get_wider): Likewise.
4559 (mode_iterator::get_known_wider): Likewise.
4560 (mode_iterator::get_2xwider): Likewise.
4561 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4562 (FOR_EACH_MODE): Likewise.
4563 (FOR_EACH_MODE_FROM): Likewise.
4564 (FOR_EACH_MODE_UNTIL): Likewise.
4565 (FOR_EACH_WIDER_MODE): Likewise.
4566 (FOR_EACH_2XWIDER_MODE): Likewise.
4567 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4568 * combine.c (simplify_comparison): Likewise
4569 * config/i386/i386.c (type_natural_mode): Likewise.
4570 * cse.c (cse_insn): Likewise.
4571 * dse.c (find_shift_sequence): Likewise.
4572 * emit-rtl.c (init_derived_machine_modes): Likewise.
4573 (init_emit_once): Likewise.
4574 * explow.c (hard_function_value): Likewise.
4575 * expmed.c (extract_fixed_bit_field_1): Likewise.
4576 (extract_bit_field_1): Likewise.
4577 (expand_divmod): Likewise.
4578 (emit_store_flag_1): Likewise.
4579 * expr.c (init_expr_target): Likewise.
4580 (convert_move): Likewise.
4581 (alignment_for_piecewise_move): Likewise.
4582 (widest_int_mode_for_size): Likewise.
4583 (emit_block_move_via_movmem): Likewise.
4584 (copy_blkmode_to_reg): Likewise.
4585 (set_storage_via_setmem): Likewise.
4586 (compress_float_constant): Likewise.
4587 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4588 * optabs-query.c (get_best_extraction_insn): Likewise.
4589 * optabs.c (expand_binop): Likewise.
4590 (expand_twoval_unop): Likewise.
4591 (expand_twoval_binop): Likewise.
4592 (widen_leading): Likewise.
4593 (widen_bswap): Likewise.
4594 (expand_parity): Likewise.
4595 (expand_unop): Likewise.
4596 (prepare_cmp_insn): Likewise.
4597 (prepare_float_lib_cmp): Likewise.
4598 (expand_float): Likewise.
4599 (expand_fix): Likewise.
4600 (expand_sfix_optab): Likewise.
4601 * postreload.c (move2add_use_add2_insn): Likewise.
4602 * reg-stack.c (reg_to_stack): Likewise.
4603 * reginfo.c (choose_hard_reg_mode): Likewise.
4604 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4605 * stor-layout.c (mode_for_size): Likewise.
4606 (smallest_mode_for_size): Likewise.
4607 (mode_for_vector): Likewise.
4608 (finish_bitfield_representative): Likewise.
4609 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4610 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4611 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4612 * var-tracking.c (prepare_call_arguments): Likewise.
4614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4615 Alan Hayward <alan.hayward@arm.com>
4616 David Sherwood <david.sherwood@arm.com>
4618 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4619 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4620 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4621 * machmode.h (mode_size): Move earlier in file.
4622 (mode_precision): Likewise.
4623 (mode_inner): Likewise.
4624 (mode_nunits): Likewise.
4625 (mode_unit_size): Likewise.
4626 (unit_unit_precision): Likewise.
4627 (mode_wider): Likewise.
4628 (mode_2xwider): Likewise.
4629 (machine_mode): New class.
4630 (mode_to_bytes): New function.
4631 (mode_to_bits): Likewise.
4632 (mode_to_precision): Likewise.
4633 (mode_to_inner): Likewise.
4634 (mode_to_unit_size): Likewise.
4635 (mode_to_unit_precision): Likewise.
4636 (mode_to_nunits): Likewise.
4637 (GET_MODE_SIZE): Use mode_to_bytes.
4638 (GET_MODE_BITSIZE): Use mode_to_bits.
4639 (GET_MODE_PRECISION): Use mode_to_precision.
4640 (GET_MODE_INNER): Use mode_to_inner.
4641 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4642 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
4643 (GET_MODE_NUNITS): Use mode_to_nunits.
4644 * system.h (ALWAYS_INLINE): New macro.
4645 * config/powerpcspe/powerpcspe-c.c
4646 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
4647 int for arg1_mode and arg2_mode.
4649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4650 Alan Hayward <alan.hayward@arm.com>
4651 David Sherwood <david.sherwood@arm.com>
4653 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
4654 Prefix mode names with E_ in case statements.
4655 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4656 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
4657 (aarch64_split_simd_move): Likewise.
4658 (aarch64_gen_storewb_pair): Likewise.
4659 (aarch64_gen_loadwb_pair): Likewise.
4660 (aarch64_gen_store_pair): Likewise.
4661 (aarch64_gen_load_pair): Likewise.
4662 (aarch64_get_condition_code_1): Likewise.
4663 (aarch64_constant_pool_reload_icode): Likewise.
4664 (get_rsqrte_type): Likewise.
4665 (get_rsqrts_type): Likewise.
4666 (get_recpe_type): Likewise.
4667 (get_recps_type): Likewise.
4668 (aarch64_gimplify_va_arg_expr): Likewise.
4669 (aarch64_simd_container_mode): Likewise.
4670 (aarch64_emit_load_exclusive): Likewise.
4671 (aarch64_emit_store_exclusive): Likewise.
4672 (aarch64_expand_compare_and_swap): Likewise.
4673 (aarch64_gen_atomic_cas): Likewise.
4674 (aarch64_emit_bic): Likewise.
4675 (aarch64_emit_atomic_swap): Likewise.
4676 (aarch64_emit_atomic_load_op): Likewise.
4677 (aarch64_evpc_trn): Likewise.
4678 (aarch64_evpc_uzp): Likewise.
4679 (aarch64_evpc_zip): Likewise.
4680 (aarch64_evpc_ext): Likewise.
4681 (aarch64_evpc_rev): Likewise.
4682 (aarch64_evpc_dup): Likewise.
4683 (aarch64_gen_ccmp_first): Likewise.
4684 (aarch64_gen_ccmp_next): Likewise.
4685 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4686 (alpha_emit_xfloating_libcall): Likewise.
4687 (emit_insxl): Likewise.
4688 (alpha_arg_type): Likewise.
4689 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4690 (arc_preferred_simd_mode): Likewise.
4691 (arc_secondary_reload): Likewise.
4692 (get_arc_condition_code): Likewise.
4693 (arc_print_operand): Likewise.
4694 (arc_legitimate_constant_p): Likewise.
4695 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4696 * config/arc/arc.md (casesi_load): Likewise.
4697 (casesi_compact_jump): Likewise.
4698 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4699 (cc_use_register): Likewise.
4700 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4701 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4702 (arm_init_iwmmxt_builtins): Likewise.
4703 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4704 (neon_expand_vector_init): Likewise.
4705 (arm_attr_length_move_neon): Likewise.
4706 (maybe_get_arm_condition_code): Likewise.
4707 (arm_emit_vector_const): Likewise.
4708 (arm_preferred_simd_mode): Likewise.
4709 (arm_output_iwmmxt_tinsr): Likewise.
4710 (thumb1_output_casesi): Likewise.
4711 (thumb2_output_casesi): Likewise.
4712 (arm_emit_load_exclusive): Likewise.
4713 (arm_emit_store_exclusive): Likewise.
4714 (arm_expand_compare_and_swap): Likewise.
4715 (arm_evpc_neon_vuzp): Likewise.
4716 (arm_evpc_neon_vzip): Likewise.
4717 (arm_evpc_neon_vrev): Likewise.
4718 (arm_evpc_neon_vtrn): Likewise.
4719 (arm_evpc_neon_vext): Likewise.
4720 (arm_validize_comparison): Likewise.
4721 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4722 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4723 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4724 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4725 (c6x_preferred_simd_mode): Likewise.
4726 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4727 (epiphany_rtx_costs): Likewise.
4728 * config/epiphany/predicates.md (proper_comparison_operator):
4730 * config/frv/frv.c (condexec_memory_operand): Likewise.
4731 (frv_emit_move): Likewise.
4732 (output_move_single): Likewise.
4733 (output_condmove_single): Likewise.
4734 (frv_hard_regno_mode_ok): Likewise.
4735 (frv_matching_accg_mode): Likewise.
4736 * config/h8300/h8300.c (split_adds_subs): Likewise.
4737 (h8300_rtx_costs): Likewise.
4738 (h8300_print_operand): Likewise.
4739 (compute_mov_length): Likewise.
4740 (output_logical_op): Likewise.
4741 (compute_logical_op_length): Likewise.
4742 (compute_logical_op_cc): Likewise.
4743 (h8300_shift_needs_scratch_p): Likewise.
4744 (output_a_shift): Likewise.
4745 (compute_a_shift_length): Likewise.
4746 (compute_a_shift_cc): Likewise.
4747 (expand_a_rotate): Likewise.
4748 (output_a_rotate): Likewise.
4749 * config/i386/i386.c (classify_argument): Likewise.
4750 (function_arg_advance_32): Likewise.
4751 (function_arg_32): Likewise.
4752 (function_arg_64): Likewise.
4753 (function_value_64): Likewise.
4754 (ix86_gimplify_va_arg): Likewise.
4755 (ix86_legitimate_constant_p): Likewise.
4756 (put_condition_code): Likewise.
4757 (split_double_mode): Likewise.
4758 (ix86_avx256_split_vector_move_misalign): Likewise.
4759 (ix86_expand_vector_logical_operator): Likewise.
4760 (ix86_split_idivmod): Likewise.
4761 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
4762 (ix86_build_const_vector): Likewise.
4763 (ix86_build_signbit_mask): Likewise.
4764 (ix86_match_ccmode): Likewise.
4765 (ix86_cc_modes_compatible): Likewise.
4766 (ix86_expand_branch): Likewise.
4767 (ix86_expand_sse_cmp): Likewise.
4768 (ix86_expand_sse_movcc): Likewise.
4769 (ix86_expand_int_sse_cmp): Likewise.
4770 (ix86_expand_vec_perm_vpermi2): Likewise.
4771 (ix86_expand_vec_perm): Likewise.
4772 (ix86_expand_sse_unpack): Likewise.
4773 (ix86_expand_int_addcc): Likewise.
4774 (ix86_split_to_parts): Likewise.
4775 (ix86_vectorize_builtin_gather): Likewise.
4776 (ix86_vectorize_builtin_scatter): Likewise.
4777 (avx_vpermilp_parallel): Likewise.
4778 (inline_memory_move_cost): Likewise.
4779 (ix86_tieable_integer_mode_p): Likewise.
4780 (x86_maybe_negate_const_int): Likewise.
4781 (ix86_expand_vector_init_duplicate): Likewise.
4782 (ix86_expand_vector_init_one_nonzero): Likewise.
4783 (ix86_expand_vector_init_one_var): Likewise.
4784 (ix86_expand_vector_init_concat): Likewise.
4785 (ix86_expand_vector_init_interleave): Likewise.
4786 (ix86_expand_vector_init_general): Likewise.
4787 (ix86_expand_vector_set): Likewise.
4788 (ix86_expand_vector_extract): Likewise.
4789 (emit_reduc_half): Likewise.
4790 (ix86_emit_i387_round): Likewise.
4791 (ix86_mangle_type): Likewise.
4792 (ix86_expand_round_sse4): Likewise.
4793 (expand_vec_perm_blend): Likewise.
4794 (canonicalize_vector_int_perm): Likewise.
4795 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
4796 (expand_vec_perm_1): Likewise.
4797 (expand_vec_perm_interleave3): Likewise.
4798 (expand_vec_perm_even_odd_pack): Likewise.
4799 (expand_vec_perm_even_odd_1): Likewise.
4800 (expand_vec_perm_broadcast_1): Likewise.
4801 (ix86_vectorize_vec_perm_const_ok): Likewise.
4802 (ix86_expand_vecop_qihi): Likewise.
4803 (ix86_expand_mul_widen_hilo): Likewise.
4804 (ix86_expand_sse2_abs): Likewise.
4805 (ix86_expand_pextr): Likewise.
4806 (ix86_expand_pinsr): Likewise.
4807 (ix86_preferred_simd_mode): Likewise.
4808 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
4809 * config/i386/sse.md (*andnot<mode>3): Likewise.
4810 (<mask_codefor><code><mode>3<mask_name>): Likewise.
4811 (*<code><mode>3): Likewise.
4812 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
4813 (ia64_expand_atomic_op): Likewise.
4814 (ia64_arg_type): Likewise.
4815 (ia64_mode_to_int): Likewise.
4816 (ia64_scalar_mode_supported_p): Likewise.
4817 (ia64_vector_mode_supported_p): Likewise.
4818 (expand_vec_perm_broadcast): Likewise.
4819 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
4820 (iq2000_function_arg_advance): Likewise.
4821 (iq2000_function_arg): Likewise.
4822 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
4823 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
4824 (m68k_libcall_value): Likewise.
4825 (m68k_function_value): Likewise.
4826 (sched_attr_op_type): Likewise.
4827 * config/mcore/mcore.c (mcore_output_move): Likewise.
4828 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
4830 (microblaze_function_arg): Likewise.
4831 * config/mips/mips.c (mips16_build_call_stub): Likewise.
4832 (mips_print_operand): Likewise.
4833 (mips_mode_ok_for_mov_fmt_p): Likewise.
4834 (mips_vector_mode_supported_p): Likewise.
4835 (mips_preferred_simd_mode): Likewise.
4836 (mips_expand_vpc_loongson_even_odd): Likewise.
4837 (mips_expand_vec_unpack): Likewise.
4838 (mips_expand_vi_broadcast): Likewise.
4839 (mips_expand_vector_init): Likewise.
4840 (mips_expand_vec_reduc): Likewise.
4841 (mips_expand_msa_cmp): Likewise.
4842 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
4843 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
4844 (cc_flags_for_mode): Likewise.
4845 * config/msp430/msp430.c (msp430_print_operand): Likewise.
4846 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
4847 (nds32_output_casesi_pc_relative): Likewise.
4848 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4849 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
4850 (nvptx_gen_unpack): Likewise.
4851 (nvptx_gen_pack): Likewise.
4852 (nvptx_gen_shuffle): Likewise.
4853 (nvptx_gen_wcast): Likewise.
4854 (nvptx_preferred_simd_mode): Likewise.
4855 * config/pa/pa.c (pa_secondary_reload): Likewise.
4856 * config/pa/predicates.md (base14_operand): Likewise.
4857 * config/powerpcspe/powerpcspe-c.c
4858 (altivec_resolve_overloaded_builtin): Likewise.
4859 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
4861 (rs6000_preferred_simd_mode): Likewise.
4862 (output_vec_const_move): Likewise.
4863 (rs6000_expand_vector_extract): Likewise.
4864 (rs6000_split_vec_extract_var): Likewise.
4865 (reg_offset_addressing_ok_p): Likewise.
4866 (rs6000_legitimate_offset_address_p): Likewise.
4867 (rs6000_legitimize_address): Likewise.
4868 (rs6000_emit_set_const): Likewise.
4869 (rs6000_const_vec): Likewise.
4870 (rs6000_emit_move): Likewise.
4871 (spe_build_register_parallel): Likewise.
4872 (rs6000_darwin64_record_arg_recurse): Likewise.
4873 (swap_selector_for_mode): Likewise.
4874 (spe_init_builtins): Likewise.
4875 (paired_init_builtins): Likewise.
4876 (altivec_init_builtins): Likewise.
4877 (do_load_for_compare): Likewise.
4878 (rs6000_generate_compare): Likewise.
4879 (rs6000_expand_float128_convert): Likewise.
4880 (emit_load_locked): Likewise.
4881 (emit_store_conditional): Likewise.
4882 (rs6000_output_function_epilogue): Likewise.
4883 (rs6000_handle_altivec_attribute): Likewise.
4884 (rs6000_function_value): Likewise.
4885 (emit_fusion_gpr_load): Likewise.
4886 (emit_fusion_p9_load): Likewise.
4887 (emit_fusion_p9_store): Likewise.
4888 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
4889 (fusion_gpr_mem_load): Likewise.
4890 (fusion_addis_mem_combo_load): Likewise.
4891 (fusion_addis_mem_combo_store): Likewise.
4892 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
4893 (fusion_gpr_mem_load): Likewise.
4894 (fusion_addis_mem_combo_load): Likewise.
4895 (fusion_addis_mem_combo_store): Likewise.
4896 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4898 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
4899 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4900 (rs6000_preferred_simd_mode): Likewise.
4901 (output_vec_const_move): Likewise.
4902 (rs6000_expand_vector_extract): Likewise.
4903 (rs6000_split_vec_extract_var): Likewise.
4904 (reg_offset_addressing_ok_p): Likewise.
4905 (rs6000_legitimate_offset_address_p): Likewise.
4906 (rs6000_legitimize_address): Likewise.
4907 (rs6000_emit_set_const): Likewise.
4908 (rs6000_const_vec): Likewise.
4909 (rs6000_emit_move): Likewise.
4910 (rs6000_darwin64_record_arg_recurse): Likewise.
4911 (swap_selector_for_mode): Likewise.
4912 (paired_init_builtins): Likewise.
4913 (altivec_init_builtins): Likewise.
4914 (rs6000_expand_float128_convert): Likewise.
4915 (emit_load_locked): Likewise.
4916 (emit_store_conditional): Likewise.
4917 (rs6000_output_function_epilogue): Likewise.
4918 (rs6000_handle_altivec_attribute): Likewise.
4919 (rs6000_function_value): Likewise.
4920 (emit_fusion_gpr_load): Likewise.
4921 (emit_fusion_p9_load): Likewise.
4922 (emit_fusion_p9_store): Likewise.
4923 * config/rx/rx.c (rx_gen_move_template): Likewise.
4924 (flags_from_mode): Likewise.
4925 * config/s390/predicates.md (s390_alc_comparison): Likewise.
4926 (s390_slb_comparison): Likewise.
4927 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
4928 (s390_vector_mode_supported_p): Likewise.
4929 (s390_cc_modes_compatible): Likewise.
4930 (s390_match_ccmode_set): Likewise.
4931 (s390_canonicalize_comparison): Likewise.
4932 (s390_emit_compare_and_swap): Likewise.
4933 (s390_branch_condition_mask): Likewise.
4934 (s390_rtx_costs): Likewise.
4935 (s390_secondary_reload): Likewise.
4936 (__SECONDARY_RELOAD_CASE): Likewise.
4937 (s390_expand_cs): Likewise.
4938 (s390_preferred_simd_mode): Likewise.
4939 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
4940 * config/sh/sh.c (sh_print_operand): Likewise.
4941 (dump_table): Likewise.
4942 (sh_secondary_reload): Likewise.
4943 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4944 * config/sh/sh.md (casesi_worker_1): Likewise.
4945 (casesi_worker_2): Likewise.
4946 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
4947 (fcc_comparison_operator): Likewise.
4948 * config/sparc/sparc.c (sparc_expand_move): Likewise.
4949 (emit_soft_tfmode_cvt): Likewise.
4950 (sparc_preferred_simd_mode): Likewise.
4951 (output_cbranch): Likewise.
4952 (sparc_print_operand): Likewise.
4953 (sparc_expand_vec_perm_bmask): Likewise.
4954 (vector_init_bshuffle): Likewise.
4955 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4956 (spu_vector_mode_supported_p): Likewise.
4957 (spu_expand_insv): Likewise.
4958 (spu_emit_branch_or_set): Likewise.
4959 (spu_handle_vector_attribute): Likewise.
4960 (spu_builtin_splats): Likewise.
4961 (spu_builtin_extract): Likewise.
4962 (spu_builtin_promote): Likewise.
4963 (spu_expand_sign_extend): Likewise.
4964 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4965 (tilegx_simd_int): Likewise.
4966 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
4967 (tilepro_simd_int): Likewise.
4968 * config/v850/v850.c (const_double_split): Likewise.
4969 (v850_print_operand): Likewise.
4970 (ep_memory_offset): Likewise.
4971 * config/vax/vax.c (vax_rtx_costs): Likewise.
4972 (vax_output_int_move): Likewise.
4973 (vax_output_int_add): Likewise.
4974 (vax_output_int_subtract): Likewise.
4975 * config/visium/predicates.md (visium_branch_operator): Likewise.
4976 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
4977 (visium_print_operand_address): Likewise.
4978 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4979 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
4980 (xtensa_expand_conditional_branch): Likewise.
4981 (xtensa_copy_incoming_a7): Likewise.
4982 (xtensa_output_literal): Likewise.
4983 * dfp.c (decimal_real_maxval): Likewise.
4984 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4987 Alan Hayward <alan.hayward@arm.com>
4988 David Sherwood <david.sherwood@arm.com>
4990 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4991 (mode_nunits_inline): Likewise.
4992 (mode_inner_inline): Likewise.
4993 (mode_unit_size_inline): Likewise.
4994 (mode_unit_precision_inline): Likewise.
4995 (emit_insn_modes_h): Likewise. Also emit a #define of the
4997 (emit_mode_wider): Add an E_ prefix to mode names.
4998 (emit_mode_complex): Likewise.
4999 (emit_mode_inner): Likewise.
5000 (emit_mode_adjustments): Likewise.
5001 (emit_mode_int_n): Likewise.
5002 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
5003 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
5004 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
5005 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
5006 (CRC32_BUILTIN, ENTRY): Likewise.
5007 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
5008 (aarch64_pop_regs): Likewise.
5009 (aarch64_process_components): Likewise.
5010 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
5011 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
5012 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
5013 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
5014 * config/arm/arm.c (arm_init_libfuncs): Likewise.
5015 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
5017 * config/i386/i386-builtin.def (pcmpestr): Likewise.
5018 (pcmpistr): Likewise.
5019 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
5020 * config/mmix/mmix.c (mmix_output_condition): Likewise.
5021 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
5023 * config/rl78/rl78.c (mduc_regs): Likewise.
5024 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5025 (htm_expand_builtin): Likewise.
5026 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
5027 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
5028 * config/xtensa/xtensa.c (print_operand): Likewise.
5029 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
5030 (NUM_MODE_VECTOR_INT): Likewise.
5031 * genoutput.c (null_operand): Likewise.
5032 (output_operand_data): Likewise.
5033 * genrecog.c (print_parameter_value): Likewise.
5034 * lra.c (debug_operand_data): Likewise.
5036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5038 * dwarf2out.c (scompare_loc_descriptor_wide)
5039 (scompare_loc_descriptor_narrow): New functions, split out from...
5040 (scompare_loc_descriptor): ...here.
5041 * expmed.c (emit_store_flag_int): New function, split out from...
5042 (emit_store_flag): ...here.
5044 2017-08-30 Richard Biener <rguenther@suse.de>
5046 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
5047 (dwarf2out_early_finish): Move setting of AT_pubnames from
5048 early debug output to early finish.
5050 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5052 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
5053 and -mdata-region to the assembler.
5055 2017-08-30 Richard Biener <rguenther@suse.de>
5057 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
5059 (gen_subprogram_die): Add DW_AT_object_pointer only early.
5060 (dwarf2out_early_global_decl): Only generate a DIE for the
5061 abstract origin if it doesn't already exist or is a declaration DIE.
5062 (resolve_addr): Do not add the linkage name twice when
5063 generating a stub DIE for the DW_TAG_GNU_call_site target.
5065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5067 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5068 Use machine_mode rather than int for arg1_mode.
5070 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
5073 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
5074 that the second argument of the built-in functions to unpack
5075 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
5076 switch statement instead a lot of if statements.
5077 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
5078 Allow 64-bit values to be in Altivec registers as well as
5079 traditional floating point registers.
5080 (pack<mode>, FMOVE128_VSX iterator): Likewise.
5082 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
5084 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
5085 MAX_REGS_PER_ADDRESS == 1.
5087 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
5089 * config/i386/i386.opt (flag_fentry): Do not init to -1.
5090 * config/i386/i386.c (ix86_option_override_internal): Simplify
5091 setting of opts->x_flag_entry.
5093 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5094 Jakub Jelinek <jakub@redhat.com>
5095 Richard Biener <rguenther@suse.de>
5097 PR tree-optimization/81503
5098 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
5099 folded constant fits in the target type; reorder tests for clarity.
5101 2017-08-29 Martin Liska <mliska@suse.cz>
5103 * passes.def: Include pass_lower_switch.
5104 * stmt.c (dump_case_nodes): Remove and move to
5105 tree-switch-conversion.
5106 (case_values_threshold): Likewise.
5107 (expand_switch_as_decision_tree_p): Likewise.
5108 (emit_case_decision_tree): Likewise.
5109 (expand_case): Likewise.
5110 (balance_case_nodes): Likewise.
5111 (node_has_low_bound): Likewise.
5112 (node_has_high_bound): Likewise.
5113 (node_is_bounded): Likewise.
5114 (emit_case_nodes): Likewise.
5115 (struct simple_case_node): New struct.
5116 (add_case_node): Remove.
5117 (emit_case_dispatch_table): Use vector instead of case_list.
5118 (reset_out_edges_aux): Remove.
5119 (compute_cases_per_edge): Likewise.
5120 (expand_case): Build list of simple_case_node.
5121 (expand_sjlj_dispatch_table): Use it.
5122 * tree-switch-conversion.c (struct case_node): Moved from
5123 stmt.c and adjusted.
5124 (emit_case_nodes): Likewise.
5125 (node_has_low_bound): Likewise.
5126 (node_has_high_bound): Likewise.
5127 (node_is_bounded): Likewise.
5128 (case_values_threshold): Likewise.
5129 (reset_out_edges_aux): Likewise.
5130 (compute_cases_per_edge): Likewise.
5131 (add_case_node): Likewise.
5132 (dump_case_nodes): Likewise.
5133 (balance_case_nodes): Likewise.
5134 (expand_switch_as_decision_tree_p): Likewise.
5135 (emit_jump): Likewise.
5136 (emit_case_decision_tree): Likewise.
5137 (try_switch_expansion): Likewise.
5138 (do_jump_if_equal): Likewise.
5139 (emit_cmp_and_jump_insns): Likewise.
5140 (fix_phi_operands_for_edge): New function.
5141 (record_phi_operand_mapping): Likewise.
5142 (class pass_lower_switch): New pass.
5143 (pass_lower_switch::execute): New function.
5144 (make_pass_lower_switch): Likewise.
5145 (conditional_probability):
5146 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
5147 * tree-pass.h: Add make_pass_lower_switch.
5149 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5152 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
5155 2017-08-29 Richard Biener <rguenther@suse.de>
5157 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
5158 we do not add a DW_AT_inline attribute twice.
5159 (gen_subprogram_die): Remove code setting DW_AT_inline on
5160 DECL_ABSTRACT_P nodes.
5162 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
5164 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
5165 calls to internal functions.
5166 (gimplify_modify_expr): Likewise.
5167 * tree-call-cdce.c (use_internal_fn): Likewise.
5168 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5169 (convert_to_divmod): Set the nothrow flag.
5170 * tree-if-conv.c (predicate_mem_writes): Likewise.
5171 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5172 (vectorizable_call): Likewise.
5173 (vectorizable_store): Likewise.
5174 (vectorizable_load): Likewise.
5175 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
5176 (vect_recog_mask_conversion_pattern): Likewise.
5178 2017-08-29 Martin Liska <mliska@suse.cz>
5181 * gcc.c (driver_handle_option): Add new argument.
5182 * opts-common.c (handle_option): Pass
5183 target_option_override_hook.
5184 * opts-global.c (lang_handle_option): Add new option.
5185 (set_default_handlers): Add new argument.
5186 (decode_options): Likewise.
5187 * opts.c (target_handle_option): Likewise.
5188 (common_handle_option): Call target_option_override_hook.
5189 * opts.h (struct cl_option_handler_func): Add hook for
5190 target option override.
5191 (struct cl_option_handlers): Likewise.
5192 (set_default_handlers): Add new argument.
5193 (decode_options): Likewise.
5194 (common_handle_option): Likewise.
5195 (target_handle_option): Likewise.
5196 * toplev.c (toplev::main): Pass targetm.target_option.override
5199 2017-08-29 Richard Biener <rguenther@suse.de>
5200 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5202 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
5203 life to the active subtree.
5205 2017-08-28 Jeff Law <law@redhat.com>
5207 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
5208 derive_equivalences.
5209 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
5211 (edge_info::derive_equivalences): New private member function
5213 * tree-ssa-dom.c (class edge_info): Changed from a struct
5214 to a class. Add ctor/dtor, methods and data members.
5215 (edge_info::edge_info): Renamed from allocate_edge_info.
5216 Initialize additional members.
5217 (edge_info::~edge_info): New.
5218 (free_dom_edge_info): Delete the edge info.
5219 (record_edge_info): Use new class & associated member functions.
5220 Tighten forms for testing for edge equivalences.
5221 (record_temporary_equivalences): Iterate over the simple
5222 equivalences rather than assuming there's only one per edge.
5223 (cprop_into_successor_phis): Iterate over the simple
5224 equivalences rather than assuming there's only one per edge.
5225 (optimize_stmt): Use operand_equal_p rather than pointer
5226 equality for mini-DSE code.
5228 2017-08-28 Nathan Sidwell <nathan@acm.org>
5230 * gcc.c (execute): Fold SIGPIPE handling into switch
5231 statement. Adjust internal error message.
5233 2017-08-28 Richard Biener <rguenther@suse.de>
5236 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
5237 Do nothing for removed DIEs.
5239 2017-08-28 Richard Biener <rguenther@suse.de>
5241 PR tree-optimization/81977
5242 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
5245 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
5248 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
5249 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
5250 using targetm.gen_mem_thread_fence.
5252 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
5255 * config/i386/i386.md (*<btsc><mode>): Change operand 2
5256 predicate to register_operand. Reorder operands.
5257 (*btr<mode>): Ditto.
5258 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
5259 (*btr<mode>_mask): Ditto.
5261 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
5263 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
5264 * config/rs6000/xmmintrin.h: New file.
5265 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
5267 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5270 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
5271 parameter and_insn and return it.
5272 (recombine_lvx_pattern): Insert a copy to ensure availability of
5273 the base register of the copied masking operation at the point of
5274 the instruction replacement.
5275 (recombine_stvx_pattern): Likewise.
5277 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5279 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
5280 undocumented switches.
5281 (-mpower9-dform-vector): Likewise.
5282 (-mpower9-dform): Likewise.
5283 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
5284 comments to delete references to -mpower9-dform* switches.
5285 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
5286 Delete reference to -mpower9-dform* switches, test for
5287 -mpower9-vector instead.
5288 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
5289 (OTHER_P9_VECTOR_MASKS): Likewise.
5290 (POWERPC_MASKS): Likewise.
5291 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
5292 tests against -mpower9-dform* to -mpower9-vector. Delete code
5293 that checked for -mpower9-dform* consistancy with other options.
5294 Add test for -mpower9-misc to enable other power9 switches.
5295 (rs6000_init_hard_regno_mode_ok): Likewise.
5296 (rs6000_option_override_internal): Likewise.
5297 (rs6000_emit_prologue): Likewise.
5298 (rs6000_emit_epilogue): Likewise.
5299 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
5300 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
5301 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
5303 (emit_fusion_p9_store): Likewise.
5304 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
5305 resetting these macros if the assembler does not support ISA 3.0
5307 (TARGET_P9_DFORM_VECTOR): Likewise.
5308 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
5309 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
5311 2017-08-25 Alan Modra <amodra@gmail.com>
5313 PR rtl-optimization/81747
5314 * cse.c (cse_extended_basic_block): Don't attempt to record
5315 equivalences for degenerate conditional branches that branch
5316 to their fall-through.
5318 2017-08-24 Martin Sebor <msebor@redhat.com>
5321 * gimple-fold.c (size_must_be_zero_p): New function.
5322 (gimple_fold_builtin_memory_op): Call it.
5324 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
5326 * config/rs6000/mm_malloc.h: New file.
5328 2017-08-24 Bin Cheng <bin.cheng@arm.com>
5330 PR tree-optimization/81913
5331 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
5332 analysis when either IVs in condition can wrap.
5334 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
5336 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
5337 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
5339 2017-08-24 Richard Biener <rguenther@suse.de>
5342 * targhooks.c (default_target_can_inline_p): Properly
5343 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
5344 is present and always compare.
5345 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
5346 infer -mfpmath=sse from TARGET_SSE_P.
5347 (ix86_can_inline_p): Properly use target_option_default_node when
5348 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
5350 2017-08-24 Richard Biener <rguenther@suse.de>
5353 * dwarf2out.c (output_die): Handle flag_generate_offload like
5355 (output_comp_unit): Likewise.
5356 (gen_array_type_die): Likewise.
5357 (dwarf2out_early_finish): Likewise.
5358 (note_variable_value_in_expr): Likewise.
5359 (dwarf2out_finish): Likewise. Adjust assert.
5360 * cgraphunit.c (symbol_table::compile): Move setting of
5361 flag_generate_offload earlier ...
5362 (symbol_table::finalize_compilation_unit): ... here, before
5363 early debug finalization.
5365 2017-08-24 Richard Biener <rguenther@suse.de>
5367 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
5368 and ipa-fnsummary.h.
5369 (ix86_can_inline_p): When ix86_fpmath flags do not match
5370 check whether the callee uses FP math at all.
5372 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
5375 * tree-ssanames.c (get_nonzero_bits): Use element_precision
5376 instead of TYPE_PRECISION.
5378 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
5379 Alan Hayward <alan.hayward@arm.com>
5380 David Sherwood <david.sherwood@arm.com>
5382 * combine.c (make_extraction): Use subreg_offset_from_lsb.
5384 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
5386 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
5388 (ix86_frame::stack_realign_allocate): New field.
5389 (struct machine_frame_state): Modify comments.
5390 (machine_frame_state::sp_realigned_fp_end): New field.
5391 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
5393 (sp_valid_at): Add assertion to assure no attempt to access invalid
5394 offset of a realigned stack.
5395 (fp_valid_at): Likewise.
5396 (choose_baseaddr): Modify comments.
5397 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
5398 ix86_expand_prologue.
5399 (ix86_expand_prologue): Modify stack realignment and allocation.
5400 (ix86_expand_epilogue): Modify comments.
5401 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5402 avx2_runtime, avx512f, and avx512f_runtime.
5404 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5406 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5407 (mstackrealign): Do not init to -1.
5408 * config/i386/i386.c (ix86_option_override_internal):
5409 Check opts_set, not opts when setting default value of
5410 opts->x_ix86_force_align_arg_pointer.
5412 2017-08-23 Richard Biener <rguenther@suse.de>
5414 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5415 lang_hooks.decl_printable_name.
5416 * print-rtl-function.c (print_rtx_function): Likewise.
5417 * tree-pretty-print.c (dump_function_header): Likewise.
5419 2017-08-23 Richard Biener <rguenther@suse.de>
5422 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5423 -g0 at compile-time.
5425 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5428 * doc/sourcebuild.texi (Other hardware attributes):
5431 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5434 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5435 Fix single-use check.
5437 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5439 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5441 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5443 * doc/install.texi: Modify to add more details on running selected
5446 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5448 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5449 is combined with -mabi=ms.
5450 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5453 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5455 PR tree-optimization/81488
5456 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5457 and cached_basis fields.
5458 (MAX_SPREAD): New constant.
5459 (alloc_cand_and_find_basis): Initialize new fields.
5460 (clear_visited): New function.
5461 (create_phi_basis_1): Rename from create_phi_basis, set visited
5462 and cached_basis fields.
5463 (create_phi_basis): New wrapper function.
5464 (phi_add_costs_1): Rename from phi_add_costs, add spread
5465 parameter, set visited field, short-circuit when limits reached.
5466 (phi_add_costs): New wrapper function.
5467 (record_phi_increments_1): Rename from record_phi_increments, set
5469 (record_phi_increments): New wrapper function.
5470 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5471 (phi_incr_cost): New wrapper function.
5472 (all_phi_incrs_profitable_1): Rename from
5473 all_phi_incrs_profitable, set visited field.
5474 (all_phi_incrs_profitable): New wrapper function.
5476 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5477 Alan Hayward <alan.hayward@arm.com>
5478 David Sherwood <david.sherwood@arm.com>
5480 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5481 that takes the outer and inner modes.
5482 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5483 comparison as the canonical test for a paradoxical subreg.
5484 * combine.c (simplify_set): Use paradoxical_subreg_p.
5485 (make_extraction): Likewise.
5486 (force_to_mode): Likewise.
5487 (rtx_equal_for_field_assignment_p): Likewise.
5488 (gen_lowpart_for_combine): Likewise.
5489 (simplify_comparison): Likewise.
5490 * cse.c (equiv_constant): Likewise.
5491 * expmed.c (store_bit_field_1): Likewise.
5492 * final.c (alter_subreg): Likewise.
5493 * fwprop.c (propagate_rtx): Likewise.
5494 (forward_propagate_subreg): Likewise.
5495 * ira-conflicts.c (ira_build_conflicts): Likewise.
5496 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5497 * lra-constraints.c (curr_insn_transform): Likewise.
5498 (split_reg): Likewise.
5499 * lra-eliminations.c (move_plus_up): Likewise.
5500 (lra_eliminate_regs_1): Likewise.
5501 * recog.c (general_operand): Likewise.
5502 * ree.c (combine_reaching_defs): Likewise.
5503 * reload.c (push_reload): Likewise.
5504 (find_reloads): Likewise.
5505 * reload1.c (elimination_effects): Likewise.
5506 (compute_reload_subreg_offset): Likewise.
5507 (choose_reload_regs): Likewise.
5508 * rtlanal.c (subreg_lsb_1): Likewise.
5509 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5510 (simplify_subreg): Likewise.
5511 * var-tracking.c (track_loc_p): Likewise.
5512 * emit-rtl.c (byte_lowpart_offset): Likewise.
5513 (paradoxical_subreg_p): Delete out-of-line definition.
5515 2017-08-22 Jeff Law <law@redhat.com>
5517 PR tree-optimization/81741
5518 PR tree-optimization/71947
5519 * tree-ssa-dom.c: Include tree-inline.h.
5520 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5521 equivalences if one is more expensive to compute than the other.
5522 * tree-ssa-scopedtables.h (class const_or_copies): Make
5523 record_const_or_copy_raw method private.
5524 (class avail_exprs_stack): New method simplify_binary_operation.
5525 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5526 avail_exprs_stack::simplify_binary_operation as needed.
5527 (avail_exprs_stack::simplify_binary_operation): New function.
5529 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5531 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5532 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5533 (DOT_SYMBOLS): Likewise.
5534 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5535 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5536 (RS6000_ABI_NAME): Likewise.
5537 (TARGET_CMODEL): Likewise.
5538 (TOC_SECTION_ASM_OP): Likewise.
5539 (SET_CMODEL): New macro.
5540 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5542 2017-08-22 Richard Biener <rguenther@suse.de>
5544 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5545 to allow for free-lang-data replacements similar to verify_type_variant.
5547 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5548 Alan Hayward <alan.hayward@arm.com>
5549 David Sherwood <david.sherwood@arm.com>
5551 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5552 VOIDmode for the LABEL_REF.
5554 2017-08-22 Richard Biener <rguenther@suse.de>
5556 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5558 2017-08-22 Richard Biener <rguenther@suse.de>
5560 * common.opt (feliminate-dwarf2-dups): Ignore.
5561 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5562 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5563 same_die_p_wrap, compute_section_prefix,
5564 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5565 (comdat_symbol_id, comdat_symbol_number): Likewise.
5566 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5568 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5569 (output_die): Mark unreachable path unreachable.
5570 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5571 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5572 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5573 (dwarf2out_early_finish): Likewise.
5575 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5577 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5579 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5582 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5583 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5584 (avr_attribute_table) <io, io_low, address>: Initialize
5585 .decl_required with true.
5587 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5589 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5590 undocumented debugging options.
5591 (-mvsx-scalar-double): Likewise.
5592 (-mallow-df-permute): Likewise.
5593 (-mvectorize-builtins): Likewise.
5594 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5595 (rs6000_builtin_vectorized_function): Likewise.
5596 (rs6000_builtin_md_vectorized_function): Likewise.
5597 (rs6000_opt_vars): Likewise.
5599 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5602 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5603 (*btrq_imm): Rename from *btrq.
5604 (*btcq_imm): Rename from *btcq.
5605 (btsc): New code attribute.
5606 (*<btsc><mode>): New insn pattern.
5607 (*btr<mode>): Ditto.
5608 (*<btsc><mode>_mask): New insn_and_split pattern.
5609 (*btr<mode>_mask): Ditto.
5611 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5612 Alan Hayward <alan.hayward@arm.com>
5613 David Sherwood <david.sherwood@arm.com>
5615 * function.c (pad_below): Simplify padding calculation.
5617 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5618 Alan Hayward <alan.hayward@arm.com>
5619 David Sherwood <david.sherwood@arm.com>
5621 * target.def (function_prologue): Remove frame size argument.
5622 (function_epilogue): Likewise.
5623 * doc/tm.texi: Regenerate.
5624 * final.c (final_start_function): Update call to function_prologue.
5625 (final_end_function): Update call to function_epilogue.
5626 (default_function_pro_epilogue): Remove frame size argument.
5627 * output.h (default_function_pro_epilogue): Likewise.
5628 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5629 (arm_output_function_prologue): Likewise.
5630 * config/frv/frv.c (frv_function_prologue): Likewise.
5631 (frv_function_epilogue): Likewise.
5632 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5633 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5634 (ia64_output_function_epilogue): Likewise.
5635 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5636 (m32r_output_function_epilogue): Likewise.
5637 * config/microblaze/microblaze.c (microblaze_function_prologue)
5638 (microblaze_function_epilogue): Likewise.
5639 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5640 (mips_output_function_epilogue): Likewise.
5641 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5642 (mmix_target_asm_function_epilogue): Likewise.
5643 * config/msp430/msp430.c (msp430_start_function): Likewise.
5644 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
5645 (nds32_asm_function_epilogue): Likewise.
5646 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
5647 * config/pa/pa.c (pa_output_function_prologue): Likewise.
5648 (pa_output_function_epilogue): Likewise.
5649 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
5650 (rs6000_output_function_epilogue): Likewise.
5651 * config/rl78/rl78.c (rl78_start_function): Likewise.
5652 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
5653 (rs6000_output_function_epilogue): Likewise.
5654 * config/rx/rx.c (rx_output_function_prologue): Likewise.
5655 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
5656 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
5657 (sparc_asm_function_epilogue): Likewise.
5659 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5661 * tree.h (type_has_mode_precision_p): New function.
5662 * convert.c (convert_to_integer_1): Use it.
5663 * expr.c (expand_expr_real_2): Likewise.
5664 (expand_expr_real_1): Likewise.
5665 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5666 * match.pd: Likewise.
5667 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5668 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5669 * tree-tailcall.c (process_assignment): Likewise.
5670 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5671 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5672 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5673 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5674 (vectorizable_assignment): Likewise.
5675 (vectorizable_shift): Likewise.
5676 (vectorizable_operation): Likewise.
5677 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5679 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5681 * match.pd: Add pow (C, x) simplification.
5683 2017-08-21 Richard Biener <rguenther@suse.de>
5685 PR tree-optimization/81900
5686 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5687 for blocks with abnormal predecessors.
5688 (compute_antic): Do not set visited flag prematurely.
5690 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5693 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5695 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5697 * stor-layout.h (vector_type_mode): Move to...
5698 * tree.h (vector_type_mode): ...here.
5699 * stor-layout.c (vector_type_mode): Move to...
5700 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5702 2017-08-21 Richard Biener <rguenther@suse.de>
5704 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5705 register_external_die hooks.
5706 (debug_false_tree_charstarstar_uhwistar): Declare.
5707 (debug_nothing_tree_charstar_uhwi): Likewise.
5708 * debug.c (do_nothing_debug_hooks): Adjust.
5709 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5710 (debug_nothing_tree_charstar_uhwi): Likewise.
5711 * dbxout.c (dbx_debug_hooks): Adjust.
5712 (xcoff_debug_hooks): Likewise.
5713 * sdbout.c (sdb_debug_hooks): Likewise.
5714 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5715 * dwarf2out.c (macinfo_label_base): New global.
5716 (dwarf2out_register_external_die): New function for the
5717 register_external_die hook.
5718 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5719 (dwarf2_debug_hooks): Use them.
5720 (dwarf2_lineno_debug_hooks): Adjust.
5721 (struct die_struct): Add with_offset flag.
5722 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5723 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5724 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5725 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5726 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5727 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5728 defining section names for the early LTO debug variants.
5729 (reset_indirect_string): New helper.
5730 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5731 (print_dw_val): Add support for offsetted symbol references.
5732 (get_ultimate_context): Split out from is_cxx.
5733 (is_cxx): Use get_ultimate_context.
5734 (is_fortran): Add decl overload.
5735 (compute_comp_unit_symbol): Split out worker from
5736 compute_section_prefix.
5737 (compute_section_prefix): Call compute_comp_unit_symbol and
5738 set comdat_type_p here.
5739 (output_die): Skip DIE symbol output for the LTO added one.
5740 Handle DIE symbol references with offset.
5741 (output_comp_unit): Guard section name mangling properly.
5742 For LTO debug sections emit a symbol at the section beginning
5743 which we use to refer to its DIEs.
5744 (add_abstract_origin_attribute): For DIEs registered via
5745 dwarf2out_register_external_die directly refer to the early
5746 DIE rather than indirectly through the shadow one we created.
5747 Remove obsolete call to dwarf2out_abstract_function for
5748 non-function/block origins.
5749 (gen_array_type_die): When generating early LTO debug do
5750 not emit DW_AT_string_length.
5751 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
5752 late when in LTO. As suggested place a gcc_unreachable for
5753 the DECL_ABSTRACT_P case.
5754 (gen_subprogram_die): Avoid another specification DIE
5755 for early built declarations/definitions for the late LTO case.
5756 (gen_variable_die): Add type references for late duplicated VLA dies
5758 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
5759 we have the abstract instance already.
5760 (process_scope_var): Adjust decl DIE contexts in LTO which
5761 first puts them in limbo.
5762 (gen_decl_die): Do not generate type DIEs late apart from
5763 types for VLAs or for decls we do not yet have a DIE. Do not
5764 call dwarf2out_abstract_function late.
5765 (dwarf2out_early_global_decl): Make sure to create DIEs
5766 for abstract instances of a decl first.
5767 (dwarf2out_late_global_decl): Adjust comment.
5768 (output_macinfo_op): With multiple macro sections use
5769 macinfo_label_base to distinguish labels.
5770 (output_macinfo): Likewise. Update macinfo_label_base.
5771 Pass in the line info label.
5772 (note_variable_value_in_expr): When generating LTO resolve
5773 all variable values here by generating DIEs as needed.
5774 (init_sections_and_labels): Add early LTO debug flag parameter
5775 and generate different sections and names if set. Add generation
5776 counter for the labels so we can have multiple of them.
5777 (reset_dies): Helper to allow DIEs to be output multiple times.
5778 (dwarf2out_finish): When outputting DIEs to the fat part of an
5779 LTO object first reset DIEs.
5780 (dwarf2out_early_finish): Output early DIEs when generating LTO.
5781 (modified_type_die): Check for decl_ultimate_origin being self
5783 (gen_type_die_with_usage): Likewise.
5784 (gen_typedef_die): Allow decl_ultimate_origin being self.
5785 (set_decl_abstract_flags): Remove.
5786 (set_block_abstract_flags): Likewise.
5787 (dwarf2out_abstract_function): Treat the early generated DIEs
5788 as the abstract copy and only add DW_AT_inline and
5789 DW_AT_artificial here and call set_decl_origin_self.
5790 If the DIE has an abstract origin don't do anything.
5791 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
5792 if we have none yet (Go fails to build one, PR78628).
5793 (variably_modified_type_p): Prevent endless recursion for Ada
5794 cyclic pointer types.
5795 * lto-streamer-in.c: Include debug.h.
5796 (dref_queue): New global.
5797 (lto_read_tree_1): Stream in DIE references.
5798 (lto_input_tree): Register DIE references.
5799 (input_function): Stream DECL_DEBUG_ARGS.
5800 * lto-streamer-out.c: Include debug.h.
5801 (lto_write_tree_1): Output DIE references.
5802 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
5803 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
5804 (output_function): Stream DECL_DEBUG_ARGS.
5805 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
5806 Stream DECL_ABSTRACT_ORIGIN.
5807 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
5808 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
5809 DECL_CONTEXT for file-scope decls.
5810 * lto-streamer.h (struct dref_entry): Declare.
5811 (dref_queue): Likewise.
5812 * cfgexpand.c (pass_expand::execute): Do not call the
5813 outlining_inline_function hook here.
5814 * lto-wrapper.c (debug_obj): New global.
5815 (tool_cleanup): Unlink it if required.
5816 (debug_objcopy): New function.
5817 (run_gcc): Handle early debug sections in the IL files by
5818 extracting them to separate files, partially linkin them and
5819 feeding the result back as result to the linker.
5820 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
5821 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
5822 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
5823 sections into a separate segment.
5824 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
5826 (darwin_asm_dwarf_section): Likewise.
5827 (darwin_asm_output_dwarf_offset): Likewise.
5828 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
5830 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5831 Alan Hayward <alan.hayward@arm.com>
5832 David Sherwood <david.sherwood@arm.com>
5834 * read-md.h (md_reader::record_potential_iterator_use): Replace
5835 pointer argument with an rtx and an index.
5836 * read-rtl.c (iterator_group::apply_iterator): Likewise.
5837 (apply_mode_iterator): Likewise.
5838 (apply_code_iterator): Likewise.
5839 (apply_int_iterator): Likewise.
5840 (apply_subst_iterator): Likewise.
5841 (record_iterator_use): Likewise.
5842 (record_attribute_use): Likewise.
5843 (md_reader::record_potential_iterator_use): Likewise. Update calls
5844 to record_iterator_use and apply_iterator.
5845 (iterator_use): Replace ptr with x and index.
5846 (attribute_use): Likewise.
5847 (apply_attribute_uses): Update calls to apply_iterator.
5848 (apply_iterators): Likewise. Update initialization of iterator_use.
5849 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
5850 and record_potential_iterator_use.
5851 (rtx_reader::read_rtx_operand): Likewise.
5853 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5854 Alan Hayward <alan.hayward@arm.com>
5855 David Sherwood <david.sherwood@arm.com>
5857 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
5860 2017-08-21 Richard Biener <rguenther@suse.de>
5863 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
5864 at struct end conservatively when comparing common bases.
5866 2017-08-21 Richard Biener <rguenther@suse.de>
5868 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
5869 (mem_ref_in_stmt): Remove.
5870 (determine_max_movement): Use ref index to get at the reference.
5871 (invariantness_dom_walker::before_dom_children): Deal with
5872 lim data already initialized.
5873 (gather_mem_refs_stmt): Initialize lim data and record ref index.
5875 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5877 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
5878 (TARGET_ISA_ROUND): Ditto.
5879 (TARGET_ROUND): Ditto.
5880 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
5881 * config/i386/i386.md: Ditto.
5882 * config/i386/sse.md: Ditto.
5883 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
5884 with OPTION_MASK_ISA_SSE4_1.
5886 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5889 * doc/extend.texi (x86 Built-in Functions): Correct the name of
5890 __builtin_ia32_lzcnt_u16.
5892 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5895 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
5896 (rs6000_set_current_function): Rewrite function to use it.
5898 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
5901 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
5902 and TYPE_WARN_IF_NOT_ALIGN.
5903 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
5904 (handle_warn_if_not_align): New.
5905 (place_union_field): Call handle_warn_if_not_align.
5906 (place_field): Call handle_warn_if_not_align.
5907 Copy TYPE_WARN_IF_NOT_ALIGN.
5908 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
5909 (layout_type): Likewise.
5910 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
5912 (tree_decl_common): Add warn_if_not_align.
5913 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
5914 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
5915 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
5916 (DECL_WARN_IF_NOT_ALIGN): Likewise.
5917 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
5918 * doc/extend.texi: Document warn_if_not_aligned attribute.
5919 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
5921 2017-08-17 Martin Liska <mliska@suse.cz>
5924 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
5925 (get_data_dependence): Use it as pointer type.
5926 (distribute_loop): Likewise.
5928 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5930 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
5931 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
5932 (p8_vmrgow_<mode>_direct): New define_insn.
5933 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
5934 handle endianness for vmrgew and vmrgow permute patterns.
5936 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5938 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
5939 * config/rs6000/rs6000-cpus.def: Remove comment.
5940 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
5941 (POWERPC_MASKS): Likewise.
5942 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
5943 use of TARGET_VSX_TIMODE.
5944 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
5945 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
5946 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
5947 (rs6000_option_override_internal): Remove dead code.
5948 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
5949 (rs6000_legitimize_reload_address): Likewise.
5950 (rs6000_legitimate_address_p): Likewise.
5951 (rs6000_opt_masks): Delete "vsx-timode".
5952 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
5953 from function comment.
5954 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
5955 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
5956 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
5958 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
5959 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
5960 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
5963 (define_peephole2): Likewise.
5965 2017-08-17 Martin Sebor <msebor@redhat.com>
5968 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
5969 past the end of an array.
5970 (test_pp_format): Add test cases.
5972 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
5974 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
5975 missing ECF_NOTHROW flags.
5977 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5980 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
5981 operands residing in integer registers.
5982 (*vsx_le_perm_load_<mode>): Likewise.
5983 (*vsx_le_perm_store_<mode>): Likewise.
5984 (define_peephole2): Add peepholes to optimize the above.
5986 2017-08-17 Marek Polacek <polacek@redhat.com>
5989 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5990 to mimic what shorten_compare did. Change the return type to bool.
5991 (fold_cond_expr_with_comparison): Update call to
5992 operand_equal_for_comparison_p.
5993 (fold_ternary_loc): Likewise.
5995 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5997 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5999 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
6001 2017-08-17 Richard Biener <rguenther@suse.de>
6003 * tree-ssa-structalias.c (solve_graph): When propagating
6004 to successors update the graphs succ edges and avoid duplicate work.
6006 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
6009 * config/i386/i386.c (ix86_option_override_internal): Save target
6010 specific options after ix86_stack_protector_guard_reg was changed.
6012 2017-08-17 Richard Biener <rguenther@suse.de>
6014 PR tree-optimization/81827
6015 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
6016 (new_var_info): Initialize it conservatively.
6017 (get_call_vi): Mark register vars.
6018 (new_scalar_tmp_constraint_exp): Likewise.
6019 (handle_rhs_call): Likewise.
6020 (handle_const_call): Likewise.
6021 (create_function_info_for): Likewise.
6022 (solve_constraints): Sort varinfos to separate register from
6023 non-register vars to pack points-to solution bitmaps during
6026 2017-08-17 Marek Polacek <polacek@redhat.com>
6028 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
6030 2017-08-17 Richard Biener <rguenther@suse.de>
6032 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
6033 to true when overflow is undefined and we saturated the result.
6035 2017-08-17 Alan Modra <amodra@gmail.com>
6038 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
6039 Don't use store multiple if only one reg needs saving.
6040 (interesting_frame_related_regno): New function.
6041 (rs6000_frame_related): Don't emit frame info for regs that
6043 (rs6000_emit_epilogue): Likewise.
6045 2017-08-16 Nathan Sidwell <nathan@acm.org>
6047 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
6048 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
6049 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
6050 * tree.c (free_lang_data_in_type): Use else-if chain. Always
6051 clear TYPE_LANG_1. Remove obsolete member-function stripping.
6052 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
6053 (verify_type): Adjust for TYPE_BINFO move.
6054 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
6055 process TYPE_BINFO directly.
6056 (hash_tree): Likewise.
6057 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6059 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6062 2017-08-16 David Malcolm <dmalcolm@redhat.com>
6064 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
6066 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
6069 * config/i386/i386.md (*anddi_1_btr): Change predicates of
6070 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
6071 Add ix86_binary_operator_ok to insn constraint.
6072 (*iordi_1_bts): Ditto.
6073 (*xordi_1_btc): Ditto.
6074 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
6075 Update corresponding peephole2 pattern.
6079 2017-08-16 Bin Cheng <bin.cheng@arm.com>
6081 PR tree-optimization/81832
6082 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
6083 copy loop header which has IFN_LOOP_DIST_ALIAS call.
6085 2017-08-16 Marek Polacek <polacek@redhat.com>
6088 * fold-const.c (fold_indirect_ref_1): Restore original behavior
6089 regarding size_zero_node.
6091 2017-08-16 Martin Liska <mliska@suse.cz>
6094 * config.gcc: Respect previously set extra_objs in case
6097 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
6099 PR tree-optimization/81835
6100 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
6101 the phi SSA_NAME. Check that the condition in a COND_EXPR does
6102 not depend on the phi.
6104 2017-08-16 Alan Modra <amodra@gmail.com>
6106 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
6109 2017-08-16 Alan Modra <amodra@gmail.com>
6111 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
6112 (save_reg_p): ..into this. Update all callers.
6113 (first_reg_to_save): Simplify.
6115 2017-08-16 Alan Modra <amodra@gmail.com>
6117 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
6120 2017-08-15 Joseph Myers <joseph@codesourcery.com>
6124 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
6125 constant count if that count is less than 32.
6127 2017-08-15 Nathan Sidwell <nathan@acm.org>
6129 * gcc.c (execute): Emit friendlier message if inferior is killed
6130 by an external cause.
6132 2017-08-15 Richard Biener <rguenther@suse.de>
6134 PR tree-optimization/81790
6135 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
6136 CONSTRUCTORs from simplifying and VN.
6138 2017-08-14 Martin Sebor <msebor@redhat.com>
6140 * builtin-attrs.def: Add comments.
6142 2017-08-14 Martin Sebor <msebor@redhat.com>
6145 * doc/extend.texi (attribute nonstring): Document new attribute.
6147 2017-08-14 Martin Sebor <msebor@redhat.com>
6150 * tree-diagnostic.c (default_tree_printer): Handle %G.
6151 * gimple-pretty-print.h (percent_G_format): Declare new function.
6152 * gimple-pretty-print.c (percent_G_format): Define.
6153 * tree-pretty-print.c (percent_K_format): Add argument.
6155 2017-08-14 Martin Sebor <msebor@redhat.com>
6157 PR translation/79998
6158 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
6159 Remove a stray space.
6161 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
6164 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
6165 (*iordi_1_bts): Ditto.
6166 (*xordi_1_btc): Ditto.
6168 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6171 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
6172 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6174 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
6175 quoted strings, and make more translator-friendly.
6176 (darwin_rs6000_override_options): Likewise.
6177 (rs6000_option_override_internal): Likewise.
6178 (rs6000_return_in_memory): Fix overlong line.
6179 (init_cmulative_args): Use quoted strings, and make more
6180 translator-friendly.
6181 (rs6000_pass_by_reference): Fix overlong line.
6182 (def_builtin): Use quoted strings.
6183 (altivec_expand_predicate_builtin): Use quoted strings, and make
6184 more translator-friendly.
6185 (htm_expand_builtin): Use quoted strings.
6186 (cpu_expand_builtin): Use quoted strings, and make more
6187 translator-friendly.
6188 (altivec_expand_builtin): Likewise.
6189 (paired_expand_predicate_builtin): Likewise.
6190 (rs6000_invalid_builtin): Likewise.
6191 (builtin_function_type): Use quoted strings.
6192 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
6193 more translator-friendly.
6194 (rs6000_trampoline_init): Likewise.
6195 (rs6000_handle_altivec_attribute): Likewise.
6196 (rs6000_inner_target_options): Use quoted strings.
6197 (rs6000_disable_incompatible_switches): Likewise.
6198 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
6199 strings, and make more translator-friendly.
6200 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
6202 2017-08-14 Bin Cheng <bin.cheng@arm.com>
6204 PR tree-optimization/81799
6205 * tree-loop-distribution.c (version_loop_by_alias_check): Force
6206 cond_expr to simple gimple operand.
6208 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
6211 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
6213 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
6217 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
6218 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
6219 TARGET_GASISR_PROLOGUES.
6220 * config/avr/avr.c (avr_option_override): Same.
6221 (avr_pass_pre_proep::execute): Same.
6223 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
6226 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
6227 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
6229 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
6231 * config/i386/i386.md (*load_tp_<mode>): Redefine as
6232 define_insn_and_split. Split to a memory load from 0 in
6233 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
6234 using PTR mode iterator.
6235 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
6236 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
6237 (*add_tp_<mode>): Redefine as define_insn_and_split.
6238 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
6239 address space. Merge with *add_tp_x32 using PTR mode iterator.
6240 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
6241 Split to an add with a memory load from 0 in
6242 DEFAULT_TLS_SEG_REG address space.
6244 2017-08-12 Andrew Pinski <apinski@cavium.com>
6246 * config/aarch64/aarch64-option-extensions.def (rdma):
6247 Fix feature string to what Linux prints out in /proc/cpuinfo.
6249 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
6252 * dwarf2out.c (modified_type_die): For C typedef types that have
6253 an ultimate origin, process the ultimate origin instead of the
6255 (gen_typedef_die): Assert that input DECLs have no ultimate
6257 (gen_type_die_with_usage): For typedef variants that have an
6258 ultimate origin, just call gen_decl_die on the original DECL.
6259 (process_scope_var): Avoid creating DIEs for local typedefs and
6260 concrete static variables.
6262 2017-08-12 Alan Modra <amodra@gmail.com>
6266 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
6267 match gnu-user.h startfile.
6268 (ENDFILE_LINUX_SPEC): Similarly.
6270 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
6273 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
6275 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
6277 2017-08-11 Tamar Christina <tamar.christina@arm.com>
6278 * config/aarch64/aarch64.md (mov<mode>): Change.
6279 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6280 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
6281 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
6283 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
6285 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
6286 for storage order barriers.
6288 2017-08-11 Martin Liska <mliska@suse.cz>
6291 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
6292 variables of void type.
6294 2017-08-11 Martin Liska <mliska@suse.cz>
6296 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
6297 TARGET_SUPPORTS_ALIASES.
6298 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
6299 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
6300 (optimize_weakref): Likewise.
6301 * symtab.c (symtab_node::noninterposable_alias): Likewise.
6302 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
6303 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
6305 2017-08-11 Martin Liska <mliska@suse.cz>
6308 * config/i386/i386.c (make_resolver_func): Do complete
6309 refactoring of the function.
6311 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6314 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
6315 * config/i386/i386.c (ix86_stack_protect_guard): Use
6316 ix86_stack_protect_guard_symbol_str to generate varible declaration.
6317 * doc/invoke.texi (x86 Options): Document
6318 -mstack-protector-guard-symbol= option.
6320 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6322 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
6323 * config/i386/i386.c (ix86_split_stack_guard): New function.
6324 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
6325 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
6326 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
6327 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
6328 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
6329 (split_stack_space_check): Call ix86_split_stack_guard.
6331 2017-08-10 Martin Sebor <msebor@redhat.com>
6333 * print-tree.c (print_node): Print location using the established
6335 Replace spaces with colons.
6336 (debug_raw, debug): Ditto.
6338 2017-08-10 Martin Sebor <msebor@redhat.com>
6341 * pretty-print.c (pp_format): Correct the handling of %s precision.
6343 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
6346 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6348 (ix86_finalize_stack_frame_flags): This. Also clear
6349 frame_pointer_needed if -fno-omit-frame-pointer is used without
6351 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6352 with ix86_finalize_stack_frame_flags.
6353 (ix86_expand_epilogue): Likewise.
6354 (ix86_expand_split_stack_prologue): Likewise.
6355 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
6357 2017-08-10 Martin Liska <mliska@suse.cz>
6360 * c-attribs.c (handle_target_attribute):
6361 Report warning for an empty string argument of target attribute.
6363 2017-08-09 Jakub Jelinek <jakub@redhat.com>
6366 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
6368 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
6369 or DECL_NONLOCAL labels.
6370 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
6371 or DECL_NONLOCAL labels here.
6373 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
6375 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
6376 to indicate when early gimple folding has been disabled.
6377 (rs6000_gimple_fold_builtin): Add debug content.
6378 (rs6000_invalid_builtin): Fix whitespace.
6379 (rs6000_expand_builtin): Fix whitespace.
6380 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
6382 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
6385 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
6386 SAVE_MULTIPLE if not all the registers that saves, should be saved.
6388 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
6390 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
6391 (qdf24xx): Likewise.
6392 * config/aarch64/aarch64.md: Include falkor.md.
6393 * config/aarch64/falkor.md: New.
6395 2017-08-09 Marek Polacek <polacek@redhat.com>
6398 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
6399 * diagnostic.c (emit_diagnostic): Add a comment.
6400 (emit_diagnostic_valist): New function.
6402 2017-08-09 Marek Polacek <polacek@redhat.com>
6405 * input.c (make_location): New overload.
6406 * input.h (make_location): Declare.
6408 2017-08-08 Alan Modra <amodra@gmail.com>
6409 H.J. Lu <hongjiu.lu@intel.com>
6412 * gcc.c (NO_PIE_SPEC): Delete.
6413 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6415 (LINK_PIE_SPEC): ..to here.
6416 (LINK_COMMAND_SPEC): Support -no-pie.
6417 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6418 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6420 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6421 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6422 (ENDFILE_CRTEND_SPEC): Similarly.
6424 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6427 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6428 (mstack-protector-guard-offset=): Ditto.
6429 * config/i386/i386.c (ix86_option_override): Handle
6430 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6432 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6433 ix86_stack_protect_guard_offset variables.
6434 (TARGET_STACK_PROTECT_GUARD): Always define.
6435 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6436 and -mstack-protector-guard-offset= options.
6438 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6440 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6441 boundary case for the last candidate.
6443 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6445 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6446 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6448 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6451 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6452 * config/aarch64/aarch64-builtins.c
6453 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6454 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6455 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6457 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6458 Andrew Pinski <pinskia@gmail.com>
6461 * internal-fn.def (XORSIGN): New.
6462 * optabs.def (xorsign_optab): New.
6463 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6464 (convert_expand_mult_copysign): New.
6465 (pass_optimize_widening_mul::execute): Call
6466 convert_expand_mult_copysign.
6468 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6470 PR tree-optimization/81354
6471 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6472 Insert on edges rather than explicitly creating landing pads.
6473 (analyze_candidates_and_replace): Commit edge inserts.
6475 2017-08-08 Richard Biener <rguenther@suse.de>
6478 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6479 (expand_simple_operations): Also look through ADDR_EXPRs with
6480 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6482 2017-08-08 Richard Biener <rguenther@suse.de>
6484 PR tree-optimization/81723
6485 * tree-vect-slp.c (struct bst_traits): New hash traits.
6486 (bst_fail): New global.
6487 (vect_build_slp_tree_2): New worker, split out from ...
6488 (vect_build_slp_tree): ... this now wrapping it with using
6489 bst_fail set to cache SLP tree build fails. Properly handle
6491 (vect_analyze_slp_instance): Allocate and free bst_fail.
6493 2017-08-08 Martin Liska <mliska@suse.cz>
6496 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6497 LABEL_DECLs that can be from a different function.
6499 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6501 PR tree-optimization/81744
6502 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6503 loop's number of iterations.
6505 2017-08-08 Martin Liska <mliska@suse.cz>
6507 * asan.c: Include header files.
6508 * attribs.c (build_decl_attribute_variant): New function moved
6510 (build_type_attribute_qual_variant): Likewise.
6511 (cmp_attrib_identifiers): Likewise.
6512 (simple_cst_list_equal): Likewise.
6513 (omp_declare_simd_clauses_equal): Likewise.
6514 (attribute_value_equal): Likewise.
6515 (comp_type_attributes): Likewise.
6516 (build_type_attribute_variant): Likewise.
6517 (lookup_ident_attribute): Likewise.
6518 (remove_attribute): Likewise.
6519 (merge_attributes): Likewise.
6520 (merge_type_attributes): Likewise.
6521 (merge_decl_attributes): Likewise.
6522 (merge_dllimport_decl_attributes): Likewise.
6523 (handle_dll_attribute): Likewise.
6524 (attribute_list_equal): Likewise.
6525 (attribute_list_contained): Likewise.
6526 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6527 (lookup_attribute_by_prefix): Likewise.
6528 * bb-reorder.c: Include header files.
6529 * builtins.c: Likewise.
6530 * calls.c: Likewise.
6531 * cfgexpand.c: Likewise.
6532 * cgraph.c: Likewise.
6533 * cgraphunit.c: Likewise.
6534 * convert.c: Likewise.
6535 * dwarf2out.c: Likewise.
6536 * final.c: Likewise.
6537 * fold-const.c: Likewise.
6538 * function.c: Likewise.
6539 * gimple-expr.c: Likewise.
6540 * gimple-fold.c: Likewise.
6541 * gimple-pretty-print.c: Likewise.
6542 * gimple.c: Likewise.
6543 * gimplify.c: Likewise.
6544 * hsa-common.c: Likewise.
6545 * hsa-gen.c: Likewise.
6546 * internal-fn.c: Likewise.
6547 * ipa-chkp.c: Likewise.
6548 * ipa-cp.c: Likewise.
6549 * ipa-devirt.c: Likewise.
6550 * ipa-fnsummary.c: Likewise.
6551 * ipa-inline.c: Likewise.
6552 * ipa-visibility.c: Likewise.
6554 * lto-cgraph.c: Likewise.
6555 * omp-expand.c: Likewise.
6556 * omp-general.c: Likewise.
6557 * omp-low.c: Likewise.
6558 * omp-offload.c: Likewise.
6559 * omp-simd-clone.c: Likewise.
6560 * opts-global.c: Likewise.
6561 * passes.c: Likewise.
6562 * predict.c: Likewise.
6563 * sancov.c: Likewise.
6564 * sanopt.c: Likewise.
6565 * symtab.c: Likewise.
6566 * toplev.c: Likewise.
6567 * trans-mem.c: Likewise.
6568 * tree-chkp.c: Likewise.
6569 * tree-eh.c: Likewise.
6570 * tree-into-ssa.c: Likewise.
6571 * tree-object-size.c: Likewise.
6572 * tree-parloops.c: Likewise.
6573 * tree-profile.c: Likewise.
6574 * tree-ssa-ccp.c: Likewise.
6575 * tree-ssa-live.c: Likewise.
6576 * tree-ssa-loop.c: Likewise.
6577 * tree-ssa-sccvn.c: Likewise.
6578 * tree-ssa-structalias.c: Likewise.
6579 * tree-ssa.c: Likewise.
6580 * tree-streamer-in.c: Likewise.
6581 * tree-vectorizer.c: Likewise.
6582 * tree-vrp.c: Likewise.
6584 * ubsan.c: Likewise.
6585 * varasm.c: Likewise.
6586 * varpool.c: Likewise.
6587 * tree.c: Remove functions moved to attribs.[ch].
6589 * config/aarch64/aarch64.c: Add attrs.h header file.
6590 * config/alpha/alpha.c: Likewise.
6591 * config/arc/arc.c: Likewise.
6592 * config/arm/arm.c: Likewise.
6593 * config/avr/avr.c: Likewise.
6594 * config/bfin/bfin.c: Likewise.
6595 * config/c6x/c6x.c: Likewise.
6596 * config/cr16/cr16.c: Likewise.
6597 * config/cris/cris.c: Likewise.
6598 * config/darwin.c: Likewise.
6599 * config/epiphany/epiphany.c: Likewise.
6600 * config/fr30/fr30.c: Likewise.
6601 * config/frv/frv.c: Likewise.
6602 * config/ft32/ft32.c: Likewise.
6603 * config/h8300/h8300.c: Likewise.
6604 * config/i386/winnt.c: Likewise.
6605 * config/ia64/ia64.c: Likewise.
6606 * config/iq2000/iq2000.c: Likewise.
6607 * config/lm32/lm32.c: Likewise.
6608 * config/m32c/m32c.c: Likewise.
6609 * config/m32r/m32r.c: Likewise.
6610 * config/m68k/m68k.c: Likewise.
6611 * config/mcore/mcore.c: Likewise.
6612 * config/microblaze/microblaze.c: Likewise.
6613 * config/mips/mips.c: Likewise.
6614 * config/mmix/mmix.c: Likewise.
6615 * config/mn10300/mn10300.c: Likewise.
6616 * config/moxie/moxie.c: Likewise.
6617 * config/msp430/msp430.c: Likewise.
6618 * config/nds32/nds32-isr.c: Likewise.
6619 * config/nds32/nds32.c: Likewise.
6620 * config/nios2/nios2.c: Likewise.
6621 * config/nvptx/nvptx.c: Likewise.
6622 * config/pa/pa.c: Likewise.
6623 * config/pdp11/pdp11.c: Likewise.
6624 * config/powerpcspe/powerpcspe.c: Likewise.
6625 * config/riscv/riscv.c: Likewise.
6626 * config/rl78/rl78.c: Likewise.
6627 * config/rx/rx.c: Likewise.
6628 * config/s390/s390.c: Likewise.
6629 * config/sh/sh.c: Likewise.
6630 * config/sol2.c: Likewise.
6631 * config/sparc/sparc.c: Likewise.
6632 * config/spu/spu.c: Likewise.
6633 * config/stormy16/stormy16.c: Likewise.
6634 * config/tilegx/tilegx.c: Likewise.
6635 * config/tilepro/tilepro.c: Likewise.
6636 * config/v850/v850.c: Likewise.
6637 * config/vax/vax.c: Likewise.
6638 * config/visium/visium.c: Likewise.
6639 * config/xtensa/xtensa.c: Likewise.
6641 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6644 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
6645 constraints since the -mupper-regs-* switches have been
6647 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
6648 into a vector from a double word element that was extracted from
6649 another vector, and eliminate extra XXPERMDI instructions.
6650 (vsx_concat_<mode>_2): Likewise.
6651 (vsx_concat_<mode>_3): Likewise.
6652 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
6653 concat to allow optimizing inserts from previous extracts.
6655 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
6657 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6658 memory reference to a SSP offset in TLS address space.
6659 (ix86_print_operand) <case '@'>: Remove.
6660 (ix86_print_operand_punct_valid_p): Remove '@' code.
6661 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6663 (stack_tls_protect_set_<mode>): Remove.
6664 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6665 (stack_tls_protect_test_<mode>): Remove.
6666 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6668 2017-08-07 Olivier Hainque <hainque@adacore.com>
6671 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6673 2017-08-07 Douglas Rupp <rupp@adacore.com>
6675 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6676 variable was referenced as multidir in command.
6678 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6681 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6684 2017-08-07 Martin Liska <mliska@suse.cz>
6686 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6687 * config/rl78/rl78.c: Add include of attribs.h.
6688 * config/sh/sh.c: Likewise.
6689 * config/v850/v850.c: Likewise.
6691 2017-08-07 Tom de Vries <tom@codesourcery.com>
6694 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6696 2017-08-07 Martin Liska <mliska@suse.cz>
6698 * config/mips/mips.c: Include attribs.h.
6700 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6703 * doc/invoke.texi: Document change in behvaior for -Ofast for
6706 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6708 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6710 (aarch64_pop_regs): Likewise.
6711 (aarch64_gen_load_pair): Likewise.
6712 (aarch64_save_callee_saves): Likewise.
6713 (aarch64_restore_callee_saves): Likewise.
6715 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6717 * config/i386/i386.c: Revert the last change.
6719 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6722 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6724 (ix86_finalize_stack_frame_flags): This. Also clear
6725 frame_pointer_needed if -fno-omit-frame-pointer is used without
6727 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6728 with ix86_finalize_stack_frame_flags.
6729 (ix86_expand_epilogue): Likewise.
6730 (ix86_expand_split_stack_prologue): Likewise.
6732 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6735 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6736 to P_AES for Westmere.
6738 2017-08-07 Jonathan Yong <10walls@gmail.com>
6740 * config/i386/mingw.opt (fset-stack-executable): Removed.
6741 * config/i386/cygming.opt (fset-stack-executable): Moved
6743 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
6745 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
6747 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
6749 2017-08-07 Marek Polacek <polacek@redhat.com>
6752 * fold-const.c (fold_indirect_ref_1): Check type_domain.
6754 2017-08-07 Martin Liska <mliska@suse.cz>
6756 * attribs.h (canonicalize_attr_name): New function.
6757 (cmp_attribs): Move from c-format.c and adjusted.
6758 (is_attribute_p): Moved from tree.h.
6759 * tree-inline.c: Add new includes.
6760 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
6761 (private_is_attribute_p): Remove.
6762 (private_lookup_attribute): Likewise.
6763 (private_lookup_attribute_by_prefix): Simplify.
6764 (remove_attribute): Use is_attribute_p.
6765 * tree.h: Remove removed declarations.
6767 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6770 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
6771 instead of computing it in the function. Formatting fix.
6772 (expand_case): Don't rely on default_edge being the first edge,
6773 clear it if removing it, pass default_edge to
6774 emit_case_dispatch_table.
6775 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
6778 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
6780 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
6781 insn in the function, emit NOP after the insn.
6783 2017-08-06 Tom de Vries <tom@codesourcery.com>
6785 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
6788 2017-08-06 Tom de Vries <tom@codesourcery.com>
6790 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
6793 2017-08-04 Yury Gribov <tetra2005@gmail.com>
6795 PR tree-optimization/57371
6796 * match.pd: New pattern.
6798 2017-08-04 Marek Polacek <polacek@redhat.com>
6801 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
6802 perform the computation in offset_int.
6804 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6806 PR tree-optimization/81136
6807 * tree-vectorizer.h: Include tree-hash-traits.h.
6808 (vec_base_alignments): New typedef.
6809 (vec_info): Add a base_alignments field.
6810 (vect_record_base_alignments): Declare.
6811 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
6813 (DR_IS_CONDITIONAL_IN_STMT): New macro.
6814 (create_data_ref): Add an is_conditional_in_stmt argument.
6815 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
6816 the is_conditional_in_stmt field.
6817 (data_ref_loc): Add an is_conditional_in_stmt field.
6818 (get_references_in_stmt): Set the is_conditional_in_stmt field.
6819 (find_data_references_in_stmt): Update call to create_data_ref.
6820 (graphite_find_data_references_in_stmt): Likewise.
6821 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
6822 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
6823 (vect_record_base_alignment): New function.
6824 (vect_record_base_alignments): Likewise.
6825 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
6826 for nested statements even if we fail to compute a misalignment.
6827 Use pooled base alignments for unconditional references.
6828 (vect_find_same_alignment_drs): Compare base addresses instead
6830 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
6831 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
6833 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6835 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
6836 Add an explicit name for the enum. Use auto_vec for slp_instances
6838 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
6840 (_bb_vec_info): Add a constructor and destructor.
6841 (vinfo_for_stmt): Return NULL for uids of -1 as well.
6842 (destroy_loop_vec_info): Delete.
6843 (vect_destroy_datarefs): Likewise.
6844 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
6845 (vec_info::vec_info): New function.
6846 (vec_info::~vec_info): Likewise.
6847 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
6848 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
6849 destroy_loop_vec_info.
6850 * tree-vect-loop.c (new_loop_vec_info): Replace with...
6851 (_loop_vec_info::_loop_vec_info): ...this.
6852 (destroy_loop_vec_info): Replace with...
6853 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
6854 the stmt_vec_infos. Leave handling of vec_info information to its
6855 destructor. Remove explicit vector releases.
6856 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
6857 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
6858 * tree-vect-slp.c (new_bb_vec_info): Replace with...
6859 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
6860 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
6861 (destroy_bb_vec_info): Replace with...
6862 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
6863 information to its destructor.
6864 (vect_slp_analyze_bb_1): Use new and delete instead of
6865 new_bb_vec_info and destroy_bb_vec_info.
6866 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
6869 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6871 * tree-data-ref.h (subscript): Add access_fn field.
6872 (data_dependence_relation): Add could_be_independent_p.
6873 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
6874 (same_access_functions): Move to tree-data-ref.c.
6875 * tree-data-ref.c (ref_contains_union_access_p): New function.
6876 (access_fn_component_p): Likewise.
6877 (access_fn_components_comparable_p): Likewise.
6878 (dr_analyze_indices): Add a reference to access_fn_component_p.
6879 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
6881 (constant_access_functions): Likewise.
6882 (add_other_self_distances): Likewise.
6883 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
6884 (initialize_data_dependence_relation): Use XCNEW and remove
6885 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
6886 of access functions that have the same type. Allow the
6887 subsequence to end with different bases in some circumstances.
6888 Record the chosen access functions in SUB_ACCESS_FN.
6889 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
6890 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
6891 (subscript_dependence_tester_1): Likewise dra and drb.
6892 (build_classic_dist_vector): Update calls accordingly.
6893 (subscript_dependence_tester): Likewise.
6894 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
6895 DDR_COULD_BE_INDEPENDENT_P.
6896 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
6897 comp_alias_ddrs instead of may_alias_ddrs.
6898 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6900 (vect_analyze_data_ref_dependence): Use it if
6901 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
6902 distance vectors if that fails.
6903 (dependence_distance_ge_vf): New function.
6904 (vect_prune_runtime_alias_test_list): Use it. Don't clear
6905 LOOP_VINFO_MAY_ALIAS_DDRS.
6907 2017-08-04 Richard Biener <rguenther@suse.de>
6910 * fold-const.c (fold_binary_loc): Properly restrict
6911 minus_var0 && minus_var1 case when associating undefined overflow
6914 2017-08-04 Tom de Vries <tom@codesourcery.com>
6916 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
6918 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6920 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6921 Don't start diagnostic messages with a capital letter.
6922 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6924 (rs6000_invalid_builtin): Likewise.
6925 (rs6000_trampoline_init): Likewise.
6927 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6930 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
6931 after setting changeable df flags.
6933 2017-08-03 Richard Biener <rguenther@suse.de>
6935 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
6936 up if the use is in USE - X.
6938 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
6940 * toplev.c (dumpfile.h): New include.
6941 (internal_error_reentered): New static function. Use it...
6942 (internal_error_function): ...here to handle reentered internal_error.
6944 2017-08-03 Richard Biener <rguenther@suse.de>
6947 * fold-const.c (split_tree): Add minus_var and minus_con
6948 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
6949 here but always use minus_*.
6950 (associate_trees): Assert we never associate with MINUS_EXPR
6951 and NULL first operand. Do not recurse for PLUS_EXPR operands
6952 when associating as MINUS_EXPR either.
6953 (fold_binary_loc): Track minus_var and minus_con.
6955 2017-08-03 Tom de Vries <tom@codesourcery.com>
6958 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
6959 ACCEL_COMPILER, apply finish_options on
6960 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
6962 2017-08-03 Tom de Vries <tom@codesourcery.com>
6965 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
6966 function_entry_patch_area_size > 0.
6968 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6971 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
6972 instead of 10??LU, perform unit multiplication in wide_int,
6973 don't change alloc_object_size_limit if the limit is larger
6976 PR tree-optimization/81655
6977 PR tree-optimization/81588
6978 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
6979 the case when ranges[i].low and high are 1 for unsigned type with
6983 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
6984 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6986 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6988 * tree-vrp.h: Add include guard.
6990 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6993 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6994 (ud2): New insn pattern.
6995 * config/i386/i386.c (ix86_expand_epilogue):
6996 For naked functions, generate ud2 instead of trap insn.
6998 2017-08-02 Marek Polacek <polacek@redhat.com>
7001 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
7003 2017-08-02 Tom de Vries <tom@codesourcery.com>
7004 Cesar Philippidis <cesar@codesourcery.com>
7006 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
7007 Add missing edge probabilities.
7009 2017-08-02 Tamar Christina <tamar.christina@arm.com>
7011 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
7014 2017-08-02 Jakub Jelinek <jakub@redhat.com>
7017 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
7018 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
7019 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
7020 prologue_seq sequences - if any.
7022 2017-08-02 Richard Biener <rguenther@suse.de>
7024 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
7025 via vectors if supported, integer extracts via punning if supported
7026 or otherwise vector extracts.
7028 2017-08-02 Richard Biener <rguenther@suse.de>
7030 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
7032 (bitmap_insert_into_set): ... this.
7034 2017-08-02 Richard Biener <rguenther@suse.de>
7036 PR tree-optimization/81633
7038 2015-08-17 Alan Hayward <alan.hayward@arm.com>
7040 PR tree-optimization/71752
7041 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
7043 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
7045 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
7046 (machine_function::call_ms2sysv_pad_out): Remove field.
7047 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
7048 (ix86_compute_frame_layout): Likewise.
7050 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
7053 * config/i386/i386.c (ix86_set_func_type): Disallow naked
7054 attribute with interrupt attribute.
7056 2017-08-01 Andrew Pinski <apinski@cavium.com>
7058 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
7059 BIT_INSERT_EXPR's operand 1
7060 to see if the types precision matches.
7062 2017-08-01 Martin Liska <mliska@suse.cz>
7065 * builtins.c (expand_builtin_memcpy_args): Remove.
7066 (expand_builtin_memcpy): Call newly added function
7067 expand_builtin_memory_copy_args.
7068 (expand_builtin_memcpy_with_bounds): Likewise.
7069 (expand_builtin_mempcpy): Remove last argument.
7070 (expand_builtin_mempcpy_with_bounds): Likewise.
7071 (expand_builtin_memory_copy_args): New function created from
7072 expand_builtin_mempcpy_args with small modifications.
7073 (expand_builtin_mempcpy_args): Remove.
7074 (expand_builtin_stpcpy): Remove unused argument.
7075 (expand_builtin): Likewise.
7076 (expand_builtin_with_bounds): Likewise.
7078 2017-08-01 Martin Liska <mliska@suse.cz>
7081 Make mempcpy more optimal (PR middle-end/70140).
7083 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7086 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7087 __builtin_vec_cmpne verify both arguments are compatible vectors
7088 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
7089 verify arg1_type is a pointer or array type. For __builtin_vec_st,
7090 move computation of aligned to after checking the argument types.
7094 * config/rs6000/vsx.md (vextract_fp_from_shorth,
7095 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
7098 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
7099 Doug Rupp <rupp@adacore.com>
7100 Olivier Hainque <hainque@adacore.com>
7102 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
7103 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
7105 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
7106 for TARGET_OS_CPP_BUILTIN.
7107 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
7108 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
7110 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
7111 passing required abi options to the assembler for EABI configurations.
7112 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
7113 of .text.hot and .text.unlikely sections for kernel modules when
7114 using ARM style exceptions.
7115 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
7116 options. Add EXTRA_CC1_SPEC.
7117 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
7119 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
7121 (ARM_TARGET2_DWARF_FORMAT): Define.
7122 * config/arm/t-vxworks: Adjust multilib control to removal of the
7123 Diab command line options.
7125 2017-08-01 Martin Liska <mliska@suse.cz>
7127 PR gcov-profile/81561
7128 * gcov.c (unblock): Make unblocking safe as we need to preserve
7129 index correspondence of blocks and block_lists.
7131 2017-08-01 Richard Biener <rguenther@suse.de>
7133 PR tree-optimization/81181
7134 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7135 (compute_antic): ... end of iteration here.
7137 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
7139 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
7140 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
7141 (ftree-slp-vectorize): Likewise.
7142 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
7143 can no longer be set independent of flag_tree_loop_vectorize.
7144 * omp-general.c (emp_max_vf): Likewise.
7145 * opts.c (enable_fdo_optimizations): Remove references to
7146 flag_tree_vectorize, these are now implicit.
7147 (common_handle_option): Remove handling for OPT_ftree_vectorize,
7148 and leave it for the options machinery.
7150 2017-08-01 Martin Liska <mliska@suse.cz>
7153 * builtins.c (expand_builtin_memcpy_args): Remove.
7154 (expand_builtin_memcpy): Call newly added function
7155 expand_builtin_memory_copy_args.
7156 (expand_builtin_memcpy_with_bounds): Likewise.
7157 (expand_builtin_mempcpy): Remove last argument.
7158 (expand_builtin_mempcpy_with_bounds): Likewise.
7159 (expand_builtin_memory_copy_args): New function created from
7160 expand_builtin_mempcpy_args with small modifications.
7161 (expand_builtin_mempcpy_args): Remove.
7162 (expand_builtin_stpcpy): Remove unused argument.
7163 (expand_builtin): Likewise.
7164 (expand_builtin_with_bounds): Likewise.
7166 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7169 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
7170 print "ds:" only for immediates in generic address space.
7172 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7175 * config/i386/i386.c (ix86_funciton_naked): New prototype.
7176 (ix86_function_ok_for_sibcall): Return false for naked functions.
7178 2017-08-01 Richard Biener <rguenther@suse.de>
7180 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
7181 (compute_antic): Seed worklist with exit block predecessors.
7182 * cfganal.c (dfs_find_deadend): For a cycle return the source
7183 of the edge closing it.
7185 2017-08-01 Tamar Christina <tamar.christina@arm.com>
7187 * config/aarch64/aarch64.c
7188 (aarch64_can_const_movi_rtx_p): Move 0 check.
7190 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7192 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
7193 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
7195 * match.pd: Ditto in address comparison pattern.
7197 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7199 PR tree-optimization/81627
7200 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
7201 closed ssa form for store-store chain.
7203 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7205 PR tree-optimization/81620
7206 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
7207 for store-store chain.
7209 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7211 PR tree-optimization/81588
7212 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
7213 ranges[i].in_p, invert comparison code ccode. For >/>=,
7214 swap rhs1 and rhs2 and comparison code unconditionally,
7215 for </<= don't do that. Don't swap rhs1/rhs2 again if
7216 ranges[i].in_p, instead invert comparison code ccode if
7217 opcode or oe->rank is BIT_IOR_EXPR.
7220 * optabs.def (vec_extract_optab, vec_init_optab): Change from
7221 a direct optab to conversion optab.
7222 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7223 with GET_MODE_INNER as last argument instead of optab_handler.
7224 * expmed.c (extract_bit_field_1): Likewise. Use vector from
7225 vector extraction if possible and optab is available.
7226 * expr.c (store_constructor): Use convert_optab_handler instead
7227 of optab_handler. Use vector initialization from smaller
7228 vectors if possible and optab is available.
7229 * tree-vect-stmts.c (vectorizable_load): Likewise.
7230 * doc/md.texi (vec_extract, vec_init): Document that the optabs
7232 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
7233 of vec_init from half-sized vectors with the same element mode.
7234 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
7235 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
7236 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
7237 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
7238 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
7239 after mode in gen_vec_extract* calls.
7240 (vec_extract<mode>): Renamed to ...
7241 (vec_extract<mode><ssescalarmodelower>): ... this.
7242 (vec_extract<mode><ssehalfvecmodelower>): New expander.
7243 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
7244 element mode after mode in gen_vec_init* calls.
7245 (VEC_INIT_HALF_MODE): New mode iterator.
7246 (vec_init<mode>): Renamed to ...
7247 (vec_init<mode><ssescalarmodelower>): ... this.
7248 (vec_init<mode><ssehalfvecmodelower>): New expander.
7249 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
7250 (vec_extractv2sfsf): ... this.
7251 (vec_initv2sf): Renamed to ...
7252 (vec_initv2sfsf): ... this.
7253 (vec_extractv2si): Renamed to ...
7254 (vec_extractv2sisi): ... this.
7255 (vec_initv2si): Renamed to ...
7256 (vec_initv2sisi): ... this.
7257 (vec_extractv4hi): Renamed to ...
7258 (vec_extractv4hihi): ... this.
7259 (vec_initv4hi): Renamed to ...
7260 (vec_initv4hihi): ... this.
7261 (vec_extractv8qi): Renamed to ...
7262 (vec_extractv8qiqi): ... this.
7263 (vec_initv8qi): Renamed to ...
7264 (vec_initv8qiqi): ... this.
7265 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
7266 (vec_init<mode>): Renamed to ...
7267 (vec_init<mode><VEC_base_l>): ... this.
7268 (vec_extract<mode>): Renamed to ...
7269 (vec_extract<mode><VEC_base_l>): ... this.
7270 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
7271 (vec_initv2sfsf): ... this.
7272 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
7273 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7274 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
7275 element mode after mode in gen_vec_init* calls.
7276 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
7277 (vec_init<mode><Vel>): ... this.
7278 (vec_extract<mode>): Renamed to ...
7279 (vec_extract<mode><Vel>): ... this.
7280 * config/aarch64/iterators.md (Vel): New mode attribute.
7281 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
7282 Add element mode after mode in gen_vec_extract* calls.
7283 * config/s390/vector.md (non_vec_l): New mode attribute.
7284 (vec_extract<mode>): Renamed to ...
7285 (vec_extract<mode><non_vec_l>): ... this.
7286 (vec_init<mode>): Renamed to ...
7287 (vec_init<mode><non_vec_l>): ... this.
7288 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
7289 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
7291 * config/arm/iterators.md (V_elem_l): New mode attribute.
7292 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
7293 (vec_extract<mode><V_elem_l>): ... this.
7294 (vec_extractv2di): Renamed to ...
7295 (vec_extractv2didi): ... this.
7296 (vec_init<mode>): Renamed to ...
7297 (vec_init<mode><V_elem_l>): ... this.
7298 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
7299 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
7300 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
7301 Add element mode after gen_vec_extract* calls.
7302 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
7303 (vec_init<mode><unitmode>): ... this.
7304 (vec_extract<mode>): Renamed to ...
7305 (vec_extract<mode><unitmode>): ... this.
7306 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
7307 (vec_init<mode><unitmode>): ... this.
7308 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
7309 (vec_initv2sfsf): ... this.
7310 (vec_extractv2sf): Renamed to ...
7311 (vec_extractv2sfsf): ... this.
7312 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
7313 Add element mode after gen_vec_extract* calls.
7314 * config/mips/mips.md (unitmode): New mode iterator.
7315 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
7316 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
7317 * config/spu/spu.md (inner_l): New mode attribute.
7318 (vec_init<mode>): Renamed to ...
7319 (vec_init<mode><inner_l>): ... this.
7320 (vec_extract<mode>): Renamed to ...
7321 (vec_extract<mode><inner_l>): ... this.
7322 * config/sparc/sparc.md (veltmode): New mode iterator.
7323 (vec_init<VMALL:mode>): Renamed to ...
7324 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
7325 * config/ia64/vect.md (vec_initv2si): Renamed to ...
7326 (vec_initv2sisi): ... this.
7327 (vec_initv2sf): Renamed to ...
7328 (vec_initv2sfsf): ... this.
7329 (vec_extractv2sf): Renamed to ...
7330 (vec_extractv2sfsf): ... this.
7331 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
7332 (vec_init<mode>): Renamed to ...
7333 (vec_init<mode><VEC_base_l>): ... this.
7334 (vec_extract<mode>): Renamed to ...
7335 (vec_extract<mode><VEC_base_l>): ... this.
7336 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
7337 (vec_initv2sfsf): ... this.
7338 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
7339 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7340 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
7341 gen_vec_init* calls.
7343 2017-08-01 Richard Biener <rguenther@suse.de>
7345 PR tree-optimization/81297
7346 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
7347 TREE_OVERFLOW from INTEGER_CSTs.
7349 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
7351 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
7353 2017-07-31 Carl Love <cel@us.ibm.com>
7355 * config/rs6000/rs6000-c: Add support for built-in functions
7356 vector signed char vec_xl_be (signed long long, signed char *);
7357 vector unsigned char vec_xl_be (signed long long, unsigned char *);
7358 vector signed int vec_xl_be (signed long long, signed int *);
7359 vector unsigned int vec_xl_be (signed long long, unsigned int *);
7360 vector signed long long vec_xl_be (signed long long, signed long long *);
7361 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
7362 vector signed short vec_xl_be (signed long long, signed short *);
7363 vector unsigned short vec_xl_be (signed long long, unsigned short *);
7364 vector double vec_xl_be (signed long long, double *);
7365 vector float vec_xl_be (signed long long, float *);
7366 * config/rs6000/altivec.h (vec_xl_be): Add #define.
7367 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
7368 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
7370 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
7371 (altivec_expand_builtin): Add switch statement to call
7372 altivec_expand_xl_be for each builtin.
7373 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
7374 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
7375 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
7376 __builtin_vsx_le_be_v16qi.
7377 * doc/extend.texi: Update the built-in documentation file for the
7378 new built-in functions.
7380 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7383 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
7385 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7387 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7389 * config.gcc: Add z14.
7390 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7391 CPU model numbers for z13s and z14.
7392 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
7394 * config/s390/s390-opts.h (enum processor_type): Rename
7395 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7396 * config/s390/s390.c (processor_table): Add field for CPU name to
7397 be passed to Binutils.
7398 (s390_asm_output_machine_for_arch): Use the new field in
7399 processor_table for Binutils.
7400 (s390_expand_builtin): Replace arch12 with z14.
7401 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7402 (s390_get_sched_attrmask): Likewise.
7403 (s390_get_unit_mask): Likewise.
7404 * config/s390/s390.opt: Add z14 to processor_type enum.
7406 2017-07-31 Martin Jambor <mjambor@suse.cz>
7409 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7410 regardless of optimization level.
7412 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7413 Martin Liska <mliska@suse.cz>
7415 * predict.def: Remove old comment and adjust probability.
7416 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7419 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7422 * config/i386/i386.c (ix86_function_naked): New function.
7423 (ix86_can_use_return_insn_p): Return false for naked functions.
7424 (ix86_expand_prologue): Skip prologue for naked functions.
7425 (ix86_expand_epilogue): Skip epilogue for naked functions
7426 and emit trap instruction.
7427 (ix86_warn_func_return): New function.
7428 (ix86_attribute_table): Add "naked" attribute specification.
7429 (TARGET_WARN_FUNC_RETURN): Define.
7430 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7432 2017-07-31 Martin Liska <mliska@suse.cz>
7434 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7435 (dump_gimple_bb_header): Always dump BB info.
7436 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7438 2017-07-31 Martin Liska <mliska@suse.cz>
7441 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7442 also with current_function_decl non-null equality.
7444 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7447 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7448 change type to the element type, instead add eltype variable and
7449 use it where we are interested in the element type.
7451 PR tree-optimization/81603
7452 * ipa-polymorphic-call.c
7453 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7454 offset arithmetic in offset_int, bail out if the resulting bit offset
7455 doesn't fit into shwi.
7457 2017-07-31 Martin Liska <mliska@suse.cz>
7459 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7460 (gimplify_save_expr): Fix comment.
7462 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7465 * config/i386/i386.c (ix86_function_arg): Update arguments for
7467 (ix86_compute_frame_layout): Set the initial stack offset to
7468 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7469 INCOMING_FRAME_SP_OFFSET.
7470 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7471 stack before exception handler returns.
7472 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7473 the 'ERROR_CODE' for exception handler.
7475 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7477 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7478 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7479 (ASM_OUTPUT_REG_POP): Ditto.
7480 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7481 instead of asm_fprintf to output pure string.
7483 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7485 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7486 to imported_module_or_decl hook.
7487 (debug_nothing_tree_tree_tree_bool): Remove.
7488 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7489 * debug.c (do_nothing_debug_hooks): Use
7490 debug_nothing_tree_tree_tree_bool_bool instead of
7491 debug_nothing_tree_tree_tree_bool.
7492 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7493 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7494 * sdbout.c (sdb_debug_hooks): Likewise.
7495 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7496 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7498 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7499 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7500 attribute, don't add anything.
7502 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7504 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7505 (fold_build2_stat_loc): Likewise.
7506 (fold_build3_stat_loc): Likewise.
7507 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7508 (fold_build1_loc): Remove macro.
7509 (fold_build2_loc): Likewise.
7510 (fold_build3_loc): Likewise.
7512 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7514 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7515 (gimple_build_debug_bind_source_stat): Likewise.
7516 * gimple.h (gimple_build_debug_bind): Remove macro.
7517 (gimple_build_debug_bind_source): Likewise.
7519 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7521 * bitmap.c (bitmap_alloc): Adjust.
7522 (bitmap_gc_alloc): Likewise.
7523 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7525 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7527 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7528 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7529 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7530 (bitmap_gc_alloc_stat): Likewise.
7531 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7533 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7535 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7536 * rtl.h (shallow_copy_rtx): Remove macro.
7538 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7540 * emit-rtl.c (gen_raw_REG): Adjust.
7541 * gengenrtl.c (gendef): Likewise.
7542 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7543 * rtl.h (rtx_alloc): Remove macro.
7545 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7547 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7548 (build_tree_list_stat): Likewise.
7549 * tree.h (build_tree_list): Remove macro.
7550 (build_tree_list_vec): Likewise.
7552 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7554 * tree.c (make_vector_stat): Remove _stat from name.
7555 (build_vector_stat): Likewise.
7556 * tree.h (make_vector_stat): Remove macro.
7557 (build_vector_stat): Likewise.
7559 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7561 * tree.h (build_var_debug_value): Remove prototype.
7563 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7565 * tree.c (tree_cons_stat): Remove _stat from name.
7566 * tree.h (tree_cons): Remove macro.
7568 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7570 * tree.c (build_vl_exp_stat): Remove _stat from name.
7571 * tree.h (build_vl_exp): Remove macro.
7573 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7575 * tree.c (build_decl_stat): Remove _stat from name.
7576 * tree.h (build_decl): Remove macro.
7578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7580 * gimple.c (gimple_build_with_ops_stat): Adjust.
7581 (gimple_alloc_stat): Remove _stat from name.
7582 * gimple.h (gimple_alloc): Remove macro.
7584 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7586 * tree.c (make_tree_vec_stat): Remove _stat from name.
7587 (grow_tree_vec_stat): Likewise.
7588 * tree.h (make_tree_vec_stat): Adjust prototype.
7589 (grow_tree_vec_stat): Likewise.
7590 (make_tree_vec): Remove macro.
7591 (grow_tree_vec): Likewise.
7593 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7595 * fold-const.c (fold_build1_stat_loc): Adjust.
7596 (fold_build2_stat_loc): Likewise.
7597 (fold_build3_stat_loc): Likewise.
7598 * tree.c (build0_stat): Remove _stat from name.
7599 (build1_stat): Likewise.
7600 (build2_stat): Likewise.
7601 (build3_stat): Likewise.
7602 (build4_stat): Likewise.
7603 (build5_stat): Likewise.
7604 * tree.h (build1_loc): Remove macro, and rename _stat function
7606 (build2_loc): Likewise.
7607 (build3_loc): Likewise.
7608 (build4_loc): Likewise.
7609 (build5_loc): Likewise.
7611 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7613 * tree.c (make_int_cst_stat): Remove _stat from name.
7614 * tree.h (make_int_cst_stat): Adjust prototype.
7615 (make_int_cst): Remove macro.
7617 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7619 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7620 * tree.h (make_tree_binfo_stat): Adjust prototype.
7621 (make_tree_binfo): Remove.
7623 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7625 * tree.c (copy_node_stat): Rename to copy_node.
7626 (build_distinct_type_copy): Adjust.
7627 * tree.h (copy_node_stat): Adjust prototype.
7628 (copy_node): Remove macro.
7630 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7632 * tree.c (make_node_stat): rename to make_node.
7633 (build_tree_list_stat): Adjust.
7634 (build0_stat): Likewise.
7635 (build2_stat): Likewise.
7636 (build3_stat): Likewise.
7637 (build4_stat): Likewise.
7638 (build5_stat): Likewise.
7639 (build_decl_stat): Likewise.
7640 * tree.h (make_node_stat): Adjust prototype.
7641 (make_node): remove macro.
7643 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
7645 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
7646 (PPC_FEATURE2_SCV): Likewise.
7647 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
7649 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7651 * config/aarch64/aarch64.c
7652 (aarch64_internal_mov_immediate): Add new special pattern.
7653 * config/aarch64/aarch64.md (*movdi_aarch64):
7654 Add reg/32bit const mov case.
7656 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7657 Richard Sandiford <richard.sandiford@linaro.org>
7659 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7660 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7661 Add integer and movi cases.
7662 (movi-split-hf-df-sf split, fp16): New.
7663 (enabled): Added TARGET_FP_F16INST.
7664 * config/aarch64/iterators.md (GPF_HF): New.
7665 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7667 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7669 * config/aarch64/aarch64.c
7670 (aarch64_simd_container_mode): Add prototype.
7671 (aarch64_expand_mov_immediate): Add HI support.
7672 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7673 (aarch64_can_const_movi_rtx_p): New.
7674 (aarch64_preferred_reload_class):
7675 Remove restrictions of using FP registers for certain SIMD operations.
7676 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7677 (aarch64_valid_floating_const): Add integer move validation.
7678 (aarch64_simd_imm_scalar_p): Remove.
7679 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7680 (aarch64_legitimate_constant_p): Expand list of supported cases.
7681 * config/aarch64/aarch64-protos.h
7682 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7683 (aarch64_reinterpret_float_as_int): New.
7684 (aarch64_simd_imm_scalar_p): Remove.
7685 * config/aarch64/constraints.md (Uvi): New.
7686 (Dd): Split into Ds and new Dd.
7687 * config/aarch64/aarch64.md (*movsi_aarch64):
7689 (*movdi_aarch64): Add SIMD mov case.
7691 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7693 * tree-predcom.c: (struct chain): Handle store-store chain in which
7694 stores for elimination only store loop invariant values.
7695 (execute_pred_commoning_chain): Ditto.
7696 (prepare_initializers_chain_store_elim): Ditto.
7697 (prepare_finalizers): Ditto.
7698 (is_inv_store_elimination_chain): New function.
7699 (initialize_root_vars_store_elim_1): New function.
7701 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7703 * tree-predcom.c: Revise general description of the pass.
7704 (enum chain_type): New enum type for store elimination.
7705 (struct chain): New field supporting store elimination.
7706 (struct component): Ditto.
7707 (dump_chain): Dump store-stores chain.
7708 (release_chain): Release resources.
7709 (split_data_refs_to_components): Compute and create component
7710 contains only stores for elimination.
7711 (get_chain_last_ref_at): New function.
7712 (make_invariant_chain): Initialization.
7713 (make_rooted_chain): Specify chain type in parameter and record it.
7714 (add_looparound_copies): Skip for store-stores chain.
7715 (determine_roots_comp): Compute type of chain and pass it to
7717 (initialize_root_vars_store_elim_2): New function.
7718 (finalize_eliminated_stores): New function.
7719 (remove_stmt): Handle store for elimination.
7720 (execute_pred_commoning_chain): Execute predictive commoning on
7722 (determine_unroll_factor): Skip unroll for store-stores chain.
7723 (prepare_initializers_chain_store_elim): New function.
7724 (prepare_initializers_chain): Hanlde store-store chain.
7725 (prepare_finalizers_chain, prepare_finalizers): New function.
7726 (tree_predictive_commoning_loop): Return integer value indicating
7727 if loop is unrolled or lcssa form is corrupted.
7728 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7730 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7732 * tree-predcom.c (initialize_root): Delete.
7733 (execute_pred_commoning_chain): Initialize root vars and replace
7734 reference of non-combined chain directly, rather than call above
7737 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7739 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7740 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7742 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7744 * tree-predcom.c (struct chain): New field init_seq.
7745 (release_chain): Release init_seq.
7746 (prepare_initializers_chain): Record intialization stmts in above
7748 (insert_init_seqs): New function.
7749 (tree_predictive_commoning_loop): Call insert_init_seqs.
7751 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7753 * tree-predcom.c (determine_roots_comp): Skip trivial components.
7755 2017-07-28 Richard Biener <rguenther@suse.de>
7757 * match.pd: Remove superfluous :c.
7758 * genmatch.c (simplify::id): Add member.
7759 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
7761 (current_id): New global.
7762 (dt_node::parent): Move from ...
7763 (dt_operand::parent): ... here. Add for_id member.
7764 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
7765 (decision_tree::find_node): Relax order requirement when
7766 merging DT_TRUE nodes to ones inbetween the current simplify
7767 and the one we try to merge with. Add diagnostic whenever
7768 we need to enforce pattern order by not merging.
7769 (decision_tree::insert): Set current_id.
7770 (decision_tree::print_node): Dump parent node and for_id.
7771 (parser::last_id): Add member.
7772 (parser::push_simplify): Assign unique id.
7773 (parser::parser): Initialize last_id.
7775 2017-07-28 Martin Liska <mliska@suse.cz>
7778 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
7779 gimple_build_debug_bind.
7781 2017-07-28 Richard Biener <rguenther@suse.de>
7783 PR tree-optimization/81502
7784 * match.pd: Add pattern combining BIT_INSERT_EXPR with
7786 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
7788 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
7789 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
7790 for BIT_FIELD_REF args.
7791 * fold-const.c (make_bit_field_ref): Likewise.
7792 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7794 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7797 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
7798 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
7799 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
7800 Or it into SANITIZER_UNDEFINED.
7801 * ubsan.c: Include gimple-fold.h and varasm.h.
7802 (ubsan_expand_ptr_ifn): New function.
7803 (instrument_pointer_overflow): New function.
7804 (maybe_instrument_pointer_overflow): New function.
7805 (instrument_object_size): Formatting fix.
7806 (pass_ubsan::execute): Call instrument_pointer_overflow
7807 and maybe_instrument_pointer_overflow.
7808 * internal-fn.c (expand_UBSAN_PTR): New function.
7809 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
7810 * sanitizer.def (__ubsan_handle_pointer_overflow,
7811 __ubsan_handle_pointer_overflow_abort): New builtins.
7812 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
7813 * internal-fn.def (UBSAN_PTR): New internal function.
7814 * opts.c (sanitizer_opts): Add pointer-overflow.
7815 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
7816 * fold-const.c (build_range_check): Compute pointer range check in
7817 integral type if pointer arithmetics would be needed. Formatting
7820 2017-07-28 Martin Liska <mliska@suse.cz>
7823 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
7824 parameters that are of a variable-length.
7826 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
7828 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
7830 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
7831 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
7832 (CRT_CALL_STATIC_FUNCTION): Likewise.
7833 (ASM_DEFAULT_SPEC): New define.
7834 (ASM_SPEC32): Likewise.
7835 (ASM_SPEC64): Likewise.
7836 (ASM_SPEC_COMMON): Likewise.
7837 (ASM_SPEC): Likewise.
7838 (INVALID_64BIT): Likewise.
7839 (LINK_OS_DEFAULT_SPEC): Likewise.
7840 (LINK_OS_SPEC32): Likewise.
7841 (LINK_OS_SPEC64): Likewise.
7842 (POWERPC_LINUX): Likewise.
7843 (PTRDIFF_TYPE): Likewise.
7844 (RESTORE_FP_PREFIX): Likewise.
7845 (RESTORE_FP_SUFFIX): Likewise.
7846 (SAVE_FP_PREFIX): Likewise.
7847 (SAVE_FP_SUFFIX): Likewise.
7848 (SIZE_TYPE): Likewise.
7849 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
7850 (TARGET_64BIT): Likewise.
7851 (TARGET_64BIT): Likewise.
7852 (TARGET_AIX): Likewise.
7853 (WCHAR_TYPE_SIZE): Likewise.
7854 (WCHAR_TYPE): Undefine.
7855 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
7856 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
7857 (CPP_OS_RTEMS_SPEC): Delete.
7858 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
7859 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
7861 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
7863 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7865 PR tree-optimization/81578
7866 * tree-parloops.c (build_new_reduction): Bail out if
7867 reduction_code isn't one of the standard OpenMP reductions.
7868 Move the details printing after that decision.
7870 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7872 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
7873 related to reload_in_progress.
7874 (splat_input_operand): Likewise.
7875 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
7877 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
7879 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
7880 (TARGET_INSTANTIATE_DECLS): Likewise.
7881 (legitimate_indexed_address_p): Delete reload_in_progress code.
7882 (rs6000_debug_legitimate_address_p): Likewise.
7883 (rs6000_eliminate_indexed_memrefs): Likewise.
7884 (rs6000_emit_le_vsx_store): Likewise.
7885 (rs6000_emit_move_si_sf_subreg): Likewise.
7886 (rs6000_emit_move): Likewise.
7887 (register_to_reg_type): Likewise.
7888 (rs6000_pre_atomic_barrier): Likewise.
7889 (rs6000_machopic_legitimize_pic_address): Likewise.
7890 (rs6000_allocate_stack_temp): Likewise.
7891 (rs6000_address_for_fpconvert): Likewise.
7892 (rs6000_address_for_altivec): Likewise.
7893 (rs6000_secondary_memory_needed_rtx): Delete function.
7894 (rs6000_check_sdmode): Likewise.
7895 (rs6000_alloc_sdmode_stack_slot): Likewise.
7896 (rs6000_instantiate_decls): Likewise.
7897 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
7898 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
7899 Delete reload_in_progress.
7900 (*vec_reload_and_plus_<mptrsize>): Likewise.
7901 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
7902 (vsx_div_v2di): Likewise.
7903 (vsx_udiv_v2di): Likewise.
7905 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7907 * config/rs6000/rs6000.opt (mlra): Replace with stub.
7908 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
7909 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
7910 (rs6000_debug_reg_global): Delete print of LRA status.
7911 (rs6000_option_override_internal): Delete dead LRA related code.
7912 (rs6000_lra_p): Delete function.
7913 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
7915 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
7917 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
7918 * config/riscv/rtems.h: New file.
7920 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7921 Sudakshina Das <sudi.das@arm.com>
7923 * config/aarch64/aarch64.md
7924 (define_split for and<mode>3nr_compare): Move
7925 non aarch64_logical_operand to a register.
7926 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
7927 register immediate operand to a register.
7928 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
7930 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7933 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
7935 2017-07-27 Richard Biener <rguenther@suse.de>
7937 PR tree-optimization/81573
7938 PR tree-optimization/81494
7939 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
7940 multi defuse cycle case.
7942 2017-07-27 Richard Biener <rguenther@suse.de>
7944 PR tree-optimization/81571
7945 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
7948 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
7950 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
7951 earlier and only if MASK_FPU is set. Adjust formatting.
7953 2017-07-27 Martin Liska <mliska@suse.cz>
7955 * opt-functions.awk: Add validation of value of Init.
7956 * optc-gen.awk: Pass new argument.
7958 2017-07-27 Martin Liska <mliska@suse.cz>
7960 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
7961 Fix wrong condition.
7963 2017-07-27 Martin Liska <mliska@suse.cz>
7965 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
7966 BBs and edges seen by autoFDO.
7968 2017-07-27 Richard Biener <rguenther@suse.de>
7970 PR tree-optimization/81502
7971 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
7972 with incompatible but same sized type.
7973 (execute_update_addresses_taken): Likewise.
7975 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
7977 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
7978 flag_tree_loop_vectorize rather than flag_tree_vectorize.
7980 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7983 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
7984 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
7985 Change s_operand to memory_operand.
7987 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7989 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7990 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7991 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7992 Emit instructions rather than returning an expression. Handle TFmode
7993 and KFmode by casting to TImode.
7994 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7995 (rs6000_emit_le_vsx_store): Likewise.
7996 * config/rs6000/vsx.md (VSX_TI): New iterator.
7997 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7998 (*vsx_le_undo_permute_<mode>): Likewise.
7999 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
8000 emit the split sequence.
8001 (*vsx_le_perm_store_<mode>): Likewise.
8003 2017-07-27 Jakub Jelinek <jakub@redhat.com>
8005 PR tree-optimization/81555
8006 PR tree-optimization/81556
8007 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
8008 if true, force CHANGED for the recursive invocation.
8009 (reassociate_bb): Remember original length of ops array, pass
8010 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
8012 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
8013 attributes for noipa attribute. For naked attribute use
8014 lookup_attribute first before lookup_attribute_spec.
8015 * final.c (rest_of_handle_final): Disable IPA RA for functions with
8017 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
8018 for functions with noipa attribute.
8019 (cgraph_externally_visible_p): Return true for functions with noipa
8021 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
8022 for functions with noipa attribute.
8023 * doc/extend.texi: Document noipa function attribute.
8024 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
8025 also for functions with noipa attribute.
8026 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
8028 2017-07-26 Andrew Pinski <apinski@cavium.com>
8030 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
8031 vec_unalign_load_cost and vec_unalign_store_cost.
8033 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8035 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
8036 -mvsx-small-integer option.
8037 (ISA_3_0_MASKS_IEEE): Likewise.
8038 (OTHER_VSX_VECTOR_MASKS): Likewise.
8039 (POWERPC_MASKS): Likewise.
8040 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
8041 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
8042 code, only testing for DImode being allowed in non-VSX floating
8044 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
8045 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
8047 (rs6000_option_override_internal): Delete -mvsx-small-integer.
8048 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
8049 TARGET_P8_VECTOR test.
8050 (rs6000_secondary_reload_simple_move): Likewise.
8051 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
8052 since TARGET_P9_VECTOR was already tested.
8053 (rs6000_opt_masks): Remove -mvsx-small-integer.
8054 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
8055 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8057 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
8058 test for TARGET_VEXTRACTUB was used, and that uses
8060 (p9 extract splitter): Likewise.
8061 (vsx_extract_<mode>_di_p9): Likewise.
8062 (vsx_extract_<mode>_store_p9): Likewise.
8063 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
8064 for TARGET_P9_VECTOR was used. Delete code that is now dead with
8065 the elimination of TARGET_VSX_SMALL_INTEGER.
8066 (vsx_extract_<mode>_p8): Likewise.
8067 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
8068 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
8069 (vsx_set_<mode>_p9): Likewise.
8070 (vsx_set_v4sf_p9): Likewise.
8071 (vsx_set_v4sf_p9_zero): Likewise.
8072 (vsx_insert_extract_v4sf_p9): Likewise.
8073 (vsx_insert_extract_v4sf_p9_2): Likewise.
8074 * config/rs6000/rs6000.md (sign extend splitter): Change
8075 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
8076 (floatsi<mode>2_lfiwax_mem): Likewise.
8077 (floatunssi<mode>2_lfiwzx_mem): Likewise.
8078 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
8079 since a test for TARGET_P9_VECTOR was used.
8080 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8081 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
8082 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8083 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8084 TARGET_P8_VECTOR test.
8085 (fix_trunc<mode>si2_stfiwx): Likewise.
8086 (fix_trunc<mode>si2_internal): Likewise.
8087 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
8088 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8090 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8091 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8092 TARGET_P8_VECTOR test.
8093 (fixuns_trunc<mode>si2_stfiwx): Likewise.
8094 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
8095 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8097 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8098 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
8099 since a test for TARGET_P9_VECTOR was used.
8100 (splitter for loading small constants): Likewise.
8102 2017-07-26 Andrew Pinski <apinski@cavium.com>
8104 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
8107 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
8110 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
8111 (fp_valid_at): Likewise.
8113 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8115 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
8116 (qdf24xx_addrcost_table): Likewise.
8117 (cortexa57_tunings): Update to use generic_branch_cost.
8118 (cortexa72_tunings): Likewise.
8119 (cortexa73_tunings): Likewise.
8120 (qdf24xx_tunings): Likewise.
8122 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8124 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
8125 (thunderx2t99_branch_cost): Likewise.
8126 (cortexa35_tunings): Update to use generic_branch_cost.
8127 (cortexa53_tunings): Likewise.
8128 (cortexa57_tunings): Likewise.
8129 (cortexa72_tunings): Likewise.
8130 (cortexa73_tunings): Likewise.
8131 (thunderx2t99_tunings): Likewise.
8133 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8135 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
8136 (sparc_option_override): Honour MASK_FSMULD.
8137 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
8138 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
8139 * config/sparc/sparc.opt (mfsmuld): New option.
8140 * doc/invoke.texi (mfsmuld): Document option.
8142 2017-07-26 Marek Polacek <polacek@redhat.com>
8145 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
8147 2017-07-26 Richard Biener <rguenther@suse.de>
8149 * gimple-match-head.c (do_valueize): Return OP if valueize
8151 (get_def): New helper to get at the def stmt of a SSA name
8153 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
8154 do_valueize to get at the def stmt.
8155 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
8157 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
8160 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
8162 2017-07-26 Martin Liska <mliska@suse.cz>
8165 * function.c (expand_function_start): Make expansion of
8166 nonlocal_goto_save_area after parm_birth_insn.
8168 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8170 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
8171 from all CPU target flags enable members.
8173 2017-07-26 Richard Biener <rguenther@suse.de>
8175 * genmatch.c (dt_simplify::gen): Make iterator vars const.
8176 (decision_tree::gen): Make 'type' const.
8177 (write_predicate): Likewise.
8179 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8181 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
8182 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
8183 (rs6000_option_override_internal): Likewise.
8184 (rs6000_expand_vector_set): Likewise.
8185 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
8186 (TARGET_UPPER_REGS_SF): Likewise.
8187 (TARGET_UPPER_REGS_DI): Likewise.
8188 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
8189 (TARGET_DIRECT_MOVE_64BIT): Likewise.
8190 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
8191 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8192 (Splitters for DI constants in Altivec registers): Likewise.
8193 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
8194 (vsx_set_v4sf_p9): Likewise.
8195 (vsx_set_v4sf_p9_zero): Likewise.
8196 (vsx_insert_extract_v4sf_p9): Likewise.
8197 (vsx_insert_extract_v4sf_p9_2): Likewise.
8199 2017-07-25 Carl Love <cel@us.ibm.com>
8201 * doc/extend.texi: Update the built-in documentation file for the
8202 existing built-in functions
8203 vector signed char vec_cnttz (vector signed char);
8204 vector unsigned char vec_cnttz (vector unsigned char);
8205 vector signed short vec_cnttz (vector signed short);
8206 vector unsigned short vec_cnttz (vector unsigned short);
8207 vector signed int vec_cnttz (vector signed int);
8208 vector unsigned int vec_cnttz (vector unsigned int);
8209 vector signed long long vec_cnttz (vector signed long long);
8210 vector unsigned long long vec_cnttz (vector unsigned long long);
8212 2017-07-25 Andrew Pinski <apinski@cavium.com>
8214 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
8215 accesses where the use is for the first operand of a BIT_INSERT.
8217 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8220 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8221 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8223 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8225 * config/i386/gstabs.h: Delete.
8226 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
8228 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
8230 * config/i386/i386.c (ix86_decompose_address): Do not check for
8231 register RTX when looking at index_reg or base_reg.
8232 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
8234 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
8236 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
8237 to update EH info here.
8239 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8241 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
8243 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8245 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
8247 2017-07-25 Torsten Duwe <duwe@suse.de>
8249 * common.opt: Introduce -fpatchable-function-entry
8250 command line option, and its variables function_entry_patch_area_size
8251 and function_entry_patch_area_start.
8252 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
8253 including a two-value parser.
8254 * target.def (print_patchable_function_entry): New target hook.
8255 * targhooks.h (default_print_patchable_function_entry): New function.
8256 * targhooks.c (default_print_patchable_function_entry): Likewise.
8257 * toplev.c (process_options): Switch off IPA-RA if
8258 patchable function entries are being generated.
8259 * varasm.c (assemble_start_function): Look at the
8260 patchable-function-entry command line switch and current
8261 function attributes and maybe generate NOP instructions by
8262 calling the print_patchable_function_entry hook.
8263 * doc/extend.texi: Document patchable_function_entry attribute.
8264 * doc/invoke.texi: Document -fpatchable_function_entry
8265 command line option.
8266 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
8268 * doc/tm.texi: Re-generate.
8270 2017-07-25 Jakub Jelinek <jakub@redhat.com>
8273 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
8274 TARGET_AVX512DQ rather than TARGET_AVX512BW.
8276 2017-07-25 Tamar Christina <tamar.christina@arm.com>
8278 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
8280 2017-07-25 Richard Biener <rguenther@suse.de>
8282 PR tree-optimization/81455
8283 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
8284 not walk in cycles when looking for guards.
8286 2017-07-25 Richard Biener <rguenther@suse.de>
8288 PR tree-optimization/81529
8289 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
8290 when optimizing backedge uses.
8292 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
8294 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
8296 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
8297 to dl_section_ref. On AIX, append an expression to subtract
8298 the size of the section length to dl_section_ref.
8300 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
8302 * configure.ac: If any of the config.* scripts fail, exit 1.
8303 * configure: Regenerate.
8305 2017-07-25 Richard Biener <rguenther@suse.de>
8308 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
8309 of immediate uses, be more verbose on errors.
8311 2017-07-25 Richard Biener <rguenther@suse.de>
8313 PR tree-optimization/81510
8314 * tree-vect-loop.c (vect_is_simple_reduction): When the
8315 reduction stmt is not inside the loop bail out.
8317 2017-07-25 Richard Biener <rguenther@suse.de>
8319 PR tree-optimization/81303
8320 * tree-vect-loop-manip.c (vect_loop_versioning): Build
8321 profitability check against LOOP_VINFO_NITERSM1.
8323 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8325 * domwalk.c (cmp_bb_postorder): Simplify.
8326 (sort_bbs_postorder): New function. Use it...
8327 (dom_walker::walk): ...here to optimize common cases.
8329 2017-07-25 Martin Liska <mliska@suse.cz>
8332 * ipa-visibility.c (function_and_variable_visibility): Make the
8333 redirection just on target that supports aliasing.
8334 Fix GNU coding style.
8336 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
8339 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
8341 * config/arm/rtems.h: Likewise.
8342 * config/bfin/rtems.h: Likewise.
8343 * config/i386/rtemself.h: Likewise.
8344 * config/lm32/rtems.h: Likewise.
8345 * config/m32c/rtems.h: Likewise.
8346 * config/m68k/rtemself.h: Likewise.
8347 * config/microblaze/rtems.h: Likewise.
8348 * config/mips/rtems.h: Likewise.
8349 * config/moxie/rtems.h: Likewise.
8350 * config/nios2/rtems.h: Likewise.
8351 * config/powerpcspe/rtems.h: Likewise.
8352 * config/rs6000/rtems.h: Likewise.
8353 * config/rtems.h: Likewise.
8354 * config/sh/rtems.h: Likewise.
8355 * config/sh/rtemself.h: Likewise.
8356 * config/sparc/rtemself.h: Likewise.
8358 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8361 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
8362 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
8363 * tree-ssa-structalias.c (alias_get_name): Same.
8365 2017-07-25 Bin Cheng <bin.cheng@arm.com>
8368 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
8369 instructions if no du chain is found.
8371 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8373 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
8375 2017-07-25 Richard Biener <rguenther@suse.de>
8378 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
8381 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8383 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
8385 (ISA_2_7_MASKS_SERVER): Likewise.
8386 (ISA_3_0_MASKS_IEEE): Likewise.
8387 (OTHER_P8_VECTOR_MASKS): Likewise.
8388 (OTHER_VSX_VECTOR_MASKS): Likewise.
8389 (POWERPC_MASKS): Likewise.
8390 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
8391 duplicate list of options.
8392 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
8393 explicit -mupper-regs options.
8394 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
8395 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
8396 alias for -mupper-regs-df.
8397 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8398 (rs6000_init_hard_regno_mode_ok): Likewise.
8399 (rs6000_option_override_internal): Likewise.
8400 (rs6000_opt_masks): Likewise.
8401 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8402 options in terms of whether -mvsx or -mpower8-vector was used.
8403 (TARGET_UPPER_REGS_DI): Likewise.
8404 (TARGET_UPPER_REGS_SF): Likewise.
8405 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8406 -mupper-regs-* options.
8408 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8410 * passes.c (emergency_dump_function): Print some empty lines and a
8411 header before the RTL dump.
8413 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8415 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8417 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8420 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8421 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8423 2017-07-24 Carl Love <cel@us.ibm.com>
8425 * config/rs6000/rs6000-c.c: Add support for built-in functions
8426 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8427 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8428 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8429 vec_extract_fp_from_shortl): Add defines for the two builtins.
8430 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8431 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8433 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8434 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8435 * doc/extend.texi: Update the built-in documentation file for the
8436 new built-in function.
8438 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8441 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8443 * doc/generic.texi: Likewise.
8444 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8445 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8447 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8449 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8450 (aarch64_mls_elt_merge<mode>): Likewise.
8452 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8454 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8455 having __cxa_atexit.
8457 2017-07-23 Michael Collison <michael.collison@arm.com>
8459 * config/arm/arm.c (arm_option_override): Deprecate
8460 use of -mstructure-size-boundary.
8461 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8462 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8464 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8467 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8468 Reduce cost estimate for direct moves.
8470 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8473 * config/i386/i386.c (ix86_option_override_internal): Disable
8474 BMI, BMI2 and TBM instructions for -m16.
8476 2017-07-21 Carl Love <cel@us.ibm.com>
8478 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8479 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8480 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8481 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8482 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8483 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8484 VMULOSW): New enum "unspec" values.
8485 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8486 altivec_vmulosw): New patterns.
8487 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8488 VMULOSW): Add definitions.
8490 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8492 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8493 (qdf24xx): Likewise.
8494 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8495 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8496 (AARCH64_FL_V8_1): Renumber.
8497 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8498 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8499 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8500 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8501 rdma to feature modifiers list.
8503 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8506 * ipa-visibility (function_and_variable_visibility): Convert
8507 recursive PLT call to direct call if appropriate.
8509 2017-07-21 Andrew Pinski <apinski@cavium.com>
8511 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8512 operand 1 to see if the types precision matches.
8513 * fold-const.c (operand_equal_p): Likewise.
8515 2017-07-21 Richard Biener <rguenther@suse.de>
8517 PR tree-optimization/81303
8518 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8519 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8520 (vect_peeling_hash_get_lowest_cost): Adjust.
8521 (vect_enhance_data_refs_alignment): Likewise. Use
8522 vect_get_peeling_costs_all_drs to compute the penalty for no
8523 peeling to match up costs.
8525 2017-07-21 Richard Biener <rguenther@suse.de>
8527 PR tree-optimization/81500
8528 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8529 we didn't identify a reduction path.
8531 2017-07-21 Tom de Vries <tom@codesourcery.com>
8532 Cesar Philippidis <cesar@codesourcery.com>
8534 PR gcov-profile/81442
8535 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8538 2017-07-21 Tom de Vries <tom@codesourcery.com>
8541 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8543 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8544 nvptx_override_options_after_change.
8546 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8548 * dwarf2out.c (output_file_names): Avoid double testing for
8551 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8553 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8555 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8557 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8559 (try_crossjump_to_edge): Do not punt on partitioned functions.
8561 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8563 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8564 Put all BBs reachable only via paths crossing cold region to cold
8566 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8568 2016-07-21 Richard Biener <rguenther@suse.de>
8570 PR tree-optimization/81303
8571 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8572 into account prologue and epilogue iterations when raising
8573 min_profitable_iters to sth at least covering one vector iteration.
8575 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8577 * config/arm/arm.c (arm_test_cpu_arch_dat):
8580 2017-07-20 Nathan Sidwell <nathan@acm.org>
8582 Remove TYPE_METHODS.
8583 * tree.h (TYPE_METHODS): Delete.
8584 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8585 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8586 (dbxout_type_methods): Scan TYPE_FIELDS.
8587 (dbxout_type): Don't check TYPE_METHODS here.
8588 * function.c (use_register_for_decl): Always ignore register for
8589 class types when not optimizing.
8590 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8591 * tree.c (free_lang_data_in_type): Stitch out member functions and
8592 templates from TYPE_FIELDS.
8593 (build_distinct_type_copy, verify_type_variant,
8594 verify_type): Member fns are on TYPE_FIELDS.
8595 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8596 * tree-pretty-print.c (dump_generic_node): Likewise.
8598 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8601 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8602 V2TImode and V4TImode.
8603 (ix86_expand_vector_extract): Likewise.
8604 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8605 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8606 (ssescalarmode): Handle V4TImode and V2TImode.
8607 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8608 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8609 (VEXTRACTI128_MODE): New mode iterator.
8610 (splitter for *vec_extractv?ti first element): New.
8611 (VEC_INIT_MODE): New mode iterator.
8612 (vec_init<mode>): Consolidate 3 expanders into one using
8613 VEC_INIT_MODE mode iterator.
8615 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8617 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8618 non_spilled_static_chain_regno_p.
8620 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8622 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8624 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8626 * bb-reorder.c (connect_traces): Allow copying of blocks within
8629 2017-07-20 Richard Biener <rguenther@suse.de>
8631 * gimple.h (gimple_phi_result): Add gphi * overload.
8632 (gimple_phi_result_ptr): Likewise.
8633 (gimple_phi_arg): Likewise. Adjust index assert to only
8634 allow actual argument accesses rather than all slots available
8636 (gimple_phi_arg_def): Add gphi * overload.
8637 * tree-phinodes.c (make_phi_node): Initialize only actual
8639 (resize_phi_node): Clear memory not covered by old node,
8640 do not initialize excess argument slots.
8641 (reserve_phi_args_for_new_edge): Initialize new argument slot
8644 2017-07-20 Bin Cheng <bin.cheng@arm.com>
8646 PR tree-optimization/81388
8648 2016-07-21 Bin Cheng <bin.cheng@arm.com>
8650 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
8651 by removing computation of may_be_zero.
8653 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8654 Tom de Vries <tom@codesourcery.com>
8657 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8658 when gimple level profile disagrees with what RTL expander did.
8660 2017-07-20 Richard Biener <rguenther@suse.de>
8662 PR tree-optimization/61171
8663 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8664 (vect_analyze_stmt): Add slp instance parameter.
8665 (vectorizable_reduction): Likewise.
8666 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8667 (vect_is_simple_reduction): Deal with chains not detected
8668 as SLP reduction chain, specifically not properly associated
8669 chains containing a mix of plus/minus.
8670 (get_reduction_op): Remove.
8671 (get_initial_defs_for_reduction): Simplify, pass in whether
8672 this is a reduction chain, pass in the SLP node for the PHIs.
8673 (vect_create_epilog_for_reduction): Get the SLP instance as
8675 (vectorizable_reduction): Get the SLP instance as arg.
8676 During analysis remember the SLP node with the PHIs in the
8677 instance. Simplify getting at the vectorized reduction PHIs.
8678 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8679 through SLP instance.
8680 (vect_slp_analyze_operations): Likewise.
8681 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8682 (vect_transform_stmt): Likewise.
8684 2017-07-20 Tom de Vries <tom@codesourcery.com>
8686 PR tree-optimization/81489
8687 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8688 read of phi arg location to before loop that modifies phi.
8690 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8692 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8695 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8698 * except.c (execute): Fix ordering issue.
8700 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8702 PR rtl-optimization/81423
8703 * combine.c (make_compound_operation_int): Don't try to optimize
8704 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8706 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8708 PR rtl-optimization/81423
8709 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8710 with a constant that is -1 in the truncated to mode.
8712 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8714 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8715 (determine_unlikely_bbs): ... here.
8716 * predict.h (propagate_unlikely_bbs_forward): Declare.
8717 * cfgexpand.c (pass_expand::execute): Use it.
8718 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8720 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8721 propagate_unlikely_bbs_forward.
8723 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8726 * except.c (maybe_add_nop_after_section_switch): New function.
8729 2017-07-19 Tom de Vries <tom@codesourcery.com>
8731 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8733 2017-07-19 Tom de Vries <tom@codesourcery.com>
8735 * gimple.h (gimple_phi_arg): Make assert more strict.
8737 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8739 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8740 * config/rs6000/mmintrin.h: New file.
8741 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8743 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8745 PR tree-optimization/81346
8746 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
8748 2017-07-19 Tom de Vries <tom@codesourcery.com>
8750 * config/nvptx/nvptx.md (VECIM): Add V2DI.
8752 2017-07-19 Tom de Vries <tom@codesourcery.com>
8754 * config/nvptx/nvptx-modes.def: Add V2DImode.
8755 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
8756 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
8757 (nvptx_output_mov_insn): Handle lack of mov.b128.
8758 (nvptx_print_operand): Handle 'H' and 'L' codes.
8759 (nvptx_vector_mode_supported): Allow V2DImode.
8760 (nvptx_preferred_simd_mode): New function.
8761 (nvptx_data_alignment): New function.
8762 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
8763 nvptx_preferred_simd_mode.
8764 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
8766 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
8768 2017-07-19 Tom de Vries <tom@codesourcery.com>
8770 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
8771 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
8772 (nvptx_vector_mode_supported): New function. Allow V2SImode.
8773 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
8774 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
8775 (mov<VECIM>_insn): New define_insn.
8776 (define_expand "mov<VECIM>): New define_expand.
8778 2017-07-19 Tom de Vries <tom@codesourcery.com>
8780 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
8782 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8784 PR tree-optimization/81346
8785 * fold-const.h (fold_div_compare, range_check_type): Declare.
8786 * fold-const.c (range_check_type): New function.
8787 (build_range_check): Use range_check_type.
8788 (fold_div_compare): No longer static, rewritten into
8789 a match.pd helper function.
8790 (fold_comparison): Don't call fold_div_compare here.
8791 * match.pd (X / C1 op C2): New optimization using fold_div_compare
8794 2017-07-19 Nathan Sidwell <nathan@acm.org>
8796 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
8797 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
8798 * tree.c (find_decls_types_r, verify_type): Use
8799 TYPE_{MIN,MAX}_VALUE_RAW.
8800 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
8801 (hash_tree): Likewise.
8802 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8804 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8807 2017-07-18 Tom de Vries <tom@codesourcery.com>
8810 * omp-expand.c (expand_omp_for_static_chunk): Handle
8811 equal-argument loop exit phi.
8813 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
8816 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
8817 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
8818 operand 2 predicate.
8819 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
8820 operand 2 predicate.
8821 (ror,rol -> rorx splitters): Use const_int_operand as
8822 operand 2 predicate.
8824 2017-06-18 Richard Biener <rguenther@suse.de>
8826 PR tree-optimization/81410
8827 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
8828 the gap in the ! slp_perm SLP case after each group.
8830 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8833 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
8836 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8839 * predict.c (set_even_probabilities): Cleanup; do not affect
8840 probabilities that are already known.
8841 (combine_predictions_for_bb): Call even when count is set.
8843 2017-07-18 Nathan Sidwell <nathan@acm.org>
8845 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
8848 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8851 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
8852 optimization for loop niter analysis.
8854 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
8857 * config/avr/avr.c (avr_optimize_casesi): Don't use
8858 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
8860 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8862 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
8863 body_cost_vec from _vect_peel_extended_info.
8864 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
8865 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
8868 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8870 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
8872 2017-07-18 Richard Biener <rguenther@suse.de>
8874 PR tree-optimization/80620
8875 PR tree-optimization/81403
8876 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
8877 info when re-using a VN table entry.
8879 2017-07-18 Richard Biener <rguenther@suse.de>
8881 PR tree-optimization/81418
8882 * tree-vect-loop.c (vectorizable_reduction): Properly compute
8883 vectype_in. Verify that with lane-reducing reduction operations
8884 we have a single def-use cycle.
8886 2017-07-17 Carl Love <cel@us.ibm.com>
8888 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
8890 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8891 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8892 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8893 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8894 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8895 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8896 VMULOSW): New enum "unspec" values.
8897 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8898 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8899 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8900 altivec_vmulosw): New patterns.
8901 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8902 VMULOSW): Add definitions.
8903 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
8905 * config/alpha/alpha.c: Include predict.h.
8907 2017-07-17 Yury Gribov <tetra2005@gmail.com>
8909 * tree-vrp.c (compare_assert_loc): Fix comparison function
8910 to return predictable results.
8912 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8914 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
8917 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
8918 * doc/invoke.texi (mexpand-adddi): Update text.
8920 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8922 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
8923 that also clobbers the CC register. The old expand code is moved
8925 (*arc_clzsi2): ... here.
8926 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
8927 the CC register. The old expand code is moved to ...
8928 (arc_ctzsi2): ... here.
8930 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8932 * config/arc/arc.opt (mindexed-loads): Use initial value
8933 TARGET_INDEXED_LOADS_DEFAULT.
8934 (mauto-modify-reg): Use initial value
8935 TARGET_AUTO_MODIFY_REG_DEFAULT.
8936 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8937 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8938 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8939 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8941 2017-07-17 Martin Liska <mliska@suse.cz>
8944 * opts.c (finish_options): Do not allow -fgnu-tm
8945 w/ -fsanitize={kernel-,}address. Say sorry.
8947 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8950 * tree-loop-distribution.c (classify_partition): Only assert on
8951 numer of iterations.
8952 (merge_dep_scc_partitions): Delete prameter. Update function call.
8953 (distribute_loop): Remove code handling loop with unknown niters.
8954 (pass_loop_distribution::execute): Skip loop with unknown niters.
8956 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8959 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
8960 function sort_partitions_by_post_order.
8962 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8964 PR tree-optimization/81374
8965 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
8966 the max index of basic blocks, rather than number of basic blocks.
8968 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8970 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
8972 (arc_legitimate_pic_operand_p): Likewise.
8973 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
8975 (arc_needs_pcl_p): Likewise.
8976 (arc_legitimate_pc_offset_p): Likewise.
8977 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
8978 function is also used in constrains.md.
8979 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
8980 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
8981 PLUS. Only return true/false in known cases, otherwise assert.
8982 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
8983 is already called in arc_legitimate_constant_p.
8984 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8986 (LEGITIMATE_PIC_OPERAND_P): Use
8987 arc_raw_symbolic_reference_mentioned_p function.
8988 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8993 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8994 Andrew Burgess <andrew.burgess@embecosm.com>
8996 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8997 (arc_return_address_register): New function.
8998 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8999 (arc_handle_fndecl_attribute): Add naked attribute.
9000 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9001 (TARGET_WARN_FUNC_RETURN): Likewise.
9002 (arc_allocate_stack_slots_for_args): New function.
9003 (arc_warn_func_return): Likewise.
9004 (machine_function): Change type fn_type.
9005 (arc_compute_function_type): Consider new naked function type,
9006 change function return type.
9007 (arc_must_save_register): Adapt to handle new
9008 arc_compute_function_type's return type.
9009 (arc_expand_prologue): Likewise.
9010 (arc_expand_epilogue): Likewise.
9011 (arc_return_address_regs): Delete.
9012 (arc_return_address_register): New function.
9013 (arc_epilogue_uses): Use above function.
9014 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
9015 (arc_function_type): Change encoding, add naked type.
9016 (ARC_INTERRUPT_P): Change to handle the new encoding.
9017 (ARC_FAST_INTERRUPT_P): Likewise.
9018 (ARC_NORMAL_P): Define.
9019 (ARC_NAKED_P): Likewise.
9020 (arc_compute_function_type): Delete prototype.
9021 * config/arc/arc.md (in_ret_delay_slot): Use
9022 arc_return_address_register function.
9023 (simple_return): Likewise.
9024 (p_return_i): Likewise.
9026 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9028 PR tree-optimization/81428
9029 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
9030 can't be built for those types.
9032 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9034 Remove stuff dead since r239246.
9036 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
9037 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
9038 (avr_inform_devices): Remove dead stuff.
9040 2017-07-17 Tamar Christina <tamar.christina@arm.com>
9042 * config/arm/arm_neon.h: Fix softp typo.
9044 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9046 PR tree-optimization/81365
9047 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
9048 aggregate moves onto bb predecessor edges, make sure there are no
9049 loads that could alias the lhs in between the start of bb and the
9052 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9055 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
9056 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
9057 [LSHIFTRT, outer_code = TRUNCATE]: Same.
9059 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9061 PR tree-optimization/81396
9062 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
9063 (init_symbolic_number): Initialize it to 1.
9064 (perform_symbolic_merge): Add n_ops from both operands into the new
9066 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
9067 without base_addr as useless if they need more than one operation.
9068 (bswap_replace): Handle !bswap case for NULL base_addr.
9070 2017-07-17 Tom de Vries <tom@codesourcery.com>
9073 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
9076 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
9078 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
9079 conditional builtin define __FIX_LEON3FT_B2BST.
9081 2017-07-17 Daniel Cederman <cederman@gaisler.com>
9083 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
9084 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
9087 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9089 PR rtl-optimization/81424
9090 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
9091 to remove potential trapping from operands if -fnon-call-exceptions.
9093 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9095 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
9096 profile_proability for scalling.
9097 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
9099 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9101 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
9103 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9105 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
9106 fixpoint arithmetics.
9108 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9110 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
9111 fixpoint arithmetics.
9113 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9115 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
9116 fixpoint arithmetics.
9118 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9120 * profile-count.h (profile_probability::from_reg_br_prob_note,
9121 profile_probability::to_reg_br_prob_note): New functions.
9122 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
9123 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
9124 * predict.c (probability_reliable_p): Update.
9125 (edge_probability_reliable_p): Update.
9126 (br_prob_note_reliable_p): Update.
9127 (invert_br_probabilities): Update.
9128 (add_reg_br_prob_note): New function.
9129 (combine_predictions_for_insn): Update.
9130 * asan.c (asan_clear_shadow): Update.
9131 * cfgbuild.c (compute_outgoing_frequencies): Update.
9132 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
9133 (update_br_prob_note): Update.
9134 (rtl_verify_edges): Update.
9135 (purge_dead_edges): Update.
9136 (fixup_reorder_chain): Update.
9137 * emit-rtl.c (try_split): Update.
9138 * ifcvt.c (cond_exec_process_insns): Update.
9139 (cond_exec_process_if_block): Update.
9140 (dead_or_predicable): Update.
9141 * internal-fn.c (expand_addsub_overflow): Update.
9142 (expand_neg_overflow): Update.
9143 (expand_mul_overflow): Update.
9144 * loop-doloop.c (doloop_modify): Update.
9145 * loop-unroll.c (compare_and_jump_seq): Update.
9146 * optabs.c (emit_cmp_and_jump_insn_1): Update.
9147 * predict.h: Update.
9148 * reorg.c (mostly_true_jump): Update.
9150 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
9151 * config/alpha/alpha.c (emit_unlikely_jump): Update.
9152 * config/arc/arc.c: (emit_unlikely_jump): Update.
9153 * config/arm/arm.c: (emit_unlikely_jump): Update.
9154 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
9155 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
9156 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
9157 (ix86_print_operand): Update.
9158 (ix86_split_fp_branch): Update.
9159 (predict_jump): Update.
9160 * config/ia64/ia64.c (ia64_print_operand): Update.
9161 * config/mmix/mmix.c (mmix_print_operand): Update.
9162 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
9163 (rs6000_expand_split_stack_prologue): Update.
9164 * config/rs6000/rs6000.c: Update.
9165 * config/s390/s390.c (s390_expand_vec_strlen): Update.
9166 (s390_expand_vec_movstr): Update.
9167 (s390_expand_cs_tdsi): Update.
9168 (s390_expand_split_stack_prologue): Update.
9169 * config/sh/sh.c (sh_print_operand): Update.
9170 (expand_cbranchsi4): Update.
9171 (expand_cbranchdi4): Update.
9172 * config/sparc/sparc.c (output_v9branch): Update.
9173 * config/spu/spu.c (get_branch_target): Update.
9174 (ea_load_store_inline): Update.
9175 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
9176 * config/tilepro/tilepro.c: Update.
9178 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9180 * gimplify.c (mostly_copy_tree_r): Revert latest change.
9181 (gimplify_save_expr): Likewise.
9183 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9185 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
9187 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9189 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
9191 * timevar.def (TV_IPA_FNSUMMARY): Define.
9193 2017-07-16 Daniel Cederman <cederman@gaisler.com>
9195 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
9196 to back store errata sensitive sequence from being generated.
9197 (sqrtdf2_fix): Likewise.
9199 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9201 * tree-ssa-threadupdate.c (compute_path_counts,
9202 update_joiner_offpath_counts): Use profile_probability.
9204 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
9207 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9209 * config/arm/arm-c.c (arm_cpu_builtins): Define
9210 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9212 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9214 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9215 array entries to represent __ieee128 versions of the
9216 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
9217 scalar_extract_sig, and scalar_insert_exp built-in functions.
9218 (altivec_resolve_overloaded_builtin): Add special case handling
9219 for the __builtin_scalar_insert_exp function, as represented by
9220 the P9V_BUILTIN_VEC_VSIEDP constant.
9221 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
9222 exponent support for __ieee128 argument.
9223 (VSESQP): Add scalar extract signature support for __ieee128
9225 (VSTDCNQP): Add scalar test negative support for __ieee128
9227 (VSIEQP): Add scalar insert exponent support for __int128 argument
9228 with __ieee128 result.
9229 (VSIEQPF): Add scalar insert exponent support for __ieee128
9230 argument with __ieee128 result.
9231 (VSTDCQP): Add scalar test data class support for __ieee128
9233 (VSTDCNQP): Add overload support for scalar test negative with
9235 (VSTDCQP): Add overload support for scalar test data class
9237 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
9239 (UNSPEC_VSX_SIEXPQP): New constant.
9240 (xsxexpqp): New insn for VSX scalar extract exponent quad
9242 (xsxsigqp): New insn for VSX scalar extract significand quad
9244 (xsiexpqpf): New insn for VSX scalar insert exponent quad
9245 precision with floating point argument.
9246 (xststdcqp): New expand for VSX scalar test data class quad
9248 (xststdcnegqp): New expand for VSX scalar test negative quad
9250 (xststdcqp): New insn to match expansions for VSX scalar test data
9251 class quad precision and VSX scalar test negative quad precision.
9252 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
9253 special case operand checking to enforce that second operand of
9254 VSX scalar test data class with quad precision argument is a 7-bit
9256 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9257 prototypes and descriptions of __ieee128 versions of
9258 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
9259 scalar_test_data_class, and scalar_test_neg built-in functions.
9261 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9263 PR tree-optimization/81162
9264 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
9265 replace a negate with an add.
9267 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
9269 * doc/invoke.texi (arm/-mcpu): Document +crypto.
9271 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9273 * config/arm/arm-c.c (arm_cpu_builtins): Define
9274 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9276 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9278 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
9279 (armv8-r): Set ARM Cortex-R52 as default CPU.
9280 * config/arm/arm-tables.opt: Regenerate.
9281 * config/arm/arm-tune.md: Regenerate.
9282 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9284 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
9285 extension for -mcpu=cortex-r52.
9287 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9289 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
9290 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
9291 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
9292 (fp-armv8): Define it as FP_ARMv8 only.
9293 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
9294 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
9296 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
9297 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
9298 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
9299 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
9300 than TARGET_FPU_ARMV8.
9301 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
9302 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
9303 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
9305 * config/arm/neon.md (neon_vrint): Likewise.
9306 (neon_vcvt): Likewise.
9307 (neon_<fmaxmin_op><mode>): Likewise.
9308 (<fmaxmin><mode>3): Likewise.
9309 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
9310 * config/arm/predicates.md (arm_cond_move_operator): Check against
9311 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
9313 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
9315 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
9318 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9320 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
9321 loop in comment with memset.
9323 2017-07-14 Martin Liska <mliska@suse.cz>
9325 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
9326 * dwarf2out.c (is_java): Remove the function.
9327 (output_pubname): Remove usage of the function.
9328 (lower_bound_default): Remove usage of DW_LANG_Java.
9329 (gen_compile_unit_die): Likewise.
9330 * gcc.c: Remove compiler defaults for .java and .zip files.
9331 * gimple-expr.c (remove_suffix): Change as there's no longer
9332 extension than 4-letter one.
9333 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
9334 (gimplify_save_expr): Likewise.
9335 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
9336 as it's possible even for other languages than Java.
9337 * langhooks.h (struct lang_hooks): Remove Java from a comment.
9338 * lto-opts.c (lto_write_options): Remove reference to Java.
9339 * opts.c (strip_off_ending): Update file extension handling.
9340 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
9341 * tree-eh.c (lower_resx): Likewise.
9342 * tree.c (free_lang_data_in_type): Remove dead code.
9343 (find_decls_types_r): Likewise.
9344 (build_common_builtin_nodes): Remove Java from a comment.
9345 (verify_type): Remove dead code.
9346 * varasm.c (assemble_external): Remove Java from a comment.
9348 2017-07-14 Martin Liska <mliska@suse.cz>
9350 * opts.c (finish_options): Add quotes.
9351 (common_handle_option): Likewise.
9353 2017-07-14 Martin Liska <mliska@suse.cz>
9355 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
9356 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
9358 * dwarf2out.c (lower_bound_default): Do not handle
9360 (gen_compile_unit_die): Likewise.
9361 * gcc.c: Remove default extension binding for GNU Pascal.
9362 * stmt.c: Remove Pascal language from a comment.
9363 * xcoffout.c: Likewise.
9365 2017-07-13 David Malcolm <dmalcolm@redhat.com>
9368 * diagnostic-show-locus.c (fixit_cmp): New function.
9369 (layout::layout): Sort m_fixit_hints.
9370 (column_range::column_range): Assert that the values are valid.
9371 (struct char_span): New struct.
9372 (correction::overwrite): New method.
9373 (struct source_line): New struct.
9374 (line_corrections::add_hint): Add assertions. Reimplement memcpy
9375 calls in terms of classes source_line and char_span, and
9376 correction::overwrite.
9377 (selftest::test_overlapped_fixit_printing_2): New function.
9378 (selftest::diagnostic_show_locus_c_tests): Call it.
9380 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9382 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
9383 early if there is no lhs.
9385 2017-07-13 Martin Liska <mliska@suse.cz>
9387 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
9388 (gen_reference_type_die): Likewise.
9389 * stor-layout.c: Remove Pascal-related comment.
9391 2017-07-13 Martin Liska <mliska@suse.cz>
9393 * opts.c (finish_options): Add quotes to error messages.
9394 (parse_sanitizer_options): Likewise.
9396 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
9398 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
9400 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9402 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9404 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9406 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9408 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9410 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9413 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9414 provides the hardware capability bits, define the macro
9415 __BUILTIN_CPU_SUPPORTS__.
9416 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9417 if GLIBC does not provide the hardware capability bits. Add a
9418 gcc_unreachable call if the built-in cpu function is neither
9419 __builtin_cpu_is nor __builtin_cpu_supports.
9420 (rs6000_get_function_versions_dispatcher): Change the warning
9421 that an old GLIBC is used which does not export the capability
9422 bits to be an error.
9423 * doc/extend.texi (target_clones attribute): Document the
9424 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9425 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9426 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9427 the macros defined by GCC if the newer GLIBC is available.
9429 2017-07-12 Jeff Law <law@redhat.com>
9431 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9432 remaining includes slightly.
9433 * config/riscv/riscv-builtins.c: Include profile-count.h.
9435 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9438 * config/avr/avr.c (avr_set_current_function): In diagnostic
9439 messages: Quote keywords and (parts of) identifiers.
9440 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9443 2017-07-12 Carl Love <cel@us.ibm.com>
9445 * config/rs6000/rs6000-c.c: Add support for built-in functions
9446 vector bool char vec_revb (vector bool char);
9447 vector bool short vec_revb (vector short char);
9448 vector bool int vec_revb (vector bool int);
9449 vector bool long long vec_revb (vector bool long long);
9450 * doc/extend.texi: Update the built-in documentation file for the
9451 new built-in functions.
9453 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9455 * config/s390/s390.md: Remove movcc splitter.
9457 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9459 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9460 load/store on condition.
9462 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9465 * config/avr/avr.c (avr_encode_section_info)
9466 [progmem && !TREE_READONLY]: Error if progmem object needs
9469 2017-07-11 Michael Collison <michael.collison@arm.com>
9471 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9474 2017-07-11 Carl Love <cel@us.ibm.com>
9476 * config/rs6000/rs6000-c.c: Add support for builtins
9477 vector unsigned int vec_parity_lsbb (vector signed int);
9478 vector unsigned int vec_parity_lsbb (vector unsigned int);
9479 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9480 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9481 vector unsigned long long vec_parity_lsbb (vector signed long long);
9482 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9483 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9484 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9485 * doc/extend.texi: Update the built-in documentation file for the
9486 new built-in functions.
9488 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9490 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9491 (layout::m_primary_loc): New field.
9492 (layout::layout): Initialize new field. Move location filtering
9493 logic from here to...
9494 (layout::maybe_add_location_range): ...this new method. Add
9495 support for filtering to just the lines already specified by other
9497 (layout::will_show_line_p): New method.
9498 (gcc_rich_location::add_location_if_nearby): New method.
9499 (selftest::test_add_location_if_nearby): New test function.
9500 (selftest::diagnostic_show_locus_c_tests): Call it.
9501 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9504 2017-07-11 Tom de Vries <tom@codesourcery.com>
9506 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9507 (bb_first_real_insn): New function.
9508 (nvptx_single): Add extra initialization of broadcasted condition
9511 2017-07-11 Nathan Sidwell <nathan@acm.org>
9513 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9515 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9517 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9518 Binutils doc as TEXI will mess them up.
9519 * doc/invoke.texi (AVR Options): Same here.
9521 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9523 * config/sparc/sparc.opt (mfix-ut700): New option.
9524 (mfix-gr712rc): Likewise.
9525 (sparc_fix_b2bst): New variable.
9526 * doc/invoke.texi (SPARC options): Document them.
9527 (ARM options): Fix warnings.
9528 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9529 instructions to prevent sequences that can trigger the store-store
9530 errata for certain LEON3FT processors.
9531 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9532 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9533 * config/sparc/sparc.md (fix_b2bst): New attribute.
9534 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9536 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9539 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9541 (*rsqrtsf2_sse): Ditto.
9543 (div<mode>3): Macroize insn from divdf3 and divsf3
9544 using MODEF mode iterator.
9546 2017-07-10 Martin Sebor <msebor@redhat.com>
9548 PR tree-optimization/80397
9549 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9550 instead of testing for equality to INTEGER_TYPE.
9552 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9554 * config.gcc: Remove uclibc from arc target spec.
9556 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9558 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9560 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9563 * lto-wrapper.c (remove_option): New function.
9564 (merge_and_complain): Merge PIC/PIE options more realistically.
9566 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9568 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9572 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9573 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9574 * config.in: Regenerate.
9575 * configure: Regenerate.
9576 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9577 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9578 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9579 (TARGET_GASISR_PROLOGUES): ...target mask.
9580 * common/config/avr/avr-common.c
9581 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9582 Set -mgas-isr-prologues.
9583 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9584 INSERT_PASS_BEFORE for it.
9585 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9586 * config/avr/avr.c (avr_option_override)
9587 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9588 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9589 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9590 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9591 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9593 (make_avr_pass_pre_proep): New function.
9594 (emit_push_sfr) <treg>: Add argument to function and use it
9596 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9597 and set machine->gasisr.yes.
9598 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9599 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9600 __gcc_isr.n_pushed to .L__stack_usage.
9601 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9602 (avr_asm_final_postscan_insn): ...this new static function.
9603 * config/avr/avr.h (machine_function)
9604 <is_no_gccisr, use_L__stack_usage>: New fields.
9605 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9606 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9607 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9608 (gasisr, *gasisr): New expander and insn.
9609 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9610 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9611 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9613 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9615 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9618 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9620 Move jump-tables out of .text again.
9623 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9624 (ASM_OUTPUT_ADDR_VEC): New function.
9625 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9626 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9627 INSN_ADDRESSes as asm comment.
9628 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9629 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9630 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9631 * config/avr/avr.md (*tablejump): Adjust comment.
9632 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9633 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9635 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9636 (avr_output_addr_vec): New proto.
9637 (avr_log_t) <insn_addresses>: New field.
9639 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9642 * config/i386/i386.c (ix86_function_arg_advance): Set
9643 outgoing_args_on_stack to true if there are outgoing arguments
9645 (ix86_function_arg): Likewise.
9646 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
9647 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
9648 * config/i386/i386.h (machine_function): Add
9649 outgoing_args_on_stack.
9651 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
9653 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9655 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
9657 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9659 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9660 (REAL_H): Remove $(MACHMODE_H).
9661 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9663 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9664 $(MACHMODE_H) and double-int.h.
9665 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9667 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9670 2017-07-07 Andrew Pinski <apinski@cavium.com>
9672 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9673 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9675 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9677 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9678 Add warning if GCC was not configured to link against a GLIBC that
9679 exports the hardware capability bits.
9680 (make_resolver_func): Make resolver function private and not a
9681 COMDAT function. Create the name with clone_function_name instead
9682 of make_unique_name.
9685 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9686 correct operand in doing the split.
9688 2017-07-07 Carl Love <cel@us.ibm.com>
9690 * config/rs6000/rs6000-c: Add support for built-in function
9691 vector unsigned short vec_pack_to_short_fp32 (vector float,
9693 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9694 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9695 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9696 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9697 (convert_4f32_8i16): Add define_expand.
9698 * doc/extend.texi: Update the built-in documentation file for the
9699 new built-in function.
9701 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9703 * config/sparc/m8.md: New file.
9704 * config/sparc/sparc.md: Include m8.md.
9706 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9708 * config/sparc/sparc.opt: New option -mvis4b.
9709 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9710 (sparc_option_override): Handle VIS4B.
9711 (enum sparc_builtins): Define
9712 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9713 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9714 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9715 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9716 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9717 (check_constant_argument): New function.
9718 (sparc_vis_init_builtins): Define builtins
9719 __builtin_vis_dictunpack{8,16,32},
9720 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9721 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9722 __builtin_vis_fpcmpde{8,16,32}shl and
9723 __builtin_vis_fpcmpur{8,16,32}shl.
9724 (sparc_expand_builtin): Check that the constant operands to
9725 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9726 constant and in range.
9727 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9729 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9730 (SPARC_IMM5_P): Likewise.
9731 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9732 (enabled): Handle vis4b.
9733 (UNSPEC_DICTUNPACK): New unspec.
9734 (UNSPEC_FPCMPSHL): Likewise.
9735 (UNSPEC_FPUCMPSHL): Likewise.
9736 (UNSPEC_FPCMPDESHL): Likewise.
9737 (UNSPEC_FPCMPURSHL): Likewise.
9738 (cpu_feature): New CPU feature `vis4b'.
9739 (dictunpack{8,16,32}): New insns.
9740 (FPCSMODE): New mode iterator.
9741 (fpcscond): New code iterator.
9742 (fpcsucond): Likewise.
9743 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9744 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
9745 (fpcmpde{8,16,32}{si,di}shl): Likewise.
9746 (fpcmpur{8,16,32}{si,di}shl): Likewise.
9747 * config/sparc/constraints.md: Define constraints `q' for unsigned
9748 2-bit integer constants and `t' for unsigned 5-bit integer
9750 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
9752 (imm5_operand_dictunpack16): Likewise.
9753 (imm5_operand_dictunpack32): Likewise.
9754 (imm2_operand): Likewise.
9755 * doc/invoke.texi (SPARC Options): Document -mvis4b.
9756 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
9757 ditunpack* and fpcmp*shl builtins.
9759 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9761 * config.gcc: Handle m8 in --with-{cpu,tune} options.
9762 * config.in: Add HAVE_AS_SPARC6 define.
9763 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
9765 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
9767 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
9768 (CPP_CPU_SPEC): Handle m8.
9769 (ASM_CPU_SPEC): Likewise.
9770 * config/sparc/sparc-opts.h (enum processor_type): Add
9772 * config/sparc/sparc.c (m8_costs): New struct.
9773 (sparc_option_override): Handle TARGET_CPU_m8.
9774 (sparc32_initialize_trampoline): Likewise.
9775 (sparc64_initialize_trampoline): Likewise.
9776 (sparc_issue_rate): Likewise.
9777 (sparc_register_move_cost): Likewise.
9778 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
9779 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
9780 (ASM_CPU64_DEFAULT_SPEC): Likewise.
9781 (CPP_CPU_SPEC): Handle M8.
9782 (ASM_CPU_SPEC): Likewise.
9783 (AS_M8_FLAG): Define.
9784 * config/sparc/sparc.md: Add m8 to the cpu attribute.
9785 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
9786 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
9788 * configure: Regenerate.
9789 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
9792 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9794 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
9796 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
9797 ("*movdi_insn_sp32"): Do not set v3pipe.
9798 ("*movsi_insn"): Likewise.
9799 ("*movdi_insn_sp64"): Likewise.
9800 ("*movsf_insn"): Likewise.
9801 ("*movdf_insn_sp32"): Likewise.
9802 ("*movdf_insn_sp64"): Likewise.
9803 ("*zero_extendsidi2_insn_sp64"): Likewise.
9804 ("*sign_extendsidi2_insn"): Likewise.
9805 ("*mov<VM32:mode>_insn"): Likewise.
9806 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9807 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9808 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9809 ("<vlop:code><VL:mode>3"): Likewise.
9810 ("*not_<vlop:code><VL:mode>3"): Likewise.
9811 ("*nand<VL:mode>_vis"): Likewise.
9812 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
9813 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
9814 ("one_cmpl<VL:mode>2"): Likewise.
9815 ("faligndata<VM64:mode>_vis"): Likewise.
9816 ("alignaddrsi_vis"): Likewise.
9817 ("alignaddrdi_vis"): Likweise.
9818 ("alignaddrlsi_vis"): Likewise.
9819 ("alignaddrldi_vis"): Likewise.
9820 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9821 ("bmaskdi_vis"): Likewise.
9822 ("bmasksi_vis"): Likewise.
9823 ("bshuffle<VM64:mode>_vis"): Likewise.
9824 ("cmask8<P:mode>_vis"): Likewise.
9825 ("cmask16<P:mode>_vis"): Likewise.
9826 ("cmask32<P:mode>_vis"): Likewise.
9827 ("pdistn<P:mode>_vis"): Likewise.
9828 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9830 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9832 * config/sparc/sparc.md ("subtype"): New insn attribute.
9833 ("*wrgsr_sp64"): Set insn subtype.
9834 ("*rdgsr_sp64"): Likewise.
9835 ("alignaddrsi_vis"): Likewise.
9836 ("alignaddrdi_vis"): Likewise.
9837 ("alignaddrlsi_vis"): Likewise.
9838 ("alignaddrldi_vis"): Likewise.
9839 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9840 ("fexpand_vis"): Likewise.
9841 ("fpmerge_vis"): Likewise.
9842 ("faligndata<VM64:mode>_vis"): Likewise.
9843 ("bshuffle<VM64:mode>_vis"): Likewise.
9844 ("cmask8<P:mode>_vis"): Likewise.
9845 ("cmask16<P:mode>_vis"): Likewise.
9846 ("cmask32<P:mode>_vis"): Likewise.
9847 ("fchksm16_vis"): Likewise.
9848 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
9849 ("fmean16_vis"): Likewise.
9850 ("fp<plusminus_insn>64_vis"): Likewise.
9851 ("<plusminus_insn>v8qi3"): Likewise.
9852 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9853 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
9854 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
9855 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
9856 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
9857 ("*movqi_insn"): Likewise.
9858 ("*movhi_insn"): Likewise.
9859 ("*movsi_insn"): Likewise.
9860 ("movsi_pic_gotdata_op"): Likewise.
9861 ("*movdi_insn_sp32"): Likewise.
9862 ("*movdi_insn_sp64"): Likewise.
9863 ("movdi_pic_gotdata_op"): Likewise.
9864 ("*movsf_insn"): Likewise.
9865 ("*movdf_insn_sp32"): Likewise.
9866 ("*movdf_insn_sp64"): Likewise.
9867 ("*zero_extendhisi2_insn"): Likewise.
9868 ("*zero_extendqihi2_insn"): Likewise.
9869 ("*zero_extendqisi2_insn"): Likewise.
9870 ("*zero_extendqidi2_insn"): Likewise.
9871 ("*zero_extendhidi2_insn"): Likewise.
9872 ("*zero_extendsidi2_insn_sp64"): Likewise.
9873 ("ldfsr"): Likewise.
9874 ("prefetch_64"): Likewise.
9875 ("prefetch_32"): Likewise.
9876 ("tie_ld32"): Likewise.
9877 ("tie_ld64"): Likewise.
9878 ("*tldo_ldub_sp32"): Likewise.
9879 ("*tldo_ldub1_sp32"): Likewise.
9880 ("*tldo_ldub2_sp32"): Likewise.
9881 ("*tldo_ldub_sp64"): Likewise.
9882 ("*tldo_ldub1_sp64"): Likewise.
9883 ("*tldo_ldub2_sp64"): Likewise.
9884 ("*tldo_ldub3_sp64"): Likewise.
9885 ("*tldo_lduh_sp32"): Likewise.
9886 ("*tldo_lduh1_sp32"): Likewise.
9887 ("*tldo_lduh_sp64"): Likewise.
9888 ("*tldo_lduh1_sp64"): Likewise.
9889 ("*tldo_lduh2_sp64"): Likewise.
9890 ("*tldo_lduw_sp32"): Likewise.
9891 ("*tldo_lduw_sp64"): Likewise.
9892 ("*tldo_lduw1_sp64"): Likewise.
9893 ("*tldo_ldx_sp64"): Likewise.
9894 ("*mov<VM32:mode>_insn"): Likewise.
9895 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9896 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9898 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9900 * config/sparc/sparc.md ("type"): New insn type viscmp.
9901 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
9903 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
9904 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
9905 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9906 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
9908 ("n7_vis_logical_11cycle"): Likewise.
9909 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
9910 * config/sparc/niagara2.md ("niag3_vis": Likewise.
9911 * config/sparc/niagara.md ("niag_vis"): Likewise.
9912 * config/sparc/ultra3.md ("us3_fga"): Likewise.
9913 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
9915 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9917 * config/sparc/sparc.md: New instruction type `bmask'.
9918 (bmaskdi_vis): Use the `bmask' type.
9919 (bmasksi_vis): Likewise.
9920 * config/sparc/ultra3.md (us3_array): Likewise.
9921 * config/sparc/niagara7.md (n7_array): Likewise.
9922 * config/sparc/niagara4.md (n4_array): Likewise.
9923 * config/sparc/niagara2.md (niag2_vis): Likewise.
9924 (niag3_vis): Likewise.
9925 * config/sparc/niagara.md (niag_vis): Likewise.
9927 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9929 * ipa-comdats.c: Remove optimize check from gate.
9930 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
9931 for functions not optimized.
9932 (ipa_fn_summary_read): Skip optimize check.
9933 (ipa_fn_summary_write): Likewise.
9934 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
9936 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
9938 (can_inline_edge_p): Check flag_pcc_struct_return for match.
9939 (check_callers): Give up on caller which is not optimized.
9940 (inline_small_functions): Likewise.
9941 (ipa_inline): Do not give up when not optimizing.
9942 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
9943 away unoptimizes cdtors.
9944 (whole_program_function_and_variable_visibility): Do
9945 ipa_discover_readonly_nonaddressable_vars in LTO mode.
9946 * ipa.c (process_references): Do not check optimize.
9947 (symbol_table::remove_unreachable_nodes): Update optimize check.
9948 (set_writeonly_bit): Update optimize check.
9949 (pass_ipa_cdtor_merge::gate): Do not check optimize.
9950 (pass_ipa_single_use::gate): Remove.
9952 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9954 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
9955 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
9956 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
9957 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
9958 permute_load, permute_store, adjust_extract, adjust_splat,
9959 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
9960 replace_swap_with_copy, dump_swap_insn_table,
9961 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
9962 recombine_lvx_pattern, recombine_stvx_pattern,
9963 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
9964 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
9965 to file rs6000-p8swap.c.
9966 * config/rs6000/rs6000-p8swap.c: New file.
9967 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
9968 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
9969 and rs6000*-*-* targets.
9971 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9973 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
9975 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9977 * lto-wrapper.c (merge_and_complain): Do not merge
9978 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
9979 fsigned_zeros, ftrapping_math, fwrapv.
9980 (append_compiler_options): Do not track these options.
9981 (append_linker_options): Likewie
9983 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9985 * cgraphunit.c (cgraph_node::finalize_function): When
9986 !flag_toplevel_reorde set no_reorder flag.
9987 (varpool_node::finalize_decl): Likewise.
9988 (symbol_table::compile): Drop no toplevel reorder path.
9990 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9992 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9993 edges; zero probability is not better than uninitialized.
9995 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9997 * asan.h (asan_sanitize_allocas_p): Declare.
9998 * asan.c (asan_sanitize_allocas_p): New function.
9999 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
10000 (handle_builtin_alloca): Likewise.
10001 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
10002 if !asan_sanitize_allocas_p.
10003 * params.def (asan-instrument-allocas): Add new option.
10004 * params.h (ASAN_PROTECT_ALLOCAS): Define.
10005 * opts.c (common_handle_option): Disable allocas sanitization for
10008 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10010 * asan.c: Include gimple-fold.h.
10011 (get_last_alloca_addr): New function.
10012 (handle_builtin_stackrestore): Likewise.
10013 (handle_builtin_alloca): Likewise.
10014 (asan_emit_allocas_unpoison): Likewise.
10015 (get_mem_refs_of_builtin_call): Add new parameter, remove const
10016 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
10017 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
10018 (instrument_builtin_call): Pass gimple iterator to
10019 get_mem_refs_of_builtin_call.
10020 (last_alloca_addr): New global.
10021 * asan.h (asan_emit_allocas_unpoison): Declare.
10022 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
10023 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
10024 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
10025 if function calls alloca.
10026 * gimple-fold.c (replace_call_with_value): Remove static keyword.
10027 * gimple-fold.h (replace_call_with_value): Declare.
10028 * internal-fn.c: Include asan.h.
10029 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
10030 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
10032 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10034 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
10035 (C_SELFTEST_FLAGS): New.
10036 (CPP_SELFTEST_FLAGS): New.
10037 (SELFTEST_DEPS): New, from deps of s-selftest.
10038 (C_SELFTEST_DEPS): New, from deps of s-selftest.
10039 (CPP_SELFTEST_DEPS): New.
10040 (selftest): Add dependency on s-selftest-c++.
10041 (s-selftest): Rename to...
10042 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
10043 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
10044 than SELFTEST_FLAGS.
10045 (selftest-gdb): Rename to...
10046 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
10048 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
10049 (selftest-valgrind): Rename to...
10050 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
10052 (selftest-valgrind): Reintroduce as an alias for
10053 selftest-c-valgrind.
10054 (s-selftest-c++): New.
10055 (selftest-c++-gdb): New.
10056 (selftest-c++-valgrind): New.
10058 2017-07-06 Olivier Hainque <hainque@adacore.com>
10060 * gcc.c (process_command): When deciding if undefined variables
10061 should be ignored when processing specs, accept "gcc -v" as well.
10063 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10065 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
10066 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
10068 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10070 * config/arm/arm-cpus.in (armv8-r): Add new entry.
10071 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
10072 * config/arm/arm-tables.opt: Regenerate.
10073 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
10075 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
10077 2017-07-06 Carl Love <cel@us.ibm.com>
10079 * ChangeLog: Clean up from mid air collision
10081 2017-07-06 Carl Love <cel@us.ibm.com>
10083 * config/rs6000/rs6000-c.c: Add support for built-in functions
10084 vector signed int vec_subc (vector signed int, vector signed int);
10085 vector signed __int128 vec_subc (vector signed __int128,
10086 vector signed __int128);
10087 vector unsigned __int128 vec_subc (vector unsigned __int128,
10088 vector unsigned __int128);
10089 vector signed int vec_sube (vector signed int, vector signed int,
10090 vector signed int);
10091 vector unsigned int vec_sube (vector unsigned int,
10092 vector unsigned int,
10093 vector unsigned int);
10094 vector signed __int128 vec_sube (vector signed __int128,
10095 vector signed __int128,
10096 vector signed__int128);
10097 vector unsigned __int128 vec_sube (vector unsigned __int128,
10098 vector unsigned __int128,
10099 vector unsigned __int128);
10100 vector signed int vec_subec (vector signed int, vector signed int,
10101 vector signed int);
10102 vector unsigned int vec_subec (vector unsigned int,
10103 vector unsigned int,
10104 vector unsigned int);
10105 vector signed __int128 vec_subec (vector signed __int128,
10106 vector signed __int128,
10107 vector signed__int128);
10108 vector unsigned __int128 vec_subec (vector unsigned __int128,
10109 vector unsigned __int128,
10110 vector unsigned __int128);
10111 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
10112 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
10113 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
10114 BU_ALTIVEC_OVERLOAD_X definitions.
10115 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
10116 * doc/extend.texi: Update the built-in documentation file for the new
10117 built-in functions.
10119 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10122 * diagnostic-show-locus.c (layout::layout): Use start and finish
10123 spelling location for the start and finish of each range.
10124 * genmatch.c (linemap_client_expand_location_to_spelling_point):
10125 Add unused aspect param.
10126 * input.c (expand_location_1): Add "aspect" param, and use it
10127 to access the correct part of the location.
10128 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
10130 (expand_location_to_spelling_point): Likewise.
10131 (linemap_client_expand_location_to_spelling_point): Add "aspect"
10132 param, and pass it to expand_location_1.
10134 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
10136 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
10137 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
10138 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
10139 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
10140 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
10141 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
10142 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
10143 _mm_maskz_getmant_ss): New intrinsics.
10144 (__builtin_ia32_getexpss128_mask): Changed to ...
10145 __builtin_ia32_getexpss128_round ... this.
10146 (__builtin_ia32_getexpsd128_mask): Changed to ...
10147 __builtin_ia32_getexpsd128_round ... this.
10148 * config/i386/i386-builtin-types.def
10149 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
10150 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
10151 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
10152 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
10153 __builtin_ia32_getmantss_mask_round): New builtins.
10154 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
10155 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
10156 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
10157 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
10158 * config/i386/sse.md
10159 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
10160 avx512f_sgetexp<mode><mask_scalar_name>
10161 <round_saeonly_scalar_name> ... this.
10162 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
10163 %0, %1, %2<round_saeonly_op3>}): Changed to ...
10164 vgetexp<ssescalarmodesuffix>
10165 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10166 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
10167 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
10168 avx512f_vgetmant<mode><mask_scalar_name>
10169 <round_saeonly_scalar_name> ... this.
10170 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10171 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
10172 vgetmant<ssescalarmodesuffix>
10173 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
10174 %0<mask_scalar_operand4>, %1, %2
10175 <round_saeonly_scalar_mask_op4>, %3} ... this.
10176 * config/i386/subst.md (mask_scalar_operand4,
10177 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
10178 round_saeonly_scalar_nimm_predicate): New subst attributes.
10180 2017-07-06 Julia Koval <julia.koval@intel.com>
10182 * config/i386/i386.c (ix86_erase_embedded_rounding):
10183 Remove code for old rounding pattern.
10185 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
10187 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
10189 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
10191 * doc/sourcebuild.texi (Test Directives, Variants of
10192 dg-require-support): Add documentation for dg-require-stack-check.
10194 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
10196 * config/i386/subst.md (mask_scalar, round_scalar,
10197 round_saeonly_scalar): New meta-templates.
10198 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
10199 round_scalar_mask_operand3, round_scalar_mask_op3,
10200 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
10201 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
10202 round_saeonly_scalar_constraint,
10203 round_saeonly_scalar_prefix): New subst attribute.
10204 * config/i386/sse.md
10205 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
10206 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
10207 <round_scalar_name> ... this.
10208 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
10209 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
10210 <round_scalar_name> ... this.
10211 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
10212 <sse>_vm<code><mode>3<mask_scalar_name>
10213 <round_saeonly_scalar_name> ... this.
10214 (v<plusminus_mnemonic><ssescalarmodesuffix>
10215 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10216 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10217 v<plusminus_mnemonic><ssescalarmodesuffix>
10218 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10219 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10220 (v<multdiv_mnemonic><ssescalarmodesuffix>
10221 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10222 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10223 v<multdiv_mnemonic><ssescalarmodesuffix>
10224 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10225 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10226 (v<maxmin_float><ssescalarmodesuffix>
10227 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
10228 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
10229 v<maxmin_float><ssescalarmodesuffix>
10230 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10231 %0<mask_scalar_operand3>, %1, %<iptr>2
10232 <round_saeonly_scalar_mask_op3>} ... this.
10234 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
10236 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
10237 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
10239 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
10240 Alan Hayward <alan.hayward@arm.com>
10241 David Sherwood <david.sherwood@arm.com>
10243 * combine.c (simplify_if_then_else): Remove "enum" before
10245 * compare-elim.c (can_eliminate_compare): Likewise.
10246 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10248 (aarch64_lookup_simd_builtin_type): Likewise.
10249 (aarch64_simd_builtin_type): Likewise.
10250 (aarch64_init_simd_builtin_types): Likewise.
10251 (aarch64_simd_expand_args): Likewise.
10252 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
10254 (aarch64_reverse_mask): Likewise.
10255 (aarch64_simd_emit_reg_reg_move): Likewise.
10256 (aarch64_gen_adjusted_ldpstp): Likewise.
10257 (aarch64_ccmp_mode_to_code): Likewise.
10258 (aarch64_operands_ok_for_ldpstp): Likewise.
10259 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10260 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10262 (aarch64_min_divisions_for_recip_mul): Likewise.
10263 (aarch64_reassociation_width): Likewise.
10264 (aarch64_get_condition_code_1): Likewise.
10265 (aarch64_simd_emit_reg_reg_move): Likewise.
10266 (aarch64_simd_attr_length_rglist): Likewise.
10267 (aarch64_reverse_mask): Likewise.
10268 (aarch64_operands_ok_for_ldpstp): Likewise.
10269 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10270 (aarch64_gen_adjusted_ldpstp): Likewise.
10271 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
10273 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
10274 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10275 (arm_lookup_simd_builtin_type): Likewise.
10276 (arm_simd_builtin_type): Likewise.
10277 (arm_init_simd_builtin_types): Likewise.
10278 (arm_expand_builtin_args): Likewise.
10279 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
10280 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
10281 (ft32_setup_incoming_varargs): Likewise.
10282 (ft32_function_arg): Likewise.
10283 (ft32_function_arg_advance): Likewise.
10284 (ft32_pass_by_reference): Likewise.
10285 (ft32_arg_partial_bytes): Likewise.
10286 (ft32_valid_pointer_mode): Likewise.
10287 (ft32_addr_space_pointer_mode): Likewise.
10288 (ft32_addr_space_legitimate_address_p): Likewise.
10289 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
10291 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
10292 (ix86_emit_outlined_ms2sysv_restore): Likewise.
10293 (iamcu_alignment): Likewise.
10294 (canonicalize_vector_int_perm): Likewise.
10295 (ix86_noce_conversion_profitable_p): Likewise.
10296 (ix86_mpx_bound_mode): Likewise.
10297 (ix86_operands_ok_for_move_multiple): Likewise.
10298 * config/microblaze/microblaze-protos.h
10299 (microblaze_expand_conditional_branch_reg): Likewise.
10300 * config/microblaze/microblaze.c
10301 (microblaze_expand_conditional_branch_reg): Likewise.
10302 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10304 (rs6000_reassociation_width): Likewise.
10305 (rs6000_invalid_binary_op): Likewise.
10306 (fusion_p9_p): Likewise.
10307 (emit_fusion_p9_load): Likewise.
10308 (emit_fusion_p9_store): Likewise.
10309 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
10311 (riscv_hard_regno_mode_ok_p): Likewise.
10312 (riscv_address_insns): Likewise.
10313 (riscv_split_symbol): Likewise.
10314 (riscv_legitimize_move): Likewise.
10315 (riscv_function_value): Likewise.
10316 (riscv_hard_regno_nregs): Likewise.
10317 (riscv_expand_builtin): Likewise.
10318 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
10319 (riscv_build_integer): Likewise.
10320 (riscv_split_integer): Likewise.
10321 (riscv_legitimate_constant_p): Likewise.
10322 (riscv_cannot_force_const_mem): Likewise.
10323 (riscv_regno_mode_ok_for_base_p): Likewise.
10324 (riscv_valid_base_register_p): Likewise.
10325 (riscv_valid_offset_p): Likewise.
10326 (riscv_valid_lo_sum_p): Likewise.
10327 (riscv_classify_address): Likewise.
10328 (riscv_legitimate_address_p): Likewise.
10329 (riscv_address_insns): Likewise.
10330 (riscv_load_store_insns): Likewise.
10331 (riscv_force_binary): Likewise.
10332 (riscv_split_symbol): Likewise.
10333 (riscv_force_address): Likewise.
10334 (riscv_legitimize_address): Likewise.
10335 (riscv_move_integer): Likewise.
10336 (riscv_legitimize_const_move): Likewise.
10337 (riscv_legitimize_move): Likewise.
10338 (riscv_address_cost): Likewise.
10339 (riscv_subword): Likewise.
10340 (riscv_output_move): Likewise.
10341 (riscv_canonicalize_int_order_test): Likewise.
10342 (riscv_emit_int_order_test): Likewise.
10343 (riscv_function_arg_boundary): Likewise.
10344 (riscv_pass_mode_in_fpr_p): Likewise.
10345 (riscv_pass_fpr_single): Likewise.
10346 (riscv_pass_fpr_pair): Likewise.
10347 (riscv_get_arg_info): Likewise.
10348 (riscv_function_arg): Likewise.
10349 (riscv_function_arg_advance): Likewise.
10350 (riscv_arg_partial_bytes): Likewise.
10351 (riscv_function_value): Likewise.
10352 (riscv_pass_by_reference): Likewise.
10353 (riscv_setup_incoming_varargs): Likewise.
10354 (riscv_print_operand): Likewise.
10355 (riscv_elf_select_rtx_section): Likewise.
10356 (riscv_save_restore_reg): Likewise.
10357 (riscv_for_each_saved_reg): Likewise.
10358 (riscv_register_move_cost): Likewise.
10359 (riscv_hard_regno_mode_ok_p): Likewise.
10360 (riscv_hard_regno_nregs): Likewise.
10361 (riscv_class_max_nregs): Likewise.
10362 (riscv_memory_move_cost): Likewise.
10363 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
10364 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
10365 (rl78_addr_space_address_mode): Likewise.
10366 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10368 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10369 (rs6000_reassociation_width): Likewise.
10370 (rs6000_invalid_binary_op): Likewise.
10371 (fusion_p9_p): Likewise.
10372 (emit_fusion_p9_load): Likewise.
10373 (emit_fusion_p9_store): Likewise.
10374 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
10375 (ok_for_simple_move_operands): Likewise.
10376 (ok_for_simple_move_strict_operands): Likewise.
10377 (ok_for_simple_arith_logic_operands): Likewise.
10378 (visium_legitimize_reload_address): Likewise.
10379 (visium_select_cc_mode): Likewise.
10380 (output_cbranch): Likewise.
10381 (visium_split_double_move): Likewise.
10382 (visium_expand_copysign): Likewise.
10383 (visium_expand_int_cstore): Likewise.
10384 (visium_expand_fp_cstore): Likewise.
10385 * config/visium/visium.c (visium_pass_by_reference): Likewise.
10386 (visium_function_arg): Likewise.
10387 (visium_function_arg_advance): Likewise.
10388 (visium_libcall_value): Likewise.
10389 (visium_setup_incoming_varargs): Likewise.
10390 (visium_legitimate_constant_p): Likewise.
10391 (visium_legitimate_address_p): Likewise.
10392 (visium_legitimize_address): Likewise.
10393 (visium_secondary_reload): Likewise.
10394 (visium_register_move_cost): Likewise.
10395 (visium_memory_move_cost): Likewise.
10396 (prepare_move_operands): Likewise.
10397 (ok_for_simple_move_operands): Likewise.
10398 (ok_for_simple_move_strict_operands): Likewise.
10399 (ok_for_simple_arith_logic_operands): Likewise.
10400 (visium_function_value_1): Likewise.
10401 (rtx_ok_for_offset_p): Likewise.
10402 (visium_legitimize_reload_address): Likewise.
10403 (visium_split_double_move): Likewise.
10404 (visium_expand_copysign): Likewise.
10405 (visium_expand_int_cstore): Likewise.
10406 (visium_expand_fp_cstore): Likewise.
10407 (visium_split_cstore): Likewise.
10408 (visium_select_cc_mode): Likewise.
10409 (visium_split_cbranch): Likewise.
10410 (output_cbranch): Likewise.
10411 (visium_print_operand_address): Likewise.
10412 * expmed.c (flip_storage_order): Likewise.
10413 * expmed.h (emit_cstore): Likewise.
10414 (flip_storage_order): Likewise.
10415 * genrecog.c (validate_pattern): Likewise.
10416 * hsa-gen.c (gen_hsa_addr): Likewise.
10417 * internal-fn.c (expand_arith_overflow): Likewise.
10418 * ira-color.c (allocno_copy_cost_saving): Likewise.
10419 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10420 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10421 (process_invariant_for_inheritance): Likewise.
10422 * lra-eliminations.c (move_plus_up): Likewise.
10423 * omp-low.c (lower_oacc_reductions): Likewise.
10424 * simplify-rtx.c (simplify_subreg): Likewise.
10425 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10426 (TARGET_CHKP_BOUND_MODE): Likewise..
10427 * targhooks.c (default_chkp_bound_mode): Likewise.
10428 (default_setup_incoming_vararg_bounds): Likewise.
10429 * targhooks.h (default_chkp_bound_mode): Likewise.
10430 (default_setup_incoming_vararg_bounds): Likewise.
10431 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10432 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10433 (have_whole_vector_shift): Likewise.
10434 * tree-vect-stmts.c (vectorizable_load): Likewise.
10435 * doc/tm.texi: Regenerate.
10437 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10439 Graceful degrade if Binutils PR21472 is not available.
10442 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10443 .rodata in flash test fails.
10444 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10445 * confgure: Regenerate.
10446 * config.in: Regenerate.
10447 * config/avr/avr.c (avr_asm_named_section)
10448 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10449 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10450 (avr_asm_init_sections): Same.
10452 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10454 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10455 (fma<VH:mode>4_intrinsic): Likewise.
10456 (*fmsub<VCVTF:mode>4): Likewise.
10457 (*fmsub<VH:mode>4_intrinsic): Likewise.
10459 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10462 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10463 Don't depend on "optimize > 0".
10464 (out_movhi_r_mr, out_movqi_mr_r): Same.
10465 (out_movhi_mr_r, out_movqi_r_mr): Same.
10466 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10467 io_address_operand on "optimize > 0".
10469 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10471 * tree-loop-distribution.c: Add general explanantion on the pass.
10472 (generate_loops_for_partition): Mark distributed loop.
10473 (pg_add_dependence_edges): New parameter. Handle alias data
10474 dependence specially and record it in the parameter if asked.
10475 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10476 (init_partition_graph_vertices, add_partition_graph_edge): New.
10477 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10478 (free_partition_graph_vdata, build_partition_graph): New.
10479 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10480 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10481 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10482 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10483 (version_for_distribution_p, finalize_partitions): New.
10484 (distribute_loop): Handle alias data dependence specially. Factor
10485 out loop fusion code as functions and call these functions.
10487 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10489 * tree-loop-distribution.c (classify_partition): New parameter and
10490 better handle reduction statement.
10491 (rdg_build_partitions): Revise comment.
10492 (distribute_loop): Compute statements in all partitions and pass it
10493 to classify_partition.
10495 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10497 * tree-loop-distribution.c (enum partition_type): New.
10498 (struct partition): New field type.
10499 (partition_merge_into): Add parameter. Update partition type.
10500 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10501 (build_rdg_partition_for_vertex): Compute partition type.
10502 (rdg_build_partitions): Dump partition type.
10503 (distribute_loop): Update calls to partition_merge_into.
10505 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10507 * tree-loop-distribution.c (struct ddr_hasher): New.
10508 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10510 (classify_partition): Call get_data_dependence.
10511 (pg_add_dependence_edges): Ditto.
10512 (distribute_loop): Release data dependence hash table.
10514 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10516 * tree-loop-distribution.c (ref_base_address): Delete.
10517 (similar_memory_accesses): Rename ...
10518 (share_memory_accesses): ... to this. Check if partitions access
10519 the same memory reference.
10520 (distribute_loop): Call share_memory_accesses.
10522 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10524 * tree-loop-distribution.c (struct partition): New field recording
10525 its data reference.
10526 (partition_alloc, partition_free): Init and release data refs.
10527 (partition_merge_into): Merge data refs.
10528 (build_rdg_partition_for_vertex): Collect data refs for partition.
10529 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10531 (distribute_loop): Remve data refs from vertice data of partition
10534 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10536 * tree-loop-distribution.c (params.h): Include header file.
10537 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10538 (datarefs_vec): New global var.
10539 (create_rdg_vertices): Use datarefs_vec directly.
10540 (free_rdg): Don't free data references.
10541 (build_rdg): Update use. Don't free data references.
10542 (distribute_loop): Compute global variable for data references.
10543 Bail out if there are too many data references.
10545 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10547 * tree-loop-distribution.c (loop_nest): New global var.
10548 (build_rdg): Use loop directly, rather than loop nest.
10549 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10551 (distribute_loop): Compute global variable loop nest. Update use.
10553 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10555 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10556 (partition_merge_into): New parameter. Dump reason for fusion.
10557 (distribute_loop): Update use of partition_merge_into.
10559 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10561 * tree-loop-distribution.c (bb_top_order_index): New.
10562 (bb_top_order_index_size, bb_top_order_cmp): New.
10563 (stmts_from_loop): Use topological order.
10564 (pass_loop_distribution::execute): Compute and release topological
10565 order for basic blocks.
10567 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10569 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10572 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10574 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10575 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10576 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10577 * internal-fn.def (LOOP_DIST_ALIAS): New.
10578 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10579 (fold_loop_internal_call): ... this.
10580 (vect_loop_dist_alias_call): New function.
10581 (set_uid_loop_bbs): Call fold_loop_internal_call.
10582 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10585 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10588 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10589 Require dead FLAGS_REG at the beginning of a peephole.
10591 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10594 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10595 arguments in the call to __builtin_ia32_sbb_u32.
10596 (_subborrow_u64): Swap _X and _Y arguments in the call to
10597 __builtin_ia32_sbb_u64.
10599 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10602 * tree-vrp.c (compare_assert_loc): Turn into a function template
10603 with stable template parameter. Only test if a->e is NULL,
10604 !a->e == !b->e has been verified already. Use e == NULL or
10605 e != NULL instead of e or ! e tests. If stable is true, don't use
10606 iterative_hash_expr, on the other side allow a or b or both NULL
10607 and sort the NULLs last.
10608 (process_assert_insertions): Sort using compare_assert_loc<false>
10609 instead of compare_assert_loc, later sort using
10610 compare_assert_loc<true> before calling process_assert_insertions_for
10611 in a loop. Use break instead of continue once seen NULL pointer.
10613 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10615 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10616 Cortex-R7 and Cortex-R8 processors.
10618 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10620 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10621 uninitialized while src is not.
10623 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10625 * common/config/arm/arm-common.c: Adjust include path for
10627 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10628 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10629 (arm-cpu-data.h): Likewise.
10630 (arm-cpu-cdata.h): Likewise.
10631 * config/arm/arm-cpu.h: Delete.
10632 * config/arm/arm-cpu-cdata.h: Delete.
10633 * config/arm/arm-cpu-data.h: Delete.
10635 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10637 * config/arm/arm-cpus.in (cortex-a55): New.
10638 (cortex-a75): Likewise.
10639 (cortex-a75.cortex-a55): Likewise.
10640 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10642 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
10643 * config/arm/arm-cpu-cdata.h: Regenerate.
10644 * config/arm/arm-cpu-data.h: Regenerate.
10645 * config/arm/arm-cpu.h: Regenerate.
10646 * config/arm/arm-tables.opt: Regenerate.
10647 * config/arm/arm-tune.md: Regenerate.
10649 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10651 * haifa-sched.c (sched_create_recovery_edges): Update profile.
10653 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10655 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10658 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10660 PR tree-optimization/81292
10661 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10662 full_string_p, also call adjust_related_strinfos if the adjustment
10663 is simple, otherwise invalidate related strinfos.
10665 2017-07-04 Martin Liska <mliska@suse.cz>
10668 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10669 newly created variable as DECL_IGNORED_P.
10671 2017-07-04 Martin Liska <mliska@suse.cz>
10674 * ipa-inline.c (inline_small_functions):
10675 Use xstrdup_for_dump.
10677 2017-07-04 Tom de Vries <tom@codesourcery.com>
10679 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10681 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10684 * config/darwin.c (darwin_function_switched_text_sections):
10687 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10689 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10691 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10693 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10695 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10697 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10698 min_profitable_iters, and th as inclusive lower bounds.
10699 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10700 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10701 for min_profitable_iters and min_profitable_estimate.
10702 (vect_transform_loop): Treat th as an inclusive lower bound.
10703 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10705 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10708 * config/darwin.c (darwin_function_switched_text_sections):
10709 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10710 in two pieces, and suppress the use of buf.
10712 2017-07-03 Nathan Sidwell <nathan@acm.org>
10714 * hash-table.h (hash_table_mod1): Fix indentation.
10716 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10718 PR middle-end/81290
10719 * predict.c (force_edge_cold): Be more careful about propagation
10721 * profile-count.h (profile_probability::guessed,
10722 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10724 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10726 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10728 * doc/invoke.texi (rcpc architecture extension): Document it.
10730 2017-07-03 Richard Biener <rguenther@suse.de>
10732 PR tree-optimization/60510
10733 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10734 the scalar reduction PHI and use it.
10735 (vectorizable_reduction): Properly guard the single_defuse_cycle
10736 path for non-SLP reduction chains where we cannot use it.
10737 Rework reduc_def/index and vector type deduction. Rework
10738 vector operand gathering during reduction op code-gen.
10739 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10740 chains dissolve the chain and leave it to non-SLP reduction
10743 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10745 * tree-data-ref.h (dr_alignment): Declare.
10746 * tree-data-ref.c (dr_alignment): New function.
10747 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
10748 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
10750 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
10752 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10754 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
10755 and base_misalignment fields.
10756 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
10757 * tree-data-ref.c: Include builtins.h.
10758 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
10759 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
10760 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
10761 * tree-vect-data-refs.c: Include tree-cfg.h.
10762 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
10763 fields instead of calculating an alignment here.
10764 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
10765 innermost_loop_behavior fields.
10767 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10769 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
10771 (DR_STEP_ALIGNMENT): New macro.
10772 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
10773 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
10774 (create_data_ref): Print it.
10775 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
10776 to tell whether the step preserves vector (mis)alignment.
10777 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10778 Move the check for an integer step and generalise to all INTEGER_CST.
10779 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
10780 Print the outer step alignment.
10782 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10784 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
10785 with offset_alignment.
10786 (DR_ALIGNED_TO): Delete.
10787 (DR_OFFSET_ALIGNMENT): New macro.
10788 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
10789 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
10790 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
10791 (create_data_ref): Likewise.
10792 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10793 (vect_analyze_data_refs): Likewise.
10794 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
10795 creating dummy innermost behavior.
10797 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10799 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
10800 with a "innermost_loop_behavior *" and refeence tree.
10801 * tree-data-ref.c (dr_analyze_innermost): Likewise.
10802 (create_data_ref): Update call accordingly.
10803 * tree-predcom.c (find_looparound_phi): Likewise.
10805 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10807 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
10808 fields with dr_wrt_vec_loop.
10809 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
10810 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
10811 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
10812 (vect_dr_behavior): New function.
10813 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10814 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
10815 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
10816 track whether the step preserves the misalignment.
10817 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10818 Use vect_dr_behavior.
10819 (vect_setup_realignment): Update call accordingly.
10820 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
10821 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
10822 call to vect_create_addr_base_for_vector_ref.
10823 (vect_create_cond_for_align_checks): Likewise.
10824 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
10825 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
10826 (vect_recog_mask_conversion_pattern): Likewise.
10827 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
10828 (new_stmt_vec_info): Remove redundant zeroing.
10830 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10832 * common/config/arm/arm-common.c (arm_be8_option): New function.
10833 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
10834 (ISA_ARMv6): Add isa_bit_be8.
10835 * config/arm/arm.h (arm_be8_option): Add prototype.
10836 (BE8_SPEC_FUNCTION): New define.
10837 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
10838 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
10839 (mlittle-endian): Similarly.
10840 (mbe8, mbe32): New options.
10841 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
10842 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
10844 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10846 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
10848 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10850 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
10851 (cleanup_tree_cfg_bb): Use it.
10852 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
10854 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
10856 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10859 * loop-doloop.c (add_test): Update profile.
10861 2017-07-03 Martin Liska <mliska@suse.cz>
10864 * sanopt.c (rewrite_usage_of_param): New function.
10865 (sanitize_rewrite_addressable_params): Likewise.
10866 (pass_sanopt::execute): Call rewrite_usage_of_param.
10868 2017-07-03 Richard Biener <rguenther@suse.de>
10870 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
10871 back to using VIEW_CONVERT_EXPR.
10873 2017-07-03 Martin Liska <mliska@suse.cz>
10876 * doc/extend.texi: Document when a resolver function is
10877 generated for target_clones.
10879 2017-07-03 Martin Liska <mliska@suse.cz>
10881 * asan.c (asan_emit_stack_protection): Unpoison just red zones
10882 and shadow memory of auto variables which are subject of
10883 use-after-scope sanitization.
10884 (asan_expand_mark_ifn): Add do set only when is_poison.
10886 2016-07-03 Richard Biener <rguenther@suse.de>
10888 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
10890 (vect_force_simple_reduction): Record reduction def -> phi mapping.
10891 (vectorizable_reduction): Perform reduction PHI creation when
10892 visiting a reduction PHI and adjust and simplify code generation
10893 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
10894 (vect_transform_loop): Visit reduction PHIs.
10895 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
10896 defs into the SLP tree.
10897 (vect_build_slp_tree): Reduction defs terminate the recursion.
10898 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
10900 (vect_get_vec_defs_for_stmt_copy): Export.
10901 (vect_get_vec_defs): Likewise.
10902 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
10904 (vect_get_vec_defs_for_stmt_copy): Declare.
10905 (vect_get_vec_defs): Likewise.
10907 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10909 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
10910 parameter with a "loop" parameter and use it instead of the
10911 loop containing DR_STMT. Don't check simple_iv when doing
10912 BB analysis. Describe the two analysis modes in the comment.
10914 2017-07-03 Tom de Vries <tom@codesourcery.com>
10916 PR tree-optimization/69468
10917 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
10918 (find_same_succ_bb): Handle ignore_edge_flags.
10920 2017-07-03 Tom de Vries <tom@codesourcery.com>
10922 PR tree-optimization/81192
10923 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
10925 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
10927 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
10929 2017-07-03 Tom de Vries <tom@codesourcery.com>
10931 PR tree-optimization/81192
10932 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
10933 BB_SAME_SUCC (bb) == NULL.
10935 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10937 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
10940 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10942 * dumpfile.c: Include profile-count.h
10943 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
10945 (insert_cond_bb): Update profile.
10946 * tree-cfg.h (insert_cond_bb): Update prototype.
10947 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
10948 * tree-dump.c: Do not include tree-cfg.
10950 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10952 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
10954 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10956 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
10959 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10961 * tree-complex.c (expand_complex_div_wide): update profile.
10963 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10964 Alan Hayward <alan.hayward@arm.com>
10965 David Sherwood <david.sherwood@arm.com>
10967 * Makefile.in (MACHMODE_H): Remove insn-modes.h
10968 (CORETYPES_H): New define.
10969 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
10970 (insn-modes-inline.h, s-modes-inline-h): New rules.
10971 (generated_files): Add insn-modes-inline.h.
10972 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
10973 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
10974 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
10975 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
10976 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
10977 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
10978 (build/gencodes.o, build/genconditions.o): Likewise.
10979 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
10980 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
10981 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
10982 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
10983 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
10984 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
10985 * coretypes.h: Include everything up to real.h for generators.
10986 Include insn-modes.h first. Include wide-int-print.h after
10987 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10988 * machmode.h: Don't include insn-modes.h here.
10989 * function-tests.c: Remove includes of signop.h, machmode.h,
10990 double-int.h and wide-int.h.
10992 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10994 * optc-save-gen.awk: Likewise.
10995 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10996 * godump.c: Remove include of wide-int-print.h.
10997 * pretty-print.h: Likewise.
10998 * wide-int-print.cc: Likewise.
10999 * wide-int.cc: Likewise.
11000 * hash-map-tests.c: Remove include of signop.h.
11001 * hash-set-tests.c: Likewise.
11002 * rtl-tests.c: Likewise.
11003 * mkconfig.sh: Remove include of machmode.h.
11004 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
11006 (emit_insn_modes_inline_h): ...this new function. Emit the code
11007 into an insn-modes-inline.h header file, adding appropriate
11008 include guards and end comments.
11009 (emit_insn_modes_c_header): Remove include of machmode.h.
11010 (emit_min_insn_modes_c_header): Include coretypes.h rather than
11012 (main): Handle -i flag and call emit_insn_modes_inline_h when
11015 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11017 * tree-ssa-strlen.c (strinfo): Rename the length field to
11018 nonzero_chars. Add a full_string_p field.
11019 (compare_nonzero_chars, zero_length_string_p): New functions.
11020 (get_addr_stridx): Add an offset_out parameter.
11021 Use compare_nonzero_chars.
11022 (get_stridx): Update accordingly. Use compare_nonzero_chars.
11023 (new_strinfo): Update after above changes to strinfo.
11024 (set_endptr_and_length): Set full_string_p.
11025 (get_string_length): Update after above changes to strinfo.
11026 (unshare_strinfo): Update call to new_strinfo.
11027 (maybe_invalidate): Likewise.
11028 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
11029 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
11030 as a uhwi instead of an shwi. Update after above changes to
11031 strinfo and new_strinfo.
11032 (zero_length_string): Assert that chainsi contains full strings.
11033 Use zero_length_string_p. Update call to new_strinfo.
11034 (adjust_related_strinfos): Update after above changes to strinfo.
11035 Copy full_string_p from origsi.
11036 (adjust_last_stmt): Use zero_length_string_p.
11037 (handle_builtin_strlen): Update after above changes to strinfo and
11038 new_strinfo. Install the lhs as the string length if the previous
11039 entry didn't describe a full string.
11040 (handle_builtin_strchr): Update after above changes to strinfo
11042 (handle_builtin_strcpy): Likewise.
11043 (handle_builtin_strcat): Likewise.
11044 (handle_builtin_malloc): Likewise.
11045 (handle_pointer_plus): Likewise.
11046 (handle_builtin_memcpy): Likewise. Track nonzero characters
11047 that aren't necessarily followed by a nul terminator.
11048 (handle_char_store): Likewise.
11050 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11052 PR tree-optimization/80769
11053 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
11054 for malloc and calloc. Document the new invariant that all related
11055 strinfos have delayed lengths or none do.
11056 (verify_related_strinfos): Move earlier in file.
11057 (set_endptr_and_length): New function, split out from...
11058 (get_string_length): ...here. Also set the lengths of related
11060 (zero_length_string): Assert that chainsi has known (rather than
11062 (adjust_related_strinfos): Likewise.
11064 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11066 PR tree-optimization/81136
11067 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
11068 assert that two references with the same misalignment have the same
11069 compile-time misalignment if those compile-time misalignments
11072 2017-07-01 Andi Kleen <ak@linux.intel.com>
11074 * print-tree.c (print_node): Print all attributes.
11076 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11078 * cfg.c (scale_bbs_frequencies): New function.
11079 * cfg.h (scale_bbs_frequencies): Declare it.
11080 * cfgloopanal.c (single_likely_exit): Cleanup.
11081 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
11083 (scale_loop_profile): Likewise.
11084 (loop_version): Likewise.
11085 (create_empty_loop_on_edge): Update.
11086 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
11087 scale_loop_frequencies, scale_loop_profile, loopify,
11088 loop_version): Update prototypes.
11089 * modulo-sched.c (sms_schedule): Update.
11090 * predict.c (unlikely_executed_edge_p): Also check probability.
11091 (probably_never_executed_edge_p): Fix typo.
11092 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11093 * tree-parloops.c (gen_parallel_loop): Update.
11094 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
11095 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11096 * tree-ssa-loop-split.c (split_loop): Update.
11097 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11098 * tree-vect-loop-manip.c (vect_do_peeling): Update.
11099 (vect_loop_versioning): Update.
11100 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11102 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11104 * trans-mem.c (split_bb_make_tm_edge): Update profile.
11106 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11108 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
11109 to keep profile consistent.
11111 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11113 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
11114 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
11115 * profile-count.h (max_safe_multiplier): Make unsigned.
11116 (profile_count::guessed_zero): New.
11118 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11120 * bb-reorder.c (fix_up_crossing_landing_pad,
11121 fix_crossing_conditional_branches): Use make_single_succ_edge
11122 to keep profile consistent.
11124 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11126 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
11129 2017-07-01 Jakub Jelinek <jakub@redhat.com>
11132 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
11133 the right scopes, make sure cond_jump isn't preserved between multiple
11134 iterations. Search for fallthru edge whenever there are 3+ edges and
11135 use find_fallthru_edge for it.
11137 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11139 Patch by Alexander Monakov <amonakov@ispras.ru>
11140 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
11141 probabilities consistently.
11143 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11145 * pa.c (pa_expand_compare_and_swap_loop): Update call of
11146 emit_cmp_and_jump_insns.
11148 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11151 * tree-inline.c (expand_call_inline): Combine profile statuses.
11153 2017-06-30 Andrew Pinski <apinski@cavium.com>
11155 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
11156 fold_stmt returned true.
11158 2017-06-30 Nathan Sidwell <nathan@acm.org>
11160 * ggc.h (empty_string): Delete.
11161 * cfgexpand.c (expand_asm_stmt): Use plain "".
11162 * optabs.c (expand_asm_memory_barrier): Likewise.
11163 * stringpool.c (empty_string): Delete.
11164 (digit_vector, digit_string): Delete.
11165 (ggc_alloc_string): Use plain "", don't optimize single digit
11166 strings. Use ggc_alloc_atomic.
11168 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
11170 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
11171 comparison set and one other set, use the cost of the non-comparison
11174 2017-06-30 Nathan Sidwell <nathan@acm.org>
11176 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
11179 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
11181 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
11182 loops. Remove now unneeded calls to gimple_switch_set_label() that
11183 just set removed labels to NULL_TREE.
11185 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
11187 * tree-ssanames.c (set_range_info_raw): Abstract from ...
11188 (set_range_info): ...here. Only call set_range_info_raw if domain
11190 (set_nonzero_bits): Call set_range_info_raw.
11191 * tree-ssanames.h (set_range_info_raw): New.
11193 2017-06-30 Jakub Jelinek <jakub@redhat.com>
11196 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
11197 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
11198 of nonimmediate_operand and <store_mask_constraint> instead of m
11199 for the input operand. For V8FI iterator, always split if input
11200 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
11201 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
11202 <store_mask_predicate> instead of register_operand and
11203 <store_mask_constraint> instead of v for the input operand. Make
11204 sure both operands aren't MEMs for if not <mask_applied>.
11206 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
11208 * lto-wrapper.c (copy_file) Close both file descriptors before
11211 2017-06-30 Martin Liska <mliska@suse.cz>
11214 * multiple_target.c (create_dispatcher_calls): Make ifunc
11215 also for function that don't have calls or are not referenced.
11217 2017-06-30 Richard Biener <rguenther@suse.de>
11219 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
11220 analyze the first scalar stmt. Move vector type computation
11221 for the BB case here from ...
11222 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
11223 live operation processing in the SLP case properly.
11225 2017-06-30 Richard Biener <rguenther@suse.de>
11227 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
11229 2017-06-30 Martin Liska <mliska@suse.cz>
11232 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
11233 before BUILT_IN_UNWIND_RESUME when ASAN is used.
11235 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
11237 * doc/invoke.texi (AArch64): Add missing options and remove redundant
11240 2017-06-30 Richard Biener <rguenther@suse.de>
11242 PR tree-optimization/81249
11243 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
11244 condition reduction result to original scalar type.
11246 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11248 * profile-count.h (enum profile_quality): Fix typos and whitespace
11251 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11253 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
11254 type for branch probabilities.
11256 2017-06-29 Julian Brown <julian@codesourcery.com>
11257 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11259 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
11260 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
11261 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
11262 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
11264 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11266 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
11267 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
11268 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
11269 CC usage from generic code to here.
11270 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
11271 CC usage into the target macros.
11273 2017-06-29 Maya Rashish <coypu@sdf.org>
11275 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
11278 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11280 * arm/arm-builtins.c: Include profile-count.h
11281 * except.c (sjlj_emit_function_enter): Use
11282 profile_probability::unlikely.
11284 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11286 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
11287 and tocrel_offset be pointer args rather than implicitly using
11289 (legitimate_constant_pool_address_p, rs6000_emit_move,
11290 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
11291 tocrel_offset and use in toc_relative_expr_p call.
11292 (print_operand, print_operand_address): Use static tocrel_base_oac
11293 and tocrel_offset_oac.
11294 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
11297 2017-06-29 Maya Rashish <coypu@sdf.org>
11299 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
11301 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
11303 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
11304 objects, take into account only the alignment of 'op0' and 'mode1' if
11307 2017-06-29 Steve Ellcey <sellcey@cavium.com>
11309 * ccmp.c (ccmp_tree_comparison_p): New function.
11310 (ccmp_candidate_p): Update to use above function.
11311 (get_compare_parts): New function.
11312 (expand_ccmp_next): Update to use new functions.
11313 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
11315 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
11316 take mode as argument.
11317 * ccmp.h (expand_ccmp_expr): Add mode as argument.
11318 * expr.c (expand_expr_real_1): Pass mode as argument.
11320 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
11322 * combine.c (combine_instructions): Print insns to dump_file, together
11325 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11327 * asan.c (asan_emit_stack_protection): Update.
11328 (create_cond_insert_point): Update.
11329 * auto-profile.c (afdo_propagate_circuit): Update.
11330 * basic-block.h (struct edge_def): Turn probability to
11331 profile_probability.
11332 (EDGE_FREQUENCY): Update.
11333 * bb-reorder.c (find_traces_1_round): Update.
11334 (better_edge_p): Update.
11335 (sanitize_hot_paths): Update.
11336 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
11337 (make_single_succ_edge): Update.
11338 (check_bb_profile): Update.
11339 (dump_edge_info): Update.
11340 (update_bb_profile_for_threading): Update.
11341 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
11342 probabilitycount to 0.
11343 * cfgbuild.c (compute_outgoing_frequencies): Update.
11344 * cfgcleanup.c (try_forward_edges): Update.
11345 (outgoing_edges_match): Update.
11346 (try_crossjump_to_edge): Update.
11347 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
11348 (expand_gimple_tailcall): Update.
11349 (construct_init_block): Use make_single_succ_edge.
11350 (construct_exit_block): Use make_single_succ_edge.
11351 * cfghooks.c (verify_flow_info): Update.
11352 (redirect_edge_succ_nodup): Update.
11353 (split_edge): Update.
11354 (account_profile_record): Update.
11355 * cfgloopanal.c (single_likely_exit): Update.
11356 * cfgloopmanip.c (scale_loop_profile): Update.
11357 (set_zero_probability): Remove.
11358 (duplicate_loop_to_header_edge): Update.
11359 * cfgloopmanip.h (loop_version): Update prototype.
11360 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11361 (force_nonfallthru_and_redirect): Update.
11362 (update_br_prob_note): Update.
11363 (rtl_verify_edges): Update.
11364 (purge_dead_edges): Update.
11365 (rtl_lv_add_condition_to_bb): Update.
11366 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
11367 * cgraphunit.c (init_lowered_empty_function): Update.
11368 (cgraph_node::expand_thunk): Update.
11369 * cilk-common.c: Include profile-count.h
11370 * dojump.c (inv): Remove.
11371 (jumpifnot): Update.
11372 (jumpifnot_1): Update.
11373 (do_jump_1): Update.
11375 (do_jump_by_parts_greater_rtx): Update.
11376 (do_compare_rtx_and_jump): Update.
11377 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
11378 do_jump_1. do_compare_rtx_and_jump): Update prototype.
11379 * dwarf2cfi.c: Include profile-count.h
11380 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
11381 (sjlj_emit_dispatch_table): Likewise.
11382 * explow.c: Include profile-count.h
11383 * expmed.c (emit_store_flag_force): Update.
11384 (do_cmp_and_jump): Update.
11385 * expr.c (compare_by_pieces_d::generate): Update.
11386 (compare_by_pieces_d::finish_mode): Update.
11387 (emit_block_move_via_loop): Update.
11388 (store_expr_with_bounds): Update.
11389 (store_constructor): Update.
11390 (expand_expr_real_2): Update.
11391 (expand_expr_real_1): Update.
11392 * expr.h (try_casesi, try_tablejump): Update prototypes.
11393 * gimple-pretty-print.c (dump_probability): Update.
11394 (dump_profile): New.
11395 (dump_gimple_label): Update.
11396 (dump_gimple_bb_header): Update.
11397 * graph.c (draw_cfg_node_succ_edges): Update.
11398 * hsa-gen.c (convert_switch_statements): Update.
11399 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
11400 (find_if_case_1): Update.
11401 (find_if_case_2): Update.
11402 * internal-fn.c (expand_arith_overflow_result_store): Update.
11403 (expand_addsub_overflow): Update.
11404 (expand_neg_overflow): Update.
11405 (expand_mul_overflow): Update.
11406 (expand_vector_ubsan_overflow): Update.
11407 * ipa-cp.c (good_cloning_opportunity_p): Update.
11408 * ipa-split.c (split_function): Use make_single_succ_edge.
11409 * ipa-utils.c (ipa_merge_profiles): Update.
11410 * loop-doloop.c (add_test): Update.
11411 (doloop_modify): Update.
11412 * loop-unroll.c (compare_and_jump_seq): Update.
11413 (unroll_loop_runtime_iterations): Update.
11414 * lra-constraints.c (lra_inheritance): Update.
11415 * lto-streamer-in.c (input_cfg): Update.
11416 * lto-streamer-out.c (output_cfg): Update.
11417 * mcf.c (adjust_cfg_counts): Update.
11418 * modulo-sched.c (sms_schedule): Update.
11419 * omp-expand.c (expand_omp_for_init_counts): Update.
11420 (extract_omp_for_update_vars): Update.
11421 (expand_omp_ordered_sink): Update.
11422 (expand_omp_for_ordered_loops): Update.
11423 (expand_omp_for_generic): Update.
11424 (expand_omp_for_static_nochunk): Update.
11425 (expand_omp_for_static_chunk): Update.
11426 (expand_cilk_for): Update.
11427 (expand_omp_simd): Update.
11428 (expand_omp_taskloop_for_outer): Update.
11429 (expand_omp_taskloop_for_inner): Update.
11430 * omp-simd-clone.c (simd_clone_adjust): Update.
11431 * optabs.c (expand_doubleword_shift): Update.
11432 (expand_abs): Update.
11433 (emit_cmp_and_jump_insn_1): Update.
11434 (expand_compare_and_swap_loop): Update.
11435 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11436 * predict.c (predictable_edge_p): Update.
11437 (edge_probability_reliable_p): Update.
11438 (set_even_probabilities): Update.
11439 (combine_predictions_for_insn): Update.
11440 (combine_predictions_for_bb): Update.
11441 (propagate_freq): Update.
11442 (estimate_bb_frequencies): Update.
11443 (force_edge_cold): Update.
11444 * profile-count.c (profile_count::dump): Add missing space into dump.
11445 (profile_count::debug): Add newline.
11446 (profile_count::differs_from_p): Explicitly convert to unsigned.
11447 (profile_count::stream_in): Update.
11448 (profile_probability::dump): New member function.
11449 (profile_probability::debug): New member function.
11450 (profile_probability::differs_from_p): New member function.
11451 (profile_probability::differs_lot_from_p): New member function.
11452 (profile_probability::stream_in): New member function.
11453 (profile_probability::stream_out): New member function.
11454 * profile-count.h (profile_count_quality): Rename to ...
11455 (profile_quality): ... this one.
11456 (profile_probability): New.
11457 (profile_count): Update.
11458 * profile.c (compute_branch_probabilities): Update.
11459 * recog.c (peep2_attempt): Update.
11460 * sched-ebb.c (schedule_ebbs): Update.
11461 * sched-rgn.c (find_single_block_region): Update.
11462 (compute_dom_prob_ps): Update.
11463 (schedule_region): Update.
11464 * sel-sched-ir.c (compute_succs_info): Update.
11465 * stmt.c (struct case_node): Update.
11466 (do_jump_if_equal): Update.
11467 (get_outgoing_edge_probs): Update.
11468 (conditional_probability): Update.
11469 (emit_case_dispatch_table): Update.
11470 (expand_case): Update.
11471 (expand_sjlj_dispatch_table): Update.
11472 (emit_case_nodes): Update.
11473 * targhooks.c: Update.
11474 * tracer.c (better_p): Update.
11475 (find_best_successor): Update.
11476 * trans-mem.c (expand_transaction): Update.
11477 * tree-call-cdce.c: Update.
11478 * tree-cfg.c (gimple_split_edge): Upate.
11479 (move_sese_region_to_fn): Upate.
11480 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11481 * tree-eh.c (lower_resx): Upate.
11482 (cleanup_empty_eh_move_lp): Upate.
11483 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11484 * tree-inline.c (copy_edges_for_bb): Update.
11485 (copy_cfg_body): Update.
11486 * tree-parloops.c (gen_parallel_loop): Update.
11487 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11488 (gimple_gen_time_profiler): Update.
11489 * tree-ssa-dce.c (remove_dead_stmt): Update.
11490 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11491 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11492 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11493 (unloop_loops): Update.
11494 (try_peel_loop): Update.
11495 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11496 * tree-ssa-loop-split.c (connect_loops): Update.
11497 (split_loop): Update.
11498 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11499 (hoist_guard): Update.
11500 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11501 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11502 (value_replacement): Update.
11503 * tree-ssa-reassoc.c (branch_fixup): Update.
11504 * tree-ssa-tail-merge.c (replace_block_by): Update.
11505 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11506 (create_edge_and_update_destination_phis): Update.
11507 (compute_path_counts): Update.
11508 (recompute_probabilities): Update.
11509 (update_joiner_offpath_counts): Update.
11510 (freqs_to_counts_path): Update.
11511 (duplicate_thread_path): Update.
11512 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11513 (struct switch_conv_info): Update.
11514 (gen_inbound_check): Update.
11515 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11516 (vect_do_peeling): Update.
11517 (vect_loop_versioning): Update.
11518 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11519 (optimize_mask_stores): Update.
11520 * ubsan.c (ubsan_expand_null_ifn): Update.
11521 * value-prof.c (gimple_divmod_fixed_value): Update.
11522 (gimple_divmod_fixed_value_transform): Update.
11523 (gimple_mod_pow2): Update.
11524 (gimple_mod_pow2_value_transform): Update.
11525 (gimple_mod_subtract): Update.
11526 (gimple_mod_subtract_transform): Update.
11527 (gimple_ic): Update.
11528 (gimple_stringop_fixed_value): Update.
11529 (gimple_stringops_transform): Update.
11530 * value-prof.h: Update.
11532 2017-06-29 Carl Love <cel@us.ibm.com>
11534 * config/rs6000/rs6000-c.c: Add support for built-in functions
11535 vector signed int vec_signed (vector float);
11536 vector signed long long vec_signed (vector double);
11537 vector signed int vec_signed2 (vector double, vector double);
11538 vector signed int vec_signede (vector double);
11539 vector signed int vec_signedo (vector double);
11540 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11541 instruction generator.
11542 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11543 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11544 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11546 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11547 vunsignede_v2df): Add define_expands.
11548 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11549 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11550 VEC_UNSIGNEDO): Add definitions.
11551 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11552 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11553 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11554 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11555 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11556 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11557 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11558 vec_unsignede, vec_unsignedo): Add builtin defines.
11559 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11561 * doc/extend.texi: Update the built-in documentation file for the
11562 new built-in functions.
11564 2017-06-29 Richard Biener <rguenther@suse.de>
11566 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11567 reduction chains to LOOP_VINFO_REDUCTIONS.
11568 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11569 SLP reductions after processing reduction chains.
11571 2017-06-29 Nathan Sidwell <nathan@acm.org>
11573 * builtins.c (fold_builtin_FUNCTION): Use
11574 lang_hooks.decl_printable_name.
11576 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11578 PR middle-end/81194
11579 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11580 with only one label.
11581 * stmt.c (expand_case): Assert NCASES is greater than one.
11583 2017-06-29 Richard Biener <rguenther@suse.de>
11585 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11587 (group_case_labels): Likewise.
11588 (find_taken_edge): Push sanity checking on val to workers...
11589 (find_taken_edge_cond_expr): ... here
11590 (find_taken_edge_switch_expr): ... and here, handle cases
11591 with just a default label.
11592 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11593 (group_case_labels): Likewise.
11594 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11595 group_case_labels does anything cleanup the CFG again.
11597 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11599 PR tree-optimization/81196
11600 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11601 exit condition comparing two IVs.
11603 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11605 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11606 profile to the dummy entry at the end of the list of architectures.
11607 * config/arm/arm-cpu-cdata.h: Regenerated.
11609 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11610 Michael Collison <michael.collison@arm.com>
11613 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11615 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11616 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11617 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11618 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11619 of shift when the shift amount is masked with constant equal to
11620 the size of the mode.
11621 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11624 2017-06-29 Martin Liska <mliska@suse.cz>
11626 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11628 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11631 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11634 (edge_growth_cache_entry::edge_growth_cache_entry): New
11636 (reset_edge_growth_cache): Update to use constructor.
11638 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11640 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11641 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11642 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11644 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
11646 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
11647 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
11649 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
11651 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
11652 (*-linux-uclibc*): Add t-uclibc tmake_file.
11653 * config/t-musl: New.
11654 * config/t-uclibc: New.
11656 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11658 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11660 (gen_comm_data): Emit architectural setting of arch_prof.
11661 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11663 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11664 (armv8-m.base, armv8-m.main): Likewise.
11665 * arm-protos.h (arm_build_target): Add profile field.
11666 (arch_option): Likewise.
11667 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11669 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11670 arm_active_target.profile.
11672 2017-06-28 Richard Biener <rguenther@suse.de>
11674 PR middle-end/81227
11675 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11676 TYPE_OVERFLOW_WRAPS.
11677 * match.pd (negate_expr_p): Likewise.
11678 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11679 fold_build2, not fold_binary.
11681 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11683 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11684 Convert memory address to Pmode.
11685 (aarch64_print_operand): Assert MEM operands are always Pmode.
11687 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11690 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11691 Remove redundant if.
11692 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11693 * config/arm/aarch-common-protos.h
11694 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11695 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11697 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11700 * multiple_target.c (create_dispatcher_calls): Set the default
11701 clone to be static, not public.
11703 2017-06-28 Richard Biener <rguenther@suse.de>
11705 * tree-vect-loop.c (vectorizable_reduction): Move special
11706 cond reduction IV var creation ...
11707 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11708 parameter. Use STMT_VINFO_VECTYPE.
11709 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11712 2017-06-28 Martin Liska <mliska@suse.cz>
11715 * ipa-visibility.c (non_local_p): Handle visibility.
11717 2017-06-28 Martin Liska <mliska@suse.cz>
11720 * common.opt: Add IntegerRange to various options.
11721 * opt-functions.awk (integer_range_info): New function.
11722 * optc-gen.awk: Add integer_range_info to cl_options struct.
11723 * opts-common.c (decode_cmdline_option): Handle
11724 CL_ERR_INT_RANGE_ARG.
11725 (cmdline_handle_error): Likewise.
11726 * opts.c (print_filtered_help): Show valid interval in
11727 when --help is provided.
11728 * opts.h (struct cl_option): Add range_min and range_max fields.
11729 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11731 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11733 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11734 (x * C EQ/NE y * C): New transformation.
11736 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11738 * genmultilib (combination_space): Accept '+' in option names.
11740 2017-06-28 Martin Liska <mliska@suse.cz>
11743 * asan.c (instrument_derefs): Bail out inner references
11744 that are hard register variables.
11746 2017-06-28 Jakub Jelinek <jakub@redhat.com>
11749 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
11750 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
11752 2017-06-28 Richard Biener <rguenther@suse.de>
11754 * tree-vectorizer.h (vect_get_vec_defs): Remove.
11755 (vect_get_slp_defs): Adjust.
11756 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
11758 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
11760 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11761 get_initial_defs_for_reduction instead of vect_get_vec_defs.
11762 (vectorizable_reduction): Adjust.
11763 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
11765 (vect_get_slp_defs): Likewise.
11766 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
11767 (vectorizable_bswap): Adjust.
11768 (vectorizable_call): Likewise.
11769 (vectorizable_conversion): Likewise.
11770 (vectorizable_assignment): Likewise.
11771 (vectorizable_shift): Likewise.
11772 (vectorizable_operation): Likewise.
11773 (vectorizable_store): Likewise.
11774 (vectorizable_condition): Likewise.
11775 (vectorizable_comparison): Likewise.
11777 2017-06-28 Michael Collison <michael.collison@arm.com>
11780 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
11782 (arm_gen_movmemqi): Removed unused variable 'i'.
11783 Convert 'for' loop into 'while' loop.
11784 (arm_expand_prologue): Remove last unnecessary set of insn.
11785 (thumb_pop): Remove unused variable 'pushed_words'.
11786 (thumb_exit): Remove last unnecessary set of regs_to_pop.
11788 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11790 * config/s390/predicates.md: Use s390_rel_address_ok_p.
11791 * config/s390/s390-protos.h: Add prototype of
11792 s390_rel_address_ok_p.
11793 * config/s390/s390.c (s390_got_symbol): New function.
11794 (s390_rel_address_ok_p): New function.
11795 (legitimize_pic_address): Use s390_rel_address_ok_p.
11796 (s390_load_got): Use s390_got_symbol.
11797 (s390_option_override): Issue error if
11798 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
11799 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
11801 * config/s390/s390.opt: New option mpic-data-is-text-relative.
11803 2017-06-27 Andrew Pinski <apinski@cavium.com>
11805 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
11806 (X * copysign (1.0, X)): New pattern.
11807 (X * copysign (1.0, -X)): New pattern.
11808 (copysign (-1.0, CST)): New pattern.
11810 2017-06-27 Joseph Myers <joseph@codesourcery.com>
11812 * genmultilib (combination_space): Remove variable.
11813 Validate reuse rules against regular expression for any sequence
11814 of multilib options in any order.
11816 2017-06-27 Michael Collison <michael.collison@arm.com>
11818 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
11819 call aarch64_split_simd_combine.
11820 * (aarch64_combine_internal<mode>): Delete pattern.
11821 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
11822 Allow register and subreg operands.
11824 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11826 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
11827 specific need, just fallback on defaults.
11828 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
11830 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11831 Olivier Hainque <hainque@adacore.com>
11833 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
11835 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
11836 targets. Pick a default if no particular attempt applied.
11837 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
11840 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11842 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
11843 (x86_64-wrs-vxworks7): Likewise.
11845 2017-06-27 Marek Polacek <polacek@redhat.com>
11848 * ubsan.c (instrument_null): Check get_base_address's result for null.
11850 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11852 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
11854 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11856 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
11857 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
11858 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
11859 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
11860 New function types.
11861 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
11862 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
11863 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
11864 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
11865 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
11866 BUILT_IN_FEUPDATEENV): New builtins.
11867 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
11868 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
11869 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
11870 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
11872 (builtin_structptr_types): Adjust size.
11873 * tree.c (builtin_structptr_types): Add four entries.
11875 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11876 Olivier Hainque <hainque@adacore.com>
11878 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
11879 (TLS_SYM): New local macro, forcing reference to __tls__ on
11880 link command lines for VxWorks 7 RTPs, triggering initialization
11882 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
11883 OS features TLS support, true for RTPs on VxWorks 7.
11884 * config/vxworks.c (vxworks_override_options): Setup emutls
11887 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11889 * predict.c (test_prediction_value_range): Use -1U instead of -1
11890 to avoid narrowing conversion warning.
11891 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
11892 to avoid narrowing conversion warning.
11893 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
11895 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
11897 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11899 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
11900 64bit configurations.
11901 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
11902 (SIZE_TYPE): Likewise.
11903 * config/vxworks.c (vxworks_emutls_var_fields): Use
11904 long_unsigned_type_node instead of unsigned_type_node as the offset
11905 field type, which is "pointer" mode in emutls.c.
11907 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11910 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
11912 PR middle-end/81207
11913 * gimple-fold.c (replace_call_with_call_and_fold): Handle
11914 gimple_vuse copying separately from gimple_vdef copying.
11916 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11918 * value-prof.c (free_hist): Remove call to memset and the enclosing if
11921 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11922 Olivier Hainque <hainque@adacore.com>
11924 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
11925 for all vxworks7 targets.
11926 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
11927 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
11928 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
11929 variations for VX6/VX7 and 32/64bits later on in ...
11930 (VXWORKS_LIB_SPEC): Leverage new macros.
11931 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
11932 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
11934 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11936 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
11937 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
11939 2017-06-26 Carl Love <cel@us.ibm.com>
11941 * config/rs6000/rs6000-c.c: Add support for built-in functions
11942 vector bool char vec_reve (vector bool char);
11943 vector signed char vec_reve (vector signed char);
11944 vector unsigned char vec_reve (vector unsigned char);
11945 vector bool int vec_reve (vector bool int);
11946 vector signed int vec_reve (vector signed int);
11947 vector unsigned int vec_reve (vector unsigned int);
11948 vector bool long long vec_reve (vector bool long long);
11949 vector signed long long vec_reve (vector signed long long);
11950 vector unsigned long long vec_reve (vector unsigned long long);
11951 vector bool short vec_reve (vector bool short);
11952 vector signed short vec_reve (vector signed short);
11953 vector double vec_reve (vector double);
11954 vector float vec_reve (vector float);
11955 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
11956 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
11957 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
11958 (altivec_vreve): New pattern.
11959 * config/rs6000/altivec.h (vec_reve): New define.
11960 * doc/extend.texi (vec_rev): Update the built-in documentation file
11961 for the new built-in functions.
11963 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11965 PR tree-optimization/71815
11966 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
11968 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
11970 (slsr_process_phi): Likewise.
11971 (replace_uncond_cands_and_profitable_phis): Don't replace a
11972 multiply candidate with a stride of 1 (copy or cast).
11973 (phi_incr_cost): Call uses_consumed_by_stmt rather than
11975 (lowest_cost_path): Likewise.
11976 (total_savings): Likewise.
11978 2017-06-26 Richard Biener <rguenther@suse.de>
11981 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
11982 Use def_builtin_pure for all gather builtins.
11984 2017-06-26 Richard Biener <rguenther@suse.de>
11986 PR tree-optimization/81203
11987 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11988 non-dominating BBs.
11990 2017-06-26 Marek Polacek <polacek@redhat.com>
11993 * doc/invoke.texi: Document -Wmultistatement-macros.
11995 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11997 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11998 arm_neon_ok_no_float_abi effective target.
12000 2017-06-26 Richard Biener <rguenther@suse.de>
12002 PR tree-optimization/80928
12003 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
12004 (copy_bbs): Set BB_DUPLICATED flag early.
12005 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
12007 (execute_on_shrinking_pred): Likewise.
12008 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
12009 BB_DUPLICATED blocks.
12010 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
12011 iterate over all PHIs considering removal of *gsi.
12013 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
12015 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
12018 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12020 * config/rs6000/rs6000-string.c: (expand_block_clear,
12021 do_load_for_compare, select_block_compare_mode,
12022 compute_current_alignment, expand_block_compare,
12023 expand_strncmp_align_check, expand_strn_compare,
12024 expand_block_move, rs6000_output_load_multiple)
12025 Move functions related to string/block move/compare
12026 to a separate file.
12027 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
12028 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
12029 for this function which is now used in two files.
12030 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
12031 * config.gcc: Add rs6000-string.o to extra_objs for
12032 targets powerpc*-*-* and rs6000*-*-*.
12034 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
12037 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
12038 32-bit, since indexed is not valid for DImode.
12039 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
12040 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
12041 (define_peephole2 for Altivec d-form load): Add 32-bit support.
12042 (define_peephole2 for Altivec d-form store): Likewise.
12045 * multiple_target.c (create_dispatcher_calls): Only create the
12046 dispatcher call if the function is the default clone of a
12047 versioned function.
12049 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
12051 PR middle-end/80902
12052 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
12053 a call, force the call to not be a tail call.
12055 2017-06-23 Jeff Law <law@redhat.com>
12057 * doc/contrib.texi: Add entry for Steven Pemberton's work on
12060 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
12062 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
12063 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
12064 handling for early expansion of vector shifts (sl,sr,sra,rl).
12065 (builtin_function_type): Add vector shift right instructions
12066 to the unsigned argument list.
12068 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12070 rtl-optimizatoin/79286
12071 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
12072 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
12073 trap. PIC register plus a const unspec without offset can never trap.
12075 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
12077 * tree.h (builtin_structptr_type): New type.
12078 (builtin_structptr_types): Declare new array.
12079 * tree.c (builtin_structptr_types): New array.
12080 (free_lang_data, build_common_tree_nodes): Use it.
12082 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
12085 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
12088 2017-06-22 Matt Turner <mattst88@gmail.com>
12090 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
12091 Lake models to skylake case. Assume skylake for unknown
12092 models with clflushopt.
12094 2017-06-22 Jeff Law <law@redhat.com>
12096 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
12097 frame sizes that do not satisfy aarch64_uimm12_shift.
12099 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
12101 * profile-count.h (apply_probability,
12102 apply_scale, probability_in): Fix checks for zero.
12104 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12106 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
12107 * doc/cppdiropts.texi (-I @var{dir}): Document it.
12109 2016-06-22 Richard Biener <rguenther@suse.de>
12111 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12112 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
12113 REDUC_MAX_EXPR support.
12114 (vectorizable_reduction): Likewise.
12115 (vect_create_epilog_for_reduction): Likewise.
12117 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
12119 * match.pd (A / (1 << B) -> A >> B): New.
12120 * generic-match-head.c: Include optabs-tree.h.
12121 * gimple-match-head.c: Likewise.
12122 * optabs-tree.h (target_supports_op_p): New.
12123 * optabs-tree.c (target_supports_op_p): New.
12125 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12127 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
12128 $gcc_cv_ld --help output.
12129 (gcc_cv_ld_demangle): Likewise.
12130 (gcc_cv_ld_eh_frame_hdr): Likewise.
12131 (gcc_cv_ld_pie): Likewise.
12132 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
12133 (gcc_cv_ld_buildid): Likewise.
12134 (gcc_cv_ld_sysroot): Likewise.
12135 (ld_bndplt_support): Likewise.
12136 (ld_pushpopstate_support): Likewise.
12137 * configure: Regenerate.
12138 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
12140 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12143 * config/i386/sse.md (round<mode>2): Renumber match_dup and
12144 operands indexes to avoid gap between operands and match_dups.
12146 2017-06-21 Andrew Pinski <apinski@cavium.com>
12148 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
12149 Increment Arith_shift and Arith_shift_reg by 1.
12150 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
12152 * config/aarch64/aarch64.c (thunderx_tunings): Enable
12153 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
12154 (aarch64_strip_extend): Add new argument and test for it.
12155 (aarch64_cheap_mult_shift_p): New function.
12156 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
12157 add a cost if it is true.
12158 Update calls to aarch64_strip_extend.
12159 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
12161 2017-06-21 Andrew Pinski <apinski@cavium.com>
12163 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
12165 (thunderxt88): Likewise.
12166 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
12167 (thunderx_prefetch_tune): New variable.
12168 (thunderx2t99_prefetch_tune): Update for the correct values.
12169 (thunderxt88_tunings): New variable.
12170 (thunderx_tunings): Use thunderx_prefetch_tune instead of
12171 generic_prefetch_tune.
12172 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
12174 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12176 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
12177 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
12178 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
12179 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
12180 (aarch64_atomic_cas<mode>, GPI): Likewise.
12182 2017-06-21 Martin Liska <mliska@suse.cz>
12184 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
12185 statements on cold and hot labels.
12186 * predict.c (tree_estimate_probability_bb): Remove the
12187 prediction from this place.
12189 2017-06-21 Martin Liska <mliska@suse.cz>
12191 PR tree-optimization/79489
12192 * gimplify.c (maybe_add_early_return_predict_stmt): New
12194 (gimplify_return_expr): Call the function.
12195 * predict.c (tree_estimate_probability_bb): Remove handling
12197 * predict.def: Update comment about early return predictor.
12198 * gimple-predict.h (is_gimple_predict): New function.
12199 * predict.def: Change default value of early return to 66.
12200 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
12202 * passes.def: Put pass_strip_predict_hints to the beginning of
12205 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
12207 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
12208 FUNCTION_DECL declarations.
12209 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
12211 (dwaf2out_decl): Likewise.
12212 * godump.c (go_early_global_decl): Skip call to the real debug hook
12213 for FUNCTION_DECL declarations.
12214 * passes.c (rest_of_decl_compilation): Skip call to the
12215 early_global_decl debug hook for FUNCTION_DECL declarations, unless
12216 -fdump-go-spec is passed.
12218 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12220 * config/i386/i386.c (struct builtin_isa): New field pure_p.
12221 Reorder for compactness.
12222 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
12223 (def_builtin_pure, def_builtin_pure2): New functions.
12224 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
12226 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12228 * match.pd (nop_convert): New predicate.
12229 ((A +- CST1) +- CST2): Allow some NOP conversions.
12231 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12234 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
12235 with ctors/dtors if GOVD_SHARED is set.
12237 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12239 * config/aarch64/aarch64.md (movti_aarch64):
12240 Emit mov rather than orr.
12241 (movtf_aarch64): Likewise.
12242 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
12243 Emit mov rather than orr.
12245 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12247 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
12248 Swap alternatives, make integer dup more expensive.
12250 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12252 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
12253 Return true for non-tls symbols.
12255 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
12257 * config/aarch64/aarch64-cores.def (cortex-a55): New.
12258 (cortex-a75): Likewise.
12259 (cortex-a75.cortex-a55): Likewise.
12260 * config/aarch64/aarch64-tune.md: Regenerate.
12261 * doc/invoke.texi (-mtune): Document new values for -mtune.
12263 2017-06-21 Tom de Vries <tom@codesourcery.com>
12265 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
12266 stack_size feature.
12267 (Effective-Target Keywords, Other attributes): Suggest using
12268 dg-add-options stack_size feature to get stack limit in stack_size
12269 effective target documentation.
12271 2017-06-21 Julian Brown <julian@codesourcery.com>
12272 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12274 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
12275 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
12276 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
12278 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
12279 attribute type list for neon_multiply.
12280 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
12281 attribute type list for neon_multiply.
12282 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
12283 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
12284 attribute type list for neon_multiply.
12285 * config/arm/types.md (crypto_pmull): Add.
12286 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
12287 attribute type list.
12289 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
12291 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
12294 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12296 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
12297 to make sure not to dereference a NULL cost_classes_ptr pointer.
12299 2017-06-20 Carl Love <cel@us.ibm.com>
12301 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12302 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12303 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12304 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12305 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12306 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12307 VMULOSW): New enum "unspec" values.
12308 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12309 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12310 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12311 altivec_vmulosw): New patterns.
12312 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12313 VMULOSW): Add definitions.
12315 2017-06-20 Julia Koval <julia.koval@intel.com>
12317 * config/i386/i386.c: Fix rounding expand for new pattern.
12318 * config/i386/subst.md: Fix pattern (parallel -> unspec).
12320 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12322 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
12323 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
12325 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12327 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
12330 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12332 * config/aarch64/aarch64-cores.def: Rearrange to sort by
12333 architecture, then by implementer ID.
12334 * config/aarch64/aarch64-tune.md: Regenerate.
12336 2017-06-20 Richard Biener <rguenther@suse.de>
12338 PR middle-end/81097
12339 * fold-const.c (split_tree): Fold to type before negating.
12341 2017-06-20 David Malcolm <dmalcolm@redhat.com>
12343 * diagnostic-show-locus.c
12344 (selftest::test_fixit_deletion_affecting_newline): New function.
12345 (selftest::diagnostic_show_locus_c_tests): Call it.
12347 2017-06-20 Andreas Schwab <schwab@suse.de>
12350 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
12353 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12355 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
12356 __ARM_FEATURE_COPROC according to support.
12358 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12360 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12361 Rewritten to avoid overflow for > 32-bit pointers.
12364 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
12365 by removing enum keyword.
12366 (ubsan_type_descriptor): Likewise. Formatting fix.
12369 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
12370 splitter): Require TARGET_SSE2 in the condition.
12372 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12375 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
12376 for doing vector set of SFmode on ISA 3.0.
12377 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
12378 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
12380 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
12381 SFmode value into a V4SF variable that was extracted from another
12382 V4SF variable without converting the element to double precision
12383 and back to single precision vector format.
12384 (vsx_insert_extract_v4sf_p9_2): Likewise.
12386 2017-06-19 Jakub Jelinek <jakub@redhat.com>
12388 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
12389 in UWHI to avoid undefined overflow.
12392 * ubsan.h (enum ubsan_encode_value_phase): New.
12393 (ubsan_encode_value): Change second argument to
12394 enum ubsan_encode_value_phase with default value of
12395 UBSAN_ENCODE_VALUE_GENERIC.
12396 * ubsan.c (ubsan_encode_value): Change second argument to
12397 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
12398 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
12399 create_tmp_var_raw instead of create_tmp_var and use a
12401 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12402 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12403 ubsan_encode_value callers.
12406 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12407 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12408 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12410 2017-06-19 Richard Biener <rguenther@suse.de>
12412 PR middle-end/81118
12413 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12414 estimates if we changed anything.
12416 2017-06-19 Richard Biener <rguenther@suse.de>
12418 PR tree-optimization/80887
12419 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12420 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12421 simplified lookups, then reset mprts_hook.
12422 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12424 (try_to_simplify): Likewise.
12426 2017-06-19 Martin Liska <mliska@suse.cz>
12429 * gimplify.c (gimplify_switch_expr):
12430 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12432 2017-06-19 Martin Liska <mliska@suse.cz>
12434 * doc/install.texi: Document that PGO runs in 4 stages.
12436 2017-06-19 Martin Liska <mliska@suse.cz>
12439 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12440 to dispatcher function name.
12441 * multiple_target.c (replace_function_decl): New function.
12442 (create_dispatcher_calls): Redirect both edges and references.
12444 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12446 * profile-count.c (profile_count::dump): Dump quality.
12447 (profile_count::differs_from_p): Update for unsigned val.
12448 * profile-count.h (profile_count_quality): New enum.
12449 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12451 2017-06-19 Richard Biener <rguenther@suse.de>
12453 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12454 struct function as arg.
12455 (estimate_numbers_of_iterations): Export overload with loop arg.
12456 (free_numbers_of_iterations_estimates_loop): Use an overload of
12457 free_numbers_of_iterations_estimates instead.
12458 * tree-cfg.c (remove_bb): Adjust.
12459 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12460 * tree-parloops.c (gen_parallel_loop): Likewise.
12461 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12463 (tree_unroll_loops_completely): Likewise.
12464 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12465 Use an overload instead and export.
12466 (estimated_loop_iterations): Adjust.
12467 (max_loop_iterations): Likewise.
12468 (likely_max_loop_iterations): Likewise.
12469 (estimate_numbers_of_iterations): Take struct function as arg
12471 (loop_exits_before_overflow): Adjust.
12472 (free_numbers_of_iterations_estimates_loop): Use an overload.
12473 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12474 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12476 2017-06-19 Richard Biener <rguenther@suse.de>
12479 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12480 RANGE_EXPR conservatively.
12482 2017-06-16 Carl Love <cel@us.ibm.com>
12484 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12485 definitions for vec_float, vec_float2, vec_floato,
12486 vec_floate built-ins.
12487 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12488 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12490 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12491 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12492 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12493 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12494 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12495 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12496 vec_floato): Add builtin defines.
12497 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12498 Update the built-in documentation file for the new built-in
12501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12503 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12504 (mthumb): Mark as the negative of -marm.
12506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12508 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12510 (ARM Options, -mtune): Document that this accepts the same
12511 extension options as -mcpu.
12512 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12516 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12517 permitted extensions.
12519 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12521 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12522 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12523 (armv8-m.main): Add option +nodsp.
12524 * config/arm/arm-cpu-cdata.h: Regenerated.
12526 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12528 * config/arm/t-fuchsia: New file.
12529 * config.gcc (arm*-*-fuchsia*): Use it.
12531 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12533 * config/arm/t-symbian: Rewrite for new option infrastructure.
12535 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12537 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12538 (MULTILIB_REQUIRED): Likewise.
12540 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12542 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12543 (MULTILIB_RESUE): Likewise.
12544 (MULTILIB_MATCHES): Likewise.
12545 (MULTLIB_REQUIRED): Likewise.
12547 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12549 * config/arm/t-rtems: Rewrite for new option framework.
12551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12553 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12554 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12555 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12556 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12557 * config/arm/t-multilib: ... here.
12558 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12559 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12560 armv7-a and armv8*-a when A-profile libraries have not been built.
12561 * config/arm/t-rmprofile: Rewrite.
12563 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12565 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12566 with a backslash. Remove the backslash after substituting unescaped
12568 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12570 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12572 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12573 the driver to pass through the default setting of -mfloat-abi.
12574 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12576 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12577 (all_feat_combs): New rule.
12578 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12580 * config/arm/t-aprofile: Rewrite.
12582 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12584 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12585 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12586 fpu is not specified by the user/command-line.
12587 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12588 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12589 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12590 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12591 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12592 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12594 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12596 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12597 * config/arm/t-arm-elf: Rewritten.
12599 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12601 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12602 have some floating-point instructions.
12603 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12604 (TARGET_MAYBE_HARD_FLOAT): New macro.
12605 * config/arm/arm-builtins.c (arm_init_builtins): Use
12606 TARGET_MAYBE_HARD_FLOAT.
12607 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12609 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12611 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12612 (configargs.h): Include it.
12613 (arm_print_hint_for_fpu_option): New function.
12614 (arm_parse_fpu_option): New function.
12615 (candidate_extension): New class.
12616 (arm_canon_for_multilib): New function.
12617 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12618 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12619 (ARCH_CANONICAL_SPECS): New macro.
12620 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12622 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12624 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12625 are set after handling multilib fragments. Set target_cpu_default2
12628 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12630 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12632 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12633 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12635 (check_arch): Likewise.
12636 * config/arm/arm.c (arm_configure_build_target): Handle
12637 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12638 options in the default.
12640 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12642 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
12643 when an option is an alias of another.
12644 * config/arm/parsecpu.awk (optalias): New parser token.
12645 (gen_comm_data): Mark non-alias options as such. Emit entries
12646 for extension aliases.
12647 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
12648 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
12649 (armv6kz, armv6zk, armv6t2): Likewise.
12650 (armv7): Make vfpv3-d16 an alias.
12651 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
12653 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
12654 Sort in canonical order.
12655 (armv8-a): Sort in canonical order.
12656 (armv8.1-a, armv8.2-a): Likewise.
12657 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12659 (cortex-a9): Sort in canonical order.
12660 * config/arm/arm.c (selftests.h): Include it.
12661 (arm_test_cpu_arch_data): New function.
12662 (arm_run_self_tests): New function.
12663 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12664 (targetm): Move declaration to the end of the file.
12665 * arm-cpu-cdata.h: Regenerated.
12667 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12669 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12670 call to target_mode_check describing the type of option passed.
12671 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12672 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12673 arm_parse_cpu_option_name to match parameters against list of
12675 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12676 arm_arch_core_flags data structure.
12677 * config/arm/arm-cpu_cdata.h: Regenerated.
12679 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12681 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12683 (arm_print_hint_for_cpu_option): Likewise.
12684 (arm_print_hint_for_arch_option): Likewise.
12685 (arm_parse_cpu_option_name): Likewise.
12686 (arm_parse_arch_option_name): Likewise.
12687 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12688 of entries in the all_fpus list.
12689 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12690 (arm_parse_cpu_option_name): Declare.
12691 (arm_parse_arch_option_name): Declare.
12692 (arm_parse_option_features): Declare.
12693 (arm_intialize_isa): Declare.
12694 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12696 (gen_comm_data): ... here. Make definitions non-static.
12697 * config/arm/arm-cpu-data.h: Regenerated.
12698 * config/arm/arm-cpu-cdata.h: Regenerated.
12700 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12702 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12703 (cpu_arch_extension): New structure.
12704 (cpu_arch_option, arch_option, cpu_option): New structures.
12705 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12706 architecture types.
12707 (gen_data): Generate new format data tables.
12708 * config/arm/arm.c (cpu_tune): New structure.
12709 (cpu_option, processors): Delete.
12710 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12711 (arm_print_hint_for_cpu_option): ... this and ...
12712 (arm_print_hint_for_arch_option): ... this.
12713 (arm_parse_arch_cpu_name): Delete. Replace with ...
12714 (arm_parse_cpu_option_name): ... this and ...
12715 (arm_parse_arch_option_name): ... this.
12716 (arm_unrecognized_feature): Change type of target parameter to
12718 (arm_parse_arch_cpu_features): Delete. Replace with ...
12719 (arm_parse_option_features): ... this.
12720 (arm_configure_build_target): Rework to use new configuration data
12722 (arm_print_tune_info): Rework for new configuration data tables.
12723 * config/arm/arm-cpu-data.h: Regenerated.
12724 * config/arm/arm-cpu.h: Regenerated.
12726 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12728 * Makefile.in (OBJS): Move sbitmap.o from here ...
12729 (OBJS-libcommon): ... to here.
12731 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12733 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12734 (ISA_ALL_CRYPTO): New macro.
12735 (ISA_ALL_SIMD): New macro
12736 (ISA_ALL_FP): New macro.
12737 * config/arm/arm.c (fpu_bitlist): Update initializer.
12738 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12740 (arm9e): Add fpu. Add option for nofp
12741 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12742 (arm926ej-s, arm1026ej-s): Likewise.
12743 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12744 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
12745 neon-fp16, neon-vfpv4, nofp and nosimd.
12746 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
12747 (cortex-a8): Add fpu. Add option for nofp.
12748 (cortex-a9): Add fpu. Add options for nosimd and nofp.
12749 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
12750 (cortex-r4f): Add fpu.
12751 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
12752 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
12754 (cortex-r8): Likewise.
12755 (cortex-m4): Add fpu. Add option for nofp.
12756 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
12757 (cortex-a17.cortex-a7): Likewise.
12758 (cortex-a32): Add fpu. Add options for crypto and nofp.
12759 (cortex-a35, cortex-a53): Likewise.
12760 (cortex-a57): Add fpu. Add option for crypto.
12761 (cortex-a72, cortex-a73): Likewise.
12762 (exynos-m1): Likewise.
12763 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
12764 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
12765 (cortex-m33): Add fpu. Add option for nofp.
12766 * config/arm/arm-cpu-cdata.h: Regenerated
12767 * config/arm/arm-cpu-data.h: Regenerated.
12769 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12771 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
12772 (armv5te, armv5tej): Likewise.
12773 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
12774 (armv7): Add options fp and vfpv3-d16.
12775 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
12776 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
12778 (armv7ve): Likewise.
12779 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
12780 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
12781 (armv8-a): Add nocrypto option.
12782 (armv8.1-a, armv8.2-a): Likewise.
12783 (armv8-m.main): add options fp, fp.dp and nofp.
12785 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12787 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
12789 (armv8-a+crc): Delete.
12790 (armv8.1-a): Add options simd, crypto and nofp.
12791 (armv8.2-a): Add options fp16, simd, crypto and nofp.
12792 (armv8.2-a+fp16): Delete.
12793 (armv8-m.main): Add option dsp.
12794 (armv8-m.main+dsp): Delete.
12795 (cortex-a8): Add fpu. Add nofp option.
12796 (cortex-a9): Add fpu. Add nofp and nosimd options.
12797 * config/arm/parsecpu.awk (gen_data): Generate option tables and
12798 link to main cpu and architecture data structures.
12799 (gen_comm_data): Only put isa attributes from the main architecture
12801 (option): New statement for architecture and CPU entries.
12802 * arm.c (struct cpu_option): New structure.
12803 (struct processors): Add entry for options.
12804 (arm_unrecognized_feature): New function.
12805 (arm_parse_arch_cpu_name): Ignore any characters after the first
12807 (arm_parse_arch_cpu_feature): New function.
12808 (arm_configure_build_target): Separate out any CPU and architecture
12809 features and parse separately. Don't error out if -mfpu=auto is
12810 used with only an architecture string.
12811 (arm_print_asm_arch_directives): New function.
12812 (arm_file_start): Call it.
12813 * config/arm/arm-cpu-cdata.h: Regenerated.
12814 * config/arm/arm-cpu-data.h: Likewise.
12815 * config/arm/arm-tables.opt: Likewise.
12817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12819 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
12820 assembler when it is not -mfpu=auto.
12822 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12824 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
12825 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
12826 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
12827 (ASM_CPU_SPEC): Rewrite.
12828 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
12829 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
12830 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
12831 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
12832 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
12833 copied string is NUL-terminated. Also strip any characters prefixed
12835 (arm_rewrite_selected_arch): New function.
12836 (arm_rewrite_march): New function.
12838 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12840 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
12841 (x_arm_cpu_string, x_arm_tune_string): Likewise.
12842 (march, mcpu, mtune): Convert to string-based options.
12843 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
12844 (arm_parse_arch_cpu_name): New function.
12845 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
12846 identify selected architecture or CPU.
12847 (arm_option_save): New function.
12848 (TARGET_OPTION_SAVE): Redefine.
12849 (arm_option_restore): Restore string options.
12850 (arm_option_print): Print string options.
12852 2017-06-16 Martin Sebor <msebor@redhat.com>
12854 PR tree-optimization/80933
12855 PR tree-optimization/80934
12856 * builtins.c (fold_builtin_3): Do not handle bcmp here.
12857 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
12858 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
12859 (gimple_fold_builtin): Call them.
12861 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12863 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
12864 as unlikely; update profile.
12866 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12868 * predict.c (force_edge_cold): Handle declaring edges impossible
12871 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12873 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
12875 (try_unroll_loop_completely): Fix reporting.
12877 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12879 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
12881 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
12884 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
12885 if given a non-constant argument for an intrinsic which requires a
12888 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12890 * profile.c (compare_freqs): New function.
12891 (branch_prob): Sort edge list.
12892 (find_spanning_tree): Assume that the list is priority sorted.
12894 2017-06-16 Richard Biener <rguenther@suse.de>
12896 PR tree-optimization/81090
12897 * passes.def (pass_record_bounds): Remove.
12898 * tree-pass.h (make_pass_record_bounds): Likewise.
12899 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
12900 make_pass_record_bounds): Likewise.
12901 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
12902 not free niter estimates at the beginning but at the end.
12903 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
12905 2017-06-16 Richard Biener <rguenther@suse.de>
12907 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
12908 initializer to workaround ICE in host GCC 4.8.
12910 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12912 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
12914 (clone_inlined_nodes): Update.
12916 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12918 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
12919 prefetch settings, and enable prefetching by default at -O3.
12921 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12923 * config/aarch64/aarch64.c (aarch64_override_options_internal):
12924 Set flag_prefetch_loop_arrays according to tuning data.
12926 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12928 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
12929 New tune structure.
12930 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
12931 [Unrelated to main purpose of the patch] Place the pointer field last
12932 to enable type checking errors when tune structure are wrongly merged.
12933 * config/aarch64/aarch64.c (generic_prefetch_tune,)
12934 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
12935 (thunderx2t99_prefetch_tune): New tune constants.
12936 (tune_params *_tunings): Update all tunings (no functional change).
12937 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
12938 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
12939 from tunings structures.
12941 2017-06-16 Jakub Jelinek <jakub@redhat.com>
12944 * ubsan.c (instrument_null): Add T argument, use it instead
12945 of computing it based on IS_LHS.
12946 (instrument_object_size): Likewise.
12947 (pass_ubsan::execute): Adjust instrument_null and
12948 instrument_object_size callers to pass gimple_get_lhs or
12949 gimple_assign_rhs1 result to it. Use instrument_null instead of
12950 calling get_base_address and instrument_mem_ref. Handle
12951 aggregate call arguments for object-size sanitization.
12953 2017-06-16 Yury Gribov <tetra2005@gmail.com>
12955 PR tree-optimization/81089
12956 * tree-vrp.c (is_masked_range_test): Validate operands of
12959 2017-06-15 Martin Sebor <msebor@redhat.com>
12962 * dumpfile.c (dump_register): Avoid calling memset to initialize
12963 a class with a default ctor.
12964 * gcc.c (struct compiler): Remove const qualification.
12965 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
12966 * hash-table.h: Ditto.
12967 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
12969 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
12970 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
12972 * params.h (struct param_info): Make struct members non-const.
12973 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
12974 with default initialization.
12975 * vec.h (vec_copy_construct, vec_default_construct): New helper
12977 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
12978 with vec_copy_construct.
12979 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
12980 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
12981 * doc/invoke.texi (-Wclass-memaccess): Document.
12983 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12985 * emit-rtl.h (is_leaf): Update comment about local
12986 register allocator.
12988 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12991 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12992 for a variable to have a section before checking if the section has a
12994 Set section to.persistent if persistent attribute is set.
12995 Warn if .persistent attribute is used on an automatic variable.
12997 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12999 PR rtl-optimization/80474
13000 * reorg.c (update_block): Do not ignore instructions in a delay slot.
13002 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
13004 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
13007 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
13009 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
13010 (casesi): Emit bounds checking as RTL.
13011 (casesi_internal_mips16_<mode>): Remove bounds checking.
13013 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
13015 * config/xtensa/xtensa.c (xtensa_option_override): Append
13016 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
13017 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
13018 xtensa_doloop_hooks): Define unconditionally.
13019 (xtensa_reorg_loops): Only call reorg_loops in the presence of
13021 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
13022 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
13023 for it in xtensa_option_override.
13024 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
13025 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
13027 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
13029 * doc/cppopts.texi: Document '-' special value to -MF.
13031 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
13033 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
13034 (cortex_a53_fconst): Likewise.
13035 (cortex_a53_fpmul): Likewise.
13036 (cortex_a53_f_load_64): Likewise.
13037 (cortex_a53_f_load_many): Likewise.
13038 (cortex_a53_advsimd_alu): Likewise.
13039 (cortex_a53_advsimd_alu_q): Likewise.
13040 (cortex_a53_advsimd_mul): Likewise.
13041 (cortex_a53_advsimd_mul_q): Likewise.
13042 (fpmac bypass): Add new bypass for fpmac-fpmac case.
13043 Add missing fmul, r2f_cvt and fconst cases.
13045 2017-06-14 Richard Biener <rguenther@suse.de>
13047 PR middle-end/81088
13048 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
13050 (fold_binary_loc): When associating do not treat pre-existing
13051 TREE_OVERFLOW on literal constants as a reason to allow
13052 TREE_OVERFLOW on associated literal constants.
13054 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13056 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
13057 (MASK_FEATURES): New macro.
13058 * config/sparc/sparc.c (sparc_option_override): Remove the special
13059 handling of -mfpu and generalize it to all MASK_FEATURES switches.
13061 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13063 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
13064 a division of 0 if non-call exceptions are enabled.
13066 2017-06-14 Andrew Pinski <apinski@cavium.com>
13067 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13070 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
13071 Improve vector initialization code gen for only variable case.
13073 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13075 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
13077 2017-06-14 Richard Biener <rguenther@suse.de>
13079 PR tree-optimization/81083
13080 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
13083 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13085 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
13086 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
13087 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
13088 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
13089 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
13090 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
13092 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13094 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
13095 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
13097 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13099 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
13101 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13103 * config/rs6000/t-rtems: Don't handle SPE.
13105 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13107 * config/rs6000/t-linux: Don't handle SPE.
13109 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13111 * config/rs6000/eabispe.h: Delete file.
13113 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13115 * config/rs6000/t-spe: Delete file.
13117 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13119 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
13120 (rs6000_legitimate_offset_address_p): Return false for anything in
13121 V2SImode or V2SFmode.
13123 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13125 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
13126 except V2SF and V2SI. Rearrange the vector modes, and add comments.
13127 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
13129 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
13130 (rs6000_legitimate_offset_address_p): Ditto.
13131 (rs6000_emit_move): Ditto.
13132 (rs6000_init_builtins): Remove V4HI_type_node.
13134 2017-06-13 Martin Liska <mliska@suse.cz>
13137 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
13138 (gate_asan): Likewise.
13139 * asan.h (asan_no_sanitize_address_p): Remove the function.
13140 (sanitize_flags_p): New function.
13141 * builtins.def: Fix coding style.
13142 * common.opt: Use renamed enum value.
13143 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
13144 * doc/extend.texi: Document no_sanitize attribute.
13145 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
13146 to SANITIZE_UNDEFINED_NONDEFAULT.
13147 * gcc.c (sanitize_spec_function): Use the renamed enum value.
13148 * gimple-fold.c (optimize_atomic_compare_exchange_p):
13149 Use sanitize_flags_p.
13150 * gimplify.c (gimplify_function_tree): Likewise.
13151 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
13152 * opts.c (parse_no_sanitize_attribute): New function.
13153 (common_handle_option): Use renamed enum value.
13154 * opts.h (parse_no_sanitize_attribute): Declare.
13155 * tree.c (sanitize_flags_p): New function.
13156 * tree.h: Declared here.
13157 * tsan.c: Use sanitize_flags_p.
13158 * ubsan.c (ubsan_expand_null_ifn): Likewise.
13159 (instrument_mem_ref): Likewise.
13160 (instrument_bool_enum_load): Likewise.
13161 (do_ubsan_in_current_function): Remove the function.
13162 (pass_ubsan::execute): Use sanitize_flags_p.
13163 * ubsan.h: Remove do_ubsan_in_current_function
13164 * tree-cfg.c (print_no_sanitize_attr_value): New function.
13165 (dump_function_to_file): Use it here.
13167 2017-06-13 Martin Jambor <mjambor@suse.cz>
13169 PR tree-optimization/80803
13170 PR tree-optimization/81063
13171 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
13172 (propagate_subaccesses_across_link): Enqueue subtree whenever
13173 necessary instead of relying on the caller.
13175 2017-06-13 Martin Jambor <mjambor@suse.cz>
13177 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
13178 that have a first_link.
13179 (sort_and_splice_var_accesses): Do not check first_link before
13181 (subtree_mark_written_and_enqueue): Likewise.
13182 (propagate_all_subaccesses): Likewise and do not stop at first
13183 parent with a first_link.
13185 2017-06-13 Martin Jambor <mjambor@suse.cz>
13187 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
13190 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13192 * match.pd: New pattern.
13194 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13196 * tree-vrp.c (is_masked_range_test): New function.
13197 (register_edge_assert_for): Determine ranges for
13200 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13202 PR tree-optimization/67328
13203 * fold-const.c (maskable_range_p): New function.
13204 (build_range_check): Generate bittests if possible.
13206 2017-06-13 Martin Liska <mliska@suse.cz>
13208 * gimple-pretty-print.c (dump_probability): Add new argument.
13209 (dump_edge_probability): Dump both probability and count.
13210 (dump_gimple_label): Likewise.
13211 (dump_gimple_bb_header): Likewise.
13213 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
13216 * config/avr/avr-devices.c: Fix indentation.
13217 * config/avr/gen-avr-mmcu-specs.c: Dito.
13219 2017-06-13 Richard Biener <rguenther@suse.de>
13221 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
13222 instead get vector type from stmt_info.
13223 (vectorizable_reduction): Adjust. Remove dead code.
13225 2017-06-13 Richard Biener <rguenther@suse.de>
13227 PR middle-end/81065
13228 * fold-const.c (extract_muldiv_1): Remove bogus distribution
13229 case of C * (x * C2 + C3).
13230 (fold_addr_of_array_ref_difference): Properly fold index difference.
13232 2017-06-12 David S. Miller <davem@davemloft.net>
13235 * config/sparc/sparc.md (return expander): Emit frame blockage if
13236 function uses alloca.
13238 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13240 * combine.c (make_field_assignment): Check len rather than the mode
13241 precision when calling force_to_mode.
13243 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
13245 Support multilibs and devices that see flash in RAM address range.
13248 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
13249 (avr_mcu_t) <flash_pm_offset>: New field.
13250 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
13251 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
13252 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
13253 (AVR_TINY_PM_OFFSET): Remove macro.
13254 * config/avr/avr.opt (-mshort-calls): New option.
13255 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13256 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
13257 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
13258 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
13259 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
13260 instead of avr_arch->have_jmp_call.
13261 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
13262 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
13263 avr_arch->flash_pm_offset to define.
13264 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
13265 new field flash_pm_offset. Add entry for avrxmega3.
13266 (avr_texinfo): Add entry for avrxmega3.
13267 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
13268 attiny212, attiny214,
13269 attiny412, attiny414, attiny416, attiny417,
13270 attiny814, attiny816, attiny817,
13271 attiny1614, attiny1616, attiny1617,
13272 attiny3214, attiny3216, attiny3217.
13273 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
13274 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
13275 (avr_print_operand_address) [AVR_TINY]: Same.
13276 (avr_asm_init_sections) <readonly_data_section>: Only patch
13277 callback if avr_arch->flash_pm_offset = 0.
13278 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
13279 for rodata if avr_arch->flash_pm_offset != 0.
13280 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
13281 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
13282 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
13283 (m_options): Append opt_rcall.
13284 (m_dirnames): Append dir_rcall.
13285 * config/avr/t-multilib: Regenerate.
13287 * configure.ac [target=avr]: Check whether avrxmega3 default
13288 linker description file works as needed.
13289 * configure: Regenerate.
13290 * doc/avr-mmcu.texi: Regenerate.
13291 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
13292 <__AVR_ARCH__>: Document avrxmega3 and 103.
13293 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
13294 <__AVR_SHORT_CALLS__>: Document it.
13295 <__AVR_PM_BASE_ADDRESS__>: Document it.
13296 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
13297 (AVR Variable Attributes) <progmem>: Document this is
13298 not needed for avrxmega3.
13299 (AVR Named Address Spaces) <__flash>: Dito.
13301 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
13303 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
13305 2017-06-12 Doug Rupp <rupp@adacore.com>
13307 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
13308 Append vxworks-stdint.h to the tm_file list.
13309 * config/vxworks-stdint.h: New file.
13311 2017-06-12 Martin Liska <mliska@suse.cz>
13313 PR tree-optimization/81041
13314 * tree-profile.c (gimple_gen_ic_func_profiler):
13315 Create an extra BB in profile-generate
13316 (gimple_gen_time_profiler): Likewise.
13318 2017-06-12 Jakub Jelinek <jakub@redhat.com>
13320 PR tree-optimization/81003
13321 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
13322 (update_range_test): Use it instead of force_gimple_operand_gsi.
13324 2017-06-12 Richard Biener <rguenther@suse.de>
13326 PR tree-optimization/81053
13327 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
13328 with backedge value not defined in loop. Simplify def stmt
13331 2017-06-11 Tom de Vries <tom@codesourcery.com>
13334 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
13336 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
13337 nvptx_cannot_force_const_mem.
13339 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13341 * opts.c (finish_options): Move test for flag_split_stack after
13342 it has been initialized.
13344 2017-06-11 Jason Merrill <jason@redhat.com>
13346 * tree.h (id_equal): New.
13347 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
13348 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
13349 instead of strcmp of IDENTIFIER_POINTER.
13351 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13353 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
13354 (mark_all_inlined_calls_cdtor): Fix formating.
13355 (inline_transform): Rescale profile before inlining.
13357 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13359 * cgraph.h (cgraph_edge::clone): Update prototype.
13360 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
13361 (cgraph_node::create_clone): Update.
13362 (cgraph_node::create_version_clone): Update.
13363 * tree-inline.c (copy_bb): Update.
13364 (expand_call_inline): Update.
13366 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13368 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
13369 factored out from ...
13370 (rs6000_emit_prologue): ... here.
13372 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13374 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
13375 factored out from ...
13376 (rs6000_emit_prologue): ... here.
13378 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13380 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
13382 (handle_missing_profiles): Fix computation of tp_first_run.
13383 (counts_to_freqs): Do not touch freqs when count is 0.
13385 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13387 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
13390 2017-06-10 Tom de Vries <tom@codesourcery.com>
13392 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13393 attributes): Document signal effective target.
13395 2017-06-10 Tom de Vries <tom@codesourcery.com>
13397 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13398 Document effective target stack_size.
13400 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13402 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13403 to the edit_context if they can be auto-applied.
13405 2017-06-9 Ian Lance Taylor <iant@golang.org>
13407 * opts.c (finish_options): If -fsplit-stack, disable implicit
13408 -forder-blocks-and-partition.
13409 * doc/invoke.texi (Optimize Options): Document that when using
13410 -fsplit-stack -forder-blocks-and-partition is not implicitly
13413 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13415 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13416 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13417 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13418 * builtins.def (abort, trap, unreachable): Declare cold.
13419 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13420 * tree-core.h (ECF_COLD): New.
13421 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13422 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13424 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13426 * predict.c (unlikely_executed_stmt_p): Cleanup.
13428 2017-06-09 Richard Biener <rguenther@suse.de>
13430 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13431 model if the ref is always written to.
13433 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13435 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13437 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13439 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13442 2017-06-09 Tom de Vries <tom@codesourcery.com>
13445 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13446 "target cannot support label values" when encountering LABEL_REF.
13448 2017-06-09 Martin Liska <mliska@suse.cz>
13450 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13451 (gimple_gen_ic_func_profiler): Emit direct comparison
13452 of __gcov_indirect_call_callee with NULL.
13453 (gimple_gen_time_profiler): Change probability from
13454 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13456 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13458 * profile.c (edge_gcov_counts): Turn to pointer.
13459 (compute_branch_probabilities, compute_branch_probabilities): Update.
13460 (branch_prob): Do not clear edge_gcov_count.
13461 * profile.h (edge_gcov_counts): Turn to pointer.
13462 (edge_gcov_count): Update.
13464 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13466 * gimple.h (gimple_check_failed): Mark cold.
13468 2017-06-09 Richard Biener <rguenther@suse.de>
13470 PR tree-optimization/66623
13471 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13472 refactor check_reduction into two parts, properly computing
13473 whether we have to check reduction validity for outer loop
13476 2017-06-09 Richard Biener <rguenther@suse.de>
13478 PR tree-optimization/79483
13479 * graphite-scop-detection.c (order): New global.
13480 (get_order): Compute bb to order mapping that satisfies code
13481 generation constraints.
13482 (cmp_pbbs): New helper.
13483 (build_scops): Start domwalk at entry block, sort generated
13486 2017-06-09 Richard Biener <rguenther@suse.de>
13488 PR middle-end/81007
13489 * ipa-polymorphic-call.c
13490 (ipa_polymorphic_call_context::restrict_to_inner_class):
13491 Skip FIELD_DECLs with error_mark_node type.
13492 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13495 2017-06-09 Martin Liska <mliska@suse.cz>
13497 * predict.c (struct branch_predictor): New struct.
13498 (test_prediction_value_range): New test.
13499 (predict_c_tests): New function.
13500 * selftest-run-tests.c (selftest::run_tests): Run the function.
13501 * selftest.h: Declare new tests.
13503 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13506 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13507 gen_add3_insn did not fail.
13508 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13509 r0, construct that number in a temporary reg and add that reg to r0.
13510 If asked to put the result in r0 as well, fail.
13512 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13514 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13515 for early expansion of vec_eqv.
13517 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13519 PR middle-end/81005
13520 * ubsan.c (instrument_null): Avoid pointless code temporary.
13521 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13523 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13527 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13530 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13533 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13535 * predict.c (unlikely_executed_edge_p): Move ahead.
13536 (probably_never_executed_edge_p): Use it.
13538 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13540 PR middle-end/79988
13541 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13542 gimple_call_builtin_p call.
13544 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13546 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13547 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13548 rtl_check_failed_type2, rtl_check_failed_code1,
13549 rtl_check_failed_code2, rtl_check_failed_code_mode,
13550 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13551 rtvec_check_failed_bounds, rtl_check_failed_flag,
13552 _fatal_insn_not_found, _fatal_insn): Likewise.
13553 * tree.h (tree_contains_struct_check_failed,
13554 tree_check_failed, tree_not_check_failed,
13555 tree_class_check_failed, tree_range_check_failed,
13556 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13557 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13558 tree_operand_check_failed, omp_clause_check_failed,
13559 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13562 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13564 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13565 flag_branch_probabilities.
13566 * ipa-inline.c (edge_badness): Likewise.
13567 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13568 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13569 * predict.c (maybe_hot_frequency_p): Likewise.
13570 (probably_never_executed): Likewise.
13571 * sched-ebb.c (schedule_ebbs): Likewise.
13572 * sched-rgn.c (find_single_block_region): Likewise.
13573 * tracer.c (tail_duplicate): Likewise.
13575 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13577 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13578 longer requires x_flag_profile_use.
13580 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13582 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13583 instead of flag_reorder_blocks_and_partition.
13584 * dbxout.c (dbxout_function_end): Likewise.
13585 * dwarf2out.c (gen_subprogram_die): Likewise.
13586 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13587 * hw-doloop.c (reorg_loops): Likewise.
13588 * varasm.c (assemble_start_function,
13589 assemble_end_function): Likewise.
13590 (decide_function_section): Do not check for
13591 flag_reorder_blocks_and_partition.
13593 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13595 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13597 (chkp_get_hard_register_fake_addr_expr): Ditto.
13598 (chkp_build_addr_expr): Add check for hard reg case.
13599 (chkp_parse_array_and_component_ref): Ditto.
13600 (chkp_find_bounds_1): Ditto.
13601 (chkp_process_stmt): Don't generate bounds store for
13604 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13606 * predict.c (maybe_hot_bb_p): Do not check profile status.
13607 (maybe_hot_edge_p): Likewise.
13608 (probably_never_executed): Check for zero counts even if profile
13610 (unlikely_executed_edge_p): New function.
13611 (unlikely_executed_stmt_p): New function.
13612 (unlikely_executed_bb_p): New function.
13613 (set_even_probabilities): Use unlikely predicates.
13614 (combine_predictions_for_bb): Likewise.
13615 (predict_paths_for_bb): Likewise.
13616 (predict_paths_leading_to_edge): Likewise.
13617 (determine_unlikely_bbs): New function.
13618 (estimate_bb_frequencies): Use it.
13619 (compute_function_frequency): Use zero counts even if profile is
13621 * profile-count.h: Fix typo.
13623 2017-08-08 Julia Koval <julia.koval@intel.com>
13625 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13626 _mm512_mask_cvtsepi16_storeu_epi8,
13627 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13628 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13629 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13630 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13631 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13632 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13633 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13634 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13635 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13636 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13637 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13638 __builtin_ia32_pmovuswb256mem_mask,
13639 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13640 __builtin_ia32_pmovwb512mem_mask): New builtins.
13642 2017-08-08 Julia Koval <julia.koval@intel.com>
13644 PR target/73350,80862
13645 * config/i386/subst.md (round): Fix round pattern.
13646 * config/i386/i386.c (ix86_erase_embedded_rounding):
13647 Fix erasing rounding for the fixed pattern.
13649 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13651 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
13653 2017-06-08 Martin Liska <mliska@suse.cz>
13655 PR gcov-profile/80911
13656 * gcov.c (block_info::block_info): New constructor.
13658 2017-06-07 Carl Love <cel@us.ibm.com>
13660 * config/rs6000/rs6000-c: The return type of the following
13661 built-in functions was implemented as int not long long. Fix sign
13662 of return value for the unsigned version of vec_mulo and vec_mule.
13663 vector unsigned long long vec_bperm (vector unsigned long long,
13664 vector unsigned char)
13665 vector signed long long vec_mule (vector signed int,
13667 vector unsigned long long vec_mule (vector unsigned int,
13668 vector unsigned int)
13669 vector signed long long vec_mulo (vector signed int,
13671 vector unsigned long long vec_mulo (vector unsigned int,
13672 vector unsigned int)
13673 * doc/extend.texi: Fix the documentation for the built-in
13676 2017-06-07 Carl Love <cel@us.ibm.com>
13679 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13682 2017-06-07 Carl Love <cel@us.ibm.com>
13684 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13685 support, Generate doublehv for signed int/float for BE case only.
13687 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13689 * doc/invoke.texi (mcx16): Rewrite.
13691 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13693 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13694 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13695 *mov<mode>_softfloat, and an anonymous splitter): Use
13696 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13698 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13700 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13702 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13703 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13704 (rs6000_debug_reg_global): Adjust.
13705 (rs6000_init_hard_regno_mode_ok): Adjust.
13706 (rs6000_dbx_register_number): Adjust.
13707 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13708 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13709 Remove SPE_ACC and SPEFSCR.
13710 (REG_ALLOC_ORDER): Ditto.
13711 (FRAME_POINTER_REGNUM): Change to 111.
13712 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13713 (REG_CLASS_NAMES): Ditto.
13714 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13715 (REGISTER_NAMES): Ditto.
13716 (ADDITIONAL_REG_NAMES): Ditto.
13717 (rs6000_reg_names): Ditto.
13718 * config/rs6000/rs6000.md: Renumber some register number
13721 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13723 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13725 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13726 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13728 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13729 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13730 Delete the SPE high registers.
13731 (REG_ALLOC_ORDER): Ditto.
13732 (enum reg_class): Remove SPE_HIGH_REGS.
13733 (REG_CLASS_NAMES): Ditto.
13734 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13735 (REGISTER_NAMES): Ditto.
13736 (rs6000_reg_names): Ditto.
13737 * doc/tm.texi.in: Remove SPE as example.
13738 * doc/tm.texi: Regenerate.
13740 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13742 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13743 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13744 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
13745 * config/rs6000/rs6000.md (type): Remove "brinc".
13747 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13749 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
13750 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
13751 * config/rs6000/linuxspe.h: Delete file.
13752 * config/rs6000/rs6000.md: Don't include spe.md.
13753 * config/rs6000/spe.h: Delete file.
13754 * config/rs6000/spe.md: Delete file.
13755 * config/rs6000/t-rs6000: Remove spe.md.
13757 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13759 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
13760 (reg_or_none500mem_operand): Delete.
13761 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
13762 instead of reg_or_none500mem_operand.
13764 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13766 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13767 handling of SPE flags.
13768 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
13770 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13772 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13774 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
13775 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
13776 paired_divv2sf3): Similar.
13777 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
13778 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
13779 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
13781 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
13782 Rename the paired_* instruction patterns.
13783 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
13785 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
13786 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
13787 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
13788 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
13789 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
13790 PAIRED_VECTOR_MODE.
13791 (struct machine_function): Delete field spe_insn_chain_scanned_p.
13792 (spe_func_has_64bit_regs_p): Delete.
13793 (spe_expand_predicate_builtin): Delete.
13794 (spe_expand_evsel_builtin): Delete.
13795 (TARGET_DWARF_REGISTER_SPAN): Do not define.
13796 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
13797 (invalid_e500_subreg): Delete.
13798 (rs6000_legitimize_address): Always force_reg op2 as well, for
13799 paired single memory accesses.
13800 (rs6000_member_type_forces_blk): Delete.
13801 (rs6000_spe_function_arg): Delete.
13802 (rs6000_expand_unop_builtin): Delete SPE handling.
13803 (rs6000_expand_binop_builtin): Ditto.
13804 (spe_expand_stv_builtin): Delete.
13805 (bdesc_2arg_spe): Delete.
13806 (spe_expand_builtin): Delete.
13807 (spe_expand_predicate_builtin): Delete.
13808 (spe_expand_evsel_builtin): Delete.
13809 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
13810 (spe_init_builtins): Delete.
13811 (spe_func_has_64bit_regs_p): Delete.
13812 (savres_routine_name): Delete "info" parameter. Adjust callers.
13813 (rs6000_emit_stack_reset): Ditto.
13814 (rs6000_dwarf_register_span): Delete.
13815 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
13816 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
13817 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
13819 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
13821 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
13822 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
13823 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
13824 mulv2sf3, divv2sf3): Delete expanders.
13826 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13828 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
13830 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13832 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
13833 * config/rs6000/rs6000.c: Ditto.
13835 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13837 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
13838 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
13839 comparison_operator.
13841 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13843 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
13844 * config/rs6000/rs6000.opt: Ditto.
13845 * config/rs6000/t-rtems: Ditto.
13847 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13849 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
13850 TARGET_E500_SINGLE by 0, simplify.
13851 * config/rs6000/rs6000.c: Ditto.
13852 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
13853 (spe_build_register_parallel): Delete.
13854 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
13855 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
13856 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
13857 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
13858 (E500_CONVERT): Delete.
13859 * config/rs6000/spe.md: Remove many patterns and all define_constants.
13861 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13863 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
13864 * config/rs6000/dfp.md: Ditto.
13865 (negdd2, *negdd2_fpr): Merge.
13866 (absdd2, *absdd2_fpr): Merge.
13867 (negtd2, *negtd2_fpr): Merge.
13868 (abstd2, *abstd2_fpr): Merge.
13869 * config/rs6000/e500.h: Delete file.
13870 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
13871 TARGET_FPRS by 1 and simplify.
13872 * config/rs6000/rs6000-c.c: Ditto.
13873 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
13874 TARGET_DF_SPE by 0.
13875 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
13877 * config/rs6000/rs6000.md: Ditto.
13878 (floatdidf2, *floatdidf2_fpr): Merge.
13879 (move_from_CR_gt_bit): Delete.
13880 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
13881 (E500_CR_IOR_COMPARE): Delete.
13882 (All patterns that require !TARGET_FPRS): Delete.
13883 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
13885 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13887 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
13889 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13891 * graphds.c (add_edge): Intitialize edge's attached data.
13892 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
13893 pointer parameter. Call pointed function on each edge during
13894 graph traversing. Skip traversing the edge when the function
13896 (graphds_dfs, graphds_scc): Ditto.
13897 (for_each_edge): New parameter. Pass the new parameter to callback
13899 * graphds.h (skip_edge_callback): New function pointer type.
13900 (graphds_dfs, graphds_scc): New function pointer parameter.
13901 (graphds_edge_callback, for_each_edge): New parameter.
13903 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13905 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
13906 out code checking if runtime alias check is possible to below ...
13907 Call the new function.
13908 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
13909 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
13911 2017-06-07 Marek Polacek <polacek@redhat.com>
13914 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
13915 TYPE_OVERFLOW_WRAPS check.
13917 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13919 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
13920 if versioning is required.
13921 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
13922 peeling with the check for versioning.
13924 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13926 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
13927 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
13928 Set true to new parameter if new ssa variable is defined.
13929 (vect_gen_vector_loop_niters): Refactor. Set range information
13930 for the new vector loop bound variable.
13931 (vect_do_peeling): Ditto.
13933 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13935 * tree-affine.c (ssa.h): Include header file.
13936 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
13937 has wrapping overflow behavior.
13939 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13941 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
13943 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13945 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
13946 (tree_to_aff_combination): ... here.
13948 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13950 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
13951 reg_pressure model function.
13952 (ivopts_global_cost_for_size): Delete.
13953 (determine_set_costs, iv_ca_recount_cost): Call new model function
13954 ivopts_estimate_reg_pressure.
13956 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13958 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
13959 expensive than udiv. Remove floating point cases from mod.
13961 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13963 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
13964 Increase idiv cost.
13966 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13968 * config/aarch64/aarch64.md
13969 (copysignsf3): Fix mask generation.
13971 2017-06-07 Jakub Jelinek <jakub@redhat.com>
13973 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
13975 (class dump_manager): Add register_dumps method.
13976 * dumpfile.c: Include langhooks.h.
13977 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
13978 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
13979 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
13980 (dump_manager::dump_register): Start with 512 entries instead of 32.
13981 (dump_manager::register_dumps): New method.
13982 * toplev.c (general_init): Instead of invoking register_dumps
13983 langhook, invoke register_dumps method on the dump manager.
13984 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13987 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13989 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13990 Say that # requires an associated define_split to exist, and that
13991 the define_split must be suitable for use after register allocation.
13993 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13995 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13996 (compute_outgoing_frequencies): Also initialize zero counts.
13997 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13998 around loops; preserve more of profile when nothing changes.
14000 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
14002 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
14004 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
14005 * config/arm/arm-cpu-cdata.h: Regenerate.
14006 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
14007 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
14008 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
14009 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
14010 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
14012 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
14014 * config/arm/t-rmprofile: Likewise.
14015 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
14017 2017-06-06 David S. Miller <davem@davemloft.net>
14020 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
14021 blockage if function uses alloca.
14023 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14025 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
14026 New "uid" fields to hold pretty-print IDs of group and ref.
14027 Memory references are now identified as <group_id>:<ref_id>
14028 instead of using [random] addresses.
14029 (dump_mem_details): Simplify, no functional change.
14030 (dump_mem_ref): Simplify and make output more concise.
14031 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
14032 (find_or_create_group): Initialize group uid.
14033 (record_ref): Initialize ref uid. Improve debug output.
14034 (prune_group_by_reuse, should_issue_prefetch_p,)
14035 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
14036 (mark_nontemporal_store, determine_loop_nest_reuse):
14037 Improve debug output.
14039 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14041 * dbgcnt.def (prefetch): New debug counter.
14042 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
14043 (schedule_prefetches): Stop issueing prefetches if debug counter
14046 2017-06-06 Tom de Vries <tom@codesourcery.com>
14048 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
14049 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
14051 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14053 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
14054 Use aarch64_reg_or_zero predicate for operand 4.
14055 (aarch64_compare_and_swap<mode> define_insn_and_split):
14056 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
14057 (aarch64_store_exclusive<mode>): Likewise for operand 2.
14059 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14061 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
14062 (arm_compute_save_core_reg_mask): This.
14063 (thumb1_compute_save_reg_mask): Rename into ...
14064 (thumb1_compute_save_core_reg_mask): This.
14065 (arm_compute_save_reg0_reg12_mask): Adapt comment.
14066 (arm_compute_frame_layout): Likewise.
14068 2017-06-06 Richard Biener <rguenther@suse.de>
14070 PR tree-optimization/80974
14071 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
14072 keep or clear leaders SSA info.
14074 2017-06-06 Tom de Vries <tom@codesourcery.com>
14076 * config/nvptx/nvptx.c (split_mode_p): New function.
14077 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
14079 2017-06-06 Tom de Vries <tom@codesourcery.com>
14081 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
14083 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14086 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
14089 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14091 * shrink-wrap.c (handle_simple_exit): Update profile.
14092 (try_shrink_wrapping): Upate profile.
14094 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14096 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
14097 (tree_guess_outgoing_edge_probabilities): New.
14098 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
14099 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
14101 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14103 * ipa-split.c (split_function): Initialize return bb profile.
14105 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14107 * profile.c (compute_branch_probabilities): Also initialize
14108 EXIT_BLOCK profile.
14110 2017-06-06 Richard Biener <rguenther@suse.de>
14112 PR tree-optimization/80928
14113 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
14114 (vect_analyze_loop_operations): Properly guard analysis for
14116 (vect_transform_loop): Likewise.
14117 (vect_analyze_loop_2): Also reset SLP type on PHIs.
14118 (vect_model_induction_cost): Do not cost for pure SLP.
14119 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
14120 of induction in inner loop vectorization.
14121 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
14122 (vect_get_and_check_slp_defs): Handle vect_induction_def.
14123 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
14125 (vect_analyze_slp_cost_1): Cost induction.
14126 (vect_detect_hybrid_slp_stmts): Handle PHIs.
14127 (vect_get_slp_vect_defs): Likewise.
14128 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
14129 (vect_transform_stmt): Handle SLP reductions.
14130 * tree-vectorizer.h (vectorizable_induction): Adjust.
14132 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14134 * config/rs6000/rs6000.c (make_resolver_func): Update
14135 init_lowered_empty_function call.
14137 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
14139 * doc/invoke.texi: Document the -fprofile-abs-path option.
14140 * common.opt (fprofile-abs-path): New option.
14141 * gcov-io.h (gcov_write_filename): Declare.
14142 * gcov-io.c (gcov_write_filename): New function.
14143 * coverage.c (coverage_begin_function): Use gcov_write_filename.
14144 * profile.c (output_location): Likewise.
14146 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14148 * shring-wrap.c: Revert accidental commit.
14150 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
14152 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
14154 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14156 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
14158 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
14159 profile in callgraph edge.
14160 * profile-count.h (apply_probability): If THIS is 0, then result is 0
14161 (apply_scale): Likewise.
14162 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
14163 Also scale profile when inlining function with zero profile.
14164 (initialize_cfun): Update exit block profile even when it is zero.
14165 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
14166 when profile is read.
14168 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14170 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
14171 (CLONE_*): New constants to define the processors we can generate
14172 code for with the target_clone attribute.
14173 (rs6000_clone_map): New array to identify which clone processors
14174 the current program is running on.
14175 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
14176 target_clone attribute.
14177 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
14178 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
14179 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
14180 (cpu_expand_builtin): Add support for target_clone attribute.
14181 (rs6000_valid_attribute_p): Allow "default" attribute.
14182 (get_decl_name): New debug function to simplify printing the
14183 current function name in debugging statements.
14184 (rs6000_clone_priority): New functions to support the target_clone
14185 attribute, and be able to generate code to switch between ISA 2.05
14186 through ISA 3.0 (power6 through power9).
14187 (rs6000_compare_version_priority): Likewise.
14188 (rs6000_get_function_versions_dispatcher): Likewise.
14189 (make_resolver_func): Likewise.
14190 (add_condition_to_bb): Likewise.
14191 (dispatch_function_versions): Likewise.
14192 (rs6000_generate_version_dispatcher_body): Likewise.
14193 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
14194 (fusion_gpr_load_p): Fix a spacing issue.
14195 * doc/extend.texi (Common Function Attributes): Document that the
14196 PowerPC supports the target_clone attribute.
14198 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14200 * config/arm/arm.h: explain F symbol found in description of ARM
14201 register allocation in its legend.
14203 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14205 * config/mips/frame-header-opt.c: Include profile-count.h.
14206 * config/riscv/riscv.c: Include profile-count.h
14208 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14210 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
14212 (sm_set_flag_if_changed): Add bbs field.
14213 (execute_sm_if_changed_flag_set): Pass BBS.
14214 (execute_sm): Update.
14216 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14218 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
14221 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14223 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
14224 (peephole2): New peephole2 to emit the above.
14225 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
14227 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14229 * config/aarch64/aarch64.c (define_peephole2 above
14230 *sub_<shift>_<mode>): New peephole.
14232 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14234 * config/i386/i386.c (make_resolver_func): Update.
14235 * Makefile.in: Add profile-count.h and profile-count.o
14236 * auto-profile.c (afdo_indirect_call): Update to new API.
14237 (afdo_set_bb_count): Update.
14238 (afdo_propagate_edge): Update.
14239 (afdo_propagate_circuit): Update.
14240 (afdo_calculate_branch_prob): Update.
14241 (afdo_annotate_cfg): Update.
14242 * basic-block.h: Include profile-count.h
14243 (struct edge_def): Turn count to profile_count.
14244 (struct basic_block_def): Likewie.
14245 (REG_BR_PROB_BASE): Move to profile-count.h
14246 (RDIV): Move to profile-count.h
14247 * bb-reorder.c (max_entry_count): Turn to profile_count.
14248 (find_traces): Update.
14249 (rotate_loop):Update.
14250 (connect_traces):Update.
14251 (sanitize_hot_paths):Update.
14252 * bt-load.c (migrate_btr_defs): Update.
14253 * cfg.c (RDIV): Remove.
14254 (init_flow): Use alloc_block.
14255 (alloc_block): Uninitialize count.
14256 (unchecked_make_edge): Uninitialize count.
14257 (check_bb_profile): Update.
14258 (dump_edge_info): Update.
14259 (dump_bb_info): Update.
14260 (update_bb_profile_for_threading): Update.
14261 (scale_bbs_frequencies_int): Update.
14262 (scale_bbs_frequencies_gcov_type): Update.
14263 (scale_bbs_frequencies_profile_count): New.
14264 * cfg.h (update_bb_profile_for_threading): Update.
14265 (scale_bbs_frequencies_profile_count): Declare.
14266 * cfgbuild.c (compute_outgoing_frequencies): Update.
14267 (find_many_sub_basic_blocks): Update.
14268 * cfgcleanup.c (try_forward_edges): Update.
14269 (try_crossjump_to_edge): Update.
14270 * cfgexpand.c (expand_gimple_tailcall): Update.
14271 (construct_exit_block): Update.
14272 * cfghooks.c (verify_flow_info): Update.
14273 (dump_bb_for_graph): Update.
14274 (split_edge): Update.
14275 (make_forwarder_block): Update.
14276 (duplicate_block): Update.
14277 (account_profile_record): Update.
14278 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
14279 (get_estimated_loop_iterations): Update.
14280 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
14281 (single_likely_exit): Update.
14282 * cfgloopmanip.c (scale_loop_profile): Update.
14284 (set_zero_probability): Update.
14285 (lv_adjust_loop_entry_edge): Update.
14286 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14287 (purge_dead_edges): Update.
14288 (rtl_account_profile_record): Update.
14289 * cgraph.c (cgraph_node::create): Uninitialize count.
14290 (symbol_table::create_edge): Uninitialize count.
14291 (cgraph_update_edges_for_call_stmt_node): Update.
14292 (cgraph_edge::dump_edge_flags): Update.
14293 (cgraph_node::dump): Update.
14294 (cgraph_edge::maybe_hot_p): Update.
14295 * cgraph.h: Include profile-count.h
14296 (create_clone), create_edge, create_indirect_edge): Update.
14297 (cgraph_node): Turn count to profile_count.
14298 (cgraph_edge0: Likewise.
14299 (make_speculative, clone): Update.
14300 (create_edge): Update.
14301 (init_lowered_empty_function): Update.
14302 * cgraphclones.c (cgraph_edge::clone): Update.
14303 (duplicate_thunk_for_node): Update.
14304 (cgraph_node::create_clone): Update.
14305 * cgraphunit.c (cgraph_node::analyze): Update.
14306 (cgraph_node::expand_thunk): Update.
14307 * final.c (dump_basic_block_info): Update.
14308 * gimple-streamer-in.c (input_bb): Update.
14309 * gimple-streamer-out.c (output_bb): Update.
14310 * graphite.c (print_global_statistics): Update.
14311 (print_graphite_scop_statistics): Update.
14312 * hsa-brig.c: Include basic-block.h.
14313 * hsa-dump.c: Include basic-block.h.
14314 * hsa-gen.c (T sum_slice): Update.
14315 (convert_switch_statements):Update.
14316 * hsa-regalloc.c: Include basic-block.h.
14317 * ipa-chkp.c (chkp_produce_thunks): Update.
14318 * ipa-cp.c (struct caller_statistics): Update.
14319 (init_caller_stats): Update.
14320 (gather_caller_stats): Update.
14321 (ipcp_cloning_candidate_p): Update.
14322 (good_cloning_opportunity_p): Update.
14323 (get_info_about_necessary_edges): Update.
14324 (dump_profile_updates): Update.
14325 (update_profiling_info): Update.
14326 (update_specialized_profile): Update.
14327 (perhaps_add_new_callers): Update.
14328 (decide_about_value): Update.
14329 (ipa_cp_c_finalize): Update.
14330 * ipa-devirt.c (struct odr_type_warn_count): Update.
14331 (struct decl_warn_count): Update.
14332 (struct final_warning_record): Update.
14333 (possible_polymorphic_call_targets): Update.
14334 (ipa_devirt): Update.
14335 * ipa-fnsummary.c (redirect_to_unreachable): Update.
14336 * ipa-icf.c (sem_function::merge): Update.
14337 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
14338 * ipa-inline.c (compute_uninlined_call_time): Update.
14339 (compute_inlined_call_time): Update.
14340 (want_inline_small_function_p): Update.
14341 (want_inline_self_recursive_call_p): Update.
14342 (edge_badness): Update.
14343 (lookup_recursive_calls): Update.
14344 (recursive_inlining): Update.
14345 (inline_small_functions): Update.
14346 (dump_overall_stats): Update.
14347 (dump_inline_stats): Update.
14348 * ipa-profile.c (ipa_profile_generate_summary): Update.
14349 (ipa_propagate_frequency): Update.
14350 (ipa_profile): Update.
14351 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14352 * ipa-utils.c (ipa_merge_profiles): Update.
14353 * loop-doloop.c (doloop_modify): Update.
14354 * loop-unroll.c (report_unroll): Update.
14355 (unroll_loop_runtime_iterations): Update.
14356 * lto-cgraph.c (lto_output_edge): Update.
14357 (lto_output_node): Update.
14358 (input_node): Update.
14359 (input_edge): Update.
14360 (merge_profile_summaries): Update.
14361 * lto-streamer-in.c (input_cfg): Update.
14362 * lto-streamer-out.c (output_cfg): Update.
14363 * mcf.c (create_fixup_graph): Update.
14364 (adjust_cfg_counts): Update.
14365 (sum_edge_counts): Update.
14366 * modulo-sched.c (sms_schedule): Update.
14367 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
14368 * predict.c (maybe_hot_count_p): Update.
14369 (probably_never_executed): Update.
14370 (dump_prediction): Update.
14371 (combine_predictions_for_bb): Update.
14372 (propagate_freq): Update.
14373 (handle_missing_profiles): Update.
14374 (counts_to_freqs): Update.
14375 (rebuild_frequencies): Update.
14376 (force_edge_cold): Update.
14377 * predict.h: Include profile-count.h
14378 (maybe_hot_count_p, counts_to_freqs): UPdate.
14379 * print-rtl-function.c: Do not include cfg.h
14380 * print-rtl.c: Include basic-block.h
14381 * profile-count.c: New file.
14382 * profile-count.h: New file.
14383 * profile.c (is_edge_inconsistent): Update.
14384 (correct_negative_edge_counts): Update.
14385 (is_inconsistent): Update.
14386 (set_bb_counts): Update.
14387 (read_profile_edge_counts): Update.
14388 (compute_frequency_overlap): Update.
14389 (compute_branch_probabilities): Update; Initialize and deinitialize
14391 (branch_prob): Update.
14392 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
14393 (edge_gcov_count): New.
14394 (bb_gcov_count): New.
14395 * shrink-wrap.c (try_shrink_wrapping): Update.
14396 * tracer.c (better_p): Update.
14397 * trans-mem.c (expand_transaction): Update.
14398 (ipa_tm_insert_irr_call): Update.
14399 (ipa_tm_insert_gettmclone_call): Update.
14400 * tree-call-cdce.c: Update.
14401 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14402 (gimple_duplicate_sese_tail): Update.
14403 (gimple_account_profile_record): Update.
14404 (execute_fixup_cfg): Update.
14405 * tree-inline.c (copy_bb): Update.
14406 (copy_edges_for_bb): Update.
14407 (initialize_cfun): Update.
14408 (freqs_to_counts): Update.
14409 (copy_cfg_body): Update.
14410 (expand_call_inline): Update.
14411 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14412 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14413 (try_unroll_loop_completely): Update.
14414 (try_peel_loop): Update.
14415 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14416 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14417 * tree-ssa-loop-split.c (connect_loops): Update.
14418 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14419 * tree-ssa-reassoc.c (branch_fixup): Update.
14420 * tree-ssa-tail-merge.c (replace_block_by): Update.
14421 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14422 (compute_path_counts): Update.
14423 (update_profile): Update.
14424 (recompute_probabilities): Update.
14425 (update_joiner_offpath_counts): Update.
14426 (estimated_freqs_path): Update.
14427 (freqs_to_counts_path): Update.
14428 (clear_counts_path): Update.
14429 (ssa_fix_duplicate_block_edges): Update.
14430 (duplicate_thread_path): Update.
14431 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14432 (struct switch_conv_info): Update.
14433 * tree-tailcall.c (decrease_profile): Update.
14434 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14435 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14436 * value-prof.c (check_counter): Update.
14437 (gimple_divmod_fixed_value): Update.
14438 (gimple_mod_pow2): Update.
14439 (gimple_mod_subtract): Update.
14440 (gimple_ic_transform): Update.
14441 (gimple_stringop_fixed_value): Update.
14442 * value-prof.h (gimple_ic): Update.
14444 2017-06-02 Carl Love <cel@us.ibm.com>
14446 * config/rs6000/rs6000-c: Add support for built-in functions
14447 vector double vec_doublee (vector signed int);
14448 vector double vec_doublee (vector unsigned int);
14449 vector double vec_doublee (vector float);
14450 vector double vec_doubleh (vector signed int);
14451 vector double vec_doubleh (vector unsigned int);
14452 vector double vec_doubleh (vector float);
14453 vector double vec_doublel (vector signed int);
14454 vector double vec_doublel (vector unsigned int);
14455 vector double vec_doublel (vector float);
14456 vector double vec_doubleo (vector signed int);
14457 vector double vec_doubleo (vector unsigned int);
14458 vector double vec_doubleo (vector float);.
14459 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14460 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14462 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14463 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14464 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14466 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14467 vec_doublel, vec_doubleh.
14468 * doc/extend.texi: Update the built-in documentation file for the
14469 new built-in functions.
14471 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14474 * ipa-inline-analysis.c (free_growth_caches): Set
14475 edge_removal_hook_holder to NULL after removing it.
14477 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14479 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14480 comparision with zero.
14482 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14483 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14484 for early expansion of vec_min and vec_max builtins.
14485 (builtin_function_type): Add min/max unsigned variants to those
14486 identified as having unsigned arguments.
14488 2017-06-02 Olivier Hainque <hainque@adacore.com>
14490 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14492 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14494 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14495 Use VALL_F16 iterator rather than VALL.
14497 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14499 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14500 Emit CBNZ inside loop when doing a strong exchange and comparing
14501 against zero. Generate the CC flags after the loop.
14503 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14505 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14506 (dl_section_ref): New.
14507 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14508 On AIX, append an expression to subtract the size of the
14509 section length to dl_section_ref.
14511 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14513 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14514 for early expansion of vector absolute builtins.
14516 2017-06-02 Richard Biener <rguenther@suse.de>
14518 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14519 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14521 2017-06-02 Richard Biener <rguenther@suse.de>
14523 PR tree-optimization/80948
14524 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14525 stmt order as well.
14527 2017-06-02 Richard Biener <rguenther@suse.de>
14529 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14531 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14532 for inductions as relevant.
14534 2017-06-02 Richard Biener <rguenther@suse.de>
14536 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14537 (vectorizable_induction): ... this. Remove dead code.
14539 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14541 * builtins. (expand_builtin_alloca): Remove second parameter and
14542 infer its value from the first parameter instead.
14543 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14545 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14547 PR rtl-optimization/80903
14548 * loop-doloop.c (add_test): Unshare sequence.
14550 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14552 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14554 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14556 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14558 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14559 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14560 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14561 (xlogue_layout::get_stub_rtx): Make static.
14562 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14563 (xlogue_layout::compute_stub_managed_regs): Rename to...
14564 (xlogue_layout::count_stub_managed_regs): ...this.
14565 (xlogue_layout::is_stub_managed_reg): New function.
14566 (xlogue_layout::m_stub_names): Rename to...
14567 (xlogue_layout::s_stub_names): ...this, make static.
14568 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14569 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14570 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14571 xlogue_layout::s_stub_names): Instantiate statics.
14572 (stub_managed_regs): Remove.
14573 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14574 (disable_call_ms2sysv_xlogues): Rename to...
14575 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14576 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14578 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14579 change after reload_completed.
14580 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14582 (ix86_expand_prologue): Likewise.
14583 (ix86_expand_epilogue): Likewise.
14584 (ix86_expand_split_stack_prologue): Likewise.
14585 (ix86_compute_frame_layout): Remove frame parameter ...
14586 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14587 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14589 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14590 (ix86_frame): Move from here ...
14591 * config/i386/i386.h (ix86_frame): ... to here.
14592 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14593 complete ix86_frame data structure instead. Remove some_ld_name.
14595 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14597 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14598 symbols that hold a DECL_VALUE_EXPR.
14600 2017-06-01 Martin Jambor <mjambor@suse.cz>
14602 PR tree-optimization/80898
14603 * tree-sra.c (process_subtree_disqualification): Removed.
14604 (disqualify_candidate): Do not acll
14605 process_subtree_disqualification.
14606 (subtree_mark_written_and_enqueue): New function.
14607 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14608 RHS has been disqualified and re-queue LHS if necessary. Apart
14609 from that, ignore disqualified RHS.
14611 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14613 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14614 address fetch for z10 or later.
14616 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14618 * config/arc/arc.md (tst_movb): Add guard when splitting.
14620 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14622 * config/arc/arc.c (arc_can_eliminate): Test against
14623 arc_frame_pointer_needed.
14625 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14627 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14628 to prevent store reordering.
14629 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14630 (type): Add block type.
14631 (stack_tie): Define special instruction to be used in
14634 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14636 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14637 constraint. It is not valid for the pattern.
14638 (noncommutative_binary_comparison): Likewise.
14640 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14642 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
14645 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14647 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
14648 be used by the reg-alloc.
14650 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14652 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
14653 reg-alloc when having mul64 or mul32x16 instructions.
14654 (mulsidi3): Likewise.
14655 (umulsidi3): Likewise.
14656 (mulsi32x16): New pattern.
14657 (mulsi64): Likewise.
14658 (mulsidi64): Likewise.
14659 (umulsidi64): Likewise.
14660 (MUL32x16_REG): Define.
14661 (mul64_600): Use MUL32x16_REG.
14662 (mac64_600): Likewise.
14663 (umul64_600): Likewise.
14664 (umac64_600): Likewise.
14666 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14668 * config/arc/arc.md (mulsi3_700): Make it commutative.
14670 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14672 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14674 (*sign_extendsidi2_insn): Likewise for movstosw.
14676 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14678 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14679 the type of the input discriminant value. Convert the
14680 discriminant value of signedness vary.
14682 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14684 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14685 Add to -Wall section.
14687 2017-06-01 Richard Biener <rguenther@suse.de>
14689 PR middle-end/66313
14690 * fold-const.c (fold_plusminus_mult_expr): If the factored
14691 factor may be zero use a wrapping type for the inner operation.
14692 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14693 and handle moved defs.
14694 (process_assignment): Properly guard the unary op case. Return a
14695 tri-state indicating that moving the stmt before the call may allow
14696 to continue. Pass through to_move.
14697 (find_tail_calls): Handle moving unrelated defs before
14700 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14703 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14704 splitter result in the canonical way.
14706 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14708 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14709 also for 32bit target. Update insn attributes.
14710 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14712 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14714 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14715 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14716 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14717 (_mm_maskz_min_ss): New intrinsics.
14719 2017-05-31 Martin Liska <mliska@suse.cz>
14721 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14722 Change comment style to one we normally use.
14723 (vectorizable_reduction): Likewise.
14724 (vectorizable_induction): Likewise.
14725 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14726 (vectorizable_call): Likewise.
14727 (vectorizable_simd_clone_call): Likewise.
14728 (vectorizable_conversion): Likewise.
14729 (vectorizable_assignment): Likewise.
14730 (vectorizable_shift): Likewise.
14731 (vectorizable_operation): Likewise.
14732 (vectorizable_store): Likewise.
14733 (vectorizable_load): Likewise.
14734 * tree-vectorizer.h: Likewise.
14736 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14738 * passes.c (emergency_dump_function): New.
14739 * tree-pass.h (emergency_dump_function): Declare.
14740 * plugin.c (plugins_internal_error_function): Remove.
14741 * plugin.h (plugins_internal_error_function): Remove declaration.
14742 * toplev.c (internal_error_function): New static function. Use it...
14743 (general_init): ...here.
14745 2017-05-31 Graham Markall <graham.markall@embecosm.com>
14747 * config/arc/arc.c (arc_print_operand): Handle constant operands.
14748 (arc_rtx_costs): Add costs for new patterns.
14749 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
14750 * config/arc/predicates.md: Add _1_2_3_operand predicate.
14752 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14754 * tree-ssa-strlen.c (get_next_strinfo): New function.
14755 (get_stridx_plus_constant): Use it.
14756 (zero_length_string): Likewise.
14757 (adjust_related_strinfos): Likewise.
14758 (adjust_last_stmt): Likewise.
14760 2017-05-31 Richard Biener <rguenther@suse.de>
14763 * config/i386/i386.c (ix86_expand_builtin): Remove assert
14764 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
14766 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14768 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
14769 loop_vinfo argument and use of dependence distance vectors.
14770 Check instead whether the two references differ only in their
14771 initial value and assume that they have the same alignment if the
14772 difference is a multiple of the vector alignment.
14773 (vect_analyze_data_refs_alignment): Update call accordingly.
14775 2017-05-31 Martin Liska <mliska@suse.cz>
14778 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
14780 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14782 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
14783 (create_intersect_range_checks): Move from ...
14784 * tree-data-ref.c (create_intersect_range_checks_index)
14785 (create_intersect_range_checks): ... to here.
14786 (create_runtime_alias_checks): New function factored from ...
14787 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
14788 here. Call above function.
14789 * tree-data-ref.h (create_runtime_alias_checks): New function.
14791 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14793 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
14794 segment length for dr_b and compute it in wide_int.
14796 2017-05-31 Richard Biener <rguenther@suse.de>
14798 PR tree-optimization/80906
14799 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
14800 and pass through iv_map.
14801 (copy_bb_and_scalar_dependences): Adjust.
14802 (translate_pending_phi_nodes): Likewise.
14803 (copy_loop_close_phi_args): Handle code-generating IVs instead
14806 2017-05-30 David Malcolm <dmalcolm@redhat.com>
14808 * diagnostic-color.c (color_dict): Add "type-diff".
14809 (parse_gcc_colors): Update comment.
14810 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
14811 -fdiagnostics-show-template-tree and -fno-elide-type.
14812 (GCC_COLORS): Add type-diff to example.
14814 (-fdiagnostics-show-template-tree): New.
14815 (-fno-elide-type): New.
14816 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
14817 the pp_format_decoder callback. Call any m_format_postprocessor's
14819 (pretty_printer::pretty_printer): Initialize
14820 m_format_postprocessor.
14821 (pretty_printer::~pretty_printer): Delete any
14822 m_format_postprocessor.
14823 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
14824 (class format_postprocessor): New class.
14825 (struct pretty_printer::format_decoder): Document the new parameters.
14826 (struct pretty_printer::m_format_postprocessor): New field.
14827 * tree-diagnostic.c (default_tree_printer): Update for new
14828 bool and const char ** params.
14829 * tree-diagnostic.h (default_tree_printer): Likewise.
14831 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
14833 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
14834 (lwa_operand): Delete rs6000_gen_cell_microcode test.
14835 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14836 rs6000_gen_cell_microcode code.
14837 (rs6000_final_prescan_insn): Delete.
14838 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
14839 "warn-cell-microcode" entries.
14840 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
14841 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
14842 throughout. Change cc_reg_not_micro_cr0_operand to
14843 cc_reg_not_cr0_operand throughout.
14844 (*extendhi<mode>2_noload): Delete.
14845 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
14846 (mwarn-cell-microcode): Delete.
14847 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
14848 -mgen-cell-microcode and -mwarn-cell-microcode.
14850 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
14853 * config/i386/constraints.md (Yd): New constraint.
14855 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
14856 and (?Yd, r) alternatives. Update insn attributes.
14857 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
14858 and (?*Yd, r) alternatives. Update insn attributes.
14859 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
14861 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
14863 * gimplify.c (gimplify_modify_expr): Don't create a
14864 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
14867 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14869 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
14871 2017-05-30 Richard Biener <rguenther@suse.de>
14873 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
14874 and reduc_def fields.
14875 (STMT_VINFO_REDUC_TYPE): New define.
14876 (STMT_VINFO_REDUC_DEF): Likewise.
14877 (vect_force_simple_reduction): Adjust prototype.
14878 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
14879 (vect_is_simple_reduction): Remove check_reduction argument.
14880 (vect_force_simple_reduction): Adjust and set
14881 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14882 (vectorizable_reduction): Do not re-do reduction analysis
14883 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14884 * tree-parloops.c (gather_scalar_reductions): Adjust.
14886 2017-05-30 Richard Biener <rguenther@suse.de>
14888 PR middle-end/80901
14889 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
14892 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
14894 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
14895 Introduce unknown_misalignment parameter and remove vf.
14896 (vect_peeling_hash_get_lowest_cost):
14897 Pass unknown_misalignment parameter.
14898 (vect_enhance_data_refs_alignment):
14899 Fix unsupportable data ref treatment.
14901 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14903 * tree-vect-data-refs.c (vect_get_data_access_cost):
14904 Workaround for SLP handling.
14905 (vect_enhance_data_refs_alignment):
14906 Compute costs for doing no peeling at all, compare to the best
14907 peeling costs so far and avoid peeling if cheaper.
14909 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14911 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
14912 Return peeling info and set costs to zero for unlimited cost
14914 (vect_enhance_data_refs_alignment): Also inspect all datarefs
14915 with unknown misalignment. Compute and costs for unknown
14916 misalignment, compare them to the costs for known misalignment
14917 and choose the cheapest for peeling.
14919 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14921 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
14922 (vect_get_peeling_costs_all_drs): Create function.
14923 (vect_peeling_hash_get_lowest_cost):
14924 Use vect_get_peeling_costs_all_drs.
14925 (vect_peeling_supportable): Create function.
14926 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
14928 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14930 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
14931 DR_HAS_NEGATIVE_STEP.
14932 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
14933 (vect_enhance_data_refs_alignment): Use.
14934 (vect_duplicate_ssa_name_ptr_info): Use.
14935 * tree-vectorizer.h (dr_misalignment): Use.
14936 (known_alignment_for_access_p): Use.
14938 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14941 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
14943 (has_section_name): New function.
14945 2017-05-30 Martin Liska <mliska@suse.cz>
14948 * auto-profile.c (get_function_decl_from_block): Fix
14951 2017-05-30 Richard Biener <rguenther@suse.de>
14953 PR middle-end/80876
14954 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
14956 2017-05-30 Martin Liska <mliska@suse.cz>
14958 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
14959 * dumpfile.h (struct dump_file_info): Remove ctors.
14961 2017-05-30 Martin Liska <mliska@suse.cz>
14963 * predict.def: Fix GNU coding style.
14965 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14967 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14968 Mark 'to' argument with ATTRIBUTE_UNUSED.
14970 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14972 * config/xtensa/xtensa.c (xtensa_emit_call): Use
14973 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
14974 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
14977 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
14979 * doc/install.texi (Options specification): Restore entry of
14980 --enable-sjlj-exceptions.
14982 2017-05-27 Michael Eager <eager@eagercon.com>
14985 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14987 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14989 * config/microblaze/microblaze.h
14990 (FIXED_REGISTERS): Update in macro.
14991 (CALL_USED_REGISTERS): Update in macro.
14993 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14995 * doc/install.texi: Add links to macOS binary distributions.
14997 2017-05-27 Jakub Jelinek <jakub@redhat.com>
15001 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15003 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15005 2017-05-26 Martin Liska <mliska@suse.cz>
15007 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
15009 2017-05-26 Martin Liska <mliska@suse.cz>
15011 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
15012 always leading ';; '.
15013 (dump_bb_info): Likewise.
15014 (brief_dump_cfg): Likewise.
15015 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
15016 * dumpfile.c: Remove usage of TDF_VERBOSE.
15017 * dumpfile.h (enum dump_kind): Likewise.
15018 (dump_gimple_bb_header): Do not use TDF_COMMENT.
15019 * print-tree.c (debug_verbose): Remove.
15020 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
15021 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
15022 * tree-diagnostic.c (default_tree_printer): Replace
15023 TDF_DIAGNOSTIC with TDF_SLIM.
15025 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15027 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
15028 in parameter loop, rather than loop_vinfo.
15029 (create_intersect_range_checks): Ditto.
15030 (vect_create_cond_for_alias_checks): Update call to above functions.
15032 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15034 PR tree-optimization/80815
15035 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
15036 for merging runtime alias checks. Handle negative DR_STEPs.
15038 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15040 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
15042 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
15043 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
15044 out code pruning runtime alias checks.
15045 * tree-data-ref.c (prune_runtime_alias_test_list): New function
15046 factored out from above.
15047 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15049 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15051 (prune_runtime_alias_test_list): New decalaration.
15053 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15055 * tree-vect-data-refs.c (compare_tree): Rename and move ...
15056 * tree-data-ref.c (data_ref_compare_tree): ... to here.
15057 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
15058 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
15059 (operator==, comp_dr_with_seg_len_pair): Ditto.
15060 (vect_prune_runtime_alias_test_list): Ditto.
15062 2017-05-26 Martin Liska <mliska@suse.cz>
15065 * params.def: Bound partial-inlining-entry-probability param.
15067 2017-05-26 Marek Polacek <polacek@redhat.com>
15070 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
15073 2017-05-26 Richard Biener <rguenther@suse.de>
15075 PR tree-optimization/80842
15076 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
15079 2017-05-26 Richard Biener <rguenther@suse.de>
15081 PR tree-optimization/80844
15082 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
15084 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
15086 * doc/md.texi (Machine Constraints): Update x86 family
15087 machine constraints section to match 'config/i386/constraints.md'.
15089 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
15091 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
15093 2017-05-25 Nathan Sidwell <nathan@acm.org>
15095 * doc/invoke.texi (--enable-languages): Update documentation.
15097 2017-05-25 Martin Liska <mliska@suse.cz>
15099 * dumpfile.c: Add TDF_FOLDING.
15100 * dumpfile.h (enum dump_kind): Likewise.
15101 * genmatch.c (dt_simplify::gen_1): Use it.
15103 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15105 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
15107 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15109 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15110 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
15112 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15114 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
15115 * match.pd (X == C): Rewrite it here.
15116 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
15117 with_certain_nonzero_bits2): New predicates.
15118 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
15120 2017-05-24 Nathan Sidwell <nathan@acm.org>
15122 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
15125 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
15128 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15130 * config/powerpcspe: New port. Files are copied from the rs6000
15131 port, with "rs6000" in filenames replaced by "powerpcspe".
15133 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
15135 PR rtl-optimization/80754
15136 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
15138 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
15140 * config/sparc/sparc.md (length): Return the correct value for -mflat
15141 sibcalls to match output_sibcall.
15143 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15147 * config/rs6000/rs6000.c (struct machine_function): Add new field
15149 (rs6000_get_separate_components): Init that field, use it.
15150 (rs6000_components_for_bb): Use the field.
15152 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
15154 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
15156 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
15158 PR middle-end/80823
15159 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
15161 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15164 * config/s390/s390.c (s390_check_qrst_address): Check incoming
15165 address against address_operand predicate.
15166 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
15168 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
15170 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
15171 parameters passed indirectly.
15173 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
15175 * config/i386/i386.md (*movdi_internal): Remove SSE4
15176 alternative 18 (?r, *v). Update insn attributes.
15177 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
15178 Update insn attributes.
15179 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
15180 Update insn attributes.
15181 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
15182 alternative 1 (r, v). Remove isa attribute.
15183 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
15184 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
15185 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
15187 2017-05-23 Tom de Vries <tom@codesourcery.com>
15189 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
15192 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15194 * cgraphunit.c (symbol_table::process_new_functions): Update.
15195 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
15196 (inline_generate_summary): Rename to ...
15197 (ipa_fn_summary_generate): ... this one.
15198 (inline_read_summary): Rename to ...
15199 (ipa_fn_summary_read): ... this one.
15200 (inline_write_summary): Rename to ...
15201 (ipa_fn_summary_write): ... this one.
15202 (inline_free_summary): Rename to ...
15203 (ipa_free_fn_summary): ... this one.
15204 (pass_data_local_fn_summary, pass_local_fn_summary,
15205 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
15206 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
15207 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
15208 make_pass_ipa_fn_summary): New.
15209 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
15210 inline_write_summary, inline_free_summary): Remove.
15211 (ipa_free_fn_summary) : New.
15212 * ipa-inline.c (ipa_inline): Update.
15213 (pass_ipa_inline): Do not generate summaries.
15214 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
15216 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
15217 and add pass_ipa_fn_summary.
15218 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
15220 (make_pass_inline_parameters): Remove.
15222 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
15224 * omp-low.c (struct omp_context): Remove "default_kind" member.
15227 * omp-offload.c (execute_oacc_device_lower): Remove the
15228 parallelism dimensions function attributes for unparallelized
15229 OpenACC kernels constructs.
15231 2017-05-23 Martin Liska <mliska@suse.cz>
15233 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
15235 (cgraph_edge::make_speculative): Likewise.
15236 (cgraph_edge::resolve_speculation): Likewise.
15237 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15238 (cgraph_node::dump): Likewise.
15239 * cgraph.h: Likewise.
15240 * cgraphunit.c (analyze_functions): Likewise.
15241 (symbol_table::compile): Likewise.
15242 * ipa-cp.c (print_all_lattices): Likewise.
15243 (determine_versionability): Likewise.
15244 (initialize_node_lattices): Likewise.
15245 (ipcp_verify_propagated_values): Likewise.
15246 (estimate_local_effects): Likewise.
15247 (update_profiling_info): Likewise.
15248 (create_specialized_node): Likewise.
15249 (perhaps_add_new_callers): Likewise.
15250 (decide_about_value): Likewise.
15251 (decide_whether_version_node): Likewise.
15252 (identify_dead_nodes): Likewise.
15253 (ipcp_store_bits_results): Likewise.
15254 * ipa-devirt.c (dump_targets): Likewise.
15255 (ipa_devirt): Likewise.
15256 * ipa-icf.c (sem_item::dump): Likewise.
15257 (sem_function::equals): Likewise.
15258 (sem_variable::equals): Likewise.
15259 (sem_item_optimizer::read_section): Likewise.
15260 (sem_item_optimizer::execute): Likewise.
15261 (congruence_class::dump): Likewise.
15262 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
15263 (dump_inline_summary): Likewise.
15264 (estimate_node_size_and_time): Likewise.
15265 (inline_analyze_function): Likewise.
15266 * ipa-inline-transform.c (inline_call): Likewise.
15267 * ipa-inline.c (report_inline_failed_reason): Likewise.
15268 (want_early_inline_function_p): Likewise.
15269 (edge_badness): Likewise.
15270 (update_edge_key): Likewise.
15271 (inline_small_functions): Likewise.
15272 * ipa-profile.c (ipa_profile): Likewise.
15273 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
15274 (ipa_make_edge_direct_to_target): Likewise.
15275 (remove_described_reference): Likewise.
15276 (ipa_impossible_devirt_target): Likewise.
15277 (propagate_controlled_uses): Likewise.
15278 (ipa_print_node_params): Likewise.
15279 (ipcp_transform_function): Likewise.
15280 * ipa-pure-const.c (pure_const_read_summary): Likewise.
15281 (propagate_pure_const): Likewise.
15282 * ipa-reference.c (generate_summary): Likewise.
15283 (read_write_all_from_decl): Likewise.
15284 (propagate): Likewise.
15285 (ipa_reference_read_optimization_summary): Likewise.
15286 * ipa-utils.c (ipa_merge_profiles): Likewise.
15287 * ipa.c (walk_polymorphic_call_targets): Likewise.
15288 (symbol_table::remove_unreachable_nodes): Likewise.
15289 (ipa_single_use): Likewise.
15290 * passes.c (execute_todo): Likewise.
15291 * predict.c (drop_profile): Likewise.
15292 * symtab.c (symtab_node::get_dump_name): New function.
15293 (symtab_node::dump_name): Likewise.
15294 (symtab_node::dump_asm_name): Likewise.
15295 (symtab_node::dump_references): Likewise.
15296 (symtab_node::dump_referring): Likewise.
15297 (symtab_node::dump_base): Likewise.
15298 (symtab_node::debug_symtab): Likewise.
15299 * tree-sra.c (convert_callers_for_node): Likewise.
15300 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
15301 * value-prof.c (init_node_map): Likewise.
15303 2017-05-23 Martin Liska <mliska@suse.cz>
15305 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
15306 and symtab_node::debug_symtab to symbol_table::debug.
15307 * cgraphunit.c (analyze_functions): Use the renamed function.
15308 (symbol_table::compile): Likewise.
15309 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
15310 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
15311 * passes.c (execute_todo): Likewise.
15312 * symtab.c (symbol_table::dump): New function.
15313 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
15315 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15317 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
15318 that nonconst implies exec.
15320 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15322 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15323 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15324 (inline_edge_summary_vec): Turn into ...
15325 (ipa_call_summaries): ... this one.
15326 (redirect_to_unreachable, edge_set_predicate,
15327 evaluate_properties_for_edge, inline_summary_alloc,
15328 reset_ipa_call_summary, reset_inline_summary,
15329 inline_summary_t::duplicate): Update.
15330 (inline_edge_duplication_hook): Turn to ...
15331 (ipa_call_summary_t::duplicate): ... this one.
15332 (inline_edge_removal_hook): Turn to ...
15333 (ipa_call_summary_t::remove): ... this one.
15334 (dump_inline_edge_summary): Turn to ...
15335 (dump_ipa_call_summary): ... this one.
15336 (estimate_function_body_sizes): Update.
15337 (inline_update_callee_summaries): Update.
15338 (remap_edge_change_prob): Update.
15339 (remap_edge_summaries): Update.
15340 (inline_merge_summary): Update.
15341 (do_estimate_edge_time): Update.
15342 (inline_generate_summary): Update.
15343 (inline_read_section): Update.
15344 (inline_read_summary): Update.
15345 (inline_free_summary): Update.
15346 * ipa-inline.c (can_inline_edge_p): Update.
15347 (compute_inlined_call_time): Update.
15348 (want_inline_small_function_p): Update.
15349 (edge_badness): Update.
15350 (early_inliner): Update.
15351 * ipa-inline.h (inline_edge_summary): Turn to ...
15352 (ipa_call_summary): ... this one.
15353 (ipa_call_summary_t): New class.
15354 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15355 (ipa_call_summaries): New.
15356 (inline_edge_summary): Remove.
15357 (estimate_edge_growth): Update.
15358 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15359 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15360 * ipa-split.c (execute_split_functions): Update.
15361 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15363 2017-05-23 Tom de Vries <tom@codesourcery.com>
15365 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15366 attributes): Document rdrand effective target.
15368 2017-05-23 Tom de Vries <tom@codesourcery.com>
15370 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15371 attributes): Sort alphabetically.
15373 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
15375 * config/avr/genmultilib.awk: Use gsub instead of gensub.
15377 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15380 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
15381 V2DF/V2DI splat into two separate patterns, one that handles
15382 registers, and the other that only handles memory. Drop support
15383 for splatting from a GPR on ISA 2.07 and then splitting the
15384 splat into direct move and splat.
15385 (vsx_splat_<mode>_reg): Likewise.
15386 (vsx_splat_<mode>_mem): Likewise.
15388 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
15390 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
15392 2017-05-22 Jakub Jelinek <jakub@redhat.com>
15394 PR middle-end/80809
15395 * omp-low.c (finish_taskreg_remap): New function.
15396 (finish_taskreg_scan): If unit size of ctx->record_type
15397 is non-constant, unshare the size expression and replace
15398 decls in it with possible outer var refs.
15400 PR middle-end/80809
15401 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15402 GOVD_SHARED rather than GOVD_PRIVATE with it.
15403 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15404 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15406 PR middle-end/80853
15407 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15408 as last argument to build_outer_var_ref for pointer bases of array
15409 section reductions.
15411 2017-05-19 Martin Sebor <msebor@redhat.com>
15413 * print-tree.c (print_node): Print DECL_READ_P flag.
15415 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15417 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15418 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15419 * cgraph.c: Likewise.
15420 * cgraphunit.c: Likewise.
15421 * gengtype.c: Likewise.
15422 * ipa-cp.c: Likewise.
15423 * ipa-devirt.c: Likewise.
15424 * ipa-icf.c: Likewise.
15425 * ipa-predicate.c: Likewise.
15426 * ipa-profile.c: Likewise.
15427 * ipa-prop.c: Likewise.
15428 * ipa-split.c: Likewise.
15430 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15431 edge_predicate_pool, dump_inline_hints,
15432 inline_summary::account_size_time, redirect_to_unreachable,
15433 edge_set_predicate, set_hint_predicate,
15434 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15435 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15436 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15437 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15438 ipa_call_summary_t::remove, initialize_growth_caches,
15439 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15440 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15441 mark_modified, unmodified_parm_1, unmodified_parm,
15442 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15443 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15444 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15445 will_be_nonconstant_predicate, record_modified_bb_info,
15446 get_minimal_bb, record_modified, param_change_prob,
15447 phi_result_unknown_predicate, predicate_for_phi_result,
15448 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15449 estimate_function_body_sizes, compute_inline_parameters,
15450 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15451 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15452 inline_update_callee_summaries, remap_edge_change_prob,
15453 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15454 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15455 inline_analyze_function, inline_summary_t::insert,
15456 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15457 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15458 inline_free_summary): Move to ipa-fnsummary.h
15459 (predicate_t): Remove.
15460 * ipa-fnsummary.c: New file.
15461 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15462 (enum inline_hints_vals, inline_hints, agg_position_info,
15463 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15464 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15465 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15466 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15467 inline_read_summary, inline_write_summary, inline_free_summary,
15468 inline_analyze_function, initialize_inline_failed,
15469 inline_merge_summary, inline_update_overall_summary,
15470 compute_inline_parameters): Move to ipa-fnsummary.h
15471 * ipa-fnsummary.h: New file.
15472 * ipa-inline-transform.h: Include ipa-inline.h.
15473 * ipa-inline.c: LIkewise.
15475 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15477 * ipa-inline.c (edge_badness): Use inlined_time instead of
15478 inline_summaries->get.
15480 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15482 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15484 2017-05-22 Nathan Sidwell <nathan@acm.org>
15486 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15487 (fdump-lang): Document 'raw' option.
15488 * dumpfile.h (TDI_tu): Delete.
15489 * dumpfile.c (dump_files): Remove translation-unit.
15490 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15492 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15494 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15495 command option from $(AWK) call.
15496 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15497 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15498 [FORMAT]: Remove handling of variable.
15499 * config/avr/t-multilib: Regenerate.
15501 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15503 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15505 (dump_inline_summary): Do not print self_time.
15506 (estimate_function_body_sizes): Do not set self_time.
15507 (compute_inline_parameters): Likewise.
15508 (inline_read_section, inline_write_summary): Do not stream self_time.
15509 * ipa-inline.h (inline_summary): Drop self_time.
15511 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15513 * ipa-inline-analysis.c (account_size_time): Rename to ...
15514 (inline_summary::account_size_time): ... this one.
15515 (reset_ipa_call_summary): Turn to ...
15516 (ipa_call_summary::reset): ... this one.
15517 (reset_inline_summary): Turn to ...
15518 (inline_summary::reset): ... this one.
15519 (inline_summary_t::remove): Update.
15520 (inline_summary_t::duplicate): Update.
15521 (ipa_call_summary_t::remove): Update.
15522 (dump_inline_summary): Update.
15523 (estimate_function_body_sizes): Update.
15524 (compute_inline_parameters): Update.
15525 (estimate_node_size_and_time): Update.
15526 (inline_merge_summary): Update.
15527 (inline_update_overall_summary): Update.
15528 (inline_read_section): Update.
15529 (inline_write_summary): Update.
15530 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15531 add account_size_time and reset member functions.
15532 (ipa_call_summary): Add reset function.
15533 * ipa-predicate.h (predicate::operator &): Constify.
15535 2017-05-22 Richard Biener <rguenther@suse.de>
15537 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15539 2017-05-19 Jason Merrill <jason@redhat.com>
15541 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15543 2017-05-19 Marek Polacek <polacek@redhat.com>
15546 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15547 TYPE_OVERFLOW_WRAPS checks.
15549 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15551 * tree-core.h (enum omp_clause_default_kind): Add
15552 "OMP_CLAUSE_DEFAULT_PRESENT".
15553 * tree-pretty-print.c (dump_omp_clause): Handle it.
15554 * gimplify.c (enum gimplify_omp_var_data): Add
15555 "GOVD_MAP_FORCE_PRESENT".
15556 (gimplify_adjust_omp_clauses_1): Map it to
15557 "GOMP_MAP_FORCE_PRESENT".
15558 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15560 * gimplify.c (oacc_default_clause): Clarify.
15562 2017-05-19 Nathan Sidwell <nathan@acm.org>
15564 LANG_HOOK_REGISTER_DUMPS
15565 * toplev.c (general_init): Call register dump lang hook.
15566 * doc/invoke.texi: Document -fdump-lang option family.
15567 * dumpfile.c (dump_files): Remove class dump here.
15568 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15569 * dumpfile.h (tree_dump_index): Remove TDI_class.
15570 * langhooks-def.h (lhd_register_dumps): Declare.
15571 (LANG_HOOKS_REGISTER_DUMPS): Define.
15572 (LANG_HOOKS_INITIALIZER): Add it.
15573 * langhooks.c (lhd_register_dumps): Define.
15574 * langhooks.h (struct lang_hooks): Add register_dumps.
15576 2017-05-19 Nathan Sidwell <nathan@acm.org>
15578 * context.h (context::set_passes): New.
15579 * context.c (context::context): Do not create pass manager.
15580 * toplev.c (general_init): Create pass manager here.
15582 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15584 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15585 use this splitter if two add or or instructions would also work for
15586 the constant we want to generate.
15588 2017-05-19 Richard Biener <rguenther@suse.de>
15591 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15592 predicate evaluation.
15594 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15596 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15598 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15601 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15603 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15604 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15605 (inline_edge_summary_vec): Turn into ...
15606 (ipa_call_summaries): ... this one.
15607 (redirect_to_unreachable, edge_set_predicate,
15608 evaluate_properties_for_edge, inline_summary_alloc,
15609 reset_ipa_call_summary, reset_inline_summary,
15610 inline_summary_t::duplicate): Update.
15611 (inline_edge_duplication_hook): Turn to ...
15612 (ipa_call_summary_t::duplicate): ... this one.
15613 (inline_edge_removal_hook): Turn to ...
15614 (ipa_call_summary_t::remove): ... this one.
15615 (dump_inline_edge_summary): Turn to ...
15616 (dump_ipa_call_summary): ... this one.
15617 (estimate_function_body_sizes): Update.
15618 (inline_update_callee_summaries): Update.
15619 (remap_edge_change_prob): Update.
15620 (remap_edge_summaries): Update.
15621 (inline_merge_summary): Update.
15622 (do_estimate_edge_time): Update.
15623 (inline_generate_summary): Update.
15624 (inline_read_section): Update.
15625 (inline_read_summary): Update.
15626 (inline_free_summary): Update.
15627 * ipa-inline.c (can_inline_edge_p): Update.
15628 (compute_inlined_call_time): Update.
15629 (want_inline_small_function_p): Update.
15630 (edge_badness): Update.
15631 (early_inliner): Update.
15632 * ipa-inline.h (inline_edge_summary): Turn to ...
15633 (ipa_call_summary): ... this one.
15634 (ipa_call_summary_t): New class.
15635 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15636 (ipa_call_summaries): New.
15637 (inline_edge_summary): Remove.
15638 (estimate_edge_growth): Update.
15639 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15640 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15641 * ipa-split.c (execute_split_functions): Update.
15642 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15644 2017-05-19 Richard Biener <rguenther@suse.de>
15646 PR middle-end/80764
15647 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
15649 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
15651 * config/rs6000/rs6000.c (struct machine_function): Add field
15652 fpr_is_wrapped_separately.
15653 (rs6000_get_separate_components): Use 64 components. Handle the
15654 new FPR components.
15655 (rs6000_components_for_bb): Handle the FPR components.
15656 (rs6000_emit_prologue_components): Handle the FPR components.
15657 (rs6000_emit_epilogue_components): Handle the FPR components.
15658 (rs6000_set_handled_components): Handle the FPR components.
15659 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15660 that are already separately shrink-wrapped.
15661 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15662 that are already separately shrink-wrapped.
15664 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15667 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15670 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15671 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15672 cases where the register allocator uses a move and an offsettable
15673 memory operation to/from a FPR register on ISA 2.06/2.07.
15674 (define_peephole2 for Altivec d-form store): Likewise.
15676 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15679 * config/i386/mmx.md (*mov<mode>_internal): Enable
15680 alternatives 11, 12, 13 and 14 also for 32bit targets.
15681 Remove alternatives 15, 16, 17 and 18.
15682 * config/i386/sse.md (vec_concatv2di): Change
15683 alternative (!x, *y) to (x, ?!*Yn).
15685 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15687 * dumpfile.h (enum dump_kind): Remove stray comma.
15689 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15691 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15692 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15693 predicate::num_conditions
15694 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15695 (CHANGED): turn into predicate::changed.
15696 (agg_position_info): Move to ipa-predicate.h
15697 (add_condition, predicate::add_clause, predicate::operator &=,
15698 predicate::or_with, predicate::evaluate, predicate::probability,
15699 dump_condition, dump_clause, predicate::dump,
15700 predicate::remap_after_duplication, predicate::remap_after_inlining,
15701 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15702 (evaluate_conditions_for_known_args): Update.
15703 (set_cond_stmt_execution_predicate): Update.
15704 * ipa-inline.h: Include ipa-predicate.h
15705 (condition, inline_param_summary, conditions, agg_position_info,
15706 predicate): Move to ipa-predicate.h
15707 * ipa-predicate.c: New file.
15708 * ipa-predicate.h: New file.
15710 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15712 * final.c (leaf_function_p): Check we are not in a sequence.
15714 2017-05-18 Martin Liska <mliska@suse.cz>
15716 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15717 * dumpfile.c (dump_register): Use new enum dump_kind.
15718 (get_dump_file_name): Likewise.
15719 (dump_enable_all): Likewise.
15720 (dump_switch_p_1): Likewise.
15721 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15722 * dumpfile.h (enum dump_kind): New enum type.
15723 (struct dump_file_info): Create constructor and
15724 format fields and comments.
15725 * passes.c (pass_manager::register_one_dump_file):
15727 * statistics.c (statistics_early_init): Likewise.
15728 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15729 TDF_TREE with TDF_SLIM.
15730 (gather_memory_references_ref): Likewise.
15732 2017-05-18 Martin Liska <mliska@suse.cz>
15734 * vec.h (struct vnull): Use it.
15736 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15738 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15739 (true_predicate, false_predicate, true_predicate_p,
15740 false_predicate_p): Remove.
15741 (single_cond_predicate, not_inlined_predicate): Turn to member function
15743 (add_condition): Update.
15744 (add_clause): Turn to...
15745 (predicate::add_clause): ... this one; update; allow passing NULL
15747 (and_predicates): Turn to ...
15748 (predicate::operator &=): ... this one.
15749 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
15750 (or_predicates): Turn to ...
15751 (predicate::or_with): ... this one.
15752 (evaluate_predicate): Turn to ...
15753 (predicate::evaluate): ... this one.
15754 (predicate_probability): Turn to ...
15755 (predicate::probability): ... this one.
15756 (dump_condition): Update.
15757 (dump_predicate): Turn to ...
15758 (predicate::dump): ... this one.
15759 (account_size_time): Update.
15760 (edge_set_predicate): Update.
15761 (set_hint_predicate): UPdate.
15762 (evaluate_conditions_for_known_args): Update.
15763 (evaluate_properties_for_edge): Update.
15764 (remap_predicate_after_duplication): Turn to...
15765 (predicate::remap_after_duplication): ... this one.
15766 (remap_hint_predicate_after_duplication): Update.
15767 (inline_summary_t::duplicate): UPdate.
15768 (dump_inline_edge_summary): Update.
15769 (dump_inline_summary): Update.
15770 (set_cond_stmt_execution_predicate): Update.
15771 (set_switch_stmt_execution_predicate): Update.
15772 (compute_bb_predicates): Update.
15773 (will_be_nonconstant_expr_predicate): Update.
15774 (will_be_nonconstant_predicate): Update.
15775 (phi_result_unknown_predicate): Update.
15776 (predicate_for_phi_result): Update.
15777 (array_index_predicate): Update.
15778 (estimate_function_body_sizes): Update.
15779 (estimate_node_size_and_time): Update.
15780 (estimate_ipcp_clone_size_and_time): Update.
15781 (remap_predicate): Rename to ...
15782 (predicate::remap_after_inlining): ... this one.
15783 (remap_hint_predicate): Update.
15784 (inline_merge_summary): Update.
15785 (inline_update_overall_summary): Update.
15786 (estimate_size_after_inlining): Update.
15787 (read_predicate): Rename to ...
15788 (predicate::stream_in): ... this one.
15789 (read_inline_edge_summary): Update.
15790 (write_predicate): Rename to ...
15791 (predicate::stream_out): ... this one.
15792 (write_inline_edge_summary): Update.
15793 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
15794 (clause_t): Turn to uint32_t
15795 (predicate): Turn to class; implement constructor and operators
15797 (size_time_entry): Update.
15798 (inline_summary): Update.
15799 (inline_edge_summary): Update.
15801 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
15803 * fold-const.c (fold_binary_loc): Move transformation...
15804 * match.pd (C - X CMP X): ... here.
15806 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
15808 * config/sparc/sparc.c (sparc_option_override): Set function
15809 alignment for -mcpu=niagara7 to 64 to match the I$ line.
15810 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
15812 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
15814 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
15816 2017-05-18 Marek Polacek <polacek@redhat.com>
15819 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
15820 (pass_ubsan::execute): Call gimple_assign_single_p instead of
15821 gimple_assign_load_p.
15823 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
15825 PR middle-end/80692
15826 * real.c (do_compare): Give decimal_do_compare preference over
15827 comparing just the signs.
15829 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
15831 * doc/md.texi (Canonicalization of Instructions): Describe the
15832 canonical form of instructions that inherently set a condition
15835 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
15837 PR middle-end/80775
15838 * tree-cfg.c: Move deletion of unreachable case statements to after
15839 the merging of consecutive case labels.
15841 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
15843 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
15844 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
15845 restoring of callee-saved registers.
15847 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
15849 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
15850 * config/visium/visium.c (single_set_and_flags): Likewise.
15851 * config/visium/visium.md (Substitutions): Likewise.
15853 2017-05-17 Martin Liska <mliska@suse.cz>
15855 * cfg.c: Introduce dump_flags_t type and
15856 use it instead of int type.
15858 * cfghooks.c: Likewise.
15859 * cfghooks.h (struct cfg_hooks): Likewise.
15860 * cfgrtl.c: Likewise.
15861 * cfgrtl.h: Likewise.
15862 * cgraph.c (cgraph_node::get_body): Likewise.
15863 * coretypes.h: Likewise.
15864 * domwalk.c: Likewise.
15865 * domwalk.h: Likewise.
15866 * dumpfile.c (struct dump_option_value_info): Likewise.
15867 (dump_enable_all): Likewise.
15868 (dump_switch_p_1): Likewise.
15869 (opt_info_switch_p): Likewise.
15870 * dumpfile.h (enum tree_dump_index): Likewise.
15871 (struct dump_file_info): Likewise.
15872 * genemit.c: Likewise.
15873 * generic-match-head.c: Likewise.
15874 * gengtype.c (open_base_files): Likewise.
15875 * gimple-pretty-print.c: Likewise.
15876 * gimple-pretty-print.h: Likewise.
15877 * graph.c (print_graph_cfg): Likewise.
15878 * graphite-scop-detection.c (dot_all_sese): Likewise.
15879 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
15880 * loop-unroll.c (report_unroll): Likewise.
15881 * passes.c (pass_manager::register_one_dump_file): Likewise.
15882 * print-tree.c: Likewise.
15883 * statistics.c: Likewise.
15884 * tree-cfg.c: Likewise.
15885 * tree-cfg.h: Likewise.
15886 * tree-dfa.c: Likewise.
15887 * tree-dfa.h: Likewise.
15888 * tree-dump.c (dump_function): Likewise.
15889 * tree-dump.h (struct dump_info): Likewise.
15890 * tree-pretty-print.c: Likewise.
15891 * tree-pretty-print.h: Likewise.
15892 * tree-ssa-live.c: Likewise.
15893 * tree-ssa-live.h: Likewise.
15894 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
15895 * tree-vect-loop.c: Likewise.
15896 * tree-vect-slp.c: Likewise.
15898 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
15899 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15901 PR tree-optimization/80457
15902 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
15903 of all arguments to a statement as scalar_to_vec operations.
15904 (vectorizable_call): Adjust call to vect_model_simple_cost for
15906 (vectorizable_conversion): Likewise.
15907 (vectorizable_assignment): Likewise.
15908 (vectorizable_shift): Likewise.
15909 (vectorizable_operation): Likewise.
15910 (vectorizable_comparison): Likewise.
15911 (vect_is_simple_cond): Record the def types for operands.
15912 (vectorizable_condition): Likewise, call vect_model_simple_cost.
15913 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
15914 for statement argument count.
15916 2017-05-16 Carl Love <cel@us.ibm.com>
15918 * config/rs6000/rs6000-c: Add support for built-in functions
15919 vector unsigned long long vec_bperm (vector unsigned long long,
15920 vector unsigned char)
15921 vector signed long long vec_mule (vector signed int,
15923 vector unsigned long long vec_mule (vector unsigned int,
15924 vector unsigned int)
15925 vector signed long long vec_mulo (vector signed int,
15927 vector unsigned long long vec_mulo (vector unsigned int,
15928 vector unsigned int)
15929 vector signed char vec_sldw (vector signed char,
15930 vector signed char,
15932 vector unsigned char vec_sldw (vector unsigned char,
15933 vector unsigned char,
15935 vector signed short vec_sldw (vector signed short,
15936 vector signed short,
15938 vector unsigned short vec_sldw (vector unsigned short,
15939 vector unsigned short,
15941 vector signed int vec_sldw (vector signed int,
15944 vector unsigned int vec_sldw (vector unsigned int,
15945 vector unsigned int,
15947 vector signed long long vec_sldw (vector signed long long,
15948 vector signed long long,
15950 vector unsigned long long vec_sldw (vector unsigned long long,
15951 vector unsigned long long,
15953 * config/rs6000/rs6000-c: Add support for built-in functions
15954 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
15955 * config/rs6000/altivec.h: Add defintion for vec_sldw.
15956 * doc/extend.texi: Update the built-in documentation for the
15957 new built-in functions.
15959 2017-05-16 Marek Polacek <polacek@redhat.com>
15963 * tree.c (save_expr): Don't fold the expression.
15965 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
15967 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
15968 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
15969 and (?*y,m). Update insn attributes.
15971 2017-05-16 Martin Liska <mliska@suse.cz>
15973 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
15974 flags argument of print_gimple_stmt, print_gimple_expr,
15975 print_generic_stmt and print_generic_expr.
15976 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
15977 * coretypes.h: Likewise.
15978 * except.c (dump_eh_tree): Likewise.
15979 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15980 * gimple-pretty-print.h: Likewise.
15981 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
15982 (backprop::push_to_worklist): Likewise.
15983 (backprop::pop_from_worklist): Likewise.
15984 (backprop::process_use): Likewise.
15985 (backprop::intersect_uses): Likewise.
15986 (note_replacement): Likewise.
15987 * gimple-ssa-store-merging.c
15988 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15989 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15990 (pass_store_merging::execute): Likewise.
15991 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15992 (ssa_base_cand_dump_callback): Likewise.
15993 (dump_incr_vec): Likewise.
15994 (replace_refs): Likewise.
15995 (replace_mult_candidate): Likewise.
15996 (create_add_on_incoming_edge): Likewise.
15997 (create_phi_basis): Likewise.
15998 (insert_initializers): Likewise.
15999 (all_phi_incrs_profitable): Likewise.
16000 (introduce_cast_before_cand): Likewise.
16001 (replace_one_candidate): Likewise.
16002 * gimplify.c (gimplify_expr): Likewise.
16003 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
16004 (set_rename): Likewise.
16005 (rename_uses): Likewise.
16006 (copy_loop_phi_nodes): Likewise.
16007 (add_close_phis_to_merge_points): Likewise.
16008 (copy_loop_close_phi_args): Likewise.
16009 (copy_cond_phi_args): Likewise.
16010 (graphite_copy_stmts_from_block): Likewise.
16011 (translate_pending_phi_nodes): Likewise.
16012 * graphite-poly.c (print_pdr): Likewise.
16013 (dump_gbb_cases): Likewise.
16014 (dump_gbb_conditions): Likewise.
16015 (print_scop_params): Likewise.
16016 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
16017 (build_cross_bb_scalars_use): Likewise.
16018 (gather_bbs::before_dom_children): Likewise.
16019 * hsa-dump.c (dump_hsa_immed): Likewise.
16020 * ipa-cp.c (print_ipcp_constant_value): Likewise.
16021 (get_replacement_map): Likewise.
16022 * ipa-inline-analysis.c (dump_condition): Likewise.
16023 (estimate_function_body_sizes): Likewise.
16024 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
16025 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
16026 * ipa-prop.c (ipa_dump_param): Likewise.
16027 (ipa_print_node_jump_functions_for_edge): Likewise.
16028 (ipa_modify_call_arguments): Likewise.
16029 (ipa_modify_expr): Likewise.
16030 (ipa_dump_param_adjustments): Likewise.
16031 (ipa_dump_agg_replacement_values): Likewise.
16032 (ipcp_modif_dom_walker::before_dom_children): Likewise.
16033 * ipa-pure-const.c (check_stmt): Likewise.
16034 (pass_nothrow::execute): Likewise.
16035 * ipa-split.c (execute_split_functions): Likewise.
16036 * omp-offload.c (dump_oacc_loop_part): Likewise.
16037 (dump_oacc_loop): Likewise.
16038 * trans-mem.c (tm_log_emit): Likewise.
16039 (tm_memopt_accumulate_memops): Likewise.
16040 (dump_tm_memopt_set): Likewise.
16041 (dump_tm_memopt_transform): Likewise.
16042 * tree-cfg.c (gimple_verify_flow_info): Likewise.
16043 (print_loop): Likewise.
16044 * tree-chkp-opt.c (chkp_print_addr): Likewise.
16045 (chkp_gather_checks_info): Likewise.
16046 (chkp_get_check_result): Likewise.
16047 (chkp_remove_check_if_pass): Likewise.
16048 (chkp_use_outer_bounds_if_possible): Likewise.
16049 (chkp_reduce_bounds_lifetime): Likewise.
16050 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
16051 (chkp_mark_completed_bounds): Likewise.
16052 (chkp_register_incomplete_bounds): Likewise.
16053 (chkp_mark_invalid_bounds): Likewise.
16054 (chkp_maybe_copy_and_register_bounds): Likewise.
16055 (chkp_build_returned_bound): Likewise.
16056 (chkp_get_bound_for_parm): Likewise.
16057 (chkp_build_bndldx): Likewise.
16058 (chkp_get_bounds_by_definition): Likewise.
16059 (chkp_generate_extern_var_bounds): Likewise.
16060 (chkp_get_bounds_for_decl_addr): Likewise.
16061 * tree-chrec.c (chrec_apply): Likewise.
16062 * tree-data-ref.c (dump_data_reference): Likewise.
16063 (dump_subscript): Likewise.
16064 (dump_data_dependence_relation): Likewise.
16065 (analyze_overlapping_iterations): Likewise.
16066 * tree-inline.c (expand_call_inline): Likewise.
16067 (tree_function_versioning): Likewise.
16068 * tree-into-ssa.c (dump_defs_stack): Likewise.
16069 (dump_currdefs): Likewise.
16070 (dump_names_replaced_by): Likewise.
16071 (dump_update_ssa): Likewise.
16072 (update_ssa): Likewise.
16073 * tree-object-size.c (pass_object_sizes::execute): Likewise.
16074 * tree-parloops.c (build_new_reduction): Likewise.
16075 (try_create_reduction_list): Likewise.
16076 (ref_conflicts_with_region): Likewise.
16077 (oacc_entry_exit_ok_1): Likewise.
16078 (oacc_entry_exit_single_gang): Likewise.
16079 * tree-pretty-print.h: Likewise.
16080 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
16081 (get_scalar_evolution): Likewise.
16082 (add_to_evolution): Likewise.
16083 (get_loop_exit_condition): Likewise.
16084 (analyze_evolution_in_loop): Likewise.
16085 (analyze_initial_condition): Likewise.
16086 (analyze_scalar_evolution): Likewise.
16087 (instantiate_scev): Likewise.
16088 (number_of_latch_executions): Likewise.
16089 (gather_chrec_stats): Likewise.
16090 (final_value_replacement_loop): Likewise.
16091 (scev_const_prop): Likewise.
16092 * tree-sra.c (dump_access): Likewise.
16093 (disqualify_candidate): Likewise.
16094 (create_access): Likewise.
16095 (reject): Likewise.
16096 (maybe_add_sra_candidate): Likewise.
16097 (create_access_replacement): Likewise.
16098 (analyze_access_subtree): Likewise.
16099 (analyze_all_variable_accesses): Likewise.
16100 (sra_modify_assign): Likewise.
16101 (initialize_constant_pool_replacements): Likewise.
16102 (find_param_candidates): Likewise.
16103 (decide_one_param_reduction): Likewise.
16104 (replace_removed_params_ssa_names): Likewise.
16105 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
16106 * tree-ssa-copy.c (dump_copy_of): Likewise.
16107 (copy_prop_visit_cond_stmt): Likewise.
16108 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
16109 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
16110 (record_equivalences_from_stmt): Likewise.
16111 * tree-ssa-dse.c (compute_trims): Likewise.
16112 (delete_dead_call): Likewise.
16113 (delete_dead_assignment): Likewise.
16114 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
16115 (forward_propagate_into_cond): Likewise.
16116 (pass_forwprop::execute): Likewise.
16117 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16118 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
16120 (move_computations_worker): Likewise.
16121 (execute_sm): Likewise.
16122 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
16123 (remove_exits_and_undefined_stmts): Likewise.
16124 (remove_redundant_iv_tests): Likewise.
16125 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
16126 (adjust_iv_update_pos): Likewise.
16127 * tree-ssa-math-opts.c (bswap_replace): Likewise.
16128 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
16129 (value_replacement): Likewise.
16130 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
16131 * tree-ssa-pre.c (print_pre_expr): Likewise.
16132 (get_representative_for): Likewise.
16133 (create_expression_by_pieces): Likewise.
16134 (insert_into_preds_of_block): Likewise.
16135 (eliminate_insert): Likewise.
16136 (eliminate_dom_walker::before_dom_children): Likewise.
16137 (eliminate): Likewise.
16138 (remove_dead_inserted_code): Likewise.
16139 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
16140 * tree-ssa-reassoc.c (get_rank): Likewise.
16141 (eliminate_duplicate_pair): Likewise.
16142 (eliminate_plus_minus_pair): Likewise.
16143 (eliminate_not_pairs): Likewise.
16144 (undistribute_ops_list): Likewise.
16145 (eliminate_redundant_comparison): Likewise.
16146 (update_range_test): Likewise.
16147 (optimize_range_tests_var_bound): Likewise.
16148 (optimize_vec_cond_expr): Likewise.
16149 (rewrite_expr_tree): Likewise.
16150 (rewrite_expr_tree_parallel): Likewise.
16151 (linearize_expr): Likewise.
16152 (break_up_subtract): Likewise.
16153 (linearize_expr_tree): Likewise.
16154 (attempt_builtin_powi): Likewise.
16155 (attempt_builtin_copysign): Likewise.
16156 (transform_stmt_to_copy): Likewise.
16157 (transform_stmt_to_multiply): Likewise.
16158 (dump_ops_vector): Likewise.
16159 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
16160 (print_scc): Likewise.
16161 (set_ssa_val_to): Likewise.
16162 (visit_reference_op_store): Likewise.
16163 (visit_use): Likewise.
16164 (sccvn_dom_walker::before_dom_children): Likewise.
16165 (run_scc_vn): Likewise.
16166 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16168 (expr_hash_elt::print): Likewise.
16169 (const_and_copies::pop_to_marker): Likewise.
16170 (const_and_copies::record_const_or_copy_raw): Likewise.
16171 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
16172 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
16173 (dump_predicates): Likewise.
16174 (find_uninit_use): Likewise.
16175 (warn_uninitialized_phi): Likewise.
16176 (pass_late_warn_uninitialized::execute): Likewise.
16177 * tree-ssa.c (verify_vssa): Likewise.
16178 (verify_ssa): Likewise.
16179 (maybe_optimize_var): Likewise.
16180 * tree-vrp.c (dump_value_range): Likewise.
16181 (dump_all_value_ranges): Likewise.
16182 (dump_asserts_for): Likewise.
16183 (register_edge_assert_for_2): Likewise.
16184 (vrp_visit_cond_stmt): Likewise.
16185 (vrp_visit_switch_stmt): Likewise.
16186 (vrp_visit_stmt): Likewise.
16187 (vrp_visit_phi_node): Likewise.
16188 (simplify_cond_using_ranges_1): Likewise.
16189 (fold_predicate_in): Likewise.
16190 (evrp_dom_walker::before_dom_children): Likewise.
16191 (evrp_dom_walker::push_value_range): Likewise.
16192 (evrp_dom_walker::pop_value_range): Likewise.
16193 (execute_early_vrp): Likewise.
16195 2017-05-16 Richard Biener <rguenther@suse.de>
16197 * dwarf2out.c (loc_list_from_tree_1): Do not create
16198 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
16200 2017-05-16 Richard Biener <rguenther@suse.de>
16202 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
16204 (note_variable_value_in_expr): If we resolved the decl ref
16205 do not push to the stack.
16207 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
16209 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
16210 operations in fast-math mode.
16211 (vaddq_f16): Likewise.
16212 (vmul_f16): Likewise.
16213 (vmulq_f16): Likewise.
16214 (vsub_f16): Likewise.
16215 (vsubq_f16): Likewise.
16216 * config/arm/neon.md (add<mode>3): New.
16218 (fma:<VH:mode>3): New. Also remove outdated comment.
16221 2017-05-16 Martin Liska <mliska@suse.cz>
16225 * ipa-devirt.c (warn_types_mismatch): Fix typo.
16226 (odr_types_equivalent_p): Likewise.
16228 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
16230 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
16232 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
16235 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
16236 non-interunit SSE move alternatives with '?'.
16237 (zero-extendsidi peephole2): New peephole to skip intermediate
16238 general register in SSE zero-extend sequence.
16240 2017-05-15 Jeff Law <law@redhat.com>
16242 * reorg.c (relax_delay_slots): Create a new variable to hold
16243 the temporary target rather than clobbering TARGET_LABEL.
16245 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
16246 missing argument to extract_bit_field call.
16247 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
16249 2017-05-15 Martin Liska <mliska@suse.cz>
16252 * gcc.c (process_command): Do not allow empty argument of -o option.
16254 2017-05-15 Renlin Li <renlin.li@arm.com>
16256 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
16257 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
16258 * config/aarch64/constraints.md (Usf): Add long call check.
16259 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
16260 (call_value): Likewise.
16261 (sibcall): Likewise.
16262 (sibcall_value): Likewise.
16264 (call_value_insn): New.
16265 (sibcall_insn): Update rtx pattern.
16266 (sibcall_value_insn): Likewise.
16267 (call_internal): Remove.
16268 (call_value_internal): Likewise.
16269 (sibcall_internal): Likewise.
16270 (sibcall_value_internal): Likewise.
16271 (call_reg): Likewise.
16272 (call_symbol): Likewise.
16273 (call_value_reg): Likewise.
16274 (call_value_symbol): Likewise.
16276 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
16279 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
16281 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
16283 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
16284 compatible with CCGOCmode and with CCZmode.
16286 2017-05-14 Martin Sebor <msebor@redhat.com>
16288 PR middle-end/77671
16289 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
16290 (gimple_fold_builtin_snprintf): Same.
16291 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
16292 (gimple_fold_builtin_snprintf): Same.
16293 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
16294 of character types.
16295 (is_call_safe): New function.
16296 (try_substitute_return_value): Call it.
16297 (try_simplify_call): New function.
16298 (pass_sprintf_length::handle_gimple_call): Call it.
16300 2017-05-14 Martin Sebor <msebor@redhat.com>
16302 PR middle-end/80669
16303 * builtins.c (expand_builtin_stpncpy): Simplify.
16305 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
16307 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
16308 * config/i386/i386.h
16309 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
16310 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
16311 (struct machine_function): Add new members call_ms2sysv,
16312 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
16313 (struct machine_frame_state): New fields sp_realigned and
16314 sp_realigned_offset.
16315 * config/i386/i386.c
16316 (enum xlogue_stub): New enum.
16317 (enum xlogue_stub_sets): New enum.
16318 (class xlogue_layout): New class.
16319 (struct ix86_frame): New fields stack_realign_allocate_offset,
16320 stack_realign_offset and outlined_save_offset. Modify comments to
16321 detail stack layout when using out-of-line stubs.
16322 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
16323 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
16324 -mcall-ms2sysv-xlogues.
16325 (stub_managed_regs): New static variable.
16326 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
16327 registers managed by out-of-line stub.
16328 (disable_call_ms2sysv_xlogues): New function.
16329 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
16330 m->call_ms2sysv when appropriate and compute frame layout for
16332 (sp_valid_at, fp_valid_at): New inline functions.
16333 (choose_basereg): New function.
16334 (choose_baseaddr): Add align parameter, use choose_basereg and modify
16336 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
16337 Use align parameter of choose_baseaddr to generated aligned SSE movs
16339 (pro_epilogue_adjust_stack): Modify to track
16340 machine_frame_state::sp_realigned.
16341 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
16342 (ix86_nsaved_sseregs): Likewise.
16343 (ix86_emit_save_regs): Likewise.
16344 (ix86_emit_save_regs_using_mov): Likewise.
16345 (ix86_emit_save_sse_regs_using_mov): Likewise.
16346 (get_scratch_register_on_entry): Likewise.
16347 (gen_frame_set): New function.
16348 (gen_frame_load): Likewise.
16349 (gen_frame_store): Likewise.
16350 (emit_outlined_ms2sysv_save): Likewise.
16351 (emit_outlined_ms2sysv_restore): Likewise.
16352 (ix86_expand_prologue): Modify stack re-alignment code and call
16353 emit_outlined_ms2sysv_save when appropriate.
16354 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
16355 parameter rtx_insn *insn, which allows the function to be used to only
16356 generate the notes.
16357 (ix86_expand_epilogue): Modify validity checks of frame and stack
16358 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
16359 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
16360 * config/i386/predicates.md
16361 (save_multiple): New predicate.
16362 (restore_multiple): Likewise.
16363 * config/i386/sse.md
16364 (save_multiple<mode>): New pattern.
16365 (save_multiple_realign<mode>): Likewise.
16366 (restore_multiple<mode>): Likewise.
16367 (restore_multiple_and_return<mode>): Likewise.
16368 (restore_multiple_leave_return<mode>): Likewise.
16369 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
16371 2017-05-14 Julia Koval <julia.koval@intel.com>
16373 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
16374 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
16375 (__builtin_ia32_xsetbv): New builtins.
16376 * config/i386/i386.c (ix86_expand_special_args_builtin):
16378 (ix86_expand_builtin): Special expand for new intrinsics.
16379 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
16380 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
16381 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
16383 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16385 * cfganal.c (inverted_post_order_compute): Change argument type
16387 * cfganal.h (inverted_post_order_compute): Adjust prototype.
16388 * df-core.c (rest_of_handle_df_initialize): Adjust.
16389 (rest_of_handle_df_finish): Likewise.
16390 (df_analyze_1): Likewise.
16391 (df_analyze): Likewise.
16392 (loop_inverted_post_order_compute): Change argument to be a vec *.
16393 (df_analyze_loop): Adjust.
16394 (df_get_n_blocks): Likewise.
16395 (df_get_postorder): Likewise.
16396 * df.h (struct df_d): Change field to be a vec.
16397 * lcm.c (compute_laterin): Adjust.
16398 (compute_available): Likewise.
16399 * lra-lives.c (lra_create_live_ranges_1): Likewise.
16400 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
16401 * tree-ssa-pre.c (compute_antic): Likewise.
16403 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16405 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16406 (depth_first_search::depth_first_search): Change structure init
16407 function to this constructor.
16408 (depth_first_search::add_bb): Rename function to this member.
16409 (depth_first_search::execute): Likewise.
16410 (flow_dfs_compute_reverse_finish): Adjust.
16412 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16414 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16415 (longest_simple_path): Likewise.
16416 * shrink-wrap.c (spread_components): Likewise.
16417 (disqualify_problematic_components): Likewise.
16418 (emit_common_heads_for_components): Likewise.
16419 (emit_common_tails_for_components): Likewise.
16420 (insert_prologue_epilogue_for_components): Likewise.
16422 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16424 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16427 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16429 * df-core.c (df_set_blocks): Start using auto_bitmap.
16430 (df_compact_blocks): Likewise.
16431 * df-problems.c (df_rd_confluence_n): Likewise.
16432 * df-scan.c (df_insn_rescan_all): Likewise.
16433 (df_process_deferred_rescans): Likewise.
16434 (df_update_entry_block_defs): Likewise.
16435 (df_update_exit_block_uses): Likewise.
16436 (df_entry_block_bitmap_verify): Likewise.
16437 (df_exit_block_bitmap_verify): Likewise.
16438 (df_scan_verify): Likewise.
16439 * lra-constraints.c (lra_constraints): Likewise.
16440 (undo_optional_reloads): Likewise.
16441 (lra_undo_inheritance): Likewise.
16442 * lra-remat.c (calculate_gen_cands): Likewise.
16443 (do_remat): Likewise.
16444 * lra-spills.c (assign_spill_hard_regs): Likewise.
16445 (spill_pseudos): Likewise.
16446 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16447 (bitmap_set_subtract_values): Likewise.
16449 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16451 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16452 management with auto_bitmap.
16453 (fix_inter_tick): Likewise.
16454 (fix_recovery_deps): Likewise.
16455 * ira.c (add_store_equivs): Likewise.
16456 (find_moveable_pseudos): Likewise.
16457 (split_live_ranges_for_shrink_wrap): Likewise.
16458 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16459 (rtx_reuse_manager::seen_def_p): Likewise.
16460 (rtx_reuse_manager::set_seen_def): Likewise.
16461 * print-rtl.h (class rtx_reuse_manager): Likewise.
16463 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16465 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16467 (migrate_btr_def): Likewise.
16468 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16469 * df-core.c (loop_post_order_compute): Likewise.
16470 (loop_inverted_post_order_compute): Likewise.
16471 * hsa-common.h: Likewise.
16472 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16473 * init-regs.c (initialize_uninitialized_regs): Likewise.
16474 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16475 (inline_small_functions): Likewise.
16476 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16477 * ira.c (combine_and_move_insns): Likewise.
16478 (build_insn_chain): Likewise.
16479 * loop-invariant.c (find_invariants): Likewise.
16480 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16481 * predict.c (tree_predict_by_opcode): Likewise.
16482 (predict_paths_leading_to): Likewise.
16483 (predict_paths_leading_to_edge): Likewise.
16484 (estimate_loops_at_level): Likewise.
16485 (estimate_loops): Likewise.
16486 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16487 (spread_components): Likewise.
16488 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16489 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16490 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16491 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16492 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16493 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16494 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16495 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16496 (mark_threaded_blocks): Likewise.
16497 (thread_through_all_blocks): Likewise.
16498 * tree-ssa.c (verify_ssa): Likewise.
16499 (execute_update_addresses_taken): Likewise.
16500 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16502 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16504 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16506 (post_order_compute): Likewise.
16507 (inverted_post_order_compute): Likewise.
16508 (pre_and_rev_post_order_compute_fn): Likewise.
16510 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16512 * genrecog.c (int_set::int_set): Explicitly construct our
16513 auto_vec base class.
16514 * vec.h (auto_vec::auto_vec): New constructor.
16516 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16518 * bitmap.h (class auto_bitmap): New constructor taking
16519 bitmap_obstack * argument.
16521 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16523 * bitmap.h (class auto_bitmap): Change type of m_bits to
16524 bitmap_head, and adjust ctor / dtor and member operators.
16526 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16528 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16529 when returned register mode doesn't match original mode.
16531 2017-05-12 Jeff Law <law@redhat.com>
16532 Jakub Jelinek <jakub@redhat.com>
16534 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16535 we look for cc setter after the compare-elim changes.
16536 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16537 within the vector to match what compare-elim now expects.
16538 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16539 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16541 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16542 after the compare-elim changes.
16543 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16544 the vector to match what compare-elim now expects.
16545 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16546 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16547 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16548 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16549 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16551 * config/visium/visium.c (single_set_and_flags): Fix where
16552 we look for cc setter after the compare-elim changes.
16553 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16554 with the vector to match what compare-elim now expects.
16555 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16556 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16557 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16558 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16559 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16560 (neg<mode>2_insn_set_overflow): Likewise.
16562 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16564 PR middle-end/79794
16565 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16566 maybe_expand_insn call, set ops[0].target. If still set after call,
16567 set alt_rtl. Add extra arg to recursive calls.
16568 (extract_bit_field): Add alt_rtl argument. Pass to
16570 * expmed.h (extract_bit_field): Fix prototype.
16571 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16572 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16573 to extract_bit_field_calls.
16574 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16575 Pass alt_rtl to extract_bit_field calls.
16576 * calls.c (store_unaligned_arguments_into_psuedos)
16577 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16578 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16580 * optabs.h (struct expand_operand): Add target bitfield.
16582 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16584 * compare-elim.c (try_eliminate_compare): Canonicalize
16585 operation with embedded compare to
16586 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16587 (set (reg) (operation)].
16589 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16591 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16594 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16595 cost of adding a carry flag for ADC instruction.
16596 [case MINUS]: Ignore the cost of subtracting a carry flag
16597 for SBB instruction.
16599 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16601 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16603 * config/rs6000/bmiintrin.h: New file.
16604 * config/rs6000/bmi2intrin.h: New file.
16605 * config/rs6000/x86intrin.h: New file.
16607 2017-05-12 Jeff Law <law@redhat.com>
16609 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16612 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16614 PR middle-end/80707
16615 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16617 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16619 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16620 early expansion of vector divide builtins.
16621 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16622 builtins identified as having unsigned arguments.
16624 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16626 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16627 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16628 expansion of vector logical operations (and, andc, or, xor,
16631 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16633 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16634 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16636 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16638 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16639 early GIMPLE expansion of vector multiplies.
16641 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16643 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
16644 TARGET_HAVE_MOVT conditional.
16645 (movt splitter): Likewise.
16647 2017-05-12 Richard Biener <rguenther@suse.de>
16649 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
16650 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16651 Fold all stmts not inplace.
16653 2017-05-12 Richard Biener <rguenther@suse.de>
16655 PR tree-optimization/80713
16656 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
16657 inserted_exprs bit for not removed stmts.
16659 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16661 PR middle-end/69921
16662 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16663 parallelized" attribute for parallelized OpenACC kernels.
16664 * omp-offload.c (execute_oacc_device_lower): Use it.
16666 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16667 Set "oacc kernels" attribute.
16668 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16669 parameter. Adjust all users.
16670 (oacc_fn_attrib_kernels_p): Remove function.
16671 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16672 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16673 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16674 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16675 assert "oacc kernels" attribute is set.
16677 2017-05-11 Carl Love <cel@us.ibm.com>
16679 * config/rs6000/rs6000-c: Add support for built-in functions
16680 vector unsigned char vec_popcnt (vector signed char)
16681 vector unsigned char vec_popcnt (vector unsigned char)
16682 vector unsigned short vec_popcnt (vector signed short)
16683 vector unsigned short vec_popcnt (vector unsigned short)
16684 vector unsigned int vec_popcnt (vector signed int)
16685 vector unsigned int vec_popcnt (vector unsigned int)
16686 vector unsigned long long vec_popcnt (vector signed long long)
16687 vector unsigned long long vec_popcnt (vector unsigned long long)
16688 vector signed long long vec_slo (vector signed long long,
16689 vector signed char)
16690 vector signed long long vec_slo (vector signed long long,
16691 vector unsigned char)
16692 vector unsigned long long vec_slo (vector unsigned long long,
16693 vector signed char)
16694 vector unsigned long long vec_slo (vector unsigned long long,
16695 vector unsigned char)
16696 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16697 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16698 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16699 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16700 * doc/extend.texi: Update the built-in documentation file for the
16701 new built-in functions.
16703 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16705 * attribs.h (sorted_attr_string): Move machine independent
16706 functions for target clone support from the i386 port to common
16707 code. Rename ix86_function_versions to common_function_versions.
16708 Rename make_name to make_unique_name.
16709 (common_function_versions): Likewise.
16710 (make_unique_name): Likewise.
16711 (make_dispatcher_decl): Likewise.
16712 (is_function_default_version): Likewise.
16713 * attribs.c (attr_strcmp): Likewise.
16714 (sorted_attr_string): Likewise.
16715 (common_function_versions): Likewise.
16716 (make_unique_name): Likewise.
16717 (make_dispatcher_decl): Likewise.
16718 (is_function_default_version): Likewise.
16719 * config/i386/i386.c (attr_strcmp): Likewise.
16720 (sorted_attr_string): Likewise.
16721 (ix86_function_versions): Likewise.
16722 (make_name): Likewise.
16723 (make_dispatcher_decl): Likewise.
16724 (is_function_default_version): Likewise.
16725 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16727 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16730 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16731 Account for direct move costs for vec_construct of integer
16734 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16737 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16738 (UNSPEC_STX_ATOMIC): Ditto.
16739 (loaddi_via_sse): New insn.
16740 (storedi_via_sse): Ditto.
16741 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16742 Update corresponding peephole2 patterns.
16743 (atomic_storedi_fpu): Ditto.
16745 2017-05-11 Julia Koval <julia.koval@intel.com>
16747 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
16748 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
16750 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
16751 (__builtin_ia32_rsqrt14ss_mask): New builtins.
16752 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
16754 2017-05-11 Nathan Sidwell <nathan@acm.org>
16756 * graphite-poly.c: Include dumpfile.h.
16758 * dumpfle.h (dump_function): Declare here ...
16759 * tree-dump.h (dump_function): ... not here.
16760 * dumpfile.c: #include tree-cfg.h.
16761 (dump_function): Move here from ...
16762 * tree-dump.c (dump_function): ... here.
16763 * gimplify.c: #include splay-tree.h, not tree-dump.h.
16764 * graphite-poly.c: Don't include tree-dump.h.
16765 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
16766 * print-tree.c: Likewise.
16767 * stor-layout.c: Likewise.
16768 * tree-nested.c: Likewise.
16770 * dumpfile.c (dump_start): Use TDF_FLAGS.
16771 (dump_enable_all): Fix TDF_KIND check thinko.
16773 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16775 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16776 array entries to represent two legal parameterizations of the
16777 overloaded __builtin_cmpb function, as represented by the
16778 P6_OV_BUILTIN_CMPB constant.
16779 (altivec_resolve_overloaded_builtin): Add special case handling
16780 for the __builtin_cmpb function, as represented by the
16781 P6_OV_BUILTIN_CMPB constant.
16782 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
16783 (BU_P6_64BIT_2): New macro.
16784 (BU_P6_OVERLOAD_2): New macro
16785 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
16786 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
16787 (CMPB): Add overload support to represent both 32-bit and 64-bit
16788 compare-bytes function.
16789 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
16790 support for TARGET_CMPB.
16791 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
16792 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16793 documentation of the __builtin_cmpb overloaded built-in function.
16795 2017-05-11 Richard Biener <rguenther@suse.de>
16797 PR tree-optimization/80705
16798 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
16799 bases are not vectorizable.
16801 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16803 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
16804 when counting register pressure.
16806 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16808 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
16809 (iv_ca_more_deps): Renamed to ...
16810 (iv_ca_compare_deps): ... this.
16811 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
16813 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16815 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
16817 (determine_group_iv_costs): ... here.
16818 (find_inv_vars_cb): Record inv var if it's not recorded before.
16820 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16822 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
16823 (get_shiftadd_cost): Ditto.
16825 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16827 * tree-ssa-address.c: Include header file.
16828 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
16830 (add_to_parts): Refactor.
16831 (addr_to_parts): New parameter. Update use of move_hint_to_base.
16832 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
16835 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16837 PR tree-optimization/53090
16838 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
16840 (extract_cond_operands): Detect condition with IV on both sides
16841 and return COMP_IV_EXPR_2.
16842 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
16843 (rewrite_use_compare): Simplify by removing call to function
16844 extract_cond_operands.
16846 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16848 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
16849 (extract_cond_operands): Detect condition comparing against non-
16850 invariant bound and return appropriate enum value.
16851 (find_interesting_uses_cond): Update use of extract_cond_operands.
16852 Handle its return value accordingly.
16853 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
16855 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16857 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
16858 nonlinear iv_use computation in loop invariant sensitive way.
16860 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16862 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
16863 (find_iv_candidates): Call relate_compare_use_with_all_cands.
16865 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16867 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
16868 (dump_cand): Support iv_cand.inv_exprs.
16869 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
16871 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
16874 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16876 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
16878 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
16879 as local function. Include necessary header files.
16880 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
16882 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16884 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
16886 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16888 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
16889 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
16890 RSHIFT_EXPR and BIT_NOT_EXPR.
16892 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16894 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
16895 (adjust_setup_cost): New parameter supporting round up adjustment.
16896 (struct address_cost_data): Delete.
16897 (force_expr_to_var_cost): Don't bound cost with spill_cost.
16898 (split_address_cost, ptr_difference_cost): Delete.
16899 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
16900 (struct ainc_cost_data): New struct.
16901 (get_address_cost_ainc): New function.
16902 (get_address_cost, get_computation_cost): Reimplement.
16903 (determine_group_iv_cost_address): Record inv_expr for all uses of
16905 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
16906 (iv_ca_has_deps): Reimplemented to ...
16907 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
16909 (iv_ca_extend): Call iv_ca_more_deps.
16911 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16913 * tree-ssa-address.c (struct mem_address): Move to header file.
16914 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
16915 * tree-ssa-address.h (struct mem_address): Move from C file.
16916 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
16918 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16920 * tree-affine.h (aff_combination_type): New interface.
16921 (aff_combination_zero_p): Remove static.
16922 (aff_combination_const_p): New interface.
16923 (aff_combination_singleton_var_p): New interfaces.
16925 2017-05-11 Richard Biener <rguenther@suse.de>
16927 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16928 Skip unreachable blocks and destinations.
16929 (eliminate): Move stmt removal and fixup ...
16930 (fini_eliminate): ... here. Skip inserted exprs.
16931 (pass_pre::execute): Move fini_pre after fini_eliminate.
16932 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
16933 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
16934 PRE to get rid of dead code that has invalid SSA form and
16935 split critical edges again.
16937 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16939 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16941 2017-05-11 Richard Biener <rguenther@suse.de>
16943 * passes.c (execute_function_todo): Verify loops if they are
16944 said to be up-to-date.
16945 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
16946 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
16948 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
16951 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
16952 handle calling assemble_external ourself.
16955 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
16956 modes with zero size. Enhance comment.
16958 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16960 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
16961 built-ins for vec_xl and vec_xst with short and char pointer
16964 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16966 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
16967 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
16968 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
16969 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
16970 (_mm_maskz_min_round_ss): New intrinsics.
16971 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
16972 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
16973 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
16974 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
16975 (__builtin_ia32_minss_mask_round): New builtins.
16976 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16977 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16978 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
16980 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
16981 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
16983 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16986 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16988 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16989 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16990 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16991 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16992 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16993 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16994 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16995 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16996 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16997 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16998 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16999 (__builtin_ia32_mulss_mask_round): New builtins.
17000 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
17001 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
17002 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
17004 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
17005 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17007 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17010 2017-05-10 Julia Koval <julia.koval@intel.com>
17012 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
17013 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
17014 (_mm256_setr_m128i): New intrinsics.
17016 2017-05-10 Julia Koval <julia.koval@intel.com>
17018 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
17019 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
17020 (_mm_maskz_rcp14_ss): New intrinsics.
17021 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
17022 (__builtin_ia32_rcp14ss_mask): New builtins.
17023 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
17025 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
17027 PR tree-optimization/51513
17028 * tree-cfg.c (gimple_seq_unreachable_p): New function.
17029 (assert_unreachable_fallthru_edge_p): Use it.
17030 (group_case_labels_stmt): Likewise.
17031 * tree-cfg.h: Prototype it.
17032 * stmt.c: Include cfghooks.h and tree-cfg.h.
17033 (emit_case_dispatch_table) <gap_label>: New local variable.
17034 Use it to fill dispatch table gaps.
17035 Test for default_label before updating probabilities.
17036 (expand_case) <default_label>: Remove unneeded initialization.
17037 Test for unreachable default case statement and remove its edge.
17038 Set default_label accordingly.
17039 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
17041 2017-05-10 Carl Love <cel@us.ibm.com>
17043 * config/rs6000/rs6000-c: Add support for built-in functions
17044 vector signed char vec_neg (vector signed char)
17045 vector signed short int vec_neg (vector short int)
17046 vector signed int vec_neg (vector signed int)
17047 vector signed long long vec_neg (vector signed long long)
17048 vector float vec_neg (vector float)
17049 vector double vec_neg (vector double)
17050 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
17052 * config/rs6000/altivec.h: Add define for vec_neg
17053 * doc/extend.texi: Update the built-in documentation for the
17054 new built-in functions.
17056 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17058 PR tree-optimization/77644
17059 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
17061 2017-05-10 Nathan Sidwell <nathan@acm.org>
17063 * dumpfile.h (TDI_lang_all): New.
17064 (TDF_KIND): New. Renumber others
17065 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
17067 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
17069 (get_dump_file_name): Adjust suffix generation.
17070 (dump_enable_all): Use TDF_KIND.
17071 * doc/invoke.texi (-fdump-lang-all): Document.
17073 * dumpfile.h: Tabify.
17075 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
17078 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
17079 Move member access before delete.
17081 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
17083 * tree-inline.c (expand_call_inline): Split block at stmt
17086 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
17089 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
17090 are now unused after splitting mov{sf,sd}_hardfloat.
17091 (f32_lr2): Likewise.
17092 (f32_lm): Likewise.
17093 (f32_lm2): Likewise.
17094 (f32_li): Likewise.
17095 (f32_li2): Likewise.
17096 (f32_lv): Likewise.
17097 (f32_sr): Likewise.
17098 (f32_sr2): Likewise.
17099 (f32_sm): Likewise.
17100 (f32_sm2): Likewise.
17101 (f32_si): Likewise.
17102 (f32_si2): Likewise.
17103 (f32_sv): Likewise.
17104 (f32_dm): Likewise.
17105 (f32_vsx): Likewise.
17106 (f32_av): Likewise.
17107 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
17108 For movsf, order stores so the VSX stores occur before the GPR
17109 store which encourages the register allocator to use a traditional
17110 FPR instead of a GPR. For movsd, order the stores so that the GPR
17111 store comes before the VSX stores to allow the power6 to work.
17112 This is due to the power6 not having a 32-bit integer store
17113 instruction from a FPR.
17114 (movsf_hardfloat): Likewise.
17115 (movsd_hardfloat): Likewise.
17117 2017-05-09 Martin Sebor <msebor@redhat.com>
17119 PR translation/80280
17120 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
17123 PR translation/80280
17124 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
17125 data member added in r247778.
17126 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
17128 2017-05-09 Nathan Sidwell <nathan@acm.org>
17130 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
17132 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
17135 2017-05-09 Marek Polacek <polacek@redhat.com>
17137 * doc/invoke.texi: Fix typo.
17139 2017-05-09 Richard Biener <rguenther@suse.de>
17141 * tree-vrp.c (vrp_val_is_max): Adjust comment.
17142 (vrp_val_is_min): Likewise.
17143 (set_value_range_to_value): Likewise.
17144 (set_value_range_to_nonnegative): Likewise.
17145 (gimple_assign_nonzero_p): Likewise.
17146 (gimple_stmt_nonzero_p): Likewise.
17147 (vrp_int_const_binop): Likewise. Remove unreachable case.
17148 (adjust_range_with_scev): Adjust comments.
17149 (compare_range_with_value): Likewise.
17150 (extract_range_from_phi_node): Likewise.
17151 (test_for_singularity): Likewise.
17153 2017-05-09 Richard Biener <rguenther@suse.de>
17155 * tree-vrp.c (get_single_symbol): Add assert that we don't
17156 get overflowed constants as invariant part.
17157 (compare_values_warnv): Add comment before the TREE_NO_WARNING
17158 checks. Use wi::cmp instead of recursing for integer constants.
17159 (compare_values): Just ignore whether we assumed undefined
17160 overflow instead of failing the compare.
17161 (extract_range_for_var_from_comparison_expr): Add comment before the
17162 TREE_NO_WARNING sets.
17163 (test_for_singularity): Likewise.
17164 (extract_range_from_comparison): Do not disable optimization
17165 when we assumed undefined overflow.
17166 (extract_range_basic): Remove init of unused var.
17168 2017-05-09 Richard Biener <rguenther@suse.de>
17170 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
17171 (extract_range_from_multiplicative_op_1): Adjust.
17172 (extract_range_from_binary_expr_1): Use int_const_binop.
17174 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17177 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
17178 rs6000_store_data_bypass_p in seven define_bypass directives and
17179 in several comments.
17180 * config/rs6000/rs6000-protos.h: Add prototype for
17181 rs6000_store_data_bypass_p function.
17182 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
17183 function implements slightly different (rs6000-specific) semantics
17184 than store_data_bypass_p, returning false rather than aborting
17185 with assertion error when arguments do not satisfy the
17186 requirements of store data bypass.
17187 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
17188 rs6000_store_data_bypass_p.
17190 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
17192 * config/xtensa/xtensa-protos.h
17193 (xtensa_initial_elimination_offset): New declaration.
17194 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17195 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
17196 macro definition, add case for FRAME_POINTER_REGNUM when
17197 FRAME_GROWS_DOWNWARD.
17198 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
17199 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
17200 xtensa_initial_elimination_offset.
17202 2017-05-08 Nathan Sidwell <nathan@acm.org>
17204 * doc/invoke.texi: Alphabetize -fdump options.
17206 2017-05-08 Martin Sebor <msebor@redhat.com>
17208 PR translation/80280
17209 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
17211 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
17213 * target.def (compute_frame_layout): New optional target hook.
17214 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
17215 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
17216 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
17218 * reload1.c (verify_initial_elim_offsets): Likewise.
17219 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
17220 (use_simple_return_p): Call arm_compute_frame_layout if needed.
17221 (arm_get_frame_offsets): Split up into this ...
17222 (arm_compute_frame_layout): ... and this function.
17224 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
17226 * config/aarch64/constraints.md (Usa): New constraint.
17227 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
17229 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
17231 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
17232 with_multilib_list after it has been checked.
17234 2017-05-08 Richard Biener <rguenther@suse.de>
17236 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
17237 (bitmap_set_subtract_values): Likewise.
17239 2017-05-08 Richard Biener <rguenther@suse.de>
17241 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
17242 (gimple_assign_nonzero): ... this and remove strict_overflow_p
17244 (gimple_stmt_nonzero_warnv_p): Rename to ...
17245 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
17247 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
17248 (extract_range_basic): Adjust, do not disable propagation on
17249 strict overflow sensitive simplification.
17250 (vrp_visit_cond_stmt): Likewise.
17252 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
17254 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
17255 body size unconditionally.
17257 2017-05-07 Jeff Law <law@redhat.com>
17260 2017-05-06 Jeff Law <law@redhat.com>
17261 PR tree-optimization/78496
17262 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17265 PR tree-optimization/78496
17266 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17267 (simplify_stmt_using_ranges): Call it.
17268 (vrp_dom_walker::before_dom_children): Extract equivalences
17269 from an ASSERT_EXPR with an equality comparison against a
17272 2017-05-06 Jeff Law <law@redhat.com>
17274 PR tree-optimization/78496
17275 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17278 PR tree-optimization/78496
17279 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17280 (simplify_stmt_using_ranges): Call it.
17281 (vrp_dom_walker::before_dom_children): Extract equivalences
17282 from an ASSERT_EXPR with an equality comparison against a
17285 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17287 * lra-constraints.c (lra_copy_reg_equiv): New function.
17288 (split_reg): Use it to copy equivalence information from the
17289 original register to the spill register.
17291 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17293 PR rtl-optimization/75964
17294 * simplify-rtx.c (simplify_const_relational_operation): Remove
17295 invalid handling of comparisons of integer ABS.
17297 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
17299 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
17300 initialize to zero.
17301 (init_regs): Remove declaration.
17302 (function_arg_advance_32): Initialize error_p as boolean variable.
17304 2017-05-05 Nathan Sidwell <nathan@acm.org>
17306 * store-motion.c (remove_reachable_equiv_notes): Reformat long
17307 lines. Use for (;;).
17309 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17311 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
17312 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
17313 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
17314 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
17315 VF=2 that require versioning.
17317 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17319 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
17322 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17324 * diagnostic.h (diagnostic_override_option_index): Convert from
17325 macro to inline function.
17327 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17329 * diagnostic.c (last_module_changed_p): New function.
17330 (set_last_module): New function.
17331 (diagnostic_report_current_module): Convert macro usage to
17332 the above functions.
17333 * diagnostic.h (diagnostic_context::last_module): Strengthen
17334 from const line_map * to const line_map_ordinary *.
17335 (diagnostic_last_module_changed): Delete macro.
17336 (diagnostic_set_last_module): Delete macro.
17338 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17340 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
17341 with diagnostic_report_diagnostic.
17342 (diagnostic_n_impl_richloc): Likewise.
17343 * diagnostic.h (report_diagnostic): Delete macro.
17344 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
17345 with diagnostic_report_diagnostic.
17346 * substring-locations.c (format_warning_va): Likewise.
17348 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17350 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
17351 save/restor of format_spec. Move option-printing code to...
17352 (print_option_information): ...this new function, and
17353 reimplement by simply printing to the pretty_printer,
17354 rather than appending to the format string.
17356 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17358 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
17359 handling logic into...
17360 (update_effective_level_from_pragmas): ...this new function.
17362 2017-05-04 Andrew Waterman <andrew@sifive.com>
17364 * config/riscv/riscv.opt (mstrict-align): New option.
17365 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
17366 (SLOW_UNALIGNED_ACCESS): Define.
17367 (riscv_slow_unaligned_access): Declare.
17368 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
17370 (riscv_slow_unaligned_access): New variable.
17371 (rocket_tune_info): Set slow_unaligned_access to true.
17372 (optimize_size_tune_info): Set slow_unaligned_access to false.
17373 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
17374 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
17375 (riscv_option_override): Set riscv_slow_unaligned_access.
17376 * doc/invoke.texi: Add -mstrict-align to RISC-V.
17378 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
17380 * config/riscv/riscv.md: Unify indentation.
17382 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17387 * config/rs6000/rs6000.md (u code attribute): Add FIX and
17389 (extendsi<mode>2): Add support for doing sign extension via
17390 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
17391 don't have ISA 3.0 instructions.
17392 (extendsi<mode>2 splitter): Likewise.
17393 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
17394 generate the normal insns since SImode can now go in vector
17395 registers. Disallow the special UNSPECs needed for previous
17396 machines to hide SImode being used. Add new insns
17397 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
17398 (fix_trunc<mode>si2_stfiwx): Likewise.
17399 (fix_trunc<mode>si2_internal): Likewise.
17400 (fixuns_trunc<mode>si2): Likewise.
17401 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17402 (fctiw<u>z_<mode>_smallint): Likewise.
17403 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17404 of floating point to 32-bit integer from doing a direct move to
17405 the GPR registers to do a store.
17406 (fctiwz_<mode>): Break long line.
17408 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17410 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17411 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17412 (addr_list, addr_offset_valid_p): New.
17413 (split_address_groups): Check offset validity with above function.
17414 (gt-tree-ssa-loop-ivopts.h): Include header file.
17416 2017-05-05 Nathan Sidwell <nathan@acm.org>
17418 * config.gcc (arm*-*-*): Add missing 'fi'.
17420 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17422 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17423 included in -fopt-info does not matter.
17424 * doc/optinfo.texi (-fopt-info): Fix description of default
17425 behavour. Explicitly say order of options included in -fopt-info
17428 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17430 * config.gcc: Allow combinations of aprofile and rmprofile values for
17431 --with-multilib-list.
17432 * config/arm/t-multilib: New file.
17433 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17434 variables. Remove setting of ISA and floating-point ABI in
17435 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17436 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17437 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17438 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17440 * config/arm/t-rmprofile: Likewise except for the matches changes.
17441 * doc/install.texi (--with-multilib-list): Document the combination of
17442 aprofile and rmprofile values and warn about pitfalls in doing that.
17444 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17446 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17447 (movdi_aarch64): Likewise.
17449 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17451 PR tree-optimization/80632
17452 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17454 (build_arrays): Initialize it for virtual phis.
17455 (fix_phi_nodes): Use it for virtual phis.
17457 PR tree-optimization/80558
17458 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17459 [x, y] op z into [x op, y op z] for op & or | if conditions
17462 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17463 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17466 * config/arm/arm.md (use_literal_pool): Remove.
17467 (64-bit immediate split): No longer takes cost into consideration
17468 if arm_disable_literal_pool is enabled.
17469 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17470 used when arm_disable_literal_pool is enabled.
17471 (arm_max_const_double_inline_cost): Remove use of
17472 arm_disable_literal_pool.
17473 (push_minipool_fix): Add assert.
17474 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17475 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17476 (no_literal_pool_sf_immediate): New.
17478 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17480 PR tree-optimization/80613
17481 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17482 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17484 2017-05-05 Richard Biener <rguenther@suse.de>
17486 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17488 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17490 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17491 of this flag from insn conditions due to removal from r247495.
17493 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17495 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17497 (arm_early_store_addr_dep_ptr): Likewise.
17498 * config/arm/aarch-common-protos.h
17499 (arm_early_load_addr_dep_ptr): Add prototype.
17500 (arm_early_store_addr_dep_ptr): Likewise.
17501 * config/arm/cortex-a53.md: Add new bypasses.
17503 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17505 * tree.c (next_type_uid): Change type to unsigned.
17506 (type_hash_canon): Decrement back next_type_uid if
17507 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17508 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17511 2017-05-04 Martin Sebor <msebor@redhat.com>
17513 * builtins.c: Fix a trivial typo in a comment.
17515 PR middle-end/79234
17516 * builtins.c (check_sizes): Adjust to handle reading past the end.
17517 Avoid printing excessive upper bound of ranges. Use %E to print
17518 tree nodes instead of converting them to %wu.
17519 (expand_builtin_memchr): New function.
17520 (compute_dest_size): Rename...
17521 (compute_objsize): ...to this.
17522 (expand_builtin_memcpy): Adjust.
17523 (expand_builtin_mempcpy): Adjust.
17524 (expand_builtin_strcat): Adjust.
17525 (expand_builtin_strcpy): Adjust.
17526 (check_strncat_sizes): Adjust.
17527 (expand_builtin_strncat): Adjust.
17528 (expand_builtin_strncpy): Adjust and simplify.
17529 (expand_builtin_memset): Adjust.
17530 (expand_builtin_bzero): Adjust.
17531 (expand_builtin_memcmp): Adjust.
17532 (expand_builtin): Handle memcmp.
17533 (maybe_emit_chk_warning): Check strncat just once.
17535 2017-05-04 Martin Sebor <msebor@redhat.com>
17537 PR preprocessor/79214
17538 PR middle-end/79222
17539 PR middle-end/79223
17540 * builtins.c (check_sizes): Add inlining context and issue
17541 warnings even when -Wno-system-headers is set.
17542 (check_strncat_sizes): Same.
17543 (expand_builtin_strncat): Same.
17544 (expand_builtin_memmove): New function.
17545 (expand_builtin_stpncpy): Same.
17546 (expand_builtin): Handle memmove and stpncpy.
17548 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17550 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17551 which is not used any more.
17553 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17555 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17557 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17559 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17560 (cortexa53_tunings): Likewise.
17561 (cortexa57_tunings): Likewise.
17562 (cortexa72_tunings): Likewise.
17563 (cortexa73_tunings): Likewise.
17565 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17567 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17568 Set loop alignment to 8.
17570 2017-05-04 Martin Sebor <msebor@redhat.com>
17572 PR translation/80280
17573 * builtins.c (expand_builtin_object_size): Add missing quoting to
17574 %D and like directives.
17575 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17576 (hsa_type_for_tree_type): Same.
17577 (verify_function_arguments): Same.
17578 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17579 * varasm.c (get_section): Same.
17582 2017-05-04 Martin Sebor <msebor@redhat.com>
17584 PR translation/80280
17585 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17587 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17589 * config/aarch64/aarch64.c (generic_addrcost_table):
17590 Change HI/TI mode setting.
17592 2017-05-04 Martin Jambor <mjambor@suse.cz>
17594 PR tree-optimization/80622
17595 * tree-sra.c (comes_initialized_p): New function.
17596 (build_accesses_from_assign): Only set write lazily when
17597 comes_initialized_p is false.
17598 (analyze_access_subtree): Use comes_initialized_p.
17599 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17600 instead of testing for PARM_DECL.
17602 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17604 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17605 constraint on operand 0 to allow more general addressing modes.
17606 Adjust output template.
17607 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17609 * config/aarch64/aarch64-protos.h
17610 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17611 * config/aarch64/constraints.md (Dp): New address constraint.
17612 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17615 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17617 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17618 update use of estimate_ipcp_clone_size_and_time.
17619 (estimate_local_effects): Update use of
17620 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17621 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17622 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17623 Return nonspecialized time.
17625 2017-05-04 Richard Biener <rguenther@suse.de>
17627 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17628 for the last VUSE which def dominates the PHI. Directly call
17630 (get_continuation_for_phi_1): Remove.
17632 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17634 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17635 to explain the use of truncating division. Cap the number of
17636 iterations to the maximum given by nb_iterations_upper_bound,
17639 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17641 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17642 * configure: Regenerate.
17643 * config.in: Regenerate.
17644 * config/i386/driver-mingw32.c: new file.
17645 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
17646 * config.host: Link driver-mingw32.o on MinGW host.
17647 * doc/install.texi: Document new --enable-mingw-wildcard configure
17650 2017-05-04 Marek Polacek <polacek@redhat.com>
17652 PR tree-optimization/80612
17653 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
17655 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17656 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17658 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17659 (movt splitter): Likewise.
17660 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17661 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17662 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17663 block for Thumb-1 with MOVT.
17664 (thumb2_legitimate_address_p): Move code block ...
17665 (can_avoid_literal_pool_for_label_p): ... into this new function.
17666 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17668 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17669 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17670 "M-profile targets with the MOVT instruction".
17672 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17674 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17675 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17676 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17678 2017-05-04 Martin Liska <mliska@suse.cz>
17680 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17681 variable cond_code.
17683 2017-05-04 Richard Biener <rguenther@suse.de>
17685 * tree.c (array_at_struct_end_p): Handle arrays at struct
17686 end with flexarrays more conservatively. Refactor and treat
17687 arrays of arrays or aggregates more strict. Fix
17688 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17689 * tree.c (array_at_struct_end_p): Adjust prototype.
17690 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17691 * gimple-fold.c (get_range_strlen): Likewise.
17692 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17694 2017-05-04 Richard Biener <rguenther@suse.de>
17696 PR tree-optimization/31130
17697 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17699 (supports_overflow_infinity): Likewise.
17700 (is_negative_overflow_infinity): Likewise.
17701 (is_positive_overflow_infinity): Likewise.
17702 (is_overflow_infinity): Likewise.
17703 (stmt_overflow_infinity): Likewise.
17704 (overflow_infinity_range_p): Likewise.
17705 (usable_range_p): Remove as always returning true.
17706 (make_overflow_infinity): Remove.
17707 (negative_overflow_infinity): Likewise.
17708 (positive_overflow_infinity): Likewise.
17709 (avoid_overflow_infinity): Likewise.
17710 (set_value_range): Adjust accordingly.
17711 (set_value_range_to_nonnegative): Likewise, remove now unused
17712 overflow_infinity arg.
17713 (vrp_operand_equal_p): Adjust.
17714 (update_value_range): Likewise.
17715 (range_int_cst_singleton_p): Likewise.
17716 (operand_less_p): Likewise.
17717 (compare_values_warnv): Likewise.
17718 (extract_range_for_var_from_comparison_expr): Likewise.
17719 (vrp_int_const_binop): Likewise.
17720 (zero_nonzero_bits_from_vr): Likewise.
17721 (extract_range_from_multiplicative_op_1): Likewise.
17722 (extract_range_from_binary_expr_1): Likewise.
17723 (extract_range_from_unary_expr): Likewise.
17724 (extract_range_from_comparison): Likewise.
17725 (extract_range_basic): Likewise.
17726 (adjust_range_with_scev): Likewise.
17727 (compare_ranges): Likewise.
17728 (compare_range_with_value): Likewise.
17729 (dump_value_range): Likewise.
17730 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17732 (simplify_cond_using_ranges): Adjust.
17734 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17736 * brig-builtins.def: Added a builtin for class_f64.
17737 * builtin-types.def: Added a builtin type needed by class_f64.
17739 2017-05-03 Jason Merrill <jason@redhat.com>
17741 * timevar.def: Add TV_CONSTEXPR.
17743 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17745 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
17747 2017-05-03 Martin Jambor <mjambor@suse.cz>
17749 * ipa-prop.c (ipa_update_after_lto_read): Removed.
17750 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
17751 * ipa-cp.c (ipcp_propagate_stage): Do not call
17752 ipa_update_after_lto_read.
17753 * ipa-inline.c (ipa_inline): Likewise.
17755 2017-05-03 Martin Jambor <mjambor@suse.cz>
17757 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
17758 tag. Added a default constructor and a destructor.
17759 (ipa_edge_args_sum_t): New class;
17760 (ipa_edge_args_sum): Declare.
17761 (ipa_edge_args_vector): Remove declaration.
17762 (IPA_EDGE_REF): Use ipa_edge_args_sum.
17763 (ipa_free_edge_args_substructures): Remove declaration.
17764 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
17765 (ipa_edge_args_info_available_for_edge_p): Likewise.
17766 * ipa-prop.c (ipa_edge_args_vector): Removed.
17767 (edge_removal_hook_holder): Likewise.
17768 (edge_duplication_hook_holder): Likewise.
17769 (ipa_edge_args_sum): New variable.
17770 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
17771 ipa_edge_args_vector.
17772 (ipa_free_edge_args_substructures): Likewise.
17773 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
17774 ipa_edge_args_vector.
17775 (ipa_edge_removal_hook): Turned into method
17776 ipa_edge_args_sum_t::remove.
17777 (ipa_edge_duplication_hook): Turned into method
17778 ipa_edge_args_sum_t::duplicate.
17779 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
17780 registering edge hooks.
17781 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
17782 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
17783 ipa_edge_args_sum instead of ipa_edge_args_vector.
17784 * ipa-profile.c (ipa_profile): Likewise.
17786 2017-05-03 Martin Jambor <mjambor@suse.cz>
17788 * symbol-summary.h (function_summary): New method exists.
17789 (function_summary::symtab_removal): Deallocate through release.
17790 (call_summary): New class.
17791 (gt_ggc_mx): New overload.
17792 (gt_pch_nx): Likewise.
17793 (gt_pch_nx): Likewise.
17795 2017-05-03 Jeff Law <law@redhat.com>
17797 PR tree-optimization/78496
17798 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
17799 from simplify_cond_using_ranges. Split off code to walk
17800 backwards through casts into ...
17801 (simplify_cond_using_ranges_2): New function.
17802 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
17803 (execute_vrp): After identifying jump threads, call
17804 simplify_cond_using_ranges_2.
17806 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
17809 * ipa-inline.h (inline_summary): Add ctor.
17810 (create_ggc): Do not use ggc_cleared_alloc.
17812 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
17813 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17815 * gcc.c (handle_braces): Support escaping in switch matching
17817 * doc/invoke.texi (Spec Files): Document it.
17818 Remove superfluous @code markup in items.
17820 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17822 * diagnostic-show-locus.c (struct column_range): New struct.
17823 (get_affected_columns): New function.
17824 (get_printed_columns): New function.
17825 (struct correction): New struct.
17826 (correction::ensure_capacity): New function.
17827 (correction::ensure_terminated): New function.
17828 (struct line_corrections): New struct.
17829 (line_corrections::~line_corrections): New dtor.
17830 (line_corrections::add_hint): New function.
17831 (layout::print_trailing_fixits): Reimplement in terms of the new
17833 (selftest::test_overlapped_fixit_printing): New function.
17834 (selftest::diagnostic_show_locus_c_tests): Call it.
17836 2017-05-03 Nathan Sidwell <nathan@acm.org>
17838 Canonicalize canonical type hashing
17839 * tree.h (type_hash_canon_hash): Declare.
17840 * tree.c (type_hash_list, attribute_hash_list): Move into
17841 type_hash_canon_hash.
17842 (build_type_attribute_qual_variant): Break out hash code calc into
17843 type_hash_canon_hash.
17844 (type_hash_canon_hash): New. Generic type hash computation.
17845 (build_range_type_1, build_array_type_1, build_function_type,
17846 build_method_type_directly, build_offset_type, build_complex_type,
17847 make_vector_type): Call it.
17849 2017-05-03 Richard Biener <rguenther@suse.de>
17851 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
17852 When all DRs have unknown misaligned do not always peel
17853 when there is a store but apply the same costing model as if
17854 there were only loads.
17856 2017-05-03 Richard Biener <rguenther@suse.de>
17859 PR tree-optimization/80492
17860 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17861 compare_base_decls returning dont-know properly.
17863 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
17865 * config/arm/iterators.md (CCSI): New mode iterator.
17866 (arch): New mode attribute.
17867 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
17868 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
17869 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
17870 code iterator for success result mode.
17871 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
17872 the corresponding new insn generators.
17874 2017-05-03 Bin Cheng <bin.cheng@arm.com>
17877 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17878 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17880 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
17882 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
17883 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
17884 (DDR_A): Wrap DDR argument in brackets.
17885 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
17886 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
17887 (DDR_REVERSED_P): Likewise.
17889 2017-05-03 Jakub Jelinek <jakub@redhat.com>
17891 PR tree-optimization/79472
17892 * tree-switch-conversion.c (struct switch_conv_info): Add
17893 contiguous_range and default_case_nonstandard fields.
17894 (collect_switch_conv_info): Compute contiguous_range and
17895 default_case_nonstandard fields, don't clear final_bb if
17896 contiguous_range and only the default case doesn't have the required
17898 (check_all_empty_except_final): Set default_case_nonstandard instead
17899 of failing if contiguous_range and the default case doesn't have empty
17901 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
17902 and only the default case doesn't have the required constants. Skip
17904 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
17905 if default_case_nonstandard.
17906 (build_constructors): Build constant 1 just once. Assert that default
17907 values aren't inserted in between cases if contiguous_range. Skip
17909 (build_arrays): Skip virtual phis.
17910 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
17911 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
17912 Handle virtual phis.
17913 (gen_inbound_check): Handle default_case_nonstandard case.
17914 (process_switch): Adjust check_final_bb caller. Call
17915 gather_default_values with the first non-default case instead of
17916 default case if default_case_nonstandard.
17918 2017-05-02 Nathan Sidwell <nathan@acm.org>
17920 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
17921 check. Fix formatting.
17923 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
17925 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
17926 errors when comparing specialized and unspecialized times.
17928 2017-05-02 David Malcolm <dmalcolm@redhat.com>
17930 * diagnostic-show-locus.c
17931 (layout::should_print_annotation_line_p): Make private.
17932 (layout::print_annotation_line): Make private.
17933 (layout::annotation_line_showed_range_p): Make private.
17934 (layout::show_ruler): Make private.
17935 (layout::print_source_line): Make private. Pass in line and
17936 line_width, rather than calling location_get_source_line. Drop
17938 (layout::print_leading_fixits): New method.
17939 (layout::print_any_fixits): Rename to...
17940 (layout::print_trailing_fixits): ...this, and make private.
17941 Don't print newline fixits.
17942 (diagnostic_show_locus): Move logic for printing one row into...
17943 (layout::print_line): ...this new function. Move the
17944 location_get_source_line call and error-handling from
17945 print_source_line to here. Call print_leading_fixits, and rename
17946 print_any_fixits to print_trailing_fixits.
17947 (selftest::test_fixit_insert_containing_newline): Update now that
17948 newlines are partially supported.
17949 (selftest::test_fixit_insert_containing_newline_2): New test.
17950 (selftest::test_fixit_replace_containing_newline): Update comments.
17951 (selftest::diagnostic_show_locus_c_tests): Call the new test.
17952 * edit-context.c (class added_line): New class.
17953 (class edited_line): Describe newline handling in comment.
17954 (edited_line::actually_edited_p): New method.
17955 (edited_line::print_content): Delete redundant decl.
17956 (edited_line::m_predecessors): New field.
17957 (edited_file::print_content): Call edited_line::print_content.
17958 (edited_file::print_diff): Update to support newlines.
17959 (edited_file::print_diff_hunk): Likewise.
17960 (edited_file::print_run_of_changed_lines): New function.
17961 (edited_file::print_diff_line): Convert to...
17962 (print_diff_line): ...this.
17963 (edited_file::get_effective_line_count): New function.
17964 (edited_line::edited_line): Initialize new field m_predecessors.
17965 (edited_line::~edited_line): Clean up m_predecessors.
17966 (edited_line::apply_fixit): Handle newlines.
17967 (edited_line::get_effective_line_count): New function.
17968 (edited_line::print_content): New function.
17969 (edited_line::print_diff_lines): New function.
17970 (selftest::test_applying_fixits_insert_containing_newline): New
17972 (selftest::test_applying_fixits_replace_containing_newline): New
17974 (selftest::insert_line): New function.
17975 (selftest::test_applying_fixits_multiple_lines): Add example of
17977 (selftest::edit_context_c_tests): Call the new tests.
17979 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17981 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
17982 parameter cand. Update dump information.
17983 (get_computation_cost): Update uses.
17985 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17987 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17988 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17989 (get_computation_at, rewrite_use_address): Update use of
17990 get_computation_aff.
17992 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17994 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17995 (get_computation): Delete.
17996 (get_computation_cost): Implement like get_computation_cost_at.
17997 Use get_computation_at.
17998 (get_computation_cost_at): Delete.
17999 (rewrite_use_nonlinear_expr): Use get_computation_at.
18000 (rewrite_use_compare, remove_unused_ivs): Ditto.
18002 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18004 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
18006 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18008 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
18009 (ivopts_global_cost_for_size): Rename parameter and update uses.
18010 (iv_ca_recount_cost): Update uses.
18011 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
18012 candidates seperately in n_invs and n_cands.
18013 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
18015 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18017 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
18018 (find_inv_vars_cb): New.
18019 (find_depends): Renamed to ...
18020 (find_inv_vars): ... this.
18021 (add_candidate_1, force_var_cost): Call find_inv_vars.
18022 (split_address_cost, determine_group_iv_cost_cond): Ditto.
18024 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18026 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
18027 inv_vars. Add inv_exprs.
18028 (struct iv_cand): Rename depends_on to inv_vars.
18029 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
18030 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
18031 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
18032 (dump_cand): Dump inv_vars.
18033 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
18034 (record_invariant, find_depends, add_candidate_1): Ditto.
18035 (set_group_iv_cost, force_var_cost): Ditto.
18036 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
18037 (get_computation_cost_at, get_computation_cost): Ditto.
18038 (determine_group_iv_cost_generic): Ditto.
18039 (determine_group_iv_cost_address): Ditto.
18040 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
18041 (determine_group_iv_costs): Ditto.
18042 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
18043 (iv_ca_set_remove_invariants): Renamed to ...
18044 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
18045 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
18046 (iv_ca_set_add_invariants): Renamed to ...
18047 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
18048 (iv_ca_set_cp): Use iv_ca_set_add_invs.
18049 (iv_ca_has_deps): Support inv_vars and inv_exprs.
18050 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
18051 (create_new_ivs): Remove useless dump.
18053 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18055 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
18057 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
18058 (iv_ca_set_no_cp, create_new_iv): Ditto.
18060 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18062 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18064 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18066 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
18067 function tree_check2.
18069 2017-05-02 Martin Liska <mliska@suse.cz>
18071 * doc/gcov.texi: Add missing preposition.
18072 * gcov.c (function_info::function_info): Properly fill up
18073 all member variables.
18075 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18077 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
18079 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18081 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
18083 2017-05-02 Martin Liska <mliska@suse.cz>
18086 * lto-streamer-in.c (lto_read_tree_1): Remove
18087 LTO_STREAMER_DEBUG.
18088 * lto-streamer.c (struct tree_hash_entry): Likewise.
18089 (struct tree_entry_hasher): Likewise.
18090 (tree_entry_hasher::hash): Likewise.
18091 (tree_entry_hasher::equal): Likewise.
18092 (lto_streamer_init): Likewise.
18093 (lto_orig_address_map): Likewise.
18094 (lto_orig_address_get): Likewise.
18095 (lto_orig_address_remove): Likewise.
18096 * lto-streamer.h: Likewise.
18097 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
18098 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
18100 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
18102 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
18103 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
18104 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
18105 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
18106 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
18107 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
18108 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
18109 (mm_maskz_sub_ss): New intrinsics.
18110 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18111 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18112 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
18113 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
18114 (__builtin_ia32_subss_mask_round): New builtins.
18115 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18116 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18117 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
18119 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
18120 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18122 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18125 2017-05-02 Martin Jambor <mjambor@suse.cz>
18127 PR tree-optimization/78687
18128 * tree-sra.c (access): New field parent.
18129 (process_subtree_disqualification): New function.
18130 (disqualify_candidate): Call it.
18131 (build_accesses_from_assign): Reset write flag if creating an
18133 (build_access_subtree): Fill in parent field and also prpagate
18134 down grp_write flag.
18135 (create_artificial_child_access): New parameter set_grp_write, set
18136 grp_write to its value.
18137 (propagate_subaccesses_across_link): Also propagate grp_write flag
18139 (propagate_all_subaccesses): Push the closest parent back to work
18140 queue if add_access_to_work_queue returned true.
18142 2017-05-02 Richard Biener <rguenther@suse.de>
18144 * common.opt (fstrict-overflow): Alias negative to fwrapv.
18145 * doc/invoke.texi (fstrict-overflow): Remove all traces of
18146 -fstrict-overflow documentation.
18147 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
18148 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
18149 flag_strict_overflow.
18150 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
18151 * lto-opts.c (lto_write_options): Do not stream it.
18152 * lto-wrapper.c (merge_and_complain): Do not handle it.
18153 * opts.c (default_options_table): Do not set -fstrict-overflow.
18154 (finish_options): Likewise do not clear it when sanitizing.
18155 * simplify-rtx.c (simplify_const_relational_operation): Do not
18156 test flag_strict_overflow.
18158 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18160 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
18161 using enabled attribute.
18162 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
18163 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
18164 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
18165 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
18166 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
18167 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
18168 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
18169 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
18170 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
18171 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
18173 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18175 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
18177 2017-05-02 Richard Biener <rguenther@suse.de>
18179 PR tree-optimization/80591
18181 2017-04-10 Richard Biener <rguenther@suse.de>
18183 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18186 2017-05-02 Richard Biener <rguenther@suse.de>
18188 PR tree-optimization/80549
18189 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
18190 (cleanup_tree_cfg_noloop): Create forwarders to known loop
18191 headers if they do not have a preheader.
18193 2017-05-02 Martin Liska <mliska@suse.cz>
18196 * common.opt: Fix typo.
18197 * doc/invoke.texi: Likewise.
18199 2017-05-01 Jan Beulich <jbeulich@suse.com>
18201 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
18202 swapping, add (x,x,m,x,n) alternative.
18204 2017-05-01 Nathan Sidwell <nathan@acm.org>
18206 * calls.c (combine_pending_stack_adjustment_and_call): Remove
18207 unnecessary unadjusted_alignment check.
18209 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18212 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
18214 * gimplify.c (gimplify_cilk_detach): New function.
18215 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
18216 * tree-core.h: Document EXPR_CILK_SPAWN.
18217 * tree.h (EXPR_CILK_SPAWN): Define.
18219 2017-05-01 David Malcolm <dmalcolm@redhat.com>
18221 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
18222 to use new fixit_hint representation, using the "replace" logic.
18223 (get_line_span_for_fixit_hint): Likewise.
18224 (layout::print_any_fixits): Likewise.
18225 (selftest::test_one_liner_many_fixits): Rename to...
18226 (selftest::test_one_liner_many_fixits_1): ...this, and update
18227 comment and expected output to reflect that the multiple fix-it
18228 hints are now consolidated into one insertion.
18229 (selftest::test_one_liner_many_fixits_2): New test.
18230 (selftest::test_diagnostic_show_locus_one_liner): Update for
18232 (selftest::test_fixit_consolidation): Update for fix-it API
18234 * diagnostic.c (print_parseable_fixits): Likewise.
18235 * edit-context.c (edited_line::m_line_events): Convert from
18236 auto_vec <line_event *> to auto_vec <line_event>.
18237 (class line_event): Convert from abstract base class to a concrete
18238 class, taking over the role of replace_event.
18239 (class insert_event): Delete.
18240 (class replace_event): Rename to class line_event. Convert to
18242 (edit_context::add_fixits): Reimplement.
18243 (edit_context::apply_insert): Delete.
18244 (edit_context::apply_replace): Rename to...
18245 (edit_context::apply_fixit): ...this. Convert to half-open range.
18246 (edited_file::apply_insert): Delete.
18247 (edited_file::apply_replace): Rename to...
18248 (edited_file::apply_fixit): ...this.
18249 (edited_line::~edited_line): Drop deletion of events.
18250 (edited_line::apply_insert): Delete.
18251 (edited_line::apply_replace): Rename to...
18252 (edited_line::apply_fixit): ...this. Convert to half-open range.
18253 Update for change to type of m_line_events.
18254 * edit-context.h (edit_context::apply_insert): Delete.
18255 (edit_context::apply_replace): Rename to...
18256 (edit_context::apply_fixit): ...this.
18258 2017-05-01 Martin Sebor <msebor@redhat.com>
18260 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
18263 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
18266 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
18267 __get_cpuid_max returns 0.
18268 (__get_cpuid_count): Ditto.
18270 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
18272 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
18273 replacement expression is another instance of one of its arguments.
18275 2017-05-01 Jakub Jelinek <jakub@redhat.com>
18278 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
18279 check for stack push/pop autoinc.
18280 * config/i386/i386.c (ix86_agi_dependent): Return false
18281 if the only reason why modified_in_p returned true is that
18282 addr is SP based and set_insn is a push or pop.
18284 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18286 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
18289 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18292 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
18293 (account_size_time): Use two predicates - exec_pred and
18295 (evaluate_conditions_for_known_args): Compute both clause and
18297 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
18298 (inline_summary_t::duplicate): Update.
18299 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
18301 (compute_inline_parameters): Likewise.
18302 (estimate_edge_size_and_time): Update caluclation of time.
18303 (estimate_node_size_and_time): Compute both time and nonspecialized
18305 (estimate_ipcp_clone_size_and_time): Update.
18306 (inline_merge_summary): Update.
18307 (do_estimate_edge_time): Update.
18308 (do_estimate_edge_size): Update.
18309 (do_estimate_edge_hints): Update.
18310 (inline_read_section, inline_write_summary): Stream both new predicates.
18311 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
18313 (compute_inlined_call_time): Cleanup.
18314 (big_speedup_p): Update.
18315 (edge_badness): Update.
18316 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
18317 (size_time_entry): Replace predicate by exec_predicate and
18318 nonconst_predicate.
18319 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
18320 (estimate_edge_time): Return also nonspec_time.
18321 (reset_edge_growth_cache): Update.
18323 2017-04-29 Jakub Jelinek <jakub@redhat.com>
18325 PR rtl-optimization/80491
18326 * ifcvt.c (noce_process_if_block): When looking for x setter
18327 with missing else_bb, don't check only the insn right before
18328 cond_earliest, but look for the last insn that x is modified in
18329 within the same bb.
18331 PR rtl-optimization/80491
18332 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
18334 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
18336 PR tree-optimization/80487
18337 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
18339 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18341 PR tree-optimization/79697
18342 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
18343 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
18344 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18346 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
18347 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
18349 2017-04-28 Martin Sebor <msebor@redhat.com>
18351 PR tree-optimization/80523
18352 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
18353 (init_target_to_host_charmap, target_to_host, target_strtol10): New
18355 (maybe_warn, format_directive, parse_directive): Use new functions.
18356 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
18358 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
18360 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
18362 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18364 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
18365 target_header_dir): Set correctly.
18366 * configure: Regenerated.
18367 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18368 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18369 instead of SYSTEM_HEADER_DIR.
18371 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
18373 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
18374 (estimate_local_effects): Likewise.
18375 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
18376 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
18377 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
18378 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
18379 do_estimate_edge_time, estimate_edge_time): Likewise.
18380 * ipa-inline-analysis.c (estimate_node_size_and_time,
18381 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
18382 (estimate_time_after_inlining): Remove.
18384 2017-04-28 Martin Liska <mliska@suse.cz>
18386 * doc/gcov.texi: Enhance documentation of gcov.
18388 2017-04-28 Martin Liska <mliska@suse.cz>
18390 * doc/gcov.texi: Sort options in alphabetic order.
18391 * doc/gcov-dump.texi: Likewise.
18392 * doc/gcov-tool.texi: Likewise.
18393 * gcov.c (print_usage): Likewise.
18394 * gcov-dump.c (print_usage): Likewise.
18395 * gcov-tool.c (print_merge_usage_message): Likewise.
18396 (print_rewrite_usage_message): Likewise.
18397 (print_overlap_usage_message): Likewise.
18399 2017-04-28 Martin Liska <mliska@suse.cz>
18401 PR gcov-profile/53915
18402 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18404 2017-04-28 Martin Liska <mliska@suse.cz>
18406 PR gcov-profile/79891
18407 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18408 is marked by compiler as living on a line.
18409 (get_cycles_count): Remove usage of the union.
18410 (output_intermediate_file): Likewise.
18411 (find_source): Fix GNU coding style.
18412 (accumulate_line_counts): Remove old non-all block mode.
18413 (output_lines): Remove usage of the union.
18414 * profile.c (output_location): Include all BBs, even if
18415 belonging to a same line (and file) as a previous BB.
18417 2017-04-28 Martin Liska <mliska@suse.cz>
18419 * gcov.c (process_args): Handle new argument 'w'.
18420 (read_graph_file): Assign ID to BBs.
18421 (output_branch_count): Display BB # if verbose flag is set.
18422 (output_lines): Likewise for arcs.
18423 (print_usage): Add '--verbose' option help.
18424 * doc/gcov.texi: Document --verbose (-w) option.
18426 2017-04-28 Martin Liska <mliska@suse.cz>
18428 * gcov.c (struct block_location_info): New struct.
18429 (process_file): Fill up the new structure.
18430 (read_graph_file): Replace usage of encoding by the newly added
18432 (add_line_counts): Likewise.
18433 (accumulate_line_counts): Remove usage of the union.
18434 (function_info::function_info): New function.
18435 (function_info::~function_info): Likewise.
18436 (process_file): Call delete instead of release_function.
18437 (release_function): Release the function.
18438 (release_structures): Call delete instead of release_function.
18439 (solve_flow_graph): Replace usage of num_blocks.
18440 (find_exception_blocks): Likewise.
18441 (output_lines): Fix GNU coding style.
18443 2017-04-28 Martin Liska <mliska@suse.cz>
18446 * coverage.c (coverage_remove_note_file): New function.
18447 * coverage.h: Declare the function.
18448 * toplev.c (finalize): Clean if an error has been seen.
18450 2017-04-28 Martin Liska <mliska@suse.cz>
18452 PR gcov-profile/80031
18453 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18454 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18455 * gcov.c (read_graph_file): Read just number of blocks.
18456 * profile.c (branch_prob): Do not stream 0 flags per a basic
18459 2017-04-28 Martin Liska <mliska@suse.cz>
18461 * gcov-dump.c (tag_*): Add new argument to declarations.
18462 (dump_gcov_file): Likewise.
18463 (tag_blocks): Add and use new argument depth.
18464 (tag_arcs): Likewise.
18465 (tag_lines): Likewise.
18466 (tag_counters): Likewise.
18467 (tag_summary): Likewise.
18468 (dump_working_sets): Use depth to do a proper indentation.
18470 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18473 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18474 * symtab.c (symtab_node::debug_symtab): Move definition here.
18476 2017-04-28 Richard Biener <rguenther@suse.de>
18478 * lto-streamer.h (LTO_major_version): Bump to 7.
18480 2017-04-28 Richard Biener <rguenther@suse.de>
18482 * tree-vrp.c (assert_info): New struct.
18483 (add_assert_info): New helper.
18484 (register_edge_assert_for_2): Refactor to add asserts to a vector
18486 (register_edge_assert_for_1): Likewise.
18487 (register_edge_assert_for): Likewise.
18488 (finish_register_edge_assert_for): New helper actually registering
18489 asserts where live on edge.
18490 (find_conditional_asserts): Adjust.
18491 (find_switch_asserts): Likewise.
18492 (evrp_dom_walker::try_find_new_range): Generalize.
18493 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18495 2017-04-27 Marek Polacek <polacek@redhat.com>
18498 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18499 arg10 and arg11 to itype.
18501 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18503 * doc/extend.texi (Object Size Checking): Improve grammar.
18505 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18508 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18509 that the logic for permitting reciprocal estimates matches that
18512 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18515 * tree.c (type_cache_hasher::equal): Only compare
18516 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18517 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18518 non-aggregate element types.
18519 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18520 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18522 2017-04-27 Richard Biener <rguenther@suse.de>
18524 PR middle-end/80533
18525 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18526 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18527 keeping a reference to a trailing array.
18529 2017-04-27 Richard Biener <rguenther@suse.de>
18531 PR middle-end/80539
18532 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18533 being in loop-closed SSA form conservatively.
18534 (chrec_fold_multiply_poly_poly): Likewise.
18536 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18538 PR middle-end/79665
18539 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18540 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18542 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18545 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18546 (aarch64_function_arg_alignment): Return unsigned int again, but still
18547 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18548 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18549 Don't emit -Wpsabi note.
18550 (aarch64_function_arg_boundary): Likewise.
18551 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18554 2017-04-26 Nathan Sidwell <nathan@acm.org>
18556 * tree.h (crc32_unsigned_n): Declare.
18557 (crc32_unsigned, crc32_unsigned): Make inline.
18558 * tree.c (crc32_unsigned_bits): Replace with ...
18559 (crc32_unsigned_n): ... this.
18560 (crc32_unsigned, crc32_byte): Remove.
18561 (crc32_string): Remove unnecessary braces.
18563 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18565 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18566 * ipa-inline-analysis.c (MAX_TIME): Remove.
18567 (account_size_time): Use sreal for time.
18568 (dump_inline_summary): Update.
18569 (estimate_function_body_sizes): Update.
18570 (estimate_edge_size_and_time): Update.
18571 (estimate_calls_size_and_time): Update.
18572 (estimate_node_size_and_time): Update.
18573 (inline_merge_summary): Update.
18574 (inline_update_overall_summary): Update.
18575 (estimate_time_after_inlining): Update.
18576 (inline_read_section): Update.
18577 (inline_write_summary): Update.
18578 * ipa-inline.c (compute_uninlined_call_time): Update.
18579 (compute_inlined_call_time): Update.
18580 (recursive_inlining): Update.
18581 (inline_small_functions): Update.
18582 (dump_overall_stats): Update.
18583 * ipa-inline.h: Include sreal.h.
18584 (size_time_entry): Turn time to sreal.
18585 (inline_summary): Turn self_time nad time to sreal.
18587 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18589 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18591 (sreal::stream_out, sreal::stream_in): New.
18592 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18594 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18596 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18599 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18602 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18604 (dimode_scalar_chain::compute_convert_gain): Ditto.
18605 (dimode_scalar_chain::make_vector_copies): Ditto.
18606 (dimode_scalar_chain::convert_reg): Ditto.
18607 (dimode_scalar_chain::convert_insn): Ditto.
18608 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18609 (VI248_AVX512BW_1): New mode iterator.
18610 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18611 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18614 2017-04-25 Martin Sebor <msebor@redhat.com>
18616 PR tree-optimization/80497
18617 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18618 constants are representable in HOST_WIDE_INT.
18619 (parse_directive): Ditto.
18621 2017-04-25 Martin Sebor <msebor@redhat.com>
18624 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18625 (new_zero_array): Adjust signature.
18626 (dom_info::dom_init): Used unsigned rather that size_t.
18627 (dom_info::dom_info): Same.
18629 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18630 Jakub Jelinek <jakub@redhat.com>
18633 * config/arm/arm.c: Include gimple.h.
18634 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18635 returns negative, increment ncrn only if it returned positive.
18636 (arm_needs_doubleword_align): Return int instead of bool,
18637 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18638 members, but if there is any such non-FIELD_DECL
18639 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18640 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18641 returns negative, increment nregs only if it returned positive.
18642 (arm_setup_incoming_varargs): Likewise.
18643 (arm_function_arg_boundary): Emit -Wpsabi note if
18644 arm_needs_doubleword_align returns negative, return
18645 DOUBLEWORD_ALIGNMENT only if it returned positive.
18647 2017-04-25 Marek Polacek <polacek@redhat.com>
18650 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18651 first argument to type.
18653 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
18656 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
18657 type checks to test for compatibility instead of equality.
18659 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18660 Jakub Jelinek <jakub@redhat.com>
18663 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18665 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18666 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18667 the alignment computation, but return their maximum in warn_alignment.
18668 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18669 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18671 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18672 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18675 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18677 * config/arc/simdext.md (dmpyh): Fix typo.
18679 2017-04-25 Richard Biener <rguenther@suse.de>
18681 PR tree-optimization/80492
18682 * alias.c (compare_base_decls): Handle registers with asm
18683 specification conservatively.
18684 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18685 compare_base_decls returning dont-know properly.
18687 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18689 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18690 (legitimate_offset_address_p): New function.
18691 (arc_legitimate_address_p): Use above function.
18693 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18695 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18697 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18699 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18700 ACCH registers whenever they are available.
18702 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18704 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18705 double regs fix when not used.
18707 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18709 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18711 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18712 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18714 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18716 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18717 when emitting short ADD instructions.
18719 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18721 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18723 (cmpsi_cc_c_insn): Likewise.
18724 (cbranchsi4_scratch): Compute proper instruction length using
18725 compact_hreg_operand.
18726 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18728 2017-04-25 Richard Biener <rguenther@suse.de>
18730 PR middle-end/80509
18731 * passes.c (pass_manager::pass_manager): Initialize
18732 m_name_to_pass_map.
18734 2017-04-25 Richard Biener <rguenther@suse.de>
18736 PR tree-optimization/79201
18737 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18739 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18742 * config/s390/vector.md: Split MEM->GPR vector moves for
18743 non-s_operand addresses.
18745 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18748 * config/s390/predicates.md (reload_const_wide_int_operand): New
18750 * config/s390/s390.md ("movti"): Remove d/P alternative.
18751 ("movti_bigconst"): New pattern definition.
18753 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18756 * s390-protos.h (s390_expand_cs_hqi): Removed.
18757 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
18758 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
18759 modes as well as CCZ1mode and CCZmode.
18760 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
18761 signature of s390_emit_compare_and_swap.
18762 (s390_expand_cs_hqi): Likewise, make static.
18763 (s390_expand_cs_tdsi): Generate an explicit compare before trying
18764 compare-and-swap, in some cases.
18765 (s390_expand_cs): Wrapper function.
18766 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
18768 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
18769 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
18770 patterns for small and large integers. Forbid symref memory operands.
18771 Move expander to s390.c. Require cc register.
18772 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
18773 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
18774 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
18775 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
18776 symref memory operands. Remove CC mode and call s390_match_ccmode
18778 ("atomic_exchange<mode>"): Allow and implement all integer modes.
18780 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18782 * config/s390/s390.md (define_peephole2): New peephole to help
18783 combining the load-and-test pattern with volatile memory.
18785 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18787 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
18788 with CCZmode for TARGET_Z196.
18790 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18792 PR rtl-optimization/80501
18793 * combine.c (make_compound_operation_int): Set subreg_code to SET
18794 even for AND with mask of the sign bit of mode.
18796 PR rtl-optimization/80500
18797 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
18798 sum's initial value.
18800 2017-04-25 Julian Brown <julian@codesourcery.com>
18801 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18803 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
18805 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
18807 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
18809 2017-04-25 Julian Brown <julian@codesourcery.com>
18810 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18812 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
18813 (thunderx2t99_sha): New Reservation.
18815 2017-04-25 Julian Brown <julian@codesourcery.com>
18816 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18818 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
18819 type for 1-element load.
18821 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
18823 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
18825 2017-04-24 Martin Jambor <mjambor@suse.cz>
18827 PR tree-optimization/80293
18828 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
18829 char arrays not totally scalarizable if it is false.
18830 (analyze_all_variable_accesses): Pass correct value in the new
18831 parameter. Add a statistics counter.
18833 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
18835 PR middle-end/79931
18836 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
18838 2017-04-24 Richard Biener <rguenther@suse.de>
18840 PR tree-optimization/80494
18841 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
18842 out for complex types.
18844 2017-04-24 Richard Biener <rguenther@suse.de>
18846 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
18847 * tree-ssa-sccvn.c (print_scc): Print SCC size.
18848 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
18849 (DFS): Adjust and never fail.
18850 (sccvn_dom_walker::fail): Remove.
18851 (sccvn_dom_walker::before_dom_children): Adjust.
18852 (run_scc_vn): Likewise and never fail.
18853 * tree-ssa-pre.c (pass_pre::execute): Adjust.
18854 (pass_fre::execute): Likewise.
18856 2017-04-24 Richard Biener <rguenther@suse.de>
18858 PR tree-optimization/79725
18859 * tree-ssa-sink.c (statement_sink_location): Return whether
18860 failure reason was zero uses. Move that check later.
18861 (sink_code_in_bb): Deal with zero uses by removing the stmt
18864 2017-04-24 Richard Biener <rguenther@suse.de>
18867 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
18868 pointer-based references.
18870 2017-04-24 Richard Biener <rguenther@suse.de>
18873 * pass_manager.h (pass_manager::operator new): Remove.
18874 (pass_manager::operator delete): Likewise.
18875 * passes.c (pass_manager::operator new): Remove.
18876 (pass_manager::operator delete): Likewise.
18877 (pass_manager::pass_manager): Zero individual pass members.
18879 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
18882 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
18883 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
18884 Check "XEXP (src, 1)" operand here.
18885 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
18886 Check "XEXP (src, 1)" operand here.
18887 (dimode_scalar_chain::make_vector_copies): Detect count register
18888 of a shift instruction. Zero extend count register from QImode
18889 to DImode to satisfy vector shift pattern count operand predicate.
18890 Substitute vector shift count operand with a DImode copy.
18891 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
18894 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
18896 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
18897 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18898 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
18899 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18900 (UNSPEC_NOREX_MEM): Remove definition.
18902 2017-04-21 Richard Biener <rguenther@suse.de>
18904 PR tree-optimization/79547
18905 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18906 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
18907 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
18908 without any constraints.
18910 2017-04-21 Richard Biener <rguenther@suse.de>
18912 PR tree-optimization/78847
18913 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
18915 2017-04-21 Richard Biener <rguenther@suse.de>
18917 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
18918 (build_distinct_type_copy): Likewise.
18919 (build_variant_type_copy): Likewise.
18920 * tree.c (build_qualified_type): Pass down mem-stat info.
18921 (build_distinct_type_copy): Likewise.
18922 (build_variant_type_copy): Likewise.
18924 2017-04-21 Richard Biener <rguenther@suse.de>
18926 PR tree-optimization/80237
18927 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
18929 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
18930 for a simplified result.
18932 2016-04-21 Richard Biener <rguenther@suse.de>
18934 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
18935 sth as strict as a simple_iv but a chrec without symbols and an
18936 operand defined in the loop we are peeling (and not some subloop).
18937 (propagate_constants_for_unrolling): Propagate all constants.
18939 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18942 * config/i386/i386.c (print_reg): Remove assert for disalowed
18943 regno values, call output_operand_lossage instead.
18945 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18948 * config/i386/constraints.md (Yc): New register constraint.
18949 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
18950 Use Yc constraint for alternative 2 of operand 0. Remove
18951 preferred_for_speed attribute.
18953 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18955 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
18956 lastprivate clauses in SIMT case.
18958 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
18960 * doc/invoke.texi (-Wextra-semi): Document new warning option.
18962 2017-04-20 Richard Biener <rguenther@suse.de>
18964 PR tree-optimization/57796
18965 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
18966 as N scalar stores.
18967 (vect_model_load_cost): Cost gathers as N scalar loads.
18969 2017-04-20 Richard Biener <rguenther@suse.de>
18971 * ggc-page.c (ggc_allocated_p): Rename to ...
18972 (safe_lookup_page_table_entry): ... this and return the lookup
18974 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
18976 2017-04-20 Richard Biener <rguenther@suse.de>
18978 PR tree-optimization/80453
18979 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
18980 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
18981 from the conditions.
18982 (vn_phi_eq): Pass them down.
18983 (vn_phi_lookup): Record them.
18984 (vn_phi_insert): Likewise.
18986 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18988 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18989 uninitialized variable warning to avoid buffer overrun.
18991 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18994 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18995 is suppressed for '{ 0 }' in C.
18997 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18999 * BASE-VER: Set to 8.0.0.
19001 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
19003 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
19004 priority .init_array and .fini_array section with SECTION_NOTYPE
19007 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19009 PR middle-end/80423
19010 * tree.h (build_array_type): Add typeless_storage default argument.
19011 * tree.c (type_cache_hasher::equal): Also compare
19012 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
19013 (build_array_type): Add typeless_storage argument, set
19014 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
19016 (build_nonshared_array_type): Adjust build_array_type_1 caller.
19017 (build_array_type): Likewise. Add typeless_storage argument.
19019 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19020 Jakub Jelinek <jakub@redhat.com>
19022 PR tree-optimization/80426
19023 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
19024 operation on symbolic operands, also compute the overflow for the
19025 invariant part when the operation degenerates into a negation.
19027 2017-04-19 Jakub Jelinek <jakub@redhat.com>
19030 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
19031 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
19034 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
19036 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
19039 * config/avr/avr.c (tree.h): Include it.
19040 (cgraph.h): Include it.
19041 (avr_encode_section_info): Don't warn for uninitialized progmem
19042 variable if it's just an alias.
19044 2017-04-19 Richard Biener <rguenther@suse.de>
19047 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
19048 when needed by AutoPGO.
19050 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
19053 * doc/lto.texi: Remove an extra 'that'.
19055 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
19057 PR rtl-optimization/80429
19058 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
19059 are only used in debug insns.
19061 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19062 Vladimir Makarov <vmakarov@redhat.com>
19064 * config/sparc/predicates.md (input_operand): Add comment. Return
19065 true for any memory operand when LRA is in progress.
19066 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
19068 2017-04-18 Jeff Law <law@redhat.com>
19071 * mips.md ({return,simple_return}_internal): Do not overwrite
19074 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19076 PR tree-optimization/80443
19077 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
19078 instead of adding 1, subtract -1 and similarly instead of subtracting
19081 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
19083 PR rtl-optimization/80357
19084 * haifa-sched.c (tmp_bitmap): New variable.
19085 (model_recompute): Handle duplicate use records.
19086 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
19087 (free_global_sched_pressure_data): Free it.
19089 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
19092 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19093 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19094 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19095 instead of SYSTEM_HEADER_DIR.
19097 2017-04-18 Jeff Law <law@redhat.com>
19099 PR middle-end/80422
19100 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
19101 predecessors after walking up the insn chain.
19103 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19106 * dwarf2out.c (modified_type_die): Try harder not to emit internal
19107 sizetype type into debug info.
19109 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19112 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
19113 unneeded test for TARGET_UPPER_REGS_SF.
19114 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
19116 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19119 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
19120 instead of gsi_after_labels.
19122 2017-04-18 Jeff Law <law@redhat.com>
19124 * regcprop.c (maybe_mode_change): Avoid creating copies of the
19128 2017-04-13 Jeff Law <law@redhat.com>
19129 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19130 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19132 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
19135 * config/avr/avr.c (intl.h): Include it.
19136 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
19138 2017-04-18 Martin Liska <mliska@suse.cz>
19140 PR gcov-profile/78783
19141 * gcov-tool.c (gcov_output_files): Validate that destination
19142 file is either removed by the tool or by a user.
19144 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19145 Guy Benyei <guybe@mellanox.com>
19147 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
19148 block, and do not negate it, the stored id is already negative.
19150 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19152 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
19154 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19157 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
19158 masks of options that should be turned off if the VSX vector
19159 options are turned off.
19160 (OTHER_P8_VECTOR_MASKS): Likewise.
19161 (OTHER_VSX_VECTOR_MASKS): Likewise.
19162 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
19163 rs6000_disable_incompatible_switches to validate no type switches
19165 (rs6000_incompatible_switch): New function to disallow turning on
19166 other vector options if -mno-vsx, -mno-power8-vector, or
19167 -mno-power9-vector are specified.
19169 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19171 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
19173 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19175 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
19176 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
19177 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
19178 (ARG_POINTER_CFA_OFFSET): Likewise.
19180 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19182 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
19183 conditions to take advantage of various optimizations.
19185 2017-04-13 Jeff Law <law@redhat.com>
19187 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19188 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19189 (zero_extendsidi2_dext): Likewise.
19191 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19194 * fold-const.c (fold_ternary_loc): Revert
19195 use op0 instead of fold_convert_loc (loc, type, arg0) part of
19198 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
19200 PR rtl-optimization/80343
19201 * lra-remat.c (update_scratch_ops): Assign original hard reg to
19202 new scratch pseudo.
19204 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
19207 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
19208 to ubsan_encode_value.
19210 2017-04-13 Jeff Law <law@redhat.com>
19212 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
19213 appearing in DEBUG_INSNs.
19215 2017-04-13 Martin Liska <mliska@suse.cz>
19217 PR gcov-profile/80413
19218 * gcov-io.c (gcov_write_string): Copy to buffer just when
19219 allocated size is greater than zero.
19221 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19224 * dwarf2out.c (decls_for_scope): Ignore declarations of
19225 current_function_decl in BLOCK_NONLOCALIZED_VARS.
19227 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
19230 * ipa-visibility.c (non_local_p): Fix typos.
19231 (localize_node): When localizing symbol in same comdat group,
19232 dissolve the group only when we know external symbols are going
19234 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
19236 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19238 PR tree-optimization/79390
19239 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
19240 order does not result in usable sequence, retry with reversed operand
19246 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
19247 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
19248 op0 instead of fold_convert_loc (loc, type, arg0).
19250 2017-04-12 Jeff Law <law@redhat.com>
19252 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
19253 has a delay slot in the generated code.
19255 * config/cris/cris.md (cris_preferred_reload_class): Return
19256 GENNONACR_REGS rather than GENERAL_REGS.
19258 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19261 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
19262 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
19263 signedness of the result type.
19265 2017-04-12 Richard Biener <rguenther@suse.de>
19266 Jeff Law <law@redhat.com>
19268 PR tree-optimization/80359
19269 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
19270 trim stores to TARGET_MEM_REFs.
19272 2017-04-12 Richard Biener <rguenther@suse.de>
19274 PR tree-optimization/79390
19275 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
19276 threading case even more.
19278 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19281 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
19282 for quad_address_p for TImode, instead of just not indexed_address.
19284 2017-04-12 Richard Biener <rguenther@suse.de>
19285 Bernd Edlinger <bernd.edlinger@hotmail.de>
19287 PR middle-end/79671
19288 * alias.c (component_uses_parent_alias_set_from): Handle
19289 TYPE_TYPELESS_STORAGE.
19290 (get_alias_set): Likewise.
19291 * tree-core.h (tree_type_common): Add typeless_storage flag.
19292 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
19293 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
19294 for types containing members with TYPE_TYPELESS_STORAGE.
19295 (place_field): Likewise.
19296 (layout_type): Likewise for ARRAY_TYPE.
19297 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
19298 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
19299 TYPE_TYPELESS_STORAGE.
19300 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
19302 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19305 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19306 first argument to type.
19308 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19312 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
19313 CONST0_RTX (mode) rather than const0_rtx where appropriate.
19314 (rs6000_expand_binop_builtin): Likewise.
19315 (rs6000_expand_ternop_builtin): Likewise; also add missing
19316 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
19317 vshasigma built-ins.
19318 * doc/extend.texi: Document that vec_xxpermdi's third argument
19319 must be a constant.
19321 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
19323 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
19324 Use shift_const cost parameter when calculating gain of STV shifts.
19326 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
19328 PR rtl-optimization/70478
19329 * lra-constraints.c (process_alt_operands): Check memory for
19330 disfavoring memory insn operand.
19332 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19334 PR middle-end/80100
19335 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
19336 left shift in unsigned HOST_WIDE_INT type.
19338 PR rtl-optimization/80385
19339 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
19340 (not (neg X)) into (plus X -1) for complex or non-integral modes.
19343 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
19344 if they have any depend clauses.
19346 2017-04-11 Martin Liska <mliska@suse.cz>
19349 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
19350 * ipa-split.c (split_function): Create a local comdat symbol
19351 if caller is in a comdat group.
19353 2017-04-11 Martin Liska <mliska@suse.cz>
19356 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
19359 2017-04-11 Martin Sebor <msebor@redhat.com>
19361 PR middle-end/80364
19362 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
19363 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
19365 (directive::set_width, directive::set_precision, format_character):
19367 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
19370 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
19373 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
19374 conflict, set target->arch_name instead of target->cpu_name.
19376 2017-04-11 Richard Biener <rguenther@suse.de>
19378 PR tree-optimization/80374
19379 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
19380 build_zero_cst, remove fold_convertible_p check again.
19382 2017-04-11 Martin Liska <mliska@suse.cz>
19385 * ubsan.c (instrument_object_size): Do not instrument register
19388 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19391 * config/i386/i386-builtin-types.def
19392 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
19393 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
19394 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
19395 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
19396 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
19397 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
19398 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
19399 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
19400 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
19401 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19402 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19403 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19404 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19405 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19406 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19407 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19408 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19409 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19410 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19411 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19412 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19413 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19414 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19415 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19416 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19417 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19418 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19419 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19420 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19421 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19422 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19423 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19424 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19425 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19426 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19427 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19428 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19429 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19430 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19431 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19432 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19433 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19434 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19435 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19436 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19437 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19439 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19440 flag to second_arg_count, handle 4 argument function type _COUNT
19441 aliases, handle second_arg_count on second argument rather than last.
19443 2017-04-10 Jeff Law <law@redhat.com>
19445 PR tree-optimization/80374
19446 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19447 record anything if we can not convert integer_zero_node to the
19450 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19453 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19454 Enhance special handling given to the TARGET_P9_MINMAX option in
19455 relation to certain other options.
19457 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19459 PR tree-optimization/80153
19460 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19461 remove POINTER_PLUS_EXPR's base part directly, rather than through
19464 2017-04-10 Richard Biener <rguenther@suse.de>
19465 Bin Cheng <bin.cheng@arm.com>
19467 PR tree-optimization/80153
19468 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19469 the first element of pointer type aff_tree. Build result expr in
19471 (add_elt_to_tree): Convert to type unconditionally. Remove other
19472 fold_convert calls.
19473 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19474 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19476 2017-04-10 Richard Biener <rguenther@suse.de>
19478 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19481 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19483 PR rtl-optimization/70478
19484 * lra-constraints.c (curr_small_class_check): New.
19485 (update_and_check_small_class_inputs): New.
19486 (process_alt_operands): Update curr_small_class_check. Disfavor
19487 alternative insn memory operands. Check available regs for small
19490 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19493 * config/mips/mips.opt (-mvirt): Update description.
19494 * doc/invoke.texi (-mvirt): Likewise.
19496 2017-04-10 Richard Biener <rguenther@suse.de>
19498 PR middle-end/80362
19499 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19500 looking for NEGATE_EXPR in -A / -B to A / B folding.
19502 2017-04-10 Martin Liska <mliska@suse.cz>
19504 PR gcov-profile/80224
19505 * gcov.c (print_usage): Fix usage string.
19506 (get_gcov_intermediate_filename): Remove.
19507 (output_gcov_file): Use both for normal and intermediate format.
19508 (generate_results): Do not initialize special file for
19509 intermediate format.
19511 2017-04-10 Richard Biener <rguenther@suse.de>
19513 PR tree-optimization/80304
19514 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19517 2017-04-10 Nathan Sidwell <nathan@acm.org>
19520 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19521 (rs6000_init_builtins): Use it.
19523 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19525 * config/arm/arm.md (<mrc>): Add mode to SET source.
19526 (<mrrc>): Likewise.
19528 2017-04-10 Richard Biener <rguenther@suse.de>
19530 PR middle-end/80344
19531 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19533 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19536 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19537 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19538 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19539 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19540 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19541 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19542 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19543 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19544 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19545 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19546 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19547 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19548 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19549 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19550 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19551 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19552 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19553 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19554 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19555 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19556 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19557 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19558 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19560 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19562 PR rtl-optimization/70478
19563 * lra-constraints.c: Reverse the last patch.
19565 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19567 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19568 Add comment for WCHAR_T.
19570 2017-04-08 Martin Liska <mliska@suse.cz>
19573 2017-04-07 Martin Liska <mliska@suse.cz>
19576 * ipa-split.c (split_function): Add function part to a same comdat
19579 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19582 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19584 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19586 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19587 (vec_pairing): ...to this.
19588 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19589 (rs6000_sched_init): Adjust for name change.
19590 (struct rs6000_sched_context): Likewise.
19591 (rs6000_init_sched_context): Likewise.
19592 (rs6000_set_sched_context): Likewise.
19594 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19600 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19602 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19603 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19604 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19606 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19608 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19610 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19612 PR rtl-optimization/70703
19613 * ira-color.c (update_conflict_hard_regno_costs): Use
19614 int64_t instead of HOST_WIDE_INT.
19616 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19618 PR rtl-optimization/70478
19619 * lra-constraints.c (process_alt_operands): Disfavor alternative
19620 insn memory operands.
19622 2017-04-07 Jeff Law <law@redhat.com>
19624 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19625 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19627 2017-04-07 Martin Liska <mliska@suse.cz>
19630 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19631 Show error message instead of an ICE.
19633 2017-04-07 Martin Liska <mliska@suse.cz>
19636 * ipa-split.c (split_function): Add function part to a same comdat
19639 2017-04-07 Richard Biener <rguenther@suse.de>
19641 PR middle-end/80341
19642 * tree.c (get_unwidened): Also handle ! for_type case for
19644 * convert.c (do_narrow): Split out from ...
19645 (convert_to_integer_1): ... here. Do not pass final truncation
19646 type to get_unwidened for TRUNC_DIV_EXPR.
19648 2017-04-07 Richard Biener <rguenther@suse.de>
19650 * tree-affine.c (wide_int_ext_for_comb): Take type rather
19652 (aff_combination_const): Adjust.
19653 (aff_combination_scale): Likewise.
19654 (aff_combination_add_elt): Likewise.
19655 (aff_combination_add_cst): Likewise.
19656 (aff_combination_convert): Likewise.
19657 (add_elt_to_tree): Likewise. Remove unused argument.
19658 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19660 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19662 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19664 * config/arm/arm.c (arm_default_short_enums): Use
19665 ARM_DEFAULT_SHORT_ENUMS.
19666 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19668 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19671 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19672 members with redundant out-of-class redeclaration.
19674 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19677 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19678 * config/i386/i386.md (*zero_extendsidi2):
19679 Add (?*x,*x) and (?*v,*v) alternatives.
19681 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19684 * config/i386/i386.c (ix86_expand_builtin)
19685 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19686 mode from insn data. Convert operands to insn operand mode.
19687 Copy operands that don't satisfy insn predicate to a register.
19689 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19691 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19694 2017-04-06 Richard Biener <rguenther@suse.de>
19696 PR tree-optimization/80334
19697 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19698 preserve alignment of accesses.
19700 2017-04-06 Richard Biener <rguenther@suse.de>
19702 PR tree-optimization/80262
19703 * tree-sra.c (build_ref_for_offset): Preserve address-space
19705 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19706 Drop useless address-space information on MEM_REF offsets.
19708 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19710 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19712 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19714 PR rtl-optimization/70703
19715 * ira-color.c (update_conflict_hard_regno_costs): Use
19716 HOST_WIDE_INT instead of long.
19718 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19721 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19722 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19724 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19725 for x86_64 target. Handle -m3dnowa option.
19727 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19729 PR rtl-optimization/70703
19730 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19731 (update_conflict_hard_regno_costs): Use long instead of unsigned
19732 arithmetic for cost calculation.
19734 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19735 Bernd Edlinger <bernd.edlinger@hotmail.de>
19738 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19741 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19744 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
19745 ptr_mode with Pmode throughout.
19746 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
19747 into probe_stack_range and use DImode.
19749 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
19752 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
19753 call_eh_return is true.
19755 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19757 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
19758 Initialize last_match_fntype_index.
19760 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19763 * tree-nvr.c: Include internal-fn.h.
19764 (pass_return_slot::execute): Ignore internal calls without
19767 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19768 Richard Biener <rguenther@suse.de>
19771 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
19772 captures used multiple times, except for the last use.
19773 * generic-match-head.c: Include gimplify.h.
19775 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19777 PR tree-optimization/79390
19778 * target.h (struct noce_if_info): Declare.
19779 * targhooks.h (default_noce_conversion_profitable_p): Declare.
19780 * target.def (noce_conversion_profitable_p): New target hook.
19781 * ifcvt.h (struct noce_if_info): New type, moved from ...
19782 * ifcvt.c (struct noce_if_info): ... here.
19783 (noce_conversion_profitable_p): Renamed to ...
19784 (default_noce_conversion_profitable_p): ... this. No longer
19786 (noce_try_store_flag_constants, noce_try_addcc,
19787 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
19788 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
19789 instead of noce_conversion_profitable_p.
19790 * config/i386/i386.c: Include ifcvt.h.
19791 (ix86_option_override_internal): Don't override
19792 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
19793 (ix86_noce_conversion_profitable_p): New function.
19794 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
19795 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
19796 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
19797 * doc/tm.texi: Regenerated.
19799 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19801 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
19804 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19807 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
19808 instructions for small multiply cores.
19810 2017-04-04 Jeff Law <law@redhat.com>
19812 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
19814 (mips_expand_vec_perm_const): Initialize elements in orig_perm
19815 that are not set by the loop over the elements.
19817 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19820 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
19821 int mode, convert_modes it to mode as unsigned, otherwise use
19822 lowpart_subreg to mode rather than SImode.
19823 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
19824 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
19825 Use DImode instead of SImode for the shift count operand.
19826 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
19829 2017-04-04 Richard Biener <rguenther@suse.de>
19831 PR middle-end/80281
19832 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
19833 arithmetic done for the negate or the plus. Simplify.
19834 (A - (-B) -> A + B): Likewise.
19835 * fold-const.c (split_tree): Make sure to not negate pointers.
19837 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
19839 PR rtl-optimization/60818
19840 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
19841 a compare of comparisons with the thing compared if this results
19842 in a different machine mode.
19844 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
19846 * alias.c (base_alias_check): Fix typo in comment.
19847 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
19848 * cgraphunit.c (symbol_table::compile): Likewise.
19849 * collect2.c (maybe_run_lto_and_relink): Likewise.
19850 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
19851 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
19852 * config/avr/avr.c (avr_map_op_t): Likewise.
19853 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
19854 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
19855 * config/epiphany/epiphany.md (movcc): Likewise.
19856 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
19857 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
19859 * config/mips/mips.c (mips_save_restore_reg): Likewise.
19860 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
19861 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
19862 * config/sh/sh.c (sh_rtx_costs): Likewise.
19863 * fold-const.c (fold_truth_andor): Likewise.
19864 * genautomata.c (collapse_flag): Likewise.
19865 * gengtype.h (struct type::u::s): Likewise.
19866 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
19867 * input.c (FORMAT_AMOUNT): Likewise.
19868 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
19869 (known_aggs_to_agg_replacement_list): Likewise.
19870 * ipa-inline-analysis.c: Likewise.
19871 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
19872 * ipa-polymorphic-call.c
19873 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
19874 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
19875 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
19877 * modulo-sched.c (apply_reg_moves): Likewise.
19878 * omp-expand.c (build_omp_regions_1): Likewise.
19879 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
19880 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
19881 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
19882 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19883 * value-prof.c: Likewise.
19884 * var-tracking.c (val_reset): Likewise.
19886 2017-04-03 Richard Biener <rguenther@suse.de>
19888 PR tree-optimization/80275
19889 * fold-const.c (split_address_to_core_and_offset): Handle
19892 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
19894 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
19895 descriptors is at least equal to that of functions.
19897 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19899 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
19901 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19904 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
19905 (mov<IMOD4:mode>): New expander.
19906 (*mov<IMOD4:mode>_internal): New insn and split pattern.
19908 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19910 PR rtl-optimization/79405
19911 * fwprop.c (propagations_left): New variable.
19912 (forward_propagate_into): Decrement it.
19913 (fwprop_init): Initialize it.
19914 (fw_prop): If the variable has reached zero, stop propagating.
19915 (fwprop_addr): Ditto.
19917 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19920 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
19921 a FUNCTION_DECL, pass it as decl instead of origin to
19924 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
19926 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
19929 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
19932 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
19933 TARGET_VSX_SMALL_INTEGER.
19935 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19937 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19938 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
19940 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19942 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
19943 extraction from odd-numbered MSA register.
19945 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19947 PR middle-end/80173
19948 * expmed.c (store_bit_field_1): Don't attempt to create
19949 a word subreg out of hard registers wider than word if they
19950 have HARD_REGNO_NREGS of 1 for their mode.
19952 PR middle-end/80163
19953 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
19954 conversions to integer types wider than word and pointer.
19957 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
19958 (rtx_equal_for_cselib_p): Pass 0 to it.
19959 * cselib.c (cselib_hasher::equal): Likewise.
19960 (rtx_equal_for_cselib_1): Add depth argument. If depth
19961 is 128, don't look up VALUE locs and punt. Increment
19962 depth in recursive calls when walking VALUE locs.
19964 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
19966 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
19967 (make_gcov_file_name): Use the canonical path name for generating
19969 (read_line): Fix handling of files with ascii null bytes.
19971 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
19973 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
19974 to initialise a vector register instead
19975 of using a const_int.
19977 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19979 PR translation/80189
19980 * gimplify.c (omp_default_clause): Use %qs instead of %s in
19981 diagnostic messages.
19983 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19986 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19987 (dfp_diex_<mode>): Update mode of operand 1.
19988 * doc/extend.texi (dxex, dxexq): Document change to return type.
19989 (diex, diexq): Document change to argument type.
19991 2017-03-30 Martin Jambor <mjambor@suse.cz>
19994 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19995 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19996 it reflects the signature changes performed at the callee side.
19997 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19998 to cgraph_build_function_type_skip_args.
19999 (build_function_decl_skip_args): Adjust call to the above function.
20001 2017-03-30 Jakub Jelinek <jakub@redhat.com>
20004 * config/i386/sse.md
20005 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
20006 register as dest whenever it is a MEM not rtx_equal_p to the
20007 corresponding dup operand, and when forcing into reg move the
20008 reg into the memory afterwards.
20009 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
20010 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
20011 for the force_reg mode.
20012 (avx512vl_vextractf128<mode>): Use register as dest either
20013 always when a MEM, or when it is a MEM not rtx_equal_p to the
20014 corresponding dup operand, or even not when it is a CONST_VECTOR
20015 depending on the mode and lo vs. hi.
20016 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
20018 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
20019 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
20020 Likewise. Require that operands[2] is even.
20021 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
20022 Remove extraneous parens. Require that operands[2] is a multiple
20024 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
20025 operands[0] is a MEM if <mask_applied>, the predicates/constraints
20026 disallow memory then.
20028 2017-03-30 Richard Biener <rguenther@suse.de>
20030 PR tree-optimization/77498
20031 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
20032 to non-constants over backedges.
20034 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
20036 PR rtl-optimization/80233
20037 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
20038 as last_combined_insn. Do not test for BARRIER_P separately.
20040 2017-03-29 Andreas Schwab <schwab@suse.de>
20043 * calls.c (prepare_call_address): Convert funexp to Pmode before
20044 copying to temp reg.
20046 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20048 PR tree-optimization/80158
20049 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
20050 Handle possible future case of more than one alternate
20052 (replace_rhs_if_not_dup): Likewise.
20053 (replace_one_candidate): Likewise.
20055 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
20057 PR rtl-optimization/80193
20058 * ira.c (ira): Do not check allocation for LRA.
20060 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
20062 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
20063 (nvptx_output_simt_exit): Declare.
20064 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
20065 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
20066 (init_softstack_frame): Move initialization of crtl->is_leaf to...
20067 (nvptx_declare_function_name): ...here. Emit declaration of local
20068 memory space buffer for omp_simt_enter insn.
20069 (nvptx_output_unisimt_switch): New.
20070 (nvptx_output_softstack_switch): New.
20071 (nvptx_output_simt_enter): New.
20072 (nvptx_output_simt_exit): New.
20073 * config/nvptx/nvptx.h (struct machine_function): New fields
20074 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
20075 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
20076 (UNSPECV_SIMT_EXIT): Ditto.
20077 (omp_simt_enter_insn): New insn.
20078 (omp_simt_enter): New expansion.
20079 (omp_simt_exit): New insn.
20080 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
20082 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
20083 (expand_GOMP_SIMT_ENTER_ALLOC): New.
20084 (expand_GOMP_SIMT_EXIT): New.
20085 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
20086 (GOMP_SIMT_ENTER_ALLOC): Ditto.
20087 (GOMP_SIMT_EXIT): Ditto.
20088 * target-insns.def (omp_simt_enter): New insn.
20089 (omp_simt_exit): Ditto.
20090 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
20092 (lower_rec_simd_input_clauses): Implement SIMT privatization.
20093 (lower_rec_input_clauses): Likewise.
20094 (lower_lastprivate_clauses): Handle SIMT privatization.
20096 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
20097 (ompdevlow_adjust_simt_enter): New.
20098 (find_simtpriv_var_op): New.
20099 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
20100 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
20102 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
20103 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
20104 (copy_decl_for_dup_finish): Ditto.
20106 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
20108 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
20111 * config/i386/i386.c (ix86_option_override_internal): Always
20112 allow -mpreferred-stack-boundary=3 for 64-bit targets.
20114 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20116 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
20118 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20120 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
20121 mark new edge's irreducible flag accordign to it.
20122 (vect_do_peeling): Check loop preheader edge's irreducible flag
20123 and pass it to function slpeel_add_loop_guard.
20125 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
20127 PR tree-optimization/80218
20128 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
20129 Update block frequencies and counts.
20131 2017-03-28 Richard Biener <rguenther@suse.de>
20133 PR tree-optimization/78644
20134 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
20135 of a simplification result we may not use it at all.
20137 2017-03-28 Richard Biener <rguenther@suse.de>
20140 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
20141 without arguments, generate default definition of a SSA name.
20143 2017-03-28 Richard Biener <rguenther@suse.de>
20145 PR middle-end/80222
20146 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
20147 TYPE_REF_CAN_ALIAS_ALL references.
20148 * fold-const.c (fold_indirect_ref_1): Likewise.
20150 2017-03-28 Martin Liska <mliska@suse.cz>
20153 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
20154 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
20156 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20157 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
20159 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
20160 (EXTRA_SPECS): Define.
20161 (SUBTARGET_EXTRA_SPECS): Likewise.
20162 (SUBTARGET_CPP_SPEC): Likewise.
20163 * config/arc/elf.h (EXTRA_SPECS): Renamed to
20164 SUBTARGET_EXTRA_SPECS.
20165 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
20167 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20169 * config/arc/simdext.md (vst64_insn): Update pattern.
20170 (vld32wh_insn): Likewise.
20171 (vld32wl_insn): Likewise.
20172 (vld64_insn): Likewise.
20173 (vld32_insn): Likewise.
20175 2017-03-28 Marek Polacek <polacek@redhat.com>
20178 * fold-const.c (fold_comparison): Use protected_set_expr_location
20179 instead of SET_EXPR_LOCATION.
20181 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
20183 * tree.c (add_expr): Avoid name lookup warning.
20185 2017-03-27 Jeff Law <law@redhat.com>
20187 PR tree-optimization/80216
20188 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
20189 function name. Limit recursion depth.
20190 (record_temporary_equivalences): Corresponding changes.
20192 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
20194 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
20197 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20200 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
20202 * cfgcleanup.c (reg_note_cfa_p): New array.
20203 (insns_have_identical_cfa_notes): New function.
20204 (old_insns_match_p): Don't cross-jump in between /f
20205 and non-/f instructions. If both i1 and i2 are frame related,
20206 verify all CFA notes, their order and content.
20208 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20211 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
20212 HImode and SImode with zero extend to DImode to one insn.
20213 (bswap<mode>2_extenddi): Likewise.
20214 (bswapsi2_extenddi): Likewise.
20215 (bswaphi2_extendsi): Likewise.
20216 (bswaphi2): Combine bswap HImode and SImode into one insn.
20217 Separate memory insns from swapping register.
20218 (bswapsi2): Likewise.
20219 (bswap<mode>2): Likewise.
20220 (bswaphi2_internal): Delete, no longer used.
20221 (bswapsi2_internal): Likewise.
20222 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
20223 store, and gpr<-gpr swap insns.
20224 (bswap<mode>2_store): Likewise.
20225 (bswaphi2_reg): Register only splitter, combine with the splitter.
20226 (bswaphi2 splitter): Likewise.
20227 (bswapsi2_reg): Likewise.
20228 (bswapsi2 splitter): Likewise.
20229 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
20230 the insns into load, store, and register/register insns.
20231 (bswapdi2_ldbrx): Likewise.
20232 (bswapdi2_load): Likewise.
20233 (bswapdi2_store): Likewise.
20234 (bswapdi2_reg): Likewise.
20236 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
20238 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
20239 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
20241 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
20244 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
20246 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20247 special handling for target option conflicts between dform
20248 options (-mpower9-dform, -mpower9-dform-vector,
20249 -mpower9-dform-scalar) and -mno-direct-move.
20251 2017-03-27 Richard Biener <rguenther@suse.de>
20253 PR tree-optimization/80181
20254 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
20256 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20258 * config/arc/predicates.md (move_double_src_operand): Replace the
20259 call to move_double_src_operand with a call to address_operand.
20261 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20263 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
20264 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
20265 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
20267 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20269 * config/arc/predicates.md (long_immediate_loadstore_operand):
20270 Consider scaled addresses cases.
20272 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20274 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
20275 restored when in interrupt.
20276 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
20277 doesn't have delay slot.
20279 2017-03-27 Richard Biener <rguenther@suse.de>
20282 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
20283 inlined thunk clones.
20285 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20288 * asan.c (instrument_derefs): Copy over last operand from
20289 original COMPONENT_REF to the new COMPONENT_REF with
20290 DECL_BIT_FIELD_REPRESENTATIVE.
20291 * ubsan.c (instrument_object_size): Likewise.
20293 2017-03-27 Richard Biener <rguenther@suse.de>
20295 PR tree-optimization/80170
20296 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
20297 sure DR/SCEV didnt fold in constants we do not see when looking
20298 at the reference base alignment.
20300 2017-03-27 Richard Biener <rguenther@suse.de>
20302 PR middle-end/80171
20303 * gimple-fold.c (fold_ctor_reference): Properly guard against
20304 NULL return value from canonicalize_constructor_val.
20306 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
20309 * config/i386/i386.c (ix86_expand_builtin)
20310 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
20311 flags reg setting and flags reg using instructions.
20312 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
20313 clobbering instructions to zero extend op2.
20315 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
20317 * doc/install.texi (Configuration) <--with-aix-soname>:
20318 Update link to AIX ld.
20320 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
20322 PR rtl-optimization/80160
20323 PR rtl-optimization/80159
20324 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
20325 reg_alternate_class into account.
20327 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
20330 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
20331 to consider in curr_insn_transform.
20333 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20335 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
20336 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
20337 and emit_mode_inner.
20339 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20341 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
20342 argument to the overloaded builtin variants. Use the new flag to
20343 deprecate certain builtin variants.
20344 * config/s390/s390-builtin-types.def: Add new builtin types.
20345 * config/s390/s390-builtins.h: Support new flags field for
20346 overloaded builtins.
20347 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
20348 (s390_macro_to_expand): Enable vector float data type.
20349 (s390_cpu_cpp_builtins_internal): Indicate support of the new
20350 builtins by incrementing the __VEC__ version number.
20351 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
20353 (s390_resolve_overloaded_builtin): Emit error messages depending
20354 on the builtin flags.
20355 * config/s390/s390.c (s390_expand_builtin): Support additional
20356 flags argument. Change error message to match the messages
20357 emitted in s390-c.c.
20358 * config/s390/s390.md: New UNSPEC_* constants.
20359 (op_type): Add new instruction types.
20360 * config/s390/vecintrin.h: Add new builtins and test data class
20362 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
20363 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
20364 (VEC_INEXACT, VEC_NOINEXACT): New constants.
20365 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
20366 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
20367 ("vec_mergel<mode>"): V_HW -> VEC_HW.
20369 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
20370 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
20371 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
20372 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
20374 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
20375 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
20376 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
20377 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
20379 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
20380 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
20381 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
20382 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
20383 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
20384 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
20385 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
20387 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
20388 ("vec_scatter_element<V_HW_4:mode>_DI")
20389 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
20390 ("vec_fpint<mode>", "vflls")
20391 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
20392 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
20393 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
20394 ("*vec_cmphe<mode>_cc"): ... these.
20396 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
20397 mode constant instead of magic value.
20399 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20401 * config/s390/s390.c (s390_expand_vec_compare): Support other
20402 vector floating point modes than just V2DF.
20403 (s390_expand_vcond): Likewise.
20404 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20405 (s390_cannot_change_mode_class): Prevent mode changes between TF
20406 and V1TF in vector registers.
20407 * config/s390/s390.md (DF, SF): New mode attributes.
20408 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20409 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20410 SFmode support for VRs.
20411 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20413 (VFT, VF_HW): New mode iterators.
20414 (vw, sdx): New mode attributes.
20415 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20416 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20417 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20418 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20419 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20420 also the new vector floating point modes. Renaming to ...
20422 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20423 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20424 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20425 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20426 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20427 ("vec_unordered<mode>"): ... these.
20429 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20430 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20431 ("*vec_extendv2df"): New insn definitions.
20433 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20435 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20436 ("mulditi3_2", "*muldi3_sign"): New patterns.
20437 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20438 rename the pattern definition.
20440 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20442 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20444 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20446 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20448 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20449 instruction if possible.
20450 * config/s390/vector.md (vec_halfnumelts): New mode
20452 ("*vec_vllezlf<mode>"): New pattern.
20454 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20456 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20457 ("popcountv4si2", "popcountv2di2"): Rename to ...
20458 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20459 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20461 ("popcount<mode>2_vxe"): New pattern.
20463 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20465 * common/config/s390/s390-common.c (processor_flags_table): Add
20467 * config.gcc: Add arch12.
20468 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20469 Default to arch12 for unknown CPU model numbers.
20470 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20471 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20472 PROCESSOR_max sanity check.
20473 * config/s390/s390-opts.h (enum processor_type): Add
20475 * config/s390/s390.c (processor_table): Add arch12.
20476 (s390_expand_builtin): Add check for B_VXE flag.
20477 (s390_issue_rate): Add PROCESSOR_ARCH12.
20478 (s390_get_sched_attrmask): Likewise.
20479 (s390_get_unit_mask): Likewise.
20480 (s390_sched_score): Enable z13 scheduling for arch12.
20481 (s390_sched_reorder): Likewise.
20482 (s390_sched_variable_issue): Likewise.
20483 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20485 (s390_tune_attr): Use z13 scheduling also for arch12.
20486 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20487 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20488 (TARGET_VXE_P): New macros.
20489 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20490 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20491 * config/s390/s390.opt: Add arch12 as processor_type.
20493 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20495 * config/s390/s390.md
20496 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20497 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20498 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20500 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20501 Rename expanders to ...
20503 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20504 ("fixuns_truncdddi2_emu"): ... these.
20506 ("fixuns_trunc<mode>si2_emu"): New expander.
20508 ("*fixuns_truncdfdi2_z13"): Rename to ...
20509 ("*fixuns_truncdfdi2_vx"): ... this.
20511 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20513 * config/s390/2964.md: Remove the single element vector compare
20514 instructions which are no longer used.
20515 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20517 (s390_canonicalize_comparison): Remove handling of DFmode
20519 (s390_expand_vec_compare_scalar): Remove function.
20520 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20521 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20523 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20525 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20527 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20529 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20530 will anyway by matched by mov<mode>_64dfp.
20532 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20534 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20535 vlef/vstef. Add missing operand to vleif.
20537 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20539 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20540 pair for all vector types with 64 bit elements.
20541 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20542 * config/s390/vector.md (V_HW_64): ... here.
20543 (V_128_NOSINGLE): New mode iterator.
20544 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20545 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20546 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20547 ("*vec_load_pairv2di"): Change to ...
20548 ("*vec_load_pair<mode>"): ... this one.
20550 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20552 * config/s390/constraints.md: Add comments.
20553 (jKK): Reject element sizes > 8 bytes.
20554 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20556 * config/s390/s390.md: Add the s_operand checks formerly in
20557 s390_split_ok_p to various splitters where they are still
20559 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20560 for 128 bit vectors. Plus two splitters.
20562 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20564 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20567 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20570 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20571 error if the boundary argument is not constant.
20573 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20575 PR rtl-optimization/80112
20576 * loop-doloop.c (doloop_condition_get): Don't check condition
20577 if cmp isn't SET with IF_THEN_ELSE src.
20579 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20581 PR tree-optimization/80158
20582 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20583 replacing a candidate statement, also replace it for the
20584 candidate's alternate interpretation.
20585 (replace_rhs_if_not_dup): Likewise.
20586 (replace_one_candidate): Likewise.
20588 2017-03-24 Richard Biener <rguenther@suse.de>
20590 PR tree-optimization/80167
20591 * graphite-isl-ast-to-gimple.c
20592 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20594 (translate_isl_ast_to_gimple::get_rename): Likewise.
20596 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20598 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20599 handling of certain combinations of target options, including the
20600 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20601 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20603 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20606 * config/arm/arm.md (*load_multiple): Add reload_completed to
20607 matching condition.
20609 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20610 Richard Biener <rguenth@suse.de>
20612 PR tree-optimization/79908
20613 PR tree-optimization/80136
20614 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20615 been cast away, gimplify_and_add suffices.
20617 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20619 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20621 2017-03-23 Richard Biener <rguenther@suse.de>
20623 PR tree-optimization/80032
20624 * gimplify.c (gimple_push_cleanup): Forced unconditional
20625 cleanups still have to go to the conditional_cleanups
20628 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20630 PR tree-optimization/80072
20631 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20633 (next_operand_entry_id): Change type to unsigned int.
20634 (sort_by_operand_rank): Make sure to return the right return value
20635 even if unsigned fields are bigger than INT_MAX.
20636 (struct oecount): Change cnt and id type to unsigned int.
20637 (oecount_hasher::equal): Formatting fix.
20638 (oecount_cmp): Make sure to return the right return value
20639 even if unsigned fields are bigger than INT_MAX.
20640 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
20643 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
20644 TREE_READONLY on result if writing it more than once.
20647 * doc/invoke.texi (-fsanitize=thread): Document that with
20648 -fnon-call-exceptions atomics are not able to throw
20652 * tsan.c: Include tree-eh.h.
20653 (instrument_builtin_call): Call maybe_clean_eh_stmt or
20654 maybe_clean_or_replace_eh_stmt where needed.
20655 (instrument_memory_accesses): Add cfg_changed argument.
20656 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
20657 if it returned true.
20658 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20660 PR rtl-optimization/63191
20661 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20662 wrapper function, moved the whole old content into ...
20663 (ix86_delegitimize_address_1): ... this. New inline function.
20664 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20665 true as last argument instead of ix86_delegitimize_address.
20667 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20669 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20670 cortexa57_branch_cost.
20672 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20674 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20676 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20679 * doc/md.texi (Constraints): Document wA constraint.
20680 * config/rs6000/constraints.md (wA): New.
20681 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20682 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20683 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20684 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20686 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20689 * gimplify.c (is_oacc_declared): New function.
20690 (oacc_default_clause): Use it to set default flags for acc declared
20691 variables inside parallel regions.
20692 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20693 declared variables.
20694 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20695 declare attribute to any decl as necessary.
20697 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20700 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20701 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20702 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20703 (arm_arch_lpae): This.
20704 * config/arm/arm.c (arm_arch7ve): Rename into ...
20705 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20706 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20709 2017-03-22 Martin Liska <mliska@suse.cz>
20712 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20713 error message instead of an ICE.
20715 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20717 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20719 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20721 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20723 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20726 2017-03-21 Martin Sebor <msebor@redhat.com>
20728 * doc/extend.texi: Use "cannot" instead of "can't."
20729 * doc/hostconfig.texi: Same.
20730 * doc/install.texi: Same.
20731 * doc/invoke.texi: Same.
20732 * doc/loop.texi: Same.
20733 * doc/md.texi: Same.
20734 * doc/objc.texi: Same.
20735 * doc/rtl.texi: Same.
20736 * doc/tm.texi: Same.
20737 * doc/tm.texi.in: Same.
20738 * doc/trouble.texi: Same.
20740 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20743 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20744 (collect_checksum_attributes): Set it.
20745 (die_checksum_ordered): Use it.
20747 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20749 PR tree-optimization/79908
20750 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
20751 change: For a VA_ARG whose LHS has been cast away, use
20752 force_gimple_operand to construct the side effects.
20754 2017-03-21 David Malcolm <dmalcolm@redhat.com>
20756 PR translation/80001
20757 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
20758 more amenable to translation.
20759 (oacc_loop_auto_partitions): Likewise.
20761 2017-03-21 Marek Polacek <polacek@redhat.com>
20762 Martin Sebor <msebor@redhat.com>
20764 PR tree-optimization/80109
20765 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
20766 on INTEGRAL_TYPE_P.
20768 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20769 Segher Boessenkool <segher@kernel.crashing.org>
20772 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
20773 check reg_used_between_p between insn and one of succ or succ2
20774 depending on if succ is artificial insn not inserted into insn
20777 2017-03-21 Martin Liska <mliska@suse.cz>
20779 PR gcov-profile/80081
20780 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
20781 * doc/gcc.texi: Include gcov-dump stuff.
20782 * doc/gcov-dump.texi: New file.
20784 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
20786 PR rtl-optimization/79150
20787 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
20788 conditional jump, if the jump is the last insn of the loop.
20790 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20791 Richard Biener <rguenth@suse.de>
20793 PR tree-optimization/79908
20794 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20795 been cast away, use force_gimple_operand to construct the side
20798 2017-03-21 Martin Liska <mliska@suse.cz>
20800 PR libfortran/79956
20801 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
20804 2017-03-21 Brad Spengler <spender@grsecurity.net>
20807 * plugin.c (htab_hash_plugin): New function.
20808 (add_new_plugin): Use it and adjust.
20809 (parse_plugin_arg_opt): Adjust.
20810 (init_one_plugin): Likewise.
20812 2017-03-21 Richard Biener <rguenther@suse.de>
20814 PR tree-optimization/80032
20815 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
20816 if set force the cleanup to happen unconditionally.
20817 (gimplify_target_expr): Push inserted clobbers with force_uncond
20818 to avoid them being removed by control-dependent DCE.
20820 2017-03-21 Richard Biener <rguenther@suse.de>
20822 PR tree-optimization/80122
20823 * tree-inline.c (copy_bb): Do not expans va-arg packs or
20824 va_arg_pack_len when the inlined call stmt requires pack
20826 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
20828 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20831 * tsan.c (instrument_builtin_call): If the memory model argument
20832 is not a constant, assume it is valid.
20835 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
20838 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20840 PR rtl-optimization/79910
20841 * combine.c (can_combine_p): Do not allow combining an I0 or I1
20842 if its dest is used by an insn before I2 (other than the combined
20843 insns themselves, which are properly handled already).
20845 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20848 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20850 * combine.c (record_used_regs): New static function.
20851 (try_combine): Handle situations where there is an additional
20852 instruction between I2 and I3 which needs to have a LOG_LINK
20856 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20858 * combine.c (try_combine): Delete redundant i1 test. Call
20859 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20861 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20864 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
20865 alternatives 13/14.
20867 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20869 PR tree-optimization/80054
20870 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
20871 the optimization if a PHI or any of its arguments is not dominated
20872 by the candidate's basis. Use gphi* rather than gimple* as
20874 (replace_profitable_candidates): Clean up a gimple* variable that
20875 should be a gphi* variable.
20877 2017-03-20 Martin Sebor <msebor@redhat.com>
20880 * doc/extend.texi (attribute constructor): Document present limitation.
20882 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
20885 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
20886 __POWER9_VECTOR__ #ifdef control, change template definition to
20887 use Power9-specific built-in function.
20888 (vec_any_eq): Likewise.
20889 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
20890 to control outcomes from this test.
20891 (vector_ae_<mode>p): For VEC_F modes, likewise.
20893 2017-03-20 Ian Lance Taylor <iant@google.com>
20895 * config/i386/i386.c (ix86_function_regparm): Save an extra
20896 register for -fsplit-stack with DECL_STATIC_CHAIN.
20898 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20901 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
20902 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
20904 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20906 * config/riscv/riscv.c (riscv_print_operand): Use "fence
20908 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
20911 2017-03-20 Marek Polacek <polacek@redhat.com>
20914 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
20916 2017-03-20 Richard Biener <rguenther@suse.de>
20918 PR tree-optimization/80113
20919 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
20920 allocate extra SSA name for PHI def.
20921 (add_close_phis_to_outer_loops): Likewise.
20922 (add_close_phis_to_merge_points): Likewise.
20923 (copy_loop_close_phi_args): Likewise.
20924 (copy_cond_phi_nodes): Likewise.
20926 2017-03-20 Martin Liska <mliska@suse.cz>
20928 PR middle-end/79753
20929 * tree-chkp.c (chkp_build_returned_bound): Do not build
20930 returned bounds for a LHS that's not a BOUNDED_P type.
20932 2017-03-20 Martin Liska <mliska@suse.cz>
20936 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
20937 COMPLEX_CST and VECTOR_CST.
20939 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20942 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
20943 target operand. A new splitter adds the clobber statement in case
20944 the target operand is dead anyway.
20946 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
20948 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
20949 to age-old versions of binutils and glibc.
20951 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
20953 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
20955 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20957 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
20959 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20961 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
20962 requirement for binutils 2.13.
20964 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20966 * combine.c (try_combine): Delete redundant i1 test. Call
20967 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20969 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
20971 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
20972 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
20974 <riscv64-*-elf>: Re-arrange section
20975 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
20976 <riscv32-*-linux>: Likewise.
20977 <riscv64-*-elf>: Likewise
20978 <riscv64-*-linux>: Likewise.
20980 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
20983 * aarch64.opt(verbose-cost-dump): Fix typo.
20985 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20988 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20989 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20991 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20993 * reload.c (find_reloads): When reloading a nonoffsettable address,
20994 use RELOAD_OTHER for it and its address reloads.
20996 PR rtl-optimization/79910
20997 * combine.c (record_used_regs): New static function.
20998 (try_combine): Handle situations where there is an additional
20999 instruction between I2 and I3 which needs to have a LOG_LINK
21002 2017-03-17 Jeff Law <law@redhat.com>
21004 PR tree-optimization/71437
21005 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
21006 conditional in the hash table first.
21007 (vrp_dom_walker::before_dom_children): Extract condition from
21008 ASSERT_EXPR. Record condition, its inverion and any implied
21009 conditions as well.
21011 2017-03-17 Marek Polacek <polacek@redhat.com>
21012 Markus Trippelsdorf <markus@trippelsdorf.de>
21014 PR tree-optimization/80079
21015 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
21018 2017-03-17 Richard Biener <rguenther@suse.de>
21020 PR middle-end/80075
21021 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
21022 Properly verify the LHS before the RHS possibly claims to be
21024 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
21027 2017-03-17 Martin Jambor <mjambor@suse.cz>
21029 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
21030 (List of -O2 options): Likewise.
21031 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
21034 2017-03-17 Tom de Vries <tom@codesourcery.com>
21036 * gcov-dump.c (print_usage): Print bug_report_url.
21038 2017-03-17 Richard Biener <rguenther@suse.de>
21040 PR middle-end/80050
21041 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
21042 (parser::peek): Likewise.
21044 2017-03-17 Richard Biener <rguenther@suse.de>
21046 PR tree-optimization/80048
21047 * sese.c (free_sese_info): Properly release rename_map and
21048 copied_bb_map elements.
21050 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
21052 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
21053 Add linked-list forward and backlinks. Insert on
21054 construction, remove on destruction.
21055 (class pass_store_merging): Add m_stores_head field.
21056 (pass_store_merging::terminate_and_process_all_chains):
21057 Iterate over m_stores_head list.
21058 (pass_store_merging::terminate_all_aliasing_chains):
21060 (pass_store_merging::execute): Check for debug stmts first.
21061 Push new chains onto the m_stores_head stack.
21063 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21066 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
21067 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
21068 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
21070 2017-03-16 Jeff Law <law@redhat.com>
21072 PR tree-optimization/71437
21073 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
21074 member function. Implementation moved into after_dom_children
21075 member function and into the threader's thread_outgoing_edges
21077 (dom_opt_dom_walker::after_dom_children): Simplify by moving
21078 some code into new thread_outgoing_edges.
21079 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
21080 definition. Simplify marker handling (do it here). Assume we always
21081 have the available expression and the const/copies tables.
21082 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
21084 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
21085 * tree-vrp.c (equiv_stack): No longer file scoped.
21086 (vrp_dom_walker): New class.
21087 (vrp_dom_walker::before_dom_children): New member function.
21088 (vrp_dom_walker::after_dom_children): Likewise.
21089 (identify_jump_threads): Setup domwalker. Use it rather than
21090 walking edges in a random order by hand. Simplify setup/finalization.
21091 (finalize_jump_threads): Remove.
21092 (vrp_finalize): Do not call identify_jump_threads here.
21093 (execute_vrp): Do it here instead and call thread_through_all_blocks
21096 PR tree-optimization/71437
21097 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
21099 (simplify_stmt_for_jump_threading): Add basic_block argument. All
21101 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
21102 (dom_opt_dom_walker::thread_across_edge): Remove
21103 handle_dominating_asserts argument. All callers changed.
21104 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
21105 changes. Remove calls to lhs_of_dominating_assert. Other
21106 uses of handle_dominating_asserts turn into unconditional code
21107 (simplify_control_stmt_condition_1): Likewise.
21108 (simplify_control_stmt_condition): Likewise.
21109 (thread_through_normal_block, thread_across_edge): Likewise.
21110 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
21111 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
21112 object if it is not an SSA_NAME.
21113 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
21114 before calling into the VRP specific simplifiers.
21115 (identify_jump_threads): Remove handle_dominating_asserts
21118 2017-03-16 Jakub Jelinek <jakub@redhat.com>
21121 * tree-diagnostic.c (default_tree_printer): No longer static.
21122 * tree-diagnostic.h (default_tree_printer): New prototype.
21124 2017-03-16 Tamar Christina <tamar.christina@arm.com>
21126 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
21127 Change ins into fmov.
21129 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21131 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
21132 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
21133 Use h_con constraint for operand 1.
21134 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
21135 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
21137 2017-03-15 Jeff Law <law@redhat.com>
21139 PR tree-optimization/71437
21140 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
21141 (record_temporary_equivalences): Use it.
21143 PR tree-optimization/71437
21144 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
21145 tree-ssa-scopedtables.
21146 (lookup_avail_expr, build_and_record_new_cond): Likewise.
21147 (record_conditions, record_cond, vuse_eq): Likewise.
21148 (record_edge_info): Adjust to API tweak of record_conditions.
21149 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
21150 (record_temporary_equivalences, optimize_stmt): Likewise.
21151 (eliminate_redundant_computations): Likewise.
21152 (record_equivalences_from_stmt): Likewise.
21153 * tree-ssa-scopedtables.c: Include options.h and params.h.
21154 (vuse_eq): New function, moved from tree-ssa-dom.c
21155 (build_and_record_new_cond): Likewise.
21156 (record_conditions): Likewise. Accept vector of conditions rather
21157 than edge_equivalence structure for first argument.
21158 for the first argument.
21159 (avail_exprs_stack::lookup_avail_expr): New member function, moved
21160 from tree-ssa-dom.c.
21161 (avail_exprs_stack::record_cond): Likewise.
21162 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
21163 from tree-ssa-dom.c.
21164 (avail_exprs_stack): Add new member functions lookup_avail_expr
21166 (record_conditions): Declare.
21168 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
21171 * lra-constraints.c (process_alt_operands): Increase reject for
21172 reloading an input/output operand.
21174 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
21177 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
21178 insns to convert from signed/unsigned char/short to IEEE 128-bit
21180 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
21182 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
21185 * config/i386/i386.c (ix86_vector_duplicate_value): Create
21186 subreg of inner mode for values already in registers.
21188 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
21190 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
21191 iteration reg is used after the loop.
21193 2017-03-14 Martin Sebor <msebor@redhat.com>
21195 PR tree-optimization/79800
21196 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
21197 precision in negative-positive range.
21198 (format_floating): Call non-const overload with adjusted precision.
21200 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21203 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
21206 2017-03-14 Martin Sebor <msebor@redhat.com>
21208 PR middle-end/80020
21209 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
21210 * builtins.def (aligned_alloc): Use it.
21213 * Makefile.in (GTFILES): Add calls.c.
21214 * calls.c: Include "gt-calls.h".
21216 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
21218 PR rtl-optimization/79728
21219 * regs.h (struct target_regs): New field
21220 x_contains_allocatable_regs_of_mode.
21221 (contains_allocatable_regs_of_mode): New macro.
21222 * reginfo.c (init_reg_sets_1): Initialize it, and change
21223 contains_reg_of_mode so it includes global regs as well.
21224 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
21225 rather than contains_regs_of_mode.
21227 2017-03-14 Martin Liska <mliska@suse.cz>
21229 * doc/invoke.texi: Document options that can't be combined with
21230 -fcheck-pointer-bounds.
21232 2017-03-14 Martin Liska <mliska@suse.cz>
21234 PR middle-end/79831
21235 * doc/invoke.texi (-Wchkp): Document the option.
21237 2017-03-14 Martin Liska <mliska@suse.cz>
21239 * Makefile.in: Install gcov-dump.
21241 2017-03-14 Martin Liska <mliska@suse.cz>
21243 * multiple_target.c (expand_target_clones): Bail out for
21244 an invalid attribute.
21246 2017-03-14 Richard Biener <rguenther@suse.de>
21248 * alias.c (struct alias_set_entry): Pack properly.
21249 * cfgloop.h (struct loop): Likewise.
21250 * cse.c (struct set): Likewise.
21251 * ipa-utils.c (struct searchc_env): Likewise.
21252 * loop-invariant.c (struct invariant): Likewise.
21253 * lra-remat.c (struct cand): Likewise.
21254 * recog.c (struct change_t): Likewise.
21255 * rtl.h (struct address_info): Likewise.
21256 * symbol-summary.h (function_summary): Likewise.
21257 * tree-loop-distribution.c (struct partition): Likewise.
21258 * tree-object-size.c (struct object_size_info): Likewise.
21259 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
21260 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
21261 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
21262 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
21263 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
21264 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
21265 (struct _stmt_vec_info): Likewise.
21267 2017-03-14 Martin Liska <mliska@suse.cz>
21270 * multiple_target.c (create_dispatcher_calls): Check that
21271 a target can create a function dispatcher.
21273 2017-03-14 Martin Liska <mliska@suse.cz>
21276 * multiple_target.c (expand_target_clones): Drop local.local
21277 flag for default implementation.
21279 2017-03-14 Richard Biener <rguenther@suse.de>
21281 PR tree-optimization/80030
21282 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
21284 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
21286 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
21287 gcc_fallthrough() instead of __attribute__((fallthrough));
21289 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21291 * doc/gcc.texi: Remove "up" link to (DIR).
21292 * doc/gccint.texi: Ditto.
21294 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21296 * doc/install.texi (Specific) <avr>: Remove reference to
21299 2017-03-13 Jeff Law <law@redhat.com>
21301 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
21302 attribute rather than comments.
21304 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
21305 match_scratch operand is highest.
21307 2017-03-13 Martin Liska <mliska@suse.cz>
21309 PR middle-end/78339
21310 * ipa-pure-const.c (warn_function_noreturn): If the declarations
21311 is a CHKP clone, use original declaration.
21313 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21315 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
21316 (arc_conditional_register_usage): Use a different allocation order
21317 when optimizing for size.
21318 * common/config/arc/arc-common.c (arc_option_optimization_table):
21319 Section anchors default on when optimizing for size.
21321 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21323 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
21325 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21327 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
21328 * config/arc/arc.md (cpu_facility): Add cd variant.
21329 (*movqi_insn): Add code density variant.
21330 (*movhi_insn): Likewise.
21331 (*movqi_insn): Likewise.
21332 (*addsi3_mixed): Likewise.
21333 (subsi3_insn): Likewise.
21335 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21337 * config/arc/arc.md (movsi_cond_exec): Update constraint.
21339 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21341 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
21342 expressions with MINUS and UNARY ops.
21344 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21347 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
21349 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
21350 between vec_select and vector argument.
21351 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
21352 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
21353 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
21354 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
21355 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
21356 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
21358 2017-03-13 Richard Biener <rguenther@suse.de>
21361 * params.def (vect-max-peeling-for-alignment): Fix typo.
21363 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21365 * doc/install.texi (Specific) <mips-*-*>: Remove description of
21366 issue that only occurred with binutils below 2.18.
21368 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21370 * doc/install.texi (Specific) <cris-axis-elf>: No longer
21371 refer to binutils 2.11/2.12 minimum.
21373 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21375 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
21376 ftp.kernel.org and simplify binutils requirement.
21378 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
21380 * doc/invoke.texi (Warning Options): Fix spelling of link-time
21382 (Optimize Options): Ditto. Also remove redundancy.
21384 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21386 PR translation/79848
21387 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
21389 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
21390 to G_ to avoid double translation.
21392 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21394 PR translation/79923
21395 * auto-profile.c (get_combined_location): Convert leading
21396 character of diagnostics to lower case and remove trailing period.
21397 (read_profile): Likewise for various diagnostics.
21398 * config/arm/arm.c (arm_option_override): Remove trailing period
21399 from various diagnostics.
21400 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
21401 (msp430_expand_delay_cycles): Likewise.
21403 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21406 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21407 full command-line argument, rather than just "str".
21408 (aarch64_validate_march): Likewise.
21409 (aarch64_validate_mtune): Likewise.
21411 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21413 PR rtl-optimization/78911
21414 * lra-assigns.c (must_not_spill_p): New function.
21415 (spill_for): Use it.
21417 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21419 PR tree-optimization/79981
21420 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21421 ATOMIC_COMPARE_EXCHANGE ifn result.
21422 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21423 IFN_ATOMIC_COMPARE_EXCHANGE.
21425 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21428 * opts.c (parse_sanitizer_options): Add missing question mark to
21429 "did you mean" message.
21431 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21433 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21435 (VMULEUH_UNS): Likewise.
21436 (VMULOUB_UNS): Likewise.
21437 (VMULOUH_UNS): Likewise.
21438 * config/rs6000/rs6000.c (builtin_function_type): Remove
21439 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21441 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21444 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21445 x with result of extra_parsing_for_operand_code_0.
21446 (function_reader::extra_parsing_for_operand_code_0): Convert
21447 return type from void to rtx, returning x. When reading
21448 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21449 larger size containing struct block_symbol.
21451 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21453 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21454 -mfloat128-hardware without -m64.
21456 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21459 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21460 entries to the case statement that marks unsigned arguments to
21461 overloaded functions.
21463 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21465 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21466 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21468 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21471 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21472 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21474 2017-03-10 Martin Liska <mliska@suse.cz>
21478 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21479 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21482 2017-03-10 Olivier Hainque <hainque@adacore.com>
21484 * tree-switch-conversion (array_value_type): Start by resetting
21485 candidate type to it's main variant.
21487 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21489 PR rtl-optimization/79909
21490 * combine.c (try_combine): Use simplify_replace_rtx on individual
21491 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21492 of the whole CALL_INSN_FUNCTION_USAGE.
21494 PR tree-optimization/79972
21495 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21496 get_range_info on SSA_NAMEs. Formatting fixes.
21498 2017-03-10 Richard Biener <rguenther@suse.de>
21499 Jakub Jelinek <jakub@redhat.com>
21501 PR tree-optimization/77975
21502 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21503 edge to be constant.
21504 (get_val_for): For constant x return it. Formatting fix.
21505 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21506 would use the same bases as the current one.
21508 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21510 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21511 instead of vec_select for V1TImode.
21512 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21514 (VSX_LE_128): Add V1TI to this mode iterator.
21515 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21516 (*vsx_le_perm_store_<mode>): Likewise.
21517 (pre-reload splitter for VSX stores): Likewise.
21518 (post-reload splitter for VSX stores): Likewise.
21519 (*vsx_xxpermdi2_le_<mode>): Likewise.
21520 (*vsx_lxvd2x2_le_<mode>): Likewise.
21521 (*vsx_stxvd2x2_le_<mode>): Likewise.
21523 2017-03-09 Michael Eager <eager@eagercon.com>
21525 Correct failures with --enable-checking=yes,rtl.
21527 * config/microblaze/microblaze.c (microblaze_expand_shift):
21528 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21529 test for const0_rtx.
21530 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21531 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21533 2017-03-09 Richard Biener <rguenther@suse.de>
21535 PR tree-optimization/79977
21536 * graphite-scop-detection.c (scop_detection::merge_sese):
21537 Handle the case of extra exits to blocks dominating the entry.
21539 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21541 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21544 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21546 PR rtl-optimization/79949
21547 * lra-constraints.c (process_alt_operands): Check memory when
21548 trying to predict a cycle. Print about the overall increase.
21550 2017-03-09 Richard Biener <rguenther@suse.de>
21552 PR middle-end/79971
21553 * gimple-expr.c (useless_type_conversion_p): Preserve
21554 TYPE_SATURATING for fixed-point types.
21556 2017-03-09 Richard Biener <rguenther@suse.de>
21559 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21560 alignment of BLKmode params.
21562 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21565 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21566 (VALL_NO_V2Q): Likewise.
21568 * config/aarch64/aarch64-simd.md
21569 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21571 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21572 VALL_NO_V2Q mode iterator.
21573 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21575 2017-03-09 Martin Liska <mliska@suse.cz>
21577 PR tree-optimization/79631
21578 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21579 tree_int_cst_sign_bit just for INTEGER constants.
21581 2017-03-09 Martin Liska <mliska@suse.cz>
21585 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21588 2017-03-09 Marek Polacek <polacek@redhat.com>
21591 * tree.c (inchash::add_expr): Handle TREE_VEC.
21593 2017-03-09 Martin Liska <mliska@suse.cz>
21596 (chkp_narrow_size_and_offset): New function.
21597 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21598 (void chkp_parse_bit_field_ref): New function.
21599 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21600 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21602 2017-03-09 Martin Liska <mliska@suse.cz>
21605 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21606 (chkp_find_bounds_1): Remove gcc_unreachable.
21608 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21611 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21612 BUILT_IN_SYNC*, determine the access type from the size suffix and
21613 always build a MEM_REF with that type. Handle forgotten
21614 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21617 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21618 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21619 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21620 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21621 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21622 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21623 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21624 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21625 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21626 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21627 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21628 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21629 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21630 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21631 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21632 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21633 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21634 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21635 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21636 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21637 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21638 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21639 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21640 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21641 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21642 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
21643 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
21644 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
21645 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
21646 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
21647 definitions outside of __OPTIMIZE__ guarded section.
21650 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
21651 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
21652 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
21653 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
21656 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21658 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21659 ("vfenez<mode>"): Add missing constraints.
21661 2017-03-08 Martin Sebor <msebor@redhat.com>
21664 * config/nds32/nds32.c (nds32_option_override):
21665 Fix misspelled diagnostic.
21667 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21670 * gimplify.c (gimplify_omp_for): Replace index var in outer
21671 taskloop statement with an artificial variable and add
21672 OMP_CLAUSE_PRIVATE clause for it.
21674 2017-03-08 Richard Biener <rguenther@suse.de>
21676 PR tree-optimization/79955
21677 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21678 for accesses that are completely outside of the variable.
21680 2017-03-08 Andrew Haley <aph@redhat.com>
21682 PR tree-optimization/79943
21683 * tree-ssa-loop-split.c (compute_new_first_bound): When
21684 calculating the new upper bound, (END-BEG) should be added, not
21687 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21689 * config/avr/avr.md (setmemhi): Make sure match_dup
21690 operand number comes before match_scratch.
21692 2017-03-08 Richard Biener <rguenther@suse.de>
21694 PR tree-optimization/79920
21695 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21696 with ncopies == 1 to ...
21697 (vect_transform_slp_perm_load): ... here. Properly compute
21698 all element loads by iterating VF times over the group. Do
21699 not handle ncopies (computed in a broken way) in
21700 vect_create_mask_and_perm.
21702 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21705 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21706 is a uniform vector, use uniform_vector_p return value instead of
21707 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21709 2017-03-07 Marek Polacek <polacek@redhat.com>
21711 PR middle-end/79809
21712 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21713 (alloca_call_type): Likewise.
21715 2017-03-07 Martin Liska <mliska@suse.cz>
21717 * gcov.c (process_args): Put comment to correct location.
21719 2017-03-07 Martin Liska <mliska@suse.cz>
21721 PR middle-end/68270
21722 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21723 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21724 (chkp_narrow_bounds_for_field): Likewise.
21725 (chkp_parse_array_and_component_ref): Pass one more argument to
21728 2017-03-07 Richard Biener <rguenther@suse.de>
21730 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21733 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21735 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21736 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21738 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21741 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21742 to end of description.
21743 (PARAM_MAX_STORES_TO_MERGE): Likewise.
21745 2017-03-07 Jakub Jelinek <jakub@redhat.com>
21747 PR rtl-optimization/79901
21748 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
21750 (*avx512f_<code><mode>3<mask_name>): ... this.
21751 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
21752 iterator instead of VI8_AVX2_AVX512BW.
21754 PR rtl-optimization/79901
21755 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
21756 min/max expander, expand it using expand_vec_cond_expr.
21759 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
21762 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21765 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
21766 to void * for PCH reasons.
21767 * dwarf2out.c (output_loc_operands, output_die): Cast
21768 v.val_vec.array to unsigned char *.
21770 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
21773 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
21776 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
21778 PR rtl-optimization/79571
21779 * lra-constraints.c (process_alt_operands): Calculate static
21780 reject and subtract it from overall when only addresses will be
21783 2017-03-06 Julia Koval <julia.koval@intel.com>
21786 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
21787 incoming stack boundary to 128 for 64-bit targets.
21789 2017-03-06 Richard Biener <rguenther@suse.de>
21791 PR tree-optimization/79894
21792 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
21793 to NULL after folding it.
21795 2017-03-06 Richard Biener <rguenther@suse.de>
21797 PR tree-optimization/79824
21798 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
21799 check disabling peeling for gaps.
21801 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
21803 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21804 attributes): Document gettimeofday.
21806 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21808 * config/s390/s390.c (s390_option_override_internal): Set
21809 PARAM_MIN_VECT_LOOP_BOUND
21811 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21813 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
21814 * config/s390/s390.md: Likewise.
21816 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21819 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
21820 (<avx2_avx512>_perm<mode>): Rename to ...
21821 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
21823 (<avx512>_perm<mode>_mask): Rename to ...
21824 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
21826 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
21827 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
21828 instead of VI8F_256_512.
21829 (avx512f_perm<mode>): New define_expand.
21830 (avx512f_perm<mode>_mask): Likewise.
21831 (avx512f_perm<mode>_1<mask_name>): New define_insn.
21832 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
21834 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21836 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
21837 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
21839 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
21841 2017-03-06 Martin Liska <mliska@suse.cz>
21844 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
21845 when having a SSA NAME w/o VAR_DECL assigned to it.
21847 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21849 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
21850 msa_dpsub_<su>_d): Fix MODE for vec_select.
21852 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21854 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
21856 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
21858 2017-03-06 Richard Biener <rguenther@suse.de>
21860 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
21861 * plugin.c (register_plugin_info): Likewise.
21862 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
21864 2017-03-05 Jakub Jelinek <jakub@redhat.com>
21866 * config/i386/sse.md (sse_storehps, sse_storelps,
21867 avx_<castmode><avxsizesuffix>_<castmode>,
21868 avx512f_<castmode><avxsizesuffix>_<castmode>,
21869 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
21870 in condition that at least one operand is not a MEM.
21872 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21874 PR middle-end/79805
21875 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
21876 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
21878 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
21879 gimple_call_nothrow_p flag based on whether original builtin can throw.
21880 If it can, emit following stmts on the fallthrough edge.
21881 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
21882 don't create new bb if inserting just debug stmts on the edge, try to
21883 insert them on the fallthru bb or just reset debug stmts.
21885 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
21888 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
21889 restore recog_data (including the operand rtxes inside it) around
21890 the call to get_insn_template.
21892 2017-03-03 Martin Sebor <msebor@redhat.com>
21894 PR tree-optimization/79699
21895 * context.c (context::~context): Free MPFR caches to avoid
21896 a memory leak on program exit.
21898 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21900 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
21901 Use wide_int::ulow () instead of .elt (0).
21903 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21905 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
21906 (*pushxf): Limit oF constraint to 32bit targets and add oC
21907 constraint for 64bit targets.
21908 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
21909 (*pushdf): Change rmF constraint to rmC.
21911 2017-03-03 Martin Liska <mliska@suse.cz>
21913 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
21914 Remove unused variable.
21916 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21919 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
21920 is a memory operand, increase num_memory.
21921 (ix86_expand_args_builtin): Likewise.
21923 2017-03-03 Jan Hubicka <jh@suse.cz>
21926 * ipa-devirt.c (maybe_record_node): Properly handle
21927 __cxa_pure_virtual visibility.
21929 2017-03-03 Martin Liska <mliska@suse.cz>
21931 PR tree-optimization/79803
21932 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
21934 (pass_loop_prefetch::execute): Disabled optimization if an
21935 assumption about L1 cache size is not met.
21937 2017-03-03 Martin Liska <mliska@suse.cz>
21939 PR rtl-optimization/79574
21940 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
21941 (hash_scan_set): Likewise.
21942 (dump_hash_table): Likewise.
21943 (hoist_code): Likewise.
21945 2017-03-03 Richard Biener <rguenther@suse.de>
21947 * fixed-value.c (fixed_from_string): Restore use of elt (1)
21948 in place of uhigh ().
21949 (fixed_convert_from_real): Likewise.
21951 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21954 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
21956 2017-03-03 Richard Biener <rguenther@suse.de>
21958 PR middle-end/79818
21959 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
21960 TYPE_OVERFLOW_UNDEFINED check.
21962 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21964 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
21966 (vector_ae_<mode>_p): Likewise.
21967 (vector_nez_<mode>_p): Likewise.
21968 (vector_ne_v2di_p): Likewise.
21969 (vector_ae_v2di_p): Likewise.
21970 (vector_ne_<mode>_p): Likewise.
21971 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
21973 (vsx_tsqrt<mode>2_fe): Likewise.
21975 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
21978 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
21980 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21982 PR rtl-optimization/79780
21983 * cprop.c (one_cprop_pass): When second and further conditional trap
21984 in a single basic block is turned into an unconditional trap, turn it
21985 into a deleted note to avoid RTL verification failures.
21987 2017-03-02 Richard Biener <rguenther@suse.de>
21989 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21991 2017-03-02 Richard Biener <rguenther@suse.de>
21993 PR tree-optimization/79345
21995 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21996 param and abort the walk, returning -1 if it is hit.
21997 (walk_aliased_vdefs): Take a limit param and pass it on.
21998 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21999 defaulting to 0 and return a signed int.
22000 * tree-ssa-uninit.c (struct check_defs_data): New struct.
22001 (check_defs): New helper.
22002 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
22003 about uninitialized memory.
22004 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
22005 bogus uninitialized warning.
22006 (fixed_convert_from_real): Likewise.
22008 2017-03-02 Bin Cheng <bin.cheng@arm.com>
22010 PR tree-optimization/66768
22011 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
22012 iv_use if base object can't be determined.
22014 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22016 PR tree-optimization/79345
22017 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
22018 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
22019 (get_pattern_stats): Initialize it.
22020 * genemit.c (gen_expand): Verify match_scratch numbers come after
22021 match_operand/match_dup numbers.
22022 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
22023 match_scratch numbers.
22024 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
22026 * config/s390/s390.md (trunctdsd2): Likewise.
22028 2017-03-02 Richard Biener <rguenther@suse.de>
22030 * wide-int.h (wide_int_storage::operator=): Implement in terms
22033 2017-03-02 Richard Biener <rguenther@suse.de>
22035 PR tree-optimization/79777
22036 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
22037 the to insert expression to sth existing.
22039 2017-03-01 Martin Sebor <msebor@redhat.com>
22041 PR middle-end/79692
22042 * gimple-ssa-sprintf.c
22043 (directive::known_width_and_precision): New function.
22044 (format_integer): Use it.
22045 (get_mpfr_format_length): Consider the full range of precision
22046 when computing %g output with the # flag. Set the likely byte
22047 count to 3 rather than 1 when precision is indeterminate.
22048 (format_floating): Correct the lower bound of precision.
22050 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22052 * doc/invoke.texi: Document default code model for 64-bit Linux.
22054 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22057 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
22058 udiv rather than div since input pattern is unsigned.
22060 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
22062 * config/i386/i386.c (print_reg): Warn for values of
22063 unsupported size in integer register.
22065 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
22068 * config/rs6000/predicates.md (current_file_function_operand): Do
22069 not allow self calls to be local if the function is replaceable.
22071 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
22074 * config/rs6000/altivec.h (vec_ctz and others): Change the
22075 preprocessor macro that controls conditional compilation from
22076 _ARCH_PWR9 to __POWER9_VECTOR__.
22077 (vec_all_ne): Change parameterization of __altivec_scalar_pred
22078 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
22079 control (instead of _ARCH_PWR9 control) so that template
22080 definition uses power9-specific function.
22081 (vec_any_eq): Likewise.
22082 (vec_all_ne): Change macro definition to use a power9-specific
22083 expansion under #ifdef __POWER9_VECTOR__ control (instead of
22084 _ARCH_PWR9 control).
22085 (vec_any_eq) Likewise.
22086 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
22087 expansion for CMPNEF to remove support for xvcmpnesp instruction.
22088 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
22089 support for xvcmpnedp instruction.
22090 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
22091 macro expansion so that Power9 implementation of vec_all_ne does
22092 not use the AltiVec predicate framework.
22093 (VCMPNEH_P): Likewise.
22094 (VCMPNEW_P): Likewise.
22095 (VCMPNED_P): Likewise.
22096 (VCMPNEFP_P): Likewise.
22097 (VCMPNEDP_P): Likewise.
22098 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
22099 implementation of vec_any_eq to not use AltiVec predicate
22101 (VCMPAEH_P): Likewise.
22102 (VCMPAEW_P): Likewise.
22103 (VCMPAED_P): Likewise.
22104 (VCMPAEFP_P): Likewise.
22105 (VCMPAEDP_P): Likewise.
22106 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
22107 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
22108 not use the AltiVec predicate framework.
22109 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
22110 of vec_any_eq to not use AltiVec predicate framework.
22111 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22112 support for predefined __POWER9_VECTOR__ macro to indicate that
22113 Power9 instruction selection is enabled.
22114 (altivec_overloaded_builtins): Remove extraneous
22115 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
22116 function argument types RS6000_BTI_bool_V16QI and
22117 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
22118 entry for overloaded function argument types RS6000_BTI_bool_V4SI
22119 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
22120 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
22121 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
22122 Power9 for implementations of vec_cmpne. Change the signature for
22123 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
22124 (representing vec_all_ne) to remove the previously described first
22125 argument of type RS6000_BTI_INTSI, as this was an artifact of
22126 reliance on the AltiVec predicate framework, which is no longer
22127 used in the implementation of these functions. Add
22128 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
22129 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
22130 since, unlike the AltiVec predicate framework implementation, we
22131 do not share function descriptors between vec_alle and vec_anyeq.
22132 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
22133 set of modes that receive special treatment even when
22134 TARGET_P9_VECTOR is true. The special treatment emits code that
22135 does not depend on Power9 instructions.
22136 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
22137 define_expand to not rely on AltiVec predicate framework.
22138 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22140 (vector_ne_v2di_p): Change this define_expand to not rely on
22141 AltiVec predicate framework.
22142 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
22144 (vector_ne_<mode>_p): Change this define_expand to not rely on
22145 AltiVec predicate framework.
22146 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22148 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
22149 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
22150 define_insn pattern.
22151 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
22152 define_insn pattern because the xvcmpne<VSs>. instruction is not
22154 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
22155 instruction is not supported.
22157 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22159 * config/nvptx/nvptx.c: Include intl.h.
22161 2017-03-01 Martin Jambor <mjambor@suse.cz>
22164 * ipa-prop.h (ipa_bits): Removed field known.
22165 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
22166 to pointers. Adjusted their comments to warn about their sharing.
22167 (ipcp_transformation_summary): Change bits to a vector of pointers.
22168 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
22169 (ipa_get_ipa_bits_for_value): Declare.
22170 * tree-vrp.h (value_range): Mark as GTY((for_user)).
22171 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
22172 (ipa_bits_hash_table): Likewise.
22173 (ipa_vr_ggc_hash_traits): Likewise.
22174 (ipa_vr_hash_table): Likewise.
22175 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
22176 being pointers and vr_known being removed.
22177 (ipa_set_jf_unknown): Likewise.
22178 (ipa_get_ipa_bits_for_value): New function.
22179 (ipa_set_jfunc_bits): Likewise.
22180 (ipa_get_value_range): New overloaded functions.
22181 (ipa_set_jfunc_vr): Likewise.
22182 (ipa_compute_jump_functions_for_edge): Use the above functions to
22183 construct bits and vr parts of jump functions.
22184 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
22185 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22187 (ipcp_grow_transformations_if_necessary): Also allocate
22188 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22190 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
22191 them. Fix too long lines.
22192 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
22193 vr_known being removed.
22194 (ipa_read_jump_function): Use new setter functions to construct bits
22195 and vr parts of jump functions or set them to NULL.
22196 (write_ipcp_transformation_info): Adjust for bits being pointers.
22197 (read_ipcp_transformation_info): Likewise.
22198 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
22200 Include gt-ipa-prop.h.
22201 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
22203 (ipcp_store_bits_results): Likewise.
22204 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
22205 Do not write to existing jump functions but use a temporary instead.
22207 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22210 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
22211 attempt to use its first operand as BIT_FIELD_REF base.
22213 2017-03-01 Richard Biener <rguenther@suse.de>
22215 PR middle-end/79721
22216 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
22217 interpolating formula in wrapping arithmetic.
22218 (chrec_apply): Convert chrec_evaluate return value to wanted type.
22220 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22222 PR tree-optimization/79734
22223 * tree-vect-generic.c (expand_vector_condition): Optimize
22224 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
22225 Handle VEC_COND_EXPR where comparison has different inner width from
22226 type's inner width.
22228 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
22230 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
22231 markup, and similar issues. Remove @opindex entries for things
22232 that aren't options. Add missing -mmpy-option entries.
22234 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22236 PR tree-optimization/79737
22237 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
22238 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
22239 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
22240 instead of byte_size. Formatting fix.
22241 (shift_bytes_in_array_right): Formatting fix.
22243 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
22246 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
22247 condition on optimize for the leaf function test.
22249 2017-02-28 Martin Liska <mliska@suse.cz>
22252 * read-rtl-function.c (function_reader::handle_unknown_directive):
22253 Bail out when one uses -flto.
22255 2017-02-28 Martin Liska <mliska@suse.cz>
22257 * common.opt: Replace space with tabular for options of <number>
22259 * config/i386/i386.opt: Show <number> value for
22260 -mlarge-data-threshold.
22261 * opts.c (print_filtered_help): Do not display number in hexadecimal
22264 2017-02-28 Martin Liska <mliska@suse.cz>
22266 * common.opt: Fix --help=option -Q for options which are of
22269 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
22271 * config/i386/i386.c (print_reg): Error out for values
22272 of 8-bit size in invalid integer register.
22274 2017-02-28 Martin Sebor <msebor@redhat.com>
22276 PR tree-optimization/79691
22277 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
22279 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22282 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
22283 gcc_unreachable with output_operand_lossage.
22285 2017-02-28 Richard Biener <rguenther@suse.de>
22287 PR tree-optimization/79740
22288 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
22290 (visit_nary_op): Insert the nary into the hashtable if we
22291 pattern-matched sth.
22292 * tree-ssa-pre.c (eliminate_insert): Robustify.
22294 2017-02-28 Richard Biener <rguenther@suse.de>
22296 PR middle-end/79731
22297 * fold-const.c (decode_field_reference): Reject out-of-bound
22300 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22302 * config/i386/i386.c: Include intl.h.
22303 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
22304 instead of just cond ? "..." : "...".
22305 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
22306 * coverage.c (read_counts_file): Likewise.
22307 * omp-offload.c: Include intl.h.
22308 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
22309 of just cond ? "..." : "...".
22310 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
22311 of just cond ? "..." : "...".
22313 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
22316 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
22318 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
22319 'tune for' CPU name.
22320 * config/arm/arm-cpu-data.h: Regenerated.
22322 2017-02-28 Richard Biener <rguenther@suse.de>
22324 PR tree-optimization/79732
22325 * tree-inline.c (expand_call_inline): Do not shadow var.
22327 2017-02-28 Richard Biener <rguenther@suse.de>
22329 PR tree-optimization/79723
22330 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
22331 address-space properly.
22333 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
22335 * doc/optinfo.texi (Optimization groups): Fix option used for
22337 * doc/invoke.texi (-fopt-info): Document "omp".
22338 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
22339 (OPTGROUP_ALL): Add OPTGROUP_OMP.
22340 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
22341 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
22342 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
22344 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
22346 * dumpfile.c (optgroup_options): Instead of "openmp", associate
22347 OPTGROUP_OMP with "omp".
22349 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
22352 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
22353 for arithmetic shift of unsigned V2DI.
22355 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
22357 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
22358 arc/linux.h headers.
22359 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
22360 (LINK_SPEC): Likewise.
22361 (ARC_TLS_EXTRA_START_SPEC): Likewise.
22362 (EXTRA_SPECS): Likewise.
22363 (STARTFILE_SPEC): Likewise.
22364 (ENDFILE_SPEC): Likewise.
22365 (LIB_SPEC): Likewise.
22366 (TARGET_SDATA_DEFAULT): Likewise.
22367 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
22368 (MULTILIB_DEFAULTS): Likewise.
22369 (DWARF2_UNWIND_INFO): Likewise.
22370 * config/arc/big.h: New file.
22371 * config/arc/elf.h: Likewise.
22372 * config/arc/linux.h: Likewise.
22373 * config/arc/t-uClibc: Remove.
22375 2017-02-27 Bin Cheng <bin.cheng@arm.com>
22377 PR tree-optimization/77536
22378 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
22379 (tree_transform_and_unroll_loop): Use above function to compute the
22380 estimated niter of unrolled loop and use it when scaling profile.
22381 Also use count info rather than frequency if it's non-zero.
22382 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
22383 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
22384 (vect_transform_loop): Call above function.
22386 2017-02-27 Richard Biener <rguenther@suse.de>
22388 PR tree-optimization/45397
22389 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
22390 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
22391 (visit_nary_op): Add pattern matching for CSEing sign-changed
22392 or truncated operations with wider ones.
22394 2017-02-27 Richard Biener <rguenther@suse.de>
22396 PR tree-optimization/79690
22397 * tree-vect-stmts.c (vectorizable_store): Use vector type
22398 built from the DR with address-space.
22400 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22402 * doc/invoke.texi (Optimize Options): Refine the description
22403 of asan-use-after-return.
22405 2017-02-25 Alan Modra <amodra@gmail.com>
22407 PR rtl-optimization/79584
22408 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22409 base, not ad->base_term, the reg within base. Remove assertion
22410 that ad->base == ad->base_term. Replace gen_int_mode using
22411 bogus mode with const0_rtx.
22413 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22415 PR middle-end/79396
22416 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22417 FMA_EXPR like tcc_binary or tcc_unary.
22419 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22422 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22424 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22425 (output_loc_operands): Handle DW_OP_call_ref and
22426 DW_OP_GNU_variable_value.
22427 (struct variable_value_struct): New type.
22428 (struct variable_value_hasher): Likewise.
22429 (variable_value_hash): New variable.
22430 (string_types): Remove.
22431 (copy_loc_descr): New function.
22432 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22433 (prepend_loc_descr_to_each): New function.
22434 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22435 instead of add_loc_descr_to_each if the first argument is single
22436 location list and the second has multiple.
22437 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22438 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22439 when looking for variable value which doesn't have other location info.
22440 (loc_list_from_tree): Formatting fix.
22441 (gen_array_type_die): Simplify DW_AT_string_length handling.
22442 (adjust_string_types): Remove.
22443 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22444 string_types. Call resolve_variable_values.
22445 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22446 (resolve_addr_in_expr): Likewise. Add A argument.
22447 (copy_deref_exprloc): Remove deref argument. Adjust for the
22448 original expression being DW_OP_GNU_variable_value with optionally
22449 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22450 optionally after it.
22451 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22452 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22453 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22454 (variable_value_hasher::hash, variable_value_hasher::equal): New
22456 (resolve_variable_value_in_expr, resolve_variable_value,
22457 resolve_variable_values, note_variable_value_in_expr,
22458 note_variable_value): New functions.
22459 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22462 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22465 * opts.h (handle_generated_option): Add GENERATED_P argument.
22466 * opts-common.c (handle_option): Adjust function comment.
22467 (handle_generated_option): Add GENERATED_P argument, pass it to
22469 (control_warning_option): Pass false to handle_generated_option
22471 * opts.c (maybe_default_option): Pass true to handle_generated_option
22473 * optc-gen.awk: Likewise.
22475 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22477 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22478 a REG, look at the REG it is a SUBREG of.
22479 (splitter for cmpeqsi_t): Ditto.
22481 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22483 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22484 the special USEs with the pattern of the insn, not the insn itself.
22486 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22489 * doc/invoke.texi: Document -mload-store-pairs.
22491 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22492 Sandra Loosemore <sandra@codesourcery.com>
22494 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22495 argument isn't a CONST_INT.
22496 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22497 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22498 (nios2_validate_compare): Bypass alternate compare logic if *op2
22499 is not a CONST_INT.
22500 (ldstwm_operation_p): Return false if first_base is not a REG or
22501 if first_offset is not a CONST_INT.
22503 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22505 * config/cris/cris.md: Use correct operand in a define_peephole2.
22507 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22509 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22511 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22513 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22514 this_insn if it is an INSN or JUMP_INSN.
22515 (force_offsettable): Look at base, not at addr.
22516 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22517 on things that aren't necessarily CONST_INTs.
22519 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22521 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22522 -mfpmath=sse is the default also for x86-32 targets with SSE2
22523 instruction set when @option{-ffast-math} is enabled
22525 2017-02-24 Jeff Law <law@redhat.com>
22527 PR rtl-optimizatoin/79286
22528 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22531 2017-02-24 Richard Biener <rguenther@suse.de>
22533 PR tree-optimization/79389
22534 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22537 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22539 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22540 function comment to reflect reality.
22541 (loop_exits_before_overflow): Fix typo in function description.
22543 2017-02-24 Richard Biener <rguenther@suse.de>
22545 PR tree-optimization/79389
22546 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22547 properly that a threading opportunity exists. Detect conditional
22548 copy/constant propagation opportunities.
22550 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22552 * config/visium/visium.md (type): Add trap.
22553 (b): New mode attribute.
22554 (*btst): Rename into...
22555 (*btst<mode>): ...this and adjust.
22556 (*cbranchsi4_btst_insn): Rename into...
22557 (*cbranch<mode>4_btst_insn): ...this and adjust.
22558 (trap): New define_insn.
22560 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22562 PR tree-optimization/79389
22563 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22564 (noce_reversed_cond_code): New function.
22565 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22566 reversed_comparison_code. Formatting fix.
22567 (noce_try_store_flag): Test rev_cond != NULL in addition to
22568 reversed_comparison_code.
22569 (noce_try_store_flag_constants): Likewise.
22570 (noce_try_store_flag_mask): Likewise.
22571 (noce_try_addcc): Use rev_cond if non-NULL instead of
22572 reversed_comparison_code.
22573 (noce_try_cmove_arith): Likewise. Formatting fixes.
22574 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22575 (noce_find_if_block): Initialize rev_cond.
22576 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22577 instead of false as last argument never attempt to reverse it
22580 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22582 PR tree-optimization/79663
22583 * tree-predcom.c (combine_chains): Process refs in reverse order
22584 only for ZERO length chains, and add explaining comment.
22586 2017-02-23 Jeff Law <law@redhat.com>
22588 PR tree-optimization/79578
22589 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22590 in call to operand_equal_p.
22592 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22595 * config/i386/cpuid.h: Fix another undefined behavior.
22597 2017-02-23 Richard Biener <rguenther@suse.de>
22599 PR tree-optimization/79683
22600 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22601 vector types for data-refs.
22603 2017-02-23 Martin Liska <mliska@suse.cz>
22605 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22607 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22609 PR middle-end/79665
22610 * internal-fn.c (get_range_pos_neg): Moved to ...
22611 * tree.c (get_range_pos_neg): ... here. No longer static.
22612 * tree.h (get_range_pos_neg): New prototype.
22613 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22614 are known to be in between 0 and signed maximum inclusive, try to
22615 expand both unsigned and signed divmod and use the cheaper one from
22618 2017-02-22 Jeff Law <law@redhat.com>
22620 PR tree-optimization/79578
22621 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22622 to compare base operands.
22624 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22627 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22628 gpc_reg_operand instead of fpr_reg_operand.
22630 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22632 * config/mips/mips.c (mips_return_in_memory): Force FP
22633 vector types to be returned in memory for o32 ABI.
22635 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22637 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22638 instead of DW_TAG_member for static data member declarations and don't
22639 set no_linkage_name for static inline data members.
22640 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
22643 2017-02-22 Martin Liska <mliska@suse.cz>
22645 * doc/invoke.texi: Replace inequality signs with square brackets
22648 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22650 PR tree-optimization/68644
22651 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
22653 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
22656 * lra-constraints.c (simplify_operand_subreg): Handle
22657 WORD_REGISTER_OPERATIONS targets.
22659 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22662 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22663 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22664 elimination by swapping fld*.
22666 2017-02-22 Richard Biener <rguenther@suse.de>
22668 PR tree-optimization/79673
22669 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22670 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22671 irrelevant address-space qualifiers and avoiding a
22672 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22674 2017-02-22 Richard Biener <rguenther@suse.de>
22676 PR tree-optimization/79666
22677 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22678 to not symbolically negate if that may introduce undefined
22681 2017-02-22 Martin Liska <mliska@suse.cz>
22684 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22685 * data-streamer-out.c (streamer_write_gcov_count_stream):
22687 * value-prof.c (stream_out_histogram_value): Make assert more
22688 precise based on type of counter.
22690 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22693 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22694 Use nonimmediate_operand instead of memory_operand for operand 1.
22695 (float-extend standard_x87sse_constant_load splitter): Ditto.
22697 2017-02-21 Jeff Law <law@redhat.com>
22699 PR tree-optimization/79621
22700 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22701 blocks with edges to themselves.
22703 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22706 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22707 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22708 Use gimple_call_builtin_p.
22711 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22712 on temporarily removed DEBUG_INSNs.
22714 PR tree-optimization/79649
22715 * tree-loop-distribution.c (classify_partition): Give up on
22716 non-generic address space loads/stores.
22718 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22720 * doc/loop.texi (Loop manipulation): Remove nonexistent
22721 tree_ssa_loop_version from the documentation.
22722 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22724 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22727 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22728 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22729 * config/rs6000/rs6000.c: Include except.h.
22730 (rs6000_expand_split_stack_prologue): Call
22731 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22733 2017-02-21 Martin Jambor <mjambor@suse.cz>
22736 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22737 have been analyzed.
22739 2017-02-21 Martin Jambor <mjambor@suse.cz>
22741 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22742 for backward compatibility only.
22743 * doc/invoke.texi (Option Summary): Remove all references to
22744 -fipa-cp-alignment.
22746 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
22750 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22752 * lra-constraints.c (curr_insn_transform): Handle
22753 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22755 2017-02-21 Martin Liska <mliska@suse.cz>
22757 * config/i386/i386.opt: Replace -masm-dialect with -masm.
22759 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
22761 PR translation/79638
22762 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
22764 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
22767 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
22768 (arm_function_ok_for_sibcall): Return false for an indirect call by
22769 descriptor if all the argument registers are used.
22770 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
22771 alignment of the function.
22773 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22775 PR tree-optimization/61441
22776 * simplify-rtx.c (simplify_const_unary_operation): For
22777 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
22778 the sNaN unmodified.
22780 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22782 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22783 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22784 instead of SYSTEM_HEADER_DIR.
22786 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
22787 Martin Liška <mliska@suse.cz>
22789 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
22790 Fix typos and grammar, use active voice, and clarify.
22792 2017-02-20 Marek Polacek <polacek@redhat.com>
22794 PR middle-end/79537
22795 * gimplify.c (gimplify_expr): Handle unused *&&L;.
22798 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
22800 2017-02-20 Jakub Jelinek <jakub@redhat.com>
22803 * config/i386/i386.c (ix86_expand_builtin): Handle
22804 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
22805 ix86_builtins_isa[fcode].isa as a requirement of those
22806 flags and any other flag in the bitmask.
22807 (ix86_init_mmx_sse_builtins): Use 0 instead of
22808 ~OPTION_MASK_ISA_64BIT as mask.
22809 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
22810 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
22811 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
22812 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
22814 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22817 * lra-constraints.c (split_reg): Check requested split mode
22818 is supported by the register.
22820 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22822 * lra-constraints.c (simplify_operand_subreg): Remove early
22825 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22828 * lra-constraints.c (curr_insn_transform): Tighten condition
22829 for converting SUBREG reloads from OP_OUT to OP_INOUT.
22831 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22834 * lra-constraints.c (curr_insn_transform): Handle
22835 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22837 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
22840 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
22842 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
22844 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
22847 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
22850 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22852 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
22855 2017-02-18 Jakub Jelinek <jakub@redhat.com>
22857 * final.c (last_columnnum, override_columnnum): New variables.
22858 (final_start_function): Set last_columnnum, pass it to begin_prologue
22859 hook and pass 0 to dwarf2out_begin_prologue.
22860 (final_scan_insn): Update override_columnnum. Pass last_columnnum
22861 to source_line debug hook.
22862 (notice_source_line): Compute last_columnnum and for debug_column_info
22863 return true on column changes.
22864 * debug.h (struct gcc_debug_hooks): Add column argument to
22865 source_line and begin_prologue hooks.
22866 (debug_nothing_int_charstar_int_bool): Remove prototype.
22867 (debug_nothing_int_int_charstar,
22868 debug_nothing_int_int_charstar_int_bool): New prototypes.
22869 (dwarf2out_begin_prologue): Add column argument.
22870 * debug.c (do_nothing_debug_hooks): Adjust source_line and
22871 begin_prologue hooks.
22872 (debug_nothing_int_charstar_int_bool): Remove.
22873 (debug_nothing_int_int_charstar,
22874 debug_nothing_int_int_charstar_int_bool): New functions.
22875 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
22876 through to dwarf2out_source_line.
22877 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
22878 (dwarf2out_source_line): Add column argument, emit it if requested.
22879 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
22881 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22882 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22883 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
22884 through to dwarf2out_begin_prologue.
22885 (vmsdbgout_source_line): Add column argument, pass it through to
22886 dwarf2out_source_line.
22887 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
22888 dbxout_source_line caller.
22889 (dbxout_source_line): Add column argument.
22891 * common.opt (gno-column-info, gcolumn-info): New options.
22892 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
22893 (check_die): Also test for multiple DW_AT_decl_column attributes.
22894 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
22895 DW_AT_decl_column if requested.
22896 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
22898 (gen_variable_die): Likewise.
22899 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
22900 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
22903 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
22904 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
22905 (ix86_handle_option): Handle OPT_m3dnowa.
22906 * doc/invoke.texi (-m3dnowa): Document.
22907 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
22908 -m3dnowa instead of -m3dnow -march=athlon.
22911 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
22912 instead of gcc_assert for K, r and R code checks. Formatting fixes.
22914 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22917 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
22918 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
22919 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
22920 generator for vsx_xxpermdi_<mode>_be.
22921 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
22922 force big-endian semantics.
22923 (vsx_xxpermdi_<mode>_be): New define_expand with same
22924 implementation as previous version of vsx_xxpermdi_<mode>.
22926 2017-02-17 Jakub Jelinek <jakub@redhat.com>
22928 PR tree-optimization/79327
22929 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
22930 variable, its initialization and use.
22932 2017-02-17 Julia Koval <julia.koval@intel.com>
22934 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
22935 (OPTION_MASK_ISA_PKU_UNSET): New.
22936 (ix86_handle_option): Handle -mrdpid.
22937 * config/i386/cpuid.h (bit_RDPID): New.
22938 * config/i386/driver-i386.c (host_detect_local_cpu):
22939 Detect RDPID feature.
22940 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
22941 * config/i386/i386-c.c (ix86_target_macros_internal):
22943 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
22944 (ix86_valid_target_attribute_inner_p): Add "rdpid".
22945 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22946 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
22947 * config/i386/i386.md (define_insn "rdpid"): New.
22948 * config/i386/i386.opt Add -mrdpid.
22949 * config/i386/immintrin.h (_rdpid_u32): New.
22951 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
22953 PR rtl-optimization/79541
22954 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
22955 instead of transforming it into USE.
22957 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
22959 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
22960 If HONOR_SNANS (SFmode) force the input to a register.
22961 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
22962 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
22963 an frsp or similar insn.
22965 2017-02-17 Martin Liska <mliska@suse.cz>
22967 PR rtl-optimization/79577
22968 * params.def (selsched-max-sched-times): Increase minimum to 1.
22970 2017-02-17 Martin Liska <mliska@suse.cz>
22972 PR rtl-optimization/79574
22973 * gcse.c (want_to_gcse_p): Prevent integer overflow.
22975 2017-02-17 Martin Liska <mliska@suse.cz>
22977 PR tree-optimization/79529
22978 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
22979 ssa_defined_default_def_p to handle cases which are implicitly
22981 * tree-ssa.c (ssa_defined_default_def_p): New function.
22982 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
22983 which are implicitly defined.
22984 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22986 2017-02-17 Richard Biener <rguenther@suse.de>
22988 PR middle-end/79576
22989 * params.def (max-ssa-name-query-depth): Limit to 10.
22991 2017-02-17 Richard Biener <rguenther@suse.de>
22993 PR tree-optimization/79552
22994 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22997 2017-02-17 Richard Biener <rguenther@suse.de>
23000 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
23002 2017-02-17 Marek Polacek <polacek@redhat.com>
23004 PR middle-end/79536
23005 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
23006 (fold_negate_expr): New wrapper.
23008 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
23010 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
23011 Correct terminology and de-emphasize pre-standard behavior.
23013 2017-02-16 Alan Modra <amodra@gmail.com>
23015 PR rtl-optimization/79286
23016 * ira.c (def_dominates_uses): New function.
23017 (update_equiv_regs): Don't create an equivalence for insns that
23018 may trap where the register def does not dominate the use.
23020 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
23022 PR rtl-optimization/78127
23023 * lra.c (lra): Call lra_eliminate before finish the loop after
23026 2017-02-16 Richard Biener <rguenther@suse.de>
23028 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
23030 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
23031 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
23032 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
23033 (isl_val_int_from_wi): New function.
23034 (extract_affine_gmp): Rename to ...
23035 (extract_affine_wi): ... this, take a widest_int.
23036 (extract_affine_int): Just wrap extract_affine_wi.
23037 (add_param_constraints): Use isl_val_int_from_wi.
23038 (add_loop_constraints): Likewise, and extract_affine_wi.
23040 2017-02-15 Jeff Law <law@redhat.com>
23042 PR middle-end/79521
23043 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
23044 ira_init_register_move_cost_if_necessary.
23046 2017-02-15 Martin Sebor <msebor@redhat.com>
23048 PR middle-end/32003
23049 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
23050 removed in a prior commit.
23052 2017-02-15 Bin Cheng <bin.cheng@arm.com>
23054 PR tree-optimization/79347
23055 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
23056 counters during peeling.
23058 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
23060 * Makefile.in (site.exp): Remove "set ISLVER".
23062 2017-02-15 Jakub Jelinek <jakub@redhat.com>
23065 * real.c (real_from_integer): Call real_convert even for decimal.
23067 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
23070 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
23072 2017-02-14 Andrew Pinski <apinski@cavium.com>
23074 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
23075 cores and change the partno/implementer to be correct.
23076 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
23077 the 'B" as the implementer.
23078 * config/aarch64/aarch64-tune.md: Regenerate.
23080 2017-02-14 Carl Love <cel@us.ibm.com>
23082 * config/rs6000/rs6000.c: Add case statement entry to make the
23083 xvcvuxdsp built-in argument unsigned.
23084 * config/rs6000/vsx.md: Fix the source and return operand types so they
23085 match the instruction definitions from the ISA document. Fix typo
23086 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
23089 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
23092 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
23093 member early_clobber_alts.
23094 * lra-lives.c (reg_early_clobber_p): New.
23095 (process_bb_lives): Use it.
23096 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
23097 (debug_operand_data): Initialize early_clobber_alts.
23098 (setup_operand_alternative): Set up early_clobber_alts.
23099 (collect_non_operand_hard_regs): Ditto. Pass early clobber
23100 alternatives to new_insn_reg.
23101 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
23103 (lra_update_insn_regno_info): Pass the new arg.
23105 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23107 PR middle-end/79505
23108 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
23109 (new_oacc_loop_raw): Don't clear already cleared fields.
23112 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
23113 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
23114 _mm512_prefetch_i64gather_ps): New inline functions and macros.
23116 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
23119 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
23121 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
23124 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
23125 the extra instruction to the right place to store 128-bit constant
23128 2017-02-14 Martin Sebor <msebor@redhat.com>
23130 PR middle-end/79448
23131 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
23132 warning for strings of unknown length.
23134 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23136 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
23138 2017-02-14 Jeff Law <law@redhat.com>
23141 * ira-costs.c (scan_one_insn): Initialize register move costs
23142 for pseudos seen in USE/CLOBBER insns.
23144 PR tree-optimization/79095
23145 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
23146 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
23147 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
23148 if the operands are known to be not equal, then the resulting range
23150 (intersect_ranges): If the new range is ~[0,0] and the old range is
23151 wide, then prefer ~[0,0].
23152 * tree-vrp.c (overflow_comparison_p_1): New function.
23153 (overflow_comparison_p): New function.
23154 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
23155 if NAME is used in an overflow test.
23156 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
23157 overflow check that can be expressed as an equality test, then adjust
23158 ops to be that equality test.
23160 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23162 * config/s390/s390-builtin-types.def: Remove flags argument.
23163 * config/s390/s390.c (s390_init_builtins): Likewise.
23165 2017-02-14 Martin Liska <mliska@suse.cz>
23167 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
23168 vector. Fix trailing white spaces.
23170 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
23172 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
23175 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23177 PR rtl-optimization/68664
23178 * config/arm/arm.c (arm_sched_can_speculate_insn):
23179 New function. Declare prototype.
23180 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23182 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23184 PR rtl-optimization/68664
23185 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
23187 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23189 2017-02-14 Amit Pawar <amit.pawar@amd.com>
23191 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
23192 max skip bytes for function, loop and jump.
23194 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23196 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
23197 ABS_EXPR for gimple dump.
23199 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23202 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
23204 PR tree-optimization/79408
23205 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
23206 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
23207 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
23208 also if rhs1 is INTEGER_CST.
23210 2017-02-14 Richard Biener <rguenther@suse.de>
23212 PR middle-end/79432
23213 * tree-into-ssa.c (insert_phi_nodes): When the function can
23214 have abnormal edges rewrite SSA names with broken use-def
23215 dominance out of SSA and register them for PHI insertion.
23217 2017-02-13 Martin Sebor <msebor@redhat.com>
23219 PR middle-end/79496
23220 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
23221 clearing info.nowrite flag when snprintf size argument is a range.
23223 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23225 * cprop.c (cprop_jump): Add missing space in string literal.
23226 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
23227 (get_constraint_for_component_ref): Likewise.
23228 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
23229 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
23230 * lra-constraints.c (process_alt_operands): Likewise.
23231 * ipa-inline.c (inline_small_functions): Likewise.
23232 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
23233 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23234 * trans-mem.c (diagnose_tm_1_op): Likewise.
23235 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
23236 (grid_parallel_clauses_gridifiable): Likewise.
23238 * config/nvptx/mkoffload.c (process): Add space in between
23241 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
23242 "MOD4_SSE_REGS" and "ALL_REGS".
23244 * spellcheck.c (test_data): Add , in between "foo" and "food".
23246 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23249 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
23250 boundary crossing check and subsequent code generation agree.
23252 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23254 * config/aarch64/aarch64.c (has_memory_op): Delete.
23255 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
23258 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23260 PR rtl-optimization/79388
23261 PR rtl-optimization/79450
23262 * combine.c (distribute_notes): When removing TEM_INSN for which
23263 corresponding dest has last value recorded, invalidate that last
23266 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23268 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
23269 of explicit '@'. Add missing assembly comment marker on branch costs
23272 2017-02-13 Nathan Sidwell <nathan@acm.org>
23274 * gengtype-lex.l (<in_struct>): Add '/'.
23276 2017-02-13 Martin Liska <mliska@suse.cz>
23279 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
23281 2017-02-13 Richard Biener <rguenther@suse.de>
23283 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
23285 * configure: Re-generate.
23286 * config.in: Likewise.
23287 * graphite-dependences.c: Simplify as if
23288 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
23289 * graphite-isl-ast-to-gimple.c: Likewise.
23290 * graphite-optimize-isl.c: Likewise.
23291 * graphite-poly.c: Likewise.
23292 * graphite-sese-to-poly.c: Likewise.
23293 * graphite.h: Likewise.
23294 * toplev.c: Include isl/version.h and use isl_version () for
23295 printing the ISL version.
23296 * doc/install.texi: Update ISL requirement.
23298 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23300 * doc/standards.texi (Standards): Update reference to
23303 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23305 * doc/extend.texi (Named Address Spaces): sourceware.org now
23307 * doc/install.texi (Binaries): Ditto.
23310 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23312 * doc/cpp.texi: Replace "stringify"/"stringification" with C
23313 standard terminology "stringize"/"stringizing" throughout.
23314 * doc/cppinternals.texi: Likewise.
23316 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23318 * doc/extend.texi: Fix some spelling mistakes and typos.
23319 * doc/invoke.texi: Likewise.
23321 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23324 * params.def (inline-min-speedup) Change from 10 to 8.
23326 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23328 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
23331 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23334 * ipa-inline-analysis.c (get_minimal_bb): New function.
23335 (record_modified): Use it.
23336 (remap_edge_change_prob): Handle also ancestor functions.
23338 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
23340 * doc/contrib.texi (Contributors): Remove broken link into
23341 the Mauve CVS repository.
23343 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23345 PR middle-end/79454
23346 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
23347 result computation whenever lhs doesn't have vector mode, not
23348 just when it has BLKmode.
23350 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23352 * doc/makefile.texi (profiledbootstrap): Refer to the
23353 installation instructions only in textual form.
23355 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23358 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
23360 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23362 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
23363 (Specific): Update mingw-w64 reference.
23365 (Specific): Remove broken link to Renesas RX processor.
23367 2017-02-10 Richard Biener <rguenther@suse.de>
23369 * toplev.c (process_options): Do not mention obsolete graphite
23370 options when printing sorry message about missing graphite support.
23371 Mention -floop-nest-optimize.
23373 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
23375 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
23376 (vtst_p16): Likewise.
23377 (vtstq_p8): Likewise.
23378 (vtstq_p16): Likewise.
23380 (vtstq_p64): Likewise.
23381 * config/arm/arm_neon.h (vgetq_lane_p64): New.
23382 (vset_lane_p64): New.
23383 (vsetq_lane_p64): New.
23385 2017-02-10 Jakub Jelinek <jakub@redhat.com>
23387 PR tree-optimization/79411
23388 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
23389 stmt operands are SSA_NAMEs used in abnormal phis.
23390 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
23393 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23396 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
23399 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23401 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23403 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23405 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23409 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23410 not arbitrary TREE_CONSTANT.
23413 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23414 "omp declare target link" attribute unless is_global_var.
23415 * omp-offload.c (find_link_var_op): Likewise.
23417 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23418 Chung-Lin Tang <cltang@codesourcery.com>
23420 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23422 (gimplify_adjust_omp_clauses): Don't delete TILE.
23423 (gimplify_omp_for): Deal with TILE.
23424 * internal-fn.c (expand_GOACC_TILE): New function.
23425 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23427 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23428 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23430 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23431 avoid DIV for outermost collapse var.
23432 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23433 Remove out of date comments, fix whitespace.
23434 * omp-general.c (omp_extract_for_data): Deal with tiling.
23435 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23436 adjust OLF_DIM_BASE value.
23437 (struct omp_for_data): Add tiling field.
23438 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23439 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23440 for auto loops. Remove default auto determining, moved to
23441 oacc_loop_fixed_partitions.
23442 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23443 stmts, add e_mask field.
23444 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23445 (oacc_thread_numbers): Use oacc_dim_call.
23446 (oacc_xform_tile): New.
23447 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23448 (finish_oacc_loop): Adjust for ifns vector.
23449 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23450 add case for GOACC_TILE fns.
23451 (oacc_loop_xform_loop): Delete.
23452 (oacc_loop_process): Iterate over call list directly, and add
23453 handling for GOACC_TILE fns.
23454 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23456 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23457 vector partitioning to outer loops. Assign 2 partitions to loops
23458 when available. Add TILE handling.
23459 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23460 (execite_oacc_device_lower): Process GOACC_TILE fns,
23461 ignore unknown specs.
23462 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23463 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23464 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23466 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23468 * configure.ac (ACX_BUGURL): Update.
23469 * configure: Regenerate.
23471 2017-02-09 Richard Biener <rguenther@suse.de>
23473 PR tree-optimization/69823
23474 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23475 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23477 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23479 * config/arc/arc-c.def: Add __NPS400__ definition.
23480 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23481 (TARGET_NPS400): Define.
23483 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23485 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23487 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23488 pointer, arch_info.
23489 (arc_cpu_types): Fill the arch_info field with a pointer into the
23490 arc_arch_types table.
23491 (arc_selected_cpu): Declare.
23492 * config/arc/arc.c (arc_selected_cpu): Make global.
23493 (arc_selected_arch): Delete.
23494 (arc_base_cpu): Delete.
23495 (arc_override_options): Remove references to deleted variables,
23496 update access to arch information.
23497 (ARC_OPT): Update access to arch information.
23498 (ARC_OPTX): Likewise.
23499 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23500 (TARGET_ARC600): Update access to arch information.
23501 (TARGET_ARC601): Likewise.
23502 (TARGET_ARC700): Likewise.
23503 (TARGET_EM): Likewise.
23504 (TARGET_HS): Likewise.
23505 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23508 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23511 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23512 condition/operands for integer GE/LE/GEU/LEU operations.
23514 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23516 PR translation/79397
23517 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23520 2017-02-08 Martin Jambor <mjambor@suse.cz>
23523 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23524 whether allocation happened.
23525 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23526 nothing was allocated.
23528 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23530 PR tree-optimization/79408
23531 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23532 constant, but SSA_NAME with a known integer range, use the minimum
23533 of that range instead of op1 to determine if modulo can be replaced
23534 with its first operand.
23536 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23538 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23540 2017-02-08 Richard Biener <rguenther@suse.de>
23542 PR tree-optimization/71824
23543 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23544 Check all loops contained in the merged region.
23546 2017-02-07 Andrew Pinski <apinski@cavium.com>
23548 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23550 2017-02-07 Andrew Pinski <apinski@cavium.com>
23552 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23553 (thunderxt88): Likewise.
23554 (thunderxt81): Disable LSE and change v8.1 to v8.
23555 (thunderxt83): Likewise.
23557 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23558 Richard Biener <rguenther@suse.de>
23560 PR middle-end/79399
23561 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23562 type from int to size_t.
23563 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23565 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23567 PR rtl-optimization/79386
23568 * cprop.c (bypass_conditional_jumps): Initialize
23569 bypass_last_basic_block already before splitting bbs after
23570 unconditional traps...
23571 (bypass_conditional_jumps): ... rather than here.
23574 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23575 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23576 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23577 fix -masm=intel patterns.
23579 2017-02-07 Richard Biener <rguenther@suse.de>
23581 PR tree-optimization/79256
23582 PR middle-end/79278
23583 * builtins.c (get_object_alignment_2): Use min_align_of_type
23584 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23585 and ADJUST_FIELD_ALIGN.
23587 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23589 * doc/tm.texi: Regenerate.
23590 * stor-layout.c (layout_decl): Adjust.
23591 (update_alignment_for_field): Likewise.
23592 (place_field): Likewise.
23593 (min_align_of_type): Likewise.
23594 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23595 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23596 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23597 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23598 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23599 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23600 * config/i386/i386.c (x86_field_alignment): Likewise.
23601 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23602 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23603 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23604 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23605 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23606 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23610 2017-01-30 Richard Biener <rguenther@suse.de>
23612 PR tree-optimization/79256
23613 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23614 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23617 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23619 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23620 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23621 builtins to SImode and emit a zero-extend, if necessary.
23623 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23625 * docs/invoke.texi (RISC-V Options): Alphabetize.
23627 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23629 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23632 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23634 * config/riscv/riscv.c: New file.
23635 * common/config/riscv/riscv-common.c: Likewise.
23636 * config.gcc: Likewise.
23637 * config/riscv/constraints.md: Likewise.
23638 * config/riscv/elf.h: Likewise.
23639 * config/riscv/generic.md: Likewise.
23640 * config/riscv/linux.h: Likewise.
23641 * config/riscv/multilib-generator: Likewise.
23642 * config/riscv/peephole.md: Likewise.
23643 * config/riscv/pic.md: Likewise.
23644 * config/riscv/predicates.md: Likewise.
23645 * config/riscv/riscv-builtins.c: Likewise.
23646 * config/riscv/riscv-c.c: Likewise.
23647 * config/riscv/riscv-ftypes.def: Likewise.
23648 * config/riscv/riscv-modes.def: Likewise.
23649 * config/riscv/riscv-opts.h: Likewise.
23650 * config/riscv/riscv-protos.h: Likewise.
23651 * config/riscv/riscv.h: Likewise.
23652 * config/riscv/riscv.md: Likewise.
23653 * config/riscv/riscv.opt: Likewise.
23654 * config/riscv/sync.md: Likewise.
23655 * config/riscv/t-elf-multilib: Likewise.
23656 * config/riscv/t-linux: Likewise.
23657 * config/riscv/t-linux-multilib: Likewise.
23658 * config/riscv/t-riscv: Likewise.
23659 * configure.ac: Likewise.
23660 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23661 Waterman as RISC-V maintainers.
23662 * doc/install.texi: Add RISC-V entries.
23663 * doc/invoke.texi: Add RISC-V options section.
23664 * doc/md.texi: Add RISC-V constraints section.
23665 * configure: Regenerated.
23667 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23670 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23671 false values to be constant vectors with all 0 or all 1 bits set.
23672 (vcondu<mode><mode>): Likewise.
23673 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23675 (fpmask_comparison_operator): Update comment.
23676 (vecint_comparison_operator): New predicate.
23677 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23678 vector conditionals when the true and false values are constant
23679 vectors with all 0 bits or all 1 bits set.
23681 2017-02-06 Martin Sebor <msebor@redhat.com>
23683 PR tree-optimization/79376
23684 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23686 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23688 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23689 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23690 to simplify split condition.
23692 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23694 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23695 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23698 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23700 PR rtl-optimization/68664
23701 * target.def (can_speculate_insn): New hook.
23702 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23703 * doc/tm.texi: Regenerate.
23704 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23705 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23706 (rs6000_sched_can_speculate_insn): New function.
23708 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23710 PR tree-optimization/79284
23711 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23712 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23713 vectorizable_mask_load_store, vectorizable_operation,
23714 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23715 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23716 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23717 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23718 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23719 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23720 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23721 is_gimple_assign (stmt). Replace another such test with
23722 is_gimple_assign (stmt).
23724 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23727 * config/avr/avr.c (rtl-iter.h): Include it.
23728 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23729 (avr_legitimate_combined_insn): ...and implementation.
23731 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23733 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23734 * config/s390/s390.c (s390_const_operand_ok)
23735 (s390_canonicalize_comparison, s390_extract_part)
23736 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23737 (s390_contiguous_bitmask_p, s390_rtx_costs)
23738 (legitimize_pic_address): Likewise.
23739 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23740 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23741 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23742 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23743 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
23745 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23747 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
23748 REGNO($0) == REGNO($1).
23750 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23752 * config/s390/linux.h(SIZE_TYPE): Add comment.
23754 2017-02-06 Julian Brown <julian@codesourcery.com>
23755 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23756 Virendra Pathak <virendra.pathak@broadcom.com>
23758 * config/aarch64/aarch64-cores.def: Change the scheduler
23760 * config/aarch64/aarch64.md: Include thunderx2t99.md.
23761 * config/aarch64/thunderx2t99.md: New file.
23763 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23765 * doc/standards.texi (Go Language): Update link to language
23768 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23770 * tree-eh.c (lower_resx): Sanitize profile.
23771 (cleanup_empty_eh_move_lp): Likewise.
23773 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23776 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
23778 * cfgloopmanip.h (loop_version): Update prototype.
23779 * modulo-sched.c (sms_schedule): Update call of loop_version.
23780 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
23781 * tree-parloops.c (gen_parallel_loop): Likewise.
23782 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
23783 * tree-ssa-loop-split.c (split_loop): Likewise.
23784 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
23785 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23787 2017-02-05 Martin Liska <mliska@suse.cz>
23790 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
23792 (print_operand_address): Initialize a struct to zero.
23794 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23796 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
23797 garbage collector only in textual form.
23799 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23801 * doc/extend.texi (x86 specific memory model extensions for
23802 transactional memory): Simplify a phrase.
23804 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
23807 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
23808 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
23809 (atomic_storedi_1): Likewise.
23811 2017-02-04 Jakub Jelinek <jakub@redhat.com>
23813 PR tree-optimization/79338
23814 * tree-parloops.c (gather_scalar_reductions): Don't call
23815 vect_analyze_loop_form for loop->inner before destroying loop's
23818 2017-02-03 Martin Sebor <msebor@redhat.com>
23820 PR tree-optimization/79327
23821 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
23822 when precision has resulted in leading zeros.
23823 (format_integer): Adjust the likely counter to assume an unknown
23824 argument that may be zero is non-zero.
23826 2017-02-03 Jason Merrill <jason@redhat.com>
23829 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
23830 avoid copying non-taken branch.
23832 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23834 PR tree-optimization/79340
23835 * tree-vect-loop.c (vectorizable_reduction): Release
23836 vec_defs elements after safe_splicing them into other vectors.
23839 PR tree-optimization/79327
23840 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
23841 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
23843 (format_integer): Use wide_int_to_tree instead of build_int_cst
23844 + to_?hwi. If argmin is NULL, just set argmin and argmax to
23845 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
23846 of shortest and longest sequence.
23848 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
23850 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
23851 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
23853 2017-02-03 Walter Lee <walt@tilera.com>
23856 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
23857 after initial stackframe link reg save.
23858 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
23860 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23863 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
23864 wu for stxssp alternative.
23866 2017-02-03 Martin Sebor <msebor@redhat.com>
23868 PR tree-optimization/79352
23869 * gimple-fold.c (get_range_strlen): Add argument.
23870 (get_range_strlen): Change return type to bool.
23871 (get_maxval_strlen): Pass in a dummy argument.
23872 * gimple-fold.h (get_range_strlen): Change return type to bool.
23873 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
23874 * tree.h (array_at_struct_end_p): Add argument.
23875 * tree.c (array_at_struct_end_p): Handle it.
23877 2017-02-03 Martin Liska <mliska@suse.cz>
23880 * multiple_target.c (create_dispatcher_calls): Redirect edge
23881 from a caller of a dispatcher.
23882 (expand_target_clones): Make the clones local.
23883 (ipa_target_clone): Do both target clones and resolvers.
23884 (ipa_dispatcher_calls): Remove the pass.
23885 (pass_dispatcher_calls::gate): Likewise.
23886 (make_pass_dispatcher_calls): Likewise.
23887 * passes.def (pass_target_clone): Put as very first IPA early
23890 2017-02-03 Martin Liska <mliska@suse.cz>
23892 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
23893 in case of a function with ifunc attribute.
23895 2017-02-03 Martin Liska <mliska@suse.cz>
23897 * cgraph.c (cgraph_node::dump): Dump function version info.
23898 * symtab.c (symtab_node::dump_base): Add missing new line.
23900 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23902 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
23903 (ifcombine_ifandif): Use it.
23905 2017-02-03 Martin Liska <mliska@suse.cz>
23907 * doc/invoke.texi: Document default value for
23908 use-after-scope-direct-emission-threshold.
23910 2017-02-03 Martin Liska <mliska@suse.cz>
23912 PR tree-optimization/79339
23913 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
23914 (format_floating): Likewise.
23916 2017-02-03 Martin Liska <mliska@suse.cz>
23919 * ipa-prop.c (ipa_node_params_t::insert): Remove current
23921 (ipa_node_params_t::remove): Likewise.
23922 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
23923 initialization from removed ipa_node_params_t::insert.
23924 (ipa_node_params::~ipa_node_params): Move from removed
23925 ipa_node_params_t::release.
23926 * symbol-summary.h (symbol_summary::m_released): New member.
23927 Do not release a summary twice. Do not allow to call finalizer
23928 for types of a summary that live in GGC memory.
23930 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23932 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
23935 2017-02-02 Martin Sebor <msebor@redhat.com>
23937 PR middle-end/79275
23938 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
23939 (format_string): Tighten up the range of output for non-constant
23940 strings and correct the expected range for wide non-constant strings.
23942 2017-02-02 Martin Sebor <msebor@redhat.com>
23944 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
23946 PR middle-end/32003
23947 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
23949 (-fdump-tree-@var): Add to index and document how to come up
23950 with pass-specific option and dump file names.
23951 (-fdump-passes): Clarify where to look for output.
23953 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23955 PR middle-end/77445
23956 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
23957 statistics of the analyzed path; allow threading for speed when
23958 any of BBs along the path are optimized for speed.
23960 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
23962 PR middle-end/78468
23963 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
23964 settings of the virtual registers.
23967 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
23969 * explow.c (get_dynamic_stack_size): Take known alignment of stack
23970 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
23973 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23975 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
23976 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
23978 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23980 * config/s390/s390.md: Add missing comments with the expanded
23982 * config/s390/vector.md: Likewise.
23983 * config/s390/vx-builtins.md: Likewise.
23985 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23988 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23989 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23990 conditions on a single line.
23992 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23994 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23995 __S390_VX__ to __VX__.
23997 2017-02-01 Andrew Pinski <apinski@cavium.com>
23999 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
24000 stmt_info to record_stmt_cost.
24001 (vect_get_known_peeling_cost): Pass stmt_info if known to
24003 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
24004 cpu_vector_cost field into
24005 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
24006 field into vec_int_stmt_cost and vec_fp_stmt_cost.
24007 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
24008 splitting of scalar_stmt_cost and vec_stmt_cost.
24009 (thunderx_vector_cost): Likewise.
24010 (cortexa57_vector_cost): LIkewise.
24011 (exynosm1_vector_cost): Likewise.
24012 (xgene1_vector_cost): Likewise.
24013 (thunderx2t99_vector_cost): Improve after the splitting of the two
24015 (aarch64_builtin_vectorization_cost): Update for the splitting of
24016 scalar_stmt_cost and vec_stmt_cost.
24018 2017-02-01 Torvald Riegel <triegel@redhat.com>
24019 Richard Henderson <rth@redhat.com>
24021 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
24022 conditional on existance of a fast atomic load.
24023 * optabs-query.c (can_atomic_load_p): New function.
24024 * optabs-query.h (can_atomic_load_p): Declare it.
24025 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
24026 no fast atomic load is available for the particular size of access.
24027 (expand_atomic_compare_and_swap): Likewise.
24028 (expand_atomic_load): Likewise.
24029 (expand_atomic_store): Likewise.
24030 (expand_atomic_fetch_op): Likewise.
24031 * testsuite/lib/target-supports.exp
24032 (check_effective_target_sync_int_128): Remove x86 because it provides
24033 no fast atomic load.
24034 (check_effective_target_sync_int_128_runtime): Likewise.
24036 2017-02-01 Richard Biener <rguenther@suse.de>
24038 * graphite.c: Include tree-vectorizer.h for find_loop_location.
24039 (graphite_transform_loops): Provide opt-info for optimized nests.
24040 * tree-parloop.c (parallelize_loops): Provide opt-info for
24041 parallelized loops.
24043 2017-02-01 Richard Biener <rguenther@suse.de>
24045 PR middle-end/79315
24046 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
24047 was not set before.
24049 2017-02-01 Richard Biener <rguenther@suse.de>
24051 PR tree-optimization/71824
24052 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24053 Verify the loops are valid in the merged SESE region.
24054 (scop_detection::can_represent_loop_1): Check analyzing the
24055 evolution of the number of iterations in the region succeeds.
24057 2017-01-31 Ian Lance Taylor <iant@golang.org>
24059 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
24060 REG_ARGS_SIZE note to 32-bit push insns and call insn.
24062 2017-01-31 David Malcolm <dmalcolm@redhat.com>
24064 PR preprocessor/79210
24065 * input.c (get_substring_ranges_for_loc): Replace line_width
24066 assertion with error-handling.
24068 2017-01-31 Richard Biener <rguenther@suse.de>
24070 PR tree-optimization/77318
24071 * graphite-sese-to-poly.c (extract_affine): Fix assert.
24072 (create_pw_aff_from_tree): Take loop parameter.
24073 (add_condition_to_pbb): Pass loop of the condition to
24074 create_pw_aff_from_tree.
24076 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24078 * config/s390/s390.c (s390_asan_shadow_offset): New function.
24079 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
24081 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
24085 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
24087 (convert_int_to_float128): Likewise.
24088 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
24089 (convert_int_to_float128): Likewise.
24090 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
24091 (UNSPEC_IEEE128_CONVERT): Likewise.
24092 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
24093 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
24094 Use local variables for IBM extended format.
24095 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
24096 (fix_trunc<mode>si2_fprs): Likewise.
24097 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
24098 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
24099 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
24100 to know that we can now have integers of all sizes in vector
24102 (fix<uns>_<mode>di2_hw): Likewise.
24103 (float<uns>_<mode>si2_hw): Likewise.
24104 (fix_<mode>si2_hw): Likewise.
24105 (fixuns_<mode>si2_hw): Likewise.
24106 (float<uns>_<mode>di2_hw): Likewise.
24107 (float_<mode>di2_hw): Likewise.
24108 (float_<mode>si2_hw): Likewise.
24109 (floatuns_<mode>di2_hw): Likewise.
24110 (floatuns_<mode>si2_hw): Likewise.
24111 (xscvqp<su>wz_<mode>): Delete, no longer used.
24112 (xscvqp<su>dz_<mode>): Likewise.
24113 (xscv<su>dqp_<mode>): Likewise.
24114 (ieee128_mfvsrd_64bit): Likewise.
24115 (ieee128_mfvsrd_32bit): Likewise.
24116 (ieee128_mfvsrwz): Likewise.
24117 (ieee128_mtvsrw): Likewise.
24118 (ieee128_mtvsrd_64bit): Likewise.
24119 (ieee128_mtvsrd_32bit): Likewise.
24121 2017-01-31 Martin Liska <mliska@suse.cz>
24124 * ipa-prop.c (ipa_free_all_node_params): Call release method
24125 instead of ~sumbol_summary to not to trigger double times
24128 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
24130 PR tree-optimization/71691
24131 * bitmap.h (class auto_bitmap): New.
24132 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
24133 is_maybe_undefined instead of ssa_undefined_value_p.
24135 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24137 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24138 __S390_ARCH_LEVEL__ to __ARCH__.
24140 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24142 PR tree-optimization/79267
24143 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
24144 if should_remove_lhs_p is true.
24146 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
24149 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
24150 (add_alignment_attribute): New.
24151 (base_type_die): Add alignment attribute.
24152 (subrange_type_die): Likewise.
24153 (modified_type_die): Likewise.
24154 (gen_array_type_die): Likewise.
24155 (gen_descr_array_type_die: Likewise.
24156 (gen_enumeration_type_die): Likewise.
24157 (gen_subprogram_die): Likewise.
24158 (gen_variable_die): Likewise.
24159 (gen_field_die): Likewise.
24160 (gen_ptr_to_mbr_type_die): Likewise.
24161 (gen_struct_or_union_type_die): Likewise.
24162 (gen_subroutine_type_die): Likewise.
24163 (gen_typedef_die): Likewise.
24164 (base_type_cmp): Compare alignment attribute.
24166 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24169 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
24170 (setb_unsigned) New pattern for setb with CCUNS.
24171 * config/rs6000/rs6000.c (expand_block_compare): Use a different
24172 subfc./subfe sequence to avoid overflow problems. Generate a
24173 shorter sequence with cmpld/setb for power9.
24174 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
24175 for generating subfc. instruction.
24176 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
24177 now uses this instruction.
24179 2017-01-30 Ian Lance Taylor <iant@google.com>
24182 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
24183 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
24185 2017-01-30 Martin Sebor <msebor@redhat.com>
24187 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
24188 Move constant to the right of a relational operator.
24189 (get_mpfr_format_length, format_character, format_string): Ditto.
24190 (should_warn_p, maybe_warn): Same.
24192 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
24194 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
24197 * asan.c (get_translation_unit_decl): Remove function.
24198 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
24200 2017-01-30 Martin Liska <mliska@suse.cz>
24202 PR gcov-profile/79259
24203 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
24204 -fprofile-generate.
24206 2017-01-30 Martin Liska <mliska@suse.cz>
24209 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
24210 Initialize variables with NULL value.
24212 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
24215 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
24217 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
24219 2017-01-30 Richard Biener <rguenther@suse.de>
24221 PR tree-optimization/79276
24222 * tree-vrp.c (process_assert_insertions): Properly adjust common
24223 when removing a duplicate.
24225 * gcc.dg/torture/pr79276.c: New testcase.
24227 2017-01-30 Richard Biener <rguenther@suse.de>
24229 PR tree-optimization/79256
24230 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24231 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24233 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
24235 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
24238 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
24239 ("*r<noxa>sbg_<mode>_sll_bitmask")
24240 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
24241 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
24242 Use contiguous_bitmask_nowrap_operand.
24244 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24247 * config/rs6000/altivec.h (vec_xl): Revise #define.
24248 (vec_xst): Likewise.
24250 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
24252 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
24254 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
24256 PR rtl-optimization/79194
24257 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
24258 traps before call to bypass_conditional_jumps.
24260 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24262 PR tree-optimization/71374
24263 * lra-constraints.c (check_conflict_input_operands): New.
24264 (match_reload): Use it.
24266 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24269 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
24270 account to calculate conflict_set.
24272 2017-01-27 Bin Cheng <bin.cheng@arm.com>
24274 PR rtl-optimization/78559
24275 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
24276 other_insn in combine.
24278 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24280 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
24281 uint16_type_node for BT_UINT16.
24283 2017-01-27 David Malcolm <dmalcolm@redhat.com>
24285 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
24286 "RTL Tests" to menu.
24287 (GIMPLE Tests): New node.
24288 (RTL Tests): New node.
24290 2017-01-27 Richard Biener <rguenther@suse.de>
24292 PR tree-optimization/79245
24293 * tree-loop-distribution.c (distribute_loop): Apply cost
24294 modeling also to detected patterns.
24296 2017-01-27 Richard Biener <rguenther@suse.de>
24298 PR tree-optimization/71433
24299 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
24300 (compare_assert_loc): New function.
24301 (process_assert_insertions): Sort and optimize assert locations
24302 to remove duplicates and push down identical assertions on
24303 edges to their destination block.
24305 2017-01-27 Richard Biener <rguenther@suse.de>
24307 PR tree-optimization/79244
24308 * tree-vrp.c (remove_range_assertions): Forcefully propagate
24309 out SSA names even if abnormal.
24311 2017-01-27 Jakub Jelinek <jakub@redhat.com>
24313 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
24314 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
24315 instead of MPFR_RNDN.
24317 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
24320 * arm.c (arm_option_override): Don't call build_target_option_node
24321 until after doing all option overrides.
24322 (arm_valid_target_attribute_tree): Likewise.
24324 2017-01-27 Martin Liska <mliska@suse.cz>
24326 * doc/invoke.texi (-fprofile-arcs): Document profiling support
24327 for {cd}tors and C++ {cd}tors.
24329 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
24331 * config/s390/s390.md ("*setmem_long_and")
24332 ("*setmem_long_and_31z"): Use zero_extend instead of and.
24334 2017-01-26 Martin Sebor <msebor@redhat.com>
24336 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
24339 2017-01-26 Martin Sebor <msebor@redhat.com>
24341 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
24342 HAVE_DFmode before using XFmode or DFmode.
24343 (parse_directive): Avoid using the z length modifier to avoid
24344 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
24346 PR middle-end/78703
24347 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
24348 to accept adjustment as an array.
24349 (get_int_range): New function.
24350 (struct directive): Make width and prec arrays.
24351 (directive::set_width, directive::set_precision): Call get_int_range.
24352 (format_integer, format_floating): Handle width and precision ranges.
24353 (format_string, parse_directive): Same.
24355 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24358 * dwarf2out.c (generate_skeleton_bottom_up): For children with
24359 comdat_type_p set, just clone them, but keep the children in the
24363 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
24364 which have direct callers with -fvar-tracking-assignments enabled
24366 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
24367 inside of type units.
24369 2017-01-26 Martin Sebor <msebor@redhat.com>
24371 PR middle-end/78703
24372 * gimple-ssa-sprintf.c (struct result_range): Add likely and
24374 (struct format_result): Replace number_chars, number_chars_min,
24375 and number_chars_max with a single member of struct result_range.
24377 (format_result::operator+=): Adjust.
24378 (struct fmtresult): Remove bounded. Handle likely and unlikely
24380 (fmtresult::adjust_for_width_or_precision): New function.
24381 (fmtresult:type_max_digits): New function.
24382 (bytes_remaining): Handle likely and unlikely counters.
24383 (min_bytes_remaining): Remove.
24384 (format_percent): Simplify.
24385 (format_integer, format_floating): Set likely and unlikely counters.
24386 (get_string_length, format_character, format_string): Same.
24387 (format_plain, should_warn_p): New function.
24388 (maybe_warn): Call should_warn_p. Update diagnostic messages
24389 and handle those for all directives, including plain strings.
24390 (format_directive): Handle likely and unlikely counters.
24391 Remove unnecessary quoting from diagnostics. Add an informational
24393 (add_bytes): Remove.
24394 (pass_sprintf_length::compute_format_length): Simplify.
24395 (try_substitute_return_value): Handle likely and unlikely counters.
24397 2017-01-26 Carl Love <cel@us.ibm.com>
24399 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
24400 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24402 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24405 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24406 endianess for subregs into account.
24407 * lra-constraints.c (lra_constraints): Do risky transformations
24408 always on the first iteration.
24409 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24410 last_call_used_reg_set.
24411 (process_bb_lives): Define and use last_call_used_reg_set.
24412 * lra.c (lra): Always continue after lra_constraints on the first
24415 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24417 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24419 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24421 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24423 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24424 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24425 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24426 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24427 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24428 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24429 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24430 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24431 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24433 2017-01-26 Marek Polacek <polacek@redhat.com>
24436 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24437 for the third operand.
24439 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24441 PR middle-end/79236
24442 * omp-low.c (struct omp_context): Add simt_stmt field.
24443 (scan_omp_for): Return omp_context *.
24444 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24445 context to the _simt_ SIMD stmt.
24446 (lower_omp_for): For combined SIMD with sibling _simt_
24447 SIMD, make sure to use the same decls in _looptemp_
24448 clauses as in the sibling.
24450 2017-01-26 David Sherwood <david.sherwood@arm.com>
24452 PR middle-end/79212
24453 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24456 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24459 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24460 emit fld b; fld a; if possible.
24462 * brig-builtins.def: Update copyright years.
24463 * config/arm/arm_acle_builtins.def: Update copyright years.
24465 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24468 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24469 constraint instead of o for the stxsd instruction.
24471 2017-01-25 Carl Love <cel@us.ibm.com>
24473 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24474 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24476 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24478 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24480 2017-01-25 Richard Biener <rguenther@suse.de>
24482 PR tree-optimization/69264
24483 * target.def (vector_alignment_reachable): Improve documentation.
24484 * doc/tm.texi: Regenerate.
24485 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24487 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24488 earlier changes with respect to TYPE_USER_ALIGN.
24489 (vector_alignment_reachable_p): Likewise. Improve dumping.
24491 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24494 * config/arm/arm.md (xordi3): Force constant operand into a register
24497 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24499 * doc/invoke.texi (-fstore-merging): Correct default optimization
24500 levels at which it is enabled.
24501 (-O): Move -fstore-merging from list to...
24504 2017-01-25 Richard Biener <rguenther@suse.de>
24507 * omp-expand.c: Include debug.h.
24508 (expand_omp_taskreg): Make sure to generate early debug before
24509 outlining anything from a function.
24510 (expand_omp_target): Likewise.
24511 (grid_expand_target_grid_body): Likewise.
24513 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24516 * asan.c (get_translation_unit_decl): New function.
24517 (asan_add_global): Extract modules file name from globals
24518 TRANSLATION_UNIT_DECL name.
24520 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24523 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24524 for long calls with APCS frame and VFP.
24526 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24528 * cfg.c (original_copy_tables_initialized_p): New function.
24529 * cfg.h (original_copy_tables_initialized_p): New decl.
24530 * cfgrtl.c (relink_block_chain): Guard the call to
24531 free_original_copy_tables with a call to
24532 original_copy_tables_initialized_p.
24533 * cgraph.h (symtab_node::native_rtl_p): New decl.
24534 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24535 (symtab_node::needed_p): Don't assert for early assembly output
24536 for __RTL functions.
24537 (cgraph_node::finalize_function): Set "force_output" for __RTL
24539 (cgraph_node::analyze): Bail out early for __RTL functions.
24540 (analyze_functions): Update assertion to support __RTL functions.
24541 (cgraph_node::expand): Bail out early for __RTL functions.
24542 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24544 * function.h (struct function): Update comment for field
24546 * gimple-expr.c: Include "tree-pass.h".
24547 (gimple_has_body_p): Return false for __RTL functions.
24548 * Makefile.in (OBJS): Add run-rtl-passes.o.
24549 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24551 (gcc::pass_manager::get_clean_slate): New accessor.
24552 * passes.c: Include "insn-addr.h".
24553 (should_skip_pass_p): Add logging. Update logic for running
24554 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24555 property-provider override so it is only done for gimple passes.
24557 (skip_pass): New function.
24558 (execute_one_pass): Call skip_pass when skipping passes.
24559 * read-md.c (md_reader::read_char): Support filtering
24560 the input to a subset of line numbers.
24561 (md_reader::md_reader): Initialize fields
24562 m_first_line and m_last_line.
24563 (md_reader::read_file_fragment): New function.
24564 * read-md.h (md_reader::read_file_fragment): New decl.
24565 (md_reader::m_first_line): New field.
24566 (md_reader::m_last_line): New field.
24567 * read-rtl-function.c (function_reader::create_function): Only
24568 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24569 curr_properties. Set DECL_INITIAL to a dummy block.
24570 (read_rtl_function_body_from_file_range): New function.
24571 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24573 * run-rtl-passes.c: New file.
24574 * run-rtl-passes.h: New file.
24576 2017-01-24 Jeff Law <law@redhat.com>
24578 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24581 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24583 PR tree-optimization/79159
24584 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24585 (record_nonwrapping_iv): Improve boundary using above function if no
24586 value range information.
24588 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24589 Martin Jambor <mjambor@suse.cz>
24591 * brig-builtins.def: New file.
24592 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24593 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24594 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24595 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24596 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24597 * builtin-types.def (BT_INT8): New.
24598 (BT_INT16): Likewise.
24599 (BT_UINT8): Likewise.
24600 (BT_UINT16): Likewise.
24601 (BT_FN_ULONG): Likewise.
24602 (BT_FN_UINT_INT): Likewise.
24603 (BT_FN_UINT_ULONG): Likewise.
24604 (BT_FN_UINT_LONG): Likewise.
24605 (BT_FN_UINT_PTR): Likewise.
24606 (BT_FN_ULONG_PTR): Likewise.
24607 (BT_FN_INT8_FLOAT): Likewise.
24608 (BT_FN_INT16_FLOAT): Likewise.
24609 (BT_FN_UINT32_FLOAT): Likewise.
24610 (BT_FN_UINT16_FLOAT): Likewise.
24611 (BT_FN_UINT8_FLOAT): Likewise.
24612 (BT_FN_UINT64_FLOAT): Likewise.
24613 (BT_FN_UINT16_UINT32): Likewise.
24614 (BT_FN_UINT32_UINT16): Likewise.
24615 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24616 (BT_FN_INT_PTR_INT): Likewise.
24617 (BT_FN_UINT_PTR_UINT): Likewise.
24618 (BT_FN_LONG_PTR_LONG): Likewise.
24619 (BT_FN_ULONG_PTR_ULONG): Likewise.
24620 (BT_FN_VOID_UINT64_UINT64): Likewise.
24621 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24622 (BT_FN_INT8_INT8_INT8): Likewise.
24623 (BT_FN_INT16_INT16_INT16): Likewise.
24624 (BT_FN_INT_INT_INT): Likewise.
24625 (BT_FN_UINT_FLOAT_UINT): Likewise.
24626 (BT_FN_FLOAT_UINT_UINT): Likewise.
24627 (BT_FN_ULONG_UINT_UINT): Likewise.
24628 (BT_FN_ULONG_UINT_PTR): Likewise.
24629 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24630 (BT_FN_UINT_UINT_UINT): Likewise.
24631 (BT_FN_VOID_UINT_PTR): Likewise.
24632 (BT_FN_UINT_UINT_PTR: Likewise.
24633 (BT_FN_UINT32_UINT64_PTR): Likewise.
24634 (BT_FN_INT_INT_UINT_UINT): Likewise.
24635 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24636 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24637 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24638 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24639 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24640 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24641 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24642 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
24643 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
24644 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
24645 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
24646 * doc/frontends.texi: List BRIG FE.
24647 * doc/install.texi (Testing): Add BRIG tesring requirements.
24648 * doc/invoke.texi (Overall Options): Mention BRIG.
24649 * doc/standards.texi (Standards): Doucment BRIG HSA version.
24651 2017-01-24 Richard Biener <rguenther@suse.de>
24653 PR translation/79208
24654 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
24656 2017-01-24 Martin Jambor <mjambor@suse.cz>
24659 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24660 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24661 and known_contexts.
24663 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24665 PR middle-end/79123
24666 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24667 casts from signed to unsigned really don't have a range.
24669 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24671 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24672 GMP_RNDx for compatiblity.
24674 2017-01-24 Martin Liska <mliska@suse.cz>
24677 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24678 that would prevent us to call alloca with -1 as argument.
24680 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24682 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24683 Avoid -Wformat-security warning.
24685 2017-01-23 Andrew Pinski <apinski@cavium.com>
24687 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24690 2017-01-23 Martin Sebor <msebor@redhat.com>
24692 PR middle-end/78703
24693 * gimple-ssa-sprintf.c (warn_level): New global.
24694 (format_integer): Use it here and throughout the rest of the file.
24695 Use the same switch to compute sign as base.
24696 (maybe_warn): New function.
24697 (format_directive): Factor out warnings into maybe_warn.
24698 Add debugging output. Use warn_level.
24699 (add_bytes): Use warn_level.
24700 (pass_sprintf_length::compute_format_length): Add debugging output.
24701 (try_substitute_return_value): Same.
24702 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24704 PR middle-end/78703
24705 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24706 (struct fmtresult, format_integer, format_floating): Adjust.
24707 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24708 (get_string_length, format_string,format_directive): Same.
24709 (pass_sprintf_length::compute_format_length): Same.
24710 (try_substitute_return_value): Simplify slightly.
24712 PR middle-end/78703
24713 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24714 (fmtresult::operator+=): Outlined.
24715 (struct fmtresult): Add ctors.
24716 (struct conversion_spec): Rename...
24717 (struct directive): ...to this. Add and remove data members.
24718 (directive::set_width, directive::set_precision): New functions.
24719 (format_percent): Use fmtresult ctor.
24720 (get_width_and_precision): Remove.
24721 (format_integer): Make naming changes. Avoid computing width and
24723 (format_floating): Same. Adjust indentation.
24724 (format_character, format_none): New functions.
24725 (format_string): Moved character handling to format_character.
24726 (format_directive): Remove arguments, change return type.
24727 (parse_directive): New function.
24728 (pass_sprintf_length::compute_format_length): Move directive
24729 parsing to parse_directive.
24731 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24733 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24734 (assign_assembler_name_if_needed): ... this.
24735 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24736 (assign_assembler_name_if_needed): ... this.
24737 (free_lang_data_in_cgraph): Adjust callers.
24738 * cgraphunit.c (cgraph_node::analyze): Likewise.
24739 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24742 2017-01-23 Richard Biener <rguenther@suse.de>
24744 PR tree-optimization/79088
24745 PR tree-optimization/79188
24746 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
24747 resetting loop bounds after last path deletion. Reset loop
24748 bounds of the target loop, make code match the comments.
24749 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
24750 Make sure loops need no fixups.
24752 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24754 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
24755 exponent support with double type for first argument.
24756 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
24757 type returned by __builtin_vec_extract_sig,
24758 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
24759 functions from "vector int" to "vector unsigned int" or from
24760 "vector long long int" to "vector unsigned long long int".
24761 Changed type returned by __builtin_vec_extract_exp,
24762 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
24763 functions from "vector int" to "vector unsigned int" or from
24764 "vector long long int" to "vector unsigned long long int".
24765 Changed return type of __builtin_vec_test_data_class,
24766 __builtin_vec_test_data_class_sp, and
24767 __builtin_vec_test_data_class_dp from "vector int" to
24768 "vector bool int" or from "vector long long int" to "vector bool
24769 long long int" and changed second argument type from "unsigned
24770 int" to "int". Added new overloaded function forms "vector float
24771 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
24772 "vector float __builtin_vec_insert_exp_sp (vector float, vector
24773 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
24774 double, vector unsigned long long int)" and "vector double
24775 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
24776 long int)". Changed return type of
24777 __builtin_scalar_test_data_class and
24778 __builtin_scalar_test_data_class_sp and
24779 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
24780 int" and changed second argument from "unsigned int" to "int".
24781 Changed type returned by __builtin_scalar_test_neg,
24782 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
24783 from "int" to "bool int". Added new overloaded function form
24784 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
24785 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
24786 exponent double-precision with floating point first argument.
24787 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
24788 documentation of scalar_test_data_class, scalar_test_neg,
24789 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
24790 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
24791 vec_test_data_class built-in functions to reflect refinements in
24792 their type signatures.
24794 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
24796 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
24798 (aarch64_elf_asm_destructor): Likewise.
24800 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
24802 PR rtl-optimization/78634
24803 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
24804 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
24805 * ifcvt.c (noce_try_cmove): Add missing cost check.
24807 PR rtl-optimization/71724
24808 * combine.c (if_then_else_cond): Look for situations where it is
24809 beneficial to undo the work of one of the recursive calls.
24811 2017-01-23 Bin Cheng <bin.cheng@arm.com>
24813 PR tree-optimization/70754
24814 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
24815 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
24816 combined stmt before it if not NULL.
24817 (combine_chains): Process refs reversely and compute dominance point
24820 2017-01-23 Martin Liska <mliska@suse.cz>
24822 PR tree-optimization/79196
24823 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
24824 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
24826 (strlen_optimize_stmt): Call the renamed function.
24828 2017-01-23 Michael Matz <matz@suse.de>
24830 PR tree-optimization/78384
24831 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
24833 2017-01-23 Richard Biener <rguenther@suse.de>
24835 PR tree-optimization/79186
24836 * tree-vrp.c (register_new_assert_for): Make sure we've seen
24837 both incoming edges before moving an assert.
24839 2017-01-23 Martin Jambor <mjambor@suse.cz>
24841 * ipa-prop.c (load_from_param_1): Removed.
24842 (load_from_unmodified_param): Bits from load_from_param_1 put back
24844 (load_from_param): Removed.
24845 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
24846 with stmt. Reverted back to use of load_from_unmodified_param.
24848 2017-01-23 Martin Jambor <mjambor@suse.cz>
24851 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
24852 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
24853 field a pointer to garbage collected vector, mark lattices and
24854 ipcp_orig_node with GTY((skip)).
24855 (ipa_get_param_count): Adjust to descriptors being a pointer.
24856 (ipa_get_param): Likewise.
24857 (ipa_get_type): Likewise.
24858 (ipa_get_param_move_cost): Likewise.
24859 (ipa_set_param_used): Likewise.
24860 (ipa_get_controlled_uses): Likewise.
24861 (ipa_set_controlled_uses): Likewise.
24862 (ipa_is_param_used): Likewise.
24863 (ipa_node_params_t): Move into garbage collector. New methods insert
24865 (ipa_node_params_sum): Annotate wth GTY(()).
24866 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
24868 (ipa_load_from_parm_agg): Adjust declaration.
24869 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
24870 * ipa-profile.c (ipa_profile): Likewise.
24871 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
24872 (ipa_populate_param_decls): Make descriptors parameter garbage
24874 (ipa_dump_param): Adjust to descriptors being a pointer.
24875 (ipa_alloc_node_params): Likewise.
24876 (ipa_initialize_node_params): Likewise.
24877 (load_from_param_1): Make descriptors parameter garbage collected.
24878 (load_from_unmodified_param): Likewise.
24879 (load_from_param): Likewise.
24880 (ipa_load_from_parm_agg): Likewise.
24881 (ipa_node_params::~ipa_node_params): Removed.
24882 (ipa_free_all_node_params): Remove call to delete operator.
24883 (ipa_node_params_t::insert): New.
24884 (ipa_node_params_t::remove): Likewise.
24885 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
24886 copy known_csts and known_contexts vectors.
24887 (ipa_read_node_info): Adjust to descriptors being a pointer.
24888 (ipcp_modif_dom_walker): Make m_descriptors field garbage
24890 (ipcp_transform_function): Make descriptors variable garbage
24893 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
24895 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
24896 * config/i386/avx512dqintrin.h: Ditto.
24897 * config/i386/avx512fintrin.h: Ditto.
24898 * config/i386/i386.c: Handle new builtins.
24899 * config/i386/i386-builtin.def: Add new builtins.
24900 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
24901 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
24903 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24904 Martin Liska <mliska@suse.cz>
24906 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
24907 * asan.c (asan_expand_poison_ifn): Support stores and use
24908 appropriate ASAN report function.
24909 * internal-fn.c (expand_ASAN_POISON_USE): New function.
24910 * internal-fn.def (ASAN_POISON_USE): Declare.
24911 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
24912 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
24913 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
24914 ASAN_POISON calls w/o LHS.
24915 * tree-ssa.c (execute_update_addresses_taken): Create clobber
24916 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
24917 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
24918 * gimplify.c (asan_poison_variables): Add attribute
24919 use_after_scope_memory to variables that really needs to live
24921 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
24922 having the attribute.
24924 2017-01-23 Martin Liska <mliska@suse.cz>
24926 * asan.c (create_asan_shadow_var): New function.
24927 (asan_expand_poison_ifn): Likewise.
24928 * asan.h (asan_expand_poison_ifn): New declaration.
24929 * internal-fn.c (expand_ASAN_POISON): Likewise.
24930 * internal-fn.def (ASAN_POISON): New builtin.
24931 * sanopt.c (pass_sanopt::execute): Expand
24932 asan_expand_poison_ifn.
24933 * tree-inline.c (copy_decl_for_dup_finish): Make function
24935 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
24936 * tree-ssa.c (is_asan_mark_p): New function.
24937 (execute_update_addresses_taken): Rewrite local variables
24938 (identified just by use-after-scope as addressable) into SSA.
24940 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24942 * doc/install.texi (Specific): opensource.apple.com uses https
24943 now. Remove trailing slash.
24945 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24947 * README.Portability: Remove note on an Irix compatibility issue.
24949 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
24951 * gcov.c (INCLUDE_ALGORITHM): Define.
24952 (INCLUDE_VECTOR): Define.
24953 No longer include <vector> and <algorithm> directly.
24955 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
24957 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
24959 * doc/invoke.texi (Code Gen Options): Ditto.
24961 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
24964 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
24966 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
24968 rtl-optimization/79125
24969 * cprop.c (local_cprop_pass): Handle cases where we make an
24970 unconditional trap.
24972 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
24976 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
24977 read from, for big endian.
24979 2017-01-20 Jiong Wang <jiong.wang@arm.com>
24981 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
24982 register pauth builtins for LP64 only.
24984 2017-01-20 Marek Polacek <polacek@redhat.com>
24987 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24990 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24992 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24994 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24995 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24996 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24998 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25001 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
25002 in big-endian mode when they are not a single duplicated value.
25004 2017-01-20 Richard Biener <rguenther@suse.de>
25006 * BASE-VER: Bump to 7.0.1.
25008 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25010 * omp-low.c (omplow_simd_context): New struct. Use it...
25011 (lower_rec_simd_input_clauses): ...here and...
25012 (lower_rec_input_clauses): ...here to hold common data. Adjust all
25013 references to idx, lane, max_vf, is_simt.
25015 2017-01-20 Graham Markall <graham.markall@embecosm.com>
25017 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
25020 2017-01-20 Martin Jambor <mjambor@suse.cz>
25022 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
25023 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
25025 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
25026 (GTFILES): Rename hsa.c to hsa-common.c.
25027 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
25028 * hsa-dump.c: Likewise.
25029 * hsa-gen.c: Likewise.
25030 * hsa-regalloc.c: Likewise.
25031 * ipa-hsa.c: Likewise.
25032 * omp-expand.c: Likewise.
25033 * omp-low.c: Likewise.
25034 * toplev.c: Likewise.
25036 2017-01-20 Marek Polacek <polacek@redhat.com>
25039 * doc/invoke.texi: Document -Wduplicated-branches.
25040 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
25041 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
25042 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
25043 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
25044 return 0 only when not OEP_LEXICOGRAPHIC.
25045 (fold_build_cleanup_point_expr): Use the expression
25046 location when building CLEANUP_POINT_EXPR.
25047 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
25048 * tree.c (add_expr): Handle error_mark_node.
25050 2017-01-20 Martin Liska <mliska@suse.cz>
25053 * tree-profile.c (init_ic_make_global_vars): Do not call
25055 (gimple_init_gcov_profiler): Likewise.
25057 2017-01-20 Martin Liska <mliska@suse.cz>
25060 * cgraph.h (maybe_create_reference): Remove argument and
25062 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
25064 * ipa-cp.c (create_specialized_node): Likewise.
25065 * symtab.c (symtab_node::maybe_create_reference): Handle
25066 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
25068 2017-01-20 Martin Liska <mliska@suse.cz>
25070 * read-rtl-function.c (function_reader::create_function): Use
25071 build_decl instread of build_decl_stat.
25073 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
25075 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
25076 * config/i386/avx512dqintrin.h: Ditto.
25077 * config/i386/avx512fintrin.h: Ditto.
25078 * config/i386/i386-builtin-types.def: Add new types.
25079 * config/i386/i386.c: Handle new types.
25080 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
25081 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
25082 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
25083 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
25084 (__builtin_ia32_kshiftridi): New.
25085 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
25087 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
25091 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
25092 define to rs6000_init_stack_protect_guard.
25093 (rs6000_init_stack_protect_guard): New function.
25095 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25096 Yunqiang Su <yunqiang.su@imgtec.com>
25098 * config.gcc (supported_defaults): Add madd4.
25099 (with_madd4): Add validation.
25100 (all_defaults): Add madd4.
25101 * config/mips/mips.opt (mmadd4): New option.
25102 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25104 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25106 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
25107 (ISA_HAS_FUSED_MADD4): Likewise.
25108 * doc/invoke.texi (-mmadd4): Document the new option.
25109 * doc/install.texi (--with-madd4): Document the new option.
25111 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25113 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
25114 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
25115 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
25116 (aarch64_init_pauth_hint_builtins): New.
25117 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
25118 (aarch64_expand_builtin): Expand new builtins.
25120 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25122 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
25123 * combine-stack-adj.c (no_unhandled_cfa): Handle
25124 REG_CFA_TOGGLE_RA_MANGLE.
25125 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
25126 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
25127 info for return address signing.
25128 (aarch64_expand_epilogue): Likewise.
25130 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25132 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
25133 * config/aarch64/aarch64-protos.h
25134 (aarch64_return_address_signing_enabled): New declaration.
25135 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
25137 (aarch64_expand_prologue): Sign return address before it's pushed onto
25139 (aarch64_expand_epilogue): Authenticate return address fetched from
25141 (aarch64_override_options): Sanity check for ILP32 and ISA level.
25142 (aarch64_attributes): New function attributes for "sign-return-address".
25143 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
25144 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
25145 ("*do_return"): Generate combined instructions according to key index.
25146 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
25147 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
25149 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
25150 * config/aarch64/aarch64.opt (msign-return-address=): New.
25151 * doc/extend.texi (AArch64 Function Attributes): Documents
25152 "sign-return-address=".
25153 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
25155 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25157 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
25158 overall option summary.
25160 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25162 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
25163 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
25164 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
25165 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
25167 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
25169 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
25170 -mpower9-minmax by default for -mcpu=power9.
25171 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
25172 128-bit floating point.
25174 2017-01-20 Alan Modra <amodra@gmail.com>
25176 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
25177 optimizing for size.
25179 2017-01-20 Alan Modra <amodra@gmail.com>
25182 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
25183 for strcmp and strncmp from corresponding builtin decl.
25185 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25187 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
25188 instead of i386/rtems-64.h.
25189 * config/i386/rtems-64.h: Remove.
25191 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25195 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
25197 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
25199 2017-01-19 Tamar Christina <tamar.christina@arm.com>
25201 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
25202 Change int to HOST_WIDE_INT.
25203 * config/aarch64/aarch64-protos.h
25204 (aarch64_simd_gen_const_vector_dup): Likewise.
25205 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
25207 2017-01-19 David Malcolm <dmalcolm@redhat.com>
25209 * langhooks-def.h (lhd_type_for_size): New decl.
25210 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
25211 * langhooks.c (lhd_type_for_size): New function, taken from
25214 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
25216 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
25217 define_bypass for CR latency.
25218 (power9-cracked-alu): Update bypass latency and remove power9-branch.
25219 (power9-alu2): Add define_bypass for CR latency.
25221 (power9-mul): Update insn latency.
25222 (power9-mul-compare): Update insn latency, bypass latency and remove
25225 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25227 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
25229 * config/aarch64/aarch64.md
25230 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
25231 aarch64_nopcrelative_literal_loads.
25232 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
25234 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
25236 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
25237 TARGET_LOONGSON_3A.
25238 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
25240 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
25243 * config.gcc (supported_defaults): Add lxc1-sxc1.
25244 (with_lxc1_sxc1): Add validation.
25245 (all_defaults): Add lxc1-sxc1.
25246 * config/mips/mips.opt (mlxc1-sxc1): New option.
25247 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25249 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25250 __mips_no_lxc1_sxc1.
25251 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
25252 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
25253 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
25255 2017-01-19 Richard Biener <rguenther@suse.de>
25257 PR tree-optimization/72488
25258 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
25259 sure to restore SSA info.
25260 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
25262 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
25264 PR rtl-optimization/79121
25265 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
25266 of the inner type when shifting an extended value.
25268 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25271 * symtab.c (symtab_node::equal_address_to): Fix comparing of
25272 interposable aliases.
25274 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
25277 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
25278 Use the evmergelohi instruction.
25279 (mov_si<mode>_e500_subreg4_2_le): Likewise.
25280 (mov_sitf_e500_subreg8_2_be): Likewise.
25281 (mov_sitf_e500_subreg12_2_le): Likewise.
25282 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
25283 (mov_si<mode>_e500_subreg4_2_be): Likewise.
25284 (mov_sitf_e500_subreg8_2_le): Likewise.
25285 (mov_sitf_e500_subreg12_2_be): Likewise.
25287 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25289 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
25290 attribute from vecsimple to vecperm.
25291 (altivec_vbpermq2): Likewise.
25293 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25296 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
25298 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25299 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
25300 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
25301 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
25302 case where N arg is SIZE_MAX.
25303 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
25304 (cmpstrsi): Add pattern.
25306 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25308 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25309 __builtin_vec_revb builtins.
25310 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
25311 built-in functions to support generation of the ISA 3.0 XXBR<x>
25312 vector byte reverse instructions.
25313 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
25314 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
25315 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
25316 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
25317 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
25318 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
25319 (P9V_BUILTIN_VEC_REVB): Likewise.
25320 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
25321 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
25322 (p9_xxbrq_v16qi): Likewise.
25323 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
25324 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
25325 (p9_xxbrh_v8hi): Likewise.
25326 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
25327 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
25328 vec_revb built-in functions.
25330 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
25332 PR rtl-optimization/78952
25333 * config/i386/i386.md (any_extract): New code iterator.
25334 (*insvqi_2): Use any_extract for source operand.
25335 (*insvqi_3): Use any_shiftrt for source operand.
25337 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25339 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
25341 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
25343 2017-01-18 Matthias Klose <doko@ubuntu.com>
25345 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
25347 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25349 * config/rs6000/altivec.h (vec_bperm): Change #define.
25350 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
25351 (altivec_vbpermq2): New define_insn.
25352 (altivec_vbpermd): Likewise.
25353 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
25354 function interface.
25355 (VBPERMD): Likewise.
25356 (VBPERM): New polymorphic function interface.
25357 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
25358 Add entries for P9V_BUILTIN_VEC_VBPERM.
25359 * doc/extend.texi: Add interfaces for vec_bperm.
25361 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25363 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
25364 first letter of error messages.
25365 (s390_resolve_overloaded_builtin): Likewise.
25366 * config/s390/s390.c (s390_expand_builtin): Likewise.
25367 (s390_invalid_arg_for_unprototyped_fn): Likewise.
25368 (s390_valid_target_attribute_inner_p): Likewise.
25369 * config/s390/s390.md ("tabort"): Likewise.
25371 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
25373 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
25374 (ISA_AVOID_DIV_HILO): New macro.
25375 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
25376 (ISA_HAS_DDIV): Likewise.
25378 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25380 * doc/invoke.texi (fabi-version): Correct number of occurrences.
25382 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25384 * doc/invoke.texi (fabi-version): Spelling fix.
25386 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25389 * doc/invoke.texi (fabi-version): Mention mangling fix for
25392 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25395 * doc/invoke.texi (fabi-version): Document discriminator mangling.
25397 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
25400 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
25401 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25403 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25405 (stack_protect_test): Ditto.
25406 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25407 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25409 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25410 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25411 -mstack-protector-guard-offset=.
25412 (RS/6000 and PowerPC Options): Ditto.
25414 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25416 * config/i386/i386.h (MASK_CLASS_P): New define.
25417 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25418 there are no registers from different register sets also when
25419 mask registers are used. Update function comment.
25420 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25421 to (*k/*r) and (*k/*km) alternatives.
25423 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25425 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25426 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25427 (EH_RETURN_HANDLER_RTX): New define.
25428 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25429 Force frame pointer in EH return functions.
25430 (aarch64_expand_epilogue): Add barrier for eh_return.
25431 (aarch64_final_eh_return_addr): Remove.
25432 (aarch64_eh_return_handler_rtx): New function.
25433 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25435 (aarch64_eh_return_handler_rtx): New prototype.
25437 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25439 * config/rs6000/altivec.h (vec_rlmi): New #define.
25440 (vec_vrlnm): Likewise.
25441 (vec_rlnm): Likewise.
25442 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25443 (UNSPEC_VRLNM): Likewise.
25444 (VIlong): New mode iterator.
25445 (altivec_vrl<VI_char>mi): New define_insn.
25446 (altivec_vrl<VI_char>nm): Likewise.
25447 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25449 (VRLDNM): Likewise.
25450 (RLNM): New polymorphic function entry.
25451 (VRLWMI): New monomorphic function entry.
25452 (VRLDMI): Likewise.
25453 (RLMI): New polymorphic function entry.
25454 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25455 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25456 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25459 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25462 * dwarf2out.c (field_byte_offset): Restore the
25463 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25464 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25465 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25468 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25471 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25473 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25476 * dwarf2out.c (add_data_member_location_attribute): For constant
25477 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25478 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25479 DW_AT_byte_size attributes.
25481 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25483 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25484 after forcing to constant memory when the code model is medium.
25486 2017-01-17 Julia Koval <julia.koval@intel.com>
25489 * config/i386/avx512fintrin.h
25490 (_mm512_i32gather_ps): Change __addr type to void const*.
25491 (_mm512_mask_i32gather_ps): Ditto.
25492 (_mm512_i32gather_pd): Ditto.
25493 (_mm512_mask_i32gather_pd): Ditto.
25494 (_mm512_i64gather_ps): Ditto.
25495 (_mm512_mask_i64gather_ps): Ditto.
25496 (_mm512_i64gather_pd): Ditto.
25497 (_mm512_mask_i64gather_pd): Ditto.
25498 (_mm512_i32gather_epi32): Ditto.
25499 (_mm512_mask_i32gather_epi32): Ditto.
25500 (_mm512_i32gather_epi64): Ditto.
25501 (_mm512_mask_i32gather_epi64): Ditto.
25502 (_mm512_i64gather_epi32): Ditto.
25503 (_mm512_mask_i64gather_epi32): Ditto.
25504 (_mm512_i64gather_epi64): Ditto.
25505 (_mm512_mask_i64gather_epi64): Ditto.
25506 (_mm512_i32scatter_ps): Change __addr type to void*.
25507 (_mm512_mask_i32scatter_ps): Ditto.
25508 (_mm512_i32scatter_pd): Ditto.
25509 (_mm512_mask_i32scatter_pd): Ditto.
25510 (_mm512_i64scatter_ps): Ditto.
25511 (_mm512_mask_i64scatter_ps): Ditto.
25512 (_mm512_i64scatter_pd): Ditto.
25513 (_mm512_mask_i64scatter_pd): Ditto.
25514 (_mm512_i32scatter_epi32): Ditto.
25515 (_mm512_mask_i32scatter_epi32): Ditto.
25516 (_mm512_i32scatter_epi64): Ditto.
25517 (_mm512_mask_i32scatter_epi64): Ditto.
25518 (_mm512_i64scatter_epi32): Ditto.
25519 (_mm512_mask_i64scatter_epi32): Ditto.
25520 (_mm512_i64scatter_epi64): Ditto.
25521 (_mm512_mask_i64scatter_epi64): Ditto.
25522 * config/i386/avx512pfintrin.h
25523 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25524 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25525 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25526 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25527 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25528 (_mm512_prefetch_i32scatter_ps): Ditto.
25529 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25530 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25531 (_mm512_prefetch_i64scatter_pd): Ditto.
25532 (_mm512_prefetch_i64scatter_ps): Ditto.
25533 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25534 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25535 * config/i386/avx512vlintrin.h
25536 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25537 (_mm_mmask_i32gather_ps): Ditto.
25538 (_mm256_mmask_i32gather_pd): Ditto.
25539 (_mm_mmask_i32gather_pd): Ditto.
25540 (_mm256_mmask_i64gather_ps): Ditto.
25541 (_mm_mmask_i64gather_ps): Ditto.
25542 (_mm256_mmask_i64gather_pd): Ditto.
25543 (_mm_mmask_i64gather_pd): Ditto.
25544 (_mm256_mmask_i32gather_epi32): Ditto.
25545 (_mm_mmask_i32gather_epi32): Ditto.
25546 (_mm256_mmask_i32gather_epi64): Ditto.
25547 (_mm_mmask_i32gather_epi64): Ditto.
25548 (_mm256_mmask_i64gather_epi32): Ditto.
25549 (_mm_mmask_i64gather_epi32): Ditto.
25550 (_mm256_mmask_i64gather_epi64): Ditto.
25551 (_mm_mmask_i64gather_epi64): Ditto.
25552 (_mm256_i32scatter_ps): Change __addr type to void*.
25553 (_mm256_mask_i32scatter_ps): Ditto.
25554 (_mm_i32scatter_ps): Ditto.
25555 (_mm_mask_i32scatter_ps): Ditto.
25556 (_mm256_i32scatter_pd): Ditto.
25557 (_mm256_mask_i32scatter_pd): Ditto.
25558 (_mm_i32scatter_pd): Ditto.
25559 (_mm_mask_i32scatter_pd): Ditto.
25560 (_mm256_i64scatter_ps): Ditto.
25561 (_mm256_mask_i64scatter_ps): Ditto.
25562 (_mm_i64scatter_ps): Ditto.
25563 (_mm_mask_i64scatter_ps): Ditto.
25564 (_mm256_i64scatter_pd): Ditto.
25565 (_mm256_mask_i64scatter_pd): Ditto.
25566 (_mm_i64scatter_pd): Ditto.
25567 (_mm_mask_i64scatter_pd): Ditto.
25568 (_mm256_i32scatter_epi32): Ditto.
25569 (_mm256_mask_i32scatter_epi32): Ditto.
25570 (_mm_i32scatter_epi32): Ditto.
25571 (_mm_mask_i32scatter_epi32): Ditto.
25572 (_mm256_i32scatter_epi64): Ditto.
25573 (_mm256_mask_i32scatter_epi64): Ditto.
25574 (_mm_i32scatter_epi64): Ditto.
25575 (_mm_mask_i32scatter_epi64): Ditto.
25576 (_mm256_i64scatter_epi32): Ditto.
25577 (_mm256_mask_i64scatter_epi32): Ditto.
25578 (_mm_i64scatter_epi32): Ditto.
25579 (_mm_mask_i64scatter_epi32): Ditto.
25580 (_mm256_i64scatter_epi64): Ditto.
25581 (_mm256_mask_i64scatter_epi64): Ditto.
25582 (_mm_i64scatter_epi64): Ditto.
25583 (_mm_mask_i64scatter_epi64): Ditto.
25584 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25585 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25586 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25587 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25588 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25589 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25590 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25591 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25592 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25593 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25594 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25595 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25596 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25597 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25598 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25599 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25600 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25601 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25602 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25603 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25604 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25605 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25606 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25607 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25608 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25609 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25610 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25611 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25612 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25613 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25614 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25615 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25616 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25617 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25618 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25619 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25620 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25621 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25622 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25623 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25624 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25625 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25626 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25627 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25628 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25629 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25630 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25631 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25632 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25633 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25634 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25635 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25636 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25637 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25638 definitions accordingly.
25640 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25641 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25644 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
25647 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
25650 * ira-conflicts.c (ira_build_conflicts): Update total conflict
25651 hard regs for inner regno.
25653 2017-01-17 Martin Liska <mliska@suse.cz>
25656 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
25657 assumption and add comment.
25659 2017-01-17 Nathan Sidwell <nathan@acm.org>
25661 * ipa-visibility.c (localize_node): New function, broken out of ...
25662 (function_and_variable_visibility): ... here. Call it.
25664 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25666 PR middle-end/77445
25667 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25668 correctly set frequency of oudgoing edge.
25669 (duplicate_thread_path): Fix profile updating.
25671 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25674 * configure.ac: Add GCC_BASE_VER.
25675 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25676 version from BASE-VER file.
25677 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25678 (gcc.o): Depend on $(BASEVER).
25679 * common.opt (dumpfullversion): New option.
25680 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25681 * doc/invoke.texi: Document -dumpfullversion.
25682 * doc/install.texi: Document --with-gcc-major-version-only.
25683 * configure: Regenerated.
25685 2017-01-17 Richard Biener <rguenther@suse.de>
25687 PR tree-optimization/71433
25688 * tree-vrp.c (register_new_assert_for): Merge same asserts
25689 on all incoming edges.
25690 (process_assert_insertions_for): Handle insertions at the
25693 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25695 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25696 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25698 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25701 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25704 2017-01-17 Alan Modra <amodra@gmail.com>
25707 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25708 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25709 symbolic stack limit when pic.
25711 2017-01-16 Martin Sebor <msebor@redhat.com>
25713 PR tree-optimization/78608
25714 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25716 2017-01-16 Jeff Law <law@redhat.com>
25719 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25720 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25721 for several include directories that may be relative to sysroot.
25722 * config/i386/x-mingw32 (gplus_includedir): Define.
25723 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25724 (native_system_includedir): Likewise.
25725 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25726 override if TARGET_SYSTEM_ROOT is defined.
25727 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25729 PR tree-optimization/79090
25730 PR tree-optimization/33562
25731 PR tree-optimization/61912
25732 PR tree-optimization/77485
25733 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25734 and computed trims into the dump file.
25736 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25738 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25740 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25743 * gimplify.c (gimplify_init_constructor): If want_value and
25744 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
25748 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
25749 sequence. Formatting fixes.
25750 (doloop_optimize): Formatting fixes.
25753 * gcc.c (debug_level_greater_than_spec_func): New function.
25754 (static_spec_functions): Add debug-level-gt spec function.
25755 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
25757 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25758 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25759 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
25760 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
25761 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
25762 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
25764 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
25766 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
25767 QImode fixups to general and mask registers only.
25769 2017-01-16 Carl Love <cel@us.ibm.com>
25771 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
25772 for built-in functions
25773 vector signed char vec_nabs (vector signed char)
25774 vector signed short vec_nabs (vector signed short)
25775 vector signed int vec_nabs (vector signed int)
25776 vector signed long long vec_nabs (vector signed long long)
25777 vector float vec_nabs (vector float)
25778 vector double vec_nabs (vector double)
25779 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
25781 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
25782 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
25783 * doc/extend.texi: Update the documentation file for the new built-in
25786 2017-01-16 Martin Sebor <msebor@redhat.com>
25788 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
25791 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25793 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
25794 UNSPEC_VSX__XXSPLTD to require special splat handling.
25796 2017-01-16 David Malcolm <dmalcolm@redhat.com>
25799 * system.h: Poison strndup.
25801 2017-01-16 Alan Modra <amodra@gmail.com>
25804 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
25807 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
25809 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
25811 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
25813 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
25814 call recog here. Assert that INSN_CODE (insn) is non-negative.
25816 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25819 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
25821 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
25822 in the currently scheduled RTL fragment.
25824 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25826 PR rtl-optimization/78751
25827 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
25830 2017-01-14 Jeff Law <law@redhat.com>
25832 PR tree-optimization/79090
25833 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
25834 variable length stores.
25835 (compute_trims): Delete dead assignment to *trim_tail.
25836 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
25839 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
25841 PR rtl-optimization/78626
25842 PR rtl-optimization/78727
25843 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
25844 of a block, and split such blocks after everything else is finished.
25846 2017-01-14 Alan Modra <amodra@gmail.com>
25849 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
25850 target legitimate_combined_insn.
25851 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
25852 (rs6000_legitimate_combined_insn): New function.
25853 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
25855 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
25856 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
25857 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
25859 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
25861 * doc/frontends.texi (G++ and GCC): Remove references to Java.
25863 2017-01-13 Jeff Law <law@redhat.com>
25865 PR tree-optimization/33562
25866 PR tree-optimization/61912
25867 PR tree-optimization/77485
25868 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
25870 (delete_dead_assignment): Likewise.
25871 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
25872 statement to delete_dead_call and delete_dead_assignment.
25874 2017-01-13 David Malcolm <dmalcolm@redhat.com>
25877 * substring-locations.c (format_warning_va): Strengthen case 1 so
25878 that both endpoints of the substring must be within the format
25879 range for just the substring to be printed.
25881 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
25883 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
25884 * config/i386/i386.c (ix86_target_string): Add missing options
25885 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
25886 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
25887 flags_other and ix86_target_other to flags2_other. Display unknown
25889 (ix86_valid_target_attribute_inner_p): Add missing options and
25890 reorder options by implied ISAs, as in ix86_target_string.
25892 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25894 * hash-table.h (hash_table::too_empty_p): New function.
25895 (hash_table::expand): Use it.
25896 (hash_table::traverse): Likewise.
25897 (hash_table::empty_slot): Use sizeof (value_type) instead of
25898 sizeof (PTR) to convert bytes to elements. Shrink the table
25899 if the current size is excessive for the current number of
25902 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25904 * ira-costs.c (record_reg_classes): Break from the inner loop
25905 early once alt_fail is known to be true. Update outer loop
25906 handling accordingly.
25908 2017-01-13 Jeff Law <law@redhat.com>
25910 * tree-ssa-dse.c (decrement_count): New function.
25911 (increment_start_addr, maybe_trim_memstar_call): Likewise.
25912 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
25913 when we know the partially dead statement is a mem* function.
25915 PR tree-optimization/61912
25916 PR tree-optimization/77485
25917 * tree-ssa-dse.c: Include expr.h.
25918 (maybe_trim_constructor_store): New function.
25919 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
25921 PR tree-optimization/33562
25922 PR tree-optimization/61912
25923 PR tree-optimization/77485
25924 * doc/invoke.texi: Document new dse-max-object-size param.
25925 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
25926 * tree-ssa-dse.c: Include params.h.
25927 (dse_store_status): New enum.
25928 (initialize_ao_ref_for_dse): New, partially extracted from
25930 (valid_ao_ref_for_dse, normalize_ref): New.
25931 (setup_live_bytes_from_ref, compute_trims): Likewise.
25932 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
25933 (maybe_trim_partially_dead_store): Likewise.
25934 (maybe_trim_complex_store): Likewise.
25935 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
25936 Track what bytes live from the original store. Return tri-state
25937 for dead, partially dead or live.
25938 (dse_dom_walker): Add constructor, destructor and new private members.
25939 (delete_dead_call, delete_dead_assignment): New extracted from
25941 (dse_optimize_stmt): Make a member of dse_dom_walker.
25942 Use initialize_ao_ref_for_dse.
25944 PR tree-optimization/33562
25945 PR tree-optimization/61912
25946 PR tree-optimization/77485
25947 * sbitmap.h (bitmap_count_bits): Prototype.
25948 (bitmap_clear_range, bitmap_set_range): Likewise.
25949 * sbitmap.c (bitmap_clear_range): New function.
25950 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
25952 2017-01-13 Martin Liska <mliska@suse.cz>
25955 * function.c (set_cfun): Add new argument force.
25956 * function.h (set_cfun): Likewise.
25957 * ipa-inline-transform.c (inline_call): Use the function when
25958 strict alising from is dropped for function we inline to.
25960 2017-01-13 Richard Biener <rguenther@suse.de>
25962 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
25963 for dumping GIMPLE INTEGER_CSTs.
25965 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25967 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
25968 to 201112L since C++17.
25970 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
25973 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
25974 if -fsanitize=kernel-address is present.
25976 2017-01-13 Richard Biener <rguenther@suse.de>
25978 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
25979 as _Literal ( type ) number in case usual suffixes do not
25980 preserve all information.
25982 2017-01-13 Richard Biener <rguenther@suse.de>
25984 PR tree-optimization/77283
25985 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25986 and ssa-iterators.h.
25987 (is_feasible_trace): Implement a cost model based on joiner
25990 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25993 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25994 char or short to __float128/_Float128 directly.
25996 2017-01-12 Martin Sebor <msebor@redhat.com>
25998 to -Wformat-overflow.
25999 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
26000 (min_bytes_remaining): Same.
26001 (get_string_length): Same.
26002 (format_string): Same.
26003 (format_directive): Same.
26005 (pass_sprintf_length::handle_gimple_call): Same.
26007 2017-01-12 Jakub Jelinek <jakub@redhat.com>
26009 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
26010 info.nowrite calls with no lhs that can't throw. Return bool
26011 whether gsi_remove has been called or not.
26012 (pass_sprintf_length::handle_gimple_call): Return bool whether
26013 try_substitute_return_value called gsi_remove. Formatting fix.
26014 (pass_sprintf_length::execute): Don't use gsi_remove if
26015 handle_gimple_call returned true.
26018 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
26019 be removed due to side-effects, don't remove following barrier nor
26020 turn the successor edge into fallthru edge.
26022 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26025 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
26026 element-reversing loads and stores as not swappable.
26028 2017-01-12 Nathan Sidwell <nathan@acm.org>
26029 Nicolai Stange <nicstange@gmail.com>
26031 * combine.c (try_combine): Don't ignore result of overlap checking
26032 loop. Combine overlap & asm check into single loop.
26034 2017-01-12 Richard Biener <rguenther@suse.de>
26036 * tree-pretty-print.c (dump_generic_node): Provide -gimple
26037 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
26039 2017-01-12 Richard Biener <rguenther@suse.de>
26041 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
26042 and TS_TARGET_OPTION directly derive from TS_BASE.
26043 * tree-core.h (tree_optimization_option): Derive from tree_base.
26044 (tree_target_option): Likewise.
26046 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
26048 * config/i386/i386.c (memory_address_length): Increase len
26049 only when rip_relative_addr_p returns false.
26051 2017-01-11 Julia Koval <julia.koval@intel.com>
26053 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
26054 (OPTION_MASK_ISA_SGX_SET): New.
26055 (ix86_handle_option): Handle OPT_msgx.
26056 * config.gcc: Added sgxintrin.h.
26057 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
26058 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
26059 * config/i386/i386.c (ix86_target_string): Add -msgx.
26061 (ix86_option_override_internal): Handle new options.
26062 (ix86_valid_target_attribute_inner_p): Add sgx.
26063 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
26064 * config/i386/i386.opt: Add msgx.
26065 * config/i386/sgxintrin.h: New file.
26066 * config/i386/x86intrin.h: Add sgxintrin.h.
26068 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26071 * fold-const.c (maybe_nonzero_address): Return 1 for function
26073 (tree_single_nonzero_warnv_p): Don't handle function local objects
26077 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
26080 2017-01-11 David Malcolm <dmalcolm@redhat.com>
26083 * opts.c: Include "spellcheck.h"
26084 (struct string_fragment): New struct.
26085 (struct edit_distance_traits<const string_fragment &>): New
26087 (get_closest_sanitizer_option): New function.
26088 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
26090 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26092 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
26094 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
26095 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
26096 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
26097 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
26098 for initial die_offset if dwarf_split_debug_info.
26099 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
26100 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
26102 (output_skeleton_debug_sections): Formatting fix. Use
26103 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
26104 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
26106 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
26108 * config/arm/cortex-a53.md: Add bypasses for
26109 cortex_a53_r2f_cvt.
26110 (cortex_a53_r2f): Only use for transfers.
26111 (cortex_a53_f2r): Likewise.
26112 (cortex_a53_r2f_cvt): Add reservation for conversions.
26113 (cortex_a53_f2r_cvt): Likewise.
26115 2017-01-11 Tamar Christina <tamar.christina@arm.com>
26117 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
26118 to all inlined functions, change static to extern.
26120 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
26123 * config/arm/arm.c (legitimize_pic_address): Handle reference to
26125 (arm_assemble_integer): Likewise.
26127 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
26129 * config.gcc: Use new awk script to check CPU, FPU and architecture
26130 parameters for --with-... options.
26131 * config/arm/parsecpu.awk: New file
26132 * config/arm/arm-cpus.in: New file.
26133 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
26135 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
26137 * config/arm/t-arm: Update dependency rules.
26138 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
26139 of processing .def files.
26140 * config/arm/genopt.sh: Deleted.
26141 * config/arm/gentune.sh: Deleted.
26142 * config/arm/arm-cores.def: Deleted.
26143 * config/arm/arm-arches.def: Deleted.
26144 * config/arm/arm-fpus.def: Deleted.
26145 * config/arm/arm-tune.md: Regenerated.
26146 * config/arm/arm-tables.opt: Regenerated.
26147 * config/arm/arm-cpu.h: New generated file.
26148 * config/arm/arm-cpu-data.h: New generated file.
26149 * config/arm/arm-cpu-cdata.h: New generated file.
26151 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
26154 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
26156 (input_varpool_node): Unpack dynamically_initialized bit.
26158 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
26160 PR rtl-optimization/79032
26161 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
26162 the alignment of the adjusted memory reference against that of MODE,
26163 instead of the alignment of the original memory reference.
26165 2017-01-11 Martin Jambor <mjambor@suse.cz>
26167 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
26169 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
26170 decorated functions.
26172 2017-01-11 Richard Biener <rguenther@suse.de>
26174 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
26175 set range/nonnull info for PHI results. Do not set it on
26176 stmts marked for removal.
26178 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
26180 * expr.c (store_field): In the bitfield case, fetch the return value
26181 from the registers before applying a single big-endian adjustment.
26182 Always do a final load for a BLKmode value not larger than a word.
26184 2017-01-10 David Malcolm <dmalcolm@redhat.com>
26187 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26188 that we correctly handle column numbers greater than
26189 LINE_MAP_MAX_COLUMN_NUMBER.
26191 2017-01-10 Martin Sebor <msebor@redhat.com>
26193 PR middle-end/78245
26194 * gimple-ssa-sprintf.c (get_destination_size): Call
26195 {init,fini}object_sizes.
26196 * tree-object-size.c (addr_object_size): Adjust.
26197 (pass_through_call): Adjust.
26198 (pass_object_sizes::execute): Adjust.
26199 * tree-object-size.h (fini_object_sizes): Declare.
26201 2017-01-10 Martin Sebor <msebor@redhat.com>
26203 PR tree-optimization/78775
26204 * builtins.c (get_size_range): Move...
26205 * calls.c: ...to here.
26206 (alloc_max_size): Accept zero argument.
26207 (operand_signed_p): Remove.
26208 (maybe_warn_alloc_args_overflow): Call get_size_range.
26209 * calls.h (get_size_range): Declare.
26211 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
26213 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
26214 from TI's devices.csv file as of September 2016.
26215 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
26217 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
26219 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
26220 * doc/invoke.texi: Likewise.
26221 * doc/md.texi: Likewise.
26222 * doc/objc.texi: Likewise.
26224 2017-01-10 Joshua Conner <joshconner@google.com>
26226 * config/arm/fuchsia-elf.h: New file.
26227 * config/fuchsia.h: New file.
26228 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
26229 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
26231 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
26233 2016-01-10 Richard Biener <rguenther@suse.de>
26235 PR tree-optimization/79034
26236 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26237 Propagate out degenerate PHIs in the joiner.
26239 2017-01-10 Martin Liska <mliska@suse.cz>
26241 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
26242 (sort_congruence_classes_by_decl_uid): Likewise.
26243 (sort_congruence_class_groups_by_decl_uid): Likewise.
26244 (sem_item_optimizer::merge_classes): Sort class, groups in these
26245 classes and members in the groups by DECL_UID of declarations.
26246 This would make merge operations stable.
26248 2017-01-10 Martin Liska <mliska@suse.cz>
26250 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
26251 usage of m_classes_vec.
26252 (sem_item_optimizer::~sem_item_optimizer): Likewise.
26253 (sem_item_optimizer::get_group_by_hash): Likewise.
26254 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
26255 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
26256 (sem_item_optimizer::verify_classes): Likewise.
26257 (sem_item_optimizer::process_cong_reduction): Likewise.
26258 (sem_item_optimizer::dump_cong_classes): Likewise.
26259 (sem_item_optimizer::merge_classes): Likewise.
26260 * ipa-icf.h (congruence_class_hash): Rename from
26261 congruence_class_group_hash. Remove declaration of m_classes_vec.
26263 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
26265 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
26266 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
26267 * config.gcc: Add avx512vpopcntdqintrin.h.
26268 * config/i386/avx512vpopcntdqintrin.h: New.
26269 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
26270 * config/i386/i386-builtin-types.def: Add new types.
26271 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
26272 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
26273 __builtin_ia32_vpopcountq_v8di_mask): New.
26274 * config/i386/i386-c.c (ix86_target_macros_internal): Define
26275 __AVX512VPOPCNTDQ__.
26276 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
26277 (PTA_AVX512VPOPCNTDQ): Define.
26278 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
26279 TARGET_AVX512VPOPCNTDQ_P): Define.
26280 * config/i386/i386.opt: Add mavx512vpopcntdq.
26281 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
26282 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
26284 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26286 PR middle-end/77484
26287 * predict.def (PRED_CALL): Set to 67.
26289 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
26291 * expr.c (store_field): In the bitfield case, if the value comes from
26292 a function call and is of an aggregate type returned in registers, do
26293 not modify the field mode; extract the value in all cases if the mode
26294 is BLKmode and the size is not larger than a word.
26296 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
26299 * config/i386/cpuid.h: Fix undefined behavior.
26301 2017-01-04 Jeff Law <law@redhat.com>
26303 PR tree-optimization/79007
26304 PR tree-optimization/67955
26305 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
26306 conservative for pt.null when flag_non_call_exceptions is on.
26308 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26310 PR translation/79019
26311 PR translation/79020
26312 * params.def (PARAM_INLINE_MIN_SPEEDUP,
26313 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
26314 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
26316 * config/avr/avr.opt (maccumulate-args): Likewise.
26317 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
26318 * common.opt (freport-bug): Likewise.
26319 * cif-code.def (CIF_FINAL_ERROR): Likewise.
26320 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
26321 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
26322 translatable string.
26323 * config/i386/i386.c (function_value_32): Likewise.
26324 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
26325 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
26327 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
26328 * common/config/msp430/msp430-common.c (msp430_handle_option):
26330 * symtab.c (symtab_node::verify_base): Likewise.
26331 * opts.c (set_debug_level): Likewise.
26332 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
26333 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
26334 missing whitespace to translatable strings.
26335 * config/avr/avr.md (bswapsi2): Fix typo in comment.
26336 * config/sh/superh.h: Likewise.
26337 * config/i386/xopintrin.h: Likewise.
26338 * config/i386/znver1.md: Likewise.
26339 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
26340 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
26341 * double-int.h (struct double_int): Likewise.
26342 * double-int.c (div_and_round_double): Likewise.
26343 * wide-int.cc: Likewise.
26344 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
26345 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
26346 * cfgcleanup.c (crossjumps_occured): Renamed to ...
26347 (crossjumps_occurred): ... this.
26348 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
26351 PR tree-optimization/78899
26352 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
26353 returning bool return struct loop *, NULL for failure and the new
26355 (versionable_outer_loop_p): Don't version outer loop if it has
26356 dont_vectorized bit set.
26357 (tree_if_conversion): When versioning outer loop, ensure
26358 tree_if_conversion is performed also on the inner loop of the
26359 non-vectorizable outer loop copy.
26360 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
26361 LOOP_VECTORIZED in inner loop of the scalar outer loop and
26362 prevent vectorization of it.
26363 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
26364 the outer loop vectorization of the non-scalar version is attempted
26365 before vectorization of the inner loop in scalar version. If
26366 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
26367 vectorization of its inner loop.
26368 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
26369 has 2 inner loops, rename also on edges from bb whose single pred
26370 is outer_loop->header. Fix typo in function comment.
26372 2017-01-09 Martin Sebor <msebor@redhat.com>
26375 * asan.c (asan_emit_stack_protection): Increase local buffer size
26376 to avoid snprintf truncation warning.
26378 2017-01-09 Andrew Pinski <apinski@cavium.com>
26380 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
26381 to reference thunderx2t99 for the tuning structure
26382 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
26384 (thunderx2t99_extra_costs): This.
26385 * config/aarch64/aarch64-tune.md: Regenerate.
26386 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
26387 (vulcan_addrcost_table): This.
26388 (vulcan_regmove_cost): Rename to ...
26389 (thunderx2t99_regmove_cost): This.
26390 (vulcan_vector_cost): Rename to ...
26391 (thunderx2t99_vector_cost): this.
26392 (vulcan_branch_cost): Rename to ...
26393 (thunderx2t99_branch_cost): This.
26394 (vulcan_tunings): Rename to ...
26395 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
26396 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
26398 2017-01-09 Martin Jambor <mjambor@suse.cz>
26402 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26403 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26404 (propagate_vr_accross_jump_function): Use the above function for all
26405 value range computations for pass-through jump functions and type
26406 converasion from explicit value range values.
26407 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26408 parameters from TYPE_ARG_TYPES.
26409 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26410 (ipa_write_node_info): Stream type of the actual argument.
26411 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26413 2017-01-09 Martin Liska <mliska@suse.cz>
26416 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26417 (lookup_compiler): Do not show error message with have_E.
26419 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26421 PR tree-optimization/78938
26422 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26423 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26424 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26425 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26428 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26430 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26433 2017-01-09 Richard Biener <rguenther@suse.de>
26435 PR tree-optimization/78997
26436 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26437 name condition properly.
26439 2017-01-09 Richard Biener <rguenther@suse.de>
26442 * dwarf2out.c (is_cxx): New overload with context.
26443 (is_naming_typedef_decl): Use it.
26445 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26447 * invoke.texi (Option Summary): Correct spacing in option lists
26448 and add line breaks to fix over-long lines.
26450 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26452 PR middle-end/17660
26454 * extend.texi (Common Variable Attributes): Add xref to GCC
26455 Internals manual to explain mode attribute keywords.
26457 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26460 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26461 and Preprocessor Options.
26462 (Options for Linking): Document -pthread here....
26463 (RS/6000 and PowerPC Options): ...not here.
26464 (Solaris 2 Options): ...or here.
26465 * doc/cppopts.texi: Document -pthread.
26467 2017-01-08 Martin Sebor <msebor@redhat.com>
26469 PR middle-end/77708
26470 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26471 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26472 New member functions.
26473 (format_directive): Used them.
26475 (pass_sprintf_length::handle_gimple_call): Same.
26476 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26477 to avoid truncation for any argument.
26478 (extract_affine_mul): Same.
26479 * tree.c (get_file_function_name): Same.
26481 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26483 PR middle-end/77484
26484 * predict.def (PRED_INDIR_CALL): Set to 86.
26486 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26488 PR preprocessor/54124
26489 * doc/cppopts.texi: Reformat -d subtable to list the full name
26490 of the options. Add cross-reference to the docs for the general
26491 compiler -d options.
26492 * doc/invoke.texi (Developer Options): Add cross-reference to the
26493 preprocessor-specific -d option documentation.
26495 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26497 PR preprocessor/13498
26498 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26499 redudant material, and reflect new command-line options.
26500 (System Headers): Likewise.
26502 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26504 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26505 -isystem, and -idirafter. Copy-edit.
26506 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26507 default for -ftrack-macro-expansion. Delete obsolete and
26508 badly-formatted implementation details about -fdebug-cpp output.
26509 * doc/cppwarnopts.texi: Copy-edit.
26511 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26514 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26515 that the transition from a max line width >= 1<<10 to narrower
26516 lines works correctly.
26518 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26520 * doc/options.texi (PerFunction): New.
26521 * opt-functions.awk (switch_flags): Map both Optimization and
26522 PerFunction to CL_OPTIMIZATION.
26523 * opth-gen.awk: Test for PerFunction flag along with
26525 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26526 it only when the latter is present. Skip those that don't in
26527 the hash function generator.
26528 * common.opt (fvar-tracking): Mark as PerFunction instead of
26530 (fvar-tracking-assignments): Likewise.
26531 (fvar-tracking-assignments-toggle): Likewise.
26532 (fvar-tracking-uninit): Likewise.
26534 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26536 PR translation/79018
26537 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26540 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26543 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26544 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26545 TARGET_LONG_JUMP_TABLE_OFFSETS.
26546 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26547 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26548 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26549 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26550 * config/m68k/m68k.md (tablejump expander): Likewise.
26551 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26552 TARGET_LONG_JUMP_TABLE_OFFSETS.
26553 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26554 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26556 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26557 David Holsgrove <david.holsgrove@xilinx.com>
26559 * common/config/microblaze/microblaze-common.c
26560 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26561 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26563 * config/microblaze/microblaze.c (microblaze_must_save_register)
26564 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26566 (microblaze_eh_return): New function.
26567 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26568 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26569 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26570 * config/microblaze/microblaze.md (eh_return): New pattern.
26572 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26574 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26575 GCC_DIAGNOSTIC_STRINGIFY): Define.
26577 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26579 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26581 * config/arm/arm.md (<mcrr>): New.
26583 * config/arm/arm.c (arm_arch5te): New.
26584 (arm_option_override): Set arm_arch5te.
26585 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26587 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26588 (arm_mcrr_qualifiers): ... this. New.
26589 (MRRC_QUALIFIERS): Define to...
26590 (arm_mrrc_qualifiers): ... this. New.
26591 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26593 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26594 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26595 (MRRCI, mrrc, MRRC): New.
26596 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26597 VUNSPEC_MRRC2): New.
26599 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26601 * config/arm/arm.md (<mcr>): New.
26603 * config/arm/arm.c (arm_coproc_builtin_available): Add
26604 support for mcr, mrc, mcr2 and mrc2.
26605 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26606 (arm_mcr_qualifiers): ... this. New.
26607 (MRC_QUALIFIERS): Define to ...
26608 (arm_mrc_qualifiers): ... this. New.
26609 (MCR_QUALIFIERS): Define to ...
26610 (arm_mcr_qualifiers): ... this. New.
26611 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26613 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26614 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26615 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26616 VUNSPEC_MRC2): New.
26618 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26620 * config/arm/arm.md (*ldc): New.
26624 * config/arm/arm.c (arm_coproc_builtin_available): Add
26625 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26626 (arm_coproc_ldc_stc_legitimate_address): New.
26627 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26628 'qualifier_const_pointer'.
26629 (LDC_QUALIFIERS): Define to...
26630 (arm_ldc_qualifiers): ... this. New.
26631 (STC_QUALIFIERS): Define to...
26632 (arm_stc_qualifiers): ... this. New.
26633 * config/arm/arm-protos.h
26634 (arm_coproc_ldc_stc_legitimate_address): New.
26635 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26636 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26637 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26638 stc2, stcl, stc2l): New.
26639 * config/arm/constraints.md (Uz): New.
26640 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26641 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26642 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
26643 VUNSPEC_STC2L): New.
26645 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26647 * config/arm/arm.md (<cdp>): New.
26648 * config/arm/arm.c (neon_const_bounds): Rename this ...
26649 (arm_const_bounds): ... this.
26650 (arm_coproc_builtin_available): New.
26651 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
26652 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
26653 (CDP_QUALIFIERS): Define to...
26654 (arm_cdp_qualifiers): ... this. New.
26656 (arm_expand_builtin_args): Add case for 6 arguments.
26657 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26658 (arm_const_bounds): ... this.
26659 (arm_coproc_builtin_available): New.
26660 * config/arm/arm_acle.h (__arm_cdp): New.
26662 * config/arm/arm_acle_builtins.def (cdp): New.
26664 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26665 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26666 'arm_const_bounds'.
26667 * config/arm/types.md (coproc): New.
26668 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26669 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26670 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26671 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26673 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26675 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26676 (UBINOP_QUALIFIERS): New.
26678 (acle_builtin_data): New. Change comment.
26679 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26680 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26681 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26682 arm_acle_builtins.def.
26683 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26684 (arm_init_acle_builtins): New.
26685 (CRC32_BUILTIN): Remove.
26686 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26687 crc32cb, crc32ch and crc32cw.
26688 (arm_init_crc32_builtins): Remove.
26689 (arm_init_builtins): Use arm_init_acle_builtins rather
26690 than arm_init_crc32_builtins.
26691 (arm_expand_acle_builtin): New.
26692 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26693 * config/arm/arm_acle_builtins.def: New.
26695 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26697 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26698 (arm_builtin_datum): ... this.
26699 (arm_init_neon_builtin): Rename to ...
26700 (arm_init_builtin): ... this. Add a new parameters PREFIX
26701 and USE_SIG_IN_NAME.
26702 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26703 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26704 'arm_builtin_datum'.
26705 (arm_init_vfp_builtins): Likewise.
26706 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26707 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26708 (arm_expand_neon_args): Rename to ...
26709 (arm_expand_builtin_args): ... this. Rename builtin_arg
26710 enum values and differentiate between ARG_BUILTIN_MEMORY
26711 and ARG_BUILTIN_NEON_MEMORY.
26712 (arm_expand_neon_builtin_1): Rename to ...
26713 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26714 values, arm_expand_builtin_args and add bool parameter NEON.
26715 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26716 (arm_expand_vfp_builtin): Likewise.
26717 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26719 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26721 PR middle-end/77484
26722 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26723 * predict.c (tree_estimate_probability_bb): Reverse direction of
26724 polymorphic call predictor.
26726 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26728 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26729 (determine_pass_name_match): ...this new function and...
26730 (should_skip_pass_p): ...this new function.
26732 2017-01-06 Nathan Sidwell <nathan@acm.org>
26734 * ipa-visibility.c (function_and_variable_visibility): Reformat
26735 comments and long lines. Remove extrneous if.
26736 * symtab.c (symtab_node::make_decl_local): Fix code format.
26737 (symtab_node::set_section_for_node): Fix comment typo.
26739 2017-01-06 Martin Liska <mliska@suse.cz>
26742 * lra-constraints.c: Rename invariant to lra_invariant.
26743 * predict.c (set_even_probabilities): Initialize e to NULL.
26745 2017-01-05 Martin Sebor <msebor@redhat.com>
26747 PR tree-optimization/78910
26748 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
26749 (format_integer): Correct off-by-one error in the handling
26750 of precision with negative numbers in signed conversions..
26752 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
26754 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
26756 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26758 PR tree-optimization/71016
26759 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
26760 factor_out_conditional_conversion. Formatting fix.
26761 (factor_out_conditional_conversion): Add cond_stmt argument.
26762 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
26763 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
26766 2017-01-05 David Malcolm <dmalcolm@redhat.com>
26768 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
26769 read-rtl-function.o, and selftest-rtl.o.
26770 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
26771 (selftest::aarch64_test_loading_full_dump): New function.
26772 (selftest::aarch64_run_selftests): New function.
26773 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
26774 selftest::aarch64_run_selftests.
26775 * config/i386/i386.c
26776 (selftest::ix86_test_loading_dump_fragment_1): New function.
26777 (selftest::ix86_test_loading_call_insn): New function.
26778 (selftest::ix86_test_loading_full_dump): New function.
26779 (selftest::ix86_test_loading_unspec): New function.
26780 (selftest::ix86_run_selftests): Call the new functions.
26781 * emit-rtl.c (maybe_set_max_label_num): New function.
26782 * emit-rtl.h (maybe_set_max_label_num): New decl.
26783 * function.c (instantiate_decls): Guard call to
26784 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
26785 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
26787 * gensupport.c (gen_reader::gen_reader): Pass "false"
26788 for new "compact" param of rtx_reader.
26789 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
26790 rather than an empty string for NULL strings.
26791 * read-md.c: Potentially include config.h rather than bconfig.h.
26792 Wrap include of errors.h with #ifdef GENERATOR_FILE.
26793 (have_error): New global, copied from errors.c.
26794 (md_reader::read_name): Rename to...
26795 (md_reader::read_name_1): ...this, adding "out_loc" param,
26796 and converting "missing name or number" to returning false, rather
26798 (md_reader::read_name): Reimplement in terms of read_name_1.
26799 (md_reader::read_name_or_nil): New function.
26800 (md_reader::read_string): Handle "(nil)" by returning NULL.
26801 (md_reader::md_reader): Add new param "compact".
26802 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
26803 (md_reader::read_file): New method.
26804 * read-md.h (md_reader::md_reader): Add new param "compact".
26805 (md_reader::read_file): New method.
26806 (md_reader::is_compact): New accessor.
26807 (md_reader::read_name): Convert return type from void to file_location.
26808 (md_reader::read_name_or_nil): New decl.
26809 (md_reader::read_name_1): New decl.
26810 (md_reader::m_compact): New field.
26811 (noop_reader::noop_reader): Pass "false" for new "compact" param
26813 (rtx_reader::rtx_reader): Add new "compact" param.
26814 (rtx_reader::read_rtx_operand): Make virtual and convert return
26815 type from void to rtx.
26816 (rtx_reader::read_until): New decl.
26817 (rtx_reader::handle_any_trailing_information): New virtual function.
26818 (rtx_reader::postprocess): New virtual function.
26819 (rtx_reader::finalize_string): New virtual function.
26820 (rtx_reader::m_in_call_function_usage): New field.
26821 (rtx_reader::m_reuse_rtx_by_id): New field.
26822 * read-rtl-function.c: New file.
26823 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
26824 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
26825 (selftest::verify_three_block_rtl_cfg): New decl.
26826 * read-rtl-function.h: New file.
26827 * read-rtl.c: Potentially include config.h rather than bconfig.h.
26828 For host, include function.h, memmodel.h, and emit-rtl.h.
26829 (one_time_initialization): New function.
26830 (struct compact_insn_name): New struct.
26831 (compact_insn_names): New array.
26832 (find_code): Handle insn codes in compact dumps.
26833 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
26834 (bind_subst_iter_and_attr): Likewise.
26835 (add_condition_to_string): Likewise.
26836 (add_condition_to_rtx): Likewise.
26837 (apply_attribute_uses): Likewise.
26838 (add_current_iterators): Likewise.
26839 (apply_iterators): Likewise.
26840 (initialize_iterators): Guard usage of apply_subst_iterator with
26841 #ifdef GENERATOR_FILE.
26842 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
26843 (md_reader::read_mapping): Likewise.
26844 (add_define_attr_for_define_subst): Likewise.
26845 (add_define_subst_attr): Likewise.
26846 (read_subst_mapping): Likewise.
26847 (check_code_iterator): Likewise.
26848 (rtx_reader::read_rtx): Likewise. Move one-time initialization
26850 (one_time_initialization): New function.
26851 (rtx_reader::read_until): New method.
26852 (read_flags): New function.
26853 (parse_reg_note_name): New function.
26854 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
26855 Handle reuse_rtx ids.
26856 Wrap iterator lookup within #ifdef GENERATOR_FILE.
26857 Add parsing support for RTL dumps, mirroring the special-cases in
26858 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
26859 values, and calling handle_any_trailing_information.
26860 (rtx_reader::read_rtx_operand): Convert return type from void
26861 to rtx, returning return_rtx. Handle case 'e'. Call
26862 finalize_string on XSTR and XTMPL fields.
26863 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
26864 "(nil)" values were omitted. Call the postprocess vfunc on the
26866 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
26867 class ctor. Initialize m_in_call_function_usage. Call
26868 one_time_initialization.
26869 * rtl-tests.c (selftest::test_uncond_jump): Call
26870 set_new_first_and_last_insn.
26871 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
26872 * selftest-rtl.c: New file.
26873 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
26874 (selftest::get_insn_by_uid): New decl.
26875 * selftest-run-tests.c (selftest::run_tests): Call
26876 read_rtl_function_c_tests.
26877 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
26878 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
26881 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
26883 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
26884 operands in a special way. Assert that pos+len <= mode precision.
26886 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26888 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
26889 3 argument Alias with unlimited for the negative form.
26890 (fno-vect-cost-model): Removed.
26892 2017-01-05 Martin Liska <mliska@suse.cz>
26894 * hsa-gen.c (gen_hsa_divmod): New function.
26895 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
26897 2017-01-05 Martin Liska <mliska@suse.cz>
26900 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
26903 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26905 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
26906 small constant length operands.
26908 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26910 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
26911 between loop iterations.
26913 2017-01-05 Martin Liska <mliska@suse.cz>
26916 * gimplify.c (gimplify_decl_expr): Compare to
26917 asan_poisoned_variables instread of checking flags.
26918 (gimplify_target_expr): Likewise.
26919 (gimplify_expr): Likewise.
26920 (gimplify_function_tree): Conditionally initialize
26921 asan_poisoned_variables.
26923 2017-01-04 Jeff Law <law@redhat.com>
26925 PR tree-optimizatin/78812
26926 * rtl.h (contains_mem_rtx_p): Prototype.
26927 * ifcvt.c (containts_mem_rtx_p): Move from here to...
26928 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
26929 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
26930 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
26931 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
26933 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26935 * input.c (assert_char_at_range): Default-initialize actual_range.
26937 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26939 * df-scan.c (df_ref_create_structure): Make regno unsigned,
26940 to match the caller.
26942 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26944 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
26945 insns after final jump in test to emit dummy move.
26947 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26949 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
26950 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
26952 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26954 * multiple_target.c (create_dispatcher_calls): Init e_next.
26955 * tree-ssa-loop-split.c (split_loop): Init border.
26956 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
26959 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
26964 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
26965 (altivec_register_operand): Do not return true if the operand
26966 contains a SUBREG mixing SImode and SFmode.
26967 (vsx_register_operand): Likewise.
26968 (vsx_reg_sfsubreg_ok): New predicate.
26969 (vfloat_operand): Do not return true if the operand contains a
26970 SUBREG mixing SImode and SFmode.
26971 (vint_operand): Likewise.
26972 (vlogical_operand): Likewise.
26973 (gpc_reg_operand): Likewise.
26974 (int_reg_operand): Likewise.
26975 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
26976 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
26977 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
26979 (rs6000_emit_move_si_sf_subreg): New helper function.
26980 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
26981 fixup SUBREGs involving SImode and SFmode.
26982 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
26983 numbers for the new peephole2 optimization.
26984 (peephole2 for SFmode unions): New peephole2 to optimize cases in
26985 the GLIBC math library that do AND/IOR/XOR operations on single
26986 precision floating point.
26987 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26988 target macros to say whether we need to avoid SUBREGs mixing
26990 (TARGET_ALLOW_SF_SUBREG): Likewise.
26991 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26992 (UNSPEC_SI_FROM_SF): Likewise.
26993 (iorxor): Change spacing.
26994 (and_ior_xor): New iterator for AND, IOR, and XOR.
26995 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26996 (movdi_from_sf_zero_ext): Likewise.
26997 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26998 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26999 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
27000 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
27001 (fms<mode>4): Likewise.
27002 (fnma<mode>4): Likewise.
27003 (fnms<mode>4): Likewise.
27004 (nfma<mode>4): Likewise.
27005 (nfms<mode>4): Likewise.
27007 2017-01-04 Marek Polacek <polacek@redhat.com>
27010 * doc/invoke.texi: Document -Wpointer-compare.
27012 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27014 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
27017 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
27018 descriptions for -gdwarf-5 and emit them as uleb128 instead of
27021 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
27024 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
27025 documentation of the powerpc_popcntb_ok attribute.
27026 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27027 code to issue warning messages if a requested CPU configuration is
27028 not supported by the binary (assembler and loader) toolchain.
27029 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
27030 made to define a built-in function that has been disabled.
27031 (paired_init_builtins): Add assertion to prevent ICE if attempt is
27032 made to define a built-in function that has been disabled.
27033 (altivec_init_builtins): Add comment explaining why definition
27034 of the DST built-in functions is not preceded by an assertion
27035 check. Add assertions to prevent ICE if attempts are made to
27036 define an altivec predicate or an abs* built-in function that has
27038 (htm_init_builtins): Add comment explaining why definition of the
27039 htm built-in functions is not preceded by an assertion check.
27041 2017-01-04 Jeff Law <law@redhat.com>
27043 PR tree-optimizatin/67955
27044 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
27045 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
27046 the points-to solution does not include pt_null. Use DECL_PT_UID
27049 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
27051 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
27052 Use gen_int_mode instead of gen_lopwart for const_int operands.
27054 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27056 PR tree-optimization/71563
27057 * match.pd: Simplify X << Y into X if Y is known to be 0 or
27058 out of range value - has low bits known to be zero.
27060 2017-01-04 Alan Modra <amodra@gmail.com>
27062 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
27063 * configure: Regenerate.
27064 * config.in: Regenerate.
27066 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27069 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
27070 a substring of the message, but strcmp with the whole message. Ifdef
27071 ENABLE_NLS, translate the message first using dgettext.
27073 2017-01-03 Jeff Law <law@redhat.com>
27075 PR tree-optimizatin/78856
27076 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
27077 (mark_threaded_blocks): Remove code to truncate thread paths that
27078 cross multiple loop headers. Instead invalidate the cached loop
27079 iteration information and handle case of a thread path walking
27080 into an irreducible region.
27082 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
27085 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
27086 assertions. Add support for doing the signbit if the IEEE 128-bit
27087 floating point value is in a GPR.
27088 * config/rs6000/rs6000.md (Fsignbit): Delete.
27089 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
27090 Update the length attribute if the value is in a GPR.
27091 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
27092 the sign or zero extension instruction, since the value is always 0/1.
27093 (signbit<mode>2_dm2): Delete using <Fsignbit>.
27096 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
27097 extracting SImode to a GPR register so that we can generate a
27098 store, limit the vector to be in a traditional Altivec register
27099 for the vextuwrx instruction.
27101 2017-01-03 Ian Lance Taylor <iant@google.com>
27103 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
27105 2017-01-03 Martin Sebor <msebor@redhat.com>
27107 PR tree-optimization/78696
27108 * gimple-ssa-sprintf.c (format_floating): Correct handling of
27109 precision. Use MPFR for %f for greater fidelity. Correct handling
27111 (pass_sprintf_length::compute_format_length): Set width and precision
27112 specified by asrerisk to void_node for vararg functions.
27113 (try_substitute_return_value): Adjust dump output.
27115 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
27117 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
27119 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
27121 * doc/invoke.texi (SPARC options): Document -mlra as the default.
27122 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
27123 -mlra/-mno-lra was passed to the compiler.
27125 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
27127 PR rtl-optimization/65618
27128 * emit-rtl.c (try_split): Move initialization of "before" and
27129 "after" to just before the call to emit_insn_after_setloc.
27131 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
27133 * doc/md.texi (Standard Names): Remove reference to Java frontend.
27135 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
27137 * dwarf2out.c (gen_enumeration_type_die): When
27138 -gno-strict-dwarf, add a DW_AT_encoding attribute.
27140 2017-01-03 Jakub Jelinek <jakub@redhat.com>
27142 PR tree-optimization/78965
27143 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
27144 Change first argument from const call_info & to call_info &. For %n
27145 set info.nowrite to false.
27147 PR middle-end/78901
27148 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
27149 possibly throwing calls.
27151 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
27152 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
27153 and fns handling, rather than in a separate case SSA_NAME.
27155 2017-01-02 Jeff Law <law@redhat.com>
27157 * config/darwin-driver.c (darwin_driver_init): Const-correctness
27158 fixes for first_period and second_period variables.
27160 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
27163 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
27164 (*insvqi_1): New insn pattern.
27165 (*insvqi_1_mem_rex64): Ditto.
27166 (*insvqi_2): Ditto.
27167 (*insvqi_3): Rename from *insvqi.
27169 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
27171 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
27173 * doc/cfg.texi (Edges): Remove reference to Java.
27174 (Maintaining the CFG): Ditto.
27176 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27178 PR middle-end/77674
27179 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
27180 transparent aliases.
27182 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27184 PR middle-end/77484
27185 * predict.def (PRED_CALL): Update hitrate.
27186 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
27187 * predict.c (tree_estimate_probability_bb): Split CALL predictor
27188 into direct/indirect/polymorphic variants.
27190 2017-01-01 Jakub Jelinek <jakub@redhat.com>
27192 Update copyright years.
27194 * gcc.c (process_command): Update copyright notice dates.
27195 * gcov-dump.c (print_version): Ditto.
27196 * gcov.c (print_version): Ditto.
27197 * gcov-tool.c (print_version): Ditto.
27198 * gengtype.c (create_file): Ditto.
27199 * doc/cpp.texi: Bump @copying's copyright year.
27200 * doc/cppinternals.texi: Ditto.
27201 * doc/gcc.texi: Ditto.
27202 * doc/gccint.texi: Ditto.
27203 * doc/gcov.texi: Ditto.
27204 * doc/install.texi: Ditto.
27205 * doc/invoke.texi: Ditto.
27207 Copyright (C) 2017 Free Software Foundation, Inc.
27209 Copying and distribution of this file, with or without modification,
27210 are permitted in any medium without royalty provided the copyright
27211 notice and this notice are preserved.