1 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
4 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
5 switching to a new text section.
7 2017-09-18 Richard Biener <rguenther@suse.de>
9 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11 (build_alias_set): Reject aliases with no access function.
13 2017-09-18 Richard Biener <rguenther@suse.de>
15 PR tree-optimization/79622
16 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
18 (build_cross_bb_scalars_use): Likewise.
20 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
22 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
24 2017-09-18 Alan Modra <amodra@gmail.com>
27 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
28 stack_pointer_rtx for count 0. Update comments. Break up
31 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
33 config/i386/i386.c: (xlogue_layout::STUB_NAME_MAX_LEN): Increase to 20
35 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
36 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
37 or avx version of the stub.
39 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
42 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
43 compute the minimum stack alignment. Also update preferred stack
44 boundary for leaf functions.
46 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
48 PR tree-optimization/82228
49 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
52 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
54 * common/config/nds32/nds32-common.c
55 (nds32_option_optimization_table): Refine formatting.
56 (nds32_option_optimization_table): Use -fsched-pressure and
57 -fomit-frame-pointer for specific optimization level.
59 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
61 * config/nds32/nds32.c: Refine formatting and comments.
62 * config/nds32/nds32.h: Likewise.
63 * config/nds32/nds32.md: Likewise.
64 * config/nds32/nds32-cost.c: Likewise.
65 * config/nds32/nds32-isr.c: Likewise.
66 * config/nds32/nds32-md-auxiliary.c: Likewise.
67 * config/nds32/nds32-multiple.md: Likewise.
68 * config/nds32/nds32-predicates.c: Likewise.
70 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
71 Jakub Jelinek <jakub@redhat.com>
73 Add support for -std=c++2a.
74 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
76 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
78 2017-09-15 Steve Ellcey <sellcey@cavium.com>
81 * doc/extend.texi (Common Function Attributes): Add
82 references to ARM, AArch64, and S/390 specific attributes.
83 (Function Specific Option Pragmas): Add AArch64 and S/390
84 to list of back ends that support the target pragma.
86 2017-09-15 Nathan Sidwell <nathan@acm.org>
88 * doc/standards.texi: Fix C++17 description. Update URLs for
91 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
93 * common.opt (Wcast-align=strict): New warning option.
94 * doc/invoke.texi: Document -Wcast-align=strict.
96 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
98 * cgraph.h (cgraph_thunk_info): Add comments.
99 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
100 assert for VIRTUAL_* arguments stricter.
102 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
104 PR tree-optimization/71026
105 * match.pd: Move RDIV patterns from fold-const.c
106 * fold-const.c (distribute_real_division): Removed.
107 (fold_binary_loc): Remove calls to distribute_real_divison.
109 2017-09-15 Jakub Jelinek <jakub@redhat.com>
111 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
112 c++1z and gnu++1z as deprecated. Change other references to
113 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
114 Change -Wc++1z-compat to -Wc++17-compat.
115 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
116 * dwarf2out.c (highest_c_language): Handle C++17.
117 (gen_compile_unit_die): Likewise.
119 2017-09-15 Jakub Jelinek <jakub@redhat.com>
121 PR rtl-optimization/82192
122 * combine.c (make_extraction): Don't look through non-paradoxical
123 SUBREGs or TRUNCATE if pos + len is or might be bigger than
126 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
127 Alan Hayward <alan.hayward@arm.com>
128 David Sherwood <david.sherwood@arm.com>
130 * target.def (function_arg_offset): New hook.
131 * targhooks.h (default_function_arg_offset): Declare.
132 * targhooks.c (default_function_arg_offset): New function.
133 * function.c (locate_and_pad_parm): Use
134 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
135 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
136 (TARGET_FUNCTION_ARG_OFFSET): ...this.
137 * doc/tm.texi: Regenerate.
138 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
139 * config/spu/spu.c (spu_function_arg_offset): New function.
140 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
141 * system.h (FUNCTION_ARG_OFFSET): Poison.
143 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
144 Alan Hayard <alan.hayward@arm.com>
145 David Sherwood <david.sherwood@arm.com>
147 * target.def (truly_noop_truncation): New hook.
148 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
149 than TRULY_NOOP_TRUNCATION.
150 * hooks.h (hook_bool_uint_uint_true): Declare.
151 * hooks.c (hook_bool_uint_uint_true): New function.
152 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
153 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
154 * doc/tm.texi: Regenerate.
155 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
156 rather than TRULY_NOOP_TRUNCATION in comments.
157 (simplify_comparison): Likewise.
158 (record_truncated_value): Likewise.
159 * expmed.c (extract_bit_field_1): Likewise.
160 (extract_split_bit_field): Likewise.
161 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
162 instead of TRULY_NOOP_TRUNCATION.
163 * function.c (assign_parm_setup_block): Likewise.
164 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
165 * rtlhooks.c: Include target.h.
166 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
167 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
168 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
169 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
170 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
171 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
172 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
173 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
174 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
175 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
176 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
177 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
178 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
179 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
180 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
181 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
182 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
183 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
184 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
185 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
186 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
187 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
188 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
189 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
190 * config/mips/mips.c (mips_truly_noop_truncation): New function.
191 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
192 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
193 rather than TRULY_NOOP_TRUNCATION in comments.
194 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
195 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
196 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
197 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
198 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
199 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
200 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
201 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
202 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
203 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
204 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
205 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
206 rather than TRULY_NOOP_TRUNCATION in comments.
207 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
208 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
209 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
210 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
211 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
212 TRULY_NOOP_TRUNCATION condition.
213 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
214 (TRULY_NOOP_TRUNCATION): Delete.
215 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
216 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
217 * config/spu/spu.c (spu_truly_noop_truncation): New function.
218 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
219 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
220 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
221 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
222 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
223 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
224 rather than TRULY_NOOP_TRUNCATION in comments.
225 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
226 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
227 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
228 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
229 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
230 * system.h (TRULY_NOOP_TRUNCATION): Poison.
232 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
235 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
236 (*cmp_ior): Likewise.
237 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
238 (*ior_scc_scc_cmp): Likewise.
239 (*and_scc_scc): Likewise.
240 (*and_scc_scc_cmp): Likewise.
242 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
243 Alan Hayard <alan.hayward@arm.com>
244 David Sherwood <david.sherwood@arm.com>
246 * target.def (can_change_mode_class): New hook.
247 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
248 (hard_regno_nregs): Likewise.
249 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
250 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
251 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
252 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
253 (LOAD_EXTEND_OP): Update accordingly.
254 * doc/tm.texi: Regenerate.
255 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
256 CANNOT_CHANGE_MODE_CLASS.
257 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
258 (REG_CAN_CHANGE_MODE_P): ...this new macro.
259 * combine.c (simplify_set): Update accordingly.
260 * emit-rtl.c (validate_subreg): Likewise.
261 * recog.c (general_operand): Likewise.
262 * regcprop.c (mode_change_ok): Likewise.
263 * reload1.c (choose_reload_regs): Likewise.
264 (inherit_piecemeal_p): Likewise.
265 * rtlanal.c (simplify_subreg_regno): Likewise.
266 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
267 instead of CANNOT_CHANGE_MODE_CLASS.
268 (reload_cse_simplify_operands): Likewise.
269 * reload.c (push_reload): Use targetm.can_change_mode_class
270 instead of CANNOT_CHANGE_MODE_CLASS.
271 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
272 REG_CANNOT_CHANGE_MODE_P.
273 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
274 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
275 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
276 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
277 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
278 (arm_can_change_mode_class): New function.
279 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
280 than CANNOT_CHANGE_MODE_CLASS in comments.
281 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
282 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
283 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
284 (ix86_can_change_mode_class): ...this new function, inverting the
285 sense of the return value.
286 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
287 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
288 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
289 (ia64_can_change_mode_class): New function.
290 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
291 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
292 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
293 (m32c_can_change_mode_class): ...this new function, inverting the
294 sense of the return value.
295 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
296 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
297 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
298 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
299 (mips_can_change_mode_class): ...this new function, inverting the
300 sense of the return value.
301 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
302 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
303 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
304 (msp430_can_change_mode_class): New function.
305 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
306 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
307 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
308 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
309 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
310 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
311 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
312 (pa_cannot_change_mode_class): Replace with...
313 (pa_can_change_mode_class): ...this new function, inverting the
314 sense of the return value.
315 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
316 than CANNOT_CHANGE_MODE_CLASS in comments.
317 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
318 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
319 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
320 (pdp11_cannot_change_mode_class): Replace with...
321 (pdp11_can_change_mode_class): ...this new function, inverting the
322 sense of the return value.
323 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
324 * config/powerpcspe/powerpcspe-protos.h
325 (rs6000_cannot_change_mode_class_ptr): Delete.
326 * config/powerpcspe/powerpcspe.c
327 (rs6000_cannot_change_mode_class_ptr): Delete.
328 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
329 (rs6000_option_override_internal): Assign to
330 targetm.can_change_mode_class instead of
331 rs6000_cannot_change_mode_class_ptr.
332 (rs6000_cannot_change_mode_class): Replace with...
333 (rs6000_can_change_mode_class): ...this new function, inverting the
334 sense of the return value.
335 (rs6000_debug_cannot_change_mode_class): Replace with...
336 (rs6000_debug_can_change_mode_class): ...this new function.
337 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
338 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
339 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
340 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
341 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
343 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
344 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
345 (rs6000_option_override_internal): Assign to
346 targetm.can_change_mode_class instead of
347 rs6000_cannot_change_mode_class_ptr.
348 (rs6000_cannot_change_mode_class): Replace with...
349 (rs6000_can_change_mode_class): ...this new function, inverting the
350 sense of the return value.
351 (rs6000_debug_cannot_change_mode_class): Replace with...
352 (rs6000_debug_can_change_mode_class): ...this new function.
353 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
354 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
355 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
356 (s390_can_change_mode_class): ...this new function, inverting the
357 sense of the return value.
358 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
359 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
360 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
361 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
362 (sh_cannot_change_mode_class): Replace with...
363 (sh_can_change_mode_class): ...this new function, inverting the
364 sense of the return value.
365 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
366 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
367 (sparc_can_change_mode_class): New function.
368 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
369 * config/spu/spu.c (spu_can_change_mode_class): New function.
370 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
371 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
372 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
373 (visium_can_change_mode_class): New function.
374 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
376 2017-09-15 Richard Biener <rguenther@suse.de>
378 PR tree-optimization/82217
379 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
380 but not undefined case.
382 2017-09-15 Jakub Jelinek <jakub@redhat.com>
385 * postreload.c (reload_cse_simplify_operands): Skip
386 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
388 2017-09-15 Richard Biener <rguenther@suse.de>
390 PR tree-optimization/68823
391 * graphite-scop-detection.c (build_alias_set): If we have a
392 possible dependence check whether we can handle them by just
393 looking at the DRs DR_ACCESS_FNs.
394 (build_scops): If build_alias_set fails, fail the SCOP.
396 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
398 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
399 to support float128 built-in functions that require the ISA 3.0
401 (BU_FLOAT128_3_HW): Likewise.
402 (SQRTF128): Add support for the IEEE 128-bit square root and fma
406 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
407 support for built-in functions that need the ISA 3.0 IEEE 128-bit
408 floating point instructions.
409 (rs6000_invalid_builtin): Likewise.
410 (rs6000_builtin_mask_names): Likewise.
411 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
412 (RS6000_BTM_FLOAT128_HW): Likewise.
413 (RS6000_BTM_COMMON): Likewise.
414 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
416 * doc/extend.texi (RS/6000 built-in functions): Document the
417 IEEE 128-bit floating point square root and fused multiply-add
420 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
422 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
423 reg (r2) isn't in the set of registers defined in the prologue.
425 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
426 Alan Hayward <alan.hayward@arm.com>
427 David Sherwood <david.sherwood@arm.com>
429 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
430 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
431 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
432 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
433 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
435 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
436 max_vectorization_factor.
437 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
439 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
440 Alan Hayward <alan.hayward@arm.com>
441 David Sherwood <david.sherwood@arm.com>
443 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
444 (vect_worthwhile_without_simd_p): Declare.
445 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
446 (vectorizable_reduction): Use it.
447 * tree-vect-stmts.c (vectorizable_shift): Likewise.
448 (vectorizable_operation): Likewise.
450 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
451 Alan Hayward <alan.hayward@arm.com>
452 David Sherwood <david.sherwood@arm.com>
454 * tree-vectorizer.h (vect_get_num_copies): New function.
455 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
456 * tree-vect-loop.c (vectorizable_reduction): Likewise.
457 (vectorizable_induction): Likewise.
458 (vectorizable_live_operation): Likewise.
459 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
460 (vectorizable_bswap): Likewise.
461 (vectorizable_call): Likewise.
462 (vectorizable_conversion): Likewise.
463 (vectorizable_assignment): Likewise.
464 (vectorizable_shift): Likewise.
465 (vectorizable_operation): Likewise.
466 (vectorizable_store): Likewise.
467 (vectorizable_load): Likewise.
468 (vectorizable_condition): Likewise.
469 (vectorizable_comparison): Likewise.
470 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
472 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
473 Alan Hayward <alan.hayward@arm.com>
474 David Sherwood <david.sherwood@arm.com>
476 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
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 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
484 an inline wrapper that provides a location.
485 (gimple_build_vector): Likewise.
486 * gimple-fold.c (gimple_build_vector_from_val): New function.
487 (gimple_build_vector): Likewise.
488 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
489 functions to build the initial value. Always return a gimple value.
490 (get_initial_defs_for_reduction): Likewise. Only compute
492 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
493 vect_init_vector on the results from get_initial_def(s)_for_reduction.
494 (vectorizable_induction): Use gimple_build_vector rather than
497 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
498 Alan Hayward <alan.hayward@arm.com>
499 David Sherwood <david.sherwood@arm.com>
501 * target.h (vec_perm_indices): New typedef.
502 (auto_vec_perm_indices): Likewise.
503 * optabs-query.h: Include target.h
504 (can_vec_perm_p): Take a vec_perm_indices *.
505 * optabs-query.c (can_vec_perm_p): Likewise.
506 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
507 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
508 * tree-vect-generic.c (lower_vec_perm): Likewise.
509 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
510 (vect_grouped_load_supported): Likewise.
511 (vect_shift_permute_load_chain): Likewise.
512 (vect_permute_store_chain): Use auto_vec_perm_indices.
513 (vect_permute_load_chain): Likewise.
514 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
515 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
516 Update uses of can_vec_perm_p.
517 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
518 mode with a number of elements. Take a vec_perm_indices *.
519 (vect_create_epilog_for_reduction): Update accordingly.
520 Use auto_vec_perm_indices.
521 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
522 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
523 (vect_transform_slp_perm_load): Likewise.
524 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
525 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
526 (vect_gen_perm_mask_checked): Likewise.
527 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
528 (vect_gen_perm_mask_checked): Likewise.
529 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
530 (vectorizable_store): Likewise.
531 (vectorizable_load): Likewise.
532 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
533 (vectorizable_bswap): Likewise.
535 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
536 Alan Hayward <alan.hayward@arm.com>
537 David Sherwood <david.sherwood@arm.com>
539 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
540 * tree.c (build_vector): Likewise.
541 (build_vector_from_ctor): Update accordingly.
542 (build_vector_from_val): Likewise.
543 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
544 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
545 * tree-vect-generic.c (add_rshift): Likewise.
546 (expand_vector_divmod): Likewise.
547 (optimize_vector_constructor): Likewise.
548 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
549 (vect_transform_slp_perm_load): Likewise.
550 (vect_schedule_slp_instance): Likewise.
551 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
552 (vectorizable_call): Likewise.
553 (vect_gen_perm_mask_any): Likewise. Add elements in order.
554 * expmed.c (make_tree): Likewise.
555 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
556 a vector passed to build_vector.
557 (fold_convert_const): Likewise.
558 (exact_inverse): Likewise.
559 (fold_ternary_loc): Likewise.
560 (fold_relational_const): Likewise.
561 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
562 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
563 (const_unop): Likewise. Store the reduction accumulator in a
564 variable rather than an array.
565 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
566 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
567 the new vector, rather than constructing it after the input arrays.
568 (native_interpret_vector): Use auto_vec<tree> when building
569 a vector passed to build_vector. Add elements in order.
570 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
571 auto_vec<tree> when building a vector passed to build_vector.
572 (vect_create_epilog_for_reduction): Likewise.
573 (vectorizable_induction): Likewise.
574 (get_initial_def_for_reduction): Likewise. Fix indentation of
576 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
578 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
579 passed to build_vector.
581 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
582 Alan Hayward <alan.hayward@arm.com>
583 David Sherwood <david.sherwood@arm.com>
585 * tree-core.h (tree_base::u): Add an "nelts" field.
586 (tree_vector): Use VECTOR_CST_NELTS as the length.
587 * tree.c (tree_size): Likewise.
588 (make_vector): Initialize VECTOR_CST_NELTS.
589 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
590 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
591 TYPE_VECTOR_SUBPARTS.
592 * expr.c (const_vector_mask_from_tree): Consistently use "units"
593 as the number of units, setting it from VECTOR_CST_NELTS.
594 (const_vector_from_tree): Likewise.
595 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
596 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
597 (fold_negate_expr_1): Likewise.
598 (fold_convert_const): Likewise.
599 (const_binop): Likewise. Differentiate the number of output and
601 (const_unop): Likewise.
602 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
603 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
604 in cases that did the opposite.
606 2017-09-14 Richard Biener <rguenther@suse.de>
608 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
611 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
613 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
615 2017-09-14 Jakub Jelinek <jakub@redhat.com>
618 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
619 if and where to split a bb, except for splitting before debug insn
620 sequences followed by non-label real insn. Delete debug insns
621 in between basic blocks.
623 * combine.c (make_compound_operation_int): Formatting fixes.
625 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
626 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
627 * config/netbsd.h (LINK_EH_SPEC): Likewise.
628 * config/sol2.h (LINK_EH_SPEC): Likewise.
629 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
630 * config/s390/linux.h (LINK_SPEC): Likewise.
631 * config/freebsd.h (LINK_EH_SPEC): Likewise.
632 * config/openbsd.h (LINK_EH_SPEC): Likewise.
633 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
634 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
635 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
636 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
637 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
638 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
640 2017-09-13 Jakub Jelinek <jakub@redhat.com>
642 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
644 (ENDFILE_LINUX_SPEC): Likewise.
645 (LINK_EH_SPEC): Likewise.
646 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
647 (LINK_OS_LINUX_SPEC32): Likewise.
648 (LINK_OS_LINUX_SPEC64): Likewise.
649 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
650 (LINK_OS_LINUX_SPEC): Likewise.
652 2017-09-13 Martin Liska <mliska@suse.cz>
655 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
656 CASE_HIGH is NULL_TREE.
658 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
659 Alan Hayward <alan.hayward@arm.com>
660 David Sherwood <david.sherwood@arm.com>
662 * target.def (secondary_memory_needed): New hook.
663 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
664 instead of SECONDARY_MEMORY_NEEDED.
665 (secondary_memory_needed_mode): Likewise.
666 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
667 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
668 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
669 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
670 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
671 * doc/tm.texi: Regenerate.
672 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
673 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
674 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
675 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
676 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
677 * config/i386/i386.c (inline_secondary_memory_needed): Put the
678 mode argument first and change the reg_class arguments to reg_class_t.
679 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
680 Make static. Update the call to inline_secondary_memory_needed.
681 (ix86_register_move_cost): Update the call to
682 inline_secondary_memory_needed.
683 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
684 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
686 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
687 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
689 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
690 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
691 * config/mips/mips.c (mips_secondary_memory_needed): Make static
692 and match hook interface. Add comment from mips.h.
693 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
694 * config/mmix/mmix.md (truncdfsf2): Refer to
695 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
697 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
698 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
699 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
700 (pa_secondary_memory_needed): New function.
701 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
702 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
703 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
704 (pdp11_secondary_memory_needed): Make static and match hook interface.
705 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
706 * config/powerpcspe/powerpcspe-protos.h
707 (rs6000_secondary_memory_needed_ptr): Delete.
708 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
710 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
711 (rs6000_option_override_internal): Assign to
712 targetm.secondary_memory_needed rather than
713 rs6000_secondary_memory_needed_ptr.
714 (rs6000_secondary_memory_needed): Match hook interface.
715 (rs6000_debug_secondary_memory_needed): Likewise.
716 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
717 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
718 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
719 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
720 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
721 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
723 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
724 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
725 (rs6000_option_override_internal): Assign to
726 targetm.secondary_memory_needed rather than
727 rs6000_secondary_memory_needed_ptr.
728 (rs6000_secondary_memory_needed): Match hook interface.
729 (rs6000_debug_secondary_memory_needed): Likewise.
730 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
731 * config/s390/s390.c (s390_secondary_memory_needed): New function.
732 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
733 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
734 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
735 (sparc_secondary_memory_needed): New function.
736 * lra-constraints.c (check_and_process_move): Refer to
737 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
739 (curr_insn_transform): Likewise.
740 (process_alt_operands): Use targetm.secondary_memory_needed
741 instead of TARGET_SECONDARY_MEMORY_NEEDED.
742 (check_secondary_memory_needed_p): Likewise.
743 (choose_split_class): Likewise.
744 * reload.c: Unconditionally include code that was previously
745 conditional on SECONDARY_MEMORY_NEEDED.
746 (push_secondary_reload): Use targetm.secondary_memory_needed
747 instead of TARGET_SECONDARY_MEMORY_NEEDED.
748 (push_reload): Likewise.
749 * reload1.c: Unconditionally include code that was previously
750 conditional on SECONDARY_MEMORY_NEEDED.
751 (choose_reload_regs): Use targetm.secondary_memory_needed
752 instead of TARGET_SECONDARY_MEMORY_NEEDED.
753 (gen_reload): Likewise.
754 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
756 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
757 Alan Hayward <alan.hayward@arm.com>
758 David Sherwood <david.sherwood@arm.com>
760 * target.def (secondary_memory_needed_mode): New hook:
761 * targhooks.c (default_secondary_memory_needed_mode): Declare.
762 * targhooks.h (default_secondary_memory_needed_mode): New function.
763 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
764 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
765 * doc/tm.texi: Regenerate.
766 * lra-constraints.c (check_and_process_move): Use
767 targetm.secondary_memory_needed_mode instead of
768 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
769 (curr_insn_transform): Likewise.
770 * reload.c (get_secondary_mem): Likewise.
771 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
772 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
774 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
775 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
776 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
777 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
778 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
780 * config/powerpcspe/powerpcspe-protos.h
781 (rs6000_secondary_memory_needed_mode): Delete.
782 * config/powerpcspe/powerpcspe.c
783 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
784 (rs6000_secondary_memory_needed_mode): Make static.
785 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
786 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
788 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
790 (rs6000_secondary_memory_needed_mode): Make static.
791 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
792 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
793 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
794 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
795 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
797 (sparc_secondary_memory_needed_mode): New function.
798 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
800 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
802 * config/aarch64/constraints.md (Umq): New constraint.
803 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
805 (mov<mode>): Update condition.
807 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
809 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
810 when bitposition is the same.
812 2017-09-13 Richard Biener <rguenther@suse.de>
814 * dwarf2out.c (output_die_symbol): Remove.
815 (output_die): Do not output a DIEs symbol.
817 2017-09-13 Richard Biener <rguenther@suse.de>
820 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
821 default-def to avoid breaking iterator update with the weird
822 interaction with cgraph_update_edges_for_call_stmt_node.
824 2017-09-13 Richard Biener <rguenther@suse.de>
826 * tree-cfg.c (verify_gimple_assign_binary): Add verification
827 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
828 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
829 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
831 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
833 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
834 Disable pc relative literal load irrespective of
835 TARGET_FIX_ERR_A53_84341 for default.
837 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
839 * config/sparc/sparc.c (output_return): Output the source location of
840 the insn in the delay slot, if any.
841 (output_sibcall): Likewise.
843 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
846 * common.opt (-static-pie): New alias.
847 (shared): Negate static-pie.
848 (-no-pie): Update help text.
850 (static-pie): New option.
851 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
853 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
854 (LINK_EH_SPEC): Likewise.
855 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
856 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
857 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
858 * gcc.c (LINK_COMMAND_SPEC): Likewise.
859 (init_gcc_specs): Likewise.
860 (init_spec): Likewise.
861 (display_help): Update help message for -pie.
862 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
865 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
867 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
868 (movdi_aarch64): Likewise.
869 (movti_aarch64): Likewise.
871 2017-09-12 Simon Wright <simon@pushface.org>
874 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
875 calculation of the minor version, always output as 0.
877 2017-09-12 Jakub Jelinek <jakub@redhat.com>
880 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
881 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
882 on it early, rather than manual conversion late. For
883 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
884 instead of performing manual conversion.
886 2017-09-12 Carl Love <cel@us.ibm.com>
888 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
889 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
891 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
892 VMULOSW): Add definitions.
893 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
894 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
895 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
896 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
897 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
899 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
901 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
903 (movti_aarch64): Likewise.
904 (movdf_aarch64): Likewise.
905 (movtf_aarch64): Likewise.
906 (load_pairdi): Likewise.
907 (store_pairdi): Likewise.
908 (load_pairdf): Likewise.
909 (store_pairdf): Likewise.
910 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
911 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
912 (ldr_got_small_<mode>): Likewise.
913 (ldr_got_small_28k_<mode>): Likewise.
914 (ldr_got_tiny): Likewise.
915 * config/aarch64/iterators.md (ldst_sz): New.
916 (ldpstp_sz): Likewise.
917 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
919 (thunderx_load): Split load_8 to load_16.
920 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
922 (thunderx2t99_storepair_basic): Split store_8 to store_16.
923 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
924 (xgene1_store_pair): Split store_8 to store_16.
925 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
926 (falkor_st_0_st_sd): Split store_8 to store_16.
928 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
930 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
931 and store1/2/3/4 to store_4/8/12/16.
932 * config/aarch64/aarch64.md: Update for rename.
933 * config/arm/arm.md: Likewise.: Likewise.
934 * config/arm/arm.c: Likewise.
935 * config/arm/thumb1.md: Likewise.
936 * config/arm/thumb2.md: Likewise.
937 * config/arm/vfp.md: Likewise.
938 * config/arm/arm-generic.md: Likewise.
939 * config/arm/arm1020e.md: Likewise.
940 * config/arm/arm1026ejs.md: Likewise.
941 * config/arm/arm1136jfs.md: Likewise.
942 * config/arm/arm926ejs.md: Likewise.
943 * config/arm/cortex-a15.md: Likewise.
944 * config/arm/cortex-a17.md: Likewise.
945 * config/arm/cortex-a5.md: Likewise.
946 * config/arm/cortex-a53.md: Likewise.
947 * config/arm/cortex-a57.md: Likewise.
948 * config/arm/cortex-a7.md: Likewise.
949 * config/arm/cortex-a8.md: Likewise.
950 * config/arm/cortex-a9.md: Likewise.
951 * config/arm/cortex-m4.md: Likewise.
952 * config/arm/cortex-m7.md: Likewise.
953 * config/arm/cortex-r4.md: Likewise.
954 * config/arm/exynos-m1.md: Likewise.
955 * config/arm/fa526.md: Likewise.
956 * config/arm/fa606te.md: Likewise.
957 * config/arm/fa626te.md: Likewise.
958 * config/arm/fa726te.md: Likewise.
959 * config/arm/fmp626.md: Likewise.
960 * config/arm/iwmmxt.md: Likewise.
961 * config/arm/ldmstm.md: Likewise.
962 * config/arm/marvell-pj4.md: Likewise.
963 * config/arm/xgene1.md: Likewise.
964 * config/aarch64/thunderx.md: Likewise.
965 * config/aarch64/thunderx2t99.md: Likewise.
966 * config/aarch64/falkor.md: Likewise.
968 2017-09-12 Martin Liska <mliska@suse.cz>
970 * attribs.c (private_lookup_attribute): New function.
971 * attribs.h (private_lookup_attribute): Declared here.
972 (lookup_attribute): Called from this place.
974 2017-09-12 Richard Biener <rguenther@suse.de>
976 PR tree-optimization/82157
977 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
978 stmts with side-effects.
980 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
981 Alan Hayward <alan.hayward@arm.com>
982 David Sherwood <david.sherwood@arm.com>
984 * target.def (hard_regno_nregs): New hook.
985 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
986 * targhooks.h (default_hard_regno_nregs): Declare.
987 * targhooks.c (default_hard_regno_nregs): New function.
988 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
989 (TARGET_HARD_REGNO_NREGS): ...this hook.
990 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
991 (CLASS_MAX_NREGS): Likewise.
992 * doc/tm.texi: Regenerate.
993 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
994 instead of HARD_REGNO_NREGS.
995 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
996 HARD_REGNO_NREGS in the comment.
997 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
998 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
999 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
1000 Return an unsigned int.
1001 (TARGET_HARD_REGNO_NREGS): Redefine.
1002 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
1003 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
1004 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1005 (arc_hard_regno_nregs): New function.
1006 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
1007 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
1008 (arm_hard_regno_nregs): New function.
1009 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
1010 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
1011 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
1012 (TARGET_HARD_REGNO_NREGS): Redefine.
1013 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
1014 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
1015 (HARD_REGNO_NREGS): Delete.
1016 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
1017 (cr16_hard_regno_nregs): New function.
1018 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
1019 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
1020 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
1021 (cris_hard_regno_nregs): New function.
1022 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
1023 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
1024 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1025 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
1026 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
1027 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
1028 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
1029 (frv_hard_regno_nregs): Make static. Take and return an
1031 (frv_class_max_nregs): Remove outdated copy of documentation.
1032 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
1033 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
1034 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
1035 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
1036 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
1037 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
1038 (TARGET_HARD_REGNO_NREGS): Redefine.
1039 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
1040 (CLASS_MAX_NREGS): Update comment.
1041 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
1042 (ia64_hard_regno_nregs): New function.
1043 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
1044 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
1045 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
1046 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
1047 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
1049 (m32c_hard_regno_nregs): Likewise. Make static.
1050 (TARGET_HARD_REGNO_NREGS): Redefine.
1051 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
1052 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
1053 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
1054 (m68k_hard_regno_nregs): New function.
1055 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
1056 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
1057 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
1058 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
1059 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
1060 Take and return an unsigned int.
1061 (TARGET_HARD_REGNO_NREGS): Redefine.
1062 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
1063 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
1064 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
1065 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
1066 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
1067 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
1068 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
1069 (msp430_hard_regno_nregs): Make static. Take and return an
1071 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
1072 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
1073 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
1074 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
1075 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
1076 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
1077 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
1078 (TARGET_HARD_REGNO_NREGS): Redefine.
1079 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
1080 (PA_HARD_REGNO_NREGS): ...this.
1081 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
1082 (PA_HARD_REGNO_NREGS): ...this.
1083 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1084 (pa_hard_regno_nregs): New function.
1085 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
1086 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
1087 (pdp11_hard_regno_nregs): New function.
1088 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
1089 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
1090 (rs6000_hard_regno_nregs_hook): New function.
1091 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
1092 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1093 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
1094 Take and return an unsigned int. Move earlier in file.
1095 (TARGET_HARD_REGNO_NREGS): Redefine.
1096 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
1097 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
1098 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
1099 (rl78_hard_regno_nregs): Make static. Take and return an
1101 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
1102 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
1103 (rs6000_hard_regno_nregs_hook): New function.
1104 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
1105 * config/rx/rx.c (rx_hard_regno_nregs): New function.
1106 (TARGET_HARD_REGNO_NREGS): Redefine.
1107 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
1108 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
1109 instead of HARD_REGNO_NREGS.
1110 (s390_hard_regno_nregs): New function.
1111 (s390_hard_regno_mode_ok): Add comment from s390.h.
1112 (TARGET_HARD_REGNO_NREGS): Redefine.
1113 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
1114 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
1115 (sh_hard_regno_nregs): New function.
1116 (sh_pass_in_reg_p): Use it.
1117 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
1118 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1119 (sparc_hard_regno_nregs): New function.
1120 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
1121 * config/spu/spu.c (spu_hard_regno_nregs): New function.
1122 (spu_function_arg_advance): Use it, supplying a valid register number.
1123 (TARGET_HARD_REGNO_NREGS): Redefine.
1124 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
1125 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
1126 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
1127 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
1128 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
1129 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
1130 (CLASS_MAX_NREGS): Remove copy of old documentation.
1131 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
1132 (visium_hard_regno_nregs): New function.
1133 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
1134 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
1135 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1136 xtensa_hard_regno_nregs): New function.
1137 * system.h (HARD_REGNO_NREGS): Poison.
1139 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1141 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
1142 hard_regno_nregs instead of HARD_REGNO_NREGS.
1143 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
1144 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
1145 (c6x_expand_epilogue): Likewise.
1146 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
1147 (frv_read_iacc_argument): Likewise.
1148 * config/sh/sh.c: Include regs.h.
1149 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1150 (regs_used): Likewise.
1151 (output_stack_adjust): Likewise.
1152 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
1153 * expmed.c: Include regs.h.
1154 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1155 * ree.c: Include regs.h.
1156 (combine_reaching_defs): Use hard_regno_nregs instead of
1158 (add_removable_extension): Likewise.
1160 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1162 * regs.h (hard_regno_nregs): Turn into a function.
1163 (end_hard_regno): Update accordingly.
1164 * caller-save.c (setup_save_areas): Likewise.
1165 (save_call_clobbered_regs): Likewise.
1166 (replace_reg_with_saved_mem): Likewise.
1167 (insert_restore): Likewise.
1168 (insert_save): Likewise.
1169 * combine.c (can_change_dest_mode): Likewise.
1170 (move_deaths): Likewise.
1171 (distribute_notes): Likewise.
1172 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
1173 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
1174 (rs6000_split_multireg_move): Likewise.
1175 (rs6000_register_move_cost): Likewise.
1176 (rs6000_memory_move_cost): Likewise.
1177 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
1178 (rs6000_split_multireg_move): Likewise.
1179 (rs6000_register_move_cost): Likewise.
1180 (rs6000_memory_move_cost): Likewise.
1181 * cselib.c (cselib_reset_table): Likewise.
1182 (cselib_lookup_1): Likewise.
1183 * emit-rtl.c (set_mode_and_regno): Likewise.
1184 * function.c (aggregate_value_p): Likewise.
1185 * ira-color.c (setup_profitable_hard_regs): Likewise.
1186 (check_hard_reg_p): Likewise.
1187 (calculate_saved_nregs): Likewise.
1188 (assign_hard_reg): Likewise.
1189 (improve_allocation): Likewise.
1190 (calculate_spill_cost): Likewise.
1191 * ira-emit.c (modify_move_list): Likewise.
1192 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
1193 (ira_hard_reg_in_set_p): Likewise.
1194 * ira.c (setup_reg_mode_hard_regset): Likewise.
1195 (clarify_prohibited_class_mode_regs): Likewise.
1196 (check_allocation): Likewise.
1197 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1198 (lra_setup_reg_renumber): Likewise.
1199 (setup_try_hard_regno_pseudos): Likewise.
1200 (spill_for): Likewise.
1201 (assign_hard_regno): Likewise.
1202 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1203 * lra-constraints.c (in_class_p): Likewise.
1204 (lra_constraint_offset): Likewise.
1205 (simplify_operand_subreg): Likewise.
1206 (lra_constraints): Likewise.
1207 (split_reg): Likewise.
1208 (split_if_necessary): Likewise.
1209 (invariant_p): Likewise.
1210 (inherit_in_ebb): Likewise.
1211 * lra-lives.c (process_bb_lives): Likewise.
1212 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
1213 (get_hard_regs): Likewise.
1214 (do_remat): Likewise.
1215 * lra-spills.c (assign_spill_hard_regs): Likewise.
1216 * mode-switching.c (create_pre_exit): Likewise.
1217 * postreload.c (reload_combine_recognize_pattern): Likewise.
1218 * recog.c (peep2_find_free_register): Likewise.
1219 * regcprop.c (kill_value_regno): Likewise.
1220 (set_value_regno): Likewise.
1221 (copy_value): Likewise.
1222 (maybe_mode_change): Likewise.
1223 (find_oldest_value_reg): Likewise.
1224 (copyprop_hardreg_forward_1): Likewise.
1225 * regrename.c (check_new_reg_p): Likewise.
1226 (regrename_do_replace): Likewise.
1227 * reload.c (push_reload): Likewise.
1228 (combine_reloads): Likewise.
1229 (find_dummy_reload): Likewise.
1230 (operands_match_p): Likewise.
1231 (find_reloads): Likewise.
1232 (find_equiv_reg): Likewise.
1233 (reload_adjust_reg_for_mode): Likewise.
1234 * reload1.c (count_pseudo): Likewise.
1235 (count_spilled_pseudo): Likewise.
1236 (find_reg): Likewise.
1237 (clear_reload_reg_in_use): Likewise.
1238 (free_for_value_p): Likewise.
1239 (allocate_reload_reg): Likewise.
1240 (choose_reload_regs): Likewise.
1241 (reload_adjust_reg_for_temp): Likewise.
1242 (emit_reload_insns): Likewise.
1243 (delete_output_reload): Likewise.
1244 * rtlanal.c (subreg_get_info): Likewise.
1245 * sched-deps.c (sched_analyze_reg): Likewise.
1246 * sel-sched.c (init_regs_for_mode): Likewise.
1247 (mark_unavailable_hard_regs): Likewise.
1248 (choose_best_reg_1): Likewise.
1249 (verify_target_availability): Likewise.
1250 * valtrack.c (dead_debug_insert_temp): Likewise.
1251 * var-tracking.c (track_loc_p): Likewise.
1252 (emit_note_insn_var_location): Likewise.
1253 * varasm.c (make_decl_rtl): Likewise.
1254 * reginfo.c (choose_hard_reg_mode): Likewise.
1255 (init_reg_modes_target): Refer directly to
1256 this_target_regs->x_hard_regno_nregs.
1258 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1260 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
1261 instead of hard_regno_nregs.
1263 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1265 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
1266 end_hard_regno instead of hard_regno_nregs.
1267 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
1268 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
1269 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
1270 * ira-color.c (improve_allocation): Likewise.
1271 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1272 * lra-lives.c (mark_regno_live): Likewise.
1273 (mark_regno_dead): Likewise.
1274 * lra-remat.c (operand_to_remat): Likewise.
1275 * lra.c (collect_non_operand_hard_regs): Likewise.
1276 * postreload.c (reload_combine_note_store): Likewise.
1277 (move2add_valid_value_p): Likewise.
1278 * reload.c (regno_clobbered_p): Likewise.
1280 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1282 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
1284 * config/v850/v850.c (v850_reorg): Likewise.
1285 * reload.c (refers_to_regno_for_reload_p): Likewise.
1286 (find_equiv_reg): Likewise.
1287 * reload1.c (reload_reg_reaches_end_p): Likewise.
1289 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1291 * caller-save.c (add_used_regs): Use REG_NREGS instead of
1293 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
1294 * config/arm/arm.c (output_move_neon): Likewise.
1295 (arm_attr_length_move_neon): Likewise.
1296 (neon_split_vcombine): Likewise.
1297 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
1298 (c6x_mark_reg_written): Likewise.
1299 (c6x_dwarf_register_span): Likewise.
1300 * config/i386/i386.c (ix86_save_reg): Likewise.
1301 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
1302 (rws_access_reg): Likewise.
1303 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
1304 * mode-switching.c (create_pre_exit): Likewise.
1305 * ree.c (combine_reaching_defs): Likewise.
1306 (add_removable_extension): Likewise.
1307 * regcprop.c (find_oldest_value_reg): Likewise.
1308 (copyprop_hardreg_forward_1): Likewise.
1309 * reload.c (reload_inner_reg_of_subreg): Likewise.
1310 (push_reload): Likewise.
1311 (combine_reloads): Likewise.
1312 (find_dummy_reload): Likewise.
1313 (reload_adjust_reg_for_mode): Likewise.
1314 * reload1.c (find_reload_regs): Likewise.
1315 (forget_old_reloads_1): Likewise.
1316 (reload_reg_free_for_value_p): Likewise.
1317 (reload_adjust_reg_for_temp): Likewise.
1318 (emit_reload_insns): Likewise.
1319 (delete_output_reload): Likewise.
1320 * sel-sched.c (choose_best_reg_1): Likewise.
1321 (choose_best_pseudo_reg): Likewise.
1323 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1324 Alan Hayward <alan.hayward@arm.com>
1325 David Sherwood <david.sherwood@arm.com>
1327 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
1328 * target.def (slow_unaligned_access): New hook.
1329 * targhooks.h (default_slow_unaligned_access): Declare.
1330 * targhooks.c (default_slow_unaligned_access): New function.
1331 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
1332 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
1333 * doc/tm.texi: Regenerate.
1334 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
1335 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
1336 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
1338 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
1339 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
1341 (rs6000_slow_unaligned_access): New function.
1342 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1343 (expand_block_compare): Likewise.
1344 (expand_strn_compare): Likewise.
1345 (rs6000_rtx_costs): Likewise.
1346 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
1347 (riscv_slow_unaligned_access): Likewise.
1348 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
1349 (riscv_slow_unaligned_access_p): ...this and make static.
1350 (riscv_option_override): Update accordingly.
1351 (riscv_slow_unaligned_access): New function.
1352 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1353 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
1354 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1355 (rs6000_slow_unaligned_access): New function.
1356 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1357 (rs6000_rtx_costs): Likewise.
1358 * config/rs6000/rs6000-string.c (expand_block_compare)
1359 (expand_strn_compare): Use targetm.slow_unaligned_access instead
1360 of SLOW_UNALIGNED_ACCESS.
1361 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
1362 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
1363 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
1364 of SLOW_UNALIGNED_ACCESS.
1365 * expmed.c (simple_mem_bitfield_p): Likewise.
1366 * expr.c (alignment_for_piecewise_move): Likewise.
1367 (emit_group_load_1): Likewise.
1368 (emit_group_store): Likewise.
1369 (copy_blkmode_from_reg): Likewise.
1370 (emit_push_insn): Likewise.
1371 (expand_assignment): Likewise.
1372 (store_field): Likewise.
1373 (expand_expr_real_1): Likewise.
1374 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1375 * lra-constraints.c (simplify_operand_subreg): Likewise.
1376 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1377 * gimple-ssa-store-merging.c: Likewise in block comment at start
1379 * tree-ssa-strlen.c: Include target.h.
1380 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
1381 of SLOW_UNALIGNED_ACCESS.
1382 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
1384 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1386 PR rtl-optimization/82185
1387 * expmed.c (emit_store_flag_int): Only test tem if it has been
1390 2017-09-12 Richard Biener <rguenther@suse.de>
1393 * match.pd ((FTYPE) N CMP CST): Fix typo.
1395 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1397 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1399 (mips_near_type_p): Add 'short_call' attribute as a synonym
1401 * doc/extend.texi (short_call): Document new function attribute.
1403 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1406 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1407 assertion check that in the condition.
1408 (get_atomic_generic_size): Likewise. Before testing if parameter
1409 has pointer type, if it has array type, call for C++
1410 default_conversion to perform array-to-pointer conversion.
1412 2017-09-12 Richard Biener <rguenther@suse.de>
1414 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1415 for operations we cannot scalarize.
1417 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1419 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1420 vectors heap vectors. Clean up comments.
1421 Make visited_bbs a reference.
1422 (profitable_jump_thread_path): Make GC
1423 vectors heap vectors. Clean up comments.
1425 (convert_and_register_jump_thread_path): Make GC vectors heap
1427 (check_subpath_and_update_thread_path): Same. Clean up comments.
1428 Make visited_bbs a reference.
1429 (handle_phi): Abstract common code to to
1430 register_jump_thread_path_if_profitable.
1431 Rename VAR_BB to DEF_BB.
1433 Make GC vectors heap vectors.
1434 Make visited_bbs a reference.
1435 (handle_assignment): Same.
1436 (register_jump_thread_path_if_profitable): New.
1437 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1439 Make GC vectors heap vectors. Clean up comments.
1440 Make visited_bbs a reference.
1441 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1442 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1445 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1448 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1449 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1451 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1453 Revert r251800 and r251799.
1455 2017-09-11 Martin Jambor <mjambor@suse.cz>
1458 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1459 arguments in advance.
1460 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1461 use it to find predecessor edges.
1462 (naive_outof_ssa): Collect vector of predecessors.
1464 2017-09-08 Jason Merrill <jason@redhat.com>
1466 PR c++/70029 - ICE with ref-qualifier and -flto
1467 * langhooks.h (struct lang_hooks_for_types): Add
1468 copy_lang_qualifiers.
1469 * attribs.c (build_type_attribute_qual_variant): Use it.
1470 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1472 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1473 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1475 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1478 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1479 (*mulsi3_sp64): New instruction.
1480 (mulsi3): New expander.
1482 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1484 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1486 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1488 * sancov.c: Include memmodel.h.
1490 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1493 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1496 2017-09-07 Carl Love <cel@us.ibm.com>
1498 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1499 the sldi instruction.
1501 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1503 * sancov.c: Include tm_p.h.
1505 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1508 * output.h (switch_to_other_text_partition): New declaration.
1509 * varasm.c (switch_to_other_text_partition): New function.
1510 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1511 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1512 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1513 to the other text partition before emitting LCL label and switch back
1514 after emitting the word after it.
1516 2017-09-07 Richard Biener <rguenther@suse.de>
1518 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1519 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1520 critical edges here, after loop init.
1521 (pass_data_pre): Remove PROP_no_crit_edges flags.
1522 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1523 for valueization of call args to avoid leaking VN_TOP.
1524 (visit_use): Assert we do not visit default defs.
1525 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1526 Use error_mark_node to more easily detect leaking VN_TOP.
1527 All default-defs are varying, not VN_TOP. Mark them visited.
1528 (run_scc_vn): Make code match comment.
1530 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1532 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1533 OPTION_MASK_FLOAT128_KEYWORD.
1534 (POWERPC_MASKS): Likewise.
1535 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1536 support for the -mfloat128-type option, and make -mfloat128
1537 default on PowerPC Linux systems. Define or undefine
1538 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1539 Define __float128 to be __ieee128 if IEEE 128-bit support is
1540 enabled, or undefine it.
1541 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1542 Delete defining __FLOAT128_TYPE__.
1543 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1544 -mfloat128-type option and make -mfloat128 default on PowerPC
1546 (TARGET_FLOAT128_TYPE): Likewise.
1547 (-mfloat128-type): Likewise.
1548 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1549 Delete the -mfloat128-type option and make -mfloat128 default on
1550 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1551 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1552 128-bit floating point is enabled. Change tests from using
1553 -mfloat128-type to -mfloat128.
1554 (rs6000_mangle_type): Use the correct mangling for the __float128
1555 type even if normal long double is restricted to 64-bits.
1556 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1558 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1559 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1560 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1562 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1563 documentation for -mfloat128.
1565 2017-09-06 Olivier Hainque <hainque@adacore.com>
1567 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1569 2017-09-06 Wish Wu <wishwu007@gmail.com>
1570 Jakub Jelinek <jakub@redhat.com>
1572 * asan.c (initialize_sanitizer_builtins): Add
1573 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1574 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1575 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1576 BT_FN_VOID_UINT64_PTR variables.
1577 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1578 (BT_FN_VOID_UINT16_UINT16): Likewise.
1579 (BT_FN_VOID_UINT32_UINT32): Likewise.
1580 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1581 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1582 (BT_FN_VOID_UINT64_PTR): Likewise.
1583 * common.opt (flag_sanitize_coverage): New variable.
1584 (fsanitize-coverage=trace-pc): Remove.
1585 (fsanitize-coverage=): Add.
1586 * flag-types.h (enum sanitize_coverage_code): New enum.
1587 * fold-const.c (fold_range_test): Disable non-short-circuit
1588 optimization if flag_sanitize_coverage.
1589 (fold_truth_andor): Likewise.
1590 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1591 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1592 (coverage_sanitizer_opts): New array.
1593 (get_closest_sanitizer_option): Add OPTS argument, handle also
1594 OPT_fsanitize_coverage_.
1595 (parse_sanitizer_options): Adjusted to also handle
1596 OPT_fsanitize_coverage_.
1597 (common_handle_option): Add OPT_fsanitize_coverage_.
1598 * sancov.c (instrument_comparison, instrument_switch): New function.
1599 (sancov_pass): Add trace-cmp support.
1600 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1601 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1602 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1603 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1604 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1605 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1606 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1607 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1608 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1609 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1611 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1613 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1614 error. Only quit immediately if parsing is complete.
1615 (BEGIN): Initialize fatal_err and parse_done.
1616 (begin fpu, end fpu): Check number of arguments.
1617 (begin arch, end arch): Likewise.
1618 (begin cpu, end cpu): Likewise.
1619 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1620 (optalias): Likewise.
1622 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1624 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1625 * config/arm/arm-isa.h: Delete. Move definitions to ...
1626 * arm-cpus.in: ... here. Use new feature and fgroup values.
1627 * config/arm/arm.c (arm_option_override): Use lower case for feature
1629 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1630 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1631 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1633 (print_isa_bits_for): New function.
1634 (gen_isa): New function.
1635 (gen_comm_data): Use print_isa_bits_for.
1636 (define feature): New keyword.
1637 (define fgroup): New keyword.
1638 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
1639 (arm-isa.h): Add rule to generate file.
1640 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1641 case for feature bit names.
1643 2017-09-06 Richard Biener <rguenther@suse.de>
1645 * tree-ssa-pre.c (NECESSARY): Remove.
1646 (create_expression_by_pieces): Do not touch pass-local flags.
1647 (insert_into_preds_of_block): Likewise.
1648 (do_pre_regular_insertion): Likewise.
1649 (eliminate_insert): Likewise.
1650 (eliminate_dom_walker::before_dom_children): Likewise.
1651 (fini_eliminate): Do not look at inserted_exprs.
1652 (mark_operand_necessary): Remove.
1653 (remove_dead_inserted_code): Replace with simple work-list
1654 algorithm based on inserted_exprs and SSA uses.
1655 (pass_pre::execute): Re-order fini_eliminate and
1656 remove_dead_inserted_code.
1658 2017-09-06 Olivier Hainque <hainque@adacore.com>
1660 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1661 for VxWorks 7. Adjust surrounding comments.
1663 2017-09-06 Richard Biener <rguenther@suse.de>
1665 * gimple-ssa-strength-reduction.c
1666 (find_candidates_dom_walker::before_dom_children): Also allow
1669 2017-09-06 Richard Biener <rguenther@suse.de>
1671 PR tree-optimization/82108
1672 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1673 for gap in the non-permutation SLP case.
1675 2017-09-06 Martin Jambor <mjambor@suse.cz>
1677 PR tree-optimization/82078
1678 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1679 add_access_to_work_queue...
1680 (build_accesses_from_assign): ...here.
1681 (propagate_all_subaccesses): Make sure racc is the group
1682 representative, if there is one.
1684 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1687 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1690 2017-09-06 Richard Biener <rguenther@suse.de>
1692 * gimple-ssa-strength-reduction.c
1693 (find_candidates_dom_walker::before_doom_children): Use a
1694 type and not a mode check.
1696 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1699 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1701 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1703 2017-09-05 Jeff Law <law@redhat.com>
1705 PR tree-optimization/64910
1706 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1707 cases where we have 3 or more operands.
1709 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1712 * omp-low.c (lower_omp_for): Recompute tree invariant if
1713 gimple_omp_for_initial/final is ADDR_EXPR.
1716 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1717 into gimple val before gimplification fo the COND_EXPR.
1719 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1721 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1722 REGION_COPY argument.
1723 (thread_through_all_blocks): Remove unused argument to
1724 duplicate_thread_path.
1726 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1727 Alan Hayward <alan.hayward@arm.com>
1728 David Sherwood <david.sherwood@arm.com>
1730 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1731 Take a scalar_mode rather than a machine_mode.
1732 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1733 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1734 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1735 (aarch64_gen_adjusted_ldpstp): Likewise.
1736 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
1738 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1739 Alan Hayward <alan.hayward@arm.com>
1740 David Sherwood <david.sherwood@arm.com>
1742 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
1743 Take a scalar_int_mode instead of a machine_mode.
1744 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1745 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1746 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1747 (aarch64_simd_attr_length_rglist): Delete.
1748 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
1749 a scalar_int_mode instead of a machine_mode.
1750 (aarch64_add_offset): Likewise.
1751 (aarch64_internal_mov_immediate): Likewise
1752 (aarch64_add_constant_internal): Likewise.
1753 (aarch64_add_constant): Likewise.
1754 (aarch64_movw_imm): Likewise.
1755 (aarch64_rtx_arith_op_extract_p): Likewise.
1756 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1757 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1758 Remove assert that the mode isn't a vector.
1759 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1760 (aarch64_expand_mov_immediate): Update calls after above changes.
1761 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
1762 (aarch64_and_bitmask_imm): Check for scalar integer modes.
1763 (aarch64_move_imm): Likewise.
1764 (aarch64_can_const_movi_rtx_p): Likewise.
1765 (aarch64_strip_extend): Likewise.
1766 (aarch64_extr_rtx_p): Likewise.
1767 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
1768 a CONST_INT when the mode parameter is VOIDmode.
1769 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
1771 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1773 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
1774 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1775 (bitwise_type_for_mode): Update accordingly.
1777 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1779 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
1780 * stor-layout.c (mode_for_size_tree): Likewise.
1781 (mode_for_array): Update accordingly.
1782 (layout_decl): Likewise.
1783 (compute_record_mode): Likewise. Only set the mode once.
1785 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1787 * target.def (get_mask_mode): Change return type to opt_mode.
1789 * doc/tm.texi: Regenerate.
1790 * targhooks.h (default_get_mask_mode): Return an opt_mode.
1791 * targhooks.c (default_get_mask_mode): Likewise.
1792 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1793 * optabs-query.c (can_vec_mask_load_store_p): Update use of
1794 targetm.get_mask_mode.
1795 * tree.c (build_truth_vector_type): Likewise.
1797 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1799 * machmode.h (mode_for_vector): Return an opt_mode.
1800 * stor-layout.c (mode_for_vector): Likewise.
1801 (mode_for_int_vector): Update accordingly.
1802 (layout_type): Likewise.
1803 * config/i386/i386.c (emit_memmov): Likewise.
1804 (ix86_expand_set_or_movmem): Likewise.
1805 (ix86_expand_vector_init): Likewise.
1806 (ix86_get_mask_mode): Likewise.
1807 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
1809 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
1810 * expmed.c (extract_bit_field_1): Likewise.
1811 * expr.c (expand_expr_real_2): Likewise.
1812 * optabs-query.c (can_vec_perm_p): Likewise.
1813 (can_vec_mask_load_store_p): Likewise.
1814 * optabs.c (expand_vec_perm): Likewise.
1815 * targhooks.c (default_get_mask_mode): Likewise.
1816 * tree-vect-stmts.c (vectorizable_store): Likewise.
1817 (vectorizable_load): Likewise.
1818 (get_vectype_for_scalar_type_and_size): Likewise.
1820 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1822 * machmode.h (mode_for_int_vector): New function.
1823 * stor-layout.c (mode_for_int_vector): Likewise.
1824 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
1825 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
1826 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1827 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1828 (s390_expand_vcond): Likewise.
1830 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1832 * machmode.h (opt_machine_mode): New type.
1833 (opt_mode<T>): Allow construction from anything that can be
1835 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
1836 (mode_for_size): Return an opt_machine_mode.
1837 * stor-layout.c (mode_for_size): Likewise.
1838 (mode_for_size_tree): Update call accordingly.
1839 (bitwise_mode_for_mode): Likewise.
1840 (make_fract_type): Likewise.
1841 (make_accum_type): Likewise.
1842 * caller-save.c (replace_reg_with_saved_mem): Update call
1844 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1845 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1846 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1847 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1848 * expmed.c (extract_bit_field_1): Likewise.
1849 * reload.c (get_secondary_mem): Likewise.
1850 * varasm.c (assemble_integer): Likewise.
1851 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
1854 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1856 * machmode.h (decimal_float_mode_for_size): New function.
1857 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
1858 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
1859 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
1860 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
1861 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
1862 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
1864 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1866 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
1867 (get_builtin_sync_mode): Likewise.
1868 (expand_ifn_atomic_compare_exchange): Likewise.
1869 (expand_builtin_atomic_clear): Likewise.
1870 (expand_builtin_atomic_test_and_set): Likewise.
1871 (fold_builtin_atomic_always_lock_free): Likewise.
1872 * calls.c (compute_argument_addresses): Likewise.
1873 (emit_library_call_value_1): Likewise.
1874 (store_one_arg): Likewise.
1875 * combine.c (combine_instructions): Likewise.
1876 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
1877 * config/arm/arm.c (arm_function_value): Likewise.
1878 (aapcs_allocate_return_reg): Likewise.
1879 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
1880 * config/i386/i386.c (construct_container): Likewise.
1881 (ix86_gimplify_va_arg): Likewise.
1882 (ix86_expand_sse_cmp): Likewise.
1883 (emit_memmov): Likewise.
1884 (emit_memset): Likewise.
1885 (expand_small_movmem_or_setmem): Likewise.
1886 (ix86_expand_pextr): Likewise.
1887 (ix86_expand_pinsr): Likewise.
1888 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
1889 * config/microblaze/microblaze.c (microblaze_block_move_straight):
1891 * config/mips/mips.c (mips_function_value_1) Likewise.
1892 (mips_block_move_straight): Likewise.
1893 (mips_expand_ins_as_unaligned_store): Likewise.
1894 * config/powerpcspe/powerpcspe.c
1895 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1896 (rs6000_darwin64_record_arg_flush): Likewise.
1897 * config/rs6000/rs6000.c
1898 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1899 (rs6000_darwin64_record_arg_flush): Likewise.
1900 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
1901 (sparc_function_value_1): Likewise.
1902 * config/spu/spu.c (adjust_operand): Likewise.
1903 (spu_emit_branch_or_set): Likewise.
1904 (arith_immediate_p): Likewise.
1905 * emit-rtl.c (gen_lowpart_common): Likewise.
1906 * expr.c (expand_expr_real_1): Likewise.
1907 * function.c (assign_parm_setup_block): Likewise.
1908 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1909 * reload1.c (alter_reg): Likewise.
1910 * stor-layout.c (mode_for_vector): Likewise.
1911 (layout_type): Likewise.
1913 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1915 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
1916 (spu_convert_move): Likewise.
1917 * lower-subreg.c (resolve_simple_move): Likewise.
1919 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1922 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
1923 define_insn to a define_expand.
1924 (altivec_vsum2sws_direct): New define_insn.
1925 (altivec_vsumsws): Convert from a define_insn to a define_expand.
1927 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1929 * config/arm/arm.c (arm_option_params_internal): Improve setting of
1932 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
1937 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
1938 realign stack if stack alignment needed is less than incoming
1941 2017-09-05 Marek Polacek <polacek@redhat.com>
1944 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
1947 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1949 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
1951 2017-09-05 Richard Biener <rguenther@suse.de>
1953 PR tree-optimization/82084
1954 * fold-const.c (can_native_encode_string_p): Handle wide characters.
1956 2017-09-05 Richard Biener <rguenther@suse.de>
1958 PR tree-optimization/82102
1959 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
1961 2017-09-05 Martin Liska <mliska@suse.cz>
1963 PR tree-optimization/82032
1964 * tree-cfg.c (generate_range_test): New function.
1965 * tree-cfg.h (generate_range_test): Declared here.
1966 * tree-cfgcleanup.c (convert_single_case_switch): New function.
1967 (cleanup_control_expr_graph): Use it.
1968 * tree-switch-conversion.c (try_switch_expansion): Remove
1970 (emit_case_nodes): Use generate_range_test.
1972 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1975 * config/i386/i386.md (*<btsc><mode>_mask): Add
1976 TARGET_USE_BT to insn constraint.
1977 (*btr<mode>_mask): Ditto.
1979 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1981 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1982 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1984 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1987 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1988 TARGET_NEON and TARGET_IWMMXT.
1989 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1990 TARGET_NEON and TARGET_IWMMXT.
1991 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
1993 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1995 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
1996 (ix86_rewrite_tls_address): Ditto.
1997 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
1998 (ix86_rewrite_tls_address_1): Ditto.
1999 (ix86_rewrite_tls_address): Ditto.
2000 * config/i386/predicates.md (tls_address_pattern): New predicate.
2001 * config/i386/i386.md (TLS address splitter): New splitter.
2003 2017-09-04 Richard Biener <rguenther@suse.de>
2005 PR tree-optimization/82084
2006 * fold-const.h (can_native_encode_string_p): Declare.
2007 * fold-const.c (can_native_encode_string_p): Factor out from ...
2008 (native_encode_string): ... here.
2009 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
2010 vectorizing stores from constants we later cannot handle.
2012 2017-09-04 Marek Polacek <polacek@redhat.com>
2015 * doc/invoke.texi: Update -Wtautological-compare documentation.
2017 2017-09-04 Jeff Law <law@redhat.com>
2019 PR tree-optimization/64910
2020 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
2021 swap the first and last operand if the last is a constant.
2023 2017-09-04 Marek Polacek <polacek@redhat.com>
2026 * convert.c (do_narrow): When sanitizing signed integer overflows,
2027 bail out for signed types.
2028 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
2030 2017-09-04 Richard Biener <rguenther@suse.de>
2032 PR tree-optimization/82060
2033 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
2034 Move devirtualization after stmt folding and before EH/AB/noreturn
2035 cleanup to get the stmt refs canonicalized. Use a bool instead
2036 of gimple_modified_p since that doesn't work for NOPs. Schedule
2037 NOPs generated by folding for removal.
2039 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2040 Alan Hayward <alan.hayward@arm.com>
2041 David Sherwood <david.sherwood@arm.com>
2043 * coretypes.h (pad_direction): New enum.
2044 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
2045 (FUNCTION_ARG_PADDING): Likewise.
2046 * target.def (function_arg_padding): New hook.
2047 * targhooks.h (default_function_arg_padding): Declare.
2048 * targhooks.c (default_function_arg_padding): New function.
2049 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
2050 (TARGET_FUNCTION_ARG_PADDING): ...this.
2051 * doc/tm.texi: Regenerate.
2052 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
2053 instead of direction.
2054 (compute_argument_addresses): Likewise.
2055 (load_register_parameters): Likewise.
2056 (emit_library_call_value_1): Likewise.
2057 (store_one_arg): Use targetm.calls.function_arg_padding instead
2058 of FUNCTION_ARG_PADDING.
2059 (must_pass_in_stack_var_size_or_pad): Likewise.
2060 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
2061 (emit_group_store): Likewise.
2062 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
2063 instead of FUNCTION_ARG_PADDING.
2064 (emit_push_insn): Likewise, and propagate enum change throughout
2066 * function.h (direction): Delete.
2067 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
2069 * function.c (assign_parm_find_stack_rtl): Likewise.
2070 (assign_parm_setup_block_p): Likewise.
2071 (assign_parm_setup_block): Likewise.
2072 (gimplify_parameters): Likewise.
2073 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
2074 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
2076 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
2077 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2078 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
2079 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
2080 (aarch64_function_arg_padding): ...this new function.
2081 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
2082 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2083 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
2084 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2085 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
2086 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2087 (arm_pad_arg_upward): Replace with...
2088 (arm_function_arg_padding): ...this new function.
2089 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
2091 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
2092 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
2093 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2094 (ia64_hpux_function_arg_padding): Replace with...
2095 (ia64_function_arg_padding): ...this new function. Use pad_direction
2096 instead of direction. Check for TARGET_HPUX.
2097 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
2098 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2099 (iq2000_function_arg_padding): New function.
2100 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
2101 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
2102 (mips_function_arg_padding): ...this new function.
2103 (mips_pad_reg_upward): Update accordingly.
2104 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2105 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
2106 targetm.calls.function_arg_padding.
2107 (FUNCTION_ARG_PADDING): Delete.
2108 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2109 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
2110 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2111 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
2112 (nios2_block_reg_padding): Return pad_direction instead of direction.
2113 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
2114 instead of direction.
2115 (nios2_function_arg_padding): Likewise. Make static.
2116 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2117 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
2118 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
2119 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
2120 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2121 (pa_function_arg_padding): Make static. Return pad_direction instead
2123 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
2124 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2125 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
2126 instead of direction. Use targetm.calls.function_arg_padding.
2127 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
2128 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2129 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
2130 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
2131 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
2133 (function_arg_padding): Rename to...
2134 (rs6000_function_arg_padding): ...this. Make static. Return
2135 pad_direction instead of direction.
2136 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2137 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
2138 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2139 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
2140 instead of direction. Use targetm.calls.function_arg_padding.
2141 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
2142 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2143 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
2144 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
2145 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2146 (function_arg_padding): Rename to...
2147 (rs6000_function_arg_padding): ...this. Make static. Return
2148 pad_direction instead of direction.
2149 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2150 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
2151 * config/s390/s390.c (s390_function_arg_padding): New function.
2152 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2153 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
2154 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
2155 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2156 (function_arg_padding): Rename to...
2157 (sparc_function_arg_padding): ...this. Make static. Return
2158 pad_direction instead of direction.
2159 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
2160 * config/spu/spu.c (spu_function_arg_padding): New function.
2161 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2162 * system.h (FUNCTION_ARG_PADDING): Poison.
2164 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2165 Alan Hayward <alan.hayward@arm.com>
2166 David Sherwood <david.sherwood@arm.com>
2168 * target.def (modes_tieable_p): New hook.
2169 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
2170 (TARGET_MODES_TIEABLE_P): ...this.
2171 * doc/tm.texi.in: Regenerate.
2172 * hooks.h (hook_bool_mode_mode_true): Declare.
2173 * hooks.c (hook_bool_mode_mode_true): New function.
2174 * combine.c (subst): Use targetm.modes_tieable_p instead of
2176 * dse.c (find_shift_sequence): Likewise.
2177 * expmed.c (extract_low_bits): Likewise.
2178 * lower-subreg.c: Include target.h.
2179 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
2181 * rtlanal.c (rtx_cost): Likewise.
2182 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
2183 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
2184 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
2185 (TARGET_MODES_TIEABLE_P): Redefine.
2186 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
2187 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
2188 (TARGET_MODES_TIEABLE_P): Redefine.
2189 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
2190 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
2191 (arc_modes_tieable_p): New function.
2192 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
2193 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
2194 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
2195 (arm_modes_tieable_p): Make static.
2196 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
2197 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
2198 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
2199 (TARGET_MODES_TIEABLE_P): Redefine.
2200 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
2201 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
2202 (TARGET_MODES_TIEABLE_P): Redefine.
2203 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
2204 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
2205 (cr16_modes_tieable_p): New function.
2206 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
2207 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
2208 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
2209 (TRULY_NOOP_TRUNCATION): Update comment.
2210 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
2211 (TRULY_NOOP_TRUNCATION): Update comment.
2212 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
2213 (frv_modes_tieable_p): New function.
2214 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
2215 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
2216 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
2217 (TARGET_MODES_TIEABLE_P): Redefine.
2218 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
2219 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
2220 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
2221 (TARGET_MODES_TIEABLE_P): Redefine.
2222 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
2223 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
2224 (ia64_modes_tieable_p): New function.
2225 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
2226 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
2227 (iq2000_modes_tieable_p): New function.
2228 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
2229 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
2230 (lm32_modes_tieable_p): New function.
2231 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
2232 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
2233 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
2234 (TARGET_MODES_TIEABLE_P): Redefine.
2235 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
2236 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
2237 (m32r_modes_tieable_p): New function.
2238 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
2239 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
2240 (m68k_modes_tieable_p): New function.
2241 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
2242 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
2243 (mcore_modes_tieable_p): New function.
2244 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
2245 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
2247 (TARGET_MODES_TIEABLE_P): Redefine.
2248 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
2249 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
2250 * config/mips/mips.c (mips_modes_tieable_p): Make static.
2251 (TARGET_MODES_TIEABLE_P): Redefine.
2252 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
2253 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
2254 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
2255 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
2256 (mn10300_modes_tieable_p): ...this and make static.
2257 (TARGET_MODES_TIEABLE_P): Redefine.
2258 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
2259 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
2260 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
2261 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
2262 (msp430_modes_tieable_p): Make static.
2263 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
2264 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
2265 (TARGET_MODES_TIEABLE_P): Redefine.
2266 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
2267 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
2268 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
2269 (TARGET_MODES_TIEABLE_P): Redefine.
2270 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
2271 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
2272 * config/pa/pa.c (pa_modes_tieable_p): Make static.
2273 (TARGET_MODES_TIEABLE_P): Redefine.
2274 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
2275 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
2276 (pdp11_modes_tieable_p): New function.
2277 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
2278 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
2279 (rs6000_modes_tieable_p): New function.
2280 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2281 * config/powerpcspe/powerpcspe.md: Update comment.
2282 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
2283 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
2284 (TARGET_MODES_TIEABLE_P): Redefine.
2285 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
2286 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
2287 (rl78_modes_tieable_p): New function.
2288 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
2289 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
2290 (rs6000_modes_tieable_p): New function.
2291 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2292 * config/rs6000/rs6000.md: Update comment.
2293 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
2294 * config/rx/rx.c (rx_modes_tieable_p): New function.
2295 (TARGET_MODES_TIEABLE_P): Redefine.
2296 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
2297 * config/s390/s390.c (s390_modes_tieable_p): New function.
2298 (TARGET_MODES_TIEABLE_P): Redefine.
2299 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
2300 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
2301 (sh_modes_tieable_p): New function.
2302 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
2303 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
2304 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
2305 (sparc_modes_tieable_p): Make static.
2306 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
2307 * config/spu/spu.c (spu_modes_tieable_p): New function.
2308 (TARGET_MODES_TIEABLE_P): Redefine.
2309 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
2310 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
2311 (TARGET_MODES_TIEABLE_P): Redefine.
2312 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
2313 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
2314 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
2315 * config/v850/v850.c (v850_modes_tieable_p): New function.
2316 (TARGET_MODES_TIEABLE_P): Redefine.
2317 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
2318 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
2319 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
2320 (visium_modes_tieable_p): New function.
2321 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
2322 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
2323 (xtensa_modes_tieable_p): New function.
2324 * system.h (MODES_TIEABLE_P): Poison.
2326 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2327 Alan Hayward <alan.hayward@arm.com>
2328 David Sherwood <david.sherwood@arm.com>
2330 * target.def (hard_regno_mode_ok): New hook.
2331 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
2332 (TARGET_HARD_REGNO_MODE_OK): ...this.
2333 * doc/tm.texi.in: Regenerate.
2334 * hooks.h (hook_bool_uint_mode_true): Declare.
2335 * hooks.c (hook_bool_uint_mode_true): New function.
2336 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
2338 * genpreds.c (write_insn_preds_c): Add an include of target.h.
2339 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
2340 instead of HARD_REGNO_MODE_OK.
2341 * caller-save.c: Include target.h.
2342 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
2344 * combine.c (can_combine_p): Likewise.
2345 (combinable_i3pat): Likewise.
2346 (can_change_dest_mode): Likewise.
2347 * expr.c (init_expr_target): Likewise.
2348 (convert_move): Likewise.
2349 (convert_modes): Likewise.
2350 * ira.c (setup_prohibited_class_mode_regs): Likewise.
2351 (setup_prohibited_mode_move_regs): Likewise.
2352 * ira.h (target_ira): Likewise.
2353 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2354 * lra-constraints.c (process_alt_operands): Likewise.
2355 (split_reg): Likewise.
2356 * recog.c (peep2_find_free_register): Likewise.
2357 * ree.c (combine_reaching_defs): Likewise.
2358 * regcprop.c (maybe_mode_change): Likewise.
2359 * reginfo.c (init_reg_sets_1): Likewise.
2360 (choose_hard_reg_mode): Likewise.
2361 (simplifiable_subregs): Likewise.
2362 * regrename.c (check_new_reg_p): Likewise.
2363 * reload.c (find_valid_class): Likewise.
2364 (find_valid_class_1): Likewise.
2365 (reload_inner_reg_of_subreg): Likewise.
2366 (push_reload): Likewise.
2367 (combine_reloads): Likewise.
2368 (find_dummy_reload): Likewise.
2369 (find_reloads): Likewise.
2370 * reload1.c (find_reg): Likewise.
2371 (set_reload_reg): Likewise.
2372 (allocate_reload_reg): Likewise.
2373 (choose_reload_regs): Likewise.
2374 (reload_adjust_reg_for_temp): Likewise.
2375 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
2376 (simplify_subreg_regno): Likewise.
2377 * sel-sched.c (init_regs_for_mode): Likewise.
2378 * varasm.c (make_decl_rtl): Likewise.
2379 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
2380 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
2382 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
2383 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
2384 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2385 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
2386 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
2387 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2388 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
2389 (arc_mode_class): Delete.
2390 (HARD_REGNO_MODE_OK): Delete.
2391 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2392 (arc_hard_regno_mode_ok): Rename old array to...
2393 (arc_hard_regno_mode_ok_modes): ...this.
2394 (arc_conditional_register_usage): Update accordingly.
2395 (arc_mode_class): Make static.
2396 (arc_hard_regno_mode_ok): New function.
2397 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2398 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2399 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2400 (arm_hard_regno_mode_ok): Make static.
2401 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2403 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2404 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2405 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2407 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2408 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2409 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2410 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2411 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2412 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2413 * config/bfin/predicates.md (valid_reg_operand): Use
2414 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2415 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2416 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2417 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2418 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2419 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2420 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2421 (cr16_hard_regno_mode_ok): Make static and return a bool.
2422 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2423 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2424 (cris_hard_regno_mode_ok): New function.
2425 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2426 (epiphany_mode_class): Delete.
2427 (HARD_REGNO_MODE_OK): Delete.
2428 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2429 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2430 (hard_regno_mode_ok): Rename to...
2431 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2432 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2433 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2435 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2436 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2437 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2438 (frv_hard_regno_mode_ok): Make static and return a bool.
2439 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2441 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2442 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2443 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2444 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2446 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2447 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2448 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2449 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2451 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2452 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2453 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2454 (ia64_hard_regno_mode_ok): New function.
2455 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2456 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2457 (iq2000_hard_regno_mode_ok): New function.
2458 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2459 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2460 (lm32_hard_regno_mode_ok): New function.
2461 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2462 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2463 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2464 instead of HARD_REGNO_MODE_OK.
2465 (m32c_hard_regno_ok): Rename to...
2466 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2467 (m32c_cannot_change_mode_class): Update accordingly.
2468 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2469 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2470 (m32r_mode_class): Delete.
2471 (HARD_REGNO_MODE_OK): Delete.
2472 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2473 (m32r_hard_regno_mode_ok): Rename to...
2474 (m32r_hard_regno_modes): ...this.
2475 (m32r_mode_class): Make static.
2476 (m32r_hard_regno_mode_ok): New function.
2477 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2478 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2479 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2480 (m68k_hard_regno_mode_ok): Make static.
2481 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2482 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2483 (mcore_hard_regno_mode_ok): New function.
2484 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2485 (HARD_REGNO_MODE_OK): Delete.
2486 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2488 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2489 (microblaze_hard_regno_mode_ok): New function.
2490 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2491 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2492 (mips_hard_regno_mode_ok): Delete.
2493 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2494 (mips_hard_regno_mode_ok_p): ...this and make static.
2495 (mips_hard_regno_mode_ok_p): Rename to...
2496 (mips_hard_regno_mode_ok_uncached): ...this.
2497 (mips_hard_regno_mode_ok): New function.
2498 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2499 of HARD_REGNO_MODE_OK.
2500 (mips_option_override): Update after above name changes.
2501 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2502 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2503 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2504 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2505 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2506 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2507 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2508 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2509 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2510 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2511 (msp430_hard_regno_mode_ok): Make static and return a bool.
2512 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2513 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2514 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2516 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2517 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2518 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2519 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2520 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2521 (PA_HARD_REGNO_MODE_OK): ...this
2522 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2523 (PA_HARD_REGNO_MODE_OK): ...this.
2524 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2525 (pa_hard_regno_mode_ok): New function.
2526 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2527 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2528 (pdp11_hard_regno_mode_ok): New function.
2529 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2530 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2532 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2534 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2535 (rs6000_hard_regno_mode_ok): Rename to...
2536 (rs6000_hard_regno_mode_ok_uncached): ...this.
2537 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2538 (rs6000_hard_regno_mode_ok): New function.
2539 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2540 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2541 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2542 (riscv_hard_regno_mode_ok): ...this and make static.
2543 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2544 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2545 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2546 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2547 (rl78_hard_regno_mode_ok): Make static and return bool.
2548 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2549 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2551 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2552 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2553 (rs6000_hard_regno_mode_ok): Rename to...
2554 (rs6000_hard_regno_mode_ok_uncached): ...this.
2555 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2556 (rs6000_hard_regno_mode_ok): New function.
2557 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2558 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2559 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2560 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2561 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2562 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2563 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2564 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2565 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2566 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2567 (sh_hard_regno_mode_ok): Make static.
2568 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2569 instead of HARD_REGNO_MODE_OK.
2570 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2571 (sparc_mode_class): Delete.
2572 (HARD_REGNO_MODE_OK): Delete.
2573 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2574 (hard_regno_mode_classes): Make static.
2575 (sparc_mode_class): Likewise.
2576 (sparc_hard_regno_mode_ok): New function.
2577 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2578 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2579 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2581 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2582 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2583 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2584 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2585 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2586 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2587 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2588 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2589 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2590 (visium_hard_regno_mode_ok): New function.
2591 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2592 instead of HARD_REGNO_MODE_OK.
2593 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2594 (HARD_REGNO_MODE_OK): Delete.
2595 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2596 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2597 (xtensa_option_override): Update accordingly.
2598 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2599 (xtensa_hard_regno_mode_ok): New function.
2600 * system.h (HARD_REGNO_MODE_OK): Poison.
2602 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2603 Alan Hayward <alan.hayward@arm.com>
2604 David Sherwood <david.sherwood@arm.com>
2606 * target.def (hard_regno_call_part_clobbered): New hook.
2607 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2608 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2609 * doc/tm.texi: Regenerate.
2610 * hooks.h (hook_bool_uint_mode_false): Declare.
2611 * hooks.c (hook_bool_uint_mode_false): New function.
2612 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2613 * cselib.c (cselib_process_insn): Use
2614 targetm.hard_regno_call_part_clobbered instead of
2615 HARD_REGNO_CALL_PART_CLOBBERED.
2616 * ira-conflicts.c (ira_build_conflicts): Likewise.
2617 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2618 * lra-constraints.c (need_for_call_save_p): Likewise.
2619 * lra-lives.c: Include target.h.
2620 (check_pseudos_live_through_calls): Use
2621 targetm.hard_regno_call_part_clobbered instead of
2622 HARD_REGNO_CALL_PART_CLOBBERED.
2623 * regcprop.c: Include target.h.
2624 (copyprop_hardreg_forward_1): Use
2625 targetm.hard_regno_call_part_clobbered instead of
2626 HARD_REGNO_CALL_PART_CLOBBERED.
2627 * reginfo.c (choose_hard_reg_mode): Likewise.
2628 * regrename.c (check_new_reg_p): Likewise.
2629 * reload.c (find_equiv_reg): Likewise.
2630 * reload1.c (emit_reload_insns): Likewise.
2631 * sched-deps.c (deps_analyze_insn): Likewise.
2632 * sel-sched.c (init_regs_for_mode): Likewise.
2633 (mark_unavailable_hard_regs): Likewise.
2634 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2635 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2636 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
2638 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2639 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2640 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
2642 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
2644 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2645 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2646 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
2648 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2649 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2650 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2652 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2653 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2655 * config/powerpcspe/powerpcspe.c
2656 (rs6000_hard_regno_call_part_clobbered): New function.
2657 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2658 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2659 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2661 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2662 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2663 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2665 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2666 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2667 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2669 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2670 Alan Hayward <alan.hayward@arm.com>
2671 David Sherwood <david.sherwood@arm.com>
2673 * rtl.h (subreg_memory_offset): Declare.
2674 * emit-rtl.c (subreg_memory_offset): New function.
2675 * expmed.c (store_bit_field_1): Use it.
2676 * expr.c (undefined_operand_subword_p): Likewise.
2677 * simplify-rtx.c (simplify_subreg): Likewise.
2679 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2681 PR rtl-optimization/57448
2684 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2685 using atomic_load pattern.
2686 (expand_atomic_store): Likewise.
2688 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2691 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2692 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2693 handling. Use replace_call_with_value with NULL instead of
2694 gsi_replace, unlink_stmt_vdef and release_defs.
2696 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2699 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2701 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2704 * rtl.h (emit_library_call_value_1): Declare.
2705 (emit_library_call): Replace declaration with a series of overloads.
2706 Remove the parameter count argument.
2707 (emit_library_call_value): Likewise.
2708 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2709 with an "rtx_mode_t *".
2710 (emit_library_call_value): Delete.
2711 (emit_library_call): Likewise.
2712 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2713 (asan_emit_allocas_unpoison): Likewise.
2714 * builtins.c (expand_builtin_powi): Likewise.
2715 (expand_asan_emit_allocas_unpoison): Likewise.
2716 * cfgexpand.c (expand_main_function): Likewise.
2717 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2718 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2719 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2720 * config/arm/arm.c (arm_trampoline_init): Likewise.
2721 (arm_call_tls_get_addr): Likewise.
2722 (arm_expand_divmod_libfunc): Likewise.
2723 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2724 (smulsi3_highpart): Likewise.
2725 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2726 (c6x_expand_compare): Likewise.
2727 (c6x_expand_movmem): Likewise.
2728 * config/frv/frv.c (frv_trampoline_init): Likewise.
2729 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2730 (ix86_expand_divmod_libfunc): Likewise.
2731 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2732 (ia64_expand_compare): Likewise.
2733 (ia64_profile_hook): Likewise.
2734 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2735 (nonlocal_goto): Likewise.
2736 (restore_stack_nonlocal): Likewise.
2737 * config/m32r/m32r.c (block_move_call): Likewise.
2738 (m32r_trampoline_init): Likewise.
2739 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
2740 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
2741 (m68k_call_m68k_read_tp): Likewise.
2742 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
2743 (microblaze_expand_divide): Likewise.
2744 * config/mips/mips.h (mips_args): Likewise.
2745 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
2746 (MIPS_ICACHE_SYNC): Likewise.
2747 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
2748 (nios2_trampoline_init): Likewise.
2749 * config/pa/pa.c (hppa_tls_call): Likewise.
2750 (pa_trampoline_init): Likewise.
2751 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
2752 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
2753 (expand_strn_compare): Likewise.
2754 (rs6000_generate_compare): Likewise.
2755 (rs6000_expand_float128_convert): Likewise.
2756 (output_profile_hook): Likewise.
2757 (rs6000_trampoline_init): Likewise.
2758 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
2759 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
2760 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
2761 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
2762 (rs6000_generate_compare): Likewise.
2763 (rs6000_expand_float128_convert): Likewise.
2764 (output_profile_hook): Likewise.
2765 (rs6000_trampoline_init): Likewise.
2766 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
2767 * config/sh/sh.c (sh_trampoline_init): Likewise.
2768 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
2769 (sparc_emit_float_lib_cmp): Likewise.
2770 (sparc32_initialize_trampoline): Likewise.
2771 (sparc64_initialize_trampoline): Likewise.
2772 (sparc_profile_hook): Likewise.
2773 * config/spu/spu.c (ea_load_store): Likewise.
2774 * config/spu/spu.md (floatunssidf2): Likewise.
2775 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
2776 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
2777 * config/visium/visium.c (expand_block_move_4): Likewise.
2778 (expand_block_move_2): Likewise.
2779 (expand_block_move_1): Likewise.
2780 (expand_block_set_4): Likewise.
2781 (expand_block_set_2): Likewise.
2782 (expand_block_set_1): Likewise.
2783 (visium_trampoline_init): Likewise.
2784 (visium_profile_hook): Likewise.
2785 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
2786 (xtensa_setup_frame_addresses): Likewise.
2787 (xtensa_trampoline_init): Likewise.
2788 * except.c (sjlj_emit_function_enter): Likewise.
2789 (sjlj_emit_function_exit): Likewise.
2790 * explow.c (allocate_dynamic_stack_space): Likewise.
2791 (probe_stack_range): Likewise.
2792 * expr.c (convert_mode_scalar): Likewise.
2793 * optabs.c (expand_binop): Likewise.
2794 (expand_twoval_binop_libfunc): Likewise.
2795 (expand_unop): Likewise.
2796 (prepare_cmp_insn): Likewise.
2797 (prepare_float_lib_cmp): Likewise.
2798 (expand_float): Likewise.
2799 (expand_fix): Likewise.
2800 (expand_fixed_convert): Likewise.
2801 (maybe_emit_sync_lock_test_and_set): Likewise.
2802 (expand_atomic_compare_and_swap): Likewise.
2803 (expand_mem_thread_fence): Likewise.
2804 (expand_atomic_fetch_op): Likewise.
2806 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
2808 * doc/generic.texi (OpenACC): Adjust URL.
2809 * doc/invoke.texi (C Dialect Options): Ditto.
2811 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
2813 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
2814 predicate for operand 1. Add (m,<S>) constraint.
2815 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
2816 Prevent memory operand 1 with register operand 2.
2818 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
2820 PR rtl-optimization/82024
2821 * combine.c (try_combine): If the combination result is a PARALLEL,
2822 and we only need to retain the SET in there that would be placed
2823 at I2, check that we can place that at I3 instead, before doing so.
2825 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2828 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
2830 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
2831 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
2834 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
2835 Jeff Law <law@redhat.com>
2837 * varasm.c (bss_initializer_p): Do not put constants into .bss
2838 (categorize_decl_for_section): Handle bss_initializer_p returning
2839 false when DECL_INITIAL is NULL.
2841 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2844 * config/s390/s390.c (s390_can_inline_p): New function.
2846 2017-09-01 Jeff Law <law@redhat.com>
2848 PR tree-optimization/82052
2849 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
2850 Always initialize the returned slot after a hash table miss
2851 when INSERT is true.
2853 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
2855 * config/s390/s390.md (mem_signal_fence): Remove.
2856 * doc/md.texi (mem_signal_fence): Remove.
2857 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
2859 * target-insns.def (mem_signal_fence): Remove.
2861 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2864 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
2867 * asan.c (create_odr_indicator): Strip name encoding from assembler
2868 name before appending it after __odr_asan_.
2870 2017-09-01 Martin Liska <mliska@suse.cz>
2872 PR tree-optimization/82059
2873 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
2874 frequency only when an edge is redirected.
2876 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2878 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
2879 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
2880 (arc_conditional_register_usage): Remove ARC600 lp_count
2882 (arc_file_start): Emit Tag_ARC_CPU_variation.
2883 (arc_can_use_doloop_p): New conditions to use ZOLs.
2884 (hwloop_fail): New function.
2885 (hwloop_optimize): Likewise.
2886 (hwloop_pattern_reg): Likewise.
2887 (arc_doloop_hooks): New struct, to be used with reorg_loops.
2888 (arc_reorg_loops): New function, calls reorg_loops.
2889 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
2890 (arc600_corereg_hazard): Remove ZOL checking, case handled by
2892 (arc_loop_hazard): Remove function, functionality moved into
2894 (arc_hazard): Remove arc_loop_hazard call.
2895 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
2896 into hwloop_optimize.
2897 (arc_label_align): Remove ZOL handling.
2898 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
2899 * config/arc/arc.md (doloop_begin): Remove pattern.
2900 (doloop_begin_i): Likewise.
2901 (doloop_end_i): Likewise.
2902 (doloop_fallback): Likewise.
2903 (doloop_fallback_m): Likewise.
2904 (doloop_end): Reimplement expand.
2905 (arc_lp): New pattern for LP instruction.
2906 (loop_end): New pattern.
2907 (loop_fail): Likewise.
2908 (decrement_and_branch_until_zero): Likewise.
2909 * config/arc/arc.opt (mlpc-width): New option.
2910 * doc/invoke.texi (mlpc-width): Document option.
2912 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2914 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
2915 (arc_ccfsm_advance): Fix checking for delay slots.
2916 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
2918 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2920 * config/arc/arc.md (movqi_insn): Add stores to save constant long
2922 (movhi_insn): Update store instruction constraint which are saving
2923 6-bit short immediates.
2924 (movsi_insn): Consider also short scaled load operations.
2925 (zero_extendhisi2_i): Use Usd constraint instead of T.
2926 (extendhisi2_i): Add q constraint.
2927 (arc_clzsi2): Add type and length attributes.
2928 (arc_ctzsi2): Likewise.
2929 * config/arc/constraints.md (Usc): Update constraint, the
2930 assembler can parse two relocations for a single instruction.
2932 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2934 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
2935 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
2937 2017-08-31 Olivier Hainque <hainque@adacore.com>
2939 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
2940 match as powerpc-wrs-vxworks*.
2942 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
2944 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
2945 register constraint for by-element operand.
2946 (aarch64_mls_elt_merge<mode>): Likewise.
2948 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2950 * config/arc/arc.c (arc_can_follow_jump): Check for short
2953 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2955 * config.gcc: Use g.opt for arc.
2956 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
2957 functionality moved to ...
2958 (legitimate_scaled_address_p): New function, ...here.
2959 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
2960 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
2961 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
2963 (arc_override_options): Handle G option.
2964 (arc_output_pic_addr_const): Correct function definition.
2965 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
2966 (arc_decl_anon_ns_mem_p): Delete.
2967 (arc_in_small_data_p): Overhaul this function to take into
2968 consideration the value given via G option.
2969 (arc_rewrite_small_data_1): Renamed and corrected old
2970 arc_rewrite_small_data function.
2971 (arc_rewrite_small_data): New function.
2972 (small_data_pattern): Don't use pic_offset_table_rtx.
2973 * config/arc/arc.h (CC1_SPEC): Recognize G option.
2974 * config/arc/simdext.md (movmisalignv2hi): Use
2975 prepare_move_operands function.
2977 (movmisalign*): Likewise.
2978 * doc/invoke.texi (ARC options): Document -G option.
2980 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2982 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2984 * config/arc/arc.c (arc_print_operand): Output scalled address for
2985 sdata whenever is possible.
2986 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2987 load/stores are available.
2988 (compact_sda_memory_operand): Check for the alignment required by
2989 code density instructions.
2990 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2992 * config/arc/constraints.md (Usd): Update constraint.
2993 (Us0): New constraint.
2994 (Usc): Update constraint.
2996 2017-08-31 Richard Biener <rguenther@suse.de>
2999 * dwarf2out.c (dwarf2out_early_global_decl): Process each
3002 2017-08-31 Tamar Christina <tamar.christina@arm.com>
3004 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
3005 Resize type_signature.
3007 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3008 Alan Hayward <alan.hayward@arm.com>
3009 David Sherwood <david.sherwood@arm.com>
3011 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
3012 subregs whose inner modes can be stored in GPRs.
3013 (aarch64_classify_index): Likewise.
3015 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3016 Alan Hayward <alan.hayward@arm.com>
3017 David Sherwood <david.sherwood@arm.com>
3019 * config/aarch64/iterators.md (V_cmp_result): Rename to...
3020 (V_INT_EQUIV): ...this.
3021 (v_cmp_result): Rename to...
3022 (v_int_equiv): ...this.
3023 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
3024 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
3025 (copysign<mode>3): Likewise.
3026 (aarch64_simd_bsl<mode>_internal): Likewise.
3027 (aarch64_simd_bsl<mode>): Likewise.
3028 (vec_cmp<mode><mode>): Likewise.
3029 (vcond<mode><mode>): Likewise.
3030 (vcond<v_cmp_mixed><mode>): Likewise.
3031 (vcondu<mode><v_cmp_mixed>): Likewise.
3032 (aarch64_cm<optab><mode>): Likewise.
3033 (aarch64_cmtst<mode>): Likewise.
3034 (aarch64_fac<optab><mode>): Likewise.
3035 (vec_perm_const<mode>): Likewise.
3036 (vcond_mask_<mode><v_cmp_result>): Rename to...
3037 (vcond_mask_<mode><v_int_equiv>): ...this.
3038 (vec_cmp<mode><v_cmp_result>): Rename to...
3039 (vec_cmp<mode><v_int_equiv>): ...this.
3041 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
3042 Alan Hayward <alan.hayward@arm.com>
3043 David Sherwood <david.sherwood@arm.com>
3045 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
3047 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
3048 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
3049 (UNSPEC_LD4_DREG): New unspecs.
3050 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
3051 (aarch64_ld2<mode>_dreg_be): Replace with...
3052 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
3054 (aarch64_ld3<mode>_dreg_le)
3055 (aarch64_ld3<mode>_dreg_be): Replace with...
3056 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
3058 (aarch64_ld4<mode>_dreg_le)
3059 (aarch64_ld4<mode>_dreg_be): Replace with...
3060 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
3063 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3065 PR tree-optimization/81987
3066 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
3067 insert an initializer in a location not dominated by the stride
3070 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3072 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
3073 on the entire header of the finally block in the fallthru case.
3075 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3077 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
3079 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
3081 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
3082 rs6000_emit_move_from_cr and call renamed function.
3083 (rs6000_emit_prologue): Call renamed functions.
3084 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
3085 movesi_from_cr, remove volatile CRs.
3087 2017-08-30 Jon Beniston <jon@beniston.com>
3088 Richard Biener <rguenther@suse.de>
3090 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
3091 of VECTOR_MODE_P check.
3092 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
3093 element vector types.
3095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3097 * df.h (df_read_modify_subreg_p): Remove in favor of...
3098 * rtl.h (read_modify_subreg_p): ...this new function. Take a
3099 const_rtx instead of an rtx.
3100 * cprop.c (local_cprop_find_used_regs): Update accordingly.
3101 * df-problems.c (df_word_lr_mark_ref): Likewise.
3102 * ira-lives.c (mark_pseudo_reg_live): Likewise.
3103 (mark_pseudo_reg_dead): Likewise.
3104 (mark_ref_dead): Likewise.
3105 * reginfo.c (init_subregs_of_mode): Likewise.
3106 * sched-deps.c (sched_analyze_1): Likewise.
3107 * df-scan.c (df_def_record_1): Likewise.
3108 (df_uses_record): Likewise.
3109 (df_read_modify_subreg_p): Remove in favor of...
3110 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
3111 const_rtx instead of an rtx.
3113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3114 Alan Hayward <alan.hayward@arm.com>
3115 David Sherwood <david.sherwood@arm.com>
3117 * rtl.h (partial_subreg_p): New function.
3118 * caller-save.c (save_call_clobbered_regs): Use it.
3119 * calls.c (expand_call): Likewise.
3120 * combine.c (combinable_i3pat): Likewise.
3121 (simplify_set): Likewise.
3122 (make_extraction): Likewise.
3123 (make_compound_operation_int): Likewise.
3124 (gen_lowpart_or_truncate): Likewise.
3125 (force_to_mode): Likewise.
3126 (make_field_assignment): Likewise.
3127 (reg_truncated_to_mode): Likewise.
3128 (record_truncated_value): Likewise.
3129 (move_deaths): Likewise.
3130 * cse.c (record_jump_cond): Likewise.
3131 (cse_insn): Likewise.
3132 * cselib.c (cselib_lookup_1): Likewise.
3133 * expmed.c (extract_bit_field_using_extv): Likewise.
3134 * function.c (assign_parm_setup_reg): Likewise.
3135 * ifcvt.c (noce_convert_multiple_sets): Likewise.
3136 * ira-build.c (create_insn_allocnos): Likewise.
3137 * lra-coalesce.c (merge_pseudos): Likewise.
3138 * lra-constraints.c (match_reload): Likewise.
3139 (simplify_operand_subreg): Likewise.
3140 (curr_insn_transform): Likewise.
3141 * lra-lives.c (process_bb_lives): Likewise.
3142 * lra.c (new_insn_reg): Likewise.
3143 (lra_substitute_pseudo): Likewise.
3144 * regcprop.c (mode_change_ok): Likewise.
3145 (maybe_mode_change): Likewise.
3146 (copyprop_hardreg_forward_1): Likewise.
3147 * reload.c (push_reload): Likewise.
3148 (find_reloads): Likewise.
3149 (find_reloads_subreg_address): Likewise.
3150 * reload1.c (alter_reg): Likewise.
3151 (eliminate_regs_1): Likewise.
3152 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3154 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
3156 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
3157 back to if statements, including unpack.
3159 2017-08-30 Martin Liska <mliska@suse.cz>
3162 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
3164 (func_checker::compare_asm_inputs_outputs): ... this function.
3165 (func_checker::compare_gimple_asm): Use the function to compare
3166 also ASM constrains.
3167 * ipa-icf-gimple.h: Rename the function.
3169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3170 Alan Hayward <alan.hayward@arm.com>
3171 David Sherwood <david.sherwood@arm.com>
3173 * coretypes.h (complex_mode): New type.
3174 * gdbhooks.py (build_pretty_printer): Handle it.
3175 * machmode.h (complex_mode): New class.
3176 (complex_mode::includes_p): New function.
3177 (is_complex_int_mode): Likewise.
3178 (is_complex_float_mode): Likewise.
3179 * genmodes.c (get_mode_class): Handle complex mode classes.
3180 * function.c (expand_function_end): Use is_complex_int_mode.
3182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3183 Alan Hayward <alan.hayward@arm.com>
3184 David Sherwood <david.sherwood@arm.com>
3186 * coretypes.h (scalar_mode_pod): New typedef.
3187 * gdbhooks.py (build_pretty_printer): Handle it.
3188 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
3189 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
3190 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
3191 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
3194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3195 Alan Hayward <alan.hayward@arm.com>
3196 David Sherwood <david.sherwood@arm.com>
3198 * machmode.h (mode_for_vector): Take a scalar_mode instead
3200 * stor-layout.c (mode_for_vector): Likewise.
3201 * explow.c (promote_mode): Use as_a <scalar_mode>.
3202 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
3204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3205 Alan Hayward <alan.hayward@arm.com>
3206 David Sherwood <david.sherwood@arm.com>
3208 * target.def (preferred_simd_mode): Take a scalar_mode
3209 instead of a machine_mode.
3210 * targhooks.h (default_preferred_simd_mode): Likewise.
3211 * targhooks.c (default_preferred_simd_mode): Likewise.
3212 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
3213 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
3214 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
3215 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
3216 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
3217 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
3218 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
3219 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
3221 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
3222 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
3223 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
3224 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
3225 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
3226 * doc/tm.texi: Regenerate.
3227 * optabs-query.c (can_vec_mask_load_store_p): Return false for
3230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3231 Alan Hayward <alan.hayward@arm.com>
3232 David Sherwood <david.sherwood@arm.com>
3234 * target.def (scalar_mode_supported_p): Take a scalar_mode
3235 instead of a machine_mode.
3236 * targhooks.h (default_scalar_mode_supported_p): Likewise.
3237 * targhooks.c (default_scalar_mode_supported_p): Likewise.
3238 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
3239 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3240 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
3241 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
3242 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
3243 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
3244 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
3245 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
3246 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
3247 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
3248 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
3249 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
3251 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
3252 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
3253 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3254 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3255 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
3257 * doc/tm.texi: Regenerate.
3259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3260 Alan Hayward <alan.hayward@arm.com>
3261 David Sherwood <david.sherwood@arm.com>
3263 * coretypes.h (opt_scalar_mode): New typedef.
3264 * gdbhooks.py (build_pretty_printers): Handle it.
3265 * machmode.h (mode_iterator::get_2xwider): Add overload for
3267 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
3269 * expr.c (convert_mode_scalar): Likewise.
3270 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3271 * optabs.c (expand_float): Likewise.
3272 (expand_fix): Likewise.
3273 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3276 Alan Hayward <alan.hayward@arm.com>
3277 David Sherwood <david.sherwood@arm.com>
3279 * optabs.c (expand_float): Explicitly check for scalars before
3280 using a branching expansion.
3281 (expand_fix): Likewise.
3283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3284 Alan Hayward <alan.hayward@arm.com>
3285 David Sherwood <david.sherwood@arm.com>
3287 * expr.c (convert_mode): Split scalar handling out into...
3288 (convert_mode_scalar): ...this new function. Treat the modes
3291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3292 Alan Hayward <alan.hayward@arm.com>
3293 David Sherwood <david.sherwood@arm.com>
3295 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
3297 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
3299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3300 Alan Hayward <alan.hayward@arm.com>
3301 David Sherwood <david.sherwood@arm.com>
3303 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
3304 rather than a machine_mode.
3305 (fixed_from_string): Likewise.
3306 (fixed_convert): Likewise.
3307 (fixed_convert_from_int): Likewise.
3308 (fixed_convert_from_real): Likewise.
3309 (real_convert_from_fixed): Likewise.
3310 * fixed-value.c (fixed_from_double_int): Likewise.
3311 (fixed_from_string): Likewise.
3312 (fixed_convert): Likewise.
3313 (fixed_convert_from_int): Likewise.
3314 (fixed_convert_from_real): Likewise.
3315 (real_convert_from_fixed): Likewise.
3316 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
3318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3319 Alan Hayward <alan.hayward@arm.com>
3320 David Sherwood <david.sherwood@arm.com>
3322 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
3323 of separate mode class checks. Do not allow vector modes here.
3324 (immed_wide_int_const): Use as_a <scalar_mode>.
3325 * explow.c (trunc_int_for_mode): Likewise.
3326 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
3327 (wi::shwi): Likewise.
3328 (wi::min_value): Likewise.
3329 (wi::max_value): Likewise.
3330 * dwarf2out.c (loc_descriptor): Likewise.
3331 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
3334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3335 Alan Hayward <alan.hayward@arm.com>
3336 David Sherwood <david.sherwood@arm.com>
3338 * tree.h (SCALAR_TYPE_MODE): New macro.
3339 * expr.c (expand_expr_addr_expr_1): Use it.
3340 (expand_expr_real_2): Likewise.
3341 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
3342 (fold_convert_const_fixed_from_int): Likewise.
3343 (fold_convert_const_fixed_from_real): Likewise.
3344 (native_encode_fixed): Likewise
3345 (native_encode_complex): Likewise
3346 (native_encode_vector): Likewise.
3347 (native_interpret_fixed): Likewise.
3348 (native_interpret_real): Likewise.
3349 (native_interpret_complex): Likewise.
3350 (native_interpret_vector): Likewise.
3351 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
3352 (simd_clone_adjust_argument_types): Likewise.
3353 (simd_clone_init_simd_arrays): Likewise.
3354 (simd_clone_adjust): Likewise.
3355 * stor-layout.c (layout_type): Likewise.
3356 * tree.c (build_minus_one_cst): Likewise.
3357 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3358 * tree-inline.c (estimate_move_cost): Likewise.
3359 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
3360 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
3361 (vectorizable_reduction): Likewise.
3362 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
3363 (vect_recog_mixed_size_cond_pattern): Likewise.
3364 (check_bool_pattern): Likewise.
3365 (adjust_bool_pattern): Likewise.
3366 (search_type_for_mask_1): Likewise.
3367 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3368 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3369 (vectorizable_load): Likewise.
3370 (vectorizable_store): Likewise.
3371 * ubsan.c (ubsan_encode_value): Likewise.
3372 * varasm.c (output_constant): Likewise.
3374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3375 Alan Hayward <alan.hayward@arm.com>
3376 David Sherwood <david.sherwood@arm.com>
3378 * coretypes.h (scalar_mode): New class.
3379 * machmode.h (scalar_mode): Likewise.
3380 (scalar_mode::includes_p): New function.
3381 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
3382 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
3383 * genmodes.c (get_mode_class): Handle remaining scalar modes.
3384 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
3385 * expmed.c (store_bit_field_1): Likewise.
3386 (extract_bit_field_1): Likewise.
3387 * expr.c (write_complex_part): Likewise.
3388 (read_complex_part): Likewise.
3389 (emit_move_complex_push): Likewise.
3390 (expand_expr_real_2): Likewise.
3391 * function.c (assign_parm_setup_reg): Likewise.
3392 (assign_parms_unsplit_complex): Likewise.
3393 * optabs.c (expand_binop): Likewise.
3394 * rtlanal.c (subreg_get_info): Likewise.
3395 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3396 * varasm.c (output_constant_pool_2): Likewise.
3398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3399 Alan Hayward <alan.hayward@arm.com>
3400 David Sherwood <david.sherwood@arm.com>
3402 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3404 (expmed_mult_highpart_optab): Likewise.
3405 (expmed_mult_highpart): Likewise.
3407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3408 Alan Hayward <alan.hayward@arm.com>
3409 David Sherwood <david.sherwood@arm.com>
3411 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3412 instead of a machine_mode.
3413 (builtin_memset_read_str): Likewise.
3414 * builtins.c (c_readstr): Likewise.
3415 (builtin_memcpy_read_str): Likewise.
3416 (builtin_strncpy_read_str): Likewise.
3417 (builtin_memset_read_str): Likewise.
3418 (builtin_memset_gen_str): Likewise.
3419 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3420 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3421 instead of a machine_mode.
3422 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3424 (make_extraction): Likewise.
3425 (try_widen_shift_mode): Take and return scalar_int_modes instead
3427 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3428 a scalar_int_mode instead of a machine_mode.
3429 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3430 (avr_addr_space_pointer_mode): Likewise.
3431 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3432 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3433 (msp430_unwind_word_mode): Likewise.
3434 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3435 (spu_addr_space_pointer_mode): Likewise.
3436 (spu_addr_space_address_mode): Likewise.
3437 (spu_libgcc_cmp_return_mode): Likewise.
3438 (spu_libgcc_shift_count_mode): Likewise.
3439 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3440 (rl78_addr_space_pointer_mode): Likewise.
3441 (fl78_unwind_word_mode): Likewise.
3442 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3444 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3445 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3446 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3447 (mips_valid_pointer_mode): Likewise.
3448 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3449 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3450 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3452 (ft32_addr_space_address_mode): Likewise.
3453 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3454 scalar_int_mode instead of a machine_mode.
3455 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3457 (m32c_addr_space_address_mode): Likewise.
3458 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3459 (rs6000_eh_return_filter_mode): Likewise.
3460 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3461 (rs6000_eh_return_filter_mode): Likewise.
3462 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3463 (s390_libgcc_shift_count_mode): Likewise.
3464 (s390_unwind_word_mode): Likewise.
3465 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3467 * target.def (mode_rep_extended): Likewise.
3468 (valid_pointer_mode): Likewise.
3469 (addr_space.valid_pointer_mode): Likewise.
3470 (eh_return_filter_mode): Return a scalar_int_mode rather than
3472 (libgcc_cmp_return_mode): Likewise.
3473 (libgcc_shift_count_mode): Likewise.
3474 (unwind_word_mode): Likewise.
3475 (addr_space.pointer_mode): Likewise.
3476 (addr_space.address_mode): Likewise.
3477 * doc/tm.texi: Regenerate.
3478 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3480 (do_jump): Use scalar_int_mode for local variables.
3481 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3482 rather than a machine_mode.
3483 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3484 (scompare_loc_descriptor_wide): Likewise.
3485 (scompare_loc_descriptor_narrow): Likewise.
3486 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3488 * except.c (sjlj_emit_dispatch_table): Likewise.
3489 (expand_builtin_eh_copy_values): Likewise.
3490 * explow.c (convert_memory_address_addr_space_1): Likewise.
3491 Take a scalar_int_mode rather than a machine_mode.
3492 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3493 than a machine_mode.
3494 (memory_address_addr_space): Use scalar_int_mode for local variables.
3495 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3496 rather than a machine_mode.
3497 * expmed.c (mask_rtx): Likewise.
3498 (init_expmed_one_conv): Likewise.
3499 (expand_mult_highpart_adjust): Likewise.
3500 (extract_high_half): Likewise.
3501 (expmed_mult_highpart_optab): Likewise.
3502 (expmed_mult_highpart): Likewise.
3503 (expand_smod_pow2): Likewise.
3504 (expand_sdiv_pow2): Likewise.
3505 (emit_store_flag_int): Likewise.
3506 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3508 (extract_low_bits): Likewise.
3509 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3511 * expr.c (pieces_addr::adjust): Likewise.
3512 (can_store_by_pieces): Likewise.
3513 (store_by_pieces): Likewise.
3514 (clear_by_pieces_1): Likewise.
3515 (expand_expr_addr_expr_1): Likewise.
3516 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3517 (expand_expr_real_1): Likewise.
3518 (try_casesi): Likewise.
3519 * final.c (shorten_branches): Likewise.
3520 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3521 type of "mode" to machine_mode.
3522 * internal-fn.c (expand_arith_overflow_result_store): Take a
3523 scalar_int_mode rather than a machine_mode.
3524 (expand_mul_overflow): Use scalar_int_mode for local variables.
3525 * loop-doloop.c (doloop_modify): Likewise.
3526 (doloop_optimize): Likewise.
3527 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3528 than a machine_mode.
3529 (expand_doubleword_shift_condmove): Likewise.
3530 (expand_doubleword_shift): Likewise.
3531 (expand_doubleword_clz): Likewise.
3532 (expand_doubleword_popcount): Likewise.
3533 (expand_doubleword_parity): Likewise.
3534 (expand_absneg_bit): Use scalar_int_mode for local variables.
3535 (prepare_float_lib_cmp): Likewise.
3536 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3537 rather than a machine_mode.
3538 (convert_memory_address_addr_space): Likewise.
3539 (get_mode_bounds): Likewise.
3540 (get_address_mode): Return a scalar_int_mode rather than a
3542 * rtlanal.c (get_address_mode): Likewise.
3543 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3544 than a machine_mode.
3545 * targhooks.c (default_mode_rep_extended): Likewise.
3546 (default_valid_pointer_mode): Likewise.
3547 (default_addr_space_valid_pointer_mode): Likewise.
3548 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3549 than a machine_mode.
3550 (default_libgcc_cmp_return_mode): Likewise.
3551 (default_libgcc_shift_count_mode): Likewise.
3552 (default_unwind_word_mode): Likewise.
3553 (default_addr_space_pointer_mode): Likewise.
3554 (default_addr_space_address_mode): Likewise.
3555 * targhooks.h (default_eh_return_filter_mode): Likewise.
3556 (default_libgcc_cmp_return_mode): Likewise.
3557 (default_libgcc_shift_count_mode): Likewise.
3558 (default_unwind_word_mode): Likewise.
3559 (default_addr_space_pointer_mode): Likewise.
3560 (default_addr_space_address_mode): Likewise.
3561 (default_mode_rep_extended): Take a scalar_int_mode rather than
3563 (default_valid_pointer_mode): Likewise.
3564 (default_addr_space_valid_pointer_mode): Likewise.
3565 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3567 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3568 rather than a machine_mode.
3569 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3570 for local variables.
3571 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3572 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3573 than a machine_mode.
3575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3576 Alan Hayward <alan.hayward@arm.com>
3577 David Sherwood <david.sherwood@arm.com>
3579 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3580 the mode argument to scalar_int_mode.
3581 (do_jump_by_parts_zero_rtx): Likewise.
3582 (do_jump_by_parts_equality_rtx): Likewise.
3583 (do_jump_by_parts_greater): Take a mode argument.
3584 (do_jump_by_parts_equality): Likewise.
3585 (do_jump_1): Update calls accordingly.
3587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3588 Alan Hayward <alan.hayward@arm.com>
3589 David Sherwood <david.sherwood@arm.com>
3591 * is-a.h (safe_dyn_cast): New function.
3592 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3593 (jump_table_for_label): Likewise.
3594 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3595 instead of an rtx_insn *.
3596 (shorten_branches): Use dyn_cast instead of LABEL_P and
3597 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3598 rtx_jump_table_data::get_data_mode.
3599 (final_scan_insn): Likewise.
3601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3602 Alan Hayward <alan.hayward@arm.com>
3603 David Sherwood <david.sherwood@arm.com>
3605 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3606 trying to combine a full-register integer set with a subreg
3609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3610 Alan Hayward <alan.hayward@arm.com>
3611 David Sherwood <david.sherwood@arm.com>
3613 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3614 that is always either address_mode or pointer_mode.
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_real_2): Use word_mode instead of innermode
3621 when the two are known to be equal.
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 * simplify-rtx.c (simplify_const_unary_operation): Use
3628 is_a <scalar_int_mode> instead of checking for a nonzero
3629 precision. Forcibly convert op_mode to a scalar_int_mode
3630 in that case. More clearly differentiate the operand and
3631 result modes and use the former when deciding what the value
3632 of a count-bits operation should be. Use is_int_mode instead
3633 of checking for a MODE_INT. Remove redundant check for whether
3634 this mode has a zero precision.
3636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3637 Alan Hayward <alan.hayward@arm.com>
3638 David Sherwood <david.sherwood@arm.com>
3640 * optabs.c (widen_leading): Change the type of the mode argument
3641 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
3642 (widen_bswap): Likewise.
3643 (expand_parity): Likewise.
3644 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
3645 (expand_ffs): Likewise.
3646 (epand_unop): Check for scalar integer modes before calling the
3649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3650 Alan Hayward <alan.hayward@arm.com>
3651 David Sherwood <david.sherwood@arm.com>
3653 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3655 (expand_expr_real_1): Update call accordingly.
3657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3658 Alan Hayward <alan.hayward@arm.com>
3659 David Sherwood <david.sherwood@arm.com>
3661 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3662 value_mode arguments. Use scalar_int_mode internally.
3663 (store_bit_field_1): Rename the new integer mode from imode
3664 to op0_mode and use it instead of GET_MODE (op0). Update calls
3665 to store_split_bit_field, store_bit_field_using_insv and
3666 store_fixed_bit_field.
3667 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3668 Use scalar_int_mode internally. Use a bit count rather than a mode
3669 when calculating the largest bit size for get_best_mode.
3670 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3671 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3672 Remove assertion that OP0 has a scalar integer mode.
3673 (store_split_bit_field): Add op0_mode and value_mode arguments.
3674 Update calls to extract_fixed_bit_field.
3675 (extract_bit_field_using_extv): Add an op0_mode argument.
3676 Use scalar_int_mode internally.
3677 (extract_bit_field_1): Rename the new integer mode from imode to
3678 op0_mode and use it instead of GET_MODE (op0). Update calls to
3679 extract_split_bit_field, extract_bit_field_using_extv and
3680 extract_fixed_bit_field.
3681 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3682 to extract_split_bit_field and extract_fixed_bit_field_1.
3683 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3684 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3686 (extract_split_bit_field): Add an op0_mode argument. Update call
3687 to extract_fixed_bit_field.
3689 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3690 Alan Hayward <alan.hayward@arm.com>
3691 David Sherwood <david.sherwood@arm.com>
3693 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3694 * explow.c (hard_function_value): Likewise.
3695 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3696 convert_to_mode call outside the loop.
3697 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3698 for the mode iterator. Require the mode specified by max_pieces
3700 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3702 (copy_blkmode_to_reg): Likewise.
3703 (set_storage_via_setmem): Likewise.
3704 * optabs.c (prepare_cmp_insn): Likewise.
3705 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3706 * stor-layout.c (finish_bitfield_representative): Likewise.
3708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3709 Alan Hayward <alan.hayward@arm.com>
3710 David Sherwood <david.sherwood@arm.com>
3712 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3713 * expr.c (convert_move): Use them.
3714 (convert_modes): Likewise.
3715 (store_expr_with_bounds): Likewise.
3717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3718 Alan Hayward <alan.hayward@arm.com>
3719 David Sherwood <david.sherwood@arm.com>
3721 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3722 parameter for the mode of "x". Remove the "known_x", "known_mode"
3723 and "known_ret" arguments. Change the type of the mode argument
3725 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3726 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3727 (reg_num_sign_bit_copies_for_combine): Likewise.
3728 * rtlanal.c (nonzero_bits1): Likewise.
3729 (num_sign_bit_copies1): Likewise.
3730 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3731 (reg_num_sign_bit_copies_general): Likewise.
3732 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3733 (reg_nonzero_bits_general): Likewise.
3735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3736 Alan Hayward <alan.hayward@arm.com>
3737 David Sherwood <david.sherwood@arm.com>
3739 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
3740 than in subroutines. Return 1 for non-integer modes.
3741 (cached_num_sign_bit_copies): Change the type of the mode parameter
3743 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
3744 classes. Handle CONST_INT_P first and then check whether X also
3745 has a scalar integer mode. Check the same thing for inner registers
3746 of a SUBREG and for values that are being extended or truncated.
3748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3749 Alan Hayward <alan.hayward@arm.com>
3750 David Sherwood <david.sherwood@arm.com>
3752 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
3753 in subroutines. Return the mode mask for non-integer modes.
3754 (cached_nonzero_bits): Change the type of the mode parameter
3756 (nonzero_bits1): Likewise. Remove early exit for other mode
3757 classes. Handle CONST_INT_P first and then check whether X
3758 also has a scalar integer mode.
3760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3761 Alan Hayward <alan.hayward@arm.com>
3762 David Sherwood <david.sherwood@arm.com>
3764 * expr.c (widest_int_mode_for_size): Make the comment match the code.
3765 Return a scalar_int_mode and assert that the size is greater than
3767 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
3768 (op_by_pieces_d::op_by_pieces_d): Likewise.
3769 (op_by_pieces_d::run): Likewise.
3770 (can_store_by_pieces): Likewise.
3772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3773 Alan Hayward <alan.hayward@arm.com>
3774 David Sherwood <david.sherwood@arm.com>
3776 * combine.c (extract_left_shift): Add a mode argument and update
3778 (make_compound_operation_int): Change the type of the mode parameter
3779 to scalar_int_mode and update the call to extract_left_shift.
3781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3782 Alan Hayward <alan.hayward@arm.com>
3783 David Sherwood <david.sherwood@arm.com>
3785 * combine.c (simplify_and_const_int): Change the type of the mode
3786 parameter to scalar_int_mode.
3787 (simplify_and_const_int_1): Likewise. Update recursive call.
3789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3790 Alan Hayward <alan.hayward@arm.com>
3791 David Sherwood <david.sherwood@arm.com>
3793 * combine.c (simplify_compare_const): Check that the mode is a
3794 scalar_int_mode (rather than VOIDmode) before testing its
3796 (simplify_comparison): Move COMPARISON_P handling out of the
3797 loop and restrict the latter part of the loop to scalar_int_modes.
3798 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
3799 and when considering SUBREG_REGs. Use is_int_mode instead of
3800 checking GET_MODE_CLASS against MODE_INT.
3802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3803 Alan Hayward <alan.hayward@arm.com>
3804 David Sherwood <david.sherwood@arm.com>
3806 * combine.c (try_widen_shift_mode): Move check for equal modes to...
3807 (simplify_shift_const_1): ...here. Use scalar_int_mode for
3808 shift_unit_mode and for modes involved in scalar shifts.
3810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3811 Alan Hayward <alan.hayward@arm.com>
3812 David Sherwood <david.sherwood@arm.com>
3814 * combine.c (force_int_to_mode): New function, split out from...
3815 (force_to_mode): ...here. Keep xmode up-to-date and use it
3816 instead of GET_MODE (x).
3818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3819 Alan Hayward <alan.hayward@arm.com>
3820 David Sherwood <david.sherwood@arm.com>
3822 * optabs-query.h (extraction_insn::struct_mode): Change type to
3823 opt_scalar_int_mode and update comment.
3824 (extraction_insn::field_mode): Change type to scalar_int_mode.
3825 (extraction_insn::pos_mode): Likewise.
3826 * combine.c (make_extraction): Update accordingly.
3827 * optabs-query.c (get_traditional_extraction_insn): Likewise.
3828 (get_optab_extraction_insn): Likewise.
3829 * recog.c (simplify_while_replacing): Likewise.
3830 * expmed.c (narrow_bit_field_mem): Change the type of the mode
3831 parameter to opt_scalar_int_mode.
3833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3834 Alan Hayward <alan.hayward@arm.com>
3835 David Sherwood <david.sherwood@arm.com>
3837 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
3838 to a scalar_int_mode instead of a machine_mode.
3839 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
3840 (get_best_mode): Return a boolean and use a pointer argument to store
3841 the selected mode. Replace the limit mode parameter with a bit limit.
3842 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
3843 for the values returned by bit_field_mode_iterator::next_mode.
3844 (store_bit_field): Update call to get_best_mode.
3845 (store_fixed_bit_field): Likewise.
3846 (extract_fixed_bit_field): Likewise.
3847 * expr.c (optimize_bitfield_assignment_op): Likewise.
3848 * fold-const.c (optimize_bit_field_compare): Likewise.
3849 (fold_truth_andor_1): Likewise.
3850 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
3851 Update for new type of m_mode.
3852 (get_best_mode): As above.
3854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3855 Alan Hayward <alan.hayward@arm.com>
3856 David Sherwood <david.sherwood@arm.com>
3858 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
3859 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
3860 (store_bit_field): Check is_a <scalar_int_mode> before calling
3861 strict_volatile_bitfield_p.
3862 (extract_bit_field): Likewise.
3864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3865 Alan Hayward <alan.hayward@arm.com>
3866 David Sherwood <david.sherwood@arm.com>
3868 * target.def (cstore_mode): Return a scalar_int_mode.
3869 * doc/tm.texi: Regenerate.
3870 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
3871 * targhooks.h (default_cstore_mode): Likewise.
3872 * targhooks.c (default_cstore_mode): Likewise, using a forced
3874 * expmed.c (emit_cstore): Expect the target of the cstore to be
3877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3878 Alan Hayward <alan.hayward@arm.com>
3879 David Sherwood <david.sherwood@arm.com>
3881 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
3883 (niter_desc): Likewise mode.
3884 (iv_analyze): Add a mode parameter.
3886 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
3887 and change its type to scalar_int_mode.
3888 * loop-iv.c: Update commentary at head of file.
3889 (iv_constant): Pass the mode paraeter before the rtx it describes
3890 and change its type to scalar_int_mode. Remove VOIDmode handling.
3891 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
3892 (iv_extend): Likewise.
3893 (shorten_into_mode): Likewise.
3894 (iv_add): Use scalar_int_mode.
3895 (iv_mult): Likewise.
3896 (iv_shift): Likewise.
3897 (canonicalize_iv_subregs): Likewise.
3898 (get_biv_step_1): Pass the outer_mode parameter before the rtx
3899 it describes and change its mode to scalar_int_mode. Also change
3900 the type of the returned inner_mode to scalar_int_mode.
3901 (get_biv_step): Likewise, turning outer_mode from a pointer
3902 into a direct parameter. Update call to get_biv_step_1.
3903 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
3904 iv_constant and get_biv_step.
3905 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
3906 and change its type to scalar_int_mode. Don't initialise iv->mode
3907 to VOIDmode and remove later checks for its still being VOIDmode.
3908 Update calls to iv_analyze_op and iv_analyze_expr. Check
3909 is_a <scalar_int_mode> when changing the mode under consideration.
3910 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
3911 Update call to iv_analyze_expr.
3912 (iv_analyze_op): Add a mode parameter. Reject subregs whose
3913 inner register is not also a scalar_int_mode. Update call to
3915 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
3916 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
3917 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
3918 separate mode class checks. Update calls to iv_analyze. Remove
3919 fix-up of VOIDmodes after iv_analyze_biv.
3920 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
3921 don't have a scalar_int_mode. Update call to biv_p.
3923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3924 Alan Hayward <alan.hayward@arm.com>
3925 David Sherwood <david.sherwood@arm.com>
3927 * cfgexpand.c (convert_debug_memory_address): Use
3928 as_a <scalar_int_mode>.
3929 * combine.c (expand_compound_operation): Likewise.
3930 (make_extraction): Likewise.
3931 (change_zero_ext): Likewise.
3932 (simplify_comparison): Likewise.
3933 * cse.c (cse_insn): Likewise.
3934 * dwarf2out.c (minmax_loc_descriptor): Likewise.
3935 (mem_loc_descriptor): Likewise.
3936 (loc_descriptor): Likewise.
3937 * expmed.c (init_expmed_one_mode): Likewise.
3938 (synth_mult): Likewise.
3939 (emit_store_flag_1): Likewise.
3940 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
3941 of a comparison with size.
3942 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
3943 (reduce_to_bit_field_precision): Likewise.
3944 * function.c (expand_function_end): Likewise.
3945 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
3946 * loop-doloop.c (doloop_modify): Likewise.
3947 * optabs.c (expand_binop): Likewise.
3948 (expand_unop): Likewise.
3949 (expand_copysign_absneg): Likewise.
3950 (prepare_cmp_insn): Likewise.
3951 (maybe_legitimize_operand): Likewise.
3952 * recog.c (const_scalar_int_operand): Likewise.
3953 * rtlanal.c (get_address_mode): Likewise.
3954 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3955 (simplify_cond_clz_ctz): Likewise.
3956 * tree-nested.c (get_nl_goto_field): Likewise.
3957 * tree.c (build_vector_type_for_mode): Likewise.
3958 * var-tracking.c (use_narrower_mode): Likewise.
3960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3961 Alan Hayward <alan.hayward@arm.com>
3962 David Sherwood <david.sherwood@arm.com>
3964 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
3965 * builtins.c (expand_builtin_signbit): Use it.
3966 * cfgexpand.c (expand_debug_expr): Likewise.
3967 * dojump.c (do_jump): Likewise.
3968 (do_compare_and_jump): Likewise.
3969 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
3970 * expmed.c (make_tree): Likewise.
3971 * expr.c (expand_expr_real_2): Likewise.
3972 (expand_expr_real_1): Likewise.
3973 (try_casesi): Likewise.
3974 * fold-const-call.c (fold_const_call_ss): Likewise.
3975 * fold-const.c (unextend): Likewise.
3976 (extract_muldiv_1): Likewise.
3977 (fold_single_bit_test): Likewise.
3978 (native_encode_int): Likewise.
3979 (native_encode_string): Likewise.
3980 (native_interpret_int): Likewise.
3981 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3982 * internal-fn.c (expand_addsub_overflow): Likewise.
3983 (expand_neg_overflow): Likewise.
3984 (expand_mul_overflow): Likewise.
3985 (expand_arith_overflow): Likewise.
3986 * match.pd: Likewise.
3987 * stor-layout.c (layout_type): Likewise.
3988 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3989 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3990 * tree-ssanames.c (get_range_info): Likewise.
3991 * tree-switch-conversion.c (array_value_type) Likewise.
3992 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
3993 (vect_recog_divmod_pattern): Likewise.
3994 (vect_recog_mixed_size_cond_pattern): Likewise.
3995 * tree-vrp.c (extract_range_basic): Likewise.
3996 (simplify_float_conversion_using_ranges): Likewise.
3997 * tree.c (int_fits_type_p): Likewise.
3998 * ubsan.c (instrument_bool_enum_load): Likewise.
3999 * varasm.c (mergeable_string_section): Likewise.
4000 (narrowing_initializer_constant_valid_p): Likewise.
4001 (output_constant): Likewise.
4003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4004 Alan Hayward <alan.hayward@arm.com>
4005 David Sherwood <david.sherwood@arm.com>
4007 * machmode.h (NARROWEST_INT_MODE): New macro.
4008 * expr.c (alignment_for_piecewise_move): Use it instead of
4009 GET_CLASS_NARROWEST_MODE (MODE_INT).
4010 (push_block): Likewise.
4011 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
4013 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4016 Alan Hayward <alan.hayward@arm.com>
4017 David Sherwood <david.sherwood@arm.com>
4019 * postreload.c (move2add_valid_value_p): Change the type of the
4020 mode parameter to scalar_int_mode.
4021 (move2add_use_add2_insn): Add a mode parameter and use it instead
4023 (move2add_use_add3_insn): Likewise.
4024 (reload_cse_move2add): Update accordingly.
4026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4027 Alan Hayward <alan.hayward@arm.com>
4028 David Sherwood <david.sherwood@arm.com>
4030 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
4032 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
4033 * optabs.c (expand_unop): Likewise.
4035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4036 Alan Hayward <alan.hayward@arm.com>
4037 David Sherwood <david.sherwood@arm.com>
4039 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
4040 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
4041 (popcount_loc_descriptor): Likewise.
4042 (bswap_loc_descriptor): Likewise.
4043 (rotate_loc_descriptor): Likewise.
4044 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
4045 calling the functions above.
4047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4048 Alan Hayward <alan.hayward@arm.com>
4049 David Sherwood <david.sherwood@arm.com>
4051 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
4053 (try_combine): Likewise.
4054 (simplify_if_then_else): Likewise.
4055 * cse.c (cse_insn): Likewise.
4056 * dwarf2out.c (mem_loc_descriptor): Likewise.
4057 * emit-rtl.c (gen_lowpart_common): Likewise.
4058 * simplify-rtx.c (simplify_truncation): Likewise.
4059 (simplify_binary_operation_1): Likewise.
4060 (simplify_const_relational_operation): Likewise.
4061 (simplify_ternary_operation): Likewise.
4062 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
4064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4065 Alan Hayward <alan.hayward@arm.com>
4066 David Sherwood <david.sherwood@arm.com>
4068 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
4069 * reload.c (push_reload): Likewise.
4070 (find_reloads): Likewise.
4072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4073 Alan Hayward <alan.hayward@arm.com>
4074 David Sherwood <david.sherwood@arm.com>
4076 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
4077 (make_compound_operation_int): Likewise.
4078 (change_zero_ext): Likewise.
4079 * expr.c (convert_move): Likewise.
4080 (convert_modes): Likewise.
4081 * fwprop.c (forward_propagate_subreg): Likewise.
4082 * loop-iv.c (get_biv_step_1): Likewise.
4083 * optabs.c (widen_operand): Likewise.
4084 * postreload.c (move2add_valid_value_p): Likewise.
4085 * recog.c (simplify_while_replacing): Likewise.
4086 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4087 (simplify_binary_operation_1): Likewise. Remove redundant
4088 mode equality check.
4090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4091 Alan Hayward <alan.hayward@arm.com>
4092 David Sherwood <david.sherwood@arm.com>
4094 * combine.c (combine_simplify_rtx): Add checks for
4095 is_a <scalar_int_mode>.
4096 (simplify_if_then_else): Likewise.
4097 (make_field_assignment): Likewise.
4098 (simplify_comparison): Likewise.
4099 * ifcvt.c (noce_try_bitop): Likewise.
4100 * loop-invariant.c (canonicalize_address_mult): Likewise.
4101 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4104 Alan Hayward <alan.hayward@arm.com>
4105 David Sherwood <david.sherwood@arm.com>
4107 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
4108 is_a <scalar_int_mode> instead of != BLKmode.
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 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
4115 instead of != VOIDmode.
4116 * combine.c (if_then_else_cond): Likewise.
4117 (change_zero_ext): Likewise.
4118 * dwarf2out.c (mem_loc_descriptor): Likewise.
4119 (loc_descriptor): Likewise.
4120 * rtlanal.c (canonicalize_condition): Likewise.
4121 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
4123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4124 Alan Hayward <alan.hayward@arm.com>
4125 David Sherwood <david.sherwood@arm.com>
4127 * simplify-rtx.c (simplify_binary_operation_1): Use
4128 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
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 * wide-int.h (int_traits<unsigned char>) New class.
4135 (int_traits<unsigned short>) Likewise.
4136 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
4137 Use GET_MODE_UNIT_PRECISION and remove redundant test for
4139 * combine.c (set_nonzero_bits_and_sign_copies): Use
4140 is_a <scalar_int_mode>.
4141 (find_split_point): Likewise.
4142 (combine_simplify_rtx): Likewise.
4143 (simplify_logical): Likewise.
4144 (expand_compound_operation): Likewise.
4145 (expand_field_assignment): Likewise.
4146 (make_compound_operation): Likewise.
4147 (extended_count): Likewise.
4148 (change_zero_ext): Likewise.
4149 (simplify_comparison): Likewise.
4150 * dwarf2out.c (scompare_loc_descriptor): Likewise.
4151 (ucompare_loc_descriptor): Likewise.
4152 (minmax_loc_descriptor): Likewise.
4153 (mem_loc_descriptor): Likewise.
4154 (loc_descriptor): Likewise.
4155 * expmed.c (init_expmed_one_mode): Likewise.
4156 * lra-constraints.c (lra_constraint_offset): Likewise.
4157 * optabs.c (prepare_libcall_arg): Likewise.
4158 * postreload.c (move2add_note_store): Likewise.
4159 * reload.c (operands_match_p): Likewise.
4160 * rtl.h (load_extend_op): Likewise.
4161 * rtlhooks.c (gen_lowpart_general): Likewise.
4162 * simplify-rtx.c (simplify_truncation): Likewise.
4163 (simplify_unary_operation_1): Likewise.
4164 (simplify_binary_operation_1): Likewise.
4165 (simplify_const_binary_operation): Likewise.
4166 (simplify_const_relational_operation): Likewise.
4167 (simplify_subreg): Likewise.
4168 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4169 * var-tracking.c (adjust_mems): Likewise.
4170 (prepare_call_arguments): Likewise.
4172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4173 Alan Hayward <alan.hayward@arm.com>
4174 David Sherwood <david.sherwood@arm.com>
4176 * machmode.h (is_int_mode): New fuction.
4177 * combine.c (find_split_point): Use it.
4178 (combine_simplify_rtx): Likewise.
4179 (simplify_if_then_else): Likewise.
4180 (simplify_set): Likewise.
4181 (simplify_shift_const_1): Likewise.
4182 (simplify_comparison): Likewise.
4183 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
4184 * cse.c (notreg_cost): Likewise.
4185 (cse_insn): Likewise.
4186 * cselib.c (cselib_lookup_1): Likewise.
4187 * dojump.c (do_jump_1): Likewise.
4188 (do_compare_rtx_and_jump): Likewise.
4189 * dse.c (get_call_args): Likewise.
4190 * dwarf2out.c (rtl_for_decl_init): Likewise.
4191 (native_encode_initializer): Likewise.
4192 * expmed.c (emit_store_flag_1): Likewise.
4193 (emit_store_flag): Likewise.
4194 * expr.c (convert_modes): Likewise.
4195 (store_field): Likewise.
4196 (expand_expr_real_1): Likewise.
4197 * fold-const.c (fold_read_from_constant_string): Likewise.
4198 * gimple-ssa-sprintf.c (get_format_string): Likewise.
4199 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
4200 * optabs.c (expand_binop): Likewise.
4201 (expand_unop): Likewise.
4202 (expand_abs_nojump): Likewise.
4203 (expand_one_cmpl_abs_nojump): Likewise.
4204 * simplify-rtx.c (mode_signbit_p): Likewise.
4205 (val_signbit_p): Likewise.
4206 (val_signbit_known_set_p): Likewise.
4207 (val_signbit_known_clear_p): Likewise.
4208 (simplify_relational_operation_1): Likewise.
4209 * tree.c (vector_type_mode): Likewise.
4211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4212 Alan Hayward <alan.hayward@arm.com>
4213 David Sherwood <david.sherwood@arm.com>
4215 * machmode.h (smallest_mode_for_size): Fix formatting.
4216 (smallest_int_mode_for_size): New function.
4217 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
4218 instead of smallest_mode_for_size.
4219 * combine.c (make_extraction): Likewise.
4220 * config/arc/arc.c (arc_expand_movmem): Likewise.
4221 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
4222 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4223 * config/s390/s390.c (s390_expand_insv): Likewise.
4224 * config/sparc/sparc.c (assign_int_registers): Likewise.
4225 * config/spu/spu.c (spu_function_value): Likewise.
4226 (spu_function_arg): Likewise.
4227 * coverage.c (get_gcov_type): Likewise.
4228 (get_gcov_unsigned_t): Likewise.
4229 * dse.c (find_shift_sequence): Likewise.
4230 * expmed.c (store_bit_field_1): Likewise.
4231 * expr.c (convert_move): Likewise.
4232 (store_field): Likewise.
4233 * internal-fn.c (expand_arith_overflow): Likewise.
4234 * optabs-query.c (get_best_extraction_insn): Likewise.
4235 * optabs.c (expand_twoval_binop_libfunc): Likewise.
4236 * stor-layout.c (layout_type): Likewise.
4237 (initialize_sizetypes): Likewise.
4238 * targhooks.c (default_get_mask_mode): Likewise.
4239 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
4241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4242 Alan Hayward <alan.hayward@arm.com>
4243 David Sherwood <david.sherwood@arm.com>
4245 * machmode.h (opt_mode::else_blk): New function.
4246 (int_mode_for_mode): Declare.
4247 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
4248 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
4250 * cfgexpand.c (expand_debug_expr): Likewise.
4251 * combine.c (gen_lowpart_or_truncate): Likewise.
4252 (gen_lowpart_for_combine): Likewise.
4253 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
4254 * config/avr/avr.c (avr_to_int_mode): Likewise.
4255 (avr_out_plus_1): Likewise.
4256 (avr_out_plus): Likewise.
4257 (avr_out_round): Likewise.
4258 * config/i386/i386.c (ix86_split_to_parts): Likewise.
4259 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4260 (s390_expand_vcond): Likewise.
4261 * config/spu/spu.c (spu_split_immediate): Likewise.
4262 (spu_expand_mov): Likewise.
4263 * dse.c (get_stored_val): Likewise.
4264 * expmed.c (store_bit_field_1): Likewise.
4265 (convert_extracted_bit_field): Use int_mode_for_mode instead of
4267 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
4268 (extract_low_bits): Likewise.
4269 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
4270 handling rather than repeating the check.
4271 (emit_group_store): Likewise.
4272 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
4273 * optabs.c (expand_absneg_bit): Likewise.
4274 (expand_copysign_absneg): Likewise.
4275 (expand_copysign_bit): Likewise.
4276 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4277 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
4278 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
4279 * var-tracking.c (prepare_call_arguments): Likewise.
4280 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
4281 int_mode_for_mode instead of mode_for_size.
4282 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4285 Alan Hayward <alan.hayward@arm.com>
4286 David Sherwood <david.sherwood@arm.com>
4288 * machmode.h (int_mode_for_size): New function.
4289 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
4290 instead of mode_for_size.
4291 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
4293 * combine.c (expand_field_assignment): Use int_mode_for_size
4294 instead of mode_for_size.
4295 (make_extraction): Likewise.
4296 (simplify_shift_const_1): Likewise.
4297 (simplify_comparison): Likewise.
4298 * dojump.c (do_jump): Likewise.
4299 * dwarf2out.c (mem_loc_descriptor): Likewise.
4300 * emit-rtl.c (init_derived_machine_modes): Likewise.
4301 * expmed.c (flip_storage_order): Likewise.
4302 (convert_extracted_bit_field): Likewise.
4303 * expr.c (copy_blkmode_from_reg): Likewise.
4304 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
4305 * internal-fn.c (expand_mul_overflow): Likewise.
4306 * lower-subreg.c (simple_move): Likewise.
4307 * optabs-libfuncs.c (init_optabs): Likewise.
4308 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4309 * tree.c (vector_type_mode): Likewise.
4310 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
4311 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
4312 * tree-vect-generic.c (expand_vector_parallel): Likewise.
4313 * tree-vect-stmts.c (vectorizable_load): Likewise.
4314 (vectorizable_store): Likewise.
4316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4317 Alan Hayward <alan.hayward@arm.com>
4318 David Sherwood <david.sherwood@arm.com>
4320 * coretypes.h (pod_mode): New type.
4321 (scalar_int_mode_pod): New typedef.
4322 * machmode.h (pod_mode): New class.
4323 (int_n_data_t::m): Change type to scalar_int_mode_pod.
4324 * genmodes.c (emit_mode_int_n): Update accordingly.
4325 * lower-subreg.h (target_lower_subreg): Change type to
4326 scalar_int_mode_pod.
4327 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
4328 scalar_int_mode_pod.
4330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4331 Alan Hayward <alan.hayward@arm.com>
4332 David Sherwood <david.sherwood@arm.com>
4334 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
4335 machine_mode to scalar_int_mode.
4336 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
4337 (rs6000_option_override_internal): Remove cast to int.
4338 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
4339 machine_mode to scalar_int_mode.
4340 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
4341 (rs6000_option_override_internal): Remove cast to int.
4342 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
4343 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
4345 * config/s390/s390.c (s390_expand_builtin): Likewise.
4346 * coretypes.h (scalar_int_mode): New type.
4347 (opt_scalar_int_mode): New typedef.
4348 * machmode.h (scalar_int_mode): New class.
4349 (scalar_int_mode::includes_p): New function.
4350 (byte_mode): Change type to scalar_int_mode.
4351 (word_mode): Likewise.
4352 (ptr_mode): Likewise.
4353 * emit-rtl.c (byte_mode): Likewise.
4354 (word_mode): Likewise.
4355 (ptr_mode): Likewise.
4356 (init_derived_machine_modes): Update accordingly.
4357 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
4358 and MODE_PARTIAL_INT.
4359 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
4360 opt_scalar_int_mode.
4362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4363 Alan Hayward <alan.hayward@arm.com>
4364 David Sherwood <david.sherwood@arm.com>
4366 * target.def (libgcc_floating_mode_supported_p): Take a
4368 * doc/tm.texi: Regenerate.
4369 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
4371 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4372 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
4375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4376 Alan Hayward <alan.hayward@arm.com>
4377 David Sherwood <david.sherwood@arm.com>
4379 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
4380 * doc/tm.texi: Regenerate.
4381 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
4382 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
4383 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
4384 * targhooks.h (default_floatn_mode): Likewise.
4385 * targhooks.c (default_floatn_mode): Likewise.
4386 * tree.c (build_common_tree_nodes): Update accordingly.
4388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4389 Alan Hayward <alan.hayward@arm.com>
4390 David Sherwood <david.sherwood@arm.com>
4392 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
4393 (mode_iterator::iterate_p): Likewise.
4394 (mode_iterator::get_wider): Likewise.
4395 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4398 Alan Hayward <alan.hayward@arm.com>
4399 David Sherwood <david.sherwood@arm.com>
4401 * coretypes.h (opt_scalar_float_mode): New typedef.
4402 * machmode.h (float_mode_for_size): New function.
4403 * emit-rtl.c (double_mode): Delete.
4404 (init_emit_once): Use float_mode_for_size.
4405 * stor-layout.c (layout_type): Likewise.
4406 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4409 Alan Hayward <alan.hayward@arm.com>
4410 David Sherwood <david.sherwood@arm.com>
4412 * output.h (assemble_real): Take a scalar_float_mode.
4413 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4414 * config/arm/arm.md (consttable_4): Likewise.
4415 (consttable_8): Likewise.
4416 (consttable_16): Likewise.
4417 * config/mips/mips.md (consttable_float): Likewise.
4418 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4419 * varasm.c (assemble_real): Take a scalar_float_mode.
4420 (output_constant_pool_2): Update accordingly.
4421 (output_constant): Likewise.
4423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4424 Alan Hayward <alan.hayward@arm.com>
4425 David Sherwood <david.sherwood@arm.com>
4427 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4428 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4429 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4430 (native_encode_real): Likewise.
4431 (native_interpret_real): Likewise.
4432 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4433 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4436 Alan Hayward <alan.hayward@arm.com>
4437 David Sherwood <david.sherwood@arm.com>
4439 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4440 <scalar_float_mode>. Simplify.
4441 (gen_extend_conv_libfunc): Likewise.
4443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4444 Alan Hayward <alan.hayward@arm.com>
4445 David Sherwood <david.sherwood@arm.com>
4447 * coretypes.h (scalar_float_mode): New type.
4448 * machmode.h (mode_traits::from_int): Use machine_mode if
4449 USE_ENUM_MODES is defined.
4450 (is_a): New function.
4452 (dyn_cast): Likewise.
4453 (scalar_float_mode): New class.
4454 (scalar_float_mode::includes_p): New function.
4455 (is_float_mode): Likewise.
4456 * gdbhooks.py (MachineModePrinter): New class.
4457 (build_pretty_printer): Use it for scalar_float_mode.
4458 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4459 (format_helper::format_helper): Turn into a template.
4460 * genmodes.c (get_mode_class): New function.
4461 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4462 or machine_mode if none.
4463 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4464 as_a <scalar_float_mode>.
4465 * dwarf2out.c (mem_loc_descriptor): Likewise.
4466 (insert_float): Likewise.
4467 (add_const_value_attribute): Likewise.
4468 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4469 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4470 (expand_unop): Update accordingly.
4471 (expand_abs_nojump): Likewise.
4472 (expand_copysign_absneg): Take a scalar_float_mode.
4473 (expand_copysign_bit): Likewise.
4474 (expand_copysign): Update accordingly.
4476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4477 Alan Hayward <alan.hayward@arm.com>
4478 David Sherwood <david.sherwood@arm.com>
4480 * coretypes.h (opt_mode): New class.
4481 * machmode.h (opt_mode): Likewise.
4482 (opt_mode::else_void): New function.
4483 (opt_mode::require): Likewise.
4484 (opt_mode::exists): Likewise.
4485 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4486 (GET_MODE_2XWIDER_MODE): Likewise.
4487 (mode_iterator::get_wider): Update accordingly.
4488 (mode_iterator::get_2xwider): Likewise.
4489 (mode_iterator::get_known_wider): Likewise, turning into a template.
4490 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4491 forcing a wider mode to exist.
4492 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4493 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4494 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4495 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4496 * lower-subreg.c (init_lower_subreg): Likewise.
4497 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4498 on the final iteration.
4499 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4500 a wider mode exists before asking for a move pattern.
4501 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4502 forcing a wider mode to exist.
4503 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4504 returning false if no such mode exists.
4505 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4506 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4507 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4508 Avoid checking for a MODE_INT if we already know the mode is not a
4510 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4511 forcing a wider mode to exist.
4512 (expmed_mult_highpart_optab): Likewise.
4513 (expmed_mult_highpart): Likewise.
4514 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4516 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4517 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4518 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4519 * internal-fn.c (expand_mul_overflow): Update use of
4520 GET_MODE_2XWIDER_MODE.
4521 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4522 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4523 GET_MODE_WIDER_MODE.
4524 (convert_plusminus_to_widen): Likewise.
4525 * tree-switch-conversion.c (array_value_type): Likewise.
4526 * var-tracking.c (emit_note_insn_var_location): Likewise.
4527 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4528 Return false inside rather than outside the loop if no wider mode
4530 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4531 and GET_MODE_2XWIDER_MODE
4532 (can_compare_p): Use else_void.
4533 * gdbhooks.py (OptMachineModePrinter): New class.
4534 (build_pretty_printer): Use it for opt_mode.
4536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4537 Alan Hayward <alan.hayward@arm.com>
4538 David Sherwood <david.sherwood@arm.com>
4540 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4541 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_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 * machmode.h (mode_traits): New structure.
4548 (get_narrowest_mode): New function.
4549 (mode_iterator::start): Likewise.
4550 (mode_iterator::iterate_p): Likewise.
4551 (mode_iterator::get_wider): Likewise.
4552 (mode_iterator::get_known_wider): Likewise.
4553 (mode_iterator::get_2xwider): Likewise.
4554 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4555 (FOR_EACH_MODE): Likewise.
4556 (FOR_EACH_MODE_FROM): Likewise.
4557 (FOR_EACH_MODE_UNTIL): Likewise.
4558 (FOR_EACH_WIDER_MODE): Likewise.
4559 (FOR_EACH_2XWIDER_MODE): Likewise.
4560 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4561 * combine.c (simplify_comparison): Likewise
4562 * config/i386/i386.c (type_natural_mode): Likewise.
4563 * cse.c (cse_insn): Likewise.
4564 * dse.c (find_shift_sequence): Likewise.
4565 * emit-rtl.c (init_derived_machine_modes): Likewise.
4566 (init_emit_once): Likewise.
4567 * explow.c (hard_function_value): Likewise.
4568 * expmed.c (extract_fixed_bit_field_1): Likewise.
4569 (extract_bit_field_1): Likewise.
4570 (expand_divmod): Likewise.
4571 (emit_store_flag_1): Likewise.
4572 * expr.c (init_expr_target): Likewise.
4573 (convert_move): Likewise.
4574 (alignment_for_piecewise_move): Likewise.
4575 (widest_int_mode_for_size): Likewise.
4576 (emit_block_move_via_movmem): Likewise.
4577 (copy_blkmode_to_reg): Likewise.
4578 (set_storage_via_setmem): Likewise.
4579 (compress_float_constant): Likewise.
4580 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4581 * optabs-query.c (get_best_extraction_insn): Likewise.
4582 * optabs.c (expand_binop): Likewise.
4583 (expand_twoval_unop): Likewise.
4584 (expand_twoval_binop): Likewise.
4585 (widen_leading): Likewise.
4586 (widen_bswap): Likewise.
4587 (expand_parity): Likewise.
4588 (expand_unop): Likewise.
4589 (prepare_cmp_insn): Likewise.
4590 (prepare_float_lib_cmp): Likewise.
4591 (expand_float): Likewise.
4592 (expand_fix): Likewise.
4593 (expand_sfix_optab): Likewise.
4594 * postreload.c (move2add_use_add2_insn): Likewise.
4595 * reg-stack.c (reg_to_stack): Likewise.
4596 * reginfo.c (choose_hard_reg_mode): Likewise.
4597 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4598 * stor-layout.c (mode_for_size): Likewise.
4599 (smallest_mode_for_size): Likewise.
4600 (mode_for_vector): Likewise.
4601 (finish_bitfield_representative): Likewise.
4602 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4603 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4604 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4605 * var-tracking.c (prepare_call_arguments): Likewise.
4607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4608 Alan Hayward <alan.hayward@arm.com>
4609 David Sherwood <david.sherwood@arm.com>
4611 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4612 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4613 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4614 * machmode.h (mode_size): Move earlier in file.
4615 (mode_precision): Likewise.
4616 (mode_inner): Likewise.
4617 (mode_nunits): Likewise.
4618 (mode_unit_size): Likewise.
4619 (unit_unit_precision): Likewise.
4620 (mode_wider): Likewise.
4621 (mode_2xwider): Likewise.
4622 (machine_mode): New class.
4623 (mode_to_bytes): New function.
4624 (mode_to_bits): Likewise.
4625 (mode_to_precision): Likewise.
4626 (mode_to_inner): Likewise.
4627 (mode_to_unit_size): Likewise.
4628 (mode_to_unit_precision): Likewise.
4629 (mode_to_nunits): Likewise.
4630 (GET_MODE_SIZE): Use mode_to_bytes.
4631 (GET_MODE_BITSIZE): Use mode_to_bits.
4632 (GET_MODE_PRECISION): Use mode_to_precision.
4633 (GET_MODE_INNER): Use mode_to_inner.
4634 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4635 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
4636 (GET_MODE_NUNITS): Use mode_to_nunits.
4637 * system.h (ALWAYS_INLINE): New macro.
4638 * config/powerpcspe/powerpcspe-c.c
4639 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
4640 int for arg1_mode and arg2_mode.
4642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4643 Alan Hayward <alan.hayward@arm.com>
4644 David Sherwood <david.sherwood@arm.com>
4646 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
4647 Prefix mode names with E_ in case statements.
4648 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4649 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
4650 (aarch64_split_simd_move): Likewise.
4651 (aarch64_gen_storewb_pair): Likewise.
4652 (aarch64_gen_loadwb_pair): Likewise.
4653 (aarch64_gen_store_pair): Likewise.
4654 (aarch64_gen_load_pair): Likewise.
4655 (aarch64_get_condition_code_1): Likewise.
4656 (aarch64_constant_pool_reload_icode): Likewise.
4657 (get_rsqrte_type): Likewise.
4658 (get_rsqrts_type): Likewise.
4659 (get_recpe_type): Likewise.
4660 (get_recps_type): Likewise.
4661 (aarch64_gimplify_va_arg_expr): Likewise.
4662 (aarch64_simd_container_mode): Likewise.
4663 (aarch64_emit_load_exclusive): Likewise.
4664 (aarch64_emit_store_exclusive): Likewise.
4665 (aarch64_expand_compare_and_swap): Likewise.
4666 (aarch64_gen_atomic_cas): Likewise.
4667 (aarch64_emit_bic): Likewise.
4668 (aarch64_emit_atomic_swap): Likewise.
4669 (aarch64_emit_atomic_load_op): Likewise.
4670 (aarch64_evpc_trn): Likewise.
4671 (aarch64_evpc_uzp): Likewise.
4672 (aarch64_evpc_zip): Likewise.
4673 (aarch64_evpc_ext): Likewise.
4674 (aarch64_evpc_rev): Likewise.
4675 (aarch64_evpc_dup): Likewise.
4676 (aarch64_gen_ccmp_first): Likewise.
4677 (aarch64_gen_ccmp_next): Likewise.
4678 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4679 (alpha_emit_xfloating_libcall): Likewise.
4680 (emit_insxl): Likewise.
4681 (alpha_arg_type): Likewise.
4682 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4683 (arc_preferred_simd_mode): Likewise.
4684 (arc_secondary_reload): Likewise.
4685 (get_arc_condition_code): Likewise.
4686 (arc_print_operand): Likewise.
4687 (arc_legitimate_constant_p): Likewise.
4688 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4689 * config/arc/arc.md (casesi_load): Likewise.
4690 (casesi_compact_jump): Likewise.
4691 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4692 (cc_use_register): Likewise.
4693 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4694 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4695 (arm_init_iwmmxt_builtins): Likewise.
4696 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4697 (neon_expand_vector_init): Likewise.
4698 (arm_attr_length_move_neon): Likewise.
4699 (maybe_get_arm_condition_code): Likewise.
4700 (arm_emit_vector_const): Likewise.
4701 (arm_preferred_simd_mode): Likewise.
4702 (arm_output_iwmmxt_tinsr): Likewise.
4703 (thumb1_output_casesi): Likewise.
4704 (thumb2_output_casesi): Likewise.
4705 (arm_emit_load_exclusive): Likewise.
4706 (arm_emit_store_exclusive): Likewise.
4707 (arm_expand_compare_and_swap): Likewise.
4708 (arm_evpc_neon_vuzp): Likewise.
4709 (arm_evpc_neon_vzip): Likewise.
4710 (arm_evpc_neon_vrev): Likewise.
4711 (arm_evpc_neon_vtrn): Likewise.
4712 (arm_evpc_neon_vext): Likewise.
4713 (arm_validize_comparison): Likewise.
4714 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4715 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4716 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4717 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4718 (c6x_preferred_simd_mode): Likewise.
4719 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4720 (epiphany_rtx_costs): Likewise.
4721 * config/epiphany/predicates.md (proper_comparison_operator):
4723 * config/frv/frv.c (condexec_memory_operand): Likewise.
4724 (frv_emit_move): Likewise.
4725 (output_move_single): Likewise.
4726 (output_condmove_single): Likewise.
4727 (frv_hard_regno_mode_ok): Likewise.
4728 (frv_matching_accg_mode): Likewise.
4729 * config/h8300/h8300.c (split_adds_subs): Likewise.
4730 (h8300_rtx_costs): Likewise.
4731 (h8300_print_operand): Likewise.
4732 (compute_mov_length): Likewise.
4733 (output_logical_op): Likewise.
4734 (compute_logical_op_length): Likewise.
4735 (compute_logical_op_cc): Likewise.
4736 (h8300_shift_needs_scratch_p): Likewise.
4737 (output_a_shift): Likewise.
4738 (compute_a_shift_length): Likewise.
4739 (compute_a_shift_cc): Likewise.
4740 (expand_a_rotate): Likewise.
4741 (output_a_rotate): Likewise.
4742 * config/i386/i386.c (classify_argument): Likewise.
4743 (function_arg_advance_32): Likewise.
4744 (function_arg_32): Likewise.
4745 (function_arg_64): Likewise.
4746 (function_value_64): Likewise.
4747 (ix86_gimplify_va_arg): Likewise.
4748 (ix86_legitimate_constant_p): Likewise.
4749 (put_condition_code): Likewise.
4750 (split_double_mode): Likewise.
4751 (ix86_avx256_split_vector_move_misalign): Likewise.
4752 (ix86_expand_vector_logical_operator): Likewise.
4753 (ix86_split_idivmod): Likewise.
4754 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
4755 (ix86_build_const_vector): Likewise.
4756 (ix86_build_signbit_mask): Likewise.
4757 (ix86_match_ccmode): Likewise.
4758 (ix86_cc_modes_compatible): Likewise.
4759 (ix86_expand_branch): Likewise.
4760 (ix86_expand_sse_cmp): Likewise.
4761 (ix86_expand_sse_movcc): Likewise.
4762 (ix86_expand_int_sse_cmp): Likewise.
4763 (ix86_expand_vec_perm_vpermi2): Likewise.
4764 (ix86_expand_vec_perm): Likewise.
4765 (ix86_expand_sse_unpack): Likewise.
4766 (ix86_expand_int_addcc): Likewise.
4767 (ix86_split_to_parts): Likewise.
4768 (ix86_vectorize_builtin_gather): Likewise.
4769 (ix86_vectorize_builtin_scatter): Likewise.
4770 (avx_vpermilp_parallel): Likewise.
4771 (inline_memory_move_cost): Likewise.
4772 (ix86_tieable_integer_mode_p): Likewise.
4773 (x86_maybe_negate_const_int): Likewise.
4774 (ix86_expand_vector_init_duplicate): Likewise.
4775 (ix86_expand_vector_init_one_nonzero): Likewise.
4776 (ix86_expand_vector_init_one_var): Likewise.
4777 (ix86_expand_vector_init_concat): Likewise.
4778 (ix86_expand_vector_init_interleave): Likewise.
4779 (ix86_expand_vector_init_general): Likewise.
4780 (ix86_expand_vector_set): Likewise.
4781 (ix86_expand_vector_extract): Likewise.
4782 (emit_reduc_half): Likewise.
4783 (ix86_emit_i387_round): Likewise.
4784 (ix86_mangle_type): Likewise.
4785 (ix86_expand_round_sse4): Likewise.
4786 (expand_vec_perm_blend): Likewise.
4787 (canonicalize_vector_int_perm): Likewise.
4788 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
4789 (expand_vec_perm_1): Likewise.
4790 (expand_vec_perm_interleave3): Likewise.
4791 (expand_vec_perm_even_odd_pack): Likewise.
4792 (expand_vec_perm_even_odd_1): Likewise.
4793 (expand_vec_perm_broadcast_1): Likewise.
4794 (ix86_vectorize_vec_perm_const_ok): Likewise.
4795 (ix86_expand_vecop_qihi): Likewise.
4796 (ix86_expand_mul_widen_hilo): Likewise.
4797 (ix86_expand_sse2_abs): Likewise.
4798 (ix86_expand_pextr): Likewise.
4799 (ix86_expand_pinsr): Likewise.
4800 (ix86_preferred_simd_mode): Likewise.
4801 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
4802 * config/i386/sse.md (*andnot<mode>3): Likewise.
4803 (<mask_codefor><code><mode>3<mask_name>): Likewise.
4804 (*<code><mode>3): Likewise.
4805 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
4806 (ia64_expand_atomic_op): Likewise.
4807 (ia64_arg_type): Likewise.
4808 (ia64_mode_to_int): Likewise.
4809 (ia64_scalar_mode_supported_p): Likewise.
4810 (ia64_vector_mode_supported_p): Likewise.
4811 (expand_vec_perm_broadcast): Likewise.
4812 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
4813 (iq2000_function_arg_advance): Likewise.
4814 (iq2000_function_arg): Likewise.
4815 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
4816 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
4817 (m68k_libcall_value): Likewise.
4818 (m68k_function_value): Likewise.
4819 (sched_attr_op_type): Likewise.
4820 * config/mcore/mcore.c (mcore_output_move): Likewise.
4821 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
4823 (microblaze_function_arg): Likewise.
4824 * config/mips/mips.c (mips16_build_call_stub): Likewise.
4825 (mips_print_operand): Likewise.
4826 (mips_mode_ok_for_mov_fmt_p): Likewise.
4827 (mips_vector_mode_supported_p): Likewise.
4828 (mips_preferred_simd_mode): Likewise.
4829 (mips_expand_vpc_loongson_even_odd): Likewise.
4830 (mips_expand_vec_unpack): Likewise.
4831 (mips_expand_vi_broadcast): Likewise.
4832 (mips_expand_vector_init): Likewise.
4833 (mips_expand_vec_reduc): Likewise.
4834 (mips_expand_msa_cmp): Likewise.
4835 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
4836 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
4837 (cc_flags_for_mode): Likewise.
4838 * config/msp430/msp430.c (msp430_print_operand): Likewise.
4839 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
4840 (nds32_output_casesi_pc_relative): Likewise.
4841 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4842 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
4843 (nvptx_gen_unpack): Likewise.
4844 (nvptx_gen_pack): Likewise.
4845 (nvptx_gen_shuffle): Likewise.
4846 (nvptx_gen_wcast): Likewise.
4847 (nvptx_preferred_simd_mode): Likewise.
4848 * config/pa/pa.c (pa_secondary_reload): Likewise.
4849 * config/pa/predicates.md (base14_operand): Likewise.
4850 * config/powerpcspe/powerpcspe-c.c
4851 (altivec_resolve_overloaded_builtin): Likewise.
4852 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
4854 (rs6000_preferred_simd_mode): Likewise.
4855 (output_vec_const_move): Likewise.
4856 (rs6000_expand_vector_extract): Likewise.
4857 (rs6000_split_vec_extract_var): Likewise.
4858 (reg_offset_addressing_ok_p): Likewise.
4859 (rs6000_legitimate_offset_address_p): Likewise.
4860 (rs6000_legitimize_address): Likewise.
4861 (rs6000_emit_set_const): Likewise.
4862 (rs6000_const_vec): Likewise.
4863 (rs6000_emit_move): Likewise.
4864 (spe_build_register_parallel): Likewise.
4865 (rs6000_darwin64_record_arg_recurse): Likewise.
4866 (swap_selector_for_mode): Likewise.
4867 (spe_init_builtins): Likewise.
4868 (paired_init_builtins): Likewise.
4869 (altivec_init_builtins): Likewise.
4870 (do_load_for_compare): Likewise.
4871 (rs6000_generate_compare): Likewise.
4872 (rs6000_expand_float128_convert): Likewise.
4873 (emit_load_locked): Likewise.
4874 (emit_store_conditional): Likewise.
4875 (rs6000_output_function_epilogue): Likewise.
4876 (rs6000_handle_altivec_attribute): Likewise.
4877 (rs6000_function_value): Likewise.
4878 (emit_fusion_gpr_load): Likewise.
4879 (emit_fusion_p9_load): Likewise.
4880 (emit_fusion_p9_store): Likewise.
4881 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
4882 (fusion_gpr_mem_load): Likewise.
4883 (fusion_addis_mem_combo_load): Likewise.
4884 (fusion_addis_mem_combo_store): Likewise.
4885 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
4886 (fusion_gpr_mem_load): Likewise.
4887 (fusion_addis_mem_combo_load): Likewise.
4888 (fusion_addis_mem_combo_store): Likewise.
4889 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4891 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
4892 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4893 (rs6000_preferred_simd_mode): Likewise.
4894 (output_vec_const_move): Likewise.
4895 (rs6000_expand_vector_extract): Likewise.
4896 (rs6000_split_vec_extract_var): Likewise.
4897 (reg_offset_addressing_ok_p): Likewise.
4898 (rs6000_legitimate_offset_address_p): Likewise.
4899 (rs6000_legitimize_address): Likewise.
4900 (rs6000_emit_set_const): Likewise.
4901 (rs6000_const_vec): Likewise.
4902 (rs6000_emit_move): Likewise.
4903 (rs6000_darwin64_record_arg_recurse): Likewise.
4904 (swap_selector_for_mode): Likewise.
4905 (paired_init_builtins): Likewise.
4906 (altivec_init_builtins): Likewise.
4907 (rs6000_expand_float128_convert): Likewise.
4908 (emit_load_locked): Likewise.
4909 (emit_store_conditional): Likewise.
4910 (rs6000_output_function_epilogue): Likewise.
4911 (rs6000_handle_altivec_attribute): Likewise.
4912 (rs6000_function_value): Likewise.
4913 (emit_fusion_gpr_load): Likewise.
4914 (emit_fusion_p9_load): Likewise.
4915 (emit_fusion_p9_store): Likewise.
4916 * config/rx/rx.c (rx_gen_move_template): Likewise.
4917 (flags_from_mode): Likewise.
4918 * config/s390/predicates.md (s390_alc_comparison): Likewise.
4919 (s390_slb_comparison): Likewise.
4920 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
4921 (s390_vector_mode_supported_p): Likewise.
4922 (s390_cc_modes_compatible): Likewise.
4923 (s390_match_ccmode_set): Likewise.
4924 (s390_canonicalize_comparison): Likewise.
4925 (s390_emit_compare_and_swap): Likewise.
4926 (s390_branch_condition_mask): Likewise.
4927 (s390_rtx_costs): Likewise.
4928 (s390_secondary_reload): Likewise.
4929 (__SECONDARY_RELOAD_CASE): Likewise.
4930 (s390_expand_cs): Likewise.
4931 (s390_preferred_simd_mode): Likewise.
4932 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
4933 * config/sh/sh.c (sh_print_operand): Likewise.
4934 (dump_table): Likewise.
4935 (sh_secondary_reload): Likewise.
4936 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4937 * config/sh/sh.md (casesi_worker_1): Likewise.
4938 (casesi_worker_2): Likewise.
4939 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
4940 (fcc_comparison_operator): Likewise.
4941 * config/sparc/sparc.c (sparc_expand_move): Likewise.
4942 (emit_soft_tfmode_cvt): Likewise.
4943 (sparc_preferred_simd_mode): Likewise.
4944 (output_cbranch): Likewise.
4945 (sparc_print_operand): Likewise.
4946 (sparc_expand_vec_perm_bmask): Likewise.
4947 (vector_init_bshuffle): Likewise.
4948 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4949 (spu_vector_mode_supported_p): Likewise.
4950 (spu_expand_insv): Likewise.
4951 (spu_emit_branch_or_set): Likewise.
4952 (spu_handle_vector_attribute): Likewise.
4953 (spu_builtin_splats): Likewise.
4954 (spu_builtin_extract): Likewise.
4955 (spu_builtin_promote): Likewise.
4956 (spu_expand_sign_extend): Likewise.
4957 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4958 (tilegx_simd_int): Likewise.
4959 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
4960 (tilepro_simd_int): Likewise.
4961 * config/v850/v850.c (const_double_split): Likewise.
4962 (v850_print_operand): Likewise.
4963 (ep_memory_offset): Likewise.
4964 * config/vax/vax.c (vax_rtx_costs): Likewise.
4965 (vax_output_int_move): Likewise.
4966 (vax_output_int_add): Likewise.
4967 (vax_output_int_subtract): Likewise.
4968 * config/visium/predicates.md (visium_branch_operator): Likewise.
4969 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
4970 (visium_print_operand_address): Likewise.
4971 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4972 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
4973 (xtensa_expand_conditional_branch): Likewise.
4974 (xtensa_copy_incoming_a7): Likewise.
4975 (xtensa_output_literal): Likewise.
4976 * dfp.c (decimal_real_maxval): Likewise.
4977 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4980 Alan Hayward <alan.hayward@arm.com>
4981 David Sherwood <david.sherwood@arm.com>
4983 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4984 (mode_nunits_inline): Likewise.
4985 (mode_inner_inline): Likewise.
4986 (mode_unit_size_inline): Likewise.
4987 (mode_unit_precision_inline): Likewise.
4988 (emit_insn_modes_h): Likewise. Also emit a #define of the
4990 (emit_mode_wider): Add an E_ prefix to mode names.
4991 (emit_mode_complex): Likewise.
4992 (emit_mode_inner): Likewise.
4993 (emit_mode_adjustments): Likewise.
4994 (emit_mode_int_n): Likewise.
4995 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
4996 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
4997 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
4998 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
4999 (CRC32_BUILTIN, ENTRY): Likewise.
5000 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
5001 (aarch64_pop_regs): Likewise.
5002 (aarch64_process_components): Likewise.
5003 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
5004 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
5005 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
5006 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
5007 * config/arm/arm.c (arm_init_libfuncs): Likewise.
5008 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
5010 * config/i386/i386-builtin.def (pcmpestr): Likewise.
5011 (pcmpistr): Likewise.
5012 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
5013 * config/mmix/mmix.c (mmix_output_condition): Likewise.
5014 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
5016 * config/rl78/rl78.c (mduc_regs): Likewise.
5017 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5018 (htm_expand_builtin): Likewise.
5019 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
5020 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
5021 * config/xtensa/xtensa.c (print_operand): Likewise.
5022 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
5023 (NUM_MODE_VECTOR_INT): Likewise.
5024 * genoutput.c (null_operand): Likewise.
5025 (output_operand_data): Likewise.
5026 * genrecog.c (print_parameter_value): Likewise.
5027 * lra.c (debug_operand_data): Likewise.
5029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5031 * dwarf2out.c (scompare_loc_descriptor_wide)
5032 (scompare_loc_descriptor_narrow): New functions, split out from...
5033 (scompare_loc_descriptor): ...here.
5034 * expmed.c (emit_store_flag_int): New function, split out from...
5035 (emit_store_flag): ...here.
5037 2017-08-30 Richard Biener <rguenther@suse.de>
5039 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
5040 (dwarf2out_early_finish): Move setting of AT_pubnames from
5041 early debug output to early finish.
5043 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5045 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
5046 and -mdata-region to the assembler.
5048 2017-08-30 Richard Biener <rguenther@suse.de>
5050 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
5052 (gen_subprogram_die): Add DW_AT_object_pointer only early.
5053 (dwarf2out_early_global_decl): Only generate a DIE for the
5054 abstract origin if it doesn't already exist or is a declaration DIE.
5055 (resolve_addr): Do not add the linkage name twice when
5056 generating a stub DIE for the DW_TAG_GNU_call_site target.
5058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5060 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5061 Use machine_mode rather than int for arg1_mode.
5063 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
5066 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
5067 that the second argument of the built-in functions to unpack
5068 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
5069 switch statement instead a lot of if statements.
5070 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
5071 Allow 64-bit values to be in Altivec registers as well as
5072 traditional floating point registers.
5073 (pack<mode>, FMOVE128_VSX iterator): Likewise.
5075 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
5077 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
5078 MAX_REGS_PER_ADDRESS == 1.
5080 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
5082 * config/i386/i386.opt (flag_fentry): Do not init to -1.
5083 * config/i386/i386.c (ix86_option_override_internal): Simplify
5084 setting of opts->x_flag_entry.
5086 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5087 Jakub Jelinek <jakub@redhat.com>
5088 Richard Biener <rguenther@suse.de>
5090 PR tree-optimization/81503
5091 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
5092 folded constant fits in the target type; reorder tests for clarity.
5094 2017-08-29 Martin Liska <mliska@suse.cz>
5096 * passes.def: Include pass_lower_switch.
5097 * stmt.c (dump_case_nodes): Remove and move to
5098 tree-switch-conversion.
5099 (case_values_threshold): Likewise.
5100 (expand_switch_as_decision_tree_p): Likewise.
5101 (emit_case_decision_tree): Likewise.
5102 (expand_case): Likewise.
5103 (balance_case_nodes): Likewise.
5104 (node_has_low_bound): Likewise.
5105 (node_has_high_bound): Likewise.
5106 (node_is_bounded): Likewise.
5107 (emit_case_nodes): Likewise.
5108 (struct simple_case_node): New struct.
5109 (add_case_node): Remove.
5110 (emit_case_dispatch_table): Use vector instead of case_list.
5111 (reset_out_edges_aux): Remove.
5112 (compute_cases_per_edge): Likewise.
5113 (expand_case): Build list of simple_case_node.
5114 (expand_sjlj_dispatch_table): Use it.
5115 * tree-switch-conversion.c (struct case_node): Moved from
5116 stmt.c and adjusted.
5117 (emit_case_nodes): Likewise.
5118 (node_has_low_bound): Likewise.
5119 (node_has_high_bound): Likewise.
5120 (node_is_bounded): Likewise.
5121 (case_values_threshold): Likewise.
5122 (reset_out_edges_aux): Likewise.
5123 (compute_cases_per_edge): Likewise.
5124 (add_case_node): Likewise.
5125 (dump_case_nodes): Likewise.
5126 (balance_case_nodes): Likewise.
5127 (expand_switch_as_decision_tree_p): Likewise.
5128 (emit_jump): Likewise.
5129 (emit_case_decision_tree): Likewise.
5130 (try_switch_expansion): Likewise.
5131 (do_jump_if_equal): Likewise.
5132 (emit_cmp_and_jump_insns): Likewise.
5133 (fix_phi_operands_for_edge): New function.
5134 (record_phi_operand_mapping): Likewise.
5135 (class pass_lower_switch): New pass.
5136 (pass_lower_switch::execute): New function.
5137 (make_pass_lower_switch): Likewise.
5138 (conditional_probability):
5139 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
5140 * tree-pass.h: Add make_pass_lower_switch.
5142 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5145 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
5148 2017-08-29 Richard Biener <rguenther@suse.de>
5150 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
5151 we do not add a DW_AT_inline attribute twice.
5152 (gen_subprogram_die): Remove code setting DW_AT_inline on
5153 DECL_ABSTRACT_P nodes.
5155 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
5157 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
5158 calls to internal functions.
5159 (gimplify_modify_expr): Likewise.
5160 * tree-call-cdce.c (use_internal_fn): Likewise.
5161 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5162 (convert_to_divmod): Set the nothrow flag.
5163 * tree-if-conv.c (predicate_mem_writes): Likewise.
5164 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5165 (vectorizable_call): Likewise.
5166 (vectorizable_store): Likewise.
5167 (vectorizable_load): Likewise.
5168 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
5169 (vect_recog_mask_conversion_pattern): Likewise.
5171 2017-08-29 Martin Liska <mliska@suse.cz>
5174 * gcc.c (driver_handle_option): Add new argument.
5175 * opts-common.c (handle_option): Pass
5176 target_option_override_hook.
5177 * opts-global.c (lang_handle_option): Add new option.
5178 (set_default_handlers): Add new argument.
5179 (decode_options): Likewise.
5180 * opts.c (target_handle_option): Likewise.
5181 (common_handle_option): Call target_option_override_hook.
5182 * opts.h (struct cl_option_handler_func): Add hook for
5183 target option override.
5184 (struct cl_option_handlers): Likewise.
5185 (set_default_handlers): Add new argument.
5186 (decode_options): Likewise.
5187 (common_handle_option): Likewise.
5188 (target_handle_option): Likewise.
5189 * toplev.c (toplev::main): Pass targetm.target_option.override
5192 2017-08-29 Richard Biener <rguenther@suse.de>
5193 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5195 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
5196 life to the active subtree.
5198 2017-08-28 Jeff Law <law@redhat.com>
5200 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
5201 derive_equivalences.
5202 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
5204 (edge_info::derive_equivalences): New private member function
5206 * tree-ssa-dom.c (class edge_info): Changed from a struct
5207 to a class. Add ctor/dtor, methods and data members.
5208 (edge_info::edge_info): Renamed from allocate_edge_info.
5209 Initialize additional members.
5210 (edge_info::~edge_info): New.
5211 (free_dom_edge_info): Delete the edge info.
5212 (record_edge_info): Use new class & associated member functions.
5213 Tighten forms for testing for edge equivalences.
5214 (record_temporary_equivalences): Iterate over the simple
5215 equivalences rather than assuming there's only one per edge.
5216 (cprop_into_successor_phis): Iterate over the simple
5217 equivalences rather than assuming there's only one per edge.
5218 (optimize_stmt): Use operand_equal_p rather than pointer
5219 equality for mini-DSE code.
5221 2017-08-28 Nathan Sidwell <nathan@acm.org>
5223 * gcc.c (execute): Fold SIGPIPE handling into switch
5224 statement. Adjust internal error message.
5226 2017-08-28 Richard Biener <rguenther@suse.de>
5229 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
5230 Do nothing for removed DIEs.
5232 2017-08-28 Richard Biener <rguenther@suse.de>
5234 PR tree-optimization/81977
5235 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
5238 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
5241 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
5242 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
5243 using targetm.gen_mem_thread_fence.
5245 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
5248 * config/i386/i386.md (*<btsc><mode>): Change operand 2
5249 predicate to register_operand. Reorder operands.
5250 (*btr<mode>): Ditto.
5251 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
5252 (*btr<mode>_mask): Ditto.
5254 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
5256 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
5257 * config/rs6000/xmmintrin.h: New file.
5258 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
5260 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5263 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
5264 parameter and_insn and return it.
5265 (recombine_lvx_pattern): Insert a copy to ensure availability of
5266 the base register of the copied masking operation at the point of
5267 the instruction replacement.
5268 (recombine_stvx_pattern): Likewise.
5270 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5272 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
5273 undocumented switches.
5274 (-mpower9-dform-vector): Likewise.
5275 (-mpower9-dform): Likewise.
5276 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
5277 comments to delete references to -mpower9-dform* switches.
5278 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
5279 Delete reference to -mpower9-dform* switches, test for
5280 -mpower9-vector instead.
5281 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
5282 (OTHER_P9_VECTOR_MASKS): Likewise.
5283 (POWERPC_MASKS): Likewise.
5284 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
5285 tests against -mpower9-dform* to -mpower9-vector. Delete code
5286 that checked for -mpower9-dform* consistancy with other options.
5287 Add test for -mpower9-misc to enable other power9 switches.
5288 (rs6000_init_hard_regno_mode_ok): Likewise.
5289 (rs6000_option_override_internal): Likewise.
5290 (rs6000_emit_prologue): Likewise.
5291 (rs6000_emit_epilogue): Likewise.
5292 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
5293 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
5294 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
5296 (emit_fusion_p9_store): Likewise.
5297 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
5298 resetting these macros if the assembler does not support ISA 3.0
5300 (TARGET_P9_DFORM_VECTOR): Likewise.
5301 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
5302 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
5304 2017-08-25 Alan Modra <amodra@gmail.com>
5306 PR rtl-optimization/81747
5307 * cse.c (cse_extended_basic_block): Don't attempt to record
5308 equivalences for degenerate conditional branches that branch
5309 to their fall-through.
5311 2017-08-24 Martin Sebor <msebor@redhat.com>
5314 * gimple-fold.c (size_must_be_zero_p): New function.
5315 (gimple_fold_builtin_memory_op): Call it.
5317 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
5319 * config/rs6000/mm_malloc.h: New file.
5321 2017-08-24 Bin Cheng <bin.cheng@arm.com>
5323 PR tree-optimization/81913
5324 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
5325 analysis when either IVs in condition can wrap.
5327 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
5329 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
5330 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
5332 2017-08-24 Richard Biener <rguenther@suse.de>
5335 * targhooks.c (default_target_can_inline_p): Properly
5336 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
5337 is present and always compare.
5338 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
5339 infer -mfpmath=sse from TARGET_SSE_P.
5340 (ix86_can_inline_p): Properly use target_option_default_node when
5341 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
5343 2017-08-24 Richard Biener <rguenther@suse.de>
5346 * dwarf2out.c (output_die): Handle flag_generate_offload like
5348 (output_comp_unit): Likewise.
5349 (gen_array_type_die): Likewise.
5350 (dwarf2out_early_finish): Likewise.
5351 (note_variable_value_in_expr): Likewise.
5352 (dwarf2out_finish): Likewise. Adjust assert.
5353 * cgraphunit.c (symbol_table::compile): Move setting of
5354 flag_generate_offload earlier ...
5355 (symbol_table::finalize_compilation_unit): ... here, before
5356 early debug finalization.
5358 2017-08-24 Richard Biener <rguenther@suse.de>
5360 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
5361 and ipa-fnsummary.h.
5362 (ix86_can_inline_p): When ix86_fpmath flags do not match
5363 check whether the callee uses FP math at all.
5365 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
5368 * tree-ssanames.c (get_nonzero_bits): Use element_precision
5369 instead of TYPE_PRECISION.
5371 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
5372 Alan Hayward <alan.hayward@arm.com>
5373 David Sherwood <david.sherwood@arm.com>
5375 * combine.c (make_extraction): Use subreg_offset_from_lsb.
5377 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
5379 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
5381 (ix86_frame::stack_realign_allocate): New field.
5382 (struct machine_frame_state): Modify comments.
5383 (machine_frame_state::sp_realigned_fp_end): New field.
5384 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
5386 (sp_valid_at): Add assertion to assure no attempt to access invalid
5387 offset of a realigned stack.
5388 (fp_valid_at): Likewise.
5389 (choose_baseaddr): Modify comments.
5390 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
5391 ix86_expand_prologue.
5392 (ix86_expand_prologue): Modify stack realignment and allocation.
5393 (ix86_expand_epilogue): Modify comments.
5394 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5395 avx2_runtime, avx512f, and avx512f_runtime.
5397 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5399 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5400 (mstackrealign): Do not init to -1.
5401 * config/i386/i386.c (ix86_option_override_internal):
5402 Check opts_set, not opts when setting default value of
5403 opts->x_ix86_force_align_arg_pointer.
5405 2017-08-23 Richard Biener <rguenther@suse.de>
5407 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5408 lang_hooks.decl_printable_name.
5409 * print-rtl-function.c (print_rtx_function): Likewise.
5410 * tree-pretty-print.c (dump_function_header): Likewise.
5412 2017-08-23 Richard Biener <rguenther@suse.de>
5415 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5416 -g0 at compile-time.
5418 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5421 * doc/sourcebuild.texi (Other hardware attributes):
5424 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5427 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5428 Fix single-use check.
5430 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5432 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5434 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5436 * doc/install.texi: Modify to add more details on running selected
5439 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5441 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5442 is combined with -mabi=ms.
5443 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5446 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5448 PR tree-optimization/81488
5449 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5450 and cached_basis fields.
5451 (MAX_SPREAD): New constant.
5452 (alloc_cand_and_find_basis): Initialize new fields.
5453 (clear_visited): New function.
5454 (create_phi_basis_1): Rename from create_phi_basis, set visited
5455 and cached_basis fields.
5456 (create_phi_basis): New wrapper function.
5457 (phi_add_costs_1): Rename from phi_add_costs, add spread
5458 parameter, set visited field, short-circuit when limits reached.
5459 (phi_add_costs): New wrapper function.
5460 (record_phi_increments_1): Rename from record_phi_increments, set
5462 (record_phi_increments): New wrapper function.
5463 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5464 (phi_incr_cost): New wrapper function.
5465 (all_phi_incrs_profitable_1): Rename from
5466 all_phi_incrs_profitable, set visited field.
5467 (all_phi_incrs_profitable): New wrapper function.
5469 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5470 Alan Hayward <alan.hayward@arm.com>
5471 David Sherwood <david.sherwood@arm.com>
5473 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5474 that takes the outer and inner modes.
5475 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5476 comparison as the canonical test for a paradoxical subreg.
5477 * combine.c (simplify_set): Use paradoxical_subreg_p.
5478 (make_extraction): Likewise.
5479 (force_to_mode): Likewise.
5480 (rtx_equal_for_field_assignment_p): Likewise.
5481 (gen_lowpart_for_combine): Likewise.
5482 (simplify_comparison): Likewise.
5483 * cse.c (equiv_constant): Likewise.
5484 * expmed.c (store_bit_field_1): Likewise.
5485 * final.c (alter_subreg): Likewise.
5486 * fwprop.c (propagate_rtx): Likewise.
5487 (forward_propagate_subreg): Likewise.
5488 * ira-conflicts.c (ira_build_conflicts): Likewise.
5489 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5490 * lra-constraints.c (curr_insn_transform): Likewise.
5491 (split_reg): Likewise.
5492 * lra-eliminations.c (move_plus_up): Likewise.
5493 (lra_eliminate_regs_1): Likewise.
5494 * recog.c (general_operand): Likewise.
5495 * ree.c (combine_reaching_defs): Likewise.
5496 * reload.c (push_reload): Likewise.
5497 (find_reloads): Likewise.
5498 * reload1.c (elimination_effects): Likewise.
5499 (compute_reload_subreg_offset): Likewise.
5500 (choose_reload_regs): Likewise.
5501 * rtlanal.c (subreg_lsb_1): Likewise.
5502 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5503 (simplify_subreg): Likewise.
5504 * var-tracking.c (track_loc_p): Likewise.
5505 * emit-rtl.c (byte_lowpart_offset): Likewise.
5506 (paradoxical_subreg_p): Delete out-of-line definition.
5508 2017-08-22 Jeff Law <law@redhat.com>
5510 PR tree-optimization/81741
5511 PR tree-optimization/71947
5512 * tree-ssa-dom.c: Include tree-inline.h.
5513 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5514 equivalences if one is more expensive to compute than the other.
5515 * tree-ssa-scopedtables.h (class const_or_copies): Make
5516 record_const_or_copy_raw method private.
5517 (class avail_exprs_stack): New method simplify_binary_operation.
5518 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5519 avail_exprs_stack::simplify_binary_operation as needed.
5520 (avail_exprs_stack::simplify_binary_operation): New function.
5522 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5524 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5525 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5526 (DOT_SYMBOLS): Likewise.
5527 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5528 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5529 (RS6000_ABI_NAME): Likewise.
5530 (TARGET_CMODEL): Likewise.
5531 (TOC_SECTION_ASM_OP): Likewise.
5532 (SET_CMODEL): New macro.
5533 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5535 2017-08-22 Richard Biener <rguenther@suse.de>
5537 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5538 to allow for free-lang-data replacements similar to verify_type_variant.
5540 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5541 Alan Hayward <alan.hayward@arm.com>
5542 David Sherwood <david.sherwood@arm.com>
5544 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5545 VOIDmode for the LABEL_REF.
5547 2017-08-22 Richard Biener <rguenther@suse.de>
5549 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5551 2017-08-22 Richard Biener <rguenther@suse.de>
5553 * common.opt (feliminate-dwarf2-dups): Ignore.
5554 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5555 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5556 same_die_p_wrap, compute_section_prefix,
5557 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5558 (comdat_symbol_id, comdat_symbol_number): Likewise.
5559 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5561 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5562 (output_die): Mark unreachable path unreachable.
5563 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5564 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5565 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5566 (dwarf2out_early_finish): Likewise.
5568 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5570 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5572 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5575 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5576 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5577 (avr_attribute_table) <io, io_low, address>: Initialize
5578 .decl_required with true.
5580 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5582 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5583 undocumented debugging options.
5584 (-mvsx-scalar-double): Likewise.
5585 (-mallow-df-permute): Likewise.
5586 (-mvectorize-builtins): Likewise.
5587 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5588 (rs6000_builtin_vectorized_function): Likewise.
5589 (rs6000_builtin_md_vectorized_function): Likewise.
5590 (rs6000_opt_vars): Likewise.
5592 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5595 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5596 (*btrq_imm): Rename from *btrq.
5597 (*btcq_imm): Rename from *btcq.
5598 (btsc): New code attribute.
5599 (*<btsc><mode>): New insn pattern.
5600 (*btr<mode>): Ditto.
5601 (*<btsc><mode>_mask): New insn_and_split pattern.
5602 (*btr<mode>_mask): Ditto.
5604 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5605 Alan Hayward <alan.hayward@arm.com>
5606 David Sherwood <david.sherwood@arm.com>
5608 * function.c (pad_below): Simplify padding calculation.
5610 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5611 Alan Hayward <alan.hayward@arm.com>
5612 David Sherwood <david.sherwood@arm.com>
5614 * target.def (function_prologue): Remove frame size argument.
5615 (function_epilogue): Likewise.
5616 * doc/tm.texi: Regenerate.
5617 * final.c (final_start_function): Update call to function_prologue.
5618 (final_end_function): Update call to function_epilogue.
5619 (default_function_pro_epilogue): Remove frame size argument.
5620 * output.h (default_function_pro_epilogue): Likewise.
5621 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5622 (arm_output_function_prologue): Likewise.
5623 * config/frv/frv.c (frv_function_prologue): Likewise.
5624 (frv_function_epilogue): Likewise.
5625 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5626 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5627 (ia64_output_function_epilogue): Likewise.
5628 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5629 (m32r_output_function_epilogue): Likewise.
5630 * config/microblaze/microblaze.c (microblaze_function_prologue)
5631 (microblaze_function_epilogue): Likewise.
5632 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5633 (mips_output_function_epilogue): Likewise.
5634 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5635 (mmix_target_asm_function_epilogue): Likewise.
5636 * config/msp430/msp430.c (msp430_start_function): Likewise.
5637 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
5638 (nds32_asm_function_epilogue): Likewise.
5639 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
5640 * config/pa/pa.c (pa_output_function_prologue): Likewise.
5641 (pa_output_function_epilogue): Likewise.
5642 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
5643 (rs6000_output_function_epilogue): Likewise.
5644 * config/rl78/rl78.c (rl78_start_function): Likewise.
5645 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
5646 (rs6000_output_function_epilogue): Likewise.
5647 * config/rx/rx.c (rx_output_function_prologue): Likewise.
5648 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
5649 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
5650 (sparc_asm_function_epilogue): Likewise.
5652 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5654 * tree.h (type_has_mode_precision_p): New function.
5655 * convert.c (convert_to_integer_1): Use it.
5656 * expr.c (expand_expr_real_2): Likewise.
5657 (expand_expr_real_1): Likewise.
5658 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5659 * match.pd: Likewise.
5660 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5661 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5662 * tree-tailcall.c (process_assignment): Likewise.
5663 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5664 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5665 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5666 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5667 (vectorizable_assignment): Likewise.
5668 (vectorizable_shift): Likewise.
5669 (vectorizable_operation): Likewise.
5670 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5672 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5674 * match.pd: Add pow (C, x) simplification.
5676 2017-08-21 Richard Biener <rguenther@suse.de>
5678 PR tree-optimization/81900
5679 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5680 for blocks with abnormal predecessors.
5681 (compute_antic): Do not set visited flag prematurely.
5683 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5686 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5688 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5690 * stor-layout.h (vector_type_mode): Move to...
5691 * tree.h (vector_type_mode): ...here.
5692 * stor-layout.c (vector_type_mode): Move to...
5693 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5695 2017-08-21 Richard Biener <rguenther@suse.de>
5697 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5698 register_external_die hooks.
5699 (debug_false_tree_charstarstar_uhwistar): Declare.
5700 (debug_nothing_tree_charstar_uhwi): Likewise.
5701 * debug.c (do_nothing_debug_hooks): Adjust.
5702 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5703 (debug_nothing_tree_charstar_uhwi): Likewise.
5704 * dbxout.c (dbx_debug_hooks): Adjust.
5705 (xcoff_debug_hooks): Likewise.
5706 * sdbout.c (sdb_debug_hooks): Likewise.
5707 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5708 * dwarf2out.c (macinfo_label_base): New global.
5709 (dwarf2out_register_external_die): New function for the
5710 register_external_die hook.
5711 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5712 (dwarf2_debug_hooks): Use them.
5713 (dwarf2_lineno_debug_hooks): Adjust.
5714 (struct die_struct): Add with_offset flag.
5715 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5716 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5717 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5718 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5719 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5720 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5721 defining section names for the early LTO debug variants.
5722 (reset_indirect_string): New helper.
5723 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5724 (print_dw_val): Add support for offsetted symbol references.
5725 (get_ultimate_context): Split out from is_cxx.
5726 (is_cxx): Use get_ultimate_context.
5727 (is_fortran): Add decl overload.
5728 (compute_comp_unit_symbol): Split out worker from
5729 compute_section_prefix.
5730 (compute_section_prefix): Call compute_comp_unit_symbol and
5731 set comdat_type_p here.
5732 (output_die): Skip DIE symbol output for the LTO added one.
5733 Handle DIE symbol references with offset.
5734 (output_comp_unit): Guard section name mangling properly.
5735 For LTO debug sections emit a symbol at the section beginning
5736 which we use to refer to its DIEs.
5737 (add_abstract_origin_attribute): For DIEs registered via
5738 dwarf2out_register_external_die directly refer to the early
5739 DIE rather than indirectly through the shadow one we created.
5740 Remove obsolete call to dwarf2out_abstract_function for
5741 non-function/block origins.
5742 (gen_array_type_die): When generating early LTO debug do
5743 not emit DW_AT_string_length.
5744 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
5745 late when in LTO. As suggested place a gcc_unreachable for
5746 the DECL_ABSTRACT_P case.
5747 (gen_subprogram_die): Avoid another specification DIE
5748 for early built declarations/definitions for the late LTO case.
5749 (gen_variable_die): Add type references for late duplicated VLA dies
5751 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
5752 we have the abstract instance already.
5753 (process_scope_var): Adjust decl DIE contexts in LTO which
5754 first puts them in limbo.
5755 (gen_decl_die): Do not generate type DIEs late apart from
5756 types for VLAs or for decls we do not yet have a DIE. Do not
5757 call dwarf2out_abstract_function late.
5758 (dwarf2out_early_global_decl): Make sure to create DIEs
5759 for abstract instances of a decl first.
5760 (dwarf2out_late_global_decl): Adjust comment.
5761 (output_macinfo_op): With multiple macro sections use
5762 macinfo_label_base to distinguish labels.
5763 (output_macinfo): Likewise. Update macinfo_label_base.
5764 Pass in the line info label.
5765 (note_variable_value_in_expr): When generating LTO resolve
5766 all variable values here by generating DIEs as needed.
5767 (init_sections_and_labels): Add early LTO debug flag parameter
5768 and generate different sections and names if set. Add generation
5769 counter for the labels so we can have multiple of them.
5770 (reset_dies): Helper to allow DIEs to be output multiple times.
5771 (dwarf2out_finish): When outputting DIEs to the fat part of an
5772 LTO object first reset DIEs.
5773 (dwarf2out_early_finish): Output early DIEs when generating LTO.
5774 (modified_type_die): Check for decl_ultimate_origin being self
5776 (gen_type_die_with_usage): Likewise.
5777 (gen_typedef_die): Allow decl_ultimate_origin being self.
5778 (set_decl_abstract_flags): Remove.
5779 (set_block_abstract_flags): Likewise.
5780 (dwarf2out_abstract_function): Treat the early generated DIEs
5781 as the abstract copy and only add DW_AT_inline and
5782 DW_AT_artificial here and call set_decl_origin_self.
5783 If the DIE has an abstract origin don't do anything.
5784 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
5785 if we have none yet (Go fails to build one, PR78628).
5786 (variably_modified_type_p): Prevent endless recursion for Ada
5787 cyclic pointer types.
5788 * lto-streamer-in.c: Include debug.h.
5789 (dref_queue): New global.
5790 (lto_read_tree_1): Stream in DIE references.
5791 (lto_input_tree): Register DIE references.
5792 (input_function): Stream DECL_DEBUG_ARGS.
5793 * lto-streamer-out.c: Include debug.h.
5794 (lto_write_tree_1): Output DIE references.
5795 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
5796 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
5797 (output_function): Stream DECL_DEBUG_ARGS.
5798 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
5799 Stream DECL_ABSTRACT_ORIGIN.
5800 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
5801 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
5802 DECL_CONTEXT for file-scope decls.
5803 * lto-streamer.h (struct dref_entry): Declare.
5804 (dref_queue): Likewise.
5805 * cfgexpand.c (pass_expand::execute): Do not call the
5806 outlining_inline_function hook here.
5807 * lto-wrapper.c (debug_obj): New global.
5808 (tool_cleanup): Unlink it if required.
5809 (debug_objcopy): New function.
5810 (run_gcc): Handle early debug sections in the IL files by
5811 extracting them to separate files, partially linkin them and
5812 feeding the result back as result to the linker.
5813 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
5814 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
5815 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
5816 sections into a separate segment.
5817 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
5819 (darwin_asm_dwarf_section): Likewise.
5820 (darwin_asm_output_dwarf_offset): Likewise.
5821 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
5823 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5824 Alan Hayward <alan.hayward@arm.com>
5825 David Sherwood <david.sherwood@arm.com>
5827 * read-md.h (md_reader::record_potential_iterator_use): Replace
5828 pointer argument with an rtx and an index.
5829 * read-rtl.c (iterator_group::apply_iterator): Likewise.
5830 (apply_mode_iterator): Likewise.
5831 (apply_code_iterator): Likewise.
5832 (apply_int_iterator): Likewise.
5833 (apply_subst_iterator): Likewise.
5834 (record_iterator_use): Likewise.
5835 (record_attribute_use): Likewise.
5836 (md_reader::record_potential_iterator_use): Likewise. Update calls
5837 to record_iterator_use and apply_iterator.
5838 (iterator_use): Replace ptr with x and index.
5839 (attribute_use): Likewise.
5840 (apply_attribute_uses): Update calls to apply_iterator.
5841 (apply_iterators): Likewise. Update initialization of iterator_use.
5842 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
5843 and record_potential_iterator_use.
5844 (rtx_reader::read_rtx_operand): Likewise.
5846 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5847 Alan Hayward <alan.hayward@arm.com>
5848 David Sherwood <david.sherwood@arm.com>
5850 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
5853 2017-08-21 Richard Biener <rguenther@suse.de>
5856 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
5857 at struct end conservatively when comparing common bases.
5859 2017-08-21 Richard Biener <rguenther@suse.de>
5861 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
5862 (mem_ref_in_stmt): Remove.
5863 (determine_max_movement): Use ref index to get at the reference.
5864 (invariantness_dom_walker::before_dom_children): Deal with
5865 lim data already initialized.
5866 (gather_mem_refs_stmt): Initialize lim data and record ref index.
5868 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5870 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
5871 (TARGET_ISA_ROUND): Ditto.
5872 (TARGET_ROUND): Ditto.
5873 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
5874 * config/i386/i386.md: Ditto.
5875 * config/i386/sse.md: Ditto.
5876 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
5877 with OPTION_MASK_ISA_SSE4_1.
5879 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5882 * doc/extend.texi (x86 Built-in Functions): Correct the name of
5883 __builtin_ia32_lzcnt_u16.
5885 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5888 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
5889 (rs6000_set_current_function): Rewrite function to use it.
5891 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
5894 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
5895 and TYPE_WARN_IF_NOT_ALIGN.
5896 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
5897 (handle_warn_if_not_align): New.
5898 (place_union_field): Call handle_warn_if_not_align.
5899 (place_field): Call handle_warn_if_not_align.
5900 Copy TYPE_WARN_IF_NOT_ALIGN.
5901 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
5902 (layout_type): Likewise.
5903 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
5905 (tree_decl_common): Add warn_if_not_align.
5906 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
5907 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
5908 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
5909 (DECL_WARN_IF_NOT_ALIGN): Likewise.
5910 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
5911 * doc/extend.texi: Document warn_if_not_aligned attribute.
5912 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
5914 2017-08-17 Martin Liska <mliska@suse.cz>
5917 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
5918 (get_data_dependence): Use it as pointer type.
5919 (distribute_loop): Likewise.
5921 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5923 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
5924 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
5925 (p8_vmrgow_<mode>_direct): New define_insn.
5926 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
5927 handle endianness for vmrgew and vmrgow permute patterns.
5929 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5931 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
5932 * config/rs6000/rs6000-cpus.def: Remove comment.
5933 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
5934 (POWERPC_MASKS): Likewise.
5935 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
5936 use of TARGET_VSX_TIMODE.
5937 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
5938 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
5939 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
5940 (rs6000_option_override_internal): Remove dead code.
5941 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
5942 (rs6000_legitimize_reload_address): Likewise.
5943 (rs6000_legitimate_address_p): Likewise.
5944 (rs6000_opt_masks): Delete "vsx-timode".
5945 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
5946 from function comment.
5947 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
5948 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
5949 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
5951 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
5952 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
5953 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
5956 (define_peephole2): Likewise.
5958 2017-08-17 Martin Sebor <msebor@redhat.com>
5961 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
5962 past the end of an array.
5963 (test_pp_format): Add test cases.
5965 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
5967 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
5968 missing ECF_NOTHROW flags.
5970 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5973 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
5974 operands residing in integer registers.
5975 (*vsx_le_perm_load_<mode>): Likewise.
5976 (*vsx_le_perm_store_<mode>): Likewise.
5977 (define_peephole2): Add peepholes to optimize the above.
5979 2017-08-17 Marek Polacek <polacek@redhat.com>
5982 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5983 to mimic what shorten_compare did. Change the return type to bool.
5984 (fold_cond_expr_with_comparison): Update call to
5985 operand_equal_for_comparison_p.
5986 (fold_ternary_loc): Likewise.
5988 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5990 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5992 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
5994 2017-08-17 Richard Biener <rguenther@suse.de>
5996 * tree-ssa-structalias.c (solve_graph): When propagating
5997 to successors update the graphs succ edges and avoid duplicate work.
5999 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
6002 * config/i386/i386.c (ix86_option_override_internal): Save target
6003 specific options after ix86_stack_protector_guard_reg was changed.
6005 2017-08-17 Richard Biener <rguenther@suse.de>
6007 PR tree-optimization/81827
6008 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
6009 (new_var_info): Initialize it conservatively.
6010 (get_call_vi): Mark register vars.
6011 (new_scalar_tmp_constraint_exp): Likewise.
6012 (handle_rhs_call): Likewise.
6013 (handle_const_call): Likewise.
6014 (create_function_info_for): Likewise.
6015 (solve_constraints): Sort varinfos to separate register from
6016 non-register vars to pack points-to solution bitmaps during
6019 2017-08-17 Marek Polacek <polacek@redhat.com>
6021 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
6023 2017-08-17 Richard Biener <rguenther@suse.de>
6025 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
6026 to true when overflow is undefined and we saturated the result.
6028 2017-08-17 Alan Modra <amodra@gmail.com>
6031 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
6032 Don't use store multiple if only one reg needs saving.
6033 (interesting_frame_related_regno): New function.
6034 (rs6000_frame_related): Don't emit frame info for regs that
6036 (rs6000_emit_epilogue): Likewise.
6038 2017-08-16 Nathan Sidwell <nathan@acm.org>
6040 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
6041 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
6042 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
6043 * tree.c (free_lang_data_in_type): Use else-if chain. Always
6044 clear TYPE_LANG_1. Remove obsolete member-function stripping.
6045 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
6046 (verify_type): Adjust for TYPE_BINFO move.
6047 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
6048 process TYPE_BINFO directly.
6049 (hash_tree): Likewise.
6050 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6052 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6055 2017-08-16 David Malcolm <dmalcolm@redhat.com>
6057 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
6059 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
6062 * config/i386/i386.md (*anddi_1_btr): Change predicates of
6063 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
6064 Add ix86_binary_operator_ok to insn constraint.
6065 (*iordi_1_bts): Ditto.
6066 (*xordi_1_btc): Ditto.
6067 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
6068 Update corresponding peephole2 pattern.
6072 2017-08-16 Bin Cheng <bin.cheng@arm.com>
6074 PR tree-optimization/81832
6075 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
6076 copy loop header which has IFN_LOOP_DIST_ALIAS call.
6078 2017-08-16 Marek Polacek <polacek@redhat.com>
6081 * fold-const.c (fold_indirect_ref_1): Restore original behavior
6082 regarding size_zero_node.
6084 2017-08-16 Martin Liska <mliska@suse.cz>
6087 * config.gcc: Respect previously set extra_objs in case
6090 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
6092 PR tree-optimization/81835
6093 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
6094 the phi SSA_NAME. Check that the condition in a COND_EXPR does
6095 not depend on the phi.
6097 2017-08-16 Alan Modra <amodra@gmail.com>
6099 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
6102 2017-08-16 Alan Modra <amodra@gmail.com>
6104 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
6105 (save_reg_p): ..into this. Update all callers.
6106 (first_reg_to_save): Simplify.
6108 2017-08-16 Alan Modra <amodra@gmail.com>
6110 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
6113 2017-08-15 Joseph Myers <joseph@codesourcery.com>
6117 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
6118 constant count if that count is less than 32.
6120 2017-08-15 Nathan Sidwell <nathan@acm.org>
6122 * gcc.c (execute): Emit friendlier message if inferior is killed
6123 by an external cause.
6125 2017-08-15 Richard Biener <rguenther@suse.de>
6127 PR tree-optimization/81790
6128 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
6129 CONSTRUCTORs from simplifying and VN.
6131 2017-08-14 Martin Sebor <msebor@redhat.com>
6133 * builtin-attrs.def: Add comments.
6135 2017-08-14 Martin Sebor <msebor@redhat.com>
6138 * doc/extend.texi (attribute nonstring): Document new attribute.
6140 2017-08-14 Martin Sebor <msebor@redhat.com>
6143 * tree-diagnostic.c (default_tree_printer): Handle %G.
6144 * gimple-pretty-print.h (percent_G_format): Declare new function.
6145 * gimple-pretty-print.c (percent_G_format): Define.
6146 * tree-pretty-print.c (percent_K_format): Add argument.
6148 2017-08-14 Martin Sebor <msebor@redhat.com>
6150 PR translation/79998
6151 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
6152 Remove a stray space.
6154 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
6157 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
6158 (*iordi_1_bts): Ditto.
6159 (*xordi_1_btc): Ditto.
6161 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6164 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
6165 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6167 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
6168 quoted strings, and make more translator-friendly.
6169 (darwin_rs6000_override_options): Likewise.
6170 (rs6000_option_override_internal): Likewise.
6171 (rs6000_return_in_memory): Fix overlong line.
6172 (init_cmulative_args): Use quoted strings, and make more
6173 translator-friendly.
6174 (rs6000_pass_by_reference): Fix overlong line.
6175 (def_builtin): Use quoted strings.
6176 (altivec_expand_predicate_builtin): Use quoted strings, and make
6177 more translator-friendly.
6178 (htm_expand_builtin): Use quoted strings.
6179 (cpu_expand_builtin): Use quoted strings, and make more
6180 translator-friendly.
6181 (altivec_expand_builtin): Likewise.
6182 (paired_expand_predicate_builtin): Likewise.
6183 (rs6000_invalid_builtin): Likewise.
6184 (builtin_function_type): Use quoted strings.
6185 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
6186 more translator-friendly.
6187 (rs6000_trampoline_init): Likewise.
6188 (rs6000_handle_altivec_attribute): Likewise.
6189 (rs6000_inner_target_options): Use quoted strings.
6190 (rs6000_disable_incompatible_switches): Likewise.
6191 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
6192 strings, and make more translator-friendly.
6193 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
6195 2017-08-14 Bin Cheng <bin.cheng@arm.com>
6197 PR tree-optimization/81799
6198 * tree-loop-distribution.c (version_loop_by_alias_check): Force
6199 cond_expr to simple gimple operand.
6201 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
6204 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
6206 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
6210 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
6211 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
6212 TARGET_GASISR_PROLOGUES.
6213 * config/avr/avr.c (avr_option_override): Same.
6214 (avr_pass_pre_proep::execute): Same.
6216 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
6219 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
6220 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
6222 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
6224 * config/i386/i386.md (*load_tp_<mode>): Redefine as
6225 define_insn_and_split. Split to a memory load from 0 in
6226 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
6227 using PTR mode iterator.
6228 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
6229 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
6230 (*add_tp_<mode>): Redefine as define_insn_and_split.
6231 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
6232 address space. Merge with *add_tp_x32 using PTR mode iterator.
6233 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
6234 Split to an add with a memory load from 0 in
6235 DEFAULT_TLS_SEG_REG address space.
6237 2017-08-12 Andrew Pinski <apinski@cavium.com>
6239 * config/aarch64/aarch64-option-extensions.def (rdma):
6240 Fix feature string to what Linux prints out in /proc/cpuinfo.
6242 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
6245 * dwarf2out.c (modified_type_die): For C typedef types that have
6246 an ultimate origin, process the ultimate origin instead of the
6248 (gen_typedef_die): Assert that input DECLs have no ultimate
6250 (gen_type_die_with_usage): For typedef variants that have an
6251 ultimate origin, just call gen_decl_die on the original DECL.
6252 (process_scope_var): Avoid creating DIEs for local typedefs and
6253 concrete static variables.
6255 2017-08-12 Alan Modra <amodra@gmail.com>
6259 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
6260 match gnu-user.h startfile.
6261 (ENDFILE_LINUX_SPEC): Similarly.
6263 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
6266 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
6268 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
6270 2017-08-11 Tamar Christina <tamar.christina@arm.com>
6271 * config/aarch64/aarch64.md (mov<mode>): Change.
6272 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6273 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
6274 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
6276 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
6278 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
6279 for storage order barriers.
6281 2017-08-11 Martin Liska <mliska@suse.cz>
6284 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
6285 variables of void type.
6287 2017-08-11 Martin Liska <mliska@suse.cz>
6289 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
6290 TARGET_SUPPORTS_ALIASES.
6291 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
6292 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
6293 (optimize_weakref): Likewise.
6294 * symtab.c (symtab_node::noninterposable_alias): Likewise.
6295 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
6296 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
6298 2017-08-11 Martin Liska <mliska@suse.cz>
6301 * config/i386/i386.c (make_resolver_func): Do complete
6302 refactoring of the function.
6304 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6307 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
6308 * config/i386/i386.c (ix86_stack_protect_guard): Use
6309 ix86_stack_protect_guard_symbol_str to generate varible declaration.
6310 * doc/invoke.texi (x86 Options): Document
6311 -mstack-protector-guard-symbol= option.
6313 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6315 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
6316 * config/i386/i386.c (ix86_split_stack_guard): New function.
6317 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
6318 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
6319 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
6320 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
6321 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
6322 (split_stack_space_check): Call ix86_split_stack_guard.
6324 2017-08-10 Martin Sebor <msebor@redhat.com>
6326 * print-tree.c (print_node): Print location using the established
6328 Replace spaces with colons.
6329 (debug_raw, debug): Ditto.
6331 2017-08-10 Martin Sebor <msebor@redhat.com>
6334 * pretty-print.c (pp_format): Correct the handling of %s precision.
6336 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
6339 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6341 (ix86_finalize_stack_frame_flags): This. Also clear
6342 frame_pointer_needed if -fno-omit-frame-pointer is used without
6344 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6345 with ix86_finalize_stack_frame_flags.
6346 (ix86_expand_epilogue): Likewise.
6347 (ix86_expand_split_stack_prologue): Likewise.
6348 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
6350 2017-08-10 Martin Liska <mliska@suse.cz>
6353 * c-attribs.c (handle_target_attribute):
6354 Report warning for an empty string argument of target attribute.
6356 2017-08-09 Jakub Jelinek <jakub@redhat.com>
6359 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
6361 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
6362 or DECL_NONLOCAL labels.
6363 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
6364 or DECL_NONLOCAL labels here.
6366 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
6368 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
6369 to indicate when early gimple folding has been disabled.
6370 (rs6000_gimple_fold_builtin): Add debug content.
6371 (rs6000_invalid_builtin): Fix whitespace.
6372 (rs6000_expand_builtin): Fix whitespace.
6373 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
6375 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
6378 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
6379 SAVE_MULTIPLE if not all the registers that saves, should be saved.
6381 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
6383 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
6384 (qdf24xx): Likewise.
6385 * config/aarch64/aarch64.md: Include falkor.md.
6386 * config/aarch64/falkor.md: New.
6388 2017-08-09 Marek Polacek <polacek@redhat.com>
6391 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
6392 * diagnostic.c (emit_diagnostic): Add a comment.
6393 (emit_diagnostic_valist): New function.
6395 2017-08-09 Marek Polacek <polacek@redhat.com>
6398 * input.c (make_location): New overload.
6399 * input.h (make_location): Declare.
6401 2017-08-08 Alan Modra <amodra@gmail.com>
6402 H.J. Lu <hongjiu.lu@intel.com>
6405 * gcc.c (NO_PIE_SPEC): Delete.
6406 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6408 (LINK_PIE_SPEC): ..to here.
6409 (LINK_COMMAND_SPEC): Support -no-pie.
6410 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6411 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6413 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6414 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6415 (ENDFILE_CRTEND_SPEC): Similarly.
6417 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6420 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6421 (mstack-protector-guard-offset=): Ditto.
6422 * config/i386/i386.c (ix86_option_override): Handle
6423 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6425 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6426 ix86_stack_protect_guard_offset variables.
6427 (TARGET_STACK_PROTECT_GUARD): Always define.
6428 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6429 and -mstack-protector-guard-offset= options.
6431 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6433 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6434 boundary case for the last candidate.
6436 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6438 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6439 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6441 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6444 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6445 * config/aarch64/aarch64-builtins.c
6446 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6447 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6448 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6450 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6451 Andrew Pinski <pinskia@gmail.com>
6454 * internal-fn.def (XORSIGN): New.
6455 * optabs.def (xorsign_optab): New.
6456 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6457 (convert_expand_mult_copysign): New.
6458 (pass_optimize_widening_mul::execute): Call
6459 convert_expand_mult_copysign.
6461 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6463 PR tree-optimization/81354
6464 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6465 Insert on edges rather than explicitly creating landing pads.
6466 (analyze_candidates_and_replace): Commit edge inserts.
6468 2017-08-08 Richard Biener <rguenther@suse.de>
6471 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6472 (expand_simple_operations): Also look through ADDR_EXPRs with
6473 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6475 2017-08-08 Richard Biener <rguenther@suse.de>
6477 PR tree-optimization/81723
6478 * tree-vect-slp.c (struct bst_traits): New hash traits.
6479 (bst_fail): New global.
6480 (vect_build_slp_tree_2): New worker, split out from ...
6481 (vect_build_slp_tree): ... this now wrapping it with using
6482 bst_fail set to cache SLP tree build fails. Properly handle
6484 (vect_analyze_slp_instance): Allocate and free bst_fail.
6486 2017-08-08 Martin Liska <mliska@suse.cz>
6489 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6490 LABEL_DECLs that can be from a different function.
6492 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6494 PR tree-optimization/81744
6495 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6496 loop's number of iterations.
6498 2017-08-08 Martin Liska <mliska@suse.cz>
6500 * asan.c: Include header files.
6501 * attribs.c (build_decl_attribute_variant): New function moved
6503 (build_type_attribute_qual_variant): Likewise.
6504 (cmp_attrib_identifiers): Likewise.
6505 (simple_cst_list_equal): Likewise.
6506 (omp_declare_simd_clauses_equal): Likewise.
6507 (attribute_value_equal): Likewise.
6508 (comp_type_attributes): Likewise.
6509 (build_type_attribute_variant): Likewise.
6510 (lookup_ident_attribute): Likewise.
6511 (remove_attribute): Likewise.
6512 (merge_attributes): Likewise.
6513 (merge_type_attributes): Likewise.
6514 (merge_decl_attributes): Likewise.
6515 (merge_dllimport_decl_attributes): Likewise.
6516 (handle_dll_attribute): Likewise.
6517 (attribute_list_equal): Likewise.
6518 (attribute_list_contained): Likewise.
6519 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6520 (lookup_attribute_by_prefix): Likewise.
6521 * bb-reorder.c: Include header files.
6522 * builtins.c: Likewise.
6523 * calls.c: Likewise.
6524 * cfgexpand.c: Likewise.
6525 * cgraph.c: Likewise.
6526 * cgraphunit.c: Likewise.
6527 * convert.c: Likewise.
6528 * dwarf2out.c: Likewise.
6529 * final.c: Likewise.
6530 * fold-const.c: Likewise.
6531 * function.c: Likewise.
6532 * gimple-expr.c: Likewise.
6533 * gimple-fold.c: Likewise.
6534 * gimple-pretty-print.c: Likewise.
6535 * gimple.c: Likewise.
6536 * gimplify.c: Likewise.
6537 * hsa-common.c: Likewise.
6538 * hsa-gen.c: Likewise.
6539 * internal-fn.c: Likewise.
6540 * ipa-chkp.c: Likewise.
6541 * ipa-cp.c: Likewise.
6542 * ipa-devirt.c: Likewise.
6543 * ipa-fnsummary.c: Likewise.
6544 * ipa-inline.c: Likewise.
6545 * ipa-visibility.c: Likewise.
6547 * lto-cgraph.c: Likewise.
6548 * omp-expand.c: Likewise.
6549 * omp-general.c: Likewise.
6550 * omp-low.c: Likewise.
6551 * omp-offload.c: Likewise.
6552 * omp-simd-clone.c: Likewise.
6553 * opts-global.c: Likewise.
6554 * passes.c: Likewise.
6555 * predict.c: Likewise.
6556 * sancov.c: Likewise.
6557 * sanopt.c: Likewise.
6558 * symtab.c: Likewise.
6559 * toplev.c: Likewise.
6560 * trans-mem.c: Likewise.
6561 * tree-chkp.c: Likewise.
6562 * tree-eh.c: Likewise.
6563 * tree-into-ssa.c: Likewise.
6564 * tree-object-size.c: Likewise.
6565 * tree-parloops.c: Likewise.
6566 * tree-profile.c: Likewise.
6567 * tree-ssa-ccp.c: Likewise.
6568 * tree-ssa-live.c: Likewise.
6569 * tree-ssa-loop.c: Likewise.
6570 * tree-ssa-sccvn.c: Likewise.
6571 * tree-ssa-structalias.c: Likewise.
6572 * tree-ssa.c: Likewise.
6573 * tree-streamer-in.c: Likewise.
6574 * tree-vectorizer.c: Likewise.
6575 * tree-vrp.c: Likewise.
6577 * ubsan.c: Likewise.
6578 * varasm.c: Likewise.
6579 * varpool.c: Likewise.
6580 * tree.c: Remove functions moved to attribs.[ch].
6582 * config/aarch64/aarch64.c: Add attrs.h header file.
6583 * config/alpha/alpha.c: Likewise.
6584 * config/arc/arc.c: Likewise.
6585 * config/arm/arm.c: Likewise.
6586 * config/avr/avr.c: Likewise.
6587 * config/bfin/bfin.c: Likewise.
6588 * config/c6x/c6x.c: Likewise.
6589 * config/cr16/cr16.c: Likewise.
6590 * config/cris/cris.c: Likewise.
6591 * config/darwin.c: Likewise.
6592 * config/epiphany/epiphany.c: Likewise.
6593 * config/fr30/fr30.c: Likewise.
6594 * config/frv/frv.c: Likewise.
6595 * config/ft32/ft32.c: Likewise.
6596 * config/h8300/h8300.c: Likewise.
6597 * config/i386/winnt.c: Likewise.
6598 * config/ia64/ia64.c: Likewise.
6599 * config/iq2000/iq2000.c: Likewise.
6600 * config/lm32/lm32.c: Likewise.
6601 * config/m32c/m32c.c: Likewise.
6602 * config/m32r/m32r.c: Likewise.
6603 * config/m68k/m68k.c: Likewise.
6604 * config/mcore/mcore.c: Likewise.
6605 * config/microblaze/microblaze.c: Likewise.
6606 * config/mips/mips.c: Likewise.
6607 * config/mmix/mmix.c: Likewise.
6608 * config/mn10300/mn10300.c: Likewise.
6609 * config/moxie/moxie.c: Likewise.
6610 * config/msp430/msp430.c: Likewise.
6611 * config/nds32/nds32-isr.c: Likewise.
6612 * config/nds32/nds32.c: Likewise.
6613 * config/nios2/nios2.c: Likewise.
6614 * config/nvptx/nvptx.c: Likewise.
6615 * config/pa/pa.c: Likewise.
6616 * config/pdp11/pdp11.c: Likewise.
6617 * config/powerpcspe/powerpcspe.c: Likewise.
6618 * config/riscv/riscv.c: Likewise.
6619 * config/rl78/rl78.c: Likewise.
6620 * config/rx/rx.c: Likewise.
6621 * config/s390/s390.c: Likewise.
6622 * config/sh/sh.c: Likewise.
6623 * config/sol2.c: Likewise.
6624 * config/sparc/sparc.c: Likewise.
6625 * config/spu/spu.c: Likewise.
6626 * config/stormy16/stormy16.c: Likewise.
6627 * config/tilegx/tilegx.c: Likewise.
6628 * config/tilepro/tilepro.c: Likewise.
6629 * config/v850/v850.c: Likewise.
6630 * config/vax/vax.c: Likewise.
6631 * config/visium/visium.c: Likewise.
6632 * config/xtensa/xtensa.c: Likewise.
6634 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6637 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
6638 constraints since the -mupper-regs-* switches have been
6640 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
6641 into a vector from a double word element that was extracted from
6642 another vector, and eliminate extra XXPERMDI instructions.
6643 (vsx_concat_<mode>_2): Likewise.
6644 (vsx_concat_<mode>_3): Likewise.
6645 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
6646 concat to allow optimizing inserts from previous extracts.
6648 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
6650 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6651 memory reference to a SSP offset in TLS address space.
6652 (ix86_print_operand) <case '@'>: Remove.
6653 (ix86_print_operand_punct_valid_p): Remove '@' code.
6654 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6656 (stack_tls_protect_set_<mode>): Remove.
6657 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6658 (stack_tls_protect_test_<mode>): Remove.
6659 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6661 2017-08-07 Olivier Hainque <hainque@adacore.com>
6664 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6666 2017-08-07 Douglas Rupp <rupp@adacore.com>
6668 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6669 variable was referenced as multidir in command.
6671 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6674 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6677 2017-08-07 Martin Liska <mliska@suse.cz>
6679 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6680 * config/rl78/rl78.c: Add include of attribs.h.
6681 * config/sh/sh.c: Likewise.
6682 * config/v850/v850.c: Likewise.
6684 2017-08-07 Tom de Vries <tom@codesourcery.com>
6687 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6689 2017-08-07 Martin Liska <mliska@suse.cz>
6691 * config/mips/mips.c: Include attribs.h.
6693 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6696 * doc/invoke.texi: Document change in behvaior for -Ofast for
6699 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6701 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6703 (aarch64_pop_regs): Likewise.
6704 (aarch64_gen_load_pair): Likewise.
6705 (aarch64_save_callee_saves): Likewise.
6706 (aarch64_restore_callee_saves): Likewise.
6708 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6710 * config/i386/i386.c: Revert the last change.
6712 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6715 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6717 (ix86_finalize_stack_frame_flags): This. Also clear
6718 frame_pointer_needed if -fno-omit-frame-pointer is used without
6720 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6721 with ix86_finalize_stack_frame_flags.
6722 (ix86_expand_epilogue): Likewise.
6723 (ix86_expand_split_stack_prologue): Likewise.
6725 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6728 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6729 to P_AES for Westmere.
6731 2017-08-07 Jonathan Yong <10walls@gmail.com>
6733 * config/i386/mingw.opt (fset-stack-executable): Removed.
6734 * config/i386/cygming.opt (fset-stack-executable): Moved
6736 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
6738 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
6740 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
6742 2017-08-07 Marek Polacek <polacek@redhat.com>
6745 * fold-const.c (fold_indirect_ref_1): Check type_domain.
6747 2017-08-07 Martin Liska <mliska@suse.cz>
6749 * attribs.h (canonicalize_attr_name): New function.
6750 (cmp_attribs): Move from c-format.c and adjusted.
6751 (is_attribute_p): Moved from tree.h.
6752 * tree-inline.c: Add new includes.
6753 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
6754 (private_is_attribute_p): Remove.
6755 (private_lookup_attribute): Likewise.
6756 (private_lookup_attribute_by_prefix): Simplify.
6757 (remove_attribute): Use is_attribute_p.
6758 * tree.h: Remove removed declarations.
6760 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6763 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
6764 instead of computing it in the function. Formatting fix.
6765 (expand_case): Don't rely on default_edge being the first edge,
6766 clear it if removing it, pass default_edge to
6767 emit_case_dispatch_table.
6768 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
6771 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
6773 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
6774 insn in the function, emit NOP after the insn.
6776 2017-08-06 Tom de Vries <tom@codesourcery.com>
6778 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
6781 2017-08-06 Tom de Vries <tom@codesourcery.com>
6783 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
6786 2017-08-04 Yury Gribov <tetra2005@gmail.com>
6788 PR tree-optimization/57371
6789 * match.pd: New pattern.
6791 2017-08-04 Marek Polacek <polacek@redhat.com>
6794 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
6795 perform the computation in offset_int.
6797 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6799 PR tree-optimization/81136
6800 * tree-vectorizer.h: Include tree-hash-traits.h.
6801 (vec_base_alignments): New typedef.
6802 (vec_info): Add a base_alignments field.
6803 (vect_record_base_alignments): Declare.
6804 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
6806 (DR_IS_CONDITIONAL_IN_STMT): New macro.
6807 (create_data_ref): Add an is_conditional_in_stmt argument.
6808 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
6809 the is_conditional_in_stmt field.
6810 (data_ref_loc): Add an is_conditional_in_stmt field.
6811 (get_references_in_stmt): Set the is_conditional_in_stmt field.
6812 (find_data_references_in_stmt): Update call to create_data_ref.
6813 (graphite_find_data_references_in_stmt): Likewise.
6814 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
6815 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
6816 (vect_record_base_alignment): New function.
6817 (vect_record_base_alignments): Likewise.
6818 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
6819 for nested statements even if we fail to compute a misalignment.
6820 Use pooled base alignments for unconditional references.
6821 (vect_find_same_alignment_drs): Compare base addresses instead
6823 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
6824 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
6826 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6828 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
6829 Add an explicit name for the enum. Use auto_vec for slp_instances
6831 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
6833 (_bb_vec_info): Add a constructor and destructor.
6834 (vinfo_for_stmt): Return NULL for uids of -1 as well.
6835 (destroy_loop_vec_info): Delete.
6836 (vect_destroy_datarefs): Likewise.
6837 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
6838 (vec_info::vec_info): New function.
6839 (vec_info::~vec_info): Likewise.
6840 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
6841 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
6842 destroy_loop_vec_info.
6843 * tree-vect-loop.c (new_loop_vec_info): Replace with...
6844 (_loop_vec_info::_loop_vec_info): ...this.
6845 (destroy_loop_vec_info): Replace with...
6846 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
6847 the stmt_vec_infos. Leave handling of vec_info information to its
6848 destructor. Remove explicit vector releases.
6849 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
6850 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
6851 * tree-vect-slp.c (new_bb_vec_info): Replace with...
6852 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
6853 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
6854 (destroy_bb_vec_info): Replace with...
6855 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
6856 information to its destructor.
6857 (vect_slp_analyze_bb_1): Use new and delete instead of
6858 new_bb_vec_info and destroy_bb_vec_info.
6859 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
6862 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6864 * tree-data-ref.h (subscript): Add access_fn field.
6865 (data_dependence_relation): Add could_be_independent_p.
6866 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
6867 (same_access_functions): Move to tree-data-ref.c.
6868 * tree-data-ref.c (ref_contains_union_access_p): New function.
6869 (access_fn_component_p): Likewise.
6870 (access_fn_components_comparable_p): Likewise.
6871 (dr_analyze_indices): Add a reference to access_fn_component_p.
6872 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
6874 (constant_access_functions): Likewise.
6875 (add_other_self_distances): Likewise.
6876 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
6877 (initialize_data_dependence_relation): Use XCNEW and remove
6878 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
6879 of access functions that have the same type. Allow the
6880 subsequence to end with different bases in some circumstances.
6881 Record the chosen access functions in SUB_ACCESS_FN.
6882 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
6883 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
6884 (subscript_dependence_tester_1): Likewise dra and drb.
6885 (build_classic_dist_vector): Update calls accordingly.
6886 (subscript_dependence_tester): Likewise.
6887 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
6888 DDR_COULD_BE_INDEPENDENT_P.
6889 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
6890 comp_alias_ddrs instead of may_alias_ddrs.
6891 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6893 (vect_analyze_data_ref_dependence): Use it if
6894 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
6895 distance vectors if that fails.
6896 (dependence_distance_ge_vf): New function.
6897 (vect_prune_runtime_alias_test_list): Use it. Don't clear
6898 LOOP_VINFO_MAY_ALIAS_DDRS.
6900 2017-08-04 Richard Biener <rguenther@suse.de>
6903 * fold-const.c (fold_binary_loc): Properly restrict
6904 minus_var0 && minus_var1 case when associating undefined overflow
6907 2017-08-04 Tom de Vries <tom@codesourcery.com>
6909 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
6911 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6913 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6914 Don't start diagnostic messages with a capital letter.
6915 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6917 (rs6000_invalid_builtin): Likewise.
6918 (rs6000_trampoline_init): Likewise.
6920 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6923 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
6924 after setting changeable df flags.
6926 2017-08-03 Richard Biener <rguenther@suse.de>
6928 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
6929 up if the use is in USE - X.
6931 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
6933 * toplev.c (dumpfile.h): New include.
6934 (internal_error_reentered): New static function. Use it...
6935 (internal_error_function): ...here to handle reentered internal_error.
6937 2017-08-03 Richard Biener <rguenther@suse.de>
6940 * fold-const.c (split_tree): Add minus_var and minus_con
6941 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
6942 here but always use minus_*.
6943 (associate_trees): Assert we never associate with MINUS_EXPR
6944 and NULL first operand. Do not recurse for PLUS_EXPR operands
6945 when associating as MINUS_EXPR either.
6946 (fold_binary_loc): Track minus_var and minus_con.
6948 2017-08-03 Tom de Vries <tom@codesourcery.com>
6951 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
6952 ACCEL_COMPILER, apply finish_options on
6953 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
6955 2017-08-03 Tom de Vries <tom@codesourcery.com>
6958 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
6959 function_entry_patch_area_size > 0.
6961 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6964 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
6965 instead of 10??LU, perform unit multiplication in wide_int,
6966 don't change alloc_object_size_limit if the limit is larger
6969 PR tree-optimization/81655
6970 PR tree-optimization/81588
6971 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
6972 the case when ranges[i].low and high are 1 for unsigned type with
6976 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
6977 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6979 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6981 * tree-vrp.h: Add include guard.
6983 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6986 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6987 (ud2): New insn pattern.
6988 * config/i386/i386.c (ix86_expand_epilogue):
6989 For naked functions, generate ud2 instead of trap insn.
6991 2017-08-02 Marek Polacek <polacek@redhat.com>
6994 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
6996 2017-08-02 Tom de Vries <tom@codesourcery.com>
6997 Cesar Philippidis <cesar@codesourcery.com>
6999 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
7000 Add missing edge probabilities.
7002 2017-08-02 Tamar Christina <tamar.christina@arm.com>
7004 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
7007 2017-08-02 Jakub Jelinek <jakub@redhat.com>
7010 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
7011 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
7012 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
7013 prologue_seq sequences - if any.
7015 2017-08-02 Richard Biener <rguenther@suse.de>
7017 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
7018 via vectors if supported, integer extracts via punning if supported
7019 or otherwise vector extracts.
7021 2017-08-02 Richard Biener <rguenther@suse.de>
7023 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
7025 (bitmap_insert_into_set): ... this.
7027 2017-08-02 Richard Biener <rguenther@suse.de>
7029 PR tree-optimization/81633
7031 2015-08-17 Alan Hayward <alan.hayward@arm.com>
7033 PR tree-optimization/71752
7034 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
7036 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
7038 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
7039 (machine_function::call_ms2sysv_pad_out): Remove field.
7040 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
7041 (ix86_compute_frame_layout): Likewise.
7043 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
7046 * config/i386/i386.c (ix86_set_func_type): Disallow naked
7047 attribute with interrupt attribute.
7049 2017-08-01 Andrew Pinski <apinski@cavium.com>
7051 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
7052 BIT_INSERT_EXPR's operand 1
7053 to see if the types precision matches.
7055 2017-08-01 Martin Liska <mliska@suse.cz>
7058 * builtins.c (expand_builtin_memcpy_args): Remove.
7059 (expand_builtin_memcpy): Call newly added function
7060 expand_builtin_memory_copy_args.
7061 (expand_builtin_memcpy_with_bounds): Likewise.
7062 (expand_builtin_mempcpy): Remove last argument.
7063 (expand_builtin_mempcpy_with_bounds): Likewise.
7064 (expand_builtin_memory_copy_args): New function created from
7065 expand_builtin_mempcpy_args with small modifications.
7066 (expand_builtin_mempcpy_args): Remove.
7067 (expand_builtin_stpcpy): Remove unused argument.
7068 (expand_builtin): Likewise.
7069 (expand_builtin_with_bounds): Likewise.
7071 2017-08-01 Martin Liska <mliska@suse.cz>
7074 Make mempcpy more optimal (PR middle-end/70140).
7076 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7079 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7080 __builtin_vec_cmpne verify both arguments are compatible vectors
7081 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
7082 verify arg1_type is a pointer or array type. For __builtin_vec_st,
7083 move computation of aligned to after checking the argument types.
7087 * config/rs6000/vsx.md (vextract_fp_from_shorth,
7088 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
7091 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
7092 Doug Rupp <rupp@adacore.com>
7093 Olivier Hainque <hainque@adacore.com>
7095 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
7096 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
7098 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
7099 for TARGET_OS_CPP_BUILTIN.
7100 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
7101 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
7103 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
7104 passing required abi options to the assembler for EABI configurations.
7105 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
7106 of .text.hot and .text.unlikely sections for kernel modules when
7107 using ARM style exceptions.
7108 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
7109 options. Add EXTRA_CC1_SPEC.
7110 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
7112 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
7114 (ARM_TARGET2_DWARF_FORMAT): Define.
7115 * config/arm/t-vxworks: Adjust multilib control to removal of the
7116 Diab command line options.
7118 2017-08-01 Martin Liska <mliska@suse.cz>
7120 PR gcov-profile/81561
7121 * gcov.c (unblock): Make unblocking safe as we need to preserve
7122 index correspondence of blocks and block_lists.
7124 2017-08-01 Richard Biener <rguenther@suse.de>
7126 PR tree-optimization/81181
7127 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7128 (compute_antic): ... end of iteration here.
7130 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
7132 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
7133 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
7134 (ftree-slp-vectorize): Likewise.
7135 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
7136 can no longer be set independent of flag_tree_loop_vectorize.
7137 * omp-general.c (emp_max_vf): Likewise.
7138 * opts.c (enable_fdo_optimizations): Remove references to
7139 flag_tree_vectorize, these are now implicit.
7140 (common_handle_option): Remove handling for OPT_ftree_vectorize,
7141 and leave it for the options machinery.
7143 2017-08-01 Martin Liska <mliska@suse.cz>
7146 * builtins.c (expand_builtin_memcpy_args): Remove.
7147 (expand_builtin_memcpy): Call newly added function
7148 expand_builtin_memory_copy_args.
7149 (expand_builtin_memcpy_with_bounds): Likewise.
7150 (expand_builtin_mempcpy): Remove last argument.
7151 (expand_builtin_mempcpy_with_bounds): Likewise.
7152 (expand_builtin_memory_copy_args): New function created from
7153 expand_builtin_mempcpy_args with small modifications.
7154 (expand_builtin_mempcpy_args): Remove.
7155 (expand_builtin_stpcpy): Remove unused argument.
7156 (expand_builtin): Likewise.
7157 (expand_builtin_with_bounds): Likewise.
7159 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7162 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
7163 print "ds:" only for immediates in generic address space.
7165 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7168 * config/i386/i386.c (ix86_funciton_naked): New prototype.
7169 (ix86_function_ok_for_sibcall): Return false for naked functions.
7171 2017-08-01 Richard Biener <rguenther@suse.de>
7173 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
7174 (compute_antic): Seed worklist with exit block predecessors.
7175 * cfganal.c (dfs_find_deadend): For a cycle return the source
7176 of the edge closing it.
7178 2017-08-01 Tamar Christina <tamar.christina@arm.com>
7180 * config/aarch64/aarch64.c
7181 (aarch64_can_const_movi_rtx_p): Move 0 check.
7183 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7185 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
7186 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
7188 * match.pd: Ditto in address comparison pattern.
7190 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7192 PR tree-optimization/81627
7193 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
7194 closed ssa form for store-store chain.
7196 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7198 PR tree-optimization/81620
7199 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
7200 for store-store chain.
7202 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7204 PR tree-optimization/81588
7205 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
7206 ranges[i].in_p, invert comparison code ccode. For >/>=,
7207 swap rhs1 and rhs2 and comparison code unconditionally,
7208 for </<= don't do that. Don't swap rhs1/rhs2 again if
7209 ranges[i].in_p, instead invert comparison code ccode if
7210 opcode or oe->rank is BIT_IOR_EXPR.
7213 * optabs.def (vec_extract_optab, vec_init_optab): Change from
7214 a direct optab to conversion optab.
7215 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7216 with GET_MODE_INNER as last argument instead of optab_handler.
7217 * expmed.c (extract_bit_field_1): Likewise. Use vector from
7218 vector extraction if possible and optab is available.
7219 * expr.c (store_constructor): Use convert_optab_handler instead
7220 of optab_handler. Use vector initialization from smaller
7221 vectors if possible and optab is available.
7222 * tree-vect-stmts.c (vectorizable_load): Likewise.
7223 * doc/md.texi (vec_extract, vec_init): Document that the optabs
7225 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
7226 of vec_init from half-sized vectors with the same element mode.
7227 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
7228 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
7229 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
7230 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
7231 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
7232 after mode in gen_vec_extract* calls.
7233 (vec_extract<mode>): Renamed to ...
7234 (vec_extract<mode><ssescalarmodelower>): ... this.
7235 (vec_extract<mode><ssehalfvecmodelower>): New expander.
7236 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
7237 element mode after mode in gen_vec_init* calls.
7238 (VEC_INIT_HALF_MODE): New mode iterator.
7239 (vec_init<mode>): Renamed to ...
7240 (vec_init<mode><ssescalarmodelower>): ... this.
7241 (vec_init<mode><ssehalfvecmodelower>): New expander.
7242 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
7243 (vec_extractv2sfsf): ... this.
7244 (vec_initv2sf): Renamed to ...
7245 (vec_initv2sfsf): ... this.
7246 (vec_extractv2si): Renamed to ...
7247 (vec_extractv2sisi): ... this.
7248 (vec_initv2si): Renamed to ...
7249 (vec_initv2sisi): ... this.
7250 (vec_extractv4hi): Renamed to ...
7251 (vec_extractv4hihi): ... this.
7252 (vec_initv4hi): Renamed to ...
7253 (vec_initv4hihi): ... this.
7254 (vec_extractv8qi): Renamed to ...
7255 (vec_extractv8qiqi): ... this.
7256 (vec_initv8qi): Renamed to ...
7257 (vec_initv8qiqi): ... this.
7258 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
7259 (vec_init<mode>): Renamed to ...
7260 (vec_init<mode><VEC_base_l>): ... this.
7261 (vec_extract<mode>): Renamed to ...
7262 (vec_extract<mode><VEC_base_l>): ... this.
7263 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
7264 (vec_initv2sfsf): ... this.
7265 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
7266 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7267 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
7268 element mode after mode in gen_vec_init* calls.
7269 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
7270 (vec_init<mode><Vel>): ... this.
7271 (vec_extract<mode>): Renamed to ...
7272 (vec_extract<mode><Vel>): ... this.
7273 * config/aarch64/iterators.md (Vel): New mode attribute.
7274 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
7275 Add element mode after mode in gen_vec_extract* calls.
7276 * config/s390/vector.md (non_vec_l): New mode attribute.
7277 (vec_extract<mode>): Renamed to ...
7278 (vec_extract<mode><non_vec_l>): ... this.
7279 (vec_init<mode>): Renamed to ...
7280 (vec_init<mode><non_vec_l>): ... this.
7281 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
7282 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
7284 * config/arm/iterators.md (V_elem_l): New mode attribute.
7285 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
7286 (vec_extract<mode><V_elem_l>): ... this.
7287 (vec_extractv2di): Renamed to ...
7288 (vec_extractv2didi): ... this.
7289 (vec_init<mode>): Renamed to ...
7290 (vec_init<mode><V_elem_l>): ... this.
7291 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
7292 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
7293 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
7294 Add element mode after gen_vec_extract* calls.
7295 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
7296 (vec_init<mode><unitmode>): ... this.
7297 (vec_extract<mode>): Renamed to ...
7298 (vec_extract<mode><unitmode>): ... this.
7299 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
7300 (vec_init<mode><unitmode>): ... this.
7301 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
7302 (vec_initv2sfsf): ... this.
7303 (vec_extractv2sf): Renamed to ...
7304 (vec_extractv2sfsf): ... this.
7305 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
7306 Add element mode after gen_vec_extract* calls.
7307 * config/mips/mips.md (unitmode): New mode iterator.
7308 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
7309 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
7310 * config/spu/spu.md (inner_l): New mode attribute.
7311 (vec_init<mode>): Renamed to ...
7312 (vec_init<mode><inner_l>): ... this.
7313 (vec_extract<mode>): Renamed to ...
7314 (vec_extract<mode><inner_l>): ... this.
7315 * config/sparc/sparc.md (veltmode): New mode iterator.
7316 (vec_init<VMALL:mode>): Renamed to ...
7317 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
7318 * config/ia64/vect.md (vec_initv2si): Renamed to ...
7319 (vec_initv2sisi): ... this.
7320 (vec_initv2sf): Renamed to ...
7321 (vec_initv2sfsf): ... this.
7322 (vec_extractv2sf): Renamed to ...
7323 (vec_extractv2sfsf): ... this.
7324 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
7325 (vec_init<mode>): Renamed to ...
7326 (vec_init<mode><VEC_base_l>): ... this.
7327 (vec_extract<mode>): Renamed to ...
7328 (vec_extract<mode><VEC_base_l>): ... this.
7329 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
7330 (vec_initv2sfsf): ... this.
7331 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
7332 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7333 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
7334 gen_vec_init* calls.
7336 2017-08-01 Richard Biener <rguenther@suse.de>
7338 PR tree-optimization/81297
7339 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
7340 TREE_OVERFLOW from INTEGER_CSTs.
7342 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
7344 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
7346 2017-07-31 Carl Love <cel@us.ibm.com>
7348 * config/rs6000/rs6000-c: Add support for built-in functions
7349 vector signed char vec_xl_be (signed long long, signed char *);
7350 vector unsigned char vec_xl_be (signed long long, unsigned char *);
7351 vector signed int vec_xl_be (signed long long, signed int *);
7352 vector unsigned int vec_xl_be (signed long long, unsigned int *);
7353 vector signed long long vec_xl_be (signed long long, signed long long *);
7354 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
7355 vector signed short vec_xl_be (signed long long, signed short *);
7356 vector unsigned short vec_xl_be (signed long long, unsigned short *);
7357 vector double vec_xl_be (signed long long, double *);
7358 vector float vec_xl_be (signed long long, float *);
7359 * config/rs6000/altivec.h (vec_xl_be): Add #define.
7360 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
7361 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
7363 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
7364 (altivec_expand_builtin): Add switch statement to call
7365 altivec_expand_xl_be for each builtin.
7366 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
7367 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
7368 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
7369 __builtin_vsx_le_be_v16qi.
7370 * doc/extend.texi: Update the built-in documentation file for the
7371 new built-in functions.
7373 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7376 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
7378 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7380 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7382 * config.gcc: Add z14.
7383 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7384 CPU model numbers for z13s and z14.
7385 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
7387 * config/s390/s390-opts.h (enum processor_type): Rename
7388 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7389 * config/s390/s390.c (processor_table): Add field for CPU name to
7390 be passed to Binutils.
7391 (s390_asm_output_machine_for_arch): Use the new field in
7392 processor_table for Binutils.
7393 (s390_expand_builtin): Replace arch12 with z14.
7394 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7395 (s390_get_sched_attrmask): Likewise.
7396 (s390_get_unit_mask): Likewise.
7397 * config/s390/s390.opt: Add z14 to processor_type enum.
7399 2017-07-31 Martin Jambor <mjambor@suse.cz>
7402 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7403 regardless of optimization level.
7405 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7406 Martin Liska <mliska@suse.cz>
7408 * predict.def: Remove old comment and adjust probability.
7409 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7412 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7415 * config/i386/i386.c (ix86_function_naked): New function.
7416 (ix86_can_use_return_insn_p): Return false for naked functions.
7417 (ix86_expand_prologue): Skip prologue for naked functions.
7418 (ix86_expand_epilogue): Skip epilogue for naked functions
7419 and emit trap instruction.
7420 (ix86_warn_func_return): New function.
7421 (ix86_attribute_table): Add "naked" attribute specification.
7422 (TARGET_WARN_FUNC_RETURN): Define.
7423 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7425 2017-07-31 Martin Liska <mliska@suse.cz>
7427 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7428 (dump_gimple_bb_header): Always dump BB info.
7429 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7431 2017-07-31 Martin Liska <mliska@suse.cz>
7434 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7435 also with current_function_decl non-null equality.
7437 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7440 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7441 change type to the element type, instead add eltype variable and
7442 use it where we are interested in the element type.
7444 PR tree-optimization/81603
7445 * ipa-polymorphic-call.c
7446 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7447 offset arithmetic in offset_int, bail out if the resulting bit offset
7448 doesn't fit into shwi.
7450 2017-07-31 Martin Liska <mliska@suse.cz>
7452 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7453 (gimplify_save_expr): Fix comment.
7455 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7458 * config/i386/i386.c (ix86_function_arg): Update arguments for
7460 (ix86_compute_frame_layout): Set the initial stack offset to
7461 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7462 INCOMING_FRAME_SP_OFFSET.
7463 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7464 stack before exception handler returns.
7465 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7466 the 'ERROR_CODE' for exception handler.
7468 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7470 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7471 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7472 (ASM_OUTPUT_REG_POP): Ditto.
7473 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7474 instead of asm_fprintf to output pure string.
7476 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7478 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7479 to imported_module_or_decl hook.
7480 (debug_nothing_tree_tree_tree_bool): Remove.
7481 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7482 * debug.c (do_nothing_debug_hooks): Use
7483 debug_nothing_tree_tree_tree_bool_bool instead of
7484 debug_nothing_tree_tree_tree_bool.
7485 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7486 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7487 * sdbout.c (sdb_debug_hooks): Likewise.
7488 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7489 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7491 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7492 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7493 attribute, don't add anything.
7495 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7497 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7498 (fold_build2_stat_loc): Likewise.
7499 (fold_build3_stat_loc): Likewise.
7500 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7501 (fold_build1_loc): Remove macro.
7502 (fold_build2_loc): Likewise.
7503 (fold_build3_loc): Likewise.
7505 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7507 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7508 (gimple_build_debug_bind_source_stat): Likewise.
7509 * gimple.h (gimple_build_debug_bind): Remove macro.
7510 (gimple_build_debug_bind_source): Likewise.
7512 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7514 * bitmap.c (bitmap_alloc): Adjust.
7515 (bitmap_gc_alloc): Likewise.
7516 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7518 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7520 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7521 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7522 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7523 (bitmap_gc_alloc_stat): Likewise.
7524 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7526 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7528 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7529 * rtl.h (shallow_copy_rtx): Remove macro.
7531 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7533 * emit-rtl.c (gen_raw_REG): Adjust.
7534 * gengenrtl.c (gendef): Likewise.
7535 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7536 * rtl.h (rtx_alloc): Remove macro.
7538 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7540 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7541 (build_tree_list_stat): Likewise.
7542 * tree.h (build_tree_list): Remove macro.
7543 (build_tree_list_vec): Likewise.
7545 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7547 * tree.c (make_vector_stat): Remove _stat from name.
7548 (build_vector_stat): Likewise.
7549 * tree.h (make_vector_stat): Remove macro.
7550 (build_vector_stat): Likewise.
7552 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7554 * tree.h (build_var_debug_value): Remove prototype.
7556 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7558 * tree.c (tree_cons_stat): Remove _stat from name.
7559 * tree.h (tree_cons): Remove macro.
7561 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7563 * tree.c (build_vl_exp_stat): Remove _stat from name.
7564 * tree.h (build_vl_exp): Remove macro.
7566 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7568 * tree.c (build_decl_stat): Remove _stat from name.
7569 * tree.h (build_decl): Remove macro.
7571 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7573 * gimple.c (gimple_build_with_ops_stat): Adjust.
7574 (gimple_alloc_stat): Remove _stat from name.
7575 * gimple.h (gimple_alloc): Remove macro.
7577 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7579 * tree.c (make_tree_vec_stat): Remove _stat from name.
7580 (grow_tree_vec_stat): Likewise.
7581 * tree.h (make_tree_vec_stat): Adjust prototype.
7582 (grow_tree_vec_stat): Likewise.
7583 (make_tree_vec): Remove macro.
7584 (grow_tree_vec): Likewise.
7586 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7588 * fold-const.c (fold_build1_stat_loc): Adjust.
7589 (fold_build2_stat_loc): Likewise.
7590 (fold_build3_stat_loc): Likewise.
7591 * tree.c (build0_stat): Remove _stat from name.
7592 (build1_stat): Likewise.
7593 (build2_stat): Likewise.
7594 (build3_stat): Likewise.
7595 (build4_stat): Likewise.
7596 (build5_stat): Likewise.
7597 * tree.h (build1_loc): Remove macro, and rename _stat function
7599 (build2_loc): Likewise.
7600 (build3_loc): Likewise.
7601 (build4_loc): Likewise.
7602 (build5_loc): Likewise.
7604 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7606 * tree.c (make_int_cst_stat): Remove _stat from name.
7607 * tree.h (make_int_cst_stat): Adjust prototype.
7608 (make_int_cst): Remove macro.
7610 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7612 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7613 * tree.h (make_tree_binfo_stat): Adjust prototype.
7614 (make_tree_binfo): Remove.
7616 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7618 * tree.c (copy_node_stat): Rename to copy_node.
7619 (build_distinct_type_copy): Adjust.
7620 * tree.h (copy_node_stat): Adjust prototype.
7621 (copy_node): Remove macro.
7623 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7625 * tree.c (make_node_stat): rename to make_node.
7626 (build_tree_list_stat): Adjust.
7627 (build0_stat): Likewise.
7628 (build2_stat): Likewise.
7629 (build3_stat): Likewise.
7630 (build4_stat): Likewise.
7631 (build5_stat): Likewise.
7632 (build_decl_stat): Likewise.
7633 * tree.h (make_node_stat): Adjust prototype.
7634 (make_node): remove macro.
7636 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
7638 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
7639 (PPC_FEATURE2_SCV): Likewise.
7640 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
7642 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7644 * config/aarch64/aarch64.c
7645 (aarch64_internal_mov_immediate): Add new special pattern.
7646 * config/aarch64/aarch64.md (*movdi_aarch64):
7647 Add reg/32bit const mov case.
7649 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7650 Richard Sandiford <richard.sandiford@linaro.org>
7652 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7653 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7654 Add integer and movi cases.
7655 (movi-split-hf-df-sf split, fp16): New.
7656 (enabled): Added TARGET_FP_F16INST.
7657 * config/aarch64/iterators.md (GPF_HF): New.
7658 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7660 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7662 * config/aarch64/aarch64.c
7663 (aarch64_simd_container_mode): Add prototype.
7664 (aarch64_expand_mov_immediate): Add HI support.
7665 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7666 (aarch64_can_const_movi_rtx_p): New.
7667 (aarch64_preferred_reload_class):
7668 Remove restrictions of using FP registers for certain SIMD operations.
7669 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7670 (aarch64_valid_floating_const): Add integer move validation.
7671 (aarch64_simd_imm_scalar_p): Remove.
7672 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7673 (aarch64_legitimate_constant_p): Expand list of supported cases.
7674 * config/aarch64/aarch64-protos.h
7675 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7676 (aarch64_reinterpret_float_as_int): New.
7677 (aarch64_simd_imm_scalar_p): Remove.
7678 * config/aarch64/constraints.md (Uvi): New.
7679 (Dd): Split into Ds and new Dd.
7680 * config/aarch64/aarch64.md (*movsi_aarch64):
7682 (*movdi_aarch64): Add SIMD mov case.
7684 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7686 * tree-predcom.c: (struct chain): Handle store-store chain in which
7687 stores for elimination only store loop invariant values.
7688 (execute_pred_commoning_chain): Ditto.
7689 (prepare_initializers_chain_store_elim): Ditto.
7690 (prepare_finalizers): Ditto.
7691 (is_inv_store_elimination_chain): New function.
7692 (initialize_root_vars_store_elim_1): New function.
7694 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7696 * tree-predcom.c: Revise general description of the pass.
7697 (enum chain_type): New enum type for store elimination.
7698 (struct chain): New field supporting store elimination.
7699 (struct component): Ditto.
7700 (dump_chain): Dump store-stores chain.
7701 (release_chain): Release resources.
7702 (split_data_refs_to_components): Compute and create component
7703 contains only stores for elimination.
7704 (get_chain_last_ref_at): New function.
7705 (make_invariant_chain): Initialization.
7706 (make_rooted_chain): Specify chain type in parameter and record it.
7707 (add_looparound_copies): Skip for store-stores chain.
7708 (determine_roots_comp): Compute type of chain and pass it to
7710 (initialize_root_vars_store_elim_2): New function.
7711 (finalize_eliminated_stores): New function.
7712 (remove_stmt): Handle store for elimination.
7713 (execute_pred_commoning_chain): Execute predictive commoning on
7715 (determine_unroll_factor): Skip unroll for store-stores chain.
7716 (prepare_initializers_chain_store_elim): New function.
7717 (prepare_initializers_chain): Hanlde store-store chain.
7718 (prepare_finalizers_chain, prepare_finalizers): New function.
7719 (tree_predictive_commoning_loop): Return integer value indicating
7720 if loop is unrolled or lcssa form is corrupted.
7721 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7723 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7725 * tree-predcom.c (initialize_root): Delete.
7726 (execute_pred_commoning_chain): Initialize root vars and replace
7727 reference of non-combined chain directly, rather than call above
7730 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7732 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7733 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7735 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7737 * tree-predcom.c (struct chain): New field init_seq.
7738 (release_chain): Release init_seq.
7739 (prepare_initializers_chain): Record intialization stmts in above
7741 (insert_init_seqs): New function.
7742 (tree_predictive_commoning_loop): Call insert_init_seqs.
7744 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7746 * tree-predcom.c (determine_roots_comp): Skip trivial components.
7748 2017-07-28 Richard Biener <rguenther@suse.de>
7750 * match.pd: Remove superfluous :c.
7751 * genmatch.c (simplify::id): Add member.
7752 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
7754 (current_id): New global.
7755 (dt_node::parent): Move from ...
7756 (dt_operand::parent): ... here. Add for_id member.
7757 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
7758 (decision_tree::find_node): Relax order requirement when
7759 merging DT_TRUE nodes to ones inbetween the current simplify
7760 and the one we try to merge with. Add diagnostic whenever
7761 we need to enforce pattern order by not merging.
7762 (decision_tree::insert): Set current_id.
7763 (decision_tree::print_node): Dump parent node and for_id.
7764 (parser::last_id): Add member.
7765 (parser::push_simplify): Assign unique id.
7766 (parser::parser): Initialize last_id.
7768 2017-07-28 Martin Liska <mliska@suse.cz>
7771 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
7772 gimple_build_debug_bind.
7774 2017-07-28 Richard Biener <rguenther@suse.de>
7776 PR tree-optimization/81502
7777 * match.pd: Add pattern combining BIT_INSERT_EXPR with
7779 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
7781 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
7782 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
7783 for BIT_FIELD_REF args.
7784 * fold-const.c (make_bit_field_ref): Likewise.
7785 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7787 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7790 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
7791 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
7792 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
7793 Or it into SANITIZER_UNDEFINED.
7794 * ubsan.c: Include gimple-fold.h and varasm.h.
7795 (ubsan_expand_ptr_ifn): New function.
7796 (instrument_pointer_overflow): New function.
7797 (maybe_instrument_pointer_overflow): New function.
7798 (instrument_object_size): Formatting fix.
7799 (pass_ubsan::execute): Call instrument_pointer_overflow
7800 and maybe_instrument_pointer_overflow.
7801 * internal-fn.c (expand_UBSAN_PTR): New function.
7802 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
7803 * sanitizer.def (__ubsan_handle_pointer_overflow,
7804 __ubsan_handle_pointer_overflow_abort): New builtins.
7805 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
7806 * internal-fn.def (UBSAN_PTR): New internal function.
7807 * opts.c (sanitizer_opts): Add pointer-overflow.
7808 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
7809 * fold-const.c (build_range_check): Compute pointer range check in
7810 integral type if pointer arithmetics would be needed. Formatting
7813 2017-07-28 Martin Liska <mliska@suse.cz>
7816 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
7817 parameters that are of a variable-length.
7819 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
7821 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
7823 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
7824 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
7825 (CRT_CALL_STATIC_FUNCTION): Likewise.
7826 (ASM_DEFAULT_SPEC): New define.
7827 (ASM_SPEC32): Likewise.
7828 (ASM_SPEC64): Likewise.
7829 (ASM_SPEC_COMMON): Likewise.
7830 (ASM_SPEC): Likewise.
7831 (INVALID_64BIT): Likewise.
7832 (LINK_OS_DEFAULT_SPEC): Likewise.
7833 (LINK_OS_SPEC32): Likewise.
7834 (LINK_OS_SPEC64): Likewise.
7835 (POWERPC_LINUX): Likewise.
7836 (PTRDIFF_TYPE): Likewise.
7837 (RESTORE_FP_PREFIX): Likewise.
7838 (RESTORE_FP_SUFFIX): Likewise.
7839 (SAVE_FP_PREFIX): Likewise.
7840 (SAVE_FP_SUFFIX): Likewise.
7841 (SIZE_TYPE): Likewise.
7842 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
7843 (TARGET_64BIT): Likewise.
7844 (TARGET_64BIT): Likewise.
7845 (TARGET_AIX): Likewise.
7846 (WCHAR_TYPE_SIZE): Likewise.
7847 (WCHAR_TYPE): Undefine.
7848 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
7849 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
7850 (CPP_OS_RTEMS_SPEC): Delete.
7851 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
7852 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
7854 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
7856 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7858 PR tree-optimization/81578
7859 * tree-parloops.c (build_new_reduction): Bail out if
7860 reduction_code isn't one of the standard OpenMP reductions.
7861 Move the details printing after that decision.
7863 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7865 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
7866 related to reload_in_progress.
7867 (splat_input_operand): Likewise.
7868 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
7870 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
7872 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
7873 (TARGET_INSTANTIATE_DECLS): Likewise.
7874 (legitimate_indexed_address_p): Delete reload_in_progress code.
7875 (rs6000_debug_legitimate_address_p): Likewise.
7876 (rs6000_eliminate_indexed_memrefs): Likewise.
7877 (rs6000_emit_le_vsx_store): Likewise.
7878 (rs6000_emit_move_si_sf_subreg): Likewise.
7879 (rs6000_emit_move): Likewise.
7880 (register_to_reg_type): Likewise.
7881 (rs6000_pre_atomic_barrier): Likewise.
7882 (rs6000_machopic_legitimize_pic_address): Likewise.
7883 (rs6000_allocate_stack_temp): Likewise.
7884 (rs6000_address_for_fpconvert): Likewise.
7885 (rs6000_address_for_altivec): Likewise.
7886 (rs6000_secondary_memory_needed_rtx): Delete function.
7887 (rs6000_check_sdmode): Likewise.
7888 (rs6000_alloc_sdmode_stack_slot): Likewise.
7889 (rs6000_instantiate_decls): Likewise.
7890 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
7891 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
7892 Delete reload_in_progress.
7893 (*vec_reload_and_plus_<mptrsize>): Likewise.
7894 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
7895 (vsx_div_v2di): Likewise.
7896 (vsx_udiv_v2di): Likewise.
7898 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7900 * config/rs6000/rs6000.opt (mlra): Replace with stub.
7901 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
7902 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
7903 (rs6000_debug_reg_global): Delete print of LRA status.
7904 (rs6000_option_override_internal): Delete dead LRA related code.
7905 (rs6000_lra_p): Delete function.
7906 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
7908 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
7910 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
7911 * config/riscv/rtems.h: New file.
7913 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7914 Sudakshina Das <sudi.das@arm.com>
7916 * config/aarch64/aarch64.md
7917 (define_split for and<mode>3nr_compare): Move
7918 non aarch64_logical_operand to a register.
7919 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
7920 register immediate operand to a register.
7921 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
7923 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7926 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
7928 2017-07-27 Richard Biener <rguenther@suse.de>
7930 PR tree-optimization/81573
7931 PR tree-optimization/81494
7932 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
7933 multi defuse cycle case.
7935 2017-07-27 Richard Biener <rguenther@suse.de>
7937 PR tree-optimization/81571
7938 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
7941 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
7943 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
7944 earlier and only if MASK_FPU is set. Adjust formatting.
7946 2017-07-27 Martin Liska <mliska@suse.cz>
7948 * opt-functions.awk: Add validation of value of Init.
7949 * optc-gen.awk: Pass new argument.
7951 2017-07-27 Martin Liska <mliska@suse.cz>
7953 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
7954 Fix wrong condition.
7956 2017-07-27 Martin Liska <mliska@suse.cz>
7958 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
7959 BBs and edges seen by autoFDO.
7961 2017-07-27 Richard Biener <rguenther@suse.de>
7963 PR tree-optimization/81502
7964 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
7965 with incompatible but same sized type.
7966 (execute_update_addresses_taken): Likewise.
7968 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
7970 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
7971 flag_tree_loop_vectorize rather than flag_tree_vectorize.
7973 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7976 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
7977 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
7978 Change s_operand to memory_operand.
7980 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7982 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7983 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7984 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7985 Emit instructions rather than returning an expression. Handle TFmode
7986 and KFmode by casting to TImode.
7987 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7988 (rs6000_emit_le_vsx_store): Likewise.
7989 * config/rs6000/vsx.md (VSX_TI): New iterator.
7990 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7991 (*vsx_le_undo_permute_<mode>): Likewise.
7992 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
7993 emit the split sequence.
7994 (*vsx_le_perm_store_<mode>): Likewise.
7996 2017-07-27 Jakub Jelinek <jakub@redhat.com>
7998 PR tree-optimization/81555
7999 PR tree-optimization/81556
8000 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
8001 if true, force CHANGED for the recursive invocation.
8002 (reassociate_bb): Remember original length of ops array, pass
8003 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
8005 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
8006 attributes for noipa attribute. For naked attribute use
8007 lookup_attribute first before lookup_attribute_spec.
8008 * final.c (rest_of_handle_final): Disable IPA RA for functions with
8010 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
8011 for functions with noipa attribute.
8012 (cgraph_externally_visible_p): Return true for functions with noipa
8014 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
8015 for functions with noipa attribute.
8016 * doc/extend.texi: Document noipa function attribute.
8017 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
8018 also for functions with noipa attribute.
8019 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
8021 2017-07-26 Andrew Pinski <apinski@cavium.com>
8023 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
8024 vec_unalign_load_cost and vec_unalign_store_cost.
8026 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8028 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
8029 -mvsx-small-integer option.
8030 (ISA_3_0_MASKS_IEEE): Likewise.
8031 (OTHER_VSX_VECTOR_MASKS): Likewise.
8032 (POWERPC_MASKS): Likewise.
8033 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
8034 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
8035 code, only testing for DImode being allowed in non-VSX floating
8037 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
8038 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
8040 (rs6000_option_override_internal): Delete -mvsx-small-integer.
8041 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
8042 TARGET_P8_VECTOR test.
8043 (rs6000_secondary_reload_simple_move): Likewise.
8044 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
8045 since TARGET_P9_VECTOR was already tested.
8046 (rs6000_opt_masks): Remove -mvsx-small-integer.
8047 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
8048 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8050 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
8051 test for TARGET_VEXTRACTUB was used, and that uses
8053 (p9 extract splitter): Likewise.
8054 (vsx_extract_<mode>_di_p9): Likewise.
8055 (vsx_extract_<mode>_store_p9): Likewise.
8056 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
8057 for TARGET_P9_VECTOR was used. Delete code that is now dead with
8058 the elimination of TARGET_VSX_SMALL_INTEGER.
8059 (vsx_extract_<mode>_p8): Likewise.
8060 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
8061 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
8062 (vsx_set_<mode>_p9): Likewise.
8063 (vsx_set_v4sf_p9): Likewise.
8064 (vsx_set_v4sf_p9_zero): Likewise.
8065 (vsx_insert_extract_v4sf_p9): Likewise.
8066 (vsx_insert_extract_v4sf_p9_2): Likewise.
8067 * config/rs6000/rs6000.md (sign extend splitter): Change
8068 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
8069 (floatsi<mode>2_lfiwax_mem): Likewise.
8070 (floatunssi<mode>2_lfiwzx_mem): Likewise.
8071 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
8072 since a test for TARGET_P9_VECTOR was used.
8073 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8074 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
8075 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8076 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8077 TARGET_P8_VECTOR test.
8078 (fix_trunc<mode>si2_stfiwx): Likewise.
8079 (fix_trunc<mode>si2_internal): Likewise.
8080 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
8081 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8083 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8084 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
8085 TARGET_P8_VECTOR test.
8086 (fixuns_trunc<mode>si2_stfiwx): Likewise.
8087 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
8088 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
8090 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
8091 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
8092 since a test for TARGET_P9_VECTOR was used.
8093 (splitter for loading small constants): Likewise.
8095 2017-07-26 Andrew Pinski <apinski@cavium.com>
8097 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
8100 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
8103 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
8104 (fp_valid_at): Likewise.
8106 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8108 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
8109 (qdf24xx_addrcost_table): Likewise.
8110 (cortexa57_tunings): Update to use generic_branch_cost.
8111 (cortexa72_tunings): Likewise.
8112 (cortexa73_tunings): Likewise.
8113 (qdf24xx_tunings): Likewise.
8115 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8117 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
8118 (thunderx2t99_branch_cost): Likewise.
8119 (cortexa35_tunings): Update to use generic_branch_cost.
8120 (cortexa53_tunings): Likewise.
8121 (cortexa57_tunings): Likewise.
8122 (cortexa72_tunings): Likewise.
8123 (cortexa73_tunings): Likewise.
8124 (thunderx2t99_tunings): Likewise.
8126 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8128 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
8129 (sparc_option_override): Honour MASK_FSMULD.
8130 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
8131 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
8132 * config/sparc/sparc.opt (mfsmuld): New option.
8133 * doc/invoke.texi (mfsmuld): Document option.
8135 2017-07-26 Marek Polacek <polacek@redhat.com>
8138 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
8140 2017-07-26 Richard Biener <rguenther@suse.de>
8142 * gimple-match-head.c (do_valueize): Return OP if valueize
8144 (get_def): New helper to get at the def stmt of a SSA name
8146 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
8147 do_valueize to get at the def stmt.
8148 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
8150 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
8153 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
8155 2017-07-26 Martin Liska <mliska@suse.cz>
8158 * function.c (expand_function_start): Make expansion of
8159 nonlocal_goto_save_area after parm_birth_insn.
8161 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8163 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
8164 from all CPU target flags enable members.
8166 2017-07-26 Richard Biener <rguenther@suse.de>
8168 * genmatch.c (dt_simplify::gen): Make iterator vars const.
8169 (decision_tree::gen): Make 'type' const.
8170 (write_predicate): Likewise.
8172 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8174 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
8175 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
8176 (rs6000_option_override_internal): Likewise.
8177 (rs6000_expand_vector_set): Likewise.
8178 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
8179 (TARGET_UPPER_REGS_SF): Likewise.
8180 (TARGET_UPPER_REGS_DI): Likewise.
8181 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
8182 (TARGET_DIRECT_MOVE_64BIT): Likewise.
8183 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
8184 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8185 (Splitters for DI constants in Altivec registers): Likewise.
8186 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
8187 (vsx_set_v4sf_p9): Likewise.
8188 (vsx_set_v4sf_p9_zero): Likewise.
8189 (vsx_insert_extract_v4sf_p9): Likewise.
8190 (vsx_insert_extract_v4sf_p9_2): Likewise.
8192 2017-07-25 Carl Love <cel@us.ibm.com>
8194 * doc/extend.texi: Update the built-in documentation file for the
8195 existing built-in functions
8196 vector signed char vec_cnttz (vector signed char);
8197 vector unsigned char vec_cnttz (vector unsigned char);
8198 vector signed short vec_cnttz (vector signed short);
8199 vector unsigned short vec_cnttz (vector unsigned short);
8200 vector signed int vec_cnttz (vector signed int);
8201 vector unsigned int vec_cnttz (vector unsigned int);
8202 vector signed long long vec_cnttz (vector signed long long);
8203 vector unsigned long long vec_cnttz (vector unsigned long long);
8205 2017-07-25 Andrew Pinski <apinski@cavium.com>
8207 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
8208 accesses where the use is for the first operand of a BIT_INSERT.
8210 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8213 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8214 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8216 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8218 * config/i386/gstabs.h: Delete.
8219 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
8221 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
8223 * config/i386/i386.c (ix86_decompose_address): Do not check for
8224 register RTX when looking at index_reg or base_reg.
8225 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
8227 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
8229 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
8230 to update EH info here.
8232 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8234 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
8236 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8238 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
8240 2017-07-25 Torsten Duwe <duwe@suse.de>
8242 * common.opt: Introduce -fpatchable-function-entry
8243 command line option, and its variables function_entry_patch_area_size
8244 and function_entry_patch_area_start.
8245 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
8246 including a two-value parser.
8247 * target.def (print_patchable_function_entry): New target hook.
8248 * targhooks.h (default_print_patchable_function_entry): New function.
8249 * targhooks.c (default_print_patchable_function_entry): Likewise.
8250 * toplev.c (process_options): Switch off IPA-RA if
8251 patchable function entries are being generated.
8252 * varasm.c (assemble_start_function): Look at the
8253 patchable-function-entry command line switch and current
8254 function attributes and maybe generate NOP instructions by
8255 calling the print_patchable_function_entry hook.
8256 * doc/extend.texi: Document patchable_function_entry attribute.
8257 * doc/invoke.texi: Document -fpatchable_function_entry
8258 command line option.
8259 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
8261 * doc/tm.texi: Re-generate.
8263 2017-07-25 Jakub Jelinek <jakub@redhat.com>
8266 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
8267 TARGET_AVX512DQ rather than TARGET_AVX512BW.
8269 2017-07-25 Tamar Christina <tamar.christina@arm.com>
8271 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
8273 2017-07-25 Richard Biener <rguenther@suse.de>
8275 PR tree-optimization/81455
8276 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
8277 not walk in cycles when looking for guards.
8279 2017-07-25 Richard Biener <rguenther@suse.de>
8281 PR tree-optimization/81529
8282 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
8283 when optimizing backedge uses.
8285 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
8287 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
8289 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
8290 to dl_section_ref. On AIX, append an expression to subtract
8291 the size of the section length to dl_section_ref.
8293 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
8295 * configure.ac: If any of the config.* scripts fail, exit 1.
8296 * configure: Regenerate.
8298 2017-07-25 Richard Biener <rguenther@suse.de>
8301 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
8302 of immediate uses, be more verbose on errors.
8304 2017-07-25 Richard Biener <rguenther@suse.de>
8306 PR tree-optimization/81510
8307 * tree-vect-loop.c (vect_is_simple_reduction): When the
8308 reduction stmt is not inside the loop bail out.
8310 2017-07-25 Richard Biener <rguenther@suse.de>
8312 PR tree-optimization/81303
8313 * tree-vect-loop-manip.c (vect_loop_versioning): Build
8314 profitability check against LOOP_VINFO_NITERSM1.
8316 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8318 * domwalk.c (cmp_bb_postorder): Simplify.
8319 (sort_bbs_postorder): New function. Use it...
8320 (dom_walker::walk): ...here to optimize common cases.
8322 2017-07-25 Martin Liska <mliska@suse.cz>
8325 * ipa-visibility.c (function_and_variable_visibility): Make the
8326 redirection just on target that supports aliasing.
8327 Fix GNU coding style.
8329 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
8332 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
8334 * config/arm/rtems.h: Likewise.
8335 * config/bfin/rtems.h: Likewise.
8336 * config/i386/rtemself.h: Likewise.
8337 * config/lm32/rtems.h: Likewise.
8338 * config/m32c/rtems.h: Likewise.
8339 * config/m68k/rtemself.h: Likewise.
8340 * config/microblaze/rtems.h: Likewise.
8341 * config/mips/rtems.h: Likewise.
8342 * config/moxie/rtems.h: Likewise.
8343 * config/nios2/rtems.h: Likewise.
8344 * config/powerpcspe/rtems.h: Likewise.
8345 * config/rs6000/rtems.h: Likewise.
8346 * config/rtems.h: Likewise.
8347 * config/sh/rtems.h: Likewise.
8348 * config/sh/rtemself.h: Likewise.
8349 * config/sparc/rtemself.h: Likewise.
8351 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8354 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
8355 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
8356 * tree-ssa-structalias.c (alias_get_name): Same.
8358 2017-07-25 Bin Cheng <bin.cheng@arm.com>
8361 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
8362 instructions if no du chain is found.
8364 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8366 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
8368 2017-07-25 Richard Biener <rguenther@suse.de>
8371 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
8374 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8376 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
8378 (ISA_2_7_MASKS_SERVER): Likewise.
8379 (ISA_3_0_MASKS_IEEE): Likewise.
8380 (OTHER_P8_VECTOR_MASKS): Likewise.
8381 (OTHER_VSX_VECTOR_MASKS): Likewise.
8382 (POWERPC_MASKS): Likewise.
8383 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
8384 duplicate list of options.
8385 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
8386 explicit -mupper-regs options.
8387 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
8388 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
8389 alias for -mupper-regs-df.
8390 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8391 (rs6000_init_hard_regno_mode_ok): Likewise.
8392 (rs6000_option_override_internal): Likewise.
8393 (rs6000_opt_masks): Likewise.
8394 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8395 options in terms of whether -mvsx or -mpower8-vector was used.
8396 (TARGET_UPPER_REGS_DI): Likewise.
8397 (TARGET_UPPER_REGS_SF): Likewise.
8398 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8399 -mupper-regs-* options.
8401 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8403 * passes.c (emergency_dump_function): Print some empty lines and a
8404 header before the RTL dump.
8406 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8408 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8410 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8413 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8414 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8416 2017-07-24 Carl Love <cel@us.ibm.com>
8418 * config/rs6000/rs6000-c.c: Add support for built-in functions
8419 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8420 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8421 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8422 vec_extract_fp_from_shortl): Add defines for the two builtins.
8423 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8424 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8426 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8427 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8428 * doc/extend.texi: Update the built-in documentation file for the
8429 new built-in function.
8431 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8434 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8436 * doc/generic.texi: Likewise.
8437 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8438 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8440 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8442 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8443 (aarch64_mls_elt_merge<mode>): Likewise.
8445 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8447 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8448 having __cxa_atexit.
8450 2017-07-23 Michael Collison <michael.collison@arm.com>
8452 * config/arm/arm.c (arm_option_override): Deprecate
8453 use of -mstructure-size-boundary.
8454 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8455 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8457 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8460 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8461 Reduce cost estimate for direct moves.
8463 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8466 * config/i386/i386.c (ix86_option_override_internal): Disable
8467 BMI, BMI2 and TBM instructions for -m16.
8469 2017-07-21 Carl Love <cel@us.ibm.com>
8471 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8472 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8473 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8474 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8475 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8476 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8477 VMULOSW): New enum "unspec" values.
8478 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8479 altivec_vmulosw): New patterns.
8480 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8481 VMULOSW): Add definitions.
8483 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8485 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8486 (qdf24xx): Likewise.
8487 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8488 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8489 (AARCH64_FL_V8_1): Renumber.
8490 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8491 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8492 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8493 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8494 rdma to feature modifiers list.
8496 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8499 * ipa-visibility (function_and_variable_visibility): Convert
8500 recursive PLT call to direct call if appropriate.
8502 2017-07-21 Andrew Pinski <apinski@cavium.com>
8504 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8505 operand 1 to see if the types precision matches.
8506 * fold-const.c (operand_equal_p): Likewise.
8508 2017-07-21 Richard Biener <rguenther@suse.de>
8510 PR tree-optimization/81303
8511 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8512 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8513 (vect_peeling_hash_get_lowest_cost): Adjust.
8514 (vect_enhance_data_refs_alignment): Likewise. Use
8515 vect_get_peeling_costs_all_drs to compute the penalty for no
8516 peeling to match up costs.
8518 2017-07-21 Richard Biener <rguenther@suse.de>
8520 PR tree-optimization/81500
8521 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8522 we didn't identify a reduction path.
8524 2017-07-21 Tom de Vries <tom@codesourcery.com>
8525 Cesar Philippidis <cesar@codesourcery.com>
8527 PR gcov-profile/81442
8528 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8531 2017-07-21 Tom de Vries <tom@codesourcery.com>
8534 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8536 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8537 nvptx_override_options_after_change.
8539 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8541 * dwarf2out.c (output_file_names): Avoid double testing for
8544 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8546 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8548 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8550 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8552 (try_crossjump_to_edge): Do not punt on partitioned functions.
8554 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8556 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8557 Put all BBs reachable only via paths crossing cold region to cold
8559 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8561 2016-07-21 Richard Biener <rguenther@suse.de>
8563 PR tree-optimization/81303
8564 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8565 into account prologue and epilogue iterations when raising
8566 min_profitable_iters to sth at least covering one vector iteration.
8568 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8570 * config/arm/arm.c (arm_test_cpu_arch_dat):
8573 2017-07-20 Nathan Sidwell <nathan@acm.org>
8575 Remove TYPE_METHODS.
8576 * tree.h (TYPE_METHODS): Delete.
8577 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8578 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8579 (dbxout_type_methods): Scan TYPE_FIELDS.
8580 (dbxout_type): Don't check TYPE_METHODS here.
8581 * function.c (use_register_for_decl): Always ignore register for
8582 class types when not optimizing.
8583 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8584 * tree.c (free_lang_data_in_type): Stitch out member functions and
8585 templates from TYPE_FIELDS.
8586 (build_distinct_type_copy, verify_type_variant,
8587 verify_type): Member fns are on TYPE_FIELDS.
8588 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8589 * tree-pretty-print.c (dump_generic_node): Likewise.
8591 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8594 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8595 V2TImode and V4TImode.
8596 (ix86_expand_vector_extract): Likewise.
8597 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8598 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8599 (ssescalarmode): Handle V4TImode and V2TImode.
8600 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8601 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8602 (VEXTRACTI128_MODE): New mode iterator.
8603 (splitter for *vec_extractv?ti first element): New.
8604 (VEC_INIT_MODE): New mode iterator.
8605 (vec_init<mode>): Consolidate 3 expanders into one using
8606 VEC_INIT_MODE mode iterator.
8608 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8610 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8611 non_spilled_static_chain_regno_p.
8613 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8615 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8617 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8619 * bb-reorder.c (connect_traces): Allow copying of blocks within
8622 2017-07-20 Richard Biener <rguenther@suse.de>
8624 * gimple.h (gimple_phi_result): Add gphi * overload.
8625 (gimple_phi_result_ptr): Likewise.
8626 (gimple_phi_arg): Likewise. Adjust index assert to only
8627 allow actual argument accesses rather than all slots available
8629 (gimple_phi_arg_def): Add gphi * overload.
8630 * tree-phinodes.c (make_phi_node): Initialize only actual
8632 (resize_phi_node): Clear memory not covered by old node,
8633 do not initialize excess argument slots.
8634 (reserve_phi_args_for_new_edge): Initialize new argument slot
8637 2017-07-20 Bin Cheng <bin.cheng@arm.com>
8639 PR tree-optimization/81388
8641 2016-07-21 Bin Cheng <bin.cheng@arm.com>
8643 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
8644 by removing computation of may_be_zero.
8646 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8647 Tom de Vries <tom@codesourcery.com>
8650 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8651 when gimple level profile disagrees with what RTL expander did.
8653 2017-07-20 Richard Biener <rguenther@suse.de>
8655 PR tree-optimization/61171
8656 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8657 (vect_analyze_stmt): Add slp instance parameter.
8658 (vectorizable_reduction): Likewise.
8659 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8660 (vect_is_simple_reduction): Deal with chains not detected
8661 as SLP reduction chain, specifically not properly associated
8662 chains containing a mix of plus/minus.
8663 (get_reduction_op): Remove.
8664 (get_initial_defs_for_reduction): Simplify, pass in whether
8665 this is a reduction chain, pass in the SLP node for the PHIs.
8666 (vect_create_epilog_for_reduction): Get the SLP instance as
8668 (vectorizable_reduction): Get the SLP instance as arg.
8669 During analysis remember the SLP node with the PHIs in the
8670 instance. Simplify getting at the vectorized reduction PHIs.
8671 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8672 through SLP instance.
8673 (vect_slp_analyze_operations): Likewise.
8674 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8675 (vect_transform_stmt): Likewise.
8677 2017-07-20 Tom de Vries <tom@codesourcery.com>
8679 PR tree-optimization/81489
8680 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8681 read of phi arg location to before loop that modifies phi.
8683 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8685 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8688 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8691 * except.c (execute): Fix ordering issue.
8693 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8695 PR rtl-optimization/81423
8696 * combine.c (make_compound_operation_int): Don't try to optimize
8697 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8699 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8701 PR rtl-optimization/81423
8702 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8703 with a constant that is -1 in the truncated to mode.
8705 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8707 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8708 (determine_unlikely_bbs): ... here.
8709 * predict.h (propagate_unlikely_bbs_forward): Declare.
8710 * cfgexpand.c (pass_expand::execute): Use it.
8711 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8713 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8714 propagate_unlikely_bbs_forward.
8716 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8719 * except.c (maybe_add_nop_after_section_switch): New function.
8722 2017-07-19 Tom de Vries <tom@codesourcery.com>
8724 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8726 2017-07-19 Tom de Vries <tom@codesourcery.com>
8728 * gimple.h (gimple_phi_arg): Make assert more strict.
8730 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8732 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8733 * config/rs6000/mmintrin.h: New file.
8734 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8736 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8738 PR tree-optimization/81346
8739 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
8741 2017-07-19 Tom de Vries <tom@codesourcery.com>
8743 * config/nvptx/nvptx.md (VECIM): Add V2DI.
8745 2017-07-19 Tom de Vries <tom@codesourcery.com>
8747 * config/nvptx/nvptx-modes.def: Add V2DImode.
8748 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
8749 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
8750 (nvptx_output_mov_insn): Handle lack of mov.b128.
8751 (nvptx_print_operand): Handle 'H' and 'L' codes.
8752 (nvptx_vector_mode_supported): Allow V2DImode.
8753 (nvptx_preferred_simd_mode): New function.
8754 (nvptx_data_alignment): New function.
8755 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
8756 nvptx_preferred_simd_mode.
8757 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
8759 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
8761 2017-07-19 Tom de Vries <tom@codesourcery.com>
8763 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
8764 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
8765 (nvptx_vector_mode_supported): New function. Allow V2SImode.
8766 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
8767 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
8768 (mov<VECIM>_insn): New define_insn.
8769 (define_expand "mov<VECIM>): New define_expand.
8771 2017-07-19 Tom de Vries <tom@codesourcery.com>
8773 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
8775 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8777 PR tree-optimization/81346
8778 * fold-const.h (fold_div_compare, range_check_type): Declare.
8779 * fold-const.c (range_check_type): New function.
8780 (build_range_check): Use range_check_type.
8781 (fold_div_compare): No longer static, rewritten into
8782 a match.pd helper function.
8783 (fold_comparison): Don't call fold_div_compare here.
8784 * match.pd (X / C1 op C2): New optimization using fold_div_compare
8787 2017-07-19 Nathan Sidwell <nathan@acm.org>
8789 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
8790 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
8791 * tree.c (find_decls_types_r, verify_type): Use
8792 TYPE_{MIN,MAX}_VALUE_RAW.
8793 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
8794 (hash_tree): Likewise.
8795 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8797 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8800 2017-07-18 Tom de Vries <tom@codesourcery.com>
8803 * omp-expand.c (expand_omp_for_static_chunk): Handle
8804 equal-argument loop exit phi.
8806 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
8809 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
8810 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
8811 operand 2 predicate.
8812 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
8813 operand 2 predicate.
8814 (ror,rol -> rorx splitters): Use const_int_operand as
8815 operand 2 predicate.
8817 2017-06-18 Richard Biener <rguenther@suse.de>
8819 PR tree-optimization/81410
8820 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
8821 the gap in the ! slp_perm SLP case after each group.
8823 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8826 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
8829 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8832 * predict.c (set_even_probabilities): Cleanup; do not affect
8833 probabilities that are already known.
8834 (combine_predictions_for_bb): Call even when count is set.
8836 2017-07-18 Nathan Sidwell <nathan@acm.org>
8838 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
8841 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8844 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
8845 optimization for loop niter analysis.
8847 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
8850 * config/avr/avr.c (avr_optimize_casesi): Don't use
8851 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
8853 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8855 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
8856 body_cost_vec from _vect_peel_extended_info.
8857 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
8858 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
8861 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8863 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
8865 2017-07-18 Richard Biener <rguenther@suse.de>
8867 PR tree-optimization/80620
8868 PR tree-optimization/81403
8869 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
8870 info when re-using a VN table entry.
8872 2017-07-18 Richard Biener <rguenther@suse.de>
8874 PR tree-optimization/81418
8875 * tree-vect-loop.c (vectorizable_reduction): Properly compute
8876 vectype_in. Verify that with lane-reducing reduction operations
8877 we have a single def-use cycle.
8879 2017-07-17 Carl Love <cel@us.ibm.com>
8881 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
8883 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8884 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8885 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8886 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8887 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8888 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8889 VMULOSW): New enum "unspec" values.
8890 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8891 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8892 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8893 altivec_vmulosw): New patterns.
8894 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8895 VMULOSW): Add definitions.
8896 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
8898 * config/alpha/alpha.c: Include predict.h.
8900 2017-07-17 Yury Gribov <tetra2005@gmail.com>
8902 * tree-vrp.c (compare_assert_loc): Fix comparison function
8903 to return predictable results.
8905 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8907 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
8910 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
8911 * doc/invoke.texi (mexpand-adddi): Update text.
8913 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8915 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
8916 that also clobbers the CC register. The old expand code is moved
8918 (*arc_clzsi2): ... here.
8919 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
8920 the CC register. The old expand code is moved to ...
8921 (arc_ctzsi2): ... here.
8923 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8925 * config/arc/arc.opt (mindexed-loads): Use initial value
8926 TARGET_INDEXED_LOADS_DEFAULT.
8927 (mauto-modify-reg): Use initial value
8928 TARGET_AUTO_MODIFY_REG_DEFAULT.
8929 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8930 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8931 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8932 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8934 2017-07-17 Martin Liska <mliska@suse.cz>
8937 * opts.c (finish_options): Do not allow -fgnu-tm
8938 w/ -fsanitize={kernel-,}address. Say sorry.
8940 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8943 * tree-loop-distribution.c (classify_partition): Only assert on
8944 numer of iterations.
8945 (merge_dep_scc_partitions): Delete prameter. Update function call.
8946 (distribute_loop): Remove code handling loop with unknown niters.
8947 (pass_loop_distribution::execute): Skip loop with unknown niters.
8949 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8952 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
8953 function sort_partitions_by_post_order.
8955 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8957 PR tree-optimization/81374
8958 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
8959 the max index of basic blocks, rather than number of basic blocks.
8961 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8963 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
8965 (arc_legitimate_pic_operand_p): Likewise.
8966 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
8968 (arc_needs_pcl_p): Likewise.
8969 (arc_legitimate_pc_offset_p): Likewise.
8970 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
8971 function is also used in constrains.md.
8972 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
8973 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
8974 PLUS. Only return true/false in known cases, otherwise assert.
8975 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
8976 is already called in arc_legitimate_constant_p.
8977 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8979 (LEGITIMATE_PIC_OPERAND_P): Use
8980 arc_raw_symbolic_reference_mentioned_p function.
8981 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8986 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8987 Andrew Burgess <andrew.burgess@embecosm.com>
8989 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8990 (arc_return_address_register): New function.
8991 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8992 (arc_handle_fndecl_attribute): Add naked attribute.
8993 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8994 (TARGET_WARN_FUNC_RETURN): Likewise.
8995 (arc_allocate_stack_slots_for_args): New function.
8996 (arc_warn_func_return): Likewise.
8997 (machine_function): Change type fn_type.
8998 (arc_compute_function_type): Consider new naked function type,
8999 change function return type.
9000 (arc_must_save_register): Adapt to handle new
9001 arc_compute_function_type's return type.
9002 (arc_expand_prologue): Likewise.
9003 (arc_expand_epilogue): Likewise.
9004 (arc_return_address_regs): Delete.
9005 (arc_return_address_register): New function.
9006 (arc_epilogue_uses): Use above function.
9007 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
9008 (arc_function_type): Change encoding, add naked type.
9009 (ARC_INTERRUPT_P): Change to handle the new encoding.
9010 (ARC_FAST_INTERRUPT_P): Likewise.
9011 (ARC_NORMAL_P): Define.
9012 (ARC_NAKED_P): Likewise.
9013 (arc_compute_function_type): Delete prototype.
9014 * config/arc/arc.md (in_ret_delay_slot): Use
9015 arc_return_address_register function.
9016 (simple_return): Likewise.
9017 (p_return_i): Likewise.
9019 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9021 PR tree-optimization/81428
9022 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
9023 can't be built for those types.
9025 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9027 Remove stuff dead since r239246.
9029 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
9030 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
9031 (avr_inform_devices): Remove dead stuff.
9033 2017-07-17 Tamar Christina <tamar.christina@arm.com>
9035 * config/arm/arm_neon.h: Fix softp typo.
9037 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9039 PR tree-optimization/81365
9040 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
9041 aggregate moves onto bb predecessor edges, make sure there are no
9042 loads that could alias the lhs in between the start of bb and the
9045 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
9048 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
9049 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
9050 [LSHIFTRT, outer_code = TRUNCATE]: Same.
9052 2017-07-17 Jakub Jelinek <jakub@redhat.com>
9054 PR tree-optimization/81396
9055 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
9056 (init_symbolic_number): Initialize it to 1.
9057 (perform_symbolic_merge): Add n_ops from both operands into the new
9059 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
9060 without base_addr as useless if they need more than one operation.
9061 (bswap_replace): Handle !bswap case for NULL base_addr.
9063 2017-07-17 Tom de Vries <tom@codesourcery.com>
9066 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
9069 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
9071 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
9072 conditional builtin define __FIX_LEON3FT_B2BST.
9074 2017-07-17 Daniel Cederman <cederman@gaisler.com>
9076 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
9077 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
9080 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9082 PR rtl-optimization/81424
9083 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
9084 to remove potential trapping from operands if -fnon-call-exceptions.
9086 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9088 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
9089 profile_proability for scalling.
9090 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
9092 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9094 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
9096 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9098 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
9099 fixpoint arithmetics.
9101 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9103 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
9104 fixpoint arithmetics.
9106 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9108 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
9109 fixpoint arithmetics.
9111 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9113 * profile-count.h (profile_probability::from_reg_br_prob_note,
9114 profile_probability::to_reg_br_prob_note): New functions.
9115 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
9116 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
9117 * predict.c (probability_reliable_p): Update.
9118 (edge_probability_reliable_p): Update.
9119 (br_prob_note_reliable_p): Update.
9120 (invert_br_probabilities): Update.
9121 (add_reg_br_prob_note): New function.
9122 (combine_predictions_for_insn): Update.
9123 * asan.c (asan_clear_shadow): Update.
9124 * cfgbuild.c (compute_outgoing_frequencies): Update.
9125 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
9126 (update_br_prob_note): Update.
9127 (rtl_verify_edges): Update.
9128 (purge_dead_edges): Update.
9129 (fixup_reorder_chain): Update.
9130 * emit-rtl.c (try_split): Update.
9131 * ifcvt.c (cond_exec_process_insns): Update.
9132 (cond_exec_process_if_block): Update.
9133 (dead_or_predicable): Update.
9134 * internal-fn.c (expand_addsub_overflow): Update.
9135 (expand_neg_overflow): Update.
9136 (expand_mul_overflow): Update.
9137 * loop-doloop.c (doloop_modify): Update.
9138 * loop-unroll.c (compare_and_jump_seq): Update.
9139 * optabs.c (emit_cmp_and_jump_insn_1): Update.
9140 * predict.h: Update.
9141 * reorg.c (mostly_true_jump): Update.
9143 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
9144 * config/alpha/alpha.c (emit_unlikely_jump): Update.
9145 * config/arc/arc.c: (emit_unlikely_jump): Update.
9146 * config/arm/arm.c: (emit_unlikely_jump): Update.
9147 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
9148 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
9149 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
9150 (ix86_print_operand): Update.
9151 (ix86_split_fp_branch): Update.
9152 (predict_jump): Update.
9153 * config/ia64/ia64.c (ia64_print_operand): Update.
9154 * config/mmix/mmix.c (mmix_print_operand): Update.
9155 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
9156 (rs6000_expand_split_stack_prologue): Update.
9157 * config/rs6000/rs6000.c: Update.
9158 * config/s390/s390.c (s390_expand_vec_strlen): Update.
9159 (s390_expand_vec_movstr): Update.
9160 (s390_expand_cs_tdsi): Update.
9161 (s390_expand_split_stack_prologue): Update.
9162 * config/sh/sh.c (sh_print_operand): Update.
9163 (expand_cbranchsi4): Update.
9164 (expand_cbranchdi4): Update.
9165 * config/sparc/sparc.c (output_v9branch): Update.
9166 * config/spu/spu.c (get_branch_target): Update.
9167 (ea_load_store_inline): Update.
9168 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
9169 * config/tilepro/tilepro.c: Update.
9171 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9173 * gimplify.c (mostly_copy_tree_r): Revert latest change.
9174 (gimplify_save_expr): Likewise.
9176 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9178 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
9180 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9182 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
9184 * timevar.def (TV_IPA_FNSUMMARY): Define.
9186 2017-07-16 Daniel Cederman <cederman@gaisler.com>
9188 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
9189 to back store errata sensitive sequence from being generated.
9190 (sqrtdf2_fix): Likewise.
9192 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9194 * tree-ssa-threadupdate.c (compute_path_counts,
9195 update_joiner_offpath_counts): Use profile_probability.
9197 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
9200 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9202 * config/arm/arm-c.c (arm_cpu_builtins): Define
9203 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9205 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9207 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9208 array entries to represent __ieee128 versions of the
9209 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
9210 scalar_extract_sig, and scalar_insert_exp built-in functions.
9211 (altivec_resolve_overloaded_builtin): Add special case handling
9212 for the __builtin_scalar_insert_exp function, as represented by
9213 the P9V_BUILTIN_VEC_VSIEDP constant.
9214 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
9215 exponent support for __ieee128 argument.
9216 (VSESQP): Add scalar extract signature support for __ieee128
9218 (VSTDCNQP): Add scalar test negative support for __ieee128
9220 (VSIEQP): Add scalar insert exponent support for __int128 argument
9221 with __ieee128 result.
9222 (VSIEQPF): Add scalar insert exponent support for __ieee128
9223 argument with __ieee128 result.
9224 (VSTDCQP): Add scalar test data class support for __ieee128
9226 (VSTDCNQP): Add overload support for scalar test negative with
9228 (VSTDCQP): Add overload support for scalar test data class
9230 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
9232 (UNSPEC_VSX_SIEXPQP): New constant.
9233 (xsxexpqp): New insn for VSX scalar extract exponent quad
9235 (xsxsigqp): New insn for VSX scalar extract significand quad
9237 (xsiexpqpf): New insn for VSX scalar insert exponent quad
9238 precision with floating point argument.
9239 (xststdcqp): New expand for VSX scalar test data class quad
9241 (xststdcnegqp): New expand for VSX scalar test negative quad
9243 (xststdcqp): New insn to match expansions for VSX scalar test data
9244 class quad precision and VSX scalar test negative quad precision.
9245 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
9246 special case operand checking to enforce that second operand of
9247 VSX scalar test data class with quad precision argument is a 7-bit
9249 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9250 prototypes and descriptions of __ieee128 versions of
9251 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
9252 scalar_test_data_class, and scalar_test_neg built-in functions.
9254 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9256 PR tree-optimization/81162
9257 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
9258 replace a negate with an add.
9260 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
9262 * doc/invoke.texi (arm/-mcpu): Document +crypto.
9264 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9266 * config/arm/arm-c.c (arm_cpu_builtins): Define
9267 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9269 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9271 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
9272 (armv8-r): Set ARM Cortex-R52 as default CPU.
9273 * config/arm/arm-tables.opt: Regenerate.
9274 * config/arm/arm-tune.md: Regenerate.
9275 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9277 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
9278 extension for -mcpu=cortex-r52.
9280 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9282 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
9283 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
9284 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
9285 (fp-armv8): Define it as FP_ARMv8 only.
9286 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
9287 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
9289 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
9290 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
9291 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
9292 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
9293 than TARGET_FPU_ARMV8.
9294 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
9295 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
9296 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
9298 * config/arm/neon.md (neon_vrint): Likewise.
9299 (neon_vcvt): Likewise.
9300 (neon_<fmaxmin_op><mode>): Likewise.
9301 (<fmaxmin><mode>3): Likewise.
9302 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
9303 * config/arm/predicates.md (arm_cond_move_operator): Check against
9304 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
9306 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
9308 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
9311 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9313 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
9314 loop in comment with memset.
9316 2017-07-14 Martin Liska <mliska@suse.cz>
9318 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
9319 * dwarf2out.c (is_java): Remove the function.
9320 (output_pubname): Remove usage of the function.
9321 (lower_bound_default): Remove usage of DW_LANG_Java.
9322 (gen_compile_unit_die): Likewise.
9323 * gcc.c: Remove compiler defaults for .java and .zip files.
9324 * gimple-expr.c (remove_suffix): Change as there's no longer
9325 extension than 4-letter one.
9326 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
9327 (gimplify_save_expr): Likewise.
9328 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
9329 as it's possible even for other languages than Java.
9330 * langhooks.h (struct lang_hooks): Remove Java from a comment.
9331 * lto-opts.c (lto_write_options): Remove reference to Java.
9332 * opts.c (strip_off_ending): Update file extension handling.
9333 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
9334 * tree-eh.c (lower_resx): Likewise.
9335 * tree.c (free_lang_data_in_type): Remove dead code.
9336 (find_decls_types_r): Likewise.
9337 (build_common_builtin_nodes): Remove Java from a comment.
9338 (verify_type): Remove dead code.
9339 * varasm.c (assemble_external): Remove Java from a comment.
9341 2017-07-14 Martin Liska <mliska@suse.cz>
9343 * opts.c (finish_options): Add quotes.
9344 (common_handle_option): Likewise.
9346 2017-07-14 Martin Liska <mliska@suse.cz>
9348 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
9349 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
9351 * dwarf2out.c (lower_bound_default): Do not handle
9353 (gen_compile_unit_die): Likewise.
9354 * gcc.c: Remove default extension binding for GNU Pascal.
9355 * stmt.c: Remove Pascal language from a comment.
9356 * xcoffout.c: Likewise.
9358 2017-07-13 David Malcolm <dmalcolm@redhat.com>
9361 * diagnostic-show-locus.c (fixit_cmp): New function.
9362 (layout::layout): Sort m_fixit_hints.
9363 (column_range::column_range): Assert that the values are valid.
9364 (struct char_span): New struct.
9365 (correction::overwrite): New method.
9366 (struct source_line): New struct.
9367 (line_corrections::add_hint): Add assertions. Reimplement memcpy
9368 calls in terms of classes source_line and char_span, and
9369 correction::overwrite.
9370 (selftest::test_overlapped_fixit_printing_2): New function.
9371 (selftest::diagnostic_show_locus_c_tests): Call it.
9373 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9375 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
9376 early if there is no lhs.
9378 2017-07-13 Martin Liska <mliska@suse.cz>
9380 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
9381 (gen_reference_type_die): Likewise.
9382 * stor-layout.c: Remove Pascal-related comment.
9384 2017-07-13 Martin Liska <mliska@suse.cz>
9386 * opts.c (finish_options): Add quotes to error messages.
9387 (parse_sanitizer_options): Likewise.
9389 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
9391 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
9393 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9395 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9397 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9399 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9401 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9403 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9406 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9407 provides the hardware capability bits, define the macro
9408 __BUILTIN_CPU_SUPPORTS__.
9409 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9410 if GLIBC does not provide the hardware capability bits. Add a
9411 gcc_unreachable call if the built-in cpu function is neither
9412 __builtin_cpu_is nor __builtin_cpu_supports.
9413 (rs6000_get_function_versions_dispatcher): Change the warning
9414 that an old GLIBC is used which does not export the capability
9415 bits to be an error.
9416 * doc/extend.texi (target_clones attribute): Document the
9417 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9418 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9419 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9420 the macros defined by GCC if the newer GLIBC is available.
9422 2017-07-12 Jeff Law <law@redhat.com>
9424 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9425 remaining includes slightly.
9426 * config/riscv/riscv-builtins.c: Include profile-count.h.
9428 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9431 * config/avr/avr.c (avr_set_current_function): In diagnostic
9432 messages: Quote keywords and (parts of) identifiers.
9433 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9436 2017-07-12 Carl Love <cel@us.ibm.com>
9438 * config/rs6000/rs6000-c.c: Add support for built-in functions
9439 vector bool char vec_revb (vector bool char);
9440 vector bool short vec_revb (vector short char);
9441 vector bool int vec_revb (vector bool int);
9442 vector bool long long vec_revb (vector bool long long);
9443 * doc/extend.texi: Update the built-in documentation file for the
9444 new built-in functions.
9446 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9448 * config/s390/s390.md: Remove movcc splitter.
9450 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9452 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9453 load/store on condition.
9455 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9458 * config/avr/avr.c (avr_encode_section_info)
9459 [progmem && !TREE_READONLY]: Error if progmem object needs
9462 2017-07-11 Michael Collison <michael.collison@arm.com>
9464 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9467 2017-07-11 Carl Love <cel@us.ibm.com>
9469 * config/rs6000/rs6000-c.c: Add support for builtins
9470 vector unsigned int vec_parity_lsbb (vector signed int);
9471 vector unsigned int vec_parity_lsbb (vector unsigned int);
9472 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9473 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9474 vector unsigned long long vec_parity_lsbb (vector signed long long);
9475 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9476 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9477 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9478 * doc/extend.texi: Update the built-in documentation file for the
9479 new built-in functions.
9481 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9483 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9484 (layout::m_primary_loc): New field.
9485 (layout::layout): Initialize new field. Move location filtering
9486 logic from here to...
9487 (layout::maybe_add_location_range): ...this new method. Add
9488 support for filtering to just the lines already specified by other
9490 (layout::will_show_line_p): New method.
9491 (gcc_rich_location::add_location_if_nearby): New method.
9492 (selftest::test_add_location_if_nearby): New test function.
9493 (selftest::diagnostic_show_locus_c_tests): Call it.
9494 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9497 2017-07-11 Tom de Vries <tom@codesourcery.com>
9499 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9500 (bb_first_real_insn): New function.
9501 (nvptx_single): Add extra initialization of broadcasted condition
9504 2017-07-11 Nathan Sidwell <nathan@acm.org>
9506 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9508 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9510 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9511 Binutils doc as TEXI will mess them up.
9512 * doc/invoke.texi (AVR Options): Same here.
9514 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9516 * config/sparc/sparc.opt (mfix-ut700): New option.
9517 (mfix-gr712rc): Likewise.
9518 (sparc_fix_b2bst): New variable.
9519 * doc/invoke.texi (SPARC options): Document them.
9520 (ARM options): Fix warnings.
9521 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9522 instructions to prevent sequences that can trigger the store-store
9523 errata for certain LEON3FT processors.
9524 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9525 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9526 * config/sparc/sparc.md (fix_b2bst): New attribute.
9527 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9529 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9532 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9534 (*rsqrtsf2_sse): Ditto.
9536 (div<mode>3): Macroize insn from divdf3 and divsf3
9537 using MODEF mode iterator.
9539 2017-07-10 Martin Sebor <msebor@redhat.com>
9541 PR tree-optimization/80397
9542 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9543 instead of testing for equality to INTEGER_TYPE.
9545 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9547 * config.gcc: Remove uclibc from arc target spec.
9549 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9551 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9553 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9556 * lto-wrapper.c (remove_option): New function.
9557 (merge_and_complain): Merge PIC/PIE options more realistically.
9559 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9561 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9565 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9566 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9567 * config.in: Regenerate.
9568 * configure: Regenerate.
9569 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9570 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9571 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9572 (TARGET_GASISR_PROLOGUES): ...target mask.
9573 * common/config/avr/avr-common.c
9574 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9575 Set -mgas-isr-prologues.
9576 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9577 INSERT_PASS_BEFORE for it.
9578 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9579 * config/avr/avr.c (avr_option_override)
9580 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9581 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9582 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9583 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9584 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9586 (make_avr_pass_pre_proep): New function.
9587 (emit_push_sfr) <treg>: Add argument to function and use it
9589 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9590 and set machine->gasisr.yes.
9591 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9592 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9593 __gcc_isr.n_pushed to .L__stack_usage.
9594 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9595 (avr_asm_final_postscan_insn): ...this new static function.
9596 * config/avr/avr.h (machine_function)
9597 <is_no_gccisr, use_L__stack_usage>: New fields.
9598 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9599 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9600 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9601 (gasisr, *gasisr): New expander and insn.
9602 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9603 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9604 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9606 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9608 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9611 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9613 Move jump-tables out of .text again.
9616 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9617 (ASM_OUTPUT_ADDR_VEC): New function.
9618 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9619 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9620 INSN_ADDRESSes as asm comment.
9621 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9622 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9623 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9624 * config/avr/avr.md (*tablejump): Adjust comment.
9625 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9626 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9628 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9629 (avr_output_addr_vec): New proto.
9630 (avr_log_t) <insn_addresses>: New field.
9632 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9635 * config/i386/i386.c (ix86_function_arg_advance): Set
9636 outgoing_args_on_stack to true if there are outgoing arguments
9638 (ix86_function_arg): Likewise.
9639 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
9640 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
9641 * config/i386/i386.h (machine_function): Add
9642 outgoing_args_on_stack.
9644 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
9646 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9648 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
9650 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9652 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9653 (REAL_H): Remove $(MACHMODE_H).
9654 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9656 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9657 $(MACHMODE_H) and double-int.h.
9658 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9660 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9663 2017-07-07 Andrew Pinski <apinski@cavium.com>
9665 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9666 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9668 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9670 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9671 Add warning if GCC was not configured to link against a GLIBC that
9672 exports the hardware capability bits.
9673 (make_resolver_func): Make resolver function private and not a
9674 COMDAT function. Create the name with clone_function_name instead
9675 of make_unique_name.
9678 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9679 correct operand in doing the split.
9681 2017-07-07 Carl Love <cel@us.ibm.com>
9683 * config/rs6000/rs6000-c: Add support for built-in function
9684 vector unsigned short vec_pack_to_short_fp32 (vector float,
9686 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9687 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9688 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9689 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9690 (convert_4f32_8i16): Add define_expand.
9691 * doc/extend.texi: Update the built-in documentation file for the
9692 new built-in function.
9694 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9696 * config/sparc/m8.md: New file.
9697 * config/sparc/sparc.md: Include m8.md.
9699 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9701 * config/sparc/sparc.opt: New option -mvis4b.
9702 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9703 (sparc_option_override): Handle VIS4B.
9704 (enum sparc_builtins): Define
9705 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9706 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9707 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9708 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9709 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9710 (check_constant_argument): New function.
9711 (sparc_vis_init_builtins): Define builtins
9712 __builtin_vis_dictunpack{8,16,32},
9713 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9714 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9715 __builtin_vis_fpcmpde{8,16,32}shl and
9716 __builtin_vis_fpcmpur{8,16,32}shl.
9717 (sparc_expand_builtin): Check that the constant operands to
9718 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9719 constant and in range.
9720 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9722 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9723 (SPARC_IMM5_P): Likewise.
9724 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9725 (enabled): Handle vis4b.
9726 (UNSPEC_DICTUNPACK): New unspec.
9727 (UNSPEC_FPCMPSHL): Likewise.
9728 (UNSPEC_FPUCMPSHL): Likewise.
9729 (UNSPEC_FPCMPDESHL): Likewise.
9730 (UNSPEC_FPCMPURSHL): Likewise.
9731 (cpu_feature): New CPU feature `vis4b'.
9732 (dictunpack{8,16,32}): New insns.
9733 (FPCSMODE): New mode iterator.
9734 (fpcscond): New code iterator.
9735 (fpcsucond): Likewise.
9736 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9737 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
9738 (fpcmpde{8,16,32}{si,di}shl): Likewise.
9739 (fpcmpur{8,16,32}{si,di}shl): Likewise.
9740 * config/sparc/constraints.md: Define constraints `q' for unsigned
9741 2-bit integer constants and `t' for unsigned 5-bit integer
9743 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
9745 (imm5_operand_dictunpack16): Likewise.
9746 (imm5_operand_dictunpack32): Likewise.
9747 (imm2_operand): Likewise.
9748 * doc/invoke.texi (SPARC Options): Document -mvis4b.
9749 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
9750 ditunpack* and fpcmp*shl builtins.
9752 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9754 * config.gcc: Handle m8 in --with-{cpu,tune} options.
9755 * config.in: Add HAVE_AS_SPARC6 define.
9756 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
9758 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
9760 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
9761 (CPP_CPU_SPEC): Handle m8.
9762 (ASM_CPU_SPEC): Likewise.
9763 * config/sparc/sparc-opts.h (enum processor_type): Add
9765 * config/sparc/sparc.c (m8_costs): New struct.
9766 (sparc_option_override): Handle TARGET_CPU_m8.
9767 (sparc32_initialize_trampoline): Likewise.
9768 (sparc64_initialize_trampoline): Likewise.
9769 (sparc_issue_rate): Likewise.
9770 (sparc_register_move_cost): Likewise.
9771 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
9772 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
9773 (ASM_CPU64_DEFAULT_SPEC): Likewise.
9774 (CPP_CPU_SPEC): Handle M8.
9775 (ASM_CPU_SPEC): Likewise.
9776 (AS_M8_FLAG): Define.
9777 * config/sparc/sparc.md: Add m8 to the cpu attribute.
9778 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
9779 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
9781 * configure: Regenerate.
9782 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
9785 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9787 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
9789 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
9790 ("*movdi_insn_sp32"): Do not set v3pipe.
9791 ("*movsi_insn"): Likewise.
9792 ("*movdi_insn_sp64"): Likewise.
9793 ("*movsf_insn"): Likewise.
9794 ("*movdf_insn_sp32"): Likewise.
9795 ("*movdf_insn_sp64"): Likewise.
9796 ("*zero_extendsidi2_insn_sp64"): Likewise.
9797 ("*sign_extendsidi2_insn"): Likewise.
9798 ("*mov<VM32:mode>_insn"): Likewise.
9799 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9800 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9801 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9802 ("<vlop:code><VL:mode>3"): Likewise.
9803 ("*not_<vlop:code><VL:mode>3"): Likewise.
9804 ("*nand<VL:mode>_vis"): Likewise.
9805 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
9806 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
9807 ("one_cmpl<VL:mode>2"): Likewise.
9808 ("faligndata<VM64:mode>_vis"): Likewise.
9809 ("alignaddrsi_vis"): Likewise.
9810 ("alignaddrdi_vis"): Likweise.
9811 ("alignaddrlsi_vis"): Likewise.
9812 ("alignaddrldi_vis"): Likewise.
9813 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9814 ("bmaskdi_vis"): Likewise.
9815 ("bmasksi_vis"): Likewise.
9816 ("bshuffle<VM64:mode>_vis"): Likewise.
9817 ("cmask8<P:mode>_vis"): Likewise.
9818 ("cmask16<P:mode>_vis"): Likewise.
9819 ("cmask32<P:mode>_vis"): Likewise.
9820 ("pdistn<P:mode>_vis"): Likewise.
9821 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9823 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9825 * config/sparc/sparc.md ("subtype"): New insn attribute.
9826 ("*wrgsr_sp64"): Set insn subtype.
9827 ("*rdgsr_sp64"): Likewise.
9828 ("alignaddrsi_vis"): Likewise.
9829 ("alignaddrdi_vis"): Likewise.
9830 ("alignaddrlsi_vis"): Likewise.
9831 ("alignaddrldi_vis"): Likewise.
9832 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9833 ("fexpand_vis"): Likewise.
9834 ("fpmerge_vis"): Likewise.
9835 ("faligndata<VM64:mode>_vis"): Likewise.
9836 ("bshuffle<VM64:mode>_vis"): Likewise.
9837 ("cmask8<P:mode>_vis"): Likewise.
9838 ("cmask16<P:mode>_vis"): Likewise.
9839 ("cmask32<P:mode>_vis"): Likewise.
9840 ("fchksm16_vis"): Likewise.
9841 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
9842 ("fmean16_vis"): Likewise.
9843 ("fp<plusminus_insn>64_vis"): Likewise.
9844 ("<plusminus_insn>v8qi3"): Likewise.
9845 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9846 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
9847 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
9848 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
9849 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
9850 ("*movqi_insn"): Likewise.
9851 ("*movhi_insn"): Likewise.
9852 ("*movsi_insn"): Likewise.
9853 ("movsi_pic_gotdata_op"): Likewise.
9854 ("*movdi_insn_sp32"): Likewise.
9855 ("*movdi_insn_sp64"): Likewise.
9856 ("movdi_pic_gotdata_op"): Likewise.
9857 ("*movsf_insn"): Likewise.
9858 ("*movdf_insn_sp32"): Likewise.
9859 ("*movdf_insn_sp64"): Likewise.
9860 ("*zero_extendhisi2_insn"): Likewise.
9861 ("*zero_extendqihi2_insn"): Likewise.
9862 ("*zero_extendqisi2_insn"): Likewise.
9863 ("*zero_extendqidi2_insn"): Likewise.
9864 ("*zero_extendhidi2_insn"): Likewise.
9865 ("*zero_extendsidi2_insn_sp64"): Likewise.
9866 ("ldfsr"): Likewise.
9867 ("prefetch_64"): Likewise.
9868 ("prefetch_32"): Likewise.
9869 ("tie_ld32"): Likewise.
9870 ("tie_ld64"): Likewise.
9871 ("*tldo_ldub_sp32"): Likewise.
9872 ("*tldo_ldub1_sp32"): Likewise.
9873 ("*tldo_ldub2_sp32"): Likewise.
9874 ("*tldo_ldub_sp64"): Likewise.
9875 ("*tldo_ldub1_sp64"): Likewise.
9876 ("*tldo_ldub2_sp64"): Likewise.
9877 ("*tldo_ldub3_sp64"): Likewise.
9878 ("*tldo_lduh_sp32"): Likewise.
9879 ("*tldo_lduh1_sp32"): Likewise.
9880 ("*tldo_lduh_sp64"): Likewise.
9881 ("*tldo_lduh1_sp64"): Likewise.
9882 ("*tldo_lduh2_sp64"): Likewise.
9883 ("*tldo_lduw_sp32"): Likewise.
9884 ("*tldo_lduw_sp64"): Likewise.
9885 ("*tldo_lduw1_sp64"): Likewise.
9886 ("*tldo_ldx_sp64"): Likewise.
9887 ("*mov<VM32:mode>_insn"): Likewise.
9888 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9889 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9891 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9893 * config/sparc/sparc.md ("type"): New insn type viscmp.
9894 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
9896 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
9897 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
9898 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9899 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
9901 ("n7_vis_logical_11cycle"): Likewise.
9902 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
9903 * config/sparc/niagara2.md ("niag3_vis": Likewise.
9904 * config/sparc/niagara.md ("niag_vis"): Likewise.
9905 * config/sparc/ultra3.md ("us3_fga"): Likewise.
9906 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
9908 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9910 * config/sparc/sparc.md: New instruction type `bmask'.
9911 (bmaskdi_vis): Use the `bmask' type.
9912 (bmasksi_vis): Likewise.
9913 * config/sparc/ultra3.md (us3_array): Likewise.
9914 * config/sparc/niagara7.md (n7_array): Likewise.
9915 * config/sparc/niagara4.md (n4_array): Likewise.
9916 * config/sparc/niagara2.md (niag2_vis): Likewise.
9917 (niag3_vis): Likewise.
9918 * config/sparc/niagara.md (niag_vis): Likewise.
9920 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9922 * ipa-comdats.c: Remove optimize check from gate.
9923 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
9924 for functions not optimized.
9925 (ipa_fn_summary_read): Skip optimize check.
9926 (ipa_fn_summary_write): Likewise.
9927 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
9929 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
9931 (can_inline_edge_p): Check flag_pcc_struct_return for match.
9932 (check_callers): Give up on caller which is not optimized.
9933 (inline_small_functions): Likewise.
9934 (ipa_inline): Do not give up when not optimizing.
9935 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
9936 away unoptimizes cdtors.
9937 (whole_program_function_and_variable_visibility): Do
9938 ipa_discover_readonly_nonaddressable_vars in LTO mode.
9939 * ipa.c (process_references): Do not check optimize.
9940 (symbol_table::remove_unreachable_nodes): Update optimize check.
9941 (set_writeonly_bit): Update optimize check.
9942 (pass_ipa_cdtor_merge::gate): Do not check optimize.
9943 (pass_ipa_single_use::gate): Remove.
9945 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9947 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
9948 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
9949 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
9950 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
9951 permute_load, permute_store, adjust_extract, adjust_splat,
9952 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
9953 replace_swap_with_copy, dump_swap_insn_table,
9954 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
9955 recombine_lvx_pattern, recombine_stvx_pattern,
9956 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
9957 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
9958 to file rs6000-p8swap.c.
9959 * config/rs6000/rs6000-p8swap.c: New file.
9960 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
9961 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
9962 and rs6000*-*-* targets.
9964 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9966 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
9968 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9970 * lto-wrapper.c (merge_and_complain): Do not merge
9971 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
9972 fsigned_zeros, ftrapping_math, fwrapv.
9973 (append_compiler_options): Do not track these options.
9974 (append_linker_options): Likewie
9976 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9978 * cgraphunit.c (cgraph_node::finalize_function): When
9979 !flag_toplevel_reorde set no_reorder flag.
9980 (varpool_node::finalize_decl): Likewise.
9981 (symbol_table::compile): Drop no toplevel reorder path.
9983 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9985 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9986 edges; zero probability is not better than uninitialized.
9988 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9990 * asan.h (asan_sanitize_allocas_p): Declare.
9991 * asan.c (asan_sanitize_allocas_p): New function.
9992 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
9993 (handle_builtin_alloca): Likewise.
9994 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
9995 if !asan_sanitize_allocas_p.
9996 * params.def (asan-instrument-allocas): Add new option.
9997 * params.h (ASAN_PROTECT_ALLOCAS): Define.
9998 * opts.c (common_handle_option): Disable allocas sanitization for
10001 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
10003 * asan.c: Include gimple-fold.h.
10004 (get_last_alloca_addr): New function.
10005 (handle_builtin_stackrestore): Likewise.
10006 (handle_builtin_alloca): Likewise.
10007 (asan_emit_allocas_unpoison): Likewise.
10008 (get_mem_refs_of_builtin_call): Add new parameter, remove const
10009 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
10010 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
10011 (instrument_builtin_call): Pass gimple iterator to
10012 get_mem_refs_of_builtin_call.
10013 (last_alloca_addr): New global.
10014 * asan.h (asan_emit_allocas_unpoison): Declare.
10015 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
10016 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
10017 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
10018 if function calls alloca.
10019 * gimple-fold.c (replace_call_with_value): Remove static keyword.
10020 * gimple-fold.h (replace_call_with_value): Declare.
10021 * internal-fn.c: Include asan.h.
10022 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
10023 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
10025 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10027 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
10028 (C_SELFTEST_FLAGS): New.
10029 (CPP_SELFTEST_FLAGS): New.
10030 (SELFTEST_DEPS): New, from deps of s-selftest.
10031 (C_SELFTEST_DEPS): New, from deps of s-selftest.
10032 (CPP_SELFTEST_DEPS): New.
10033 (selftest): Add dependency on s-selftest-c++.
10034 (s-selftest): Rename to...
10035 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
10036 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
10037 than SELFTEST_FLAGS.
10038 (selftest-gdb): Rename to...
10039 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
10041 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
10042 (selftest-valgrind): Rename to...
10043 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
10045 (selftest-valgrind): Reintroduce as an alias for
10046 selftest-c-valgrind.
10047 (s-selftest-c++): New.
10048 (selftest-c++-gdb): New.
10049 (selftest-c++-valgrind): New.
10051 2017-07-06 Olivier Hainque <hainque@adacore.com>
10053 * gcc.c (process_command): When deciding if undefined variables
10054 should be ignored when processing specs, accept "gcc -v" as well.
10056 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
10058 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
10059 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
10061 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10063 * config/arm/arm-cpus.in (armv8-r): Add new entry.
10064 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
10065 * config/arm/arm-tables.opt: Regenerate.
10066 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
10068 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
10070 2017-07-06 Carl Love <cel@us.ibm.com>
10072 * ChangeLog: Clean up from mid air collision
10074 2017-07-06 Carl Love <cel@us.ibm.com>
10076 * config/rs6000/rs6000-c.c: Add support for built-in functions
10077 vector signed int vec_subc (vector signed int, vector signed int);
10078 vector signed __int128 vec_subc (vector signed __int128,
10079 vector signed __int128);
10080 vector unsigned __int128 vec_subc (vector unsigned __int128,
10081 vector unsigned __int128);
10082 vector signed int vec_sube (vector signed int, vector signed int,
10083 vector signed int);
10084 vector unsigned int vec_sube (vector unsigned int,
10085 vector unsigned int,
10086 vector unsigned int);
10087 vector signed __int128 vec_sube (vector signed __int128,
10088 vector signed __int128,
10089 vector signed__int128);
10090 vector unsigned __int128 vec_sube (vector unsigned __int128,
10091 vector unsigned __int128,
10092 vector unsigned __int128);
10093 vector signed int vec_subec (vector signed int, vector signed int,
10094 vector signed int);
10095 vector unsigned int vec_subec (vector unsigned int,
10096 vector unsigned int,
10097 vector unsigned int);
10098 vector signed __int128 vec_subec (vector signed __int128,
10099 vector signed __int128,
10100 vector signed__int128);
10101 vector unsigned __int128 vec_subec (vector unsigned __int128,
10102 vector unsigned __int128,
10103 vector unsigned __int128);
10104 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
10105 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
10106 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
10107 BU_ALTIVEC_OVERLOAD_X definitions.
10108 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
10109 * doc/extend.texi: Update the built-in documentation file for the new
10110 built-in functions.
10112 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10115 * diagnostic-show-locus.c (layout::layout): Use start and finish
10116 spelling location for the start and finish of each range.
10117 * genmatch.c (linemap_client_expand_location_to_spelling_point):
10118 Add unused aspect param.
10119 * input.c (expand_location_1): Add "aspect" param, and use it
10120 to access the correct part of the location.
10121 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
10123 (expand_location_to_spelling_point): Likewise.
10124 (linemap_client_expand_location_to_spelling_point): Add "aspect"
10125 param, and pass it to expand_location_1.
10127 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
10129 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
10130 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
10131 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
10132 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
10133 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
10134 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
10135 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
10136 _mm_maskz_getmant_ss): New intrinsics.
10137 (__builtin_ia32_getexpss128_mask): Changed to ...
10138 __builtin_ia32_getexpss128_round ... this.
10139 (__builtin_ia32_getexpsd128_mask): Changed to ...
10140 __builtin_ia32_getexpsd128_round ... this.
10141 * config/i386/i386-builtin-types.def
10142 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
10143 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
10144 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
10145 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
10146 __builtin_ia32_getmantss_mask_round): New builtins.
10147 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
10148 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
10149 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
10150 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
10151 * config/i386/sse.md
10152 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
10153 avx512f_sgetexp<mode><mask_scalar_name>
10154 <round_saeonly_scalar_name> ... this.
10155 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
10156 %0, %1, %2<round_saeonly_op3>}): Changed to ...
10157 vgetexp<ssescalarmodesuffix>
10158 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10159 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
10160 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
10161 avx512f_vgetmant<mode><mask_scalar_name>
10162 <round_saeonly_scalar_name> ... this.
10163 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10164 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
10165 vgetmant<ssescalarmodesuffix>
10166 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
10167 %0<mask_scalar_operand4>, %1, %2
10168 <round_saeonly_scalar_mask_op4>, %3} ... this.
10169 * config/i386/subst.md (mask_scalar_operand4,
10170 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
10171 round_saeonly_scalar_nimm_predicate): New subst attributes.
10173 2017-07-06 Julia Koval <julia.koval@intel.com>
10175 * config/i386/i386.c (ix86_erase_embedded_rounding):
10176 Remove code for old rounding pattern.
10178 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
10180 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
10182 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
10184 * doc/sourcebuild.texi (Test Directives, Variants of
10185 dg-require-support): Add documentation for dg-require-stack-check.
10187 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
10189 * config/i386/subst.md (mask_scalar, round_scalar,
10190 round_saeonly_scalar): New meta-templates.
10191 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
10192 round_scalar_mask_operand3, round_scalar_mask_op3,
10193 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
10194 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
10195 round_saeonly_scalar_constraint,
10196 round_saeonly_scalar_prefix): New subst attribute.
10197 * config/i386/sse.md
10198 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
10199 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
10200 <round_scalar_name> ... this.
10201 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
10202 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
10203 <round_scalar_name> ... this.
10204 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
10205 <sse>_vm<code><mode>3<mask_scalar_name>
10206 <round_saeonly_scalar_name> ... this.
10207 (v<plusminus_mnemonic><ssescalarmodesuffix>
10208 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10209 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10210 v<plusminus_mnemonic><ssescalarmodesuffix>
10211 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10212 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10213 (v<multdiv_mnemonic><ssescalarmodesuffix>
10214 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10215 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10216 v<multdiv_mnemonic><ssescalarmodesuffix>
10217 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10218 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10219 (v<maxmin_float><ssescalarmodesuffix>
10220 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
10221 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
10222 v<maxmin_float><ssescalarmodesuffix>
10223 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10224 %0<mask_scalar_operand3>, %1, %<iptr>2
10225 <round_saeonly_scalar_mask_op3>} ... this.
10227 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
10229 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
10230 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
10232 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
10233 Alan Hayward <alan.hayward@arm.com>
10234 David Sherwood <david.sherwood@arm.com>
10236 * combine.c (simplify_if_then_else): Remove "enum" before
10238 * compare-elim.c (can_eliminate_compare): Likewise.
10239 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10241 (aarch64_lookup_simd_builtin_type): Likewise.
10242 (aarch64_simd_builtin_type): Likewise.
10243 (aarch64_init_simd_builtin_types): Likewise.
10244 (aarch64_simd_expand_args): Likewise.
10245 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
10247 (aarch64_reverse_mask): Likewise.
10248 (aarch64_simd_emit_reg_reg_move): Likewise.
10249 (aarch64_gen_adjusted_ldpstp): Likewise.
10250 (aarch64_ccmp_mode_to_code): Likewise.
10251 (aarch64_operands_ok_for_ldpstp): Likewise.
10252 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10253 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10255 (aarch64_min_divisions_for_recip_mul): Likewise.
10256 (aarch64_reassociation_width): Likewise.
10257 (aarch64_get_condition_code_1): Likewise.
10258 (aarch64_simd_emit_reg_reg_move): Likewise.
10259 (aarch64_simd_attr_length_rglist): Likewise.
10260 (aarch64_reverse_mask): Likewise.
10261 (aarch64_operands_ok_for_ldpstp): Likewise.
10262 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10263 (aarch64_gen_adjusted_ldpstp): Likewise.
10264 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
10266 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
10267 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10268 (arm_lookup_simd_builtin_type): Likewise.
10269 (arm_simd_builtin_type): Likewise.
10270 (arm_init_simd_builtin_types): Likewise.
10271 (arm_expand_builtin_args): Likewise.
10272 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
10273 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
10274 (ft32_setup_incoming_varargs): Likewise.
10275 (ft32_function_arg): Likewise.
10276 (ft32_function_arg_advance): Likewise.
10277 (ft32_pass_by_reference): Likewise.
10278 (ft32_arg_partial_bytes): Likewise.
10279 (ft32_valid_pointer_mode): Likewise.
10280 (ft32_addr_space_pointer_mode): Likewise.
10281 (ft32_addr_space_legitimate_address_p): Likewise.
10282 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
10284 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
10285 (ix86_emit_outlined_ms2sysv_restore): Likewise.
10286 (iamcu_alignment): Likewise.
10287 (canonicalize_vector_int_perm): Likewise.
10288 (ix86_noce_conversion_profitable_p): Likewise.
10289 (ix86_mpx_bound_mode): Likewise.
10290 (ix86_operands_ok_for_move_multiple): Likewise.
10291 * config/microblaze/microblaze-protos.h
10292 (microblaze_expand_conditional_branch_reg): Likewise.
10293 * config/microblaze/microblaze.c
10294 (microblaze_expand_conditional_branch_reg): Likewise.
10295 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10297 (rs6000_reassociation_width): Likewise.
10298 (rs6000_invalid_binary_op): Likewise.
10299 (fusion_p9_p): Likewise.
10300 (emit_fusion_p9_load): Likewise.
10301 (emit_fusion_p9_store): Likewise.
10302 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
10304 (riscv_hard_regno_mode_ok_p): Likewise.
10305 (riscv_address_insns): Likewise.
10306 (riscv_split_symbol): Likewise.
10307 (riscv_legitimize_move): Likewise.
10308 (riscv_function_value): Likewise.
10309 (riscv_hard_regno_nregs): Likewise.
10310 (riscv_expand_builtin): Likewise.
10311 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
10312 (riscv_build_integer): Likewise.
10313 (riscv_split_integer): Likewise.
10314 (riscv_legitimate_constant_p): Likewise.
10315 (riscv_cannot_force_const_mem): Likewise.
10316 (riscv_regno_mode_ok_for_base_p): Likewise.
10317 (riscv_valid_base_register_p): Likewise.
10318 (riscv_valid_offset_p): Likewise.
10319 (riscv_valid_lo_sum_p): Likewise.
10320 (riscv_classify_address): Likewise.
10321 (riscv_legitimate_address_p): Likewise.
10322 (riscv_address_insns): Likewise.
10323 (riscv_load_store_insns): Likewise.
10324 (riscv_force_binary): Likewise.
10325 (riscv_split_symbol): Likewise.
10326 (riscv_force_address): Likewise.
10327 (riscv_legitimize_address): Likewise.
10328 (riscv_move_integer): Likewise.
10329 (riscv_legitimize_const_move): Likewise.
10330 (riscv_legitimize_move): Likewise.
10331 (riscv_address_cost): Likewise.
10332 (riscv_subword): Likewise.
10333 (riscv_output_move): Likewise.
10334 (riscv_canonicalize_int_order_test): Likewise.
10335 (riscv_emit_int_order_test): Likewise.
10336 (riscv_function_arg_boundary): Likewise.
10337 (riscv_pass_mode_in_fpr_p): Likewise.
10338 (riscv_pass_fpr_single): Likewise.
10339 (riscv_pass_fpr_pair): Likewise.
10340 (riscv_get_arg_info): Likewise.
10341 (riscv_function_arg): Likewise.
10342 (riscv_function_arg_advance): Likewise.
10343 (riscv_arg_partial_bytes): Likewise.
10344 (riscv_function_value): Likewise.
10345 (riscv_pass_by_reference): Likewise.
10346 (riscv_setup_incoming_varargs): Likewise.
10347 (riscv_print_operand): Likewise.
10348 (riscv_elf_select_rtx_section): Likewise.
10349 (riscv_save_restore_reg): Likewise.
10350 (riscv_for_each_saved_reg): Likewise.
10351 (riscv_register_move_cost): Likewise.
10352 (riscv_hard_regno_mode_ok_p): Likewise.
10353 (riscv_hard_regno_nregs): Likewise.
10354 (riscv_class_max_nregs): Likewise.
10355 (riscv_memory_move_cost): Likewise.
10356 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
10357 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
10358 (rl78_addr_space_address_mode): Likewise.
10359 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10361 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10362 (rs6000_reassociation_width): Likewise.
10363 (rs6000_invalid_binary_op): Likewise.
10364 (fusion_p9_p): Likewise.
10365 (emit_fusion_p9_load): Likewise.
10366 (emit_fusion_p9_store): Likewise.
10367 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
10368 (ok_for_simple_move_operands): Likewise.
10369 (ok_for_simple_move_strict_operands): Likewise.
10370 (ok_for_simple_arith_logic_operands): Likewise.
10371 (visium_legitimize_reload_address): Likewise.
10372 (visium_select_cc_mode): Likewise.
10373 (output_cbranch): Likewise.
10374 (visium_split_double_move): Likewise.
10375 (visium_expand_copysign): Likewise.
10376 (visium_expand_int_cstore): Likewise.
10377 (visium_expand_fp_cstore): Likewise.
10378 * config/visium/visium.c (visium_pass_by_reference): Likewise.
10379 (visium_function_arg): Likewise.
10380 (visium_function_arg_advance): Likewise.
10381 (visium_libcall_value): Likewise.
10382 (visium_setup_incoming_varargs): Likewise.
10383 (visium_legitimate_constant_p): Likewise.
10384 (visium_legitimate_address_p): Likewise.
10385 (visium_legitimize_address): Likewise.
10386 (visium_secondary_reload): Likewise.
10387 (visium_register_move_cost): Likewise.
10388 (visium_memory_move_cost): Likewise.
10389 (prepare_move_operands): Likewise.
10390 (ok_for_simple_move_operands): Likewise.
10391 (ok_for_simple_move_strict_operands): Likewise.
10392 (ok_for_simple_arith_logic_operands): Likewise.
10393 (visium_function_value_1): Likewise.
10394 (rtx_ok_for_offset_p): Likewise.
10395 (visium_legitimize_reload_address): Likewise.
10396 (visium_split_double_move): Likewise.
10397 (visium_expand_copysign): Likewise.
10398 (visium_expand_int_cstore): Likewise.
10399 (visium_expand_fp_cstore): Likewise.
10400 (visium_split_cstore): Likewise.
10401 (visium_select_cc_mode): Likewise.
10402 (visium_split_cbranch): Likewise.
10403 (output_cbranch): Likewise.
10404 (visium_print_operand_address): Likewise.
10405 * expmed.c (flip_storage_order): Likewise.
10406 * expmed.h (emit_cstore): Likewise.
10407 (flip_storage_order): Likewise.
10408 * genrecog.c (validate_pattern): Likewise.
10409 * hsa-gen.c (gen_hsa_addr): Likewise.
10410 * internal-fn.c (expand_arith_overflow): Likewise.
10411 * ira-color.c (allocno_copy_cost_saving): Likewise.
10412 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10413 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10414 (process_invariant_for_inheritance): Likewise.
10415 * lra-eliminations.c (move_plus_up): Likewise.
10416 * omp-low.c (lower_oacc_reductions): Likewise.
10417 * simplify-rtx.c (simplify_subreg): Likewise.
10418 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10419 (TARGET_CHKP_BOUND_MODE): Likewise..
10420 * targhooks.c (default_chkp_bound_mode): Likewise.
10421 (default_setup_incoming_vararg_bounds): Likewise.
10422 * targhooks.h (default_chkp_bound_mode): Likewise.
10423 (default_setup_incoming_vararg_bounds): Likewise.
10424 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10425 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10426 (have_whole_vector_shift): Likewise.
10427 * tree-vect-stmts.c (vectorizable_load): Likewise.
10428 * doc/tm.texi: Regenerate.
10430 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10432 Graceful degrade if Binutils PR21472 is not available.
10435 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10436 .rodata in flash test fails.
10437 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10438 * confgure: Regenerate.
10439 * config.in: Regenerate.
10440 * config/avr/avr.c (avr_asm_named_section)
10441 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10442 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10443 (avr_asm_init_sections): Same.
10445 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10447 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10448 (fma<VH:mode>4_intrinsic): Likewise.
10449 (*fmsub<VCVTF:mode>4): Likewise.
10450 (*fmsub<VH:mode>4_intrinsic): Likewise.
10452 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10455 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10456 Don't depend on "optimize > 0".
10457 (out_movhi_r_mr, out_movqi_mr_r): Same.
10458 (out_movhi_mr_r, out_movqi_r_mr): Same.
10459 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10460 io_address_operand on "optimize > 0".
10462 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10464 * tree-loop-distribution.c: Add general explanantion on the pass.
10465 (generate_loops_for_partition): Mark distributed loop.
10466 (pg_add_dependence_edges): New parameter. Handle alias data
10467 dependence specially and record it in the parameter if asked.
10468 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10469 (init_partition_graph_vertices, add_partition_graph_edge): New.
10470 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10471 (free_partition_graph_vdata, build_partition_graph): New.
10472 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10473 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10474 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10475 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10476 (version_for_distribution_p, finalize_partitions): New.
10477 (distribute_loop): Handle alias data dependence specially. Factor
10478 out loop fusion code as functions and call these functions.
10480 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10482 * tree-loop-distribution.c (classify_partition): New parameter and
10483 better handle reduction statement.
10484 (rdg_build_partitions): Revise comment.
10485 (distribute_loop): Compute statements in all partitions and pass it
10486 to classify_partition.
10488 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10490 * tree-loop-distribution.c (enum partition_type): New.
10491 (struct partition): New field type.
10492 (partition_merge_into): Add parameter. Update partition type.
10493 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10494 (build_rdg_partition_for_vertex): Compute partition type.
10495 (rdg_build_partitions): Dump partition type.
10496 (distribute_loop): Update calls to partition_merge_into.
10498 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10500 * tree-loop-distribution.c (struct ddr_hasher): New.
10501 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10503 (classify_partition): Call get_data_dependence.
10504 (pg_add_dependence_edges): Ditto.
10505 (distribute_loop): Release data dependence hash table.
10507 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10509 * tree-loop-distribution.c (ref_base_address): Delete.
10510 (similar_memory_accesses): Rename ...
10511 (share_memory_accesses): ... to this. Check if partitions access
10512 the same memory reference.
10513 (distribute_loop): Call share_memory_accesses.
10515 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10517 * tree-loop-distribution.c (struct partition): New field recording
10518 its data reference.
10519 (partition_alloc, partition_free): Init and release data refs.
10520 (partition_merge_into): Merge data refs.
10521 (build_rdg_partition_for_vertex): Collect data refs for partition.
10522 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10524 (distribute_loop): Remve data refs from vertice data of partition
10527 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10529 * tree-loop-distribution.c (params.h): Include header file.
10530 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10531 (datarefs_vec): New global var.
10532 (create_rdg_vertices): Use datarefs_vec directly.
10533 (free_rdg): Don't free data references.
10534 (build_rdg): Update use. Don't free data references.
10535 (distribute_loop): Compute global variable for data references.
10536 Bail out if there are too many data references.
10538 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10540 * tree-loop-distribution.c (loop_nest): New global var.
10541 (build_rdg): Use loop directly, rather than loop nest.
10542 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10544 (distribute_loop): Compute global variable loop nest. Update use.
10546 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10548 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10549 (partition_merge_into): New parameter. Dump reason for fusion.
10550 (distribute_loop): Update use of partition_merge_into.
10552 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10554 * tree-loop-distribution.c (bb_top_order_index): New.
10555 (bb_top_order_index_size, bb_top_order_cmp): New.
10556 (stmts_from_loop): Use topological order.
10557 (pass_loop_distribution::execute): Compute and release topological
10558 order for basic blocks.
10560 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10562 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10565 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10567 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10568 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10569 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10570 * internal-fn.def (LOOP_DIST_ALIAS): New.
10571 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10572 (fold_loop_internal_call): ... this.
10573 (vect_loop_dist_alias_call): New function.
10574 (set_uid_loop_bbs): Call fold_loop_internal_call.
10575 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10578 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10581 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10582 Require dead FLAGS_REG at the beginning of a peephole.
10584 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10587 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10588 arguments in the call to __builtin_ia32_sbb_u32.
10589 (_subborrow_u64): Swap _X and _Y arguments in the call to
10590 __builtin_ia32_sbb_u64.
10592 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10595 * tree-vrp.c (compare_assert_loc): Turn into a function template
10596 with stable template parameter. Only test if a->e is NULL,
10597 !a->e == !b->e has been verified already. Use e == NULL or
10598 e != NULL instead of e or ! e tests. If stable is true, don't use
10599 iterative_hash_expr, on the other side allow a or b or both NULL
10600 and sort the NULLs last.
10601 (process_assert_insertions): Sort using compare_assert_loc<false>
10602 instead of compare_assert_loc, later sort using
10603 compare_assert_loc<true> before calling process_assert_insertions_for
10604 in a loop. Use break instead of continue once seen NULL pointer.
10606 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10608 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10609 Cortex-R7 and Cortex-R8 processors.
10611 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10613 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10614 uninitialized while src is not.
10616 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10618 * common/config/arm/arm-common.c: Adjust include path for
10620 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10621 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10622 (arm-cpu-data.h): Likewise.
10623 (arm-cpu-cdata.h): Likewise.
10624 * config/arm/arm-cpu.h: Delete.
10625 * config/arm/arm-cpu-cdata.h: Delete.
10626 * config/arm/arm-cpu-data.h: Delete.
10628 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10630 * config/arm/arm-cpus.in (cortex-a55): New.
10631 (cortex-a75): Likewise.
10632 (cortex-a75.cortex-a55): Likewise.
10633 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10635 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
10636 * config/arm/arm-cpu-cdata.h: Regenerate.
10637 * config/arm/arm-cpu-data.h: Regenerate.
10638 * config/arm/arm-cpu.h: Regenerate.
10639 * config/arm/arm-tables.opt: Regenerate.
10640 * config/arm/arm-tune.md: Regenerate.
10642 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10644 * haifa-sched.c (sched_create_recovery_edges): Update profile.
10646 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10648 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10651 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10653 PR tree-optimization/81292
10654 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10655 full_string_p, also call adjust_related_strinfos if the adjustment
10656 is simple, otherwise invalidate related strinfos.
10658 2017-07-04 Martin Liska <mliska@suse.cz>
10661 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10662 newly created variable as DECL_IGNORED_P.
10664 2017-07-04 Martin Liska <mliska@suse.cz>
10667 * ipa-inline.c (inline_small_functions):
10668 Use xstrdup_for_dump.
10670 2017-07-04 Tom de Vries <tom@codesourcery.com>
10672 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10674 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10677 * config/darwin.c (darwin_function_switched_text_sections):
10680 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10682 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10684 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10686 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10688 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10690 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10691 min_profitable_iters, and th as inclusive lower bounds.
10692 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10693 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10694 for min_profitable_iters and min_profitable_estimate.
10695 (vect_transform_loop): Treat th as an inclusive lower bound.
10696 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10698 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10701 * config/darwin.c (darwin_function_switched_text_sections):
10702 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10703 in two pieces, and suppress the use of buf.
10705 2017-07-03 Nathan Sidwell <nathan@acm.org>
10707 * hash-table.h (hash_table_mod1): Fix indentation.
10709 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10711 PR middle-end/81290
10712 * predict.c (force_edge_cold): Be more careful about propagation
10714 * profile-count.h (profile_probability::guessed,
10715 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10717 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10719 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10721 * doc/invoke.texi (rcpc architecture extension): Document it.
10723 2017-07-03 Richard Biener <rguenther@suse.de>
10725 PR tree-optimization/60510
10726 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10727 the scalar reduction PHI and use it.
10728 (vectorizable_reduction): Properly guard the single_defuse_cycle
10729 path for non-SLP reduction chains where we cannot use it.
10730 Rework reduc_def/index and vector type deduction. Rework
10731 vector operand gathering during reduction op code-gen.
10732 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10733 chains dissolve the chain and leave it to non-SLP reduction
10736 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10738 * tree-data-ref.h (dr_alignment): Declare.
10739 * tree-data-ref.c (dr_alignment): New function.
10740 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
10741 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
10743 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
10745 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10747 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
10748 and base_misalignment fields.
10749 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
10750 * tree-data-ref.c: Include builtins.h.
10751 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
10752 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
10753 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
10754 * tree-vect-data-refs.c: Include tree-cfg.h.
10755 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
10756 fields instead of calculating an alignment here.
10757 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
10758 innermost_loop_behavior fields.
10760 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10762 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
10764 (DR_STEP_ALIGNMENT): New macro.
10765 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
10766 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
10767 (create_data_ref): Print it.
10768 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
10769 to tell whether the step preserves vector (mis)alignment.
10770 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10771 Move the check for an integer step and generalise to all INTEGER_CST.
10772 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
10773 Print the outer step alignment.
10775 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10777 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
10778 with offset_alignment.
10779 (DR_ALIGNED_TO): Delete.
10780 (DR_OFFSET_ALIGNMENT): New macro.
10781 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
10782 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
10783 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
10784 (create_data_ref): Likewise.
10785 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10786 (vect_analyze_data_refs): Likewise.
10787 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
10788 creating dummy innermost behavior.
10790 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10792 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
10793 with a "innermost_loop_behavior *" and refeence tree.
10794 * tree-data-ref.c (dr_analyze_innermost): Likewise.
10795 (create_data_ref): Update call accordingly.
10796 * tree-predcom.c (find_looparound_phi): Likewise.
10798 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10800 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
10801 fields with dr_wrt_vec_loop.
10802 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
10803 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
10804 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
10805 (vect_dr_behavior): New function.
10806 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10807 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
10808 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
10809 track whether the step preserves the misalignment.
10810 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10811 Use vect_dr_behavior.
10812 (vect_setup_realignment): Update call accordingly.
10813 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
10814 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
10815 call to vect_create_addr_base_for_vector_ref.
10816 (vect_create_cond_for_align_checks): Likewise.
10817 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
10818 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
10819 (vect_recog_mask_conversion_pattern): Likewise.
10820 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
10821 (new_stmt_vec_info): Remove redundant zeroing.
10823 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10825 * common/config/arm/arm-common.c (arm_be8_option): New function.
10826 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
10827 (ISA_ARMv6): Add isa_bit_be8.
10828 * config/arm/arm.h (arm_be8_option): Add prototype.
10829 (BE8_SPEC_FUNCTION): New define.
10830 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
10831 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
10832 (mlittle-endian): Similarly.
10833 (mbe8, mbe32): New options.
10834 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
10835 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
10837 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10839 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
10841 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10843 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
10844 (cleanup_tree_cfg_bb): Use it.
10845 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
10847 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
10849 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10852 * loop-doloop.c (add_test): Update profile.
10854 2017-07-03 Martin Liska <mliska@suse.cz>
10857 * sanopt.c (rewrite_usage_of_param): New function.
10858 (sanitize_rewrite_addressable_params): Likewise.
10859 (pass_sanopt::execute): Call rewrite_usage_of_param.
10861 2017-07-03 Richard Biener <rguenther@suse.de>
10863 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
10864 back to using VIEW_CONVERT_EXPR.
10866 2017-07-03 Martin Liska <mliska@suse.cz>
10869 * doc/extend.texi: Document when a resolver function is
10870 generated for target_clones.
10872 2017-07-03 Martin Liska <mliska@suse.cz>
10874 * asan.c (asan_emit_stack_protection): Unpoison just red zones
10875 and shadow memory of auto variables which are subject of
10876 use-after-scope sanitization.
10877 (asan_expand_mark_ifn): Add do set only when is_poison.
10879 2016-07-03 Richard Biener <rguenther@suse.de>
10881 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
10883 (vect_force_simple_reduction): Record reduction def -> phi mapping.
10884 (vectorizable_reduction): Perform reduction PHI creation when
10885 visiting a reduction PHI and adjust and simplify code generation
10886 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
10887 (vect_transform_loop): Visit reduction PHIs.
10888 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
10889 defs into the SLP tree.
10890 (vect_build_slp_tree): Reduction defs terminate the recursion.
10891 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
10893 (vect_get_vec_defs_for_stmt_copy): Export.
10894 (vect_get_vec_defs): Likewise.
10895 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
10897 (vect_get_vec_defs_for_stmt_copy): Declare.
10898 (vect_get_vec_defs): Likewise.
10900 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10902 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
10903 parameter with a "loop" parameter and use it instead of the
10904 loop containing DR_STMT. Don't check simple_iv when doing
10905 BB analysis. Describe the two analysis modes in the comment.
10907 2017-07-03 Tom de Vries <tom@codesourcery.com>
10909 PR tree-optimization/69468
10910 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
10911 (find_same_succ_bb): Handle ignore_edge_flags.
10913 2017-07-03 Tom de Vries <tom@codesourcery.com>
10915 PR tree-optimization/81192
10916 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
10918 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
10920 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
10922 2017-07-03 Tom de Vries <tom@codesourcery.com>
10924 PR tree-optimization/81192
10925 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
10926 BB_SAME_SUCC (bb) == NULL.
10928 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10930 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
10933 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10935 * dumpfile.c: Include profile-count.h
10936 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
10938 (insert_cond_bb): Update profile.
10939 * tree-cfg.h (insert_cond_bb): Update prototype.
10940 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
10941 * tree-dump.c: Do not include tree-cfg.
10943 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10945 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
10947 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10949 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
10952 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10954 * tree-complex.c (expand_complex_div_wide): update profile.
10956 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10957 Alan Hayward <alan.hayward@arm.com>
10958 David Sherwood <david.sherwood@arm.com>
10960 * Makefile.in (MACHMODE_H): Remove insn-modes.h
10961 (CORETYPES_H): New define.
10962 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
10963 (insn-modes-inline.h, s-modes-inline-h): New rules.
10964 (generated_files): Add insn-modes-inline.h.
10965 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
10966 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
10967 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
10968 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
10969 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
10970 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
10971 (build/gencodes.o, build/genconditions.o): Likewise.
10972 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
10973 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
10974 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
10975 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
10976 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
10977 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
10978 * coretypes.h: Include everything up to real.h for generators.
10979 Include insn-modes.h first. Include wide-int-print.h after
10980 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10981 * machmode.h: Don't include insn-modes.h here.
10982 * function-tests.c: Remove includes of signop.h, machmode.h,
10983 double-int.h and wide-int.h.
10985 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10987 * optc-save-gen.awk: Likewise.
10988 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10989 * godump.c: Remove include of wide-int-print.h.
10990 * pretty-print.h: Likewise.
10991 * wide-int-print.cc: Likewise.
10992 * wide-int.cc: Likewise.
10993 * hash-map-tests.c: Remove include of signop.h.
10994 * hash-set-tests.c: Likewise.
10995 * rtl-tests.c: Likewise.
10996 * mkconfig.sh: Remove include of machmode.h.
10997 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
10999 (emit_insn_modes_inline_h): ...this new function. Emit the code
11000 into an insn-modes-inline.h header file, adding appropriate
11001 include guards and end comments.
11002 (emit_insn_modes_c_header): Remove include of machmode.h.
11003 (emit_min_insn_modes_c_header): Include coretypes.h rather than
11005 (main): Handle -i flag and call emit_insn_modes_inline_h when
11008 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11010 * tree-ssa-strlen.c (strinfo): Rename the length field to
11011 nonzero_chars. Add a full_string_p field.
11012 (compare_nonzero_chars, zero_length_string_p): New functions.
11013 (get_addr_stridx): Add an offset_out parameter.
11014 Use compare_nonzero_chars.
11015 (get_stridx): Update accordingly. Use compare_nonzero_chars.
11016 (new_strinfo): Update after above changes to strinfo.
11017 (set_endptr_and_length): Set full_string_p.
11018 (get_string_length): Update after above changes to strinfo.
11019 (unshare_strinfo): Update call to new_strinfo.
11020 (maybe_invalidate): Likewise.
11021 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
11022 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
11023 as a uhwi instead of an shwi. Update after above changes to
11024 strinfo and new_strinfo.
11025 (zero_length_string): Assert that chainsi contains full strings.
11026 Use zero_length_string_p. Update call to new_strinfo.
11027 (adjust_related_strinfos): Update after above changes to strinfo.
11028 Copy full_string_p from origsi.
11029 (adjust_last_stmt): Use zero_length_string_p.
11030 (handle_builtin_strlen): Update after above changes to strinfo and
11031 new_strinfo. Install the lhs as the string length if the previous
11032 entry didn't describe a full string.
11033 (handle_builtin_strchr): Update after above changes to strinfo
11035 (handle_builtin_strcpy): Likewise.
11036 (handle_builtin_strcat): Likewise.
11037 (handle_builtin_malloc): Likewise.
11038 (handle_pointer_plus): Likewise.
11039 (handle_builtin_memcpy): Likewise. Track nonzero characters
11040 that aren't necessarily followed by a nul terminator.
11041 (handle_char_store): Likewise.
11043 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11045 PR tree-optimization/80769
11046 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
11047 for malloc and calloc. Document the new invariant that all related
11048 strinfos have delayed lengths or none do.
11049 (verify_related_strinfos): Move earlier in file.
11050 (set_endptr_and_length): New function, split out from...
11051 (get_string_length): ...here. Also set the lengths of related
11053 (zero_length_string): Assert that chainsi has known (rather than
11055 (adjust_related_strinfos): Likewise.
11057 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
11059 PR tree-optimization/81136
11060 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
11061 assert that two references with the same misalignment have the same
11062 compile-time misalignment if those compile-time misalignments
11065 2017-07-01 Andi Kleen <ak@linux.intel.com>
11067 * print-tree.c (print_node): Print all attributes.
11069 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11071 * cfg.c (scale_bbs_frequencies): New function.
11072 * cfg.h (scale_bbs_frequencies): Declare it.
11073 * cfgloopanal.c (single_likely_exit): Cleanup.
11074 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
11076 (scale_loop_profile): Likewise.
11077 (loop_version): Likewise.
11078 (create_empty_loop_on_edge): Update.
11079 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
11080 scale_loop_frequencies, scale_loop_profile, loopify,
11081 loop_version): Update prototypes.
11082 * modulo-sched.c (sms_schedule): Update.
11083 * predict.c (unlikely_executed_edge_p): Also check probability.
11084 (probably_never_executed_edge_p): Fix typo.
11085 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11086 * tree-parloops.c (gen_parallel_loop): Update.
11087 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
11088 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11089 * tree-ssa-loop-split.c (split_loop): Update.
11090 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11091 * tree-vect-loop-manip.c (vect_do_peeling): Update.
11092 (vect_loop_versioning): Update.
11093 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11095 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11097 * trans-mem.c (split_bb_make_tm_edge): Update profile.
11099 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11101 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
11102 to keep profile consistent.
11104 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11106 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
11107 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
11108 * profile-count.h (max_safe_multiplier): Make unsigned.
11109 (profile_count::guessed_zero): New.
11111 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11113 * bb-reorder.c (fix_up_crossing_landing_pad,
11114 fix_crossing_conditional_branches): Use make_single_succ_edge
11115 to keep profile consistent.
11117 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11119 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
11122 2017-07-01 Jakub Jelinek <jakub@redhat.com>
11125 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
11126 the right scopes, make sure cond_jump isn't preserved between multiple
11127 iterations. Search for fallthru edge whenever there are 3+ edges and
11128 use find_fallthru_edge for it.
11130 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11132 Patch by Alexander Monakov <amonakov@ispras.ru>
11133 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
11134 probabilities consistently.
11136 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11138 * pa.c (pa_expand_compare_and_swap_loop): Update call of
11139 emit_cmp_and_jump_insns.
11141 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11144 * tree-inline.c (expand_call_inline): Combine profile statuses.
11146 2017-06-30 Andrew Pinski <apinski@cavium.com>
11148 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
11149 fold_stmt returned true.
11151 2017-06-30 Nathan Sidwell <nathan@acm.org>
11153 * ggc.h (empty_string): Delete.
11154 * cfgexpand.c (expand_asm_stmt): Use plain "".
11155 * optabs.c (expand_asm_memory_barrier): Likewise.
11156 * stringpool.c (empty_string): Delete.
11157 (digit_vector, digit_string): Delete.
11158 (ggc_alloc_string): Use plain "", don't optimize single digit
11159 strings. Use ggc_alloc_atomic.
11161 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
11163 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
11164 comparison set and one other set, use the cost of the non-comparison
11167 2017-06-30 Nathan Sidwell <nathan@acm.org>
11169 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
11172 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
11174 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
11175 loops. Remove now unneeded calls to gimple_switch_set_label() that
11176 just set removed labels to NULL_TREE.
11178 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
11180 * tree-ssanames.c (set_range_info_raw): Abstract from ...
11181 (set_range_info): ...here. Only call set_range_info_raw if domain
11183 (set_nonzero_bits): Call set_range_info_raw.
11184 * tree-ssanames.h (set_range_info_raw): New.
11186 2017-06-30 Jakub Jelinek <jakub@redhat.com>
11189 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
11190 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
11191 of nonimmediate_operand and <store_mask_constraint> instead of m
11192 for the input operand. For V8FI iterator, always split if input
11193 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
11194 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
11195 <store_mask_predicate> instead of register_operand and
11196 <store_mask_constraint> instead of v for the input operand. Make
11197 sure both operands aren't MEMs for if not <mask_applied>.
11199 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
11201 * lto-wrapper.c (copy_file) Close both file descriptors before
11204 2017-06-30 Martin Liska <mliska@suse.cz>
11207 * multiple_target.c (create_dispatcher_calls): Make ifunc
11208 also for function that don't have calls or are not referenced.
11210 2017-06-30 Richard Biener <rguenther@suse.de>
11212 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
11213 analyze the first scalar stmt. Move vector type computation
11214 for the BB case here from ...
11215 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
11216 live operation processing in the SLP case properly.
11218 2017-06-30 Richard Biener <rguenther@suse.de>
11220 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
11222 2017-06-30 Martin Liska <mliska@suse.cz>
11225 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
11226 before BUILT_IN_UNWIND_RESUME when ASAN is used.
11228 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
11230 * doc/invoke.texi (AArch64): Add missing options and remove redundant
11233 2017-06-30 Richard Biener <rguenther@suse.de>
11235 PR tree-optimization/81249
11236 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
11237 condition reduction result to original scalar type.
11239 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11241 * profile-count.h (enum profile_quality): Fix typos and whitespace
11244 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11246 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
11247 type for branch probabilities.
11249 2017-06-29 Julian Brown <julian@codesourcery.com>
11250 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11252 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
11253 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
11254 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
11255 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
11257 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11259 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
11260 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
11261 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
11262 CC usage from generic code to here.
11263 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
11264 CC usage into the target macros.
11266 2017-06-29 Maya Rashish <coypu@sdf.org>
11268 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
11271 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11273 * arm/arm-builtins.c: Include profile-count.h
11274 * except.c (sjlj_emit_function_enter): Use
11275 profile_probability::unlikely.
11277 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11279 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
11280 and tocrel_offset be pointer args rather than implicitly using
11282 (legitimate_constant_pool_address_p, rs6000_emit_move,
11283 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
11284 tocrel_offset and use in toc_relative_expr_p call.
11285 (print_operand, print_operand_address): Use static tocrel_base_oac
11286 and tocrel_offset_oac.
11287 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
11290 2017-06-29 Maya Rashish <coypu@sdf.org>
11292 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
11294 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
11296 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
11297 objects, take into account only the alignment of 'op0' and 'mode1' if
11300 2017-06-29 Steve Ellcey <sellcey@cavium.com>
11302 * ccmp.c (ccmp_tree_comparison_p): New function.
11303 (ccmp_candidate_p): Update to use above function.
11304 (get_compare_parts): New function.
11305 (expand_ccmp_next): Update to use new functions.
11306 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
11308 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
11309 take mode as argument.
11310 * ccmp.h (expand_ccmp_expr): Add mode as argument.
11311 * expr.c (expand_expr_real_1): Pass mode as argument.
11313 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
11315 * combine.c (combine_instructions): Print insns to dump_file, together
11318 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11320 * asan.c (asan_emit_stack_protection): Update.
11321 (create_cond_insert_point): Update.
11322 * auto-profile.c (afdo_propagate_circuit): Update.
11323 * basic-block.h (struct edge_def): Turn probability to
11324 profile_probability.
11325 (EDGE_FREQUENCY): Update.
11326 * bb-reorder.c (find_traces_1_round): Update.
11327 (better_edge_p): Update.
11328 (sanitize_hot_paths): Update.
11329 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
11330 (make_single_succ_edge): Update.
11331 (check_bb_profile): Update.
11332 (dump_edge_info): Update.
11333 (update_bb_profile_for_threading): Update.
11334 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
11335 probabilitycount to 0.
11336 * cfgbuild.c (compute_outgoing_frequencies): Update.
11337 * cfgcleanup.c (try_forward_edges): Update.
11338 (outgoing_edges_match): Update.
11339 (try_crossjump_to_edge): Update.
11340 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
11341 (expand_gimple_tailcall): Update.
11342 (construct_init_block): Use make_single_succ_edge.
11343 (construct_exit_block): Use make_single_succ_edge.
11344 * cfghooks.c (verify_flow_info): Update.
11345 (redirect_edge_succ_nodup): Update.
11346 (split_edge): Update.
11347 (account_profile_record): Update.
11348 * cfgloopanal.c (single_likely_exit): Update.
11349 * cfgloopmanip.c (scale_loop_profile): Update.
11350 (set_zero_probability): Remove.
11351 (duplicate_loop_to_header_edge): Update.
11352 * cfgloopmanip.h (loop_version): Update prototype.
11353 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11354 (force_nonfallthru_and_redirect): Update.
11355 (update_br_prob_note): Update.
11356 (rtl_verify_edges): Update.
11357 (purge_dead_edges): Update.
11358 (rtl_lv_add_condition_to_bb): Update.
11359 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
11360 * cgraphunit.c (init_lowered_empty_function): Update.
11361 (cgraph_node::expand_thunk): Update.
11362 * cilk-common.c: Include profile-count.h
11363 * dojump.c (inv): Remove.
11364 (jumpifnot): Update.
11365 (jumpifnot_1): Update.
11366 (do_jump_1): Update.
11368 (do_jump_by_parts_greater_rtx): Update.
11369 (do_compare_rtx_and_jump): Update.
11370 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
11371 do_jump_1. do_compare_rtx_and_jump): Update prototype.
11372 * dwarf2cfi.c: Include profile-count.h
11373 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
11374 (sjlj_emit_dispatch_table): Likewise.
11375 * explow.c: Include profile-count.h
11376 * expmed.c (emit_store_flag_force): Update.
11377 (do_cmp_and_jump): Update.
11378 * expr.c (compare_by_pieces_d::generate): Update.
11379 (compare_by_pieces_d::finish_mode): Update.
11380 (emit_block_move_via_loop): Update.
11381 (store_expr_with_bounds): Update.
11382 (store_constructor): Update.
11383 (expand_expr_real_2): Update.
11384 (expand_expr_real_1): Update.
11385 * expr.h (try_casesi, try_tablejump): Update prototypes.
11386 * gimple-pretty-print.c (dump_probability): Update.
11387 (dump_profile): New.
11388 (dump_gimple_label): Update.
11389 (dump_gimple_bb_header): Update.
11390 * graph.c (draw_cfg_node_succ_edges): Update.
11391 * hsa-gen.c (convert_switch_statements): Update.
11392 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
11393 (find_if_case_1): Update.
11394 (find_if_case_2): Update.
11395 * internal-fn.c (expand_arith_overflow_result_store): Update.
11396 (expand_addsub_overflow): Update.
11397 (expand_neg_overflow): Update.
11398 (expand_mul_overflow): Update.
11399 (expand_vector_ubsan_overflow): Update.
11400 * ipa-cp.c (good_cloning_opportunity_p): Update.
11401 * ipa-split.c (split_function): Use make_single_succ_edge.
11402 * ipa-utils.c (ipa_merge_profiles): Update.
11403 * loop-doloop.c (add_test): Update.
11404 (doloop_modify): Update.
11405 * loop-unroll.c (compare_and_jump_seq): Update.
11406 (unroll_loop_runtime_iterations): Update.
11407 * lra-constraints.c (lra_inheritance): Update.
11408 * lto-streamer-in.c (input_cfg): Update.
11409 * lto-streamer-out.c (output_cfg): Update.
11410 * mcf.c (adjust_cfg_counts): Update.
11411 * modulo-sched.c (sms_schedule): Update.
11412 * omp-expand.c (expand_omp_for_init_counts): Update.
11413 (extract_omp_for_update_vars): Update.
11414 (expand_omp_ordered_sink): Update.
11415 (expand_omp_for_ordered_loops): Update.
11416 (expand_omp_for_generic): Update.
11417 (expand_omp_for_static_nochunk): Update.
11418 (expand_omp_for_static_chunk): Update.
11419 (expand_cilk_for): Update.
11420 (expand_omp_simd): Update.
11421 (expand_omp_taskloop_for_outer): Update.
11422 (expand_omp_taskloop_for_inner): Update.
11423 * omp-simd-clone.c (simd_clone_adjust): Update.
11424 * optabs.c (expand_doubleword_shift): Update.
11425 (expand_abs): Update.
11426 (emit_cmp_and_jump_insn_1): Update.
11427 (expand_compare_and_swap_loop): Update.
11428 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11429 * predict.c (predictable_edge_p): Update.
11430 (edge_probability_reliable_p): Update.
11431 (set_even_probabilities): Update.
11432 (combine_predictions_for_insn): Update.
11433 (combine_predictions_for_bb): Update.
11434 (propagate_freq): Update.
11435 (estimate_bb_frequencies): Update.
11436 (force_edge_cold): Update.
11437 * profile-count.c (profile_count::dump): Add missing space into dump.
11438 (profile_count::debug): Add newline.
11439 (profile_count::differs_from_p): Explicitly convert to unsigned.
11440 (profile_count::stream_in): Update.
11441 (profile_probability::dump): New member function.
11442 (profile_probability::debug): New member function.
11443 (profile_probability::differs_from_p): New member function.
11444 (profile_probability::differs_lot_from_p): New member function.
11445 (profile_probability::stream_in): New member function.
11446 (profile_probability::stream_out): New member function.
11447 * profile-count.h (profile_count_quality): Rename to ...
11448 (profile_quality): ... this one.
11449 (profile_probability): New.
11450 (profile_count): Update.
11451 * profile.c (compute_branch_probabilities): Update.
11452 * recog.c (peep2_attempt): Update.
11453 * sched-ebb.c (schedule_ebbs): Update.
11454 * sched-rgn.c (find_single_block_region): Update.
11455 (compute_dom_prob_ps): Update.
11456 (schedule_region): Update.
11457 * sel-sched-ir.c (compute_succs_info): Update.
11458 * stmt.c (struct case_node): Update.
11459 (do_jump_if_equal): Update.
11460 (get_outgoing_edge_probs): Update.
11461 (conditional_probability): Update.
11462 (emit_case_dispatch_table): Update.
11463 (expand_case): Update.
11464 (expand_sjlj_dispatch_table): Update.
11465 (emit_case_nodes): Update.
11466 * targhooks.c: Update.
11467 * tracer.c (better_p): Update.
11468 (find_best_successor): Update.
11469 * trans-mem.c (expand_transaction): Update.
11470 * tree-call-cdce.c: Update.
11471 * tree-cfg.c (gimple_split_edge): Upate.
11472 (move_sese_region_to_fn): Upate.
11473 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11474 * tree-eh.c (lower_resx): Upate.
11475 (cleanup_empty_eh_move_lp): Upate.
11476 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11477 * tree-inline.c (copy_edges_for_bb): Update.
11478 (copy_cfg_body): Update.
11479 * tree-parloops.c (gen_parallel_loop): Update.
11480 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11481 (gimple_gen_time_profiler): Update.
11482 * tree-ssa-dce.c (remove_dead_stmt): Update.
11483 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11484 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11485 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11486 (unloop_loops): Update.
11487 (try_peel_loop): Update.
11488 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11489 * tree-ssa-loop-split.c (connect_loops): Update.
11490 (split_loop): Update.
11491 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11492 (hoist_guard): Update.
11493 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11494 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11495 (value_replacement): Update.
11496 * tree-ssa-reassoc.c (branch_fixup): Update.
11497 * tree-ssa-tail-merge.c (replace_block_by): Update.
11498 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11499 (create_edge_and_update_destination_phis): Update.
11500 (compute_path_counts): Update.
11501 (recompute_probabilities): Update.
11502 (update_joiner_offpath_counts): Update.
11503 (freqs_to_counts_path): Update.
11504 (duplicate_thread_path): Update.
11505 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11506 (struct switch_conv_info): Update.
11507 (gen_inbound_check): Update.
11508 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11509 (vect_do_peeling): Update.
11510 (vect_loop_versioning): Update.
11511 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11512 (optimize_mask_stores): Update.
11513 * ubsan.c (ubsan_expand_null_ifn): Update.
11514 * value-prof.c (gimple_divmod_fixed_value): Update.
11515 (gimple_divmod_fixed_value_transform): Update.
11516 (gimple_mod_pow2): Update.
11517 (gimple_mod_pow2_value_transform): Update.
11518 (gimple_mod_subtract): Update.
11519 (gimple_mod_subtract_transform): Update.
11520 (gimple_ic): Update.
11521 (gimple_stringop_fixed_value): Update.
11522 (gimple_stringops_transform): Update.
11523 * value-prof.h: Update.
11525 2017-06-29 Carl Love <cel@us.ibm.com>
11527 * config/rs6000/rs6000-c.c: Add support for built-in functions
11528 vector signed int vec_signed (vector float);
11529 vector signed long long vec_signed (vector double);
11530 vector signed int vec_signed2 (vector double, vector double);
11531 vector signed int vec_signede (vector double);
11532 vector signed int vec_signedo (vector double);
11533 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11534 instruction generator.
11535 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11536 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11537 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11539 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11540 vunsignede_v2df): Add define_expands.
11541 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11542 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11543 VEC_UNSIGNEDO): Add definitions.
11544 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11545 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11546 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11547 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11548 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11549 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11550 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11551 vec_unsignede, vec_unsignedo): Add builtin defines.
11552 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11554 * doc/extend.texi: Update the built-in documentation file for the
11555 new built-in functions.
11557 2017-06-29 Richard Biener <rguenther@suse.de>
11559 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11560 reduction chains to LOOP_VINFO_REDUCTIONS.
11561 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11562 SLP reductions after processing reduction chains.
11564 2017-06-29 Nathan Sidwell <nathan@acm.org>
11566 * builtins.c (fold_builtin_FUNCTION): Use
11567 lang_hooks.decl_printable_name.
11569 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11571 PR middle-end/81194
11572 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11573 with only one label.
11574 * stmt.c (expand_case): Assert NCASES is greater than one.
11576 2017-06-29 Richard Biener <rguenther@suse.de>
11578 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11580 (group_case_labels): Likewise.
11581 (find_taken_edge): Push sanity checking on val to workers...
11582 (find_taken_edge_cond_expr): ... here
11583 (find_taken_edge_switch_expr): ... and here, handle cases
11584 with just a default label.
11585 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11586 (group_case_labels): Likewise.
11587 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11588 group_case_labels does anything cleanup the CFG again.
11590 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11592 PR tree-optimization/81196
11593 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11594 exit condition comparing two IVs.
11596 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11598 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11599 profile to the dummy entry at the end of the list of architectures.
11600 * config/arm/arm-cpu-cdata.h: Regenerated.
11602 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11603 Michael Collison <michael.collison@arm.com>
11606 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11608 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11609 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11610 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11611 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11612 of shift when the shift amount is masked with constant equal to
11613 the size of the mode.
11614 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11617 2017-06-29 Martin Liska <mliska@suse.cz>
11619 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11621 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11624 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11627 (edge_growth_cache_entry::edge_growth_cache_entry): New
11629 (reset_edge_growth_cache): Update to use constructor.
11631 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11633 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11634 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11635 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11637 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
11639 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
11640 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
11642 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
11644 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
11645 (*-linux-uclibc*): Add t-uclibc tmake_file.
11646 * config/t-musl: New.
11647 * config/t-uclibc: New.
11649 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11651 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11653 (gen_comm_data): Emit architectural setting of arch_prof.
11654 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11656 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11657 (armv8-m.base, armv8-m.main): Likewise.
11658 * arm-protos.h (arm_build_target): Add profile field.
11659 (arch_option): Likewise.
11660 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11662 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11663 arm_active_target.profile.
11665 2017-06-28 Richard Biener <rguenther@suse.de>
11667 PR middle-end/81227
11668 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11669 TYPE_OVERFLOW_WRAPS.
11670 * match.pd (negate_expr_p): Likewise.
11671 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11672 fold_build2, not fold_binary.
11674 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11676 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11677 Convert memory address to Pmode.
11678 (aarch64_print_operand): Assert MEM operands are always Pmode.
11680 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11683 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11684 Remove redundant if.
11685 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11686 * config/arm/aarch-common-protos.h
11687 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11688 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11690 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11693 * multiple_target.c (create_dispatcher_calls): Set the default
11694 clone to be static, not public.
11696 2017-06-28 Richard Biener <rguenther@suse.de>
11698 * tree-vect-loop.c (vectorizable_reduction): Move special
11699 cond reduction IV var creation ...
11700 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11701 parameter. Use STMT_VINFO_VECTYPE.
11702 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11705 2017-06-28 Martin Liska <mliska@suse.cz>
11708 * ipa-visibility.c (non_local_p): Handle visibility.
11710 2017-06-28 Martin Liska <mliska@suse.cz>
11713 * common.opt: Add IntegerRange to various options.
11714 * opt-functions.awk (integer_range_info): New function.
11715 * optc-gen.awk: Add integer_range_info to cl_options struct.
11716 * opts-common.c (decode_cmdline_option): Handle
11717 CL_ERR_INT_RANGE_ARG.
11718 (cmdline_handle_error): Likewise.
11719 * opts.c (print_filtered_help): Show valid interval in
11720 when --help is provided.
11721 * opts.h (struct cl_option): Add range_min and range_max fields.
11722 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11724 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11726 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11727 (x * C EQ/NE y * C): New transformation.
11729 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11731 * genmultilib (combination_space): Accept '+' in option names.
11733 2017-06-28 Martin Liska <mliska@suse.cz>
11736 * asan.c (instrument_derefs): Bail out inner references
11737 that are hard register variables.
11739 2017-06-28 Jakub Jelinek <jakub@redhat.com>
11742 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
11743 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
11745 2017-06-28 Richard Biener <rguenther@suse.de>
11747 * tree-vectorizer.h (vect_get_vec_defs): Remove.
11748 (vect_get_slp_defs): Adjust.
11749 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
11751 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
11753 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11754 get_initial_defs_for_reduction instead of vect_get_vec_defs.
11755 (vectorizable_reduction): Adjust.
11756 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
11758 (vect_get_slp_defs): Likewise.
11759 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
11760 (vectorizable_bswap): Adjust.
11761 (vectorizable_call): Likewise.
11762 (vectorizable_conversion): Likewise.
11763 (vectorizable_assignment): Likewise.
11764 (vectorizable_shift): Likewise.
11765 (vectorizable_operation): Likewise.
11766 (vectorizable_store): Likewise.
11767 (vectorizable_condition): Likewise.
11768 (vectorizable_comparison): Likewise.
11770 2017-06-28 Michael Collison <michael.collison@arm.com>
11773 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
11775 (arm_gen_movmemqi): Removed unused variable 'i'.
11776 Convert 'for' loop into 'while' loop.
11777 (arm_expand_prologue): Remove last unnecessary set of insn.
11778 (thumb_pop): Remove unused variable 'pushed_words'.
11779 (thumb_exit): Remove last unnecessary set of regs_to_pop.
11781 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11783 * config/s390/predicates.md: Use s390_rel_address_ok_p.
11784 * config/s390/s390-protos.h: Add prototype of
11785 s390_rel_address_ok_p.
11786 * config/s390/s390.c (s390_got_symbol): New function.
11787 (s390_rel_address_ok_p): New function.
11788 (legitimize_pic_address): Use s390_rel_address_ok_p.
11789 (s390_load_got): Use s390_got_symbol.
11790 (s390_option_override): Issue error if
11791 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
11792 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
11794 * config/s390/s390.opt: New option mpic-data-is-text-relative.
11796 2017-06-27 Andrew Pinski <apinski@cavium.com>
11798 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
11799 (X * copysign (1.0, X)): New pattern.
11800 (X * copysign (1.0, -X)): New pattern.
11801 (copysign (-1.0, CST)): New pattern.
11803 2017-06-27 Joseph Myers <joseph@codesourcery.com>
11805 * genmultilib (combination_space): Remove variable.
11806 Validate reuse rules against regular expression for any sequence
11807 of multilib options in any order.
11809 2017-06-27 Michael Collison <michael.collison@arm.com>
11811 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
11812 call aarch64_split_simd_combine.
11813 * (aarch64_combine_internal<mode>): Delete pattern.
11814 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
11815 Allow register and subreg operands.
11817 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11819 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
11820 specific need, just fallback on defaults.
11821 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
11823 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11824 Olivier Hainque <hainque@adacore.com>
11826 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
11828 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
11829 targets. Pick a default if no particular attempt applied.
11830 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
11833 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11835 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
11836 (x86_64-wrs-vxworks7): Likewise.
11838 2017-06-27 Marek Polacek <polacek@redhat.com>
11841 * ubsan.c (instrument_null): Check get_base_address's result for null.
11843 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11845 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
11847 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11849 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
11850 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
11851 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
11852 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
11853 New function types.
11854 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
11855 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
11856 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
11857 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
11858 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
11859 BUILT_IN_FEUPDATEENV): New builtins.
11860 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
11861 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
11862 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
11863 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
11865 (builtin_structptr_types): Adjust size.
11866 * tree.c (builtin_structptr_types): Add four entries.
11868 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11869 Olivier Hainque <hainque@adacore.com>
11871 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
11872 (TLS_SYM): New local macro, forcing reference to __tls__ on
11873 link command lines for VxWorks 7 RTPs, triggering initialization
11875 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
11876 OS features TLS support, true for RTPs on VxWorks 7.
11877 * config/vxworks.c (vxworks_override_options): Setup emutls
11880 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11882 * predict.c (test_prediction_value_range): Use -1U instead of -1
11883 to avoid narrowing conversion warning.
11884 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
11885 to avoid narrowing conversion warning.
11886 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
11888 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
11890 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11892 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
11893 64bit configurations.
11894 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
11895 (SIZE_TYPE): Likewise.
11896 * config/vxworks.c (vxworks_emutls_var_fields): Use
11897 long_unsigned_type_node instead of unsigned_type_node as the offset
11898 field type, which is "pointer" mode in emutls.c.
11900 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11903 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
11905 PR middle-end/81207
11906 * gimple-fold.c (replace_call_with_call_and_fold): Handle
11907 gimple_vuse copying separately from gimple_vdef copying.
11909 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11911 * value-prof.c (free_hist): Remove call to memset and the enclosing if
11914 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11915 Olivier Hainque <hainque@adacore.com>
11917 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
11918 for all vxworks7 targets.
11919 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
11920 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
11921 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
11922 variations for VX6/VX7 and 32/64bits later on in ...
11923 (VXWORKS_LIB_SPEC): Leverage new macros.
11924 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
11925 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
11927 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11929 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
11930 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
11932 2017-06-26 Carl Love <cel@us.ibm.com>
11934 * config/rs6000/rs6000-c.c: Add support for built-in functions
11935 vector bool char vec_reve (vector bool char);
11936 vector signed char vec_reve (vector signed char);
11937 vector unsigned char vec_reve (vector unsigned char);
11938 vector bool int vec_reve (vector bool int);
11939 vector signed int vec_reve (vector signed int);
11940 vector unsigned int vec_reve (vector unsigned int);
11941 vector bool long long vec_reve (vector bool long long);
11942 vector signed long long vec_reve (vector signed long long);
11943 vector unsigned long long vec_reve (vector unsigned long long);
11944 vector bool short vec_reve (vector bool short);
11945 vector signed short vec_reve (vector signed short);
11946 vector double vec_reve (vector double);
11947 vector float vec_reve (vector float);
11948 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
11949 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
11950 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
11951 (altivec_vreve): New pattern.
11952 * config/rs6000/altivec.h (vec_reve): New define.
11953 * doc/extend.texi (vec_rev): Update the built-in documentation file
11954 for the new built-in functions.
11956 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11958 PR tree-optimization/71815
11959 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
11961 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
11963 (slsr_process_phi): Likewise.
11964 (replace_uncond_cands_and_profitable_phis): Don't replace a
11965 multiply candidate with a stride of 1 (copy or cast).
11966 (phi_incr_cost): Call uses_consumed_by_stmt rather than
11968 (lowest_cost_path): Likewise.
11969 (total_savings): Likewise.
11971 2017-06-26 Richard Biener <rguenther@suse.de>
11974 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
11975 Use def_builtin_pure for all gather builtins.
11977 2017-06-26 Richard Biener <rguenther@suse.de>
11979 PR tree-optimization/81203
11980 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11981 non-dominating BBs.
11983 2017-06-26 Marek Polacek <polacek@redhat.com>
11986 * doc/invoke.texi: Document -Wmultistatement-macros.
11988 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11990 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11991 arm_neon_ok_no_float_abi effective target.
11993 2017-06-26 Richard Biener <rguenther@suse.de>
11995 PR tree-optimization/80928
11996 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
11997 (copy_bbs): Set BB_DUPLICATED flag early.
11998 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
12000 (execute_on_shrinking_pred): Likewise.
12001 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
12002 BB_DUPLICATED blocks.
12003 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
12004 iterate over all PHIs considering removal of *gsi.
12006 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
12008 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
12011 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12013 * config/rs6000/rs6000-string.c: (expand_block_clear,
12014 do_load_for_compare, select_block_compare_mode,
12015 compute_current_alignment, expand_block_compare,
12016 expand_strncmp_align_check, expand_strn_compare,
12017 expand_block_move, rs6000_output_load_multiple)
12018 Move functions related to string/block move/compare
12019 to a separate file.
12020 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
12021 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
12022 for this function which is now used in two files.
12023 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
12024 * config.gcc: Add rs6000-string.o to extra_objs for
12025 targets powerpc*-*-* and rs6000*-*-*.
12027 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
12030 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
12031 32-bit, since indexed is not valid for DImode.
12032 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
12033 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
12034 (define_peephole2 for Altivec d-form load): Add 32-bit support.
12035 (define_peephole2 for Altivec d-form store): Likewise.
12038 * multiple_target.c (create_dispatcher_calls): Only create the
12039 dispatcher call if the function is the default clone of a
12040 versioned function.
12042 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
12044 PR middle-end/80902
12045 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
12046 a call, force the call to not be a tail call.
12048 2017-06-23 Jeff Law <law@redhat.com>
12050 * doc/contrib.texi: Add entry for Steven Pemberton's work on
12053 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
12055 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
12056 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
12057 handling for early expansion of vector shifts (sl,sr,sra,rl).
12058 (builtin_function_type): Add vector shift right instructions
12059 to the unsigned argument list.
12061 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12063 rtl-optimizatoin/79286
12064 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
12065 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
12066 trap. PIC register plus a const unspec without offset can never trap.
12068 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
12070 * tree.h (builtin_structptr_type): New type.
12071 (builtin_structptr_types): Declare new array.
12072 * tree.c (builtin_structptr_types): New array.
12073 (free_lang_data, build_common_tree_nodes): Use it.
12075 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
12078 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
12081 2017-06-22 Matt Turner <mattst88@gmail.com>
12083 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
12084 Lake models to skylake case. Assume skylake for unknown
12085 models with clflushopt.
12087 2017-06-22 Jeff Law <law@redhat.com>
12089 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
12090 frame sizes that do not satisfy aarch64_uimm12_shift.
12092 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
12094 * profile-count.h (apply_probability,
12095 apply_scale, probability_in): Fix checks for zero.
12097 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12099 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
12100 * doc/cppdiropts.texi (-I @var{dir}): Document it.
12102 2016-06-22 Richard Biener <rguenther@suse.de>
12104 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12105 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
12106 REDUC_MAX_EXPR support.
12107 (vectorizable_reduction): Likewise.
12108 (vect_create_epilog_for_reduction): Likewise.
12110 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
12112 * match.pd (A / (1 << B) -> A >> B): New.
12113 * generic-match-head.c: Include optabs-tree.h.
12114 * gimple-match-head.c: Likewise.
12115 * optabs-tree.h (target_supports_op_p): New.
12116 * optabs-tree.c (target_supports_op_p): New.
12118 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12120 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
12121 $gcc_cv_ld --help output.
12122 (gcc_cv_ld_demangle): Likewise.
12123 (gcc_cv_ld_eh_frame_hdr): Likewise.
12124 (gcc_cv_ld_pie): Likewise.
12125 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
12126 (gcc_cv_ld_buildid): Likewise.
12127 (gcc_cv_ld_sysroot): Likewise.
12128 (ld_bndplt_support): Likewise.
12129 (ld_pushpopstate_support): Likewise.
12130 * configure: Regenerate.
12131 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
12133 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12136 * config/i386/sse.md (round<mode>2): Renumber match_dup and
12137 operands indexes to avoid gap between operands and match_dups.
12139 2017-06-21 Andrew Pinski <apinski@cavium.com>
12141 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
12142 Increment Arith_shift and Arith_shift_reg by 1.
12143 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
12145 * config/aarch64/aarch64.c (thunderx_tunings): Enable
12146 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
12147 (aarch64_strip_extend): Add new argument and test for it.
12148 (aarch64_cheap_mult_shift_p): New function.
12149 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
12150 add a cost if it is true.
12151 Update calls to aarch64_strip_extend.
12152 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
12154 2017-06-21 Andrew Pinski <apinski@cavium.com>
12156 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
12158 (thunderxt88): Likewise.
12159 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
12160 (thunderx_prefetch_tune): New variable.
12161 (thunderx2t99_prefetch_tune): Update for the correct values.
12162 (thunderxt88_tunings): New variable.
12163 (thunderx_tunings): Use thunderx_prefetch_tune instead of
12164 generic_prefetch_tune.
12165 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
12167 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12169 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
12170 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
12171 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
12172 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
12173 (aarch64_atomic_cas<mode>, GPI): Likewise.
12175 2017-06-21 Martin Liska <mliska@suse.cz>
12177 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
12178 statements on cold and hot labels.
12179 * predict.c (tree_estimate_probability_bb): Remove the
12180 prediction from this place.
12182 2017-06-21 Martin Liska <mliska@suse.cz>
12184 PR tree-optimization/79489
12185 * gimplify.c (maybe_add_early_return_predict_stmt): New
12187 (gimplify_return_expr): Call the function.
12188 * predict.c (tree_estimate_probability_bb): Remove handling
12190 * predict.def: Update comment about early return predictor.
12191 * gimple-predict.h (is_gimple_predict): New function.
12192 * predict.def: Change default value of early return to 66.
12193 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
12195 * passes.def: Put pass_strip_predict_hints to the beginning of
12198 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
12200 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
12201 FUNCTION_DECL declarations.
12202 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
12204 (dwaf2out_decl): Likewise.
12205 * godump.c (go_early_global_decl): Skip call to the real debug hook
12206 for FUNCTION_DECL declarations.
12207 * passes.c (rest_of_decl_compilation): Skip call to the
12208 early_global_decl debug hook for FUNCTION_DECL declarations, unless
12209 -fdump-go-spec is passed.
12211 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12213 * config/i386/i386.c (struct builtin_isa): New field pure_p.
12214 Reorder for compactness.
12215 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
12216 (def_builtin_pure, def_builtin_pure2): New functions.
12217 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
12219 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12221 * match.pd (nop_convert): New predicate.
12222 ((A +- CST1) +- CST2): Allow some NOP conversions.
12224 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12227 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
12228 with ctors/dtors if GOVD_SHARED is set.
12230 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12232 * config/aarch64/aarch64.md (movti_aarch64):
12233 Emit mov rather than orr.
12234 (movtf_aarch64): Likewise.
12235 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
12236 Emit mov rather than orr.
12238 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12240 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
12241 Swap alternatives, make integer dup more expensive.
12243 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12245 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
12246 Return true for non-tls symbols.
12248 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
12250 * config/aarch64/aarch64-cores.def (cortex-a55): New.
12251 (cortex-a75): Likewise.
12252 (cortex-a75.cortex-a55): Likewise.
12253 * config/aarch64/aarch64-tune.md: Regenerate.
12254 * doc/invoke.texi (-mtune): Document new values for -mtune.
12256 2017-06-21 Tom de Vries <tom@codesourcery.com>
12258 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
12259 stack_size feature.
12260 (Effective-Target Keywords, Other attributes): Suggest using
12261 dg-add-options stack_size feature to get stack limit in stack_size
12262 effective target documentation.
12264 2017-06-21 Julian Brown <julian@codesourcery.com>
12265 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12267 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
12268 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
12269 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
12271 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
12272 attribute type list for neon_multiply.
12273 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
12274 attribute type list for neon_multiply.
12275 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
12276 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
12277 attribute type list for neon_multiply.
12278 * config/arm/types.md (crypto_pmull): Add.
12279 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
12280 attribute type list.
12282 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
12284 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
12287 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12289 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
12290 to make sure not to dereference a NULL cost_classes_ptr pointer.
12292 2017-06-20 Carl Love <cel@us.ibm.com>
12294 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12295 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12296 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12297 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12298 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12299 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12300 VMULOSW): New enum "unspec" values.
12301 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12302 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12303 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12304 altivec_vmulosw): New patterns.
12305 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12306 VMULOSW): Add definitions.
12308 2017-06-20 Julia Koval <julia.koval@intel.com>
12310 * config/i386/i386.c: Fix rounding expand for new pattern.
12311 * config/i386/subst.md: Fix pattern (parallel -> unspec).
12313 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12315 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
12316 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
12318 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12320 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
12323 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12325 * config/aarch64/aarch64-cores.def: Rearrange to sort by
12326 architecture, then by implementer ID.
12327 * config/aarch64/aarch64-tune.md: Regenerate.
12329 2017-06-20 Richard Biener <rguenther@suse.de>
12331 PR middle-end/81097
12332 * fold-const.c (split_tree): Fold to type before negating.
12334 2017-06-20 David Malcolm <dmalcolm@redhat.com>
12336 * diagnostic-show-locus.c
12337 (selftest::test_fixit_deletion_affecting_newline): New function.
12338 (selftest::diagnostic_show_locus_c_tests): Call it.
12340 2017-06-20 Andreas Schwab <schwab@suse.de>
12343 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
12346 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12348 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
12349 __ARM_FEATURE_COPROC according to support.
12351 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12353 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12354 Rewritten to avoid overflow for > 32-bit pointers.
12357 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
12358 by removing enum keyword.
12359 (ubsan_type_descriptor): Likewise. Formatting fix.
12362 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
12363 splitter): Require TARGET_SSE2 in the condition.
12365 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12368 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
12369 for doing vector set of SFmode on ISA 3.0.
12370 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
12371 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
12373 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
12374 SFmode value into a V4SF variable that was extracted from another
12375 V4SF variable without converting the element to double precision
12376 and back to single precision vector format.
12377 (vsx_insert_extract_v4sf_p9_2): Likewise.
12379 2017-06-19 Jakub Jelinek <jakub@redhat.com>
12381 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
12382 in UWHI to avoid undefined overflow.
12385 * ubsan.h (enum ubsan_encode_value_phase): New.
12386 (ubsan_encode_value): Change second argument to
12387 enum ubsan_encode_value_phase with default value of
12388 UBSAN_ENCODE_VALUE_GENERIC.
12389 * ubsan.c (ubsan_encode_value): Change second argument to
12390 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
12391 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
12392 create_tmp_var_raw instead of create_tmp_var and use a
12394 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12395 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12396 ubsan_encode_value callers.
12399 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12400 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12401 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12403 2017-06-19 Richard Biener <rguenther@suse.de>
12405 PR middle-end/81118
12406 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12407 estimates if we changed anything.
12409 2017-06-19 Richard Biener <rguenther@suse.de>
12411 PR tree-optimization/80887
12412 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12413 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12414 simplified lookups, then reset mprts_hook.
12415 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12417 (try_to_simplify): Likewise.
12419 2017-06-19 Martin Liska <mliska@suse.cz>
12422 * gimplify.c (gimplify_switch_expr):
12423 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12425 2017-06-19 Martin Liska <mliska@suse.cz>
12427 * doc/install.texi: Document that PGO runs in 4 stages.
12429 2017-06-19 Martin Liska <mliska@suse.cz>
12432 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12433 to dispatcher function name.
12434 * multiple_target.c (replace_function_decl): New function.
12435 (create_dispatcher_calls): Redirect both edges and references.
12437 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12439 * profile-count.c (profile_count::dump): Dump quality.
12440 (profile_count::differs_from_p): Update for unsigned val.
12441 * profile-count.h (profile_count_quality): New enum.
12442 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12444 2017-06-19 Richard Biener <rguenther@suse.de>
12446 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12447 struct function as arg.
12448 (estimate_numbers_of_iterations): Export overload with loop arg.
12449 (free_numbers_of_iterations_estimates_loop): Use an overload of
12450 free_numbers_of_iterations_estimates instead.
12451 * tree-cfg.c (remove_bb): Adjust.
12452 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12453 * tree-parloops.c (gen_parallel_loop): Likewise.
12454 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12456 (tree_unroll_loops_completely): Likewise.
12457 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12458 Use an overload instead and export.
12459 (estimated_loop_iterations): Adjust.
12460 (max_loop_iterations): Likewise.
12461 (likely_max_loop_iterations): Likewise.
12462 (estimate_numbers_of_iterations): Take struct function as arg
12464 (loop_exits_before_overflow): Adjust.
12465 (free_numbers_of_iterations_estimates_loop): Use an overload.
12466 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12467 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12469 2017-06-19 Richard Biener <rguenther@suse.de>
12472 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12473 RANGE_EXPR conservatively.
12475 2017-06-16 Carl Love <cel@us.ibm.com>
12477 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12478 definitions for vec_float, vec_float2, vec_floato,
12479 vec_floate built-ins.
12480 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12481 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12483 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12484 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12485 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12486 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12487 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12488 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12489 vec_floato): Add builtin defines.
12490 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12491 Update the built-in documentation file for the new built-in
12494 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12496 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12497 (mthumb): Mark as the negative of -marm.
12499 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12501 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12503 (ARM Options, -mtune): Document that this accepts the same
12504 extension options as -mcpu.
12505 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12507 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12509 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12510 permitted extensions.
12512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12514 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12515 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12516 (armv8-m.main): Add option +nodsp.
12517 * config/arm/arm-cpu-cdata.h: Regenerated.
12519 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12521 * config/arm/t-fuchsia: New file.
12522 * config.gcc (arm*-*-fuchsia*): Use it.
12524 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12526 * config/arm/t-symbian: Rewrite for new option infrastructure.
12528 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12530 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12531 (MULTILIB_REQUIRED): Likewise.
12533 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12535 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12536 (MULTILIB_RESUE): Likewise.
12537 (MULTILIB_MATCHES): Likewise.
12538 (MULTLIB_REQUIRED): Likewise.
12540 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12542 * config/arm/t-rtems: Rewrite for new option framework.
12544 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12546 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12547 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12548 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12549 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12550 * config/arm/t-multilib: ... here.
12551 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12552 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12553 armv7-a and armv8*-a when A-profile libraries have not been built.
12554 * config/arm/t-rmprofile: Rewrite.
12556 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12558 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12559 with a backslash. Remove the backslash after substituting unescaped
12561 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12563 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12565 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12566 the driver to pass through the default setting of -mfloat-abi.
12567 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12569 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12570 (all_feat_combs): New rule.
12571 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12573 * config/arm/t-aprofile: Rewrite.
12575 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12577 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12578 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12579 fpu is not specified by the user/command-line.
12580 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12581 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12582 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12583 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12584 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12585 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12587 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12589 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12590 * config/arm/t-arm-elf: Rewritten.
12592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12594 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12595 have some floating-point instructions.
12596 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12597 (TARGET_MAYBE_HARD_FLOAT): New macro.
12598 * config/arm/arm-builtins.c (arm_init_builtins): Use
12599 TARGET_MAYBE_HARD_FLOAT.
12600 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12602 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12604 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12605 (configargs.h): Include it.
12606 (arm_print_hint_for_fpu_option): New function.
12607 (arm_parse_fpu_option): New function.
12608 (candidate_extension): New class.
12609 (arm_canon_for_multilib): New function.
12610 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12611 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12612 (ARCH_CANONICAL_SPECS): New macro.
12613 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12615 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12617 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12618 are set after handling multilib fragments. Set target_cpu_default2
12621 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12623 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12625 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12626 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12628 (check_arch): Likewise.
12629 * config/arm/arm.c (arm_configure_build_target): Handle
12630 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12631 options in the default.
12633 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12635 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
12636 when an option is an alias of another.
12637 * config/arm/parsecpu.awk (optalias): New parser token.
12638 (gen_comm_data): Mark non-alias options as such. Emit entries
12639 for extension aliases.
12640 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
12641 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
12642 (armv6kz, armv6zk, armv6t2): Likewise.
12643 (armv7): Make vfpv3-d16 an alias.
12644 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
12646 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
12647 Sort in canonical order.
12648 (armv8-a): Sort in canonical order.
12649 (armv8.1-a, armv8.2-a): Likewise.
12650 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12652 (cortex-a9): Sort in canonical order.
12653 * config/arm/arm.c (selftests.h): Include it.
12654 (arm_test_cpu_arch_data): New function.
12655 (arm_run_self_tests): New function.
12656 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12657 (targetm): Move declaration to the end of the file.
12658 * arm-cpu-cdata.h: Regenerated.
12660 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12662 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12663 call to target_mode_check describing the type of option passed.
12664 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12665 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12666 arm_parse_cpu_option_name to match parameters against list of
12668 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12669 arm_arch_core_flags data structure.
12670 * config/arm/arm-cpu_cdata.h: Regenerated.
12672 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12674 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12676 (arm_print_hint_for_cpu_option): Likewise.
12677 (arm_print_hint_for_arch_option): Likewise.
12678 (arm_parse_cpu_option_name): Likewise.
12679 (arm_parse_arch_option_name): Likewise.
12680 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12681 of entries in the all_fpus list.
12682 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12683 (arm_parse_cpu_option_name): Declare.
12684 (arm_parse_arch_option_name): Declare.
12685 (arm_parse_option_features): Declare.
12686 (arm_intialize_isa): Declare.
12687 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12689 (gen_comm_data): ... here. Make definitions non-static.
12690 * config/arm/arm-cpu-data.h: Regenerated.
12691 * config/arm/arm-cpu-cdata.h: Regenerated.
12693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12695 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12696 (cpu_arch_extension): New structure.
12697 (cpu_arch_option, arch_option, cpu_option): New structures.
12698 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12699 architecture types.
12700 (gen_data): Generate new format data tables.
12701 * config/arm/arm.c (cpu_tune): New structure.
12702 (cpu_option, processors): Delete.
12703 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12704 (arm_print_hint_for_cpu_option): ... this and ...
12705 (arm_print_hint_for_arch_option): ... this.
12706 (arm_parse_arch_cpu_name): Delete. Replace with ...
12707 (arm_parse_cpu_option_name): ... this and ...
12708 (arm_parse_arch_option_name): ... this.
12709 (arm_unrecognized_feature): Change type of target parameter to
12711 (arm_parse_arch_cpu_features): Delete. Replace with ...
12712 (arm_parse_option_features): ... this.
12713 (arm_configure_build_target): Rework to use new configuration data
12715 (arm_print_tune_info): Rework for new configuration data tables.
12716 * config/arm/arm-cpu-data.h: Regenerated.
12717 * config/arm/arm-cpu.h: Regenerated.
12719 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12721 * Makefile.in (OBJS): Move sbitmap.o from here ...
12722 (OBJS-libcommon): ... to here.
12724 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12726 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12727 (ISA_ALL_CRYPTO): New macro.
12728 (ISA_ALL_SIMD): New macro
12729 (ISA_ALL_FP): New macro.
12730 * config/arm/arm.c (fpu_bitlist): Update initializer.
12731 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12733 (arm9e): Add fpu. Add option for nofp
12734 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12735 (arm926ej-s, arm1026ej-s): Likewise.
12736 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12737 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
12738 neon-fp16, neon-vfpv4, nofp and nosimd.
12739 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
12740 (cortex-a8): Add fpu. Add option for nofp.
12741 (cortex-a9): Add fpu. Add options for nosimd and nofp.
12742 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
12743 (cortex-r4f): Add fpu.
12744 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
12745 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
12747 (cortex-r8): Likewise.
12748 (cortex-m4): Add fpu. Add option for nofp.
12749 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
12750 (cortex-a17.cortex-a7): Likewise.
12751 (cortex-a32): Add fpu. Add options for crypto and nofp.
12752 (cortex-a35, cortex-a53): Likewise.
12753 (cortex-a57): Add fpu. Add option for crypto.
12754 (cortex-a72, cortex-a73): Likewise.
12755 (exynos-m1): Likewise.
12756 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
12757 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
12758 (cortex-m33): Add fpu. Add option for nofp.
12759 * config/arm/arm-cpu-cdata.h: Regenerated
12760 * config/arm/arm-cpu-data.h: Regenerated.
12762 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12764 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
12765 (armv5te, armv5tej): Likewise.
12766 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
12767 (armv7): Add options fp and vfpv3-d16.
12768 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
12769 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
12771 (armv7ve): Likewise.
12772 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
12773 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
12774 (armv8-a): Add nocrypto option.
12775 (armv8.1-a, armv8.2-a): Likewise.
12776 (armv8-m.main): add options fp, fp.dp and nofp.
12778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12780 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
12782 (armv8-a+crc): Delete.
12783 (armv8.1-a): Add options simd, crypto and nofp.
12784 (armv8.2-a): Add options fp16, simd, crypto and nofp.
12785 (armv8.2-a+fp16): Delete.
12786 (armv8-m.main): Add option dsp.
12787 (armv8-m.main+dsp): Delete.
12788 (cortex-a8): Add fpu. Add nofp option.
12789 (cortex-a9): Add fpu. Add nofp and nosimd options.
12790 * config/arm/parsecpu.awk (gen_data): Generate option tables and
12791 link to main cpu and architecture data structures.
12792 (gen_comm_data): Only put isa attributes from the main architecture
12794 (option): New statement for architecture and CPU entries.
12795 * arm.c (struct cpu_option): New structure.
12796 (struct processors): Add entry for options.
12797 (arm_unrecognized_feature): New function.
12798 (arm_parse_arch_cpu_name): Ignore any characters after the first
12800 (arm_parse_arch_cpu_feature): New function.
12801 (arm_configure_build_target): Separate out any CPU and architecture
12802 features and parse separately. Don't error out if -mfpu=auto is
12803 used with only an architecture string.
12804 (arm_print_asm_arch_directives): New function.
12805 (arm_file_start): Call it.
12806 * config/arm/arm-cpu-cdata.h: Regenerated.
12807 * config/arm/arm-cpu-data.h: Likewise.
12808 * config/arm/arm-tables.opt: Likewise.
12810 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12812 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
12813 assembler when it is not -mfpu=auto.
12815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12817 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
12818 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
12819 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
12820 (ASM_CPU_SPEC): Rewrite.
12821 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
12822 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
12823 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
12824 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
12825 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
12826 copied string is NUL-terminated. Also strip any characters prefixed
12828 (arm_rewrite_selected_arch): New function.
12829 (arm_rewrite_march): New function.
12831 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12833 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
12834 (x_arm_cpu_string, x_arm_tune_string): Likewise.
12835 (march, mcpu, mtune): Convert to string-based options.
12836 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
12837 (arm_parse_arch_cpu_name): New function.
12838 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
12839 identify selected architecture or CPU.
12840 (arm_option_save): New function.
12841 (TARGET_OPTION_SAVE): Redefine.
12842 (arm_option_restore): Restore string options.
12843 (arm_option_print): Print string options.
12845 2017-06-16 Martin Sebor <msebor@redhat.com>
12847 PR tree-optimization/80933
12848 PR tree-optimization/80934
12849 * builtins.c (fold_builtin_3): Do not handle bcmp here.
12850 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
12851 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
12852 (gimple_fold_builtin): Call them.
12854 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12856 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
12857 as unlikely; update profile.
12859 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12861 * predict.c (force_edge_cold): Handle declaring edges impossible
12864 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12866 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
12868 (try_unroll_loop_completely): Fix reporting.
12870 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12872 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
12874 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
12877 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
12878 if given a non-constant argument for an intrinsic which requires a
12881 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12883 * profile.c (compare_freqs): New function.
12884 (branch_prob): Sort edge list.
12885 (find_spanning_tree): Assume that the list is priority sorted.
12887 2017-06-16 Richard Biener <rguenther@suse.de>
12889 PR tree-optimization/81090
12890 * passes.def (pass_record_bounds): Remove.
12891 * tree-pass.h (make_pass_record_bounds): Likewise.
12892 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
12893 make_pass_record_bounds): Likewise.
12894 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
12895 not free niter estimates at the beginning but at the end.
12896 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
12898 2017-06-16 Richard Biener <rguenther@suse.de>
12900 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
12901 initializer to workaround ICE in host GCC 4.8.
12903 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12905 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
12907 (clone_inlined_nodes): Update.
12909 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12911 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
12912 prefetch settings, and enable prefetching by default at -O3.
12914 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12916 * config/aarch64/aarch64.c (aarch64_override_options_internal):
12917 Set flag_prefetch_loop_arrays according to tuning data.
12919 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12921 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
12922 New tune structure.
12923 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
12924 [Unrelated to main purpose of the patch] Place the pointer field last
12925 to enable type checking errors when tune structure are wrongly merged.
12926 * config/aarch64/aarch64.c (generic_prefetch_tune,)
12927 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
12928 (thunderx2t99_prefetch_tune): New tune constants.
12929 (tune_params *_tunings): Update all tunings (no functional change).
12930 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
12931 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
12932 from tunings structures.
12934 2017-06-16 Jakub Jelinek <jakub@redhat.com>
12937 * ubsan.c (instrument_null): Add T argument, use it instead
12938 of computing it based on IS_LHS.
12939 (instrument_object_size): Likewise.
12940 (pass_ubsan::execute): Adjust instrument_null and
12941 instrument_object_size callers to pass gimple_get_lhs or
12942 gimple_assign_rhs1 result to it. Use instrument_null instead of
12943 calling get_base_address and instrument_mem_ref. Handle
12944 aggregate call arguments for object-size sanitization.
12946 2017-06-16 Yury Gribov <tetra2005@gmail.com>
12948 PR tree-optimization/81089
12949 * tree-vrp.c (is_masked_range_test): Validate operands of
12952 2017-06-15 Martin Sebor <msebor@redhat.com>
12955 * dumpfile.c (dump_register): Avoid calling memset to initialize
12956 a class with a default ctor.
12957 * gcc.c (struct compiler): Remove const qualification.
12958 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
12959 * hash-table.h: Ditto.
12960 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
12962 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
12963 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
12965 * params.h (struct param_info): Make struct members non-const.
12966 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
12967 with default initialization.
12968 * vec.h (vec_copy_construct, vec_default_construct): New helper
12970 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
12971 with vec_copy_construct.
12972 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
12973 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
12974 * doc/invoke.texi (-Wclass-memaccess): Document.
12976 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12978 * emit-rtl.h (is_leaf): Update comment about local
12979 register allocator.
12981 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12984 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12985 for a variable to have a section before checking if the section has a
12987 Set section to.persistent if persistent attribute is set.
12988 Warn if .persistent attribute is used on an automatic variable.
12990 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12992 PR rtl-optimization/80474
12993 * reorg.c (update_block): Do not ignore instructions in a delay slot.
12995 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
12997 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
13000 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
13002 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
13003 (casesi): Emit bounds checking as RTL.
13004 (casesi_internal_mips16_<mode>): Remove bounds checking.
13006 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
13008 * config/xtensa/xtensa.c (xtensa_option_override): Append
13009 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
13010 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
13011 xtensa_doloop_hooks): Define unconditionally.
13012 (xtensa_reorg_loops): Only call reorg_loops in the presence of
13014 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
13015 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
13016 for it in xtensa_option_override.
13017 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
13018 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
13020 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
13022 * doc/cppopts.texi: Document '-' special value to -MF.
13024 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
13026 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
13027 (cortex_a53_fconst): Likewise.
13028 (cortex_a53_fpmul): Likewise.
13029 (cortex_a53_f_load_64): Likewise.
13030 (cortex_a53_f_load_many): Likewise.
13031 (cortex_a53_advsimd_alu): Likewise.
13032 (cortex_a53_advsimd_alu_q): Likewise.
13033 (cortex_a53_advsimd_mul): Likewise.
13034 (cortex_a53_advsimd_mul_q): Likewise.
13035 (fpmac bypass): Add new bypass for fpmac-fpmac case.
13036 Add missing fmul, r2f_cvt and fconst cases.
13038 2017-06-14 Richard Biener <rguenther@suse.de>
13040 PR middle-end/81088
13041 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
13043 (fold_binary_loc): When associating do not treat pre-existing
13044 TREE_OVERFLOW on literal constants as a reason to allow
13045 TREE_OVERFLOW on associated literal constants.
13047 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13049 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
13050 (MASK_FEATURES): New macro.
13051 * config/sparc/sparc.c (sparc_option_override): Remove the special
13052 handling of -mfpu and generalize it to all MASK_FEATURES switches.
13054 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13056 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
13057 a division of 0 if non-call exceptions are enabled.
13059 2017-06-14 Andrew Pinski <apinski@cavium.com>
13060 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13063 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
13064 Improve vector initialization code gen for only variable case.
13066 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
13068 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
13070 2017-06-14 Richard Biener <rguenther@suse.de>
13072 PR tree-optimization/81083
13073 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
13076 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13078 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
13079 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
13080 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
13081 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
13082 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
13083 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
13085 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13087 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
13088 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
13090 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13092 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
13094 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13096 * config/rs6000/t-rtems: Don't handle SPE.
13098 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13100 * config/rs6000/t-linux: Don't handle SPE.
13102 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13104 * config/rs6000/eabispe.h: Delete file.
13106 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13108 * config/rs6000/t-spe: Delete file.
13110 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13112 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
13113 (rs6000_legitimate_offset_address_p): Return false for anything in
13114 V2SImode or V2SFmode.
13116 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13118 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
13119 except V2SF and V2SI. Rearrange the vector modes, and add comments.
13120 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
13122 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
13123 (rs6000_legitimate_offset_address_p): Ditto.
13124 (rs6000_emit_move): Ditto.
13125 (rs6000_init_builtins): Remove V4HI_type_node.
13127 2017-06-13 Martin Liska <mliska@suse.cz>
13130 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
13131 (gate_asan): Likewise.
13132 * asan.h (asan_no_sanitize_address_p): Remove the function.
13133 (sanitize_flags_p): New function.
13134 * builtins.def: Fix coding style.
13135 * common.opt: Use renamed enum value.
13136 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
13137 * doc/extend.texi: Document no_sanitize attribute.
13138 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
13139 to SANITIZE_UNDEFINED_NONDEFAULT.
13140 * gcc.c (sanitize_spec_function): Use the renamed enum value.
13141 * gimple-fold.c (optimize_atomic_compare_exchange_p):
13142 Use sanitize_flags_p.
13143 * gimplify.c (gimplify_function_tree): Likewise.
13144 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
13145 * opts.c (parse_no_sanitize_attribute): New function.
13146 (common_handle_option): Use renamed enum value.
13147 * opts.h (parse_no_sanitize_attribute): Declare.
13148 * tree.c (sanitize_flags_p): New function.
13149 * tree.h: Declared here.
13150 * tsan.c: Use sanitize_flags_p.
13151 * ubsan.c (ubsan_expand_null_ifn): Likewise.
13152 (instrument_mem_ref): Likewise.
13153 (instrument_bool_enum_load): Likewise.
13154 (do_ubsan_in_current_function): Remove the function.
13155 (pass_ubsan::execute): Use sanitize_flags_p.
13156 * ubsan.h: Remove do_ubsan_in_current_function
13157 * tree-cfg.c (print_no_sanitize_attr_value): New function.
13158 (dump_function_to_file): Use it here.
13160 2017-06-13 Martin Jambor <mjambor@suse.cz>
13162 PR tree-optimization/80803
13163 PR tree-optimization/81063
13164 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
13165 (propagate_subaccesses_across_link): Enqueue subtree whenever
13166 necessary instead of relying on the caller.
13168 2017-06-13 Martin Jambor <mjambor@suse.cz>
13170 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
13171 that have a first_link.
13172 (sort_and_splice_var_accesses): Do not check first_link before
13174 (subtree_mark_written_and_enqueue): Likewise.
13175 (propagate_all_subaccesses): Likewise and do not stop at first
13176 parent with a first_link.
13178 2017-06-13 Martin Jambor <mjambor@suse.cz>
13180 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
13183 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13185 * match.pd: New pattern.
13187 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13189 * tree-vrp.c (is_masked_range_test): New function.
13190 (register_edge_assert_for): Determine ranges for
13193 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13195 PR tree-optimization/67328
13196 * fold-const.c (maskable_range_p): New function.
13197 (build_range_check): Generate bittests if possible.
13199 2017-06-13 Martin Liska <mliska@suse.cz>
13201 * gimple-pretty-print.c (dump_probability): Add new argument.
13202 (dump_edge_probability): Dump both probability and count.
13203 (dump_gimple_label): Likewise.
13204 (dump_gimple_bb_header): Likewise.
13206 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
13209 * config/avr/avr-devices.c: Fix indentation.
13210 * config/avr/gen-avr-mmcu-specs.c: Dito.
13212 2017-06-13 Richard Biener <rguenther@suse.de>
13214 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
13215 instead get vector type from stmt_info.
13216 (vectorizable_reduction): Adjust. Remove dead code.
13218 2017-06-13 Richard Biener <rguenther@suse.de>
13220 PR middle-end/81065
13221 * fold-const.c (extract_muldiv_1): Remove bogus distribution
13222 case of C * (x * C2 + C3).
13223 (fold_addr_of_array_ref_difference): Properly fold index difference.
13225 2017-06-12 David S. Miller <davem@davemloft.net>
13228 * config/sparc/sparc.md (return expander): Emit frame blockage if
13229 function uses alloca.
13231 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13233 * combine.c (make_field_assignment): Check len rather than the mode
13234 precision when calling force_to_mode.
13236 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
13238 Support multilibs and devices that see flash in RAM address range.
13241 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
13242 (avr_mcu_t) <flash_pm_offset>: New field.
13243 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
13244 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
13245 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
13246 (AVR_TINY_PM_OFFSET): Remove macro.
13247 * config/avr/avr.opt (-mshort-calls): New option.
13248 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13249 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
13250 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
13251 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
13252 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
13253 instead of avr_arch->have_jmp_call.
13254 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
13255 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
13256 avr_arch->flash_pm_offset to define.
13257 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
13258 new field flash_pm_offset. Add entry for avrxmega3.
13259 (avr_texinfo): Add entry for avrxmega3.
13260 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
13261 attiny212, attiny214,
13262 attiny412, attiny414, attiny416, attiny417,
13263 attiny814, attiny816, attiny817,
13264 attiny1614, attiny1616, attiny1617,
13265 attiny3214, attiny3216, attiny3217.
13266 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
13267 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
13268 (avr_print_operand_address) [AVR_TINY]: Same.
13269 (avr_asm_init_sections) <readonly_data_section>: Only patch
13270 callback if avr_arch->flash_pm_offset = 0.
13271 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
13272 for rodata if avr_arch->flash_pm_offset != 0.
13273 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
13274 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
13275 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
13276 (m_options): Append opt_rcall.
13277 (m_dirnames): Append dir_rcall.
13278 * config/avr/t-multilib: Regenerate.
13280 * configure.ac [target=avr]: Check whether avrxmega3 default
13281 linker description file works as needed.
13282 * configure: Regenerate.
13283 * doc/avr-mmcu.texi: Regenerate.
13284 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
13285 <__AVR_ARCH__>: Document avrxmega3 and 103.
13286 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
13287 <__AVR_SHORT_CALLS__>: Document it.
13288 <__AVR_PM_BASE_ADDRESS__>: Document it.
13289 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
13290 (AVR Variable Attributes) <progmem>: Document this is
13291 not needed for avrxmega3.
13292 (AVR Named Address Spaces) <__flash>: Dito.
13294 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
13296 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
13298 2017-06-12 Doug Rupp <rupp@adacore.com>
13300 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
13301 Append vxworks-stdint.h to the tm_file list.
13302 * config/vxworks-stdint.h: New file.
13304 2017-06-12 Martin Liska <mliska@suse.cz>
13306 PR tree-optimization/81041
13307 * tree-profile.c (gimple_gen_ic_func_profiler):
13308 Create an extra BB in profile-generate
13309 (gimple_gen_time_profiler): Likewise.
13311 2017-06-12 Jakub Jelinek <jakub@redhat.com>
13313 PR tree-optimization/81003
13314 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
13315 (update_range_test): Use it instead of force_gimple_operand_gsi.
13317 2017-06-12 Richard Biener <rguenther@suse.de>
13319 PR tree-optimization/81053
13320 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
13321 with backedge value not defined in loop. Simplify def stmt
13324 2017-06-11 Tom de Vries <tom@codesourcery.com>
13327 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
13329 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
13330 nvptx_cannot_force_const_mem.
13332 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13334 * opts.c (finish_options): Move test for flag_split_stack after
13335 it has been initialized.
13337 2017-06-11 Jason Merrill <jason@redhat.com>
13339 * tree.h (id_equal): New.
13340 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
13341 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
13342 instead of strcmp of IDENTIFIER_POINTER.
13344 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13346 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
13347 (mark_all_inlined_calls_cdtor): Fix formating.
13348 (inline_transform): Rescale profile before inlining.
13350 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13352 * cgraph.h (cgraph_edge::clone): Update prototype.
13353 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
13354 (cgraph_node::create_clone): Update.
13355 (cgraph_node::create_version_clone): Update.
13356 * tree-inline.c (copy_bb): Update.
13357 (expand_call_inline): Update.
13359 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13361 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
13362 factored out from ...
13363 (rs6000_emit_prologue): ... here.
13365 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13367 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
13368 factored out from ...
13369 (rs6000_emit_prologue): ... here.
13371 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13373 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
13375 (handle_missing_profiles): Fix computation of tp_first_run.
13376 (counts_to_freqs): Do not touch freqs when count is 0.
13378 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13380 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
13383 2017-06-10 Tom de Vries <tom@codesourcery.com>
13385 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13386 attributes): Document signal effective target.
13388 2017-06-10 Tom de Vries <tom@codesourcery.com>
13390 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13391 Document effective target stack_size.
13393 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13395 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13396 to the edit_context if they can be auto-applied.
13398 2017-06-9 Ian Lance Taylor <iant@golang.org>
13400 * opts.c (finish_options): If -fsplit-stack, disable implicit
13401 -forder-blocks-and-partition.
13402 * doc/invoke.texi (Optimize Options): Document that when using
13403 -fsplit-stack -forder-blocks-and-partition is not implicitly
13406 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13408 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13409 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13410 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13411 * builtins.def (abort, trap, unreachable): Declare cold.
13412 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13413 * tree-core.h (ECF_COLD): New.
13414 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13415 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13417 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13419 * predict.c (unlikely_executed_stmt_p): Cleanup.
13421 2017-06-09 Richard Biener <rguenther@suse.de>
13423 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13424 model if the ref is always written to.
13426 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13428 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13430 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13432 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13435 2017-06-09 Tom de Vries <tom@codesourcery.com>
13438 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13439 "target cannot support label values" when encountering LABEL_REF.
13441 2017-06-09 Martin Liska <mliska@suse.cz>
13443 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13444 (gimple_gen_ic_func_profiler): Emit direct comparison
13445 of __gcov_indirect_call_callee with NULL.
13446 (gimple_gen_time_profiler): Change probability from
13447 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13449 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13451 * profile.c (edge_gcov_counts): Turn to pointer.
13452 (compute_branch_probabilities, compute_branch_probabilities): Update.
13453 (branch_prob): Do not clear edge_gcov_count.
13454 * profile.h (edge_gcov_counts): Turn to pointer.
13455 (edge_gcov_count): Update.
13457 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13459 * gimple.h (gimple_check_failed): Mark cold.
13461 2017-06-09 Richard Biener <rguenther@suse.de>
13463 PR tree-optimization/66623
13464 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13465 refactor check_reduction into two parts, properly computing
13466 whether we have to check reduction validity for outer loop
13469 2017-06-09 Richard Biener <rguenther@suse.de>
13471 PR tree-optimization/79483
13472 * graphite-scop-detection.c (order): New global.
13473 (get_order): Compute bb to order mapping that satisfies code
13474 generation constraints.
13475 (cmp_pbbs): New helper.
13476 (build_scops): Start domwalk at entry block, sort generated
13479 2017-06-09 Richard Biener <rguenther@suse.de>
13481 PR middle-end/81007
13482 * ipa-polymorphic-call.c
13483 (ipa_polymorphic_call_context::restrict_to_inner_class):
13484 Skip FIELD_DECLs with error_mark_node type.
13485 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13488 2017-06-09 Martin Liska <mliska@suse.cz>
13490 * predict.c (struct branch_predictor): New struct.
13491 (test_prediction_value_range): New test.
13492 (predict_c_tests): New function.
13493 * selftest-run-tests.c (selftest::run_tests): Run the function.
13494 * selftest.h: Declare new tests.
13496 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13499 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13500 gen_add3_insn did not fail.
13501 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13502 r0, construct that number in a temporary reg and add that reg to r0.
13503 If asked to put the result in r0 as well, fail.
13505 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13507 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13508 for early expansion of vec_eqv.
13510 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13512 PR middle-end/81005
13513 * ubsan.c (instrument_null): Avoid pointless code temporary.
13514 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13516 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13520 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13523 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13526 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13528 * predict.c (unlikely_executed_edge_p): Move ahead.
13529 (probably_never_executed_edge_p): Use it.
13531 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13533 PR middle-end/79988
13534 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13535 gimple_call_builtin_p call.
13537 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13539 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13540 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13541 rtl_check_failed_type2, rtl_check_failed_code1,
13542 rtl_check_failed_code2, rtl_check_failed_code_mode,
13543 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13544 rtvec_check_failed_bounds, rtl_check_failed_flag,
13545 _fatal_insn_not_found, _fatal_insn): Likewise.
13546 * tree.h (tree_contains_struct_check_failed,
13547 tree_check_failed, tree_not_check_failed,
13548 tree_class_check_failed, tree_range_check_failed,
13549 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13550 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13551 tree_operand_check_failed, omp_clause_check_failed,
13552 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13555 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13557 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13558 flag_branch_probabilities.
13559 * ipa-inline.c (edge_badness): Likewise.
13560 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13561 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13562 * predict.c (maybe_hot_frequency_p): Likewise.
13563 (probably_never_executed): Likewise.
13564 * sched-ebb.c (schedule_ebbs): Likewise.
13565 * sched-rgn.c (find_single_block_region): Likewise.
13566 * tracer.c (tail_duplicate): Likewise.
13568 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13570 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13571 longer requires x_flag_profile_use.
13573 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13575 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13576 instead of flag_reorder_blocks_and_partition.
13577 * dbxout.c (dbxout_function_end): Likewise.
13578 * dwarf2out.c (gen_subprogram_die): Likewise.
13579 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13580 * hw-doloop.c (reorg_loops): Likewise.
13581 * varasm.c (assemble_start_function,
13582 assemble_end_function): Likewise.
13583 (decide_function_section): Do not check for
13584 flag_reorder_blocks_and_partition.
13586 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13588 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13590 (chkp_get_hard_register_fake_addr_expr): Ditto.
13591 (chkp_build_addr_expr): Add check for hard reg case.
13592 (chkp_parse_array_and_component_ref): Ditto.
13593 (chkp_find_bounds_1): Ditto.
13594 (chkp_process_stmt): Don't generate bounds store for
13597 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13599 * predict.c (maybe_hot_bb_p): Do not check profile status.
13600 (maybe_hot_edge_p): Likewise.
13601 (probably_never_executed): Check for zero counts even if profile
13603 (unlikely_executed_edge_p): New function.
13604 (unlikely_executed_stmt_p): New function.
13605 (unlikely_executed_bb_p): New function.
13606 (set_even_probabilities): Use unlikely predicates.
13607 (combine_predictions_for_bb): Likewise.
13608 (predict_paths_for_bb): Likewise.
13609 (predict_paths_leading_to_edge): Likewise.
13610 (determine_unlikely_bbs): New function.
13611 (estimate_bb_frequencies): Use it.
13612 (compute_function_frequency): Use zero counts even if profile is
13614 * profile-count.h: Fix typo.
13616 2017-08-08 Julia Koval <julia.koval@intel.com>
13618 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13619 _mm512_mask_cvtsepi16_storeu_epi8,
13620 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13621 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13622 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13623 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13624 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13625 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13626 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13627 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13628 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13629 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13630 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13631 __builtin_ia32_pmovuswb256mem_mask,
13632 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13633 __builtin_ia32_pmovwb512mem_mask): New builtins.
13635 2017-08-08 Julia Koval <julia.koval@intel.com>
13637 PR target/73350,80862
13638 * config/i386/subst.md (round): Fix round pattern.
13639 * config/i386/i386.c (ix86_erase_embedded_rounding):
13640 Fix erasing rounding for the fixed pattern.
13642 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13644 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
13646 2017-06-08 Martin Liska <mliska@suse.cz>
13648 PR gcov-profile/80911
13649 * gcov.c (block_info::block_info): New constructor.
13651 2017-06-07 Carl Love <cel@us.ibm.com>
13653 * config/rs6000/rs6000-c: The return type of the following
13654 built-in functions was implemented as int not long long. Fix sign
13655 of return value for the unsigned version of vec_mulo and vec_mule.
13656 vector unsigned long long vec_bperm (vector unsigned long long,
13657 vector unsigned char)
13658 vector signed long long vec_mule (vector signed int,
13660 vector unsigned long long vec_mule (vector unsigned int,
13661 vector unsigned int)
13662 vector signed long long vec_mulo (vector signed int,
13664 vector unsigned long long vec_mulo (vector unsigned int,
13665 vector unsigned int)
13666 * doc/extend.texi: Fix the documentation for the built-in
13669 2017-06-07 Carl Love <cel@us.ibm.com>
13672 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13675 2017-06-07 Carl Love <cel@us.ibm.com>
13677 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13678 support, Generate doublehv for signed int/float for BE case only.
13680 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13682 * doc/invoke.texi (mcx16): Rewrite.
13684 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13686 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13687 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13688 *mov<mode>_softfloat, and an anonymous splitter): Use
13689 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13691 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13693 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13695 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13696 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13697 (rs6000_debug_reg_global): Adjust.
13698 (rs6000_init_hard_regno_mode_ok): Adjust.
13699 (rs6000_dbx_register_number): Adjust.
13700 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13701 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13702 Remove SPE_ACC and SPEFSCR.
13703 (REG_ALLOC_ORDER): Ditto.
13704 (FRAME_POINTER_REGNUM): Change to 111.
13705 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13706 (REG_CLASS_NAMES): Ditto.
13707 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13708 (REGISTER_NAMES): Ditto.
13709 (ADDITIONAL_REG_NAMES): Ditto.
13710 (rs6000_reg_names): Ditto.
13711 * config/rs6000/rs6000.md: Renumber some register number
13714 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13716 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13718 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13719 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13721 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13722 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13723 Delete the SPE high registers.
13724 (REG_ALLOC_ORDER): Ditto.
13725 (enum reg_class): Remove SPE_HIGH_REGS.
13726 (REG_CLASS_NAMES): Ditto.
13727 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13728 (REGISTER_NAMES): Ditto.
13729 (rs6000_reg_names): Ditto.
13730 * doc/tm.texi.in: Remove SPE as example.
13731 * doc/tm.texi: Regenerate.
13733 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13735 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13736 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13737 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
13738 * config/rs6000/rs6000.md (type): Remove "brinc".
13740 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13742 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
13743 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
13744 * config/rs6000/linuxspe.h: Delete file.
13745 * config/rs6000/rs6000.md: Don't include spe.md.
13746 * config/rs6000/spe.h: Delete file.
13747 * config/rs6000/spe.md: Delete file.
13748 * config/rs6000/t-rs6000: Remove spe.md.
13750 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13752 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
13753 (reg_or_none500mem_operand): Delete.
13754 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
13755 instead of reg_or_none500mem_operand.
13757 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13759 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13760 handling of SPE flags.
13761 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
13763 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13765 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13767 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
13768 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
13769 paired_divv2sf3): Similar.
13770 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
13771 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
13772 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
13774 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
13775 Rename the paired_* instruction patterns.
13776 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
13778 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
13779 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
13780 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
13781 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
13782 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
13783 PAIRED_VECTOR_MODE.
13784 (struct machine_function): Delete field spe_insn_chain_scanned_p.
13785 (spe_func_has_64bit_regs_p): Delete.
13786 (spe_expand_predicate_builtin): Delete.
13787 (spe_expand_evsel_builtin): Delete.
13788 (TARGET_DWARF_REGISTER_SPAN): Do not define.
13789 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
13790 (invalid_e500_subreg): Delete.
13791 (rs6000_legitimize_address): Always force_reg op2 as well, for
13792 paired single memory accesses.
13793 (rs6000_member_type_forces_blk): Delete.
13794 (rs6000_spe_function_arg): Delete.
13795 (rs6000_expand_unop_builtin): Delete SPE handling.
13796 (rs6000_expand_binop_builtin): Ditto.
13797 (spe_expand_stv_builtin): Delete.
13798 (bdesc_2arg_spe): Delete.
13799 (spe_expand_builtin): Delete.
13800 (spe_expand_predicate_builtin): Delete.
13801 (spe_expand_evsel_builtin): Delete.
13802 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
13803 (spe_init_builtins): Delete.
13804 (spe_func_has_64bit_regs_p): Delete.
13805 (savres_routine_name): Delete "info" parameter. Adjust callers.
13806 (rs6000_emit_stack_reset): Ditto.
13807 (rs6000_dwarf_register_span): Delete.
13808 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
13809 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
13810 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
13812 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
13814 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
13815 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
13816 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
13817 mulv2sf3, divv2sf3): Delete expanders.
13819 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13821 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
13823 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13825 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
13826 * config/rs6000/rs6000.c: Ditto.
13828 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13830 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
13831 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
13832 comparison_operator.
13834 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13836 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
13837 * config/rs6000/rs6000.opt: Ditto.
13838 * config/rs6000/t-rtems: Ditto.
13840 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13842 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
13843 TARGET_E500_SINGLE by 0, simplify.
13844 * config/rs6000/rs6000.c: Ditto.
13845 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
13846 (spe_build_register_parallel): Delete.
13847 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
13848 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
13849 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
13850 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
13851 (E500_CONVERT): Delete.
13852 * config/rs6000/spe.md: Remove many patterns and all define_constants.
13854 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13856 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
13857 * config/rs6000/dfp.md: Ditto.
13858 (negdd2, *negdd2_fpr): Merge.
13859 (absdd2, *absdd2_fpr): Merge.
13860 (negtd2, *negtd2_fpr): Merge.
13861 (abstd2, *abstd2_fpr): Merge.
13862 * config/rs6000/e500.h: Delete file.
13863 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
13864 TARGET_FPRS by 1 and simplify.
13865 * config/rs6000/rs6000-c.c: Ditto.
13866 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
13867 TARGET_DF_SPE by 0.
13868 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
13870 * config/rs6000/rs6000.md: Ditto.
13871 (floatdidf2, *floatdidf2_fpr): Merge.
13872 (move_from_CR_gt_bit): Delete.
13873 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
13874 (E500_CR_IOR_COMPARE): Delete.
13875 (All patterns that require !TARGET_FPRS): Delete.
13876 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
13878 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13880 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
13882 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13884 * graphds.c (add_edge): Intitialize edge's attached data.
13885 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
13886 pointer parameter. Call pointed function on each edge during
13887 graph traversing. Skip traversing the edge when the function
13889 (graphds_dfs, graphds_scc): Ditto.
13890 (for_each_edge): New parameter. Pass the new parameter to callback
13892 * graphds.h (skip_edge_callback): New function pointer type.
13893 (graphds_dfs, graphds_scc): New function pointer parameter.
13894 (graphds_edge_callback, for_each_edge): New parameter.
13896 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13898 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
13899 out code checking if runtime alias check is possible to below ...
13900 Call the new function.
13901 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
13902 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
13904 2017-06-07 Marek Polacek <polacek@redhat.com>
13907 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
13908 TYPE_OVERFLOW_WRAPS check.
13910 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13912 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
13913 if versioning is required.
13914 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
13915 peeling with the check for versioning.
13917 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13919 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
13920 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
13921 Set true to new parameter if new ssa variable is defined.
13922 (vect_gen_vector_loop_niters): Refactor. Set range information
13923 for the new vector loop bound variable.
13924 (vect_do_peeling): Ditto.
13926 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13928 * tree-affine.c (ssa.h): Include header file.
13929 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
13930 has wrapping overflow behavior.
13932 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13934 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
13936 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13938 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
13939 (tree_to_aff_combination): ... here.
13941 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13943 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
13944 reg_pressure model function.
13945 (ivopts_global_cost_for_size): Delete.
13946 (determine_set_costs, iv_ca_recount_cost): Call new model function
13947 ivopts_estimate_reg_pressure.
13949 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13951 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
13952 expensive than udiv. Remove floating point cases from mod.
13954 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13956 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
13957 Increase idiv cost.
13959 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13961 * config/aarch64/aarch64.md
13962 (copysignsf3): Fix mask generation.
13964 2017-06-07 Jakub Jelinek <jakub@redhat.com>
13966 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
13968 (class dump_manager): Add register_dumps method.
13969 * dumpfile.c: Include langhooks.h.
13970 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
13971 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
13972 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
13973 (dump_manager::dump_register): Start with 512 entries instead of 32.
13974 (dump_manager::register_dumps): New method.
13975 * toplev.c (general_init): Instead of invoking register_dumps
13976 langhook, invoke register_dumps method on the dump manager.
13977 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13980 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13982 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13983 Say that # requires an associated define_split to exist, and that
13984 the define_split must be suitable for use after register allocation.
13986 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13988 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13989 (compute_outgoing_frequencies): Also initialize zero counts.
13990 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13991 around loops; preserve more of profile when nothing changes.
13993 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
13995 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
13997 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
13998 * config/arm/arm-cpu-cdata.h: Regenerate.
13999 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
14000 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
14001 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
14002 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
14003 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
14005 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
14007 * config/arm/t-rmprofile: Likewise.
14008 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
14010 2017-06-06 David S. Miller <davem@davemloft.net>
14013 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
14014 blockage if function uses alloca.
14016 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14018 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
14019 New "uid" fields to hold pretty-print IDs of group and ref.
14020 Memory references are now identified as <group_id>:<ref_id>
14021 instead of using [random] addresses.
14022 (dump_mem_details): Simplify, no functional change.
14023 (dump_mem_ref): Simplify and make output more concise.
14024 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
14025 (find_or_create_group): Initialize group uid.
14026 (record_ref): Initialize ref uid. Improve debug output.
14027 (prune_group_by_reuse, should_issue_prefetch_p,)
14028 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
14029 (mark_nontemporal_store, determine_loop_nest_reuse):
14030 Improve debug output.
14032 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14034 * dbgcnt.def (prefetch): New debug counter.
14035 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
14036 (schedule_prefetches): Stop issueing prefetches if debug counter
14039 2017-06-06 Tom de Vries <tom@codesourcery.com>
14041 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
14042 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
14044 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14046 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
14047 Use aarch64_reg_or_zero predicate for operand 4.
14048 (aarch64_compare_and_swap<mode> define_insn_and_split):
14049 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
14050 (aarch64_store_exclusive<mode>): Likewise for operand 2.
14052 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14054 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
14055 (arm_compute_save_core_reg_mask): This.
14056 (thumb1_compute_save_reg_mask): Rename into ...
14057 (thumb1_compute_save_core_reg_mask): This.
14058 (arm_compute_save_reg0_reg12_mask): Adapt comment.
14059 (arm_compute_frame_layout): Likewise.
14061 2017-06-06 Richard Biener <rguenther@suse.de>
14063 PR tree-optimization/80974
14064 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
14065 keep or clear leaders SSA info.
14067 2017-06-06 Tom de Vries <tom@codesourcery.com>
14069 * config/nvptx/nvptx.c (split_mode_p): New function.
14070 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
14072 2017-06-06 Tom de Vries <tom@codesourcery.com>
14074 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
14076 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14079 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
14082 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14084 * shrink-wrap.c (handle_simple_exit): Update profile.
14085 (try_shrink_wrapping): Upate profile.
14087 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14089 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
14090 (tree_guess_outgoing_edge_probabilities): New.
14091 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
14092 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
14094 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14096 * ipa-split.c (split_function): Initialize return bb profile.
14098 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
14100 * profile.c (compute_branch_probabilities): Also initialize
14101 EXIT_BLOCK profile.
14103 2017-06-06 Richard Biener <rguenther@suse.de>
14105 PR tree-optimization/80928
14106 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
14107 (vect_analyze_loop_operations): Properly guard analysis for
14109 (vect_transform_loop): Likewise.
14110 (vect_analyze_loop_2): Also reset SLP type on PHIs.
14111 (vect_model_induction_cost): Do not cost for pure SLP.
14112 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
14113 of induction in inner loop vectorization.
14114 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
14115 (vect_get_and_check_slp_defs): Handle vect_induction_def.
14116 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
14118 (vect_analyze_slp_cost_1): Cost induction.
14119 (vect_detect_hybrid_slp_stmts): Handle PHIs.
14120 (vect_get_slp_vect_defs): Likewise.
14121 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
14122 (vect_transform_stmt): Handle SLP reductions.
14123 * tree-vectorizer.h (vectorizable_induction): Adjust.
14125 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14127 * config/rs6000/rs6000.c (make_resolver_func): Update
14128 init_lowered_empty_function call.
14130 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
14132 * doc/invoke.texi: Document the -fprofile-abs-path option.
14133 * common.opt (fprofile-abs-path): New option.
14134 * gcov-io.h (gcov_write_filename): Declare.
14135 * gcov-io.c (gcov_write_filename): New function.
14136 * coverage.c (coverage_begin_function): Use gcov_write_filename.
14137 * profile.c (output_location): Likewise.
14139 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14141 * shring-wrap.c: Revert accidental commit.
14143 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
14145 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
14147 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14149 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
14151 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
14152 profile in callgraph edge.
14153 * profile-count.h (apply_probability): If THIS is 0, then result is 0
14154 (apply_scale): Likewise.
14155 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
14156 Also scale profile when inlining function with zero profile.
14157 (initialize_cfun): Update exit block profile even when it is zero.
14158 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
14159 when profile is read.
14161 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14163 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
14164 (CLONE_*): New constants to define the processors we can generate
14165 code for with the target_clone attribute.
14166 (rs6000_clone_map): New array to identify which clone processors
14167 the current program is running on.
14168 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
14169 target_clone attribute.
14170 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
14171 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
14172 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
14173 (cpu_expand_builtin): Add support for target_clone attribute.
14174 (rs6000_valid_attribute_p): Allow "default" attribute.
14175 (get_decl_name): New debug function to simplify printing the
14176 current function name in debugging statements.
14177 (rs6000_clone_priority): New functions to support the target_clone
14178 attribute, and be able to generate code to switch between ISA 2.05
14179 through ISA 3.0 (power6 through power9).
14180 (rs6000_compare_version_priority): Likewise.
14181 (rs6000_get_function_versions_dispatcher): Likewise.
14182 (make_resolver_func): Likewise.
14183 (add_condition_to_bb): Likewise.
14184 (dispatch_function_versions): Likewise.
14185 (rs6000_generate_version_dispatcher_body): Likewise.
14186 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
14187 (fusion_gpr_load_p): Fix a spacing issue.
14188 * doc/extend.texi (Common Function Attributes): Document that the
14189 PowerPC supports the target_clone attribute.
14191 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14193 * config/arm/arm.h: explain F symbol found in description of ARM
14194 register allocation in its legend.
14196 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14198 * config/mips/frame-header-opt.c: Include profile-count.h.
14199 * config/riscv/riscv.c: Include profile-count.h
14201 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14203 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
14205 (sm_set_flag_if_changed): Add bbs field.
14206 (execute_sm_if_changed_flag_set): Pass BBS.
14207 (execute_sm): Update.
14209 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14211 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
14214 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14216 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
14217 (peephole2): New peephole2 to emit the above.
14218 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
14220 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14222 * config/aarch64/aarch64.c (define_peephole2 above
14223 *sub_<shift>_<mode>): New peephole.
14225 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14227 * config/i386/i386.c (make_resolver_func): Update.
14228 * Makefile.in: Add profile-count.h and profile-count.o
14229 * auto-profile.c (afdo_indirect_call): Update to new API.
14230 (afdo_set_bb_count): Update.
14231 (afdo_propagate_edge): Update.
14232 (afdo_propagate_circuit): Update.
14233 (afdo_calculate_branch_prob): Update.
14234 (afdo_annotate_cfg): Update.
14235 * basic-block.h: Include profile-count.h
14236 (struct edge_def): Turn count to profile_count.
14237 (struct basic_block_def): Likewie.
14238 (REG_BR_PROB_BASE): Move to profile-count.h
14239 (RDIV): Move to profile-count.h
14240 * bb-reorder.c (max_entry_count): Turn to profile_count.
14241 (find_traces): Update.
14242 (rotate_loop):Update.
14243 (connect_traces):Update.
14244 (sanitize_hot_paths):Update.
14245 * bt-load.c (migrate_btr_defs): Update.
14246 * cfg.c (RDIV): Remove.
14247 (init_flow): Use alloc_block.
14248 (alloc_block): Uninitialize count.
14249 (unchecked_make_edge): Uninitialize count.
14250 (check_bb_profile): Update.
14251 (dump_edge_info): Update.
14252 (dump_bb_info): Update.
14253 (update_bb_profile_for_threading): Update.
14254 (scale_bbs_frequencies_int): Update.
14255 (scale_bbs_frequencies_gcov_type): Update.
14256 (scale_bbs_frequencies_profile_count): New.
14257 * cfg.h (update_bb_profile_for_threading): Update.
14258 (scale_bbs_frequencies_profile_count): Declare.
14259 * cfgbuild.c (compute_outgoing_frequencies): Update.
14260 (find_many_sub_basic_blocks): Update.
14261 * cfgcleanup.c (try_forward_edges): Update.
14262 (try_crossjump_to_edge): Update.
14263 * cfgexpand.c (expand_gimple_tailcall): Update.
14264 (construct_exit_block): Update.
14265 * cfghooks.c (verify_flow_info): Update.
14266 (dump_bb_for_graph): Update.
14267 (split_edge): Update.
14268 (make_forwarder_block): Update.
14269 (duplicate_block): Update.
14270 (account_profile_record): Update.
14271 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
14272 (get_estimated_loop_iterations): Update.
14273 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
14274 (single_likely_exit): Update.
14275 * cfgloopmanip.c (scale_loop_profile): Update.
14277 (set_zero_probability): Update.
14278 (lv_adjust_loop_entry_edge): Update.
14279 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14280 (purge_dead_edges): Update.
14281 (rtl_account_profile_record): Update.
14282 * cgraph.c (cgraph_node::create): Uninitialize count.
14283 (symbol_table::create_edge): Uninitialize count.
14284 (cgraph_update_edges_for_call_stmt_node): Update.
14285 (cgraph_edge::dump_edge_flags): Update.
14286 (cgraph_node::dump): Update.
14287 (cgraph_edge::maybe_hot_p): Update.
14288 * cgraph.h: Include profile-count.h
14289 (create_clone), create_edge, create_indirect_edge): Update.
14290 (cgraph_node): Turn count to profile_count.
14291 (cgraph_edge0: Likewise.
14292 (make_speculative, clone): Update.
14293 (create_edge): Update.
14294 (init_lowered_empty_function): Update.
14295 * cgraphclones.c (cgraph_edge::clone): Update.
14296 (duplicate_thunk_for_node): Update.
14297 (cgraph_node::create_clone): Update.
14298 * cgraphunit.c (cgraph_node::analyze): Update.
14299 (cgraph_node::expand_thunk): Update.
14300 * final.c (dump_basic_block_info): Update.
14301 * gimple-streamer-in.c (input_bb): Update.
14302 * gimple-streamer-out.c (output_bb): Update.
14303 * graphite.c (print_global_statistics): Update.
14304 (print_graphite_scop_statistics): Update.
14305 * hsa-brig.c: Include basic-block.h.
14306 * hsa-dump.c: Include basic-block.h.
14307 * hsa-gen.c (T sum_slice): Update.
14308 (convert_switch_statements):Update.
14309 * hsa-regalloc.c: Include basic-block.h.
14310 * ipa-chkp.c (chkp_produce_thunks): Update.
14311 * ipa-cp.c (struct caller_statistics): Update.
14312 (init_caller_stats): Update.
14313 (gather_caller_stats): Update.
14314 (ipcp_cloning_candidate_p): Update.
14315 (good_cloning_opportunity_p): Update.
14316 (get_info_about_necessary_edges): Update.
14317 (dump_profile_updates): Update.
14318 (update_profiling_info): Update.
14319 (update_specialized_profile): Update.
14320 (perhaps_add_new_callers): Update.
14321 (decide_about_value): Update.
14322 (ipa_cp_c_finalize): Update.
14323 * ipa-devirt.c (struct odr_type_warn_count): Update.
14324 (struct decl_warn_count): Update.
14325 (struct final_warning_record): Update.
14326 (possible_polymorphic_call_targets): Update.
14327 (ipa_devirt): Update.
14328 * ipa-fnsummary.c (redirect_to_unreachable): Update.
14329 * ipa-icf.c (sem_function::merge): Update.
14330 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
14331 * ipa-inline.c (compute_uninlined_call_time): Update.
14332 (compute_inlined_call_time): Update.
14333 (want_inline_small_function_p): Update.
14334 (want_inline_self_recursive_call_p): Update.
14335 (edge_badness): Update.
14336 (lookup_recursive_calls): Update.
14337 (recursive_inlining): Update.
14338 (inline_small_functions): Update.
14339 (dump_overall_stats): Update.
14340 (dump_inline_stats): Update.
14341 * ipa-profile.c (ipa_profile_generate_summary): Update.
14342 (ipa_propagate_frequency): Update.
14343 (ipa_profile): Update.
14344 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14345 * ipa-utils.c (ipa_merge_profiles): Update.
14346 * loop-doloop.c (doloop_modify): Update.
14347 * loop-unroll.c (report_unroll): Update.
14348 (unroll_loop_runtime_iterations): Update.
14349 * lto-cgraph.c (lto_output_edge): Update.
14350 (lto_output_node): Update.
14351 (input_node): Update.
14352 (input_edge): Update.
14353 (merge_profile_summaries): Update.
14354 * lto-streamer-in.c (input_cfg): Update.
14355 * lto-streamer-out.c (output_cfg): Update.
14356 * mcf.c (create_fixup_graph): Update.
14357 (adjust_cfg_counts): Update.
14358 (sum_edge_counts): Update.
14359 * modulo-sched.c (sms_schedule): Update.
14360 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
14361 * predict.c (maybe_hot_count_p): Update.
14362 (probably_never_executed): Update.
14363 (dump_prediction): Update.
14364 (combine_predictions_for_bb): Update.
14365 (propagate_freq): Update.
14366 (handle_missing_profiles): Update.
14367 (counts_to_freqs): Update.
14368 (rebuild_frequencies): Update.
14369 (force_edge_cold): Update.
14370 * predict.h: Include profile-count.h
14371 (maybe_hot_count_p, counts_to_freqs): UPdate.
14372 * print-rtl-function.c: Do not include cfg.h
14373 * print-rtl.c: Include basic-block.h
14374 * profile-count.c: New file.
14375 * profile-count.h: New file.
14376 * profile.c (is_edge_inconsistent): Update.
14377 (correct_negative_edge_counts): Update.
14378 (is_inconsistent): Update.
14379 (set_bb_counts): Update.
14380 (read_profile_edge_counts): Update.
14381 (compute_frequency_overlap): Update.
14382 (compute_branch_probabilities): Update; Initialize and deinitialize
14384 (branch_prob): Update.
14385 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
14386 (edge_gcov_count): New.
14387 (bb_gcov_count): New.
14388 * shrink-wrap.c (try_shrink_wrapping): Update.
14389 * tracer.c (better_p): Update.
14390 * trans-mem.c (expand_transaction): Update.
14391 (ipa_tm_insert_irr_call): Update.
14392 (ipa_tm_insert_gettmclone_call): Update.
14393 * tree-call-cdce.c: Update.
14394 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14395 (gimple_duplicate_sese_tail): Update.
14396 (gimple_account_profile_record): Update.
14397 (execute_fixup_cfg): Update.
14398 * tree-inline.c (copy_bb): Update.
14399 (copy_edges_for_bb): Update.
14400 (initialize_cfun): Update.
14401 (freqs_to_counts): Update.
14402 (copy_cfg_body): Update.
14403 (expand_call_inline): Update.
14404 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14405 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14406 (try_unroll_loop_completely): Update.
14407 (try_peel_loop): Update.
14408 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14409 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14410 * tree-ssa-loop-split.c (connect_loops): Update.
14411 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14412 * tree-ssa-reassoc.c (branch_fixup): Update.
14413 * tree-ssa-tail-merge.c (replace_block_by): Update.
14414 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14415 (compute_path_counts): Update.
14416 (update_profile): Update.
14417 (recompute_probabilities): Update.
14418 (update_joiner_offpath_counts): Update.
14419 (estimated_freqs_path): Update.
14420 (freqs_to_counts_path): Update.
14421 (clear_counts_path): Update.
14422 (ssa_fix_duplicate_block_edges): Update.
14423 (duplicate_thread_path): Update.
14424 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14425 (struct switch_conv_info): Update.
14426 * tree-tailcall.c (decrease_profile): Update.
14427 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14428 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14429 * value-prof.c (check_counter): Update.
14430 (gimple_divmod_fixed_value): Update.
14431 (gimple_mod_pow2): Update.
14432 (gimple_mod_subtract): Update.
14433 (gimple_ic_transform): Update.
14434 (gimple_stringop_fixed_value): Update.
14435 * value-prof.h (gimple_ic): Update.
14437 2017-06-02 Carl Love <cel@us.ibm.com>
14439 * config/rs6000/rs6000-c: Add support for built-in functions
14440 vector double vec_doublee (vector signed int);
14441 vector double vec_doublee (vector unsigned int);
14442 vector double vec_doublee (vector float);
14443 vector double vec_doubleh (vector signed int);
14444 vector double vec_doubleh (vector unsigned int);
14445 vector double vec_doubleh (vector float);
14446 vector double vec_doublel (vector signed int);
14447 vector double vec_doublel (vector unsigned int);
14448 vector double vec_doublel (vector float);
14449 vector double vec_doubleo (vector signed int);
14450 vector double vec_doubleo (vector unsigned int);
14451 vector double vec_doubleo (vector float);.
14452 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14453 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14455 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14456 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14457 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14459 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14460 vec_doublel, vec_doubleh.
14461 * doc/extend.texi: Update the built-in documentation file for the
14462 new built-in functions.
14464 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14467 * ipa-inline-analysis.c (free_growth_caches): Set
14468 edge_removal_hook_holder to NULL after removing it.
14470 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14472 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14473 comparision with zero.
14475 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14476 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14477 for early expansion of vec_min and vec_max builtins.
14478 (builtin_function_type): Add min/max unsigned variants to those
14479 identified as having unsigned arguments.
14481 2017-06-02 Olivier Hainque <hainque@adacore.com>
14483 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14485 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14487 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14488 Use VALL_F16 iterator rather than VALL.
14490 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14492 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14493 Emit CBNZ inside loop when doing a strong exchange and comparing
14494 against zero. Generate the CC flags after the loop.
14496 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14498 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14499 (dl_section_ref): New.
14500 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14501 On AIX, append an expression to subtract the size of the
14502 section length to dl_section_ref.
14504 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14506 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14507 for early expansion of vector absolute builtins.
14509 2017-06-02 Richard Biener <rguenther@suse.de>
14511 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14512 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14514 2017-06-02 Richard Biener <rguenther@suse.de>
14516 PR tree-optimization/80948
14517 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14518 stmt order as well.
14520 2017-06-02 Richard Biener <rguenther@suse.de>
14522 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14524 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14525 for inductions as relevant.
14527 2017-06-02 Richard Biener <rguenther@suse.de>
14529 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14530 (vectorizable_induction): ... this. Remove dead code.
14532 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14534 * builtins. (expand_builtin_alloca): Remove second parameter and
14535 infer its value from the first parameter instead.
14536 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14538 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14540 PR rtl-optimization/80903
14541 * loop-doloop.c (add_test): Unshare sequence.
14543 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14545 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14547 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14549 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14551 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14552 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14553 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14554 (xlogue_layout::get_stub_rtx): Make static.
14555 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14556 (xlogue_layout::compute_stub_managed_regs): Rename to...
14557 (xlogue_layout::count_stub_managed_regs): ...this.
14558 (xlogue_layout::is_stub_managed_reg): New function.
14559 (xlogue_layout::m_stub_names): Rename to...
14560 (xlogue_layout::s_stub_names): ...this, make static.
14561 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14562 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14563 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14564 xlogue_layout::s_stub_names): Instantiate statics.
14565 (stub_managed_regs): Remove.
14566 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14567 (disable_call_ms2sysv_xlogues): Rename to...
14568 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14569 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14571 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14572 change after reload_completed.
14573 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14575 (ix86_expand_prologue): Likewise.
14576 (ix86_expand_epilogue): Likewise.
14577 (ix86_expand_split_stack_prologue): Likewise.
14578 (ix86_compute_frame_layout): Remove frame parameter ...
14579 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14580 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14582 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14583 (ix86_frame): Move from here ...
14584 * config/i386/i386.h (ix86_frame): ... to here.
14585 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14586 complete ix86_frame data structure instead. Remove some_ld_name.
14588 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14590 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14591 symbols that hold a DECL_VALUE_EXPR.
14593 2017-06-01 Martin Jambor <mjambor@suse.cz>
14595 PR tree-optimization/80898
14596 * tree-sra.c (process_subtree_disqualification): Removed.
14597 (disqualify_candidate): Do not acll
14598 process_subtree_disqualification.
14599 (subtree_mark_written_and_enqueue): New function.
14600 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14601 RHS has been disqualified and re-queue LHS if necessary. Apart
14602 from that, ignore disqualified RHS.
14604 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14606 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14607 address fetch for z10 or later.
14609 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14611 * config/arc/arc.md (tst_movb): Add guard when splitting.
14613 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14615 * config/arc/arc.c (arc_can_eliminate): Test against
14616 arc_frame_pointer_needed.
14618 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14620 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14621 to prevent store reordering.
14622 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14623 (type): Add block type.
14624 (stack_tie): Define special instruction to be used in
14627 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14629 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14630 constraint. It is not valid for the pattern.
14631 (noncommutative_binary_comparison): Likewise.
14633 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14635 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
14638 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14640 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
14641 be used by the reg-alloc.
14643 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14645 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
14646 reg-alloc when having mul64 or mul32x16 instructions.
14647 (mulsidi3): Likewise.
14648 (umulsidi3): Likewise.
14649 (mulsi32x16): New pattern.
14650 (mulsi64): Likewise.
14651 (mulsidi64): Likewise.
14652 (umulsidi64): Likewise.
14653 (MUL32x16_REG): Define.
14654 (mul64_600): Use MUL32x16_REG.
14655 (mac64_600): Likewise.
14656 (umul64_600): Likewise.
14657 (umac64_600): Likewise.
14659 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14661 * config/arc/arc.md (mulsi3_700): Make it commutative.
14663 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14665 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14667 (*sign_extendsidi2_insn): Likewise for movstosw.
14669 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14671 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14672 the type of the input discriminant value. Convert the
14673 discriminant value of signedness vary.
14675 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14677 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14678 Add to -Wall section.
14680 2017-06-01 Richard Biener <rguenther@suse.de>
14682 PR middle-end/66313
14683 * fold-const.c (fold_plusminus_mult_expr): If the factored
14684 factor may be zero use a wrapping type for the inner operation.
14685 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14686 and handle moved defs.
14687 (process_assignment): Properly guard the unary op case. Return a
14688 tri-state indicating that moving the stmt before the call may allow
14689 to continue. Pass through to_move.
14690 (find_tail_calls): Handle moving unrelated defs before
14693 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14696 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14697 splitter result in the canonical way.
14699 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14701 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14702 also for 32bit target. Update insn attributes.
14703 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14705 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14707 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14708 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14709 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14710 (_mm_maskz_min_ss): New intrinsics.
14712 2017-05-31 Martin Liska <mliska@suse.cz>
14714 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14715 Change comment style to one we normally use.
14716 (vectorizable_reduction): Likewise.
14717 (vectorizable_induction): Likewise.
14718 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14719 (vectorizable_call): Likewise.
14720 (vectorizable_simd_clone_call): Likewise.
14721 (vectorizable_conversion): Likewise.
14722 (vectorizable_assignment): Likewise.
14723 (vectorizable_shift): Likewise.
14724 (vectorizable_operation): Likewise.
14725 (vectorizable_store): Likewise.
14726 (vectorizable_load): Likewise.
14727 * tree-vectorizer.h: Likewise.
14729 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14731 * passes.c (emergency_dump_function): New.
14732 * tree-pass.h (emergency_dump_function): Declare.
14733 * plugin.c (plugins_internal_error_function): Remove.
14734 * plugin.h (plugins_internal_error_function): Remove declaration.
14735 * toplev.c (internal_error_function): New static function. Use it...
14736 (general_init): ...here.
14738 2017-05-31 Graham Markall <graham.markall@embecosm.com>
14740 * config/arc/arc.c (arc_print_operand): Handle constant operands.
14741 (arc_rtx_costs): Add costs for new patterns.
14742 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
14743 * config/arc/predicates.md: Add _1_2_3_operand predicate.
14745 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14747 * tree-ssa-strlen.c (get_next_strinfo): New function.
14748 (get_stridx_plus_constant): Use it.
14749 (zero_length_string): Likewise.
14750 (adjust_related_strinfos): Likewise.
14751 (adjust_last_stmt): Likewise.
14753 2017-05-31 Richard Biener <rguenther@suse.de>
14756 * config/i386/i386.c (ix86_expand_builtin): Remove assert
14757 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
14759 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14761 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
14762 loop_vinfo argument and use of dependence distance vectors.
14763 Check instead whether the two references differ only in their
14764 initial value and assume that they have the same alignment if the
14765 difference is a multiple of the vector alignment.
14766 (vect_analyze_data_refs_alignment): Update call accordingly.
14768 2017-05-31 Martin Liska <mliska@suse.cz>
14771 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
14773 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14775 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
14776 (create_intersect_range_checks): Move from ...
14777 * tree-data-ref.c (create_intersect_range_checks_index)
14778 (create_intersect_range_checks): ... to here.
14779 (create_runtime_alias_checks): New function factored from ...
14780 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
14781 here. Call above function.
14782 * tree-data-ref.h (create_runtime_alias_checks): New function.
14784 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14786 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
14787 segment length for dr_b and compute it in wide_int.
14789 2017-05-31 Richard Biener <rguenther@suse.de>
14791 PR tree-optimization/80906
14792 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
14793 and pass through iv_map.
14794 (copy_bb_and_scalar_dependences): Adjust.
14795 (translate_pending_phi_nodes): Likewise.
14796 (copy_loop_close_phi_args): Handle code-generating IVs instead
14799 2017-05-30 David Malcolm <dmalcolm@redhat.com>
14801 * diagnostic-color.c (color_dict): Add "type-diff".
14802 (parse_gcc_colors): Update comment.
14803 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
14804 -fdiagnostics-show-template-tree and -fno-elide-type.
14805 (GCC_COLORS): Add type-diff to example.
14807 (-fdiagnostics-show-template-tree): New.
14808 (-fno-elide-type): New.
14809 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
14810 the pp_format_decoder callback. Call any m_format_postprocessor's
14812 (pretty_printer::pretty_printer): Initialize
14813 m_format_postprocessor.
14814 (pretty_printer::~pretty_printer): Delete any
14815 m_format_postprocessor.
14816 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
14817 (class format_postprocessor): New class.
14818 (struct pretty_printer::format_decoder): Document the new parameters.
14819 (struct pretty_printer::m_format_postprocessor): New field.
14820 * tree-diagnostic.c (default_tree_printer): Update for new
14821 bool and const char ** params.
14822 * tree-diagnostic.h (default_tree_printer): Likewise.
14824 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
14826 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
14827 (lwa_operand): Delete rs6000_gen_cell_microcode test.
14828 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14829 rs6000_gen_cell_microcode code.
14830 (rs6000_final_prescan_insn): Delete.
14831 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
14832 "warn-cell-microcode" entries.
14833 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
14834 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
14835 throughout. Change cc_reg_not_micro_cr0_operand to
14836 cc_reg_not_cr0_operand throughout.
14837 (*extendhi<mode>2_noload): Delete.
14838 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
14839 (mwarn-cell-microcode): Delete.
14840 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
14841 -mgen-cell-microcode and -mwarn-cell-microcode.
14843 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
14846 * config/i386/constraints.md (Yd): New constraint.
14848 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
14849 and (?Yd, r) alternatives. Update insn attributes.
14850 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
14851 and (?*Yd, r) alternatives. Update insn attributes.
14852 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
14854 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
14856 * gimplify.c (gimplify_modify_expr): Don't create a
14857 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
14860 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14862 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
14864 2017-05-30 Richard Biener <rguenther@suse.de>
14866 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
14867 and reduc_def fields.
14868 (STMT_VINFO_REDUC_TYPE): New define.
14869 (STMT_VINFO_REDUC_DEF): Likewise.
14870 (vect_force_simple_reduction): Adjust prototype.
14871 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
14872 (vect_is_simple_reduction): Remove check_reduction argument.
14873 (vect_force_simple_reduction): Adjust and set
14874 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14875 (vectorizable_reduction): Do not re-do reduction analysis
14876 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14877 * tree-parloops.c (gather_scalar_reductions): Adjust.
14879 2017-05-30 Richard Biener <rguenther@suse.de>
14881 PR middle-end/80901
14882 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
14885 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
14887 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
14888 Introduce unknown_misalignment parameter and remove vf.
14889 (vect_peeling_hash_get_lowest_cost):
14890 Pass unknown_misalignment parameter.
14891 (vect_enhance_data_refs_alignment):
14892 Fix unsupportable data ref treatment.
14894 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14896 * tree-vect-data-refs.c (vect_get_data_access_cost):
14897 Workaround for SLP handling.
14898 (vect_enhance_data_refs_alignment):
14899 Compute costs for doing no peeling at all, compare to the best
14900 peeling costs so far and avoid peeling if cheaper.
14902 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14904 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
14905 Return peeling info and set costs to zero for unlimited cost
14907 (vect_enhance_data_refs_alignment): Also inspect all datarefs
14908 with unknown misalignment. Compute and costs for unknown
14909 misalignment, compare them to the costs for known misalignment
14910 and choose the cheapest for peeling.
14912 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14914 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
14915 (vect_get_peeling_costs_all_drs): Create function.
14916 (vect_peeling_hash_get_lowest_cost):
14917 Use vect_get_peeling_costs_all_drs.
14918 (vect_peeling_supportable): Create function.
14919 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
14921 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14923 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
14924 DR_HAS_NEGATIVE_STEP.
14925 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
14926 (vect_enhance_data_refs_alignment): Use.
14927 (vect_duplicate_ssa_name_ptr_info): Use.
14928 * tree-vectorizer.h (dr_misalignment): Use.
14929 (known_alignment_for_access_p): Use.
14931 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14934 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
14936 (has_section_name): New function.
14938 2017-05-30 Martin Liska <mliska@suse.cz>
14941 * auto-profile.c (get_function_decl_from_block): Fix
14944 2017-05-30 Richard Biener <rguenther@suse.de>
14946 PR middle-end/80876
14947 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
14949 2017-05-30 Martin Liska <mliska@suse.cz>
14951 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
14952 * dumpfile.h (struct dump_file_info): Remove ctors.
14954 2017-05-30 Martin Liska <mliska@suse.cz>
14956 * predict.def: Fix GNU coding style.
14958 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14960 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14961 Mark 'to' argument with ATTRIBUTE_UNUSED.
14963 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14965 * config/xtensa/xtensa.c (xtensa_emit_call): Use
14966 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
14967 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
14970 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
14972 * doc/install.texi (Options specification): Restore entry of
14973 --enable-sjlj-exceptions.
14975 2017-05-27 Michael Eager <eager@eagercon.com>
14978 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14980 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14982 * config/microblaze/microblaze.h
14983 (FIXED_REGISTERS): Update in macro.
14984 (CALL_USED_REGISTERS): Update in macro.
14986 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14988 * doc/install.texi: Add links to macOS binary distributions.
14990 2017-05-27 Jakub Jelinek <jakub@redhat.com>
14994 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14996 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14998 2017-05-26 Martin Liska <mliska@suse.cz>
15000 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
15002 2017-05-26 Martin Liska <mliska@suse.cz>
15004 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
15005 always leading ';; '.
15006 (dump_bb_info): Likewise.
15007 (brief_dump_cfg): Likewise.
15008 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
15009 * dumpfile.c: Remove usage of TDF_VERBOSE.
15010 * dumpfile.h (enum dump_kind): Likewise.
15011 (dump_gimple_bb_header): Do not use TDF_COMMENT.
15012 * print-tree.c (debug_verbose): Remove.
15013 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
15014 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
15015 * tree-diagnostic.c (default_tree_printer): Replace
15016 TDF_DIAGNOSTIC with TDF_SLIM.
15018 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15020 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
15021 in parameter loop, rather than loop_vinfo.
15022 (create_intersect_range_checks): Ditto.
15023 (vect_create_cond_for_alias_checks): Update call to above functions.
15025 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15027 PR tree-optimization/80815
15028 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
15029 for merging runtime alias checks. Handle negative DR_STEPs.
15031 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15033 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
15035 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
15036 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
15037 out code pruning runtime alias checks.
15038 * tree-data-ref.c (prune_runtime_alias_test_list): New function
15039 factored out from above.
15040 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15042 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
15044 (prune_runtime_alias_test_list): New decalaration.
15046 2017-05-26 Bin Cheng <bin.cheng@arm.com>
15048 * tree-vect-data-refs.c (compare_tree): Rename and move ...
15049 * tree-data-ref.c (data_ref_compare_tree): ... to here.
15050 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
15051 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
15052 (operator==, comp_dr_with_seg_len_pair): Ditto.
15053 (vect_prune_runtime_alias_test_list): Ditto.
15055 2017-05-26 Martin Liska <mliska@suse.cz>
15058 * params.def: Bound partial-inlining-entry-probability param.
15060 2017-05-26 Marek Polacek <polacek@redhat.com>
15063 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
15066 2017-05-26 Richard Biener <rguenther@suse.de>
15068 PR tree-optimization/80842
15069 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
15072 2017-05-26 Richard Biener <rguenther@suse.de>
15074 PR tree-optimization/80844
15075 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
15077 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
15079 * doc/md.texi (Machine Constraints): Update x86 family
15080 machine constraints section to match 'config/i386/constraints.md'.
15082 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
15084 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
15086 2017-05-25 Nathan Sidwell <nathan@acm.org>
15088 * doc/invoke.texi (--enable-languages): Update documentation.
15090 2017-05-25 Martin Liska <mliska@suse.cz>
15092 * dumpfile.c: Add TDF_FOLDING.
15093 * dumpfile.h (enum dump_kind): Likewise.
15094 * genmatch.c (dt_simplify::gen_1): Use it.
15096 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15098 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
15100 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15102 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15103 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
15105 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15107 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
15108 * match.pd (X == C): Rewrite it here.
15109 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
15110 with_certain_nonzero_bits2): New predicates.
15111 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
15113 2017-05-24 Nathan Sidwell <nathan@acm.org>
15115 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
15118 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
15121 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15123 * config/powerpcspe: New port. Files are copied from the rs6000
15124 port, with "rs6000" in filenames replaced by "powerpcspe".
15126 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
15128 PR rtl-optimization/80754
15129 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
15131 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
15133 * config/sparc/sparc.md (length): Return the correct value for -mflat
15134 sibcalls to match output_sibcall.
15136 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15140 * config/rs6000/rs6000.c (struct machine_function): Add new field
15142 (rs6000_get_separate_components): Init that field, use it.
15143 (rs6000_components_for_bb): Use the field.
15145 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
15147 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
15149 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
15151 PR middle-end/80823
15152 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
15154 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15157 * config/s390/s390.c (s390_check_qrst_address): Check incoming
15158 address against address_operand predicate.
15159 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
15161 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
15163 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
15164 parameters passed indirectly.
15166 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
15168 * config/i386/i386.md (*movdi_internal): Remove SSE4
15169 alternative 18 (?r, *v). Update insn attributes.
15170 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
15171 Update insn attributes.
15172 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
15173 Update insn attributes.
15174 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
15175 alternative 1 (r, v). Remove isa attribute.
15176 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
15177 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
15178 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
15180 2017-05-23 Tom de Vries <tom@codesourcery.com>
15182 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
15185 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15187 * cgraphunit.c (symbol_table::process_new_functions): Update.
15188 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
15189 (inline_generate_summary): Rename to ...
15190 (ipa_fn_summary_generate): ... this one.
15191 (inline_read_summary): Rename to ...
15192 (ipa_fn_summary_read): ... this one.
15193 (inline_write_summary): Rename to ...
15194 (ipa_fn_summary_write): ... this one.
15195 (inline_free_summary): Rename to ...
15196 (ipa_free_fn_summary): ... this one.
15197 (pass_data_local_fn_summary, pass_local_fn_summary,
15198 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
15199 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
15200 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
15201 make_pass_ipa_fn_summary): New.
15202 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
15203 inline_write_summary, inline_free_summary): Remove.
15204 (ipa_free_fn_summary) : New.
15205 * ipa-inline.c (ipa_inline): Update.
15206 (pass_ipa_inline): Do not generate summaries.
15207 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
15209 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
15210 and add pass_ipa_fn_summary.
15211 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
15213 (make_pass_inline_parameters): Remove.
15215 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
15217 * omp-low.c (struct omp_context): Remove "default_kind" member.
15220 * omp-offload.c (execute_oacc_device_lower): Remove the
15221 parallelism dimensions function attributes for unparallelized
15222 OpenACC kernels constructs.
15224 2017-05-23 Martin Liska <mliska@suse.cz>
15226 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
15228 (cgraph_edge::make_speculative): Likewise.
15229 (cgraph_edge::resolve_speculation): Likewise.
15230 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15231 (cgraph_node::dump): Likewise.
15232 * cgraph.h: Likewise.
15233 * cgraphunit.c (analyze_functions): Likewise.
15234 (symbol_table::compile): Likewise.
15235 * ipa-cp.c (print_all_lattices): Likewise.
15236 (determine_versionability): Likewise.
15237 (initialize_node_lattices): Likewise.
15238 (ipcp_verify_propagated_values): Likewise.
15239 (estimate_local_effects): Likewise.
15240 (update_profiling_info): Likewise.
15241 (create_specialized_node): Likewise.
15242 (perhaps_add_new_callers): Likewise.
15243 (decide_about_value): Likewise.
15244 (decide_whether_version_node): Likewise.
15245 (identify_dead_nodes): Likewise.
15246 (ipcp_store_bits_results): Likewise.
15247 * ipa-devirt.c (dump_targets): Likewise.
15248 (ipa_devirt): Likewise.
15249 * ipa-icf.c (sem_item::dump): Likewise.
15250 (sem_function::equals): Likewise.
15251 (sem_variable::equals): Likewise.
15252 (sem_item_optimizer::read_section): Likewise.
15253 (sem_item_optimizer::execute): Likewise.
15254 (congruence_class::dump): Likewise.
15255 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
15256 (dump_inline_summary): Likewise.
15257 (estimate_node_size_and_time): Likewise.
15258 (inline_analyze_function): Likewise.
15259 * ipa-inline-transform.c (inline_call): Likewise.
15260 * ipa-inline.c (report_inline_failed_reason): Likewise.
15261 (want_early_inline_function_p): Likewise.
15262 (edge_badness): Likewise.
15263 (update_edge_key): Likewise.
15264 (inline_small_functions): Likewise.
15265 * ipa-profile.c (ipa_profile): Likewise.
15266 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
15267 (ipa_make_edge_direct_to_target): Likewise.
15268 (remove_described_reference): Likewise.
15269 (ipa_impossible_devirt_target): Likewise.
15270 (propagate_controlled_uses): Likewise.
15271 (ipa_print_node_params): Likewise.
15272 (ipcp_transform_function): Likewise.
15273 * ipa-pure-const.c (pure_const_read_summary): Likewise.
15274 (propagate_pure_const): Likewise.
15275 * ipa-reference.c (generate_summary): Likewise.
15276 (read_write_all_from_decl): Likewise.
15277 (propagate): Likewise.
15278 (ipa_reference_read_optimization_summary): Likewise.
15279 * ipa-utils.c (ipa_merge_profiles): Likewise.
15280 * ipa.c (walk_polymorphic_call_targets): Likewise.
15281 (symbol_table::remove_unreachable_nodes): Likewise.
15282 (ipa_single_use): Likewise.
15283 * passes.c (execute_todo): Likewise.
15284 * predict.c (drop_profile): Likewise.
15285 * symtab.c (symtab_node::get_dump_name): New function.
15286 (symtab_node::dump_name): Likewise.
15287 (symtab_node::dump_asm_name): Likewise.
15288 (symtab_node::dump_references): Likewise.
15289 (symtab_node::dump_referring): Likewise.
15290 (symtab_node::dump_base): Likewise.
15291 (symtab_node::debug_symtab): Likewise.
15292 * tree-sra.c (convert_callers_for_node): Likewise.
15293 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
15294 * value-prof.c (init_node_map): Likewise.
15296 2017-05-23 Martin Liska <mliska@suse.cz>
15298 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
15299 and symtab_node::debug_symtab to symbol_table::debug.
15300 * cgraphunit.c (analyze_functions): Use the renamed function.
15301 (symbol_table::compile): Likewise.
15302 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
15303 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
15304 * passes.c (execute_todo): Likewise.
15305 * symtab.c (symbol_table::dump): New function.
15306 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
15308 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15310 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
15311 that nonconst implies exec.
15313 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15315 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15316 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15317 (inline_edge_summary_vec): Turn into ...
15318 (ipa_call_summaries): ... this one.
15319 (redirect_to_unreachable, edge_set_predicate,
15320 evaluate_properties_for_edge, inline_summary_alloc,
15321 reset_ipa_call_summary, reset_inline_summary,
15322 inline_summary_t::duplicate): Update.
15323 (inline_edge_duplication_hook): Turn to ...
15324 (ipa_call_summary_t::duplicate): ... this one.
15325 (inline_edge_removal_hook): Turn to ...
15326 (ipa_call_summary_t::remove): ... this one.
15327 (dump_inline_edge_summary): Turn to ...
15328 (dump_ipa_call_summary): ... this one.
15329 (estimate_function_body_sizes): Update.
15330 (inline_update_callee_summaries): Update.
15331 (remap_edge_change_prob): Update.
15332 (remap_edge_summaries): Update.
15333 (inline_merge_summary): Update.
15334 (do_estimate_edge_time): Update.
15335 (inline_generate_summary): Update.
15336 (inline_read_section): Update.
15337 (inline_read_summary): Update.
15338 (inline_free_summary): Update.
15339 * ipa-inline.c (can_inline_edge_p): Update.
15340 (compute_inlined_call_time): Update.
15341 (want_inline_small_function_p): Update.
15342 (edge_badness): Update.
15343 (early_inliner): Update.
15344 * ipa-inline.h (inline_edge_summary): Turn to ...
15345 (ipa_call_summary): ... this one.
15346 (ipa_call_summary_t): New class.
15347 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15348 (ipa_call_summaries): New.
15349 (inline_edge_summary): Remove.
15350 (estimate_edge_growth): Update.
15351 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15352 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15353 * ipa-split.c (execute_split_functions): Update.
15354 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15356 2017-05-23 Tom de Vries <tom@codesourcery.com>
15358 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15359 attributes): Document rdrand effective target.
15361 2017-05-23 Tom de Vries <tom@codesourcery.com>
15363 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15364 attributes): Sort alphabetically.
15366 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
15368 * config/avr/genmultilib.awk: Use gsub instead of gensub.
15370 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15373 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
15374 V2DF/V2DI splat into two separate patterns, one that handles
15375 registers, and the other that only handles memory. Drop support
15376 for splatting from a GPR on ISA 2.07 and then splitting the
15377 splat into direct move and splat.
15378 (vsx_splat_<mode>_reg): Likewise.
15379 (vsx_splat_<mode>_mem): Likewise.
15381 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
15383 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
15385 2017-05-22 Jakub Jelinek <jakub@redhat.com>
15387 PR middle-end/80809
15388 * omp-low.c (finish_taskreg_remap): New function.
15389 (finish_taskreg_scan): If unit size of ctx->record_type
15390 is non-constant, unshare the size expression and replace
15391 decls in it with possible outer var refs.
15393 PR middle-end/80809
15394 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15395 GOVD_SHARED rather than GOVD_PRIVATE with it.
15396 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15397 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15399 PR middle-end/80853
15400 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15401 as last argument to build_outer_var_ref for pointer bases of array
15402 section reductions.
15404 2017-05-19 Martin Sebor <msebor@redhat.com>
15406 * print-tree.c (print_node): Print DECL_READ_P flag.
15408 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15410 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15411 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15412 * cgraph.c: Likewise.
15413 * cgraphunit.c: Likewise.
15414 * gengtype.c: Likewise.
15415 * ipa-cp.c: Likewise.
15416 * ipa-devirt.c: Likewise.
15417 * ipa-icf.c: Likewise.
15418 * ipa-predicate.c: Likewise.
15419 * ipa-profile.c: Likewise.
15420 * ipa-prop.c: Likewise.
15421 * ipa-split.c: Likewise.
15423 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15424 edge_predicate_pool, dump_inline_hints,
15425 inline_summary::account_size_time, redirect_to_unreachable,
15426 edge_set_predicate, set_hint_predicate,
15427 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15428 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15429 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15430 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15431 ipa_call_summary_t::remove, initialize_growth_caches,
15432 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15433 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15434 mark_modified, unmodified_parm_1, unmodified_parm,
15435 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15436 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15437 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15438 will_be_nonconstant_predicate, record_modified_bb_info,
15439 get_minimal_bb, record_modified, param_change_prob,
15440 phi_result_unknown_predicate, predicate_for_phi_result,
15441 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15442 estimate_function_body_sizes, compute_inline_parameters,
15443 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15444 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15445 inline_update_callee_summaries, remap_edge_change_prob,
15446 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15447 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15448 inline_analyze_function, inline_summary_t::insert,
15449 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15450 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15451 inline_free_summary): Move to ipa-fnsummary.h
15452 (predicate_t): Remove.
15453 * ipa-fnsummary.c: New file.
15454 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15455 (enum inline_hints_vals, inline_hints, agg_position_info,
15456 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15457 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15458 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15459 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15460 inline_read_summary, inline_write_summary, inline_free_summary,
15461 inline_analyze_function, initialize_inline_failed,
15462 inline_merge_summary, inline_update_overall_summary,
15463 compute_inline_parameters): Move to ipa-fnsummary.h
15464 * ipa-fnsummary.h: New file.
15465 * ipa-inline-transform.h: Include ipa-inline.h.
15466 * ipa-inline.c: LIkewise.
15468 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15470 * ipa-inline.c (edge_badness): Use inlined_time instead of
15471 inline_summaries->get.
15473 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15475 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15477 2017-05-22 Nathan Sidwell <nathan@acm.org>
15479 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15480 (fdump-lang): Document 'raw' option.
15481 * dumpfile.h (TDI_tu): Delete.
15482 * dumpfile.c (dump_files): Remove translation-unit.
15483 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15485 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15487 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15488 command option from $(AWK) call.
15489 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15490 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15491 [FORMAT]: Remove handling of variable.
15492 * config/avr/t-multilib: Regenerate.
15494 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15496 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15498 (dump_inline_summary): Do not print self_time.
15499 (estimate_function_body_sizes): Do not set self_time.
15500 (compute_inline_parameters): Likewise.
15501 (inline_read_section, inline_write_summary): Do not stream self_time.
15502 * ipa-inline.h (inline_summary): Drop self_time.
15504 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15506 * ipa-inline-analysis.c (account_size_time): Rename to ...
15507 (inline_summary::account_size_time): ... this one.
15508 (reset_ipa_call_summary): Turn to ...
15509 (ipa_call_summary::reset): ... this one.
15510 (reset_inline_summary): Turn to ...
15511 (inline_summary::reset): ... this one.
15512 (inline_summary_t::remove): Update.
15513 (inline_summary_t::duplicate): Update.
15514 (ipa_call_summary_t::remove): Update.
15515 (dump_inline_summary): Update.
15516 (estimate_function_body_sizes): Update.
15517 (compute_inline_parameters): Update.
15518 (estimate_node_size_and_time): Update.
15519 (inline_merge_summary): Update.
15520 (inline_update_overall_summary): Update.
15521 (inline_read_section): Update.
15522 (inline_write_summary): Update.
15523 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15524 add account_size_time and reset member functions.
15525 (ipa_call_summary): Add reset function.
15526 * ipa-predicate.h (predicate::operator &): Constify.
15528 2017-05-22 Richard Biener <rguenther@suse.de>
15530 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15532 2017-05-19 Jason Merrill <jason@redhat.com>
15534 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15536 2017-05-19 Marek Polacek <polacek@redhat.com>
15539 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15540 TYPE_OVERFLOW_WRAPS checks.
15542 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15544 * tree-core.h (enum omp_clause_default_kind): Add
15545 "OMP_CLAUSE_DEFAULT_PRESENT".
15546 * tree-pretty-print.c (dump_omp_clause): Handle it.
15547 * gimplify.c (enum gimplify_omp_var_data): Add
15548 "GOVD_MAP_FORCE_PRESENT".
15549 (gimplify_adjust_omp_clauses_1): Map it to
15550 "GOMP_MAP_FORCE_PRESENT".
15551 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15553 * gimplify.c (oacc_default_clause): Clarify.
15555 2017-05-19 Nathan Sidwell <nathan@acm.org>
15557 LANG_HOOK_REGISTER_DUMPS
15558 * toplev.c (general_init): Call register dump lang hook.
15559 * doc/invoke.texi: Document -fdump-lang option family.
15560 * dumpfile.c (dump_files): Remove class dump here.
15561 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15562 * dumpfile.h (tree_dump_index): Remove TDI_class.
15563 * langhooks-def.h (lhd_register_dumps): Declare.
15564 (LANG_HOOKS_REGISTER_DUMPS): Define.
15565 (LANG_HOOKS_INITIALIZER): Add it.
15566 * langhooks.c (lhd_register_dumps): Define.
15567 * langhooks.h (struct lang_hooks): Add register_dumps.
15569 2017-05-19 Nathan Sidwell <nathan@acm.org>
15571 * context.h (context::set_passes): New.
15572 * context.c (context::context): Do not create pass manager.
15573 * toplev.c (general_init): Create pass manager here.
15575 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15577 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15578 use this splitter if two add or or instructions would also work for
15579 the constant we want to generate.
15581 2017-05-19 Richard Biener <rguenther@suse.de>
15584 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15585 predicate evaluation.
15587 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15589 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15591 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15594 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15596 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15597 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15598 (inline_edge_summary_vec): Turn into ...
15599 (ipa_call_summaries): ... this one.
15600 (redirect_to_unreachable, edge_set_predicate,
15601 evaluate_properties_for_edge, inline_summary_alloc,
15602 reset_ipa_call_summary, reset_inline_summary,
15603 inline_summary_t::duplicate): Update.
15604 (inline_edge_duplication_hook): Turn to ...
15605 (ipa_call_summary_t::duplicate): ... this one.
15606 (inline_edge_removal_hook): Turn to ...
15607 (ipa_call_summary_t::remove): ... this one.
15608 (dump_inline_edge_summary): Turn to ...
15609 (dump_ipa_call_summary): ... this one.
15610 (estimate_function_body_sizes): Update.
15611 (inline_update_callee_summaries): Update.
15612 (remap_edge_change_prob): Update.
15613 (remap_edge_summaries): Update.
15614 (inline_merge_summary): Update.
15615 (do_estimate_edge_time): Update.
15616 (inline_generate_summary): Update.
15617 (inline_read_section): Update.
15618 (inline_read_summary): Update.
15619 (inline_free_summary): Update.
15620 * ipa-inline.c (can_inline_edge_p): Update.
15621 (compute_inlined_call_time): Update.
15622 (want_inline_small_function_p): Update.
15623 (edge_badness): Update.
15624 (early_inliner): Update.
15625 * ipa-inline.h (inline_edge_summary): Turn to ...
15626 (ipa_call_summary): ... this one.
15627 (ipa_call_summary_t): New class.
15628 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15629 (ipa_call_summaries): New.
15630 (inline_edge_summary): Remove.
15631 (estimate_edge_growth): Update.
15632 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15633 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15634 * ipa-split.c (execute_split_functions): Update.
15635 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15637 2017-05-19 Richard Biener <rguenther@suse.de>
15639 PR middle-end/80764
15640 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
15642 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
15644 * config/rs6000/rs6000.c (struct machine_function): Add field
15645 fpr_is_wrapped_separately.
15646 (rs6000_get_separate_components): Use 64 components. Handle the
15647 new FPR components.
15648 (rs6000_components_for_bb): Handle the FPR components.
15649 (rs6000_emit_prologue_components): Handle the FPR components.
15650 (rs6000_emit_epilogue_components): Handle the FPR components.
15651 (rs6000_set_handled_components): Handle the FPR components.
15652 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15653 that are already separately shrink-wrapped.
15654 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15655 that are already separately shrink-wrapped.
15657 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15660 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15663 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15664 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15665 cases where the register allocator uses a move and an offsettable
15666 memory operation to/from a FPR register on ISA 2.06/2.07.
15667 (define_peephole2 for Altivec d-form store): Likewise.
15669 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15672 * config/i386/mmx.md (*mov<mode>_internal): Enable
15673 alternatives 11, 12, 13 and 14 also for 32bit targets.
15674 Remove alternatives 15, 16, 17 and 18.
15675 * config/i386/sse.md (vec_concatv2di): Change
15676 alternative (!x, *y) to (x, ?!*Yn).
15678 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15680 * dumpfile.h (enum dump_kind): Remove stray comma.
15682 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15684 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15685 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15686 predicate::num_conditions
15687 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15688 (CHANGED): turn into predicate::changed.
15689 (agg_position_info): Move to ipa-predicate.h
15690 (add_condition, predicate::add_clause, predicate::operator &=,
15691 predicate::or_with, predicate::evaluate, predicate::probability,
15692 dump_condition, dump_clause, predicate::dump,
15693 predicate::remap_after_duplication, predicate::remap_after_inlining,
15694 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15695 (evaluate_conditions_for_known_args): Update.
15696 (set_cond_stmt_execution_predicate): Update.
15697 * ipa-inline.h: Include ipa-predicate.h
15698 (condition, inline_param_summary, conditions, agg_position_info,
15699 predicate): Move to ipa-predicate.h
15700 * ipa-predicate.c: New file.
15701 * ipa-predicate.h: New file.
15703 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15705 * final.c (leaf_function_p): Check we are not in a sequence.
15707 2017-05-18 Martin Liska <mliska@suse.cz>
15709 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15710 * dumpfile.c (dump_register): Use new enum dump_kind.
15711 (get_dump_file_name): Likewise.
15712 (dump_enable_all): Likewise.
15713 (dump_switch_p_1): Likewise.
15714 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15715 * dumpfile.h (enum dump_kind): New enum type.
15716 (struct dump_file_info): Create constructor and
15717 format fields and comments.
15718 * passes.c (pass_manager::register_one_dump_file):
15720 * statistics.c (statistics_early_init): Likewise.
15721 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15722 TDF_TREE with TDF_SLIM.
15723 (gather_memory_references_ref): Likewise.
15725 2017-05-18 Martin Liska <mliska@suse.cz>
15727 * vec.h (struct vnull): Use it.
15729 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15731 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15732 (true_predicate, false_predicate, true_predicate_p,
15733 false_predicate_p): Remove.
15734 (single_cond_predicate, not_inlined_predicate): Turn to member function
15736 (add_condition): Update.
15737 (add_clause): Turn to...
15738 (predicate::add_clause): ... this one; update; allow passing NULL
15740 (and_predicates): Turn to ...
15741 (predicate::operator &=): ... this one.
15742 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
15743 (or_predicates): Turn to ...
15744 (predicate::or_with): ... this one.
15745 (evaluate_predicate): Turn to ...
15746 (predicate::evaluate): ... this one.
15747 (predicate_probability): Turn to ...
15748 (predicate::probability): ... this one.
15749 (dump_condition): Update.
15750 (dump_predicate): Turn to ...
15751 (predicate::dump): ... this one.
15752 (account_size_time): Update.
15753 (edge_set_predicate): Update.
15754 (set_hint_predicate): UPdate.
15755 (evaluate_conditions_for_known_args): Update.
15756 (evaluate_properties_for_edge): Update.
15757 (remap_predicate_after_duplication): Turn to...
15758 (predicate::remap_after_duplication): ... this one.
15759 (remap_hint_predicate_after_duplication): Update.
15760 (inline_summary_t::duplicate): UPdate.
15761 (dump_inline_edge_summary): Update.
15762 (dump_inline_summary): Update.
15763 (set_cond_stmt_execution_predicate): Update.
15764 (set_switch_stmt_execution_predicate): Update.
15765 (compute_bb_predicates): Update.
15766 (will_be_nonconstant_expr_predicate): Update.
15767 (will_be_nonconstant_predicate): Update.
15768 (phi_result_unknown_predicate): Update.
15769 (predicate_for_phi_result): Update.
15770 (array_index_predicate): Update.
15771 (estimate_function_body_sizes): Update.
15772 (estimate_node_size_and_time): Update.
15773 (estimate_ipcp_clone_size_and_time): Update.
15774 (remap_predicate): Rename to ...
15775 (predicate::remap_after_inlining): ... this one.
15776 (remap_hint_predicate): Update.
15777 (inline_merge_summary): Update.
15778 (inline_update_overall_summary): Update.
15779 (estimate_size_after_inlining): Update.
15780 (read_predicate): Rename to ...
15781 (predicate::stream_in): ... this one.
15782 (read_inline_edge_summary): Update.
15783 (write_predicate): Rename to ...
15784 (predicate::stream_out): ... this one.
15785 (write_inline_edge_summary): Update.
15786 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
15787 (clause_t): Turn to uint32_t
15788 (predicate): Turn to class; implement constructor and operators
15790 (size_time_entry): Update.
15791 (inline_summary): Update.
15792 (inline_edge_summary): Update.
15794 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
15796 * fold-const.c (fold_binary_loc): Move transformation...
15797 * match.pd (C - X CMP X): ... here.
15799 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
15801 * config/sparc/sparc.c (sparc_option_override): Set function
15802 alignment for -mcpu=niagara7 to 64 to match the I$ line.
15803 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
15805 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
15807 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
15809 2017-05-18 Marek Polacek <polacek@redhat.com>
15812 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
15813 (pass_ubsan::execute): Call gimple_assign_single_p instead of
15814 gimple_assign_load_p.
15816 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
15818 PR middle-end/80692
15819 * real.c (do_compare): Give decimal_do_compare preference over
15820 comparing just the signs.
15822 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
15824 * doc/md.texi (Canonicalization of Instructions): Describe the
15825 canonical form of instructions that inherently set a condition
15828 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
15830 PR middle-end/80775
15831 * tree-cfg.c: Move deletion of unreachable case statements to after
15832 the merging of consecutive case labels.
15834 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
15836 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
15837 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
15838 restoring of callee-saved registers.
15840 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
15842 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
15843 * config/visium/visium.c (single_set_and_flags): Likewise.
15844 * config/visium/visium.md (Substitutions): Likewise.
15846 2017-05-17 Martin Liska <mliska@suse.cz>
15848 * cfg.c: Introduce dump_flags_t type and
15849 use it instead of int type.
15851 * cfghooks.c: Likewise.
15852 * cfghooks.h (struct cfg_hooks): Likewise.
15853 * cfgrtl.c: Likewise.
15854 * cfgrtl.h: Likewise.
15855 * cgraph.c (cgraph_node::get_body): Likewise.
15856 * coretypes.h: Likewise.
15857 * domwalk.c: Likewise.
15858 * domwalk.h: Likewise.
15859 * dumpfile.c (struct dump_option_value_info): Likewise.
15860 (dump_enable_all): Likewise.
15861 (dump_switch_p_1): Likewise.
15862 (opt_info_switch_p): Likewise.
15863 * dumpfile.h (enum tree_dump_index): Likewise.
15864 (struct dump_file_info): Likewise.
15865 * genemit.c: Likewise.
15866 * generic-match-head.c: Likewise.
15867 * gengtype.c (open_base_files): Likewise.
15868 * gimple-pretty-print.c: Likewise.
15869 * gimple-pretty-print.h: Likewise.
15870 * graph.c (print_graph_cfg): Likewise.
15871 * graphite-scop-detection.c (dot_all_sese): Likewise.
15872 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
15873 * loop-unroll.c (report_unroll): Likewise.
15874 * passes.c (pass_manager::register_one_dump_file): Likewise.
15875 * print-tree.c: Likewise.
15876 * statistics.c: Likewise.
15877 * tree-cfg.c: Likewise.
15878 * tree-cfg.h: Likewise.
15879 * tree-dfa.c: Likewise.
15880 * tree-dfa.h: Likewise.
15881 * tree-dump.c (dump_function): Likewise.
15882 * tree-dump.h (struct dump_info): Likewise.
15883 * tree-pretty-print.c: Likewise.
15884 * tree-pretty-print.h: Likewise.
15885 * tree-ssa-live.c: Likewise.
15886 * tree-ssa-live.h: Likewise.
15887 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
15888 * tree-vect-loop.c: Likewise.
15889 * tree-vect-slp.c: Likewise.
15891 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
15892 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15894 PR tree-optimization/80457
15895 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
15896 of all arguments to a statement as scalar_to_vec operations.
15897 (vectorizable_call): Adjust call to vect_model_simple_cost for
15899 (vectorizable_conversion): Likewise.
15900 (vectorizable_assignment): Likewise.
15901 (vectorizable_shift): Likewise.
15902 (vectorizable_operation): Likewise.
15903 (vectorizable_comparison): Likewise.
15904 (vect_is_simple_cond): Record the def types for operands.
15905 (vectorizable_condition): Likewise, call vect_model_simple_cost.
15906 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
15907 for statement argument count.
15909 2017-05-16 Carl Love <cel@us.ibm.com>
15911 * config/rs6000/rs6000-c: Add support for built-in functions
15912 vector unsigned long long vec_bperm (vector unsigned long long,
15913 vector unsigned char)
15914 vector signed long long vec_mule (vector signed int,
15916 vector unsigned long long vec_mule (vector unsigned int,
15917 vector unsigned int)
15918 vector signed long long vec_mulo (vector signed int,
15920 vector unsigned long long vec_mulo (vector unsigned int,
15921 vector unsigned int)
15922 vector signed char vec_sldw (vector signed char,
15923 vector signed char,
15925 vector unsigned char vec_sldw (vector unsigned char,
15926 vector unsigned char,
15928 vector signed short vec_sldw (vector signed short,
15929 vector signed short,
15931 vector unsigned short vec_sldw (vector unsigned short,
15932 vector unsigned short,
15934 vector signed int vec_sldw (vector signed int,
15937 vector unsigned int vec_sldw (vector unsigned int,
15938 vector unsigned int,
15940 vector signed long long vec_sldw (vector signed long long,
15941 vector signed long long,
15943 vector unsigned long long vec_sldw (vector unsigned long long,
15944 vector unsigned long long,
15946 * config/rs6000/rs6000-c: Add support for built-in functions
15947 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
15948 * config/rs6000/altivec.h: Add defintion for vec_sldw.
15949 * doc/extend.texi: Update the built-in documentation for the
15950 new built-in functions.
15952 2017-05-16 Marek Polacek <polacek@redhat.com>
15956 * tree.c (save_expr): Don't fold the expression.
15958 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
15960 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
15961 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
15962 and (?*y,m). Update insn attributes.
15964 2017-05-16 Martin Liska <mliska@suse.cz>
15966 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
15967 flags argument of print_gimple_stmt, print_gimple_expr,
15968 print_generic_stmt and print_generic_expr.
15969 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
15970 * coretypes.h: Likewise.
15971 * except.c (dump_eh_tree): Likewise.
15972 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15973 * gimple-pretty-print.h: Likewise.
15974 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
15975 (backprop::push_to_worklist): Likewise.
15976 (backprop::pop_from_worklist): Likewise.
15977 (backprop::process_use): Likewise.
15978 (backprop::intersect_uses): Likewise.
15979 (note_replacement): Likewise.
15980 * gimple-ssa-store-merging.c
15981 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15982 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15983 (pass_store_merging::execute): Likewise.
15984 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15985 (ssa_base_cand_dump_callback): Likewise.
15986 (dump_incr_vec): Likewise.
15987 (replace_refs): Likewise.
15988 (replace_mult_candidate): Likewise.
15989 (create_add_on_incoming_edge): Likewise.
15990 (create_phi_basis): Likewise.
15991 (insert_initializers): Likewise.
15992 (all_phi_incrs_profitable): Likewise.
15993 (introduce_cast_before_cand): Likewise.
15994 (replace_one_candidate): Likewise.
15995 * gimplify.c (gimplify_expr): Likewise.
15996 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
15997 (set_rename): Likewise.
15998 (rename_uses): Likewise.
15999 (copy_loop_phi_nodes): Likewise.
16000 (add_close_phis_to_merge_points): Likewise.
16001 (copy_loop_close_phi_args): Likewise.
16002 (copy_cond_phi_args): Likewise.
16003 (graphite_copy_stmts_from_block): Likewise.
16004 (translate_pending_phi_nodes): Likewise.
16005 * graphite-poly.c (print_pdr): Likewise.
16006 (dump_gbb_cases): Likewise.
16007 (dump_gbb_conditions): Likewise.
16008 (print_scop_params): Likewise.
16009 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
16010 (build_cross_bb_scalars_use): Likewise.
16011 (gather_bbs::before_dom_children): Likewise.
16012 * hsa-dump.c (dump_hsa_immed): Likewise.
16013 * ipa-cp.c (print_ipcp_constant_value): Likewise.
16014 (get_replacement_map): Likewise.
16015 * ipa-inline-analysis.c (dump_condition): Likewise.
16016 (estimate_function_body_sizes): Likewise.
16017 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
16018 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
16019 * ipa-prop.c (ipa_dump_param): Likewise.
16020 (ipa_print_node_jump_functions_for_edge): Likewise.
16021 (ipa_modify_call_arguments): Likewise.
16022 (ipa_modify_expr): Likewise.
16023 (ipa_dump_param_adjustments): Likewise.
16024 (ipa_dump_agg_replacement_values): Likewise.
16025 (ipcp_modif_dom_walker::before_dom_children): Likewise.
16026 * ipa-pure-const.c (check_stmt): Likewise.
16027 (pass_nothrow::execute): Likewise.
16028 * ipa-split.c (execute_split_functions): Likewise.
16029 * omp-offload.c (dump_oacc_loop_part): Likewise.
16030 (dump_oacc_loop): Likewise.
16031 * trans-mem.c (tm_log_emit): Likewise.
16032 (tm_memopt_accumulate_memops): Likewise.
16033 (dump_tm_memopt_set): Likewise.
16034 (dump_tm_memopt_transform): Likewise.
16035 * tree-cfg.c (gimple_verify_flow_info): Likewise.
16036 (print_loop): Likewise.
16037 * tree-chkp-opt.c (chkp_print_addr): Likewise.
16038 (chkp_gather_checks_info): Likewise.
16039 (chkp_get_check_result): Likewise.
16040 (chkp_remove_check_if_pass): Likewise.
16041 (chkp_use_outer_bounds_if_possible): Likewise.
16042 (chkp_reduce_bounds_lifetime): Likewise.
16043 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
16044 (chkp_mark_completed_bounds): Likewise.
16045 (chkp_register_incomplete_bounds): Likewise.
16046 (chkp_mark_invalid_bounds): Likewise.
16047 (chkp_maybe_copy_and_register_bounds): Likewise.
16048 (chkp_build_returned_bound): Likewise.
16049 (chkp_get_bound_for_parm): Likewise.
16050 (chkp_build_bndldx): Likewise.
16051 (chkp_get_bounds_by_definition): Likewise.
16052 (chkp_generate_extern_var_bounds): Likewise.
16053 (chkp_get_bounds_for_decl_addr): Likewise.
16054 * tree-chrec.c (chrec_apply): Likewise.
16055 * tree-data-ref.c (dump_data_reference): Likewise.
16056 (dump_subscript): Likewise.
16057 (dump_data_dependence_relation): Likewise.
16058 (analyze_overlapping_iterations): Likewise.
16059 * tree-inline.c (expand_call_inline): Likewise.
16060 (tree_function_versioning): Likewise.
16061 * tree-into-ssa.c (dump_defs_stack): Likewise.
16062 (dump_currdefs): Likewise.
16063 (dump_names_replaced_by): Likewise.
16064 (dump_update_ssa): Likewise.
16065 (update_ssa): Likewise.
16066 * tree-object-size.c (pass_object_sizes::execute): Likewise.
16067 * tree-parloops.c (build_new_reduction): Likewise.
16068 (try_create_reduction_list): Likewise.
16069 (ref_conflicts_with_region): Likewise.
16070 (oacc_entry_exit_ok_1): Likewise.
16071 (oacc_entry_exit_single_gang): Likewise.
16072 * tree-pretty-print.h: Likewise.
16073 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
16074 (get_scalar_evolution): Likewise.
16075 (add_to_evolution): Likewise.
16076 (get_loop_exit_condition): Likewise.
16077 (analyze_evolution_in_loop): Likewise.
16078 (analyze_initial_condition): Likewise.
16079 (analyze_scalar_evolution): Likewise.
16080 (instantiate_scev): Likewise.
16081 (number_of_latch_executions): Likewise.
16082 (gather_chrec_stats): Likewise.
16083 (final_value_replacement_loop): Likewise.
16084 (scev_const_prop): Likewise.
16085 * tree-sra.c (dump_access): Likewise.
16086 (disqualify_candidate): Likewise.
16087 (create_access): Likewise.
16088 (reject): Likewise.
16089 (maybe_add_sra_candidate): Likewise.
16090 (create_access_replacement): Likewise.
16091 (analyze_access_subtree): Likewise.
16092 (analyze_all_variable_accesses): Likewise.
16093 (sra_modify_assign): Likewise.
16094 (initialize_constant_pool_replacements): Likewise.
16095 (find_param_candidates): Likewise.
16096 (decide_one_param_reduction): Likewise.
16097 (replace_removed_params_ssa_names): Likewise.
16098 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
16099 * tree-ssa-copy.c (dump_copy_of): Likewise.
16100 (copy_prop_visit_cond_stmt): Likewise.
16101 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
16102 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
16103 (record_equivalences_from_stmt): Likewise.
16104 * tree-ssa-dse.c (compute_trims): Likewise.
16105 (delete_dead_call): Likewise.
16106 (delete_dead_assignment): Likewise.
16107 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
16108 (forward_propagate_into_cond): Likewise.
16109 (pass_forwprop::execute): Likewise.
16110 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16111 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
16113 (move_computations_worker): Likewise.
16114 (execute_sm): Likewise.
16115 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
16116 (remove_exits_and_undefined_stmts): Likewise.
16117 (remove_redundant_iv_tests): Likewise.
16118 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
16119 (adjust_iv_update_pos): Likewise.
16120 * tree-ssa-math-opts.c (bswap_replace): Likewise.
16121 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
16122 (value_replacement): Likewise.
16123 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
16124 * tree-ssa-pre.c (print_pre_expr): Likewise.
16125 (get_representative_for): Likewise.
16126 (create_expression_by_pieces): Likewise.
16127 (insert_into_preds_of_block): Likewise.
16128 (eliminate_insert): Likewise.
16129 (eliminate_dom_walker::before_dom_children): Likewise.
16130 (eliminate): Likewise.
16131 (remove_dead_inserted_code): Likewise.
16132 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
16133 * tree-ssa-reassoc.c (get_rank): Likewise.
16134 (eliminate_duplicate_pair): Likewise.
16135 (eliminate_plus_minus_pair): Likewise.
16136 (eliminate_not_pairs): Likewise.
16137 (undistribute_ops_list): Likewise.
16138 (eliminate_redundant_comparison): Likewise.
16139 (update_range_test): Likewise.
16140 (optimize_range_tests_var_bound): Likewise.
16141 (optimize_vec_cond_expr): Likewise.
16142 (rewrite_expr_tree): Likewise.
16143 (rewrite_expr_tree_parallel): Likewise.
16144 (linearize_expr): Likewise.
16145 (break_up_subtract): Likewise.
16146 (linearize_expr_tree): Likewise.
16147 (attempt_builtin_powi): Likewise.
16148 (attempt_builtin_copysign): Likewise.
16149 (transform_stmt_to_copy): Likewise.
16150 (transform_stmt_to_multiply): Likewise.
16151 (dump_ops_vector): Likewise.
16152 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
16153 (print_scc): Likewise.
16154 (set_ssa_val_to): Likewise.
16155 (visit_reference_op_store): Likewise.
16156 (visit_use): Likewise.
16157 (sccvn_dom_walker::before_dom_children): Likewise.
16158 (run_scc_vn): Likewise.
16159 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16161 (expr_hash_elt::print): Likewise.
16162 (const_and_copies::pop_to_marker): Likewise.
16163 (const_and_copies::record_const_or_copy_raw): Likewise.
16164 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
16165 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
16166 (dump_predicates): Likewise.
16167 (find_uninit_use): Likewise.
16168 (warn_uninitialized_phi): Likewise.
16169 (pass_late_warn_uninitialized::execute): Likewise.
16170 * tree-ssa.c (verify_vssa): Likewise.
16171 (verify_ssa): Likewise.
16172 (maybe_optimize_var): Likewise.
16173 * tree-vrp.c (dump_value_range): Likewise.
16174 (dump_all_value_ranges): Likewise.
16175 (dump_asserts_for): Likewise.
16176 (register_edge_assert_for_2): Likewise.
16177 (vrp_visit_cond_stmt): Likewise.
16178 (vrp_visit_switch_stmt): Likewise.
16179 (vrp_visit_stmt): Likewise.
16180 (vrp_visit_phi_node): Likewise.
16181 (simplify_cond_using_ranges_1): Likewise.
16182 (fold_predicate_in): Likewise.
16183 (evrp_dom_walker::before_dom_children): Likewise.
16184 (evrp_dom_walker::push_value_range): Likewise.
16185 (evrp_dom_walker::pop_value_range): Likewise.
16186 (execute_early_vrp): Likewise.
16188 2017-05-16 Richard Biener <rguenther@suse.de>
16190 * dwarf2out.c (loc_list_from_tree_1): Do not create
16191 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
16193 2017-05-16 Richard Biener <rguenther@suse.de>
16195 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
16197 (note_variable_value_in_expr): If we resolved the decl ref
16198 do not push to the stack.
16200 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
16202 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
16203 operations in fast-math mode.
16204 (vaddq_f16): Likewise.
16205 (vmul_f16): Likewise.
16206 (vmulq_f16): Likewise.
16207 (vsub_f16): Likewise.
16208 (vsubq_f16): Likewise.
16209 * config/arm/neon.md (add<mode>3): New.
16211 (fma:<VH:mode>3): New. Also remove outdated comment.
16214 2017-05-16 Martin Liska <mliska@suse.cz>
16218 * ipa-devirt.c (warn_types_mismatch): Fix typo.
16219 (odr_types_equivalent_p): Likewise.
16221 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
16223 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
16225 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
16228 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
16229 non-interunit SSE move alternatives with '?'.
16230 (zero-extendsidi peephole2): New peephole to skip intermediate
16231 general register in SSE zero-extend sequence.
16233 2017-05-15 Jeff Law <law@redhat.com>
16235 * reorg.c (relax_delay_slots): Create a new variable to hold
16236 the temporary target rather than clobbering TARGET_LABEL.
16238 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
16239 missing argument to extract_bit_field call.
16240 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
16242 2017-05-15 Martin Liska <mliska@suse.cz>
16245 * gcc.c (process_command): Do not allow empty argument of -o option.
16247 2017-05-15 Renlin Li <renlin.li@arm.com>
16249 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
16250 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
16251 * config/aarch64/constraints.md (Usf): Add long call check.
16252 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
16253 (call_value): Likewise.
16254 (sibcall): Likewise.
16255 (sibcall_value): Likewise.
16257 (call_value_insn): New.
16258 (sibcall_insn): Update rtx pattern.
16259 (sibcall_value_insn): Likewise.
16260 (call_internal): Remove.
16261 (call_value_internal): Likewise.
16262 (sibcall_internal): Likewise.
16263 (sibcall_value_internal): Likewise.
16264 (call_reg): Likewise.
16265 (call_symbol): Likewise.
16266 (call_value_reg): Likewise.
16267 (call_value_symbol): Likewise.
16269 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
16272 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
16274 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
16276 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
16277 compatible with CCGOCmode and with CCZmode.
16279 2017-05-14 Martin Sebor <msebor@redhat.com>
16281 PR middle-end/77671
16282 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
16283 (gimple_fold_builtin_snprintf): Same.
16284 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
16285 (gimple_fold_builtin_snprintf): Same.
16286 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
16287 of character types.
16288 (is_call_safe): New function.
16289 (try_substitute_return_value): Call it.
16290 (try_simplify_call): New function.
16291 (pass_sprintf_length::handle_gimple_call): Call it.
16293 2017-05-14 Martin Sebor <msebor@redhat.com>
16295 PR middle-end/80669
16296 * builtins.c (expand_builtin_stpncpy): Simplify.
16298 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
16300 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
16301 * config/i386/i386.h
16302 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
16303 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
16304 (struct machine_function): Add new members call_ms2sysv,
16305 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
16306 (struct machine_frame_state): New fields sp_realigned and
16307 sp_realigned_offset.
16308 * config/i386/i386.c
16309 (enum xlogue_stub): New enum.
16310 (enum xlogue_stub_sets): New enum.
16311 (class xlogue_layout): New class.
16312 (struct ix86_frame): New fields stack_realign_allocate_offset,
16313 stack_realign_offset and outlined_save_offset. Modify comments to
16314 detail stack layout when using out-of-line stubs.
16315 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
16316 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
16317 -mcall-ms2sysv-xlogues.
16318 (stub_managed_regs): New static variable.
16319 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
16320 registers managed by out-of-line stub.
16321 (disable_call_ms2sysv_xlogues): New function.
16322 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
16323 m->call_ms2sysv when appropriate and compute frame layout for
16325 (sp_valid_at, fp_valid_at): New inline functions.
16326 (choose_basereg): New function.
16327 (choose_baseaddr): Add align parameter, use choose_basereg and modify
16329 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
16330 Use align parameter of choose_baseaddr to generated aligned SSE movs
16332 (pro_epilogue_adjust_stack): Modify to track
16333 machine_frame_state::sp_realigned.
16334 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
16335 (ix86_nsaved_sseregs): Likewise.
16336 (ix86_emit_save_regs): Likewise.
16337 (ix86_emit_save_regs_using_mov): Likewise.
16338 (ix86_emit_save_sse_regs_using_mov): Likewise.
16339 (get_scratch_register_on_entry): Likewise.
16340 (gen_frame_set): New function.
16341 (gen_frame_load): Likewise.
16342 (gen_frame_store): Likewise.
16343 (emit_outlined_ms2sysv_save): Likewise.
16344 (emit_outlined_ms2sysv_restore): Likewise.
16345 (ix86_expand_prologue): Modify stack re-alignment code and call
16346 emit_outlined_ms2sysv_save when appropriate.
16347 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
16348 parameter rtx_insn *insn, which allows the function to be used to only
16349 generate the notes.
16350 (ix86_expand_epilogue): Modify validity checks of frame and stack
16351 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
16352 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
16353 * config/i386/predicates.md
16354 (save_multiple): New predicate.
16355 (restore_multiple): Likewise.
16356 * config/i386/sse.md
16357 (save_multiple<mode>): New pattern.
16358 (save_multiple_realign<mode>): Likewise.
16359 (restore_multiple<mode>): Likewise.
16360 (restore_multiple_and_return<mode>): Likewise.
16361 (restore_multiple_leave_return<mode>): Likewise.
16362 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
16364 2017-05-14 Julia Koval <julia.koval@intel.com>
16366 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
16367 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
16368 (__builtin_ia32_xsetbv): New builtins.
16369 * config/i386/i386.c (ix86_expand_special_args_builtin):
16371 (ix86_expand_builtin): Special expand for new intrinsics.
16372 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
16373 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
16374 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
16376 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16378 * cfganal.c (inverted_post_order_compute): Change argument type
16380 * cfganal.h (inverted_post_order_compute): Adjust prototype.
16381 * df-core.c (rest_of_handle_df_initialize): Adjust.
16382 (rest_of_handle_df_finish): Likewise.
16383 (df_analyze_1): Likewise.
16384 (df_analyze): Likewise.
16385 (loop_inverted_post_order_compute): Change argument to be a vec *.
16386 (df_analyze_loop): Adjust.
16387 (df_get_n_blocks): Likewise.
16388 (df_get_postorder): Likewise.
16389 * df.h (struct df_d): Change field to be a vec.
16390 * lcm.c (compute_laterin): Adjust.
16391 (compute_available): Likewise.
16392 * lra-lives.c (lra_create_live_ranges_1): Likewise.
16393 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
16394 * tree-ssa-pre.c (compute_antic): Likewise.
16396 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16398 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16399 (depth_first_search::depth_first_search): Change structure init
16400 function to this constructor.
16401 (depth_first_search::add_bb): Rename function to this member.
16402 (depth_first_search::execute): Likewise.
16403 (flow_dfs_compute_reverse_finish): Adjust.
16405 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16407 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16408 (longest_simple_path): Likewise.
16409 * shrink-wrap.c (spread_components): Likewise.
16410 (disqualify_problematic_components): Likewise.
16411 (emit_common_heads_for_components): Likewise.
16412 (emit_common_tails_for_components): Likewise.
16413 (insert_prologue_epilogue_for_components): Likewise.
16415 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16417 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16420 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16422 * df-core.c (df_set_blocks): Start using auto_bitmap.
16423 (df_compact_blocks): Likewise.
16424 * df-problems.c (df_rd_confluence_n): Likewise.
16425 * df-scan.c (df_insn_rescan_all): Likewise.
16426 (df_process_deferred_rescans): Likewise.
16427 (df_update_entry_block_defs): Likewise.
16428 (df_update_exit_block_uses): Likewise.
16429 (df_entry_block_bitmap_verify): Likewise.
16430 (df_exit_block_bitmap_verify): Likewise.
16431 (df_scan_verify): Likewise.
16432 * lra-constraints.c (lra_constraints): Likewise.
16433 (undo_optional_reloads): Likewise.
16434 (lra_undo_inheritance): Likewise.
16435 * lra-remat.c (calculate_gen_cands): Likewise.
16436 (do_remat): Likewise.
16437 * lra-spills.c (assign_spill_hard_regs): Likewise.
16438 (spill_pseudos): Likewise.
16439 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16440 (bitmap_set_subtract_values): Likewise.
16442 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16444 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16445 management with auto_bitmap.
16446 (fix_inter_tick): Likewise.
16447 (fix_recovery_deps): Likewise.
16448 * ira.c (add_store_equivs): Likewise.
16449 (find_moveable_pseudos): Likewise.
16450 (split_live_ranges_for_shrink_wrap): Likewise.
16451 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16452 (rtx_reuse_manager::seen_def_p): Likewise.
16453 (rtx_reuse_manager::set_seen_def): Likewise.
16454 * print-rtl.h (class rtx_reuse_manager): Likewise.
16456 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16458 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16460 (migrate_btr_def): Likewise.
16461 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16462 * df-core.c (loop_post_order_compute): Likewise.
16463 (loop_inverted_post_order_compute): Likewise.
16464 * hsa-common.h: Likewise.
16465 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16466 * init-regs.c (initialize_uninitialized_regs): Likewise.
16467 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16468 (inline_small_functions): Likewise.
16469 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16470 * ira.c (combine_and_move_insns): Likewise.
16471 (build_insn_chain): Likewise.
16472 * loop-invariant.c (find_invariants): Likewise.
16473 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16474 * predict.c (tree_predict_by_opcode): Likewise.
16475 (predict_paths_leading_to): Likewise.
16476 (predict_paths_leading_to_edge): Likewise.
16477 (estimate_loops_at_level): Likewise.
16478 (estimate_loops): Likewise.
16479 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16480 (spread_components): Likewise.
16481 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16482 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16483 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16484 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16485 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16486 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16487 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16488 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16489 (mark_threaded_blocks): Likewise.
16490 (thread_through_all_blocks): Likewise.
16491 * tree-ssa.c (verify_ssa): Likewise.
16492 (execute_update_addresses_taken): Likewise.
16493 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16495 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16497 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16499 (post_order_compute): Likewise.
16500 (inverted_post_order_compute): Likewise.
16501 (pre_and_rev_post_order_compute_fn): Likewise.
16503 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16505 * genrecog.c (int_set::int_set): Explicitly construct our
16506 auto_vec base class.
16507 * vec.h (auto_vec::auto_vec): New constructor.
16509 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16511 * bitmap.h (class auto_bitmap): New constructor taking
16512 bitmap_obstack * argument.
16514 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16516 * bitmap.h (class auto_bitmap): Change type of m_bits to
16517 bitmap_head, and adjust ctor / dtor and member operators.
16519 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16521 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16522 when returned register mode doesn't match original mode.
16524 2017-05-12 Jeff Law <law@redhat.com>
16525 Jakub Jelinek <jakub@redhat.com>
16527 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16528 we look for cc setter after the compare-elim changes.
16529 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16530 within the vector to match what compare-elim now expects.
16531 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16532 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16534 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16535 after the compare-elim changes.
16536 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16537 the vector to match what compare-elim now expects.
16538 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16539 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16540 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16541 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16542 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16544 * config/visium/visium.c (single_set_and_flags): Fix where
16545 we look for cc setter after the compare-elim changes.
16546 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16547 with the vector to match what compare-elim now expects.
16548 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16549 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16550 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16551 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16552 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16553 (neg<mode>2_insn_set_overflow): Likewise.
16555 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16557 PR middle-end/79794
16558 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16559 maybe_expand_insn call, set ops[0].target. If still set after call,
16560 set alt_rtl. Add extra arg to recursive calls.
16561 (extract_bit_field): Add alt_rtl argument. Pass to
16563 * expmed.h (extract_bit_field): Fix prototype.
16564 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16565 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16566 to extract_bit_field_calls.
16567 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16568 Pass alt_rtl to extract_bit_field calls.
16569 * calls.c (store_unaligned_arguments_into_psuedos)
16570 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16571 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16573 * optabs.h (struct expand_operand): Add target bitfield.
16575 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16577 * compare-elim.c (try_eliminate_compare): Canonicalize
16578 operation with embedded compare to
16579 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16580 (set (reg) (operation)].
16582 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16584 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16587 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16588 cost of adding a carry flag for ADC instruction.
16589 [case MINUS]: Ignore the cost of subtracting a carry flag
16590 for SBB instruction.
16592 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16594 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16596 * config/rs6000/bmiintrin.h: New file.
16597 * config/rs6000/bmi2intrin.h: New file.
16598 * config/rs6000/x86intrin.h: New file.
16600 2017-05-12 Jeff Law <law@redhat.com>
16602 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16605 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16607 PR middle-end/80707
16608 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16610 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16612 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16613 early expansion of vector divide builtins.
16614 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16615 builtins identified as having unsigned arguments.
16617 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16619 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16620 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16621 expansion of vector logical operations (and, andc, or, xor,
16624 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16626 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16627 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16629 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16631 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16632 early GIMPLE expansion of vector multiplies.
16634 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16636 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
16637 TARGET_HAVE_MOVT conditional.
16638 (movt splitter): Likewise.
16640 2017-05-12 Richard Biener <rguenther@suse.de>
16642 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
16643 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16644 Fold all stmts not inplace.
16646 2017-05-12 Richard Biener <rguenther@suse.de>
16648 PR tree-optimization/80713
16649 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
16650 inserted_exprs bit for not removed stmts.
16652 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16654 PR middle-end/69921
16655 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16656 parallelized" attribute for parallelized OpenACC kernels.
16657 * omp-offload.c (execute_oacc_device_lower): Use it.
16659 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16660 Set "oacc kernels" attribute.
16661 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16662 parameter. Adjust all users.
16663 (oacc_fn_attrib_kernels_p): Remove function.
16664 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16665 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16666 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16667 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16668 assert "oacc kernels" attribute is set.
16670 2017-05-11 Carl Love <cel@us.ibm.com>
16672 * config/rs6000/rs6000-c: Add support for built-in functions
16673 vector unsigned char vec_popcnt (vector signed char)
16674 vector unsigned char vec_popcnt (vector unsigned char)
16675 vector unsigned short vec_popcnt (vector signed short)
16676 vector unsigned short vec_popcnt (vector unsigned short)
16677 vector unsigned int vec_popcnt (vector signed int)
16678 vector unsigned int vec_popcnt (vector unsigned int)
16679 vector unsigned long long vec_popcnt (vector signed long long)
16680 vector unsigned long long vec_popcnt (vector unsigned long long)
16681 vector signed long long vec_slo (vector signed long long,
16682 vector signed char)
16683 vector signed long long vec_slo (vector signed long long,
16684 vector unsigned char)
16685 vector unsigned long long vec_slo (vector unsigned long long,
16686 vector signed char)
16687 vector unsigned long long vec_slo (vector unsigned long long,
16688 vector unsigned char)
16689 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16690 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16691 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16692 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16693 * doc/extend.texi: Update the built-in documentation file for the
16694 new built-in functions.
16696 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16698 * attribs.h (sorted_attr_string): Move machine independent
16699 functions for target clone support from the i386 port to common
16700 code. Rename ix86_function_versions to common_function_versions.
16701 Rename make_name to make_unique_name.
16702 (common_function_versions): Likewise.
16703 (make_unique_name): Likewise.
16704 (make_dispatcher_decl): Likewise.
16705 (is_function_default_version): Likewise.
16706 * attribs.c (attr_strcmp): Likewise.
16707 (sorted_attr_string): Likewise.
16708 (common_function_versions): Likewise.
16709 (make_unique_name): Likewise.
16710 (make_dispatcher_decl): Likewise.
16711 (is_function_default_version): Likewise.
16712 * config/i386/i386.c (attr_strcmp): Likewise.
16713 (sorted_attr_string): Likewise.
16714 (ix86_function_versions): Likewise.
16715 (make_name): Likewise.
16716 (make_dispatcher_decl): Likewise.
16717 (is_function_default_version): Likewise.
16718 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16720 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16723 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16724 Account for direct move costs for vec_construct of integer
16727 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16730 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16731 (UNSPEC_STX_ATOMIC): Ditto.
16732 (loaddi_via_sse): New insn.
16733 (storedi_via_sse): Ditto.
16734 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16735 Update corresponding peephole2 patterns.
16736 (atomic_storedi_fpu): Ditto.
16738 2017-05-11 Julia Koval <julia.koval@intel.com>
16740 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
16741 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
16743 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
16744 (__builtin_ia32_rsqrt14ss_mask): New builtins.
16745 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
16747 2017-05-11 Nathan Sidwell <nathan@acm.org>
16749 * graphite-poly.c: Include dumpfile.h.
16751 * dumpfle.h (dump_function): Declare here ...
16752 * tree-dump.h (dump_function): ... not here.
16753 * dumpfile.c: #include tree-cfg.h.
16754 (dump_function): Move here from ...
16755 * tree-dump.c (dump_function): ... here.
16756 * gimplify.c: #include splay-tree.h, not tree-dump.h.
16757 * graphite-poly.c: Don't include tree-dump.h.
16758 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
16759 * print-tree.c: Likewise.
16760 * stor-layout.c: Likewise.
16761 * tree-nested.c: Likewise.
16763 * dumpfile.c (dump_start): Use TDF_FLAGS.
16764 (dump_enable_all): Fix TDF_KIND check thinko.
16766 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16768 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16769 array entries to represent two legal parameterizations of the
16770 overloaded __builtin_cmpb function, as represented by the
16771 P6_OV_BUILTIN_CMPB constant.
16772 (altivec_resolve_overloaded_builtin): Add special case handling
16773 for the __builtin_cmpb function, as represented by the
16774 P6_OV_BUILTIN_CMPB constant.
16775 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
16776 (BU_P6_64BIT_2): New macro.
16777 (BU_P6_OVERLOAD_2): New macro
16778 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
16779 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
16780 (CMPB): Add overload support to represent both 32-bit and 64-bit
16781 compare-bytes function.
16782 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
16783 support for TARGET_CMPB.
16784 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
16785 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16786 documentation of the __builtin_cmpb overloaded built-in function.
16788 2017-05-11 Richard Biener <rguenther@suse.de>
16790 PR tree-optimization/80705
16791 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
16792 bases are not vectorizable.
16794 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16796 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
16797 when counting register pressure.
16799 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16801 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
16802 (iv_ca_more_deps): Renamed to ...
16803 (iv_ca_compare_deps): ... this.
16804 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
16806 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16808 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
16810 (determine_group_iv_costs): ... here.
16811 (find_inv_vars_cb): Record inv var if it's not recorded before.
16813 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16815 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
16816 (get_shiftadd_cost): Ditto.
16818 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16820 * tree-ssa-address.c: Include header file.
16821 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
16823 (add_to_parts): Refactor.
16824 (addr_to_parts): New parameter. Update use of move_hint_to_base.
16825 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
16828 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16830 PR tree-optimization/53090
16831 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
16833 (extract_cond_operands): Detect condition with IV on both sides
16834 and return COMP_IV_EXPR_2.
16835 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
16836 (rewrite_use_compare): Simplify by removing call to function
16837 extract_cond_operands.
16839 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16841 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
16842 (extract_cond_operands): Detect condition comparing against non-
16843 invariant bound and return appropriate enum value.
16844 (find_interesting_uses_cond): Update use of extract_cond_operands.
16845 Handle its return value accordingly.
16846 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
16848 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16850 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
16851 nonlinear iv_use computation in loop invariant sensitive way.
16853 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16855 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
16856 (find_iv_candidates): Call relate_compare_use_with_all_cands.
16858 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16860 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
16861 (dump_cand): Support iv_cand.inv_exprs.
16862 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
16864 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
16867 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16869 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
16871 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
16872 as local function. Include necessary header files.
16873 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
16875 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16877 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
16879 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16881 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
16882 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
16883 RSHIFT_EXPR and BIT_NOT_EXPR.
16885 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16887 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
16888 (adjust_setup_cost): New parameter supporting round up adjustment.
16889 (struct address_cost_data): Delete.
16890 (force_expr_to_var_cost): Don't bound cost with spill_cost.
16891 (split_address_cost, ptr_difference_cost): Delete.
16892 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
16893 (struct ainc_cost_data): New struct.
16894 (get_address_cost_ainc): New function.
16895 (get_address_cost, get_computation_cost): Reimplement.
16896 (determine_group_iv_cost_address): Record inv_expr for all uses of
16898 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
16899 (iv_ca_has_deps): Reimplemented to ...
16900 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
16902 (iv_ca_extend): Call iv_ca_more_deps.
16904 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16906 * tree-ssa-address.c (struct mem_address): Move to header file.
16907 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
16908 * tree-ssa-address.h (struct mem_address): Move from C file.
16909 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
16911 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16913 * tree-affine.h (aff_combination_type): New interface.
16914 (aff_combination_zero_p): Remove static.
16915 (aff_combination_const_p): New interface.
16916 (aff_combination_singleton_var_p): New interfaces.
16918 2017-05-11 Richard Biener <rguenther@suse.de>
16920 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16921 Skip unreachable blocks and destinations.
16922 (eliminate): Move stmt removal and fixup ...
16923 (fini_eliminate): ... here. Skip inserted exprs.
16924 (pass_pre::execute): Move fini_pre after fini_eliminate.
16925 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
16926 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
16927 PRE to get rid of dead code that has invalid SSA form and
16928 split critical edges again.
16930 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16932 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16934 2017-05-11 Richard Biener <rguenther@suse.de>
16936 * passes.c (execute_function_todo): Verify loops if they are
16937 said to be up-to-date.
16938 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
16939 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
16941 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
16944 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
16945 handle calling assemble_external ourself.
16948 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
16949 modes with zero size. Enhance comment.
16951 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16953 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
16954 built-ins for vec_xl and vec_xst with short and char pointer
16957 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16959 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
16960 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
16961 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
16962 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
16963 (_mm_maskz_min_round_ss): New intrinsics.
16964 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
16965 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
16966 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
16967 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
16968 (__builtin_ia32_minss_mask_round): New builtins.
16969 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16970 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16971 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
16973 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
16974 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
16976 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16979 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16981 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16982 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16983 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16984 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16985 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16986 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16987 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16988 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16989 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16990 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16991 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16992 (__builtin_ia32_mulss_mask_round): New builtins.
16993 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16994 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16995 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
16997 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
16998 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
17000 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
17003 2017-05-10 Julia Koval <julia.koval@intel.com>
17005 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
17006 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
17007 (_mm256_setr_m128i): New intrinsics.
17009 2017-05-10 Julia Koval <julia.koval@intel.com>
17011 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
17012 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
17013 (_mm_maskz_rcp14_ss): New intrinsics.
17014 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
17015 (__builtin_ia32_rcp14ss_mask): New builtins.
17016 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
17018 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
17020 PR tree-optimization/51513
17021 * tree-cfg.c (gimple_seq_unreachable_p): New function.
17022 (assert_unreachable_fallthru_edge_p): Use it.
17023 (group_case_labels_stmt): Likewise.
17024 * tree-cfg.h: Prototype it.
17025 * stmt.c: Include cfghooks.h and tree-cfg.h.
17026 (emit_case_dispatch_table) <gap_label>: New local variable.
17027 Use it to fill dispatch table gaps.
17028 Test for default_label before updating probabilities.
17029 (expand_case) <default_label>: Remove unneeded initialization.
17030 Test for unreachable default case statement and remove its edge.
17031 Set default_label accordingly.
17032 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
17034 2017-05-10 Carl Love <cel@us.ibm.com>
17036 * config/rs6000/rs6000-c: Add support for built-in functions
17037 vector signed char vec_neg (vector signed char)
17038 vector signed short int vec_neg (vector short int)
17039 vector signed int vec_neg (vector signed int)
17040 vector signed long long vec_neg (vector signed long long)
17041 vector float vec_neg (vector float)
17042 vector double vec_neg (vector double)
17043 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
17045 * config/rs6000/altivec.h: Add define for vec_neg
17046 * doc/extend.texi: Update the built-in documentation for the
17047 new built-in functions.
17049 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17051 PR tree-optimization/77644
17052 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
17054 2017-05-10 Nathan Sidwell <nathan@acm.org>
17056 * dumpfile.h (TDI_lang_all): New.
17057 (TDF_KIND): New. Renumber others
17058 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
17060 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
17062 (get_dump_file_name): Adjust suffix generation.
17063 (dump_enable_all): Use TDF_KIND.
17064 * doc/invoke.texi (-fdump-lang-all): Document.
17066 * dumpfile.h: Tabify.
17068 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
17071 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
17072 Move member access before delete.
17074 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
17076 * tree-inline.c (expand_call_inline): Split block at stmt
17079 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
17082 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
17083 are now unused after splitting mov{sf,sd}_hardfloat.
17084 (f32_lr2): Likewise.
17085 (f32_lm): Likewise.
17086 (f32_lm2): Likewise.
17087 (f32_li): Likewise.
17088 (f32_li2): Likewise.
17089 (f32_lv): Likewise.
17090 (f32_sr): Likewise.
17091 (f32_sr2): Likewise.
17092 (f32_sm): Likewise.
17093 (f32_sm2): Likewise.
17094 (f32_si): Likewise.
17095 (f32_si2): Likewise.
17096 (f32_sv): Likewise.
17097 (f32_dm): Likewise.
17098 (f32_vsx): Likewise.
17099 (f32_av): Likewise.
17100 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
17101 For movsf, order stores so the VSX stores occur before the GPR
17102 store which encourages the register allocator to use a traditional
17103 FPR instead of a GPR. For movsd, order the stores so that the GPR
17104 store comes before the VSX stores to allow the power6 to work.
17105 This is due to the power6 not having a 32-bit integer store
17106 instruction from a FPR.
17107 (movsf_hardfloat): Likewise.
17108 (movsd_hardfloat): Likewise.
17110 2017-05-09 Martin Sebor <msebor@redhat.com>
17112 PR translation/80280
17113 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
17116 PR translation/80280
17117 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
17118 data member added in r247778.
17119 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
17121 2017-05-09 Nathan Sidwell <nathan@acm.org>
17123 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
17125 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
17128 2017-05-09 Marek Polacek <polacek@redhat.com>
17130 * doc/invoke.texi: Fix typo.
17132 2017-05-09 Richard Biener <rguenther@suse.de>
17134 * tree-vrp.c (vrp_val_is_max): Adjust comment.
17135 (vrp_val_is_min): Likewise.
17136 (set_value_range_to_value): Likewise.
17137 (set_value_range_to_nonnegative): Likewise.
17138 (gimple_assign_nonzero_p): Likewise.
17139 (gimple_stmt_nonzero_p): Likewise.
17140 (vrp_int_const_binop): Likewise. Remove unreachable case.
17141 (adjust_range_with_scev): Adjust comments.
17142 (compare_range_with_value): Likewise.
17143 (extract_range_from_phi_node): Likewise.
17144 (test_for_singularity): Likewise.
17146 2017-05-09 Richard Biener <rguenther@suse.de>
17148 * tree-vrp.c (get_single_symbol): Add assert that we don't
17149 get overflowed constants as invariant part.
17150 (compare_values_warnv): Add comment before the TREE_NO_WARNING
17151 checks. Use wi::cmp instead of recursing for integer constants.
17152 (compare_values): Just ignore whether we assumed undefined
17153 overflow instead of failing the compare.
17154 (extract_range_for_var_from_comparison_expr): Add comment before the
17155 TREE_NO_WARNING sets.
17156 (test_for_singularity): Likewise.
17157 (extract_range_from_comparison): Do not disable optimization
17158 when we assumed undefined overflow.
17159 (extract_range_basic): Remove init of unused var.
17161 2017-05-09 Richard Biener <rguenther@suse.de>
17163 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
17164 (extract_range_from_multiplicative_op_1): Adjust.
17165 (extract_range_from_binary_expr_1): Use int_const_binop.
17167 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17170 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
17171 rs6000_store_data_bypass_p in seven define_bypass directives and
17172 in several comments.
17173 * config/rs6000/rs6000-protos.h: Add prototype for
17174 rs6000_store_data_bypass_p function.
17175 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
17176 function implements slightly different (rs6000-specific) semantics
17177 than store_data_bypass_p, returning false rather than aborting
17178 with assertion error when arguments do not satisfy the
17179 requirements of store data bypass.
17180 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
17181 rs6000_store_data_bypass_p.
17183 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
17185 * config/xtensa/xtensa-protos.h
17186 (xtensa_initial_elimination_offset): New declaration.
17187 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17188 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
17189 macro definition, add case for FRAME_POINTER_REGNUM when
17190 FRAME_GROWS_DOWNWARD.
17191 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
17192 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
17193 xtensa_initial_elimination_offset.
17195 2017-05-08 Nathan Sidwell <nathan@acm.org>
17197 * doc/invoke.texi: Alphabetize -fdump options.
17199 2017-05-08 Martin Sebor <msebor@redhat.com>
17201 PR translation/80280
17202 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
17204 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
17206 * target.def (compute_frame_layout): New optional target hook.
17207 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
17208 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
17209 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
17211 * reload1.c (verify_initial_elim_offsets): Likewise.
17212 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
17213 (use_simple_return_p): Call arm_compute_frame_layout if needed.
17214 (arm_get_frame_offsets): Split up into this ...
17215 (arm_compute_frame_layout): ... and this function.
17217 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
17219 * config/aarch64/constraints.md (Usa): New constraint.
17220 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
17222 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
17224 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
17225 with_multilib_list after it has been checked.
17227 2017-05-08 Richard Biener <rguenther@suse.de>
17229 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
17230 (bitmap_set_subtract_values): Likewise.
17232 2017-05-08 Richard Biener <rguenther@suse.de>
17234 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
17235 (gimple_assign_nonzero): ... this and remove strict_overflow_p
17237 (gimple_stmt_nonzero_warnv_p): Rename to ...
17238 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
17240 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
17241 (extract_range_basic): Adjust, do not disable propagation on
17242 strict overflow sensitive simplification.
17243 (vrp_visit_cond_stmt): Likewise.
17245 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
17247 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
17248 body size unconditionally.
17250 2017-05-07 Jeff Law <law@redhat.com>
17253 2017-05-06 Jeff Law <law@redhat.com>
17254 PR tree-optimization/78496
17255 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17258 PR tree-optimization/78496
17259 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17260 (simplify_stmt_using_ranges): Call it.
17261 (vrp_dom_walker::before_dom_children): Extract equivalences
17262 from an ASSERT_EXPR with an equality comparison against a
17265 2017-05-06 Jeff Law <law@redhat.com>
17267 PR tree-optimization/78496
17268 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17271 PR tree-optimization/78496
17272 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17273 (simplify_stmt_using_ranges): Call it.
17274 (vrp_dom_walker::before_dom_children): Extract equivalences
17275 from an ASSERT_EXPR with an equality comparison against a
17278 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17280 * lra-constraints.c (lra_copy_reg_equiv): New function.
17281 (split_reg): Use it to copy equivalence information from the
17282 original register to the spill register.
17284 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17286 PR rtl-optimization/75964
17287 * simplify-rtx.c (simplify_const_relational_operation): Remove
17288 invalid handling of comparisons of integer ABS.
17290 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
17292 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
17293 initialize to zero.
17294 (init_regs): Remove declaration.
17295 (function_arg_advance_32): Initialize error_p as boolean variable.
17297 2017-05-05 Nathan Sidwell <nathan@acm.org>
17299 * store-motion.c (remove_reachable_equiv_notes): Reformat long
17300 lines. Use for (;;).
17302 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17304 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
17305 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
17306 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
17307 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
17308 VF=2 that require versioning.
17310 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17312 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
17315 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17317 * diagnostic.h (diagnostic_override_option_index): Convert from
17318 macro to inline function.
17320 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17322 * diagnostic.c (last_module_changed_p): New function.
17323 (set_last_module): New function.
17324 (diagnostic_report_current_module): Convert macro usage to
17325 the above functions.
17326 * diagnostic.h (diagnostic_context::last_module): Strengthen
17327 from const line_map * to const line_map_ordinary *.
17328 (diagnostic_last_module_changed): Delete macro.
17329 (diagnostic_set_last_module): Delete macro.
17331 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17333 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
17334 with diagnostic_report_diagnostic.
17335 (diagnostic_n_impl_richloc): Likewise.
17336 * diagnostic.h (report_diagnostic): Delete macro.
17337 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
17338 with diagnostic_report_diagnostic.
17339 * substring-locations.c (format_warning_va): Likewise.
17341 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17343 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
17344 save/restor of format_spec. Move option-printing code to...
17345 (print_option_information): ...this new function, and
17346 reimplement by simply printing to the pretty_printer,
17347 rather than appending to the format string.
17349 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17351 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
17352 handling logic into...
17353 (update_effective_level_from_pragmas): ...this new function.
17355 2017-05-04 Andrew Waterman <andrew@sifive.com>
17357 * config/riscv/riscv.opt (mstrict-align): New option.
17358 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
17359 (SLOW_UNALIGNED_ACCESS): Define.
17360 (riscv_slow_unaligned_access): Declare.
17361 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
17363 (riscv_slow_unaligned_access): New variable.
17364 (rocket_tune_info): Set slow_unaligned_access to true.
17365 (optimize_size_tune_info): Set slow_unaligned_access to false.
17366 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
17367 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
17368 (riscv_option_override): Set riscv_slow_unaligned_access.
17369 * doc/invoke.texi: Add -mstrict-align to RISC-V.
17371 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
17373 * config/riscv/riscv.md: Unify indentation.
17375 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17380 * config/rs6000/rs6000.md (u code attribute): Add FIX and
17382 (extendsi<mode>2): Add support for doing sign extension via
17383 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
17384 don't have ISA 3.0 instructions.
17385 (extendsi<mode>2 splitter): Likewise.
17386 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
17387 generate the normal insns since SImode can now go in vector
17388 registers. Disallow the special UNSPECs needed for previous
17389 machines to hide SImode being used. Add new insns
17390 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
17391 (fix_trunc<mode>si2_stfiwx): Likewise.
17392 (fix_trunc<mode>si2_internal): Likewise.
17393 (fixuns_trunc<mode>si2): Likewise.
17394 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17395 (fctiw<u>z_<mode>_smallint): Likewise.
17396 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17397 of floating point to 32-bit integer from doing a direct move to
17398 the GPR registers to do a store.
17399 (fctiwz_<mode>): Break long line.
17401 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17403 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17404 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17405 (addr_list, addr_offset_valid_p): New.
17406 (split_address_groups): Check offset validity with above function.
17407 (gt-tree-ssa-loop-ivopts.h): Include header file.
17409 2017-05-05 Nathan Sidwell <nathan@acm.org>
17411 * config.gcc (arm*-*-*): Add missing 'fi'.
17413 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17415 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17416 included in -fopt-info does not matter.
17417 * doc/optinfo.texi (-fopt-info): Fix description of default
17418 behavour. Explicitly say order of options included in -fopt-info
17421 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17423 * config.gcc: Allow combinations of aprofile and rmprofile values for
17424 --with-multilib-list.
17425 * config/arm/t-multilib: New file.
17426 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17427 variables. Remove setting of ISA and floating-point ABI in
17428 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17429 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17430 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17431 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17433 * config/arm/t-rmprofile: Likewise except for the matches changes.
17434 * doc/install.texi (--with-multilib-list): Document the combination of
17435 aprofile and rmprofile values and warn about pitfalls in doing that.
17437 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17439 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17440 (movdi_aarch64): Likewise.
17442 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17444 PR tree-optimization/80632
17445 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17447 (build_arrays): Initialize it for virtual phis.
17448 (fix_phi_nodes): Use it for virtual phis.
17450 PR tree-optimization/80558
17451 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17452 [x, y] op z into [x op, y op z] for op & or | if conditions
17455 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17456 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17459 * config/arm/arm.md (use_literal_pool): Remove.
17460 (64-bit immediate split): No longer takes cost into consideration
17461 if arm_disable_literal_pool is enabled.
17462 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17463 used when arm_disable_literal_pool is enabled.
17464 (arm_max_const_double_inline_cost): Remove use of
17465 arm_disable_literal_pool.
17466 (push_minipool_fix): Add assert.
17467 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17468 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17469 (no_literal_pool_sf_immediate): New.
17471 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17473 PR tree-optimization/80613
17474 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17475 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17477 2017-05-05 Richard Biener <rguenther@suse.de>
17479 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17481 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17483 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17484 of this flag from insn conditions due to removal from r247495.
17486 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17488 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17490 (arm_early_store_addr_dep_ptr): Likewise.
17491 * config/arm/aarch-common-protos.h
17492 (arm_early_load_addr_dep_ptr): Add prototype.
17493 (arm_early_store_addr_dep_ptr): Likewise.
17494 * config/arm/cortex-a53.md: Add new bypasses.
17496 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17498 * tree.c (next_type_uid): Change type to unsigned.
17499 (type_hash_canon): Decrement back next_type_uid if
17500 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17501 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17504 2017-05-04 Martin Sebor <msebor@redhat.com>
17506 * builtins.c: Fix a trivial typo in a comment.
17508 PR middle-end/79234
17509 * builtins.c (check_sizes): Adjust to handle reading past the end.
17510 Avoid printing excessive upper bound of ranges. Use %E to print
17511 tree nodes instead of converting them to %wu.
17512 (expand_builtin_memchr): New function.
17513 (compute_dest_size): Rename...
17514 (compute_objsize): ...to this.
17515 (expand_builtin_memcpy): Adjust.
17516 (expand_builtin_mempcpy): Adjust.
17517 (expand_builtin_strcat): Adjust.
17518 (expand_builtin_strcpy): Adjust.
17519 (check_strncat_sizes): Adjust.
17520 (expand_builtin_strncat): Adjust.
17521 (expand_builtin_strncpy): Adjust and simplify.
17522 (expand_builtin_memset): Adjust.
17523 (expand_builtin_bzero): Adjust.
17524 (expand_builtin_memcmp): Adjust.
17525 (expand_builtin): Handle memcmp.
17526 (maybe_emit_chk_warning): Check strncat just once.
17528 2017-05-04 Martin Sebor <msebor@redhat.com>
17530 PR preprocessor/79214
17531 PR middle-end/79222
17532 PR middle-end/79223
17533 * builtins.c (check_sizes): Add inlining context and issue
17534 warnings even when -Wno-system-headers is set.
17535 (check_strncat_sizes): Same.
17536 (expand_builtin_strncat): Same.
17537 (expand_builtin_memmove): New function.
17538 (expand_builtin_stpncpy): Same.
17539 (expand_builtin): Handle memmove and stpncpy.
17541 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17543 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17544 which is not used any more.
17546 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17548 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17550 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17552 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17553 (cortexa53_tunings): Likewise.
17554 (cortexa57_tunings): Likewise.
17555 (cortexa72_tunings): Likewise.
17556 (cortexa73_tunings): Likewise.
17558 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17560 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17561 Set loop alignment to 8.
17563 2017-05-04 Martin Sebor <msebor@redhat.com>
17565 PR translation/80280
17566 * builtins.c (expand_builtin_object_size): Add missing quoting to
17567 %D and like directives.
17568 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17569 (hsa_type_for_tree_type): Same.
17570 (verify_function_arguments): Same.
17571 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17572 * varasm.c (get_section): Same.
17575 2017-05-04 Martin Sebor <msebor@redhat.com>
17577 PR translation/80280
17578 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17580 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17582 * config/aarch64/aarch64.c (generic_addrcost_table):
17583 Change HI/TI mode setting.
17585 2017-05-04 Martin Jambor <mjambor@suse.cz>
17587 PR tree-optimization/80622
17588 * tree-sra.c (comes_initialized_p): New function.
17589 (build_accesses_from_assign): Only set write lazily when
17590 comes_initialized_p is false.
17591 (analyze_access_subtree): Use comes_initialized_p.
17592 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17593 instead of testing for PARM_DECL.
17595 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17597 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17598 constraint on operand 0 to allow more general addressing modes.
17599 Adjust output template.
17600 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17602 * config/aarch64/aarch64-protos.h
17603 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17604 * config/aarch64/constraints.md (Dp): New address constraint.
17605 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17608 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17610 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17611 update use of estimate_ipcp_clone_size_and_time.
17612 (estimate_local_effects): Update use of
17613 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17614 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17615 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17616 Return nonspecialized time.
17618 2017-05-04 Richard Biener <rguenther@suse.de>
17620 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17621 for the last VUSE which def dominates the PHI. Directly call
17623 (get_continuation_for_phi_1): Remove.
17625 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17627 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17628 to explain the use of truncating division. Cap the number of
17629 iterations to the maximum given by nb_iterations_upper_bound,
17632 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17634 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17635 * configure: Regenerate.
17636 * config.in: Regenerate.
17637 * config/i386/driver-mingw32.c: new file.
17638 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
17639 * config.host: Link driver-mingw32.o on MinGW host.
17640 * doc/install.texi: Document new --enable-mingw-wildcard configure
17643 2017-05-04 Marek Polacek <polacek@redhat.com>
17645 PR tree-optimization/80612
17646 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
17648 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17649 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17651 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17652 (movt splitter): Likewise.
17653 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17654 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17655 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17656 block for Thumb-1 with MOVT.
17657 (thumb2_legitimate_address_p): Move code block ...
17658 (can_avoid_literal_pool_for_label_p): ... into this new function.
17659 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17661 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17662 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17663 "M-profile targets with the MOVT instruction".
17665 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17667 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17668 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17669 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17671 2017-05-04 Martin Liska <mliska@suse.cz>
17673 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17674 variable cond_code.
17676 2017-05-04 Richard Biener <rguenther@suse.de>
17678 * tree.c (array_at_struct_end_p): Handle arrays at struct
17679 end with flexarrays more conservatively. Refactor and treat
17680 arrays of arrays or aggregates more strict. Fix
17681 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17682 * tree.c (array_at_struct_end_p): Adjust prototype.
17683 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17684 * gimple-fold.c (get_range_strlen): Likewise.
17685 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17687 2017-05-04 Richard Biener <rguenther@suse.de>
17689 PR tree-optimization/31130
17690 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17692 (supports_overflow_infinity): Likewise.
17693 (is_negative_overflow_infinity): Likewise.
17694 (is_positive_overflow_infinity): Likewise.
17695 (is_overflow_infinity): Likewise.
17696 (stmt_overflow_infinity): Likewise.
17697 (overflow_infinity_range_p): Likewise.
17698 (usable_range_p): Remove as always returning true.
17699 (make_overflow_infinity): Remove.
17700 (negative_overflow_infinity): Likewise.
17701 (positive_overflow_infinity): Likewise.
17702 (avoid_overflow_infinity): Likewise.
17703 (set_value_range): Adjust accordingly.
17704 (set_value_range_to_nonnegative): Likewise, remove now unused
17705 overflow_infinity arg.
17706 (vrp_operand_equal_p): Adjust.
17707 (update_value_range): Likewise.
17708 (range_int_cst_singleton_p): Likewise.
17709 (operand_less_p): Likewise.
17710 (compare_values_warnv): Likewise.
17711 (extract_range_for_var_from_comparison_expr): Likewise.
17712 (vrp_int_const_binop): Likewise.
17713 (zero_nonzero_bits_from_vr): Likewise.
17714 (extract_range_from_multiplicative_op_1): Likewise.
17715 (extract_range_from_binary_expr_1): Likewise.
17716 (extract_range_from_unary_expr): Likewise.
17717 (extract_range_from_comparison): Likewise.
17718 (extract_range_basic): Likewise.
17719 (adjust_range_with_scev): Likewise.
17720 (compare_ranges): Likewise.
17721 (compare_range_with_value): Likewise.
17722 (dump_value_range): Likewise.
17723 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17725 (simplify_cond_using_ranges): Adjust.
17727 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17729 * brig-builtins.def: Added a builtin for class_f64.
17730 * builtin-types.def: Added a builtin type needed by class_f64.
17732 2017-05-03 Jason Merrill <jason@redhat.com>
17734 * timevar.def: Add TV_CONSTEXPR.
17736 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17738 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
17740 2017-05-03 Martin Jambor <mjambor@suse.cz>
17742 * ipa-prop.c (ipa_update_after_lto_read): Removed.
17743 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
17744 * ipa-cp.c (ipcp_propagate_stage): Do not call
17745 ipa_update_after_lto_read.
17746 * ipa-inline.c (ipa_inline): Likewise.
17748 2017-05-03 Martin Jambor <mjambor@suse.cz>
17750 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
17751 tag. Added a default constructor and a destructor.
17752 (ipa_edge_args_sum_t): New class;
17753 (ipa_edge_args_sum): Declare.
17754 (ipa_edge_args_vector): Remove declaration.
17755 (IPA_EDGE_REF): Use ipa_edge_args_sum.
17756 (ipa_free_edge_args_substructures): Remove declaration.
17757 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
17758 (ipa_edge_args_info_available_for_edge_p): Likewise.
17759 * ipa-prop.c (ipa_edge_args_vector): Removed.
17760 (edge_removal_hook_holder): Likewise.
17761 (edge_duplication_hook_holder): Likewise.
17762 (ipa_edge_args_sum): New variable.
17763 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
17764 ipa_edge_args_vector.
17765 (ipa_free_edge_args_substructures): Likewise.
17766 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
17767 ipa_edge_args_vector.
17768 (ipa_edge_removal_hook): Turned into method
17769 ipa_edge_args_sum_t::remove.
17770 (ipa_edge_duplication_hook): Turned into method
17771 ipa_edge_args_sum_t::duplicate.
17772 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
17773 registering edge hooks.
17774 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
17775 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
17776 ipa_edge_args_sum instead of ipa_edge_args_vector.
17777 * ipa-profile.c (ipa_profile): Likewise.
17779 2017-05-03 Martin Jambor <mjambor@suse.cz>
17781 * symbol-summary.h (function_summary): New method exists.
17782 (function_summary::symtab_removal): Deallocate through release.
17783 (call_summary): New class.
17784 (gt_ggc_mx): New overload.
17785 (gt_pch_nx): Likewise.
17786 (gt_pch_nx): Likewise.
17788 2017-05-03 Jeff Law <law@redhat.com>
17790 PR tree-optimization/78496
17791 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
17792 from simplify_cond_using_ranges. Split off code to walk
17793 backwards through casts into ...
17794 (simplify_cond_using_ranges_2): New function.
17795 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
17796 (execute_vrp): After identifying jump threads, call
17797 simplify_cond_using_ranges_2.
17799 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
17802 * ipa-inline.h (inline_summary): Add ctor.
17803 (create_ggc): Do not use ggc_cleared_alloc.
17805 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
17806 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17808 * gcc.c (handle_braces): Support escaping in switch matching
17810 * doc/invoke.texi (Spec Files): Document it.
17811 Remove superfluous @code markup in items.
17813 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17815 * diagnostic-show-locus.c (struct column_range): New struct.
17816 (get_affected_columns): New function.
17817 (get_printed_columns): New function.
17818 (struct correction): New struct.
17819 (correction::ensure_capacity): New function.
17820 (correction::ensure_terminated): New function.
17821 (struct line_corrections): New struct.
17822 (line_corrections::~line_corrections): New dtor.
17823 (line_corrections::add_hint): New function.
17824 (layout::print_trailing_fixits): Reimplement in terms of the new
17826 (selftest::test_overlapped_fixit_printing): New function.
17827 (selftest::diagnostic_show_locus_c_tests): Call it.
17829 2017-05-03 Nathan Sidwell <nathan@acm.org>
17831 Canonicalize canonical type hashing
17832 * tree.h (type_hash_canon_hash): Declare.
17833 * tree.c (type_hash_list, attribute_hash_list): Move into
17834 type_hash_canon_hash.
17835 (build_type_attribute_qual_variant): Break out hash code calc into
17836 type_hash_canon_hash.
17837 (type_hash_canon_hash): New. Generic type hash computation.
17838 (build_range_type_1, build_array_type_1, build_function_type,
17839 build_method_type_directly, build_offset_type, build_complex_type,
17840 make_vector_type): Call it.
17842 2017-05-03 Richard Biener <rguenther@suse.de>
17844 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
17845 When all DRs have unknown misaligned do not always peel
17846 when there is a store but apply the same costing model as if
17847 there were only loads.
17849 2017-05-03 Richard Biener <rguenther@suse.de>
17852 PR tree-optimization/80492
17853 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17854 compare_base_decls returning dont-know properly.
17856 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
17858 * config/arm/iterators.md (CCSI): New mode iterator.
17859 (arch): New mode attribute.
17860 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
17861 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
17862 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
17863 code iterator for success result mode.
17864 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
17865 the corresponding new insn generators.
17867 2017-05-03 Bin Cheng <bin.cheng@arm.com>
17870 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17871 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17873 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
17875 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
17876 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
17877 (DDR_A): Wrap DDR argument in brackets.
17878 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
17879 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
17880 (DDR_REVERSED_P): Likewise.
17882 2017-05-03 Jakub Jelinek <jakub@redhat.com>
17884 PR tree-optimization/79472
17885 * tree-switch-conversion.c (struct switch_conv_info): Add
17886 contiguous_range and default_case_nonstandard fields.
17887 (collect_switch_conv_info): Compute contiguous_range and
17888 default_case_nonstandard fields, don't clear final_bb if
17889 contiguous_range and only the default case doesn't have the required
17891 (check_all_empty_except_final): Set default_case_nonstandard instead
17892 of failing if contiguous_range and the default case doesn't have empty
17894 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
17895 and only the default case doesn't have the required constants. Skip
17897 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
17898 if default_case_nonstandard.
17899 (build_constructors): Build constant 1 just once. Assert that default
17900 values aren't inserted in between cases if contiguous_range. Skip
17902 (build_arrays): Skip virtual phis.
17903 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
17904 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
17905 Handle virtual phis.
17906 (gen_inbound_check): Handle default_case_nonstandard case.
17907 (process_switch): Adjust check_final_bb caller. Call
17908 gather_default_values with the first non-default case instead of
17909 default case if default_case_nonstandard.
17911 2017-05-02 Nathan Sidwell <nathan@acm.org>
17913 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
17914 check. Fix formatting.
17916 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
17918 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
17919 errors when comparing specialized and unspecialized times.
17921 2017-05-02 David Malcolm <dmalcolm@redhat.com>
17923 * diagnostic-show-locus.c
17924 (layout::should_print_annotation_line_p): Make private.
17925 (layout::print_annotation_line): Make private.
17926 (layout::annotation_line_showed_range_p): Make private.
17927 (layout::show_ruler): Make private.
17928 (layout::print_source_line): Make private. Pass in line and
17929 line_width, rather than calling location_get_source_line. Drop
17931 (layout::print_leading_fixits): New method.
17932 (layout::print_any_fixits): Rename to...
17933 (layout::print_trailing_fixits): ...this, and make private.
17934 Don't print newline fixits.
17935 (diagnostic_show_locus): Move logic for printing one row into...
17936 (layout::print_line): ...this new function. Move the
17937 location_get_source_line call and error-handling from
17938 print_source_line to here. Call print_leading_fixits, and rename
17939 print_any_fixits to print_trailing_fixits.
17940 (selftest::test_fixit_insert_containing_newline): Update now that
17941 newlines are partially supported.
17942 (selftest::test_fixit_insert_containing_newline_2): New test.
17943 (selftest::test_fixit_replace_containing_newline): Update comments.
17944 (selftest::diagnostic_show_locus_c_tests): Call the new test.
17945 * edit-context.c (class added_line): New class.
17946 (class edited_line): Describe newline handling in comment.
17947 (edited_line::actually_edited_p): New method.
17948 (edited_line::print_content): Delete redundant decl.
17949 (edited_line::m_predecessors): New field.
17950 (edited_file::print_content): Call edited_line::print_content.
17951 (edited_file::print_diff): Update to support newlines.
17952 (edited_file::print_diff_hunk): Likewise.
17953 (edited_file::print_run_of_changed_lines): New function.
17954 (edited_file::print_diff_line): Convert to...
17955 (print_diff_line): ...this.
17956 (edited_file::get_effective_line_count): New function.
17957 (edited_line::edited_line): Initialize new field m_predecessors.
17958 (edited_line::~edited_line): Clean up m_predecessors.
17959 (edited_line::apply_fixit): Handle newlines.
17960 (edited_line::get_effective_line_count): New function.
17961 (edited_line::print_content): New function.
17962 (edited_line::print_diff_lines): New function.
17963 (selftest::test_applying_fixits_insert_containing_newline): New
17965 (selftest::test_applying_fixits_replace_containing_newline): New
17967 (selftest::insert_line): New function.
17968 (selftest::test_applying_fixits_multiple_lines): Add example of
17970 (selftest::edit_context_c_tests): Call the new tests.
17972 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17974 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
17975 parameter cand. Update dump information.
17976 (get_computation_cost): Update uses.
17978 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17980 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17981 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17982 (get_computation_at, rewrite_use_address): Update use of
17983 get_computation_aff.
17985 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17987 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17988 (get_computation): Delete.
17989 (get_computation_cost): Implement like get_computation_cost_at.
17990 Use get_computation_at.
17991 (get_computation_cost_at): Delete.
17992 (rewrite_use_nonlinear_expr): Use get_computation_at.
17993 (rewrite_use_compare, remove_unused_ivs): Ditto.
17995 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17997 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
17999 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18001 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
18002 (ivopts_global_cost_for_size): Rename parameter and update uses.
18003 (iv_ca_recount_cost): Update uses.
18004 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
18005 candidates seperately in n_invs and n_cands.
18006 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
18008 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18010 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
18011 (find_inv_vars_cb): New.
18012 (find_depends): Renamed to ...
18013 (find_inv_vars): ... this.
18014 (add_candidate_1, force_var_cost): Call find_inv_vars.
18015 (split_address_cost, determine_group_iv_cost_cond): Ditto.
18017 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18019 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
18020 inv_vars. Add inv_exprs.
18021 (struct iv_cand): Rename depends_on to inv_vars.
18022 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
18023 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
18024 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
18025 (dump_cand): Dump inv_vars.
18026 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
18027 (record_invariant, find_depends, add_candidate_1): Ditto.
18028 (set_group_iv_cost, force_var_cost): Ditto.
18029 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
18030 (get_computation_cost_at, get_computation_cost): Ditto.
18031 (determine_group_iv_cost_generic): Ditto.
18032 (determine_group_iv_cost_address): Ditto.
18033 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
18034 (determine_group_iv_costs): Ditto.
18035 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
18036 (iv_ca_set_remove_invariants): Renamed to ...
18037 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
18038 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
18039 (iv_ca_set_add_invariants): Renamed to ...
18040 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
18041 (iv_ca_set_cp): Use iv_ca_set_add_invs.
18042 (iv_ca_has_deps): Support inv_vars and inv_exprs.
18043 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
18044 (create_new_ivs): Remove useless dump.
18046 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18048 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
18050 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
18051 (iv_ca_set_no_cp, create_new_iv): Ditto.
18053 2017-05-02 Bin Cheng <bin.cheng@arm.com>
18055 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18057 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18059 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
18060 function tree_check2.
18062 2017-05-02 Martin Liska <mliska@suse.cz>
18064 * doc/gcov.texi: Add missing preposition.
18065 * gcov.c (function_info::function_info): Properly fill up
18066 all member variables.
18068 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18070 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
18072 2017-05-02 Tamar Christina <tamar.christina@arm.com>
18074 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
18076 2017-05-02 Martin Liska <mliska@suse.cz>
18079 * lto-streamer-in.c (lto_read_tree_1): Remove
18080 LTO_STREAMER_DEBUG.
18081 * lto-streamer.c (struct tree_hash_entry): Likewise.
18082 (struct tree_entry_hasher): Likewise.
18083 (tree_entry_hasher::hash): Likewise.
18084 (tree_entry_hasher::equal): Likewise.
18085 (lto_streamer_init): Likewise.
18086 (lto_orig_address_map): Likewise.
18087 (lto_orig_address_get): Likewise.
18088 (lto_orig_address_remove): Likewise.
18089 * lto-streamer.h: Likewise.
18090 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
18091 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
18093 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
18095 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
18096 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
18097 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
18098 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
18099 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
18100 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
18101 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
18102 (mm_maskz_sub_ss): New intrinsics.
18103 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18104 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18105 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
18106 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
18107 (__builtin_ia32_subss_mask_round): New builtins.
18108 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18109 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18110 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
18112 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
18113 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18115 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18118 2017-05-02 Martin Jambor <mjambor@suse.cz>
18120 PR tree-optimization/78687
18121 * tree-sra.c (access): New field parent.
18122 (process_subtree_disqualification): New function.
18123 (disqualify_candidate): Call it.
18124 (build_accesses_from_assign): Reset write flag if creating an
18126 (build_access_subtree): Fill in parent field and also prpagate
18127 down grp_write flag.
18128 (create_artificial_child_access): New parameter set_grp_write, set
18129 grp_write to its value.
18130 (propagate_subaccesses_across_link): Also propagate grp_write flag
18132 (propagate_all_subaccesses): Push the closest parent back to work
18133 queue if add_access_to_work_queue returned true.
18135 2017-05-02 Richard Biener <rguenther@suse.de>
18137 * common.opt (fstrict-overflow): Alias negative to fwrapv.
18138 * doc/invoke.texi (fstrict-overflow): Remove all traces of
18139 -fstrict-overflow documentation.
18140 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
18141 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
18142 flag_strict_overflow.
18143 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
18144 * lto-opts.c (lto_write_options): Do not stream it.
18145 * lto-wrapper.c (merge_and_complain): Do not handle it.
18146 * opts.c (default_options_table): Do not set -fstrict-overflow.
18147 (finish_options): Likewise do not clear it when sanitizing.
18148 * simplify-rtx.c (simplify_const_relational_operation): Do not
18149 test flag_strict_overflow.
18151 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18153 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
18154 using enabled attribute.
18155 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
18156 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
18157 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
18158 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
18159 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
18160 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
18161 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
18162 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
18163 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
18164 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
18166 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18168 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
18170 2017-05-02 Richard Biener <rguenther@suse.de>
18172 PR tree-optimization/80591
18174 2017-04-10 Richard Biener <rguenther@suse.de>
18176 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18179 2017-05-02 Richard Biener <rguenther@suse.de>
18181 PR tree-optimization/80549
18182 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
18183 (cleanup_tree_cfg_noloop): Create forwarders to known loop
18184 headers if they do not have a preheader.
18186 2017-05-02 Martin Liska <mliska@suse.cz>
18189 * common.opt: Fix typo.
18190 * doc/invoke.texi: Likewise.
18192 2017-05-01 Jan Beulich <jbeulich@suse.com>
18194 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
18195 swapping, add (x,x,m,x,n) alternative.
18197 2017-05-01 Nathan Sidwell <nathan@acm.org>
18199 * calls.c (combine_pending_stack_adjustment_and_call): Remove
18200 unnecessary unadjusted_alignment check.
18202 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18205 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
18207 * gimplify.c (gimplify_cilk_detach): New function.
18208 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
18209 * tree-core.h: Document EXPR_CILK_SPAWN.
18210 * tree.h (EXPR_CILK_SPAWN): Define.
18212 2017-05-01 David Malcolm <dmalcolm@redhat.com>
18214 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
18215 to use new fixit_hint representation, using the "replace" logic.
18216 (get_line_span_for_fixit_hint): Likewise.
18217 (layout::print_any_fixits): Likewise.
18218 (selftest::test_one_liner_many_fixits): Rename to...
18219 (selftest::test_one_liner_many_fixits_1): ...this, and update
18220 comment and expected output to reflect that the multiple fix-it
18221 hints are now consolidated into one insertion.
18222 (selftest::test_one_liner_many_fixits_2): New test.
18223 (selftest::test_diagnostic_show_locus_one_liner): Update for
18225 (selftest::test_fixit_consolidation): Update for fix-it API
18227 * diagnostic.c (print_parseable_fixits): Likewise.
18228 * edit-context.c (edited_line::m_line_events): Convert from
18229 auto_vec <line_event *> to auto_vec <line_event>.
18230 (class line_event): Convert from abstract base class to a concrete
18231 class, taking over the role of replace_event.
18232 (class insert_event): Delete.
18233 (class replace_event): Rename to class line_event. Convert to
18235 (edit_context::add_fixits): Reimplement.
18236 (edit_context::apply_insert): Delete.
18237 (edit_context::apply_replace): Rename to...
18238 (edit_context::apply_fixit): ...this. Convert to half-open range.
18239 (edited_file::apply_insert): Delete.
18240 (edited_file::apply_replace): Rename to...
18241 (edited_file::apply_fixit): ...this.
18242 (edited_line::~edited_line): Drop deletion of events.
18243 (edited_line::apply_insert): Delete.
18244 (edited_line::apply_replace): Rename to...
18245 (edited_line::apply_fixit): ...this. Convert to half-open range.
18246 Update for change to type of m_line_events.
18247 * edit-context.h (edit_context::apply_insert): Delete.
18248 (edit_context::apply_replace): Rename to...
18249 (edit_context::apply_fixit): ...this.
18251 2017-05-01 Martin Sebor <msebor@redhat.com>
18253 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
18256 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
18259 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
18260 __get_cpuid_max returns 0.
18261 (__get_cpuid_count): Ditto.
18263 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
18265 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
18266 replacement expression is another instance of one of its arguments.
18268 2017-05-01 Jakub Jelinek <jakub@redhat.com>
18271 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
18272 check for stack push/pop autoinc.
18273 * config/i386/i386.c (ix86_agi_dependent): Return false
18274 if the only reason why modified_in_p returned true is that
18275 addr is SP based and set_insn is a push or pop.
18277 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18279 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
18282 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18285 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
18286 (account_size_time): Use two predicates - exec_pred and
18288 (evaluate_conditions_for_known_args): Compute both clause and
18290 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
18291 (inline_summary_t::duplicate): Update.
18292 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
18294 (compute_inline_parameters): Likewise.
18295 (estimate_edge_size_and_time): Update caluclation of time.
18296 (estimate_node_size_and_time): Compute both time and nonspecialized
18298 (estimate_ipcp_clone_size_and_time): Update.
18299 (inline_merge_summary): Update.
18300 (do_estimate_edge_time): Update.
18301 (do_estimate_edge_size): Update.
18302 (do_estimate_edge_hints): Update.
18303 (inline_read_section, inline_write_summary): Stream both new predicates.
18304 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
18306 (compute_inlined_call_time): Cleanup.
18307 (big_speedup_p): Update.
18308 (edge_badness): Update.
18309 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
18310 (size_time_entry): Replace predicate by exec_predicate and
18311 nonconst_predicate.
18312 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
18313 (estimate_edge_time): Return also nonspec_time.
18314 (reset_edge_growth_cache): Update.
18316 2017-04-29 Jakub Jelinek <jakub@redhat.com>
18318 PR rtl-optimization/80491
18319 * ifcvt.c (noce_process_if_block): When looking for x setter
18320 with missing else_bb, don't check only the insn right before
18321 cond_earliest, but look for the last insn that x is modified in
18322 within the same bb.
18324 PR rtl-optimization/80491
18325 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
18327 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
18329 PR tree-optimization/80487
18330 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
18332 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18334 PR tree-optimization/79697
18335 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
18336 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
18337 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18339 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
18340 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
18342 2017-04-28 Martin Sebor <msebor@redhat.com>
18344 PR tree-optimization/80523
18345 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
18346 (init_target_to_host_charmap, target_to_host, target_strtol10): New
18348 (maybe_warn, format_directive, parse_directive): Use new functions.
18349 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
18351 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
18353 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
18355 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18357 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
18358 target_header_dir): Set correctly.
18359 * configure: Regenerated.
18360 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18361 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18362 instead of SYSTEM_HEADER_DIR.
18364 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
18366 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
18367 (estimate_local_effects): Likewise.
18368 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
18369 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
18370 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
18371 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
18372 do_estimate_edge_time, estimate_edge_time): Likewise.
18373 * ipa-inline-analysis.c (estimate_node_size_and_time,
18374 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
18375 (estimate_time_after_inlining): Remove.
18377 2017-04-28 Martin Liska <mliska@suse.cz>
18379 * doc/gcov.texi: Enhance documentation of gcov.
18381 2017-04-28 Martin Liska <mliska@suse.cz>
18383 * doc/gcov.texi: Sort options in alphabetic order.
18384 * doc/gcov-dump.texi: Likewise.
18385 * doc/gcov-tool.texi: Likewise.
18386 * gcov.c (print_usage): Likewise.
18387 * gcov-dump.c (print_usage): Likewise.
18388 * gcov-tool.c (print_merge_usage_message): Likewise.
18389 (print_rewrite_usage_message): Likewise.
18390 (print_overlap_usage_message): Likewise.
18392 2017-04-28 Martin Liska <mliska@suse.cz>
18394 PR gcov-profile/53915
18395 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18397 2017-04-28 Martin Liska <mliska@suse.cz>
18399 PR gcov-profile/79891
18400 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18401 is marked by compiler as living on a line.
18402 (get_cycles_count): Remove usage of the union.
18403 (output_intermediate_file): Likewise.
18404 (find_source): Fix GNU coding style.
18405 (accumulate_line_counts): Remove old non-all block mode.
18406 (output_lines): Remove usage of the union.
18407 * profile.c (output_location): Include all BBs, even if
18408 belonging to a same line (and file) as a previous BB.
18410 2017-04-28 Martin Liska <mliska@suse.cz>
18412 * gcov.c (process_args): Handle new argument 'w'.
18413 (read_graph_file): Assign ID to BBs.
18414 (output_branch_count): Display BB # if verbose flag is set.
18415 (output_lines): Likewise for arcs.
18416 (print_usage): Add '--verbose' option help.
18417 * doc/gcov.texi: Document --verbose (-w) option.
18419 2017-04-28 Martin Liska <mliska@suse.cz>
18421 * gcov.c (struct block_location_info): New struct.
18422 (process_file): Fill up the new structure.
18423 (read_graph_file): Replace usage of encoding by the newly added
18425 (add_line_counts): Likewise.
18426 (accumulate_line_counts): Remove usage of the union.
18427 (function_info::function_info): New function.
18428 (function_info::~function_info): Likewise.
18429 (process_file): Call delete instead of release_function.
18430 (release_function): Release the function.
18431 (release_structures): Call delete instead of release_function.
18432 (solve_flow_graph): Replace usage of num_blocks.
18433 (find_exception_blocks): Likewise.
18434 (output_lines): Fix GNU coding style.
18436 2017-04-28 Martin Liska <mliska@suse.cz>
18439 * coverage.c (coverage_remove_note_file): New function.
18440 * coverage.h: Declare the function.
18441 * toplev.c (finalize): Clean if an error has been seen.
18443 2017-04-28 Martin Liska <mliska@suse.cz>
18445 PR gcov-profile/80031
18446 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18447 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18448 * gcov.c (read_graph_file): Read just number of blocks.
18449 * profile.c (branch_prob): Do not stream 0 flags per a basic
18452 2017-04-28 Martin Liska <mliska@suse.cz>
18454 * gcov-dump.c (tag_*): Add new argument to declarations.
18455 (dump_gcov_file): Likewise.
18456 (tag_blocks): Add and use new argument depth.
18457 (tag_arcs): Likewise.
18458 (tag_lines): Likewise.
18459 (tag_counters): Likewise.
18460 (tag_summary): Likewise.
18461 (dump_working_sets): Use depth to do a proper indentation.
18463 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18466 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18467 * symtab.c (symtab_node::debug_symtab): Move definition here.
18469 2017-04-28 Richard Biener <rguenther@suse.de>
18471 * lto-streamer.h (LTO_major_version): Bump to 7.
18473 2017-04-28 Richard Biener <rguenther@suse.de>
18475 * tree-vrp.c (assert_info): New struct.
18476 (add_assert_info): New helper.
18477 (register_edge_assert_for_2): Refactor to add asserts to a vector
18479 (register_edge_assert_for_1): Likewise.
18480 (register_edge_assert_for): Likewise.
18481 (finish_register_edge_assert_for): New helper actually registering
18482 asserts where live on edge.
18483 (find_conditional_asserts): Adjust.
18484 (find_switch_asserts): Likewise.
18485 (evrp_dom_walker::try_find_new_range): Generalize.
18486 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18488 2017-04-27 Marek Polacek <polacek@redhat.com>
18491 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18492 arg10 and arg11 to itype.
18494 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18496 * doc/extend.texi (Object Size Checking): Improve grammar.
18498 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18501 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18502 that the logic for permitting reciprocal estimates matches that
18505 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18508 * tree.c (type_cache_hasher::equal): Only compare
18509 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18510 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18511 non-aggregate element types.
18512 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18513 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18515 2017-04-27 Richard Biener <rguenther@suse.de>
18517 PR middle-end/80533
18518 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18519 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18520 keeping a reference to a trailing array.
18522 2017-04-27 Richard Biener <rguenther@suse.de>
18524 PR middle-end/80539
18525 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18526 being in loop-closed SSA form conservatively.
18527 (chrec_fold_multiply_poly_poly): Likewise.
18529 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18531 PR middle-end/79665
18532 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18533 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18535 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18538 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18539 (aarch64_function_arg_alignment): Return unsigned int again, but still
18540 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18541 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18542 Don't emit -Wpsabi note.
18543 (aarch64_function_arg_boundary): Likewise.
18544 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18547 2017-04-26 Nathan Sidwell <nathan@acm.org>
18549 * tree.h (crc32_unsigned_n): Declare.
18550 (crc32_unsigned, crc32_unsigned): Make inline.
18551 * tree.c (crc32_unsigned_bits): Replace with ...
18552 (crc32_unsigned_n): ... this.
18553 (crc32_unsigned, crc32_byte): Remove.
18554 (crc32_string): Remove unnecessary braces.
18556 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18558 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18559 * ipa-inline-analysis.c (MAX_TIME): Remove.
18560 (account_size_time): Use sreal for time.
18561 (dump_inline_summary): Update.
18562 (estimate_function_body_sizes): Update.
18563 (estimate_edge_size_and_time): Update.
18564 (estimate_calls_size_and_time): Update.
18565 (estimate_node_size_and_time): Update.
18566 (inline_merge_summary): Update.
18567 (inline_update_overall_summary): Update.
18568 (estimate_time_after_inlining): Update.
18569 (inline_read_section): Update.
18570 (inline_write_summary): Update.
18571 * ipa-inline.c (compute_uninlined_call_time): Update.
18572 (compute_inlined_call_time): Update.
18573 (recursive_inlining): Update.
18574 (inline_small_functions): Update.
18575 (dump_overall_stats): Update.
18576 * ipa-inline.h: Include sreal.h.
18577 (size_time_entry): Turn time to sreal.
18578 (inline_summary): Turn self_time nad time to sreal.
18580 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18582 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18584 (sreal::stream_out, sreal::stream_in): New.
18585 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18587 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18589 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18592 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18595 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18597 (dimode_scalar_chain::compute_convert_gain): Ditto.
18598 (dimode_scalar_chain::make_vector_copies): Ditto.
18599 (dimode_scalar_chain::convert_reg): Ditto.
18600 (dimode_scalar_chain::convert_insn): Ditto.
18601 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18602 (VI248_AVX512BW_1): New mode iterator.
18603 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18604 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18607 2017-04-25 Martin Sebor <msebor@redhat.com>
18609 PR tree-optimization/80497
18610 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18611 constants are representable in HOST_WIDE_INT.
18612 (parse_directive): Ditto.
18614 2017-04-25 Martin Sebor <msebor@redhat.com>
18617 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18618 (new_zero_array): Adjust signature.
18619 (dom_info::dom_init): Used unsigned rather that size_t.
18620 (dom_info::dom_info): Same.
18622 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18623 Jakub Jelinek <jakub@redhat.com>
18626 * config/arm/arm.c: Include gimple.h.
18627 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18628 returns negative, increment ncrn only if it returned positive.
18629 (arm_needs_doubleword_align): Return int instead of bool,
18630 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18631 members, but if there is any such non-FIELD_DECL
18632 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18633 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18634 returns negative, increment nregs only if it returned positive.
18635 (arm_setup_incoming_varargs): Likewise.
18636 (arm_function_arg_boundary): Emit -Wpsabi note if
18637 arm_needs_doubleword_align returns negative, return
18638 DOUBLEWORD_ALIGNMENT only if it returned positive.
18640 2017-04-25 Marek Polacek <polacek@redhat.com>
18643 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18644 first argument to type.
18646 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
18649 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
18650 type checks to test for compatibility instead of equality.
18652 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18653 Jakub Jelinek <jakub@redhat.com>
18656 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18658 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18659 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18660 the alignment computation, but return their maximum in warn_alignment.
18661 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18662 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18664 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18665 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18668 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18670 * config/arc/simdext.md (dmpyh): Fix typo.
18672 2017-04-25 Richard Biener <rguenther@suse.de>
18674 PR tree-optimization/80492
18675 * alias.c (compare_base_decls): Handle registers with asm
18676 specification conservatively.
18677 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18678 compare_base_decls returning dont-know properly.
18680 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18682 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18683 (legitimate_offset_address_p): New function.
18684 (arc_legitimate_address_p): Use above function.
18686 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18688 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18690 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18692 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18693 ACCH registers whenever they are available.
18695 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18697 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18698 double regs fix when not used.
18700 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18702 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18704 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18705 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18707 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18709 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18710 when emitting short ADD instructions.
18712 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18714 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18716 (cmpsi_cc_c_insn): Likewise.
18717 (cbranchsi4_scratch): Compute proper instruction length using
18718 compact_hreg_operand.
18719 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18721 2017-04-25 Richard Biener <rguenther@suse.de>
18723 PR middle-end/80509
18724 * passes.c (pass_manager::pass_manager): Initialize
18725 m_name_to_pass_map.
18727 2017-04-25 Richard Biener <rguenther@suse.de>
18729 PR tree-optimization/79201
18730 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18732 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18735 * config/s390/vector.md: Split MEM->GPR vector moves for
18736 non-s_operand addresses.
18738 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18741 * config/s390/predicates.md (reload_const_wide_int_operand): New
18743 * config/s390/s390.md ("movti"): Remove d/P alternative.
18744 ("movti_bigconst"): New pattern definition.
18746 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18749 * s390-protos.h (s390_expand_cs_hqi): Removed.
18750 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
18751 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
18752 modes as well as CCZ1mode and CCZmode.
18753 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
18754 signature of s390_emit_compare_and_swap.
18755 (s390_expand_cs_hqi): Likewise, make static.
18756 (s390_expand_cs_tdsi): Generate an explicit compare before trying
18757 compare-and-swap, in some cases.
18758 (s390_expand_cs): Wrapper function.
18759 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
18761 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
18762 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
18763 patterns for small and large integers. Forbid symref memory operands.
18764 Move expander to s390.c. Require cc register.
18765 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
18766 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
18767 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
18768 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
18769 symref memory operands. Remove CC mode and call s390_match_ccmode
18771 ("atomic_exchange<mode>"): Allow and implement all integer modes.
18773 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18775 * config/s390/s390.md (define_peephole2): New peephole to help
18776 combining the load-and-test pattern with volatile memory.
18778 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18780 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
18781 with CCZmode for TARGET_Z196.
18783 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18785 PR rtl-optimization/80501
18786 * combine.c (make_compound_operation_int): Set subreg_code to SET
18787 even for AND with mask of the sign bit of mode.
18789 PR rtl-optimization/80500
18790 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
18791 sum's initial value.
18793 2017-04-25 Julian Brown <julian@codesourcery.com>
18794 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18796 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
18798 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
18800 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
18802 2017-04-25 Julian Brown <julian@codesourcery.com>
18803 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18805 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
18806 (thunderx2t99_sha): New Reservation.
18808 2017-04-25 Julian Brown <julian@codesourcery.com>
18809 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18811 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
18812 type for 1-element load.
18814 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
18816 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
18818 2017-04-24 Martin Jambor <mjambor@suse.cz>
18820 PR tree-optimization/80293
18821 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
18822 char arrays not totally scalarizable if it is false.
18823 (analyze_all_variable_accesses): Pass correct value in the new
18824 parameter. Add a statistics counter.
18826 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
18828 PR middle-end/79931
18829 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
18831 2017-04-24 Richard Biener <rguenther@suse.de>
18833 PR tree-optimization/80494
18834 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
18835 out for complex types.
18837 2017-04-24 Richard Biener <rguenther@suse.de>
18839 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
18840 * tree-ssa-sccvn.c (print_scc): Print SCC size.
18841 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
18842 (DFS): Adjust and never fail.
18843 (sccvn_dom_walker::fail): Remove.
18844 (sccvn_dom_walker::before_dom_children): Adjust.
18845 (run_scc_vn): Likewise and never fail.
18846 * tree-ssa-pre.c (pass_pre::execute): Adjust.
18847 (pass_fre::execute): Likewise.
18849 2017-04-24 Richard Biener <rguenther@suse.de>
18851 PR tree-optimization/79725
18852 * tree-ssa-sink.c (statement_sink_location): Return whether
18853 failure reason was zero uses. Move that check later.
18854 (sink_code_in_bb): Deal with zero uses by removing the stmt
18857 2017-04-24 Richard Biener <rguenther@suse.de>
18860 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
18861 pointer-based references.
18863 2017-04-24 Richard Biener <rguenther@suse.de>
18866 * pass_manager.h (pass_manager::operator new): Remove.
18867 (pass_manager::operator delete): Likewise.
18868 * passes.c (pass_manager::operator new): Remove.
18869 (pass_manager::operator delete): Likewise.
18870 (pass_manager::pass_manager): Zero individual pass members.
18872 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
18875 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
18876 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
18877 Check "XEXP (src, 1)" operand here.
18878 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
18879 Check "XEXP (src, 1)" operand here.
18880 (dimode_scalar_chain::make_vector_copies): Detect count register
18881 of a shift instruction. Zero extend count register from QImode
18882 to DImode to satisfy vector shift pattern count operand predicate.
18883 Substitute vector shift count operand with a DImode copy.
18884 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
18887 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
18889 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
18890 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18891 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
18892 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18893 (UNSPEC_NOREX_MEM): Remove definition.
18895 2017-04-21 Richard Biener <rguenther@suse.de>
18897 PR tree-optimization/79547
18898 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18899 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
18900 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
18901 without any constraints.
18903 2017-04-21 Richard Biener <rguenther@suse.de>
18905 PR tree-optimization/78847
18906 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
18908 2017-04-21 Richard Biener <rguenther@suse.de>
18910 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
18911 (build_distinct_type_copy): Likewise.
18912 (build_variant_type_copy): Likewise.
18913 * tree.c (build_qualified_type): Pass down mem-stat info.
18914 (build_distinct_type_copy): Likewise.
18915 (build_variant_type_copy): Likewise.
18917 2017-04-21 Richard Biener <rguenther@suse.de>
18919 PR tree-optimization/80237
18920 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
18922 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
18923 for a simplified result.
18925 2016-04-21 Richard Biener <rguenther@suse.de>
18927 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
18928 sth as strict as a simple_iv but a chrec without symbols and an
18929 operand defined in the loop we are peeling (and not some subloop).
18930 (propagate_constants_for_unrolling): Propagate all constants.
18932 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18935 * config/i386/i386.c (print_reg): Remove assert for disalowed
18936 regno values, call output_operand_lossage instead.
18938 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18941 * config/i386/constraints.md (Yc): New register constraint.
18942 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
18943 Use Yc constraint for alternative 2 of operand 0. Remove
18944 preferred_for_speed attribute.
18946 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18948 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
18949 lastprivate clauses in SIMT case.
18951 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
18953 * doc/invoke.texi (-Wextra-semi): Document new warning option.
18955 2017-04-20 Richard Biener <rguenther@suse.de>
18957 PR tree-optimization/57796
18958 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
18959 as N scalar stores.
18960 (vect_model_load_cost): Cost gathers as N scalar loads.
18962 2017-04-20 Richard Biener <rguenther@suse.de>
18964 * ggc-page.c (ggc_allocated_p): Rename to ...
18965 (safe_lookup_page_table_entry): ... this and return the lookup
18967 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
18969 2017-04-20 Richard Biener <rguenther@suse.de>
18971 PR tree-optimization/80453
18972 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
18973 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
18974 from the conditions.
18975 (vn_phi_eq): Pass them down.
18976 (vn_phi_lookup): Record them.
18977 (vn_phi_insert): Likewise.
18979 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18981 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18982 uninitialized variable warning to avoid buffer overrun.
18984 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18987 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18988 is suppressed for '{ 0 }' in C.
18990 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18992 * BASE-VER: Set to 8.0.0.
18994 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
18996 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
18997 priority .init_array and .fini_array section with SECTION_NOTYPE
19000 2017-04-20 Jakub Jelinek <jakub@redhat.com>
19002 PR middle-end/80423
19003 * tree.h (build_array_type): Add typeless_storage default argument.
19004 * tree.c (type_cache_hasher::equal): Also compare
19005 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
19006 (build_array_type): Add typeless_storage argument, set
19007 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
19009 (build_nonshared_array_type): Adjust build_array_type_1 caller.
19010 (build_array_type): Likewise. Add typeless_storage argument.
19012 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19013 Jakub Jelinek <jakub@redhat.com>
19015 PR tree-optimization/80426
19016 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
19017 operation on symbolic operands, also compute the overflow for the
19018 invariant part when the operation degenerates into a negation.
19020 2017-04-19 Jakub Jelinek <jakub@redhat.com>
19023 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
19024 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
19027 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
19029 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
19032 * config/avr/avr.c (tree.h): Include it.
19033 (cgraph.h): Include it.
19034 (avr_encode_section_info): Don't warn for uninitialized progmem
19035 variable if it's just an alias.
19037 2017-04-19 Richard Biener <rguenther@suse.de>
19040 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
19041 when needed by AutoPGO.
19043 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
19046 * doc/lto.texi: Remove an extra 'that'.
19048 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
19050 PR rtl-optimization/80429
19051 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
19052 are only used in debug insns.
19054 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
19055 Vladimir Makarov <vmakarov@redhat.com>
19057 * config/sparc/predicates.md (input_operand): Add comment. Return
19058 true for any memory operand when LRA is in progress.
19059 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
19061 2017-04-18 Jeff Law <law@redhat.com>
19064 * mips.md ({return,simple_return}_internal): Do not overwrite
19067 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19069 PR tree-optimization/80443
19070 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
19071 instead of adding 1, subtract -1 and similarly instead of subtracting
19074 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
19076 PR rtl-optimization/80357
19077 * haifa-sched.c (tmp_bitmap): New variable.
19078 (model_recompute): Handle duplicate use records.
19079 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
19080 (free_global_sched_pressure_data): Free it.
19082 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
19085 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19086 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19087 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19088 instead of SYSTEM_HEADER_DIR.
19090 2017-04-18 Jeff Law <law@redhat.com>
19092 PR middle-end/80422
19093 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
19094 predecessors after walking up the insn chain.
19096 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19099 * dwarf2out.c (modified_type_die): Try harder not to emit internal
19100 sizetype type into debug info.
19102 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19105 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
19106 unneeded test for TARGET_UPPER_REGS_SF.
19107 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
19109 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19112 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
19113 instead of gsi_after_labels.
19115 2017-04-18 Jeff Law <law@redhat.com>
19117 * regcprop.c (maybe_mode_change): Avoid creating copies of the
19121 2017-04-13 Jeff Law <law@redhat.com>
19122 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19123 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19125 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
19128 * config/avr/avr.c (intl.h): Include it.
19129 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
19131 2017-04-18 Martin Liska <mliska@suse.cz>
19133 PR gcov-profile/78783
19134 * gcov-tool.c (gcov_output_files): Validate that destination
19135 file is either removed by the tool or by a user.
19137 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19138 Guy Benyei <guybe@mellanox.com>
19140 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
19141 block, and do not negate it, the stored id is already negative.
19143 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19145 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
19147 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19150 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
19151 masks of options that should be turned off if the VSX vector
19152 options are turned off.
19153 (OTHER_P8_VECTOR_MASKS): Likewise.
19154 (OTHER_VSX_VECTOR_MASKS): Likewise.
19155 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
19156 rs6000_disable_incompatible_switches to validate no type switches
19158 (rs6000_incompatible_switch): New function to disallow turning on
19159 other vector options if -mno-vsx, -mno-power8-vector, or
19160 -mno-power9-vector are specified.
19162 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19164 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
19166 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19168 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
19169 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
19170 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
19171 (ARG_POINTER_CFA_OFFSET): Likewise.
19173 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19175 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
19176 conditions to take advantage of various optimizations.
19178 2017-04-13 Jeff Law <law@redhat.com>
19180 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19181 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19182 (zero_extendsidi2_dext): Likewise.
19184 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19187 * fold-const.c (fold_ternary_loc): Revert
19188 use op0 instead of fold_convert_loc (loc, type, arg0) part of
19191 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
19193 PR rtl-optimization/80343
19194 * lra-remat.c (update_scratch_ops): Assign original hard reg to
19195 new scratch pseudo.
19197 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
19200 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
19201 to ubsan_encode_value.
19203 2017-04-13 Jeff Law <law@redhat.com>
19205 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
19206 appearing in DEBUG_INSNs.
19208 2017-04-13 Martin Liska <mliska@suse.cz>
19210 PR gcov-profile/80413
19211 * gcov-io.c (gcov_write_string): Copy to buffer just when
19212 allocated size is greater than zero.
19214 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19217 * dwarf2out.c (decls_for_scope): Ignore declarations of
19218 current_function_decl in BLOCK_NONLOCALIZED_VARS.
19220 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
19223 * ipa-visibility.c (non_local_p): Fix typos.
19224 (localize_node): When localizing symbol in same comdat group,
19225 dissolve the group only when we know external symbols are going
19227 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
19229 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19231 PR tree-optimization/79390
19232 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
19233 order does not result in usable sequence, retry with reversed operand
19239 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
19240 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
19241 op0 instead of fold_convert_loc (loc, type, arg0).
19243 2017-04-12 Jeff Law <law@redhat.com>
19245 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
19246 has a delay slot in the generated code.
19248 * config/cris/cris.md (cris_preferred_reload_class): Return
19249 GENNONACR_REGS rather than GENERAL_REGS.
19251 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19254 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
19255 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
19256 signedness of the result type.
19258 2017-04-12 Richard Biener <rguenther@suse.de>
19259 Jeff Law <law@redhat.com>
19261 PR tree-optimization/80359
19262 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
19263 trim stores to TARGET_MEM_REFs.
19265 2017-04-12 Richard Biener <rguenther@suse.de>
19267 PR tree-optimization/79390
19268 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
19269 threading case even more.
19271 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19274 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
19275 for quad_address_p for TImode, instead of just not indexed_address.
19277 2017-04-12 Richard Biener <rguenther@suse.de>
19278 Bernd Edlinger <bernd.edlinger@hotmail.de>
19280 PR middle-end/79671
19281 * alias.c (component_uses_parent_alias_set_from): Handle
19282 TYPE_TYPELESS_STORAGE.
19283 (get_alias_set): Likewise.
19284 * tree-core.h (tree_type_common): Add typeless_storage flag.
19285 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
19286 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
19287 for types containing members with TYPE_TYPELESS_STORAGE.
19288 (place_field): Likewise.
19289 (layout_type): Likewise for ARRAY_TYPE.
19290 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
19291 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
19292 TYPE_TYPELESS_STORAGE.
19293 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
19295 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19298 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19299 first argument to type.
19301 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19305 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
19306 CONST0_RTX (mode) rather than const0_rtx where appropriate.
19307 (rs6000_expand_binop_builtin): Likewise.
19308 (rs6000_expand_ternop_builtin): Likewise; also add missing
19309 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
19310 vshasigma built-ins.
19311 * doc/extend.texi: Document that vec_xxpermdi's third argument
19312 must be a constant.
19314 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
19316 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
19317 Use shift_const cost parameter when calculating gain of STV shifts.
19319 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
19321 PR rtl-optimization/70478
19322 * lra-constraints.c (process_alt_operands): Check memory for
19323 disfavoring memory insn operand.
19325 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19327 PR middle-end/80100
19328 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
19329 left shift in unsigned HOST_WIDE_INT type.
19331 PR rtl-optimization/80385
19332 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
19333 (not (neg X)) into (plus X -1) for complex or non-integral modes.
19336 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
19337 if they have any depend clauses.
19339 2017-04-11 Martin Liska <mliska@suse.cz>
19342 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
19343 * ipa-split.c (split_function): Create a local comdat symbol
19344 if caller is in a comdat group.
19346 2017-04-11 Martin Liska <mliska@suse.cz>
19349 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
19352 2017-04-11 Martin Sebor <msebor@redhat.com>
19354 PR middle-end/80364
19355 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
19356 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
19358 (directive::set_width, directive::set_precision, format_character):
19360 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
19363 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
19366 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
19367 conflict, set target->arch_name instead of target->cpu_name.
19369 2017-04-11 Richard Biener <rguenther@suse.de>
19371 PR tree-optimization/80374
19372 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
19373 build_zero_cst, remove fold_convertible_p check again.
19375 2017-04-11 Martin Liska <mliska@suse.cz>
19378 * ubsan.c (instrument_object_size): Do not instrument register
19381 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19384 * config/i386/i386-builtin-types.def
19385 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
19386 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
19387 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
19388 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
19389 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
19390 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
19391 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
19392 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
19393 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
19394 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19395 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19396 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19397 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19398 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19399 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19400 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19401 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19402 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19403 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19404 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19405 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19406 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19407 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19408 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19409 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19410 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19411 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19412 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19413 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19414 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19415 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19416 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19417 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19418 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19419 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19420 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19421 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19422 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19423 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19424 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19425 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19426 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19427 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19428 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19429 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19430 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19432 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19433 flag to second_arg_count, handle 4 argument function type _COUNT
19434 aliases, handle second_arg_count on second argument rather than last.
19436 2017-04-10 Jeff Law <law@redhat.com>
19438 PR tree-optimization/80374
19439 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19440 record anything if we can not convert integer_zero_node to the
19443 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19446 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19447 Enhance special handling given to the TARGET_P9_MINMAX option in
19448 relation to certain other options.
19450 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19452 PR tree-optimization/80153
19453 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19454 remove POINTER_PLUS_EXPR's base part directly, rather than through
19457 2017-04-10 Richard Biener <rguenther@suse.de>
19458 Bin Cheng <bin.cheng@arm.com>
19460 PR tree-optimization/80153
19461 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19462 the first element of pointer type aff_tree. Build result expr in
19464 (add_elt_to_tree): Convert to type unconditionally. Remove other
19465 fold_convert calls.
19466 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19467 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19469 2017-04-10 Richard Biener <rguenther@suse.de>
19471 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19474 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19476 PR rtl-optimization/70478
19477 * lra-constraints.c (curr_small_class_check): New.
19478 (update_and_check_small_class_inputs): New.
19479 (process_alt_operands): Update curr_small_class_check. Disfavor
19480 alternative insn memory operands. Check available regs for small
19483 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19486 * config/mips/mips.opt (-mvirt): Update description.
19487 * doc/invoke.texi (-mvirt): Likewise.
19489 2017-04-10 Richard Biener <rguenther@suse.de>
19491 PR middle-end/80362
19492 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19493 looking for NEGATE_EXPR in -A / -B to A / B folding.
19495 2017-04-10 Martin Liska <mliska@suse.cz>
19497 PR gcov-profile/80224
19498 * gcov.c (print_usage): Fix usage string.
19499 (get_gcov_intermediate_filename): Remove.
19500 (output_gcov_file): Use both for normal and intermediate format.
19501 (generate_results): Do not initialize special file for
19502 intermediate format.
19504 2017-04-10 Richard Biener <rguenther@suse.de>
19506 PR tree-optimization/80304
19507 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19510 2017-04-10 Nathan Sidwell <nathan@acm.org>
19513 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19514 (rs6000_init_builtins): Use it.
19516 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19518 * config/arm/arm.md (<mrc>): Add mode to SET source.
19519 (<mrrc>): Likewise.
19521 2017-04-10 Richard Biener <rguenther@suse.de>
19523 PR middle-end/80344
19524 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19526 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19529 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19530 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19531 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19532 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19533 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19534 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19535 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19536 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19537 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19538 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19539 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19540 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19541 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19542 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19543 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19544 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19545 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19546 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19547 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19548 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19549 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19550 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19551 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19553 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19555 PR rtl-optimization/70478
19556 * lra-constraints.c: Reverse the last patch.
19558 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19560 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19561 Add comment for WCHAR_T.
19563 2017-04-08 Martin Liska <mliska@suse.cz>
19566 2017-04-07 Martin Liska <mliska@suse.cz>
19569 * ipa-split.c (split_function): Add function part to a same comdat
19572 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19575 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19577 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19579 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19580 (vec_pairing): ...to this.
19581 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19582 (rs6000_sched_init): Adjust for name change.
19583 (struct rs6000_sched_context): Likewise.
19584 (rs6000_init_sched_context): Likewise.
19585 (rs6000_set_sched_context): Likewise.
19587 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19593 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19595 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19596 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19597 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19599 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19601 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19603 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19605 PR rtl-optimization/70703
19606 * ira-color.c (update_conflict_hard_regno_costs): Use
19607 int64_t instead of HOST_WIDE_INT.
19609 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19611 PR rtl-optimization/70478
19612 * lra-constraints.c (process_alt_operands): Disfavor alternative
19613 insn memory operands.
19615 2017-04-07 Jeff Law <law@redhat.com>
19617 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19618 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19620 2017-04-07 Martin Liska <mliska@suse.cz>
19623 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19624 Show error message instead of an ICE.
19626 2017-04-07 Martin Liska <mliska@suse.cz>
19629 * ipa-split.c (split_function): Add function part to a same comdat
19632 2017-04-07 Richard Biener <rguenther@suse.de>
19634 PR middle-end/80341
19635 * tree.c (get_unwidened): Also handle ! for_type case for
19637 * convert.c (do_narrow): Split out from ...
19638 (convert_to_integer_1): ... here. Do not pass final truncation
19639 type to get_unwidened for TRUNC_DIV_EXPR.
19641 2017-04-07 Richard Biener <rguenther@suse.de>
19643 * tree-affine.c (wide_int_ext_for_comb): Take type rather
19645 (aff_combination_const): Adjust.
19646 (aff_combination_scale): Likewise.
19647 (aff_combination_add_elt): Likewise.
19648 (aff_combination_add_cst): Likewise.
19649 (aff_combination_convert): Likewise.
19650 (add_elt_to_tree): Likewise. Remove unused argument.
19651 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19653 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19655 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19657 * config/arm/arm.c (arm_default_short_enums): Use
19658 ARM_DEFAULT_SHORT_ENUMS.
19659 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19661 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19664 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19665 members with redundant out-of-class redeclaration.
19667 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19670 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19671 * config/i386/i386.md (*zero_extendsidi2):
19672 Add (?*x,*x) and (?*v,*v) alternatives.
19674 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19677 * config/i386/i386.c (ix86_expand_builtin)
19678 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19679 mode from insn data. Convert operands to insn operand mode.
19680 Copy operands that don't satisfy insn predicate to a register.
19682 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19684 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19687 2017-04-06 Richard Biener <rguenther@suse.de>
19689 PR tree-optimization/80334
19690 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19691 preserve alignment of accesses.
19693 2017-04-06 Richard Biener <rguenther@suse.de>
19695 PR tree-optimization/80262
19696 * tree-sra.c (build_ref_for_offset): Preserve address-space
19698 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19699 Drop useless address-space information on MEM_REF offsets.
19701 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19703 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19705 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19707 PR rtl-optimization/70703
19708 * ira-color.c (update_conflict_hard_regno_costs): Use
19709 HOST_WIDE_INT instead of long.
19711 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19714 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19715 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19717 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19718 for x86_64 target. Handle -m3dnowa option.
19720 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19722 PR rtl-optimization/70703
19723 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19724 (update_conflict_hard_regno_costs): Use long instead of unsigned
19725 arithmetic for cost calculation.
19727 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19728 Bernd Edlinger <bernd.edlinger@hotmail.de>
19731 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19734 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19737 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
19738 ptr_mode with Pmode throughout.
19739 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
19740 into probe_stack_range and use DImode.
19742 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
19745 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
19746 call_eh_return is true.
19748 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19750 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
19751 Initialize last_match_fntype_index.
19753 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19756 * tree-nvr.c: Include internal-fn.h.
19757 (pass_return_slot::execute): Ignore internal calls without
19760 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19761 Richard Biener <rguenther@suse.de>
19764 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
19765 captures used multiple times, except for the last use.
19766 * generic-match-head.c: Include gimplify.h.
19768 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19770 PR tree-optimization/79390
19771 * target.h (struct noce_if_info): Declare.
19772 * targhooks.h (default_noce_conversion_profitable_p): Declare.
19773 * target.def (noce_conversion_profitable_p): New target hook.
19774 * ifcvt.h (struct noce_if_info): New type, moved from ...
19775 * ifcvt.c (struct noce_if_info): ... here.
19776 (noce_conversion_profitable_p): Renamed to ...
19777 (default_noce_conversion_profitable_p): ... this. No longer
19779 (noce_try_store_flag_constants, noce_try_addcc,
19780 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
19781 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
19782 instead of noce_conversion_profitable_p.
19783 * config/i386/i386.c: Include ifcvt.h.
19784 (ix86_option_override_internal): Don't override
19785 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
19786 (ix86_noce_conversion_profitable_p): New function.
19787 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
19788 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
19789 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
19790 * doc/tm.texi: Regenerated.
19792 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19794 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
19797 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19800 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
19801 instructions for small multiply cores.
19803 2017-04-04 Jeff Law <law@redhat.com>
19805 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
19807 (mips_expand_vec_perm_const): Initialize elements in orig_perm
19808 that are not set by the loop over the elements.
19810 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19813 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
19814 int mode, convert_modes it to mode as unsigned, otherwise use
19815 lowpart_subreg to mode rather than SImode.
19816 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
19817 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
19818 Use DImode instead of SImode for the shift count operand.
19819 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
19822 2017-04-04 Richard Biener <rguenther@suse.de>
19824 PR middle-end/80281
19825 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
19826 arithmetic done for the negate or the plus. Simplify.
19827 (A - (-B) -> A + B): Likewise.
19828 * fold-const.c (split_tree): Make sure to not negate pointers.
19830 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
19832 PR rtl-optimization/60818
19833 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
19834 a compare of comparisons with the thing compared if this results
19835 in a different machine mode.
19837 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
19839 * alias.c (base_alias_check): Fix typo in comment.
19840 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
19841 * cgraphunit.c (symbol_table::compile): Likewise.
19842 * collect2.c (maybe_run_lto_and_relink): Likewise.
19843 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
19844 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
19845 * config/avr/avr.c (avr_map_op_t): Likewise.
19846 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
19847 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
19848 * config/epiphany/epiphany.md (movcc): Likewise.
19849 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
19850 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
19852 * config/mips/mips.c (mips_save_restore_reg): Likewise.
19853 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
19854 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
19855 * config/sh/sh.c (sh_rtx_costs): Likewise.
19856 * fold-const.c (fold_truth_andor): Likewise.
19857 * genautomata.c (collapse_flag): Likewise.
19858 * gengtype.h (struct type::u::s): Likewise.
19859 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
19860 * input.c (FORMAT_AMOUNT): Likewise.
19861 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
19862 (known_aggs_to_agg_replacement_list): Likewise.
19863 * ipa-inline-analysis.c: Likewise.
19864 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
19865 * ipa-polymorphic-call.c
19866 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
19867 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
19868 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
19870 * modulo-sched.c (apply_reg_moves): Likewise.
19871 * omp-expand.c (build_omp_regions_1): Likewise.
19872 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
19873 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
19874 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
19875 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19876 * value-prof.c: Likewise.
19877 * var-tracking.c (val_reset): Likewise.
19879 2017-04-03 Richard Biener <rguenther@suse.de>
19881 PR tree-optimization/80275
19882 * fold-const.c (split_address_to_core_and_offset): Handle
19885 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
19887 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
19888 descriptors is at least equal to that of functions.
19890 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19892 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
19894 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19897 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
19898 (mov<IMOD4:mode>): New expander.
19899 (*mov<IMOD4:mode>_internal): New insn and split pattern.
19901 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19903 PR rtl-optimization/79405
19904 * fwprop.c (propagations_left): New variable.
19905 (forward_propagate_into): Decrement it.
19906 (fwprop_init): Initialize it.
19907 (fw_prop): If the variable has reached zero, stop propagating.
19908 (fwprop_addr): Ditto.
19910 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19913 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
19914 a FUNCTION_DECL, pass it as decl instead of origin to
19917 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
19919 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
19922 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
19925 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
19926 TARGET_VSX_SMALL_INTEGER.
19928 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19930 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19931 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
19933 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19935 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
19936 extraction from odd-numbered MSA register.
19938 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19940 PR middle-end/80173
19941 * expmed.c (store_bit_field_1): Don't attempt to create
19942 a word subreg out of hard registers wider than word if they
19943 have HARD_REGNO_NREGS of 1 for their mode.
19945 PR middle-end/80163
19946 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
19947 conversions to integer types wider than word and pointer.
19950 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
19951 (rtx_equal_for_cselib_p): Pass 0 to it.
19952 * cselib.c (cselib_hasher::equal): Likewise.
19953 (rtx_equal_for_cselib_1): Add depth argument. If depth
19954 is 128, don't look up VALUE locs and punt. Increment
19955 depth in recursive calls when walking VALUE locs.
19957 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
19959 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
19960 (make_gcov_file_name): Use the canonical path name for generating
19962 (read_line): Fix handling of files with ascii null bytes.
19964 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
19966 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
19967 to initialise a vector register instead
19968 of using a const_int.
19970 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19972 PR translation/80189
19973 * gimplify.c (omp_default_clause): Use %qs instead of %s in
19974 diagnostic messages.
19976 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19979 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19980 (dfp_diex_<mode>): Update mode of operand 1.
19981 * doc/extend.texi (dxex, dxexq): Document change to return type.
19982 (diex, diexq): Document change to argument type.
19984 2017-03-30 Martin Jambor <mjambor@suse.cz>
19987 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19988 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19989 it reflects the signature changes performed at the callee side.
19990 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19991 to cgraph_build_function_type_skip_args.
19992 (build_function_decl_skip_args): Adjust call to the above function.
19994 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19997 * config/i386/sse.md
19998 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
19999 register as dest whenever it is a MEM not rtx_equal_p to the
20000 corresponding dup operand, and when forcing into reg move the
20001 reg into the memory afterwards.
20002 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
20003 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
20004 for the force_reg mode.
20005 (avx512vl_vextractf128<mode>): Use register as dest either
20006 always when a MEM, or when it is a MEM not rtx_equal_p to the
20007 corresponding dup operand, or even not when it is a CONST_VECTOR
20008 depending on the mode and lo vs. hi.
20009 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
20011 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
20012 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
20013 Likewise. Require that operands[2] is even.
20014 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
20015 Remove extraneous parens. Require that operands[2] is a multiple
20017 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
20018 operands[0] is a MEM if <mask_applied>, the predicates/constraints
20019 disallow memory then.
20021 2017-03-30 Richard Biener <rguenther@suse.de>
20023 PR tree-optimization/77498
20024 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
20025 to non-constants over backedges.
20027 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
20029 PR rtl-optimization/80233
20030 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
20031 as last_combined_insn. Do not test for BARRIER_P separately.
20033 2017-03-29 Andreas Schwab <schwab@suse.de>
20036 * calls.c (prepare_call_address): Convert funexp to Pmode before
20037 copying to temp reg.
20039 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20041 PR tree-optimization/80158
20042 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
20043 Handle possible future case of more than one alternate
20045 (replace_rhs_if_not_dup): Likewise.
20046 (replace_one_candidate): Likewise.
20048 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
20050 PR rtl-optimization/80193
20051 * ira.c (ira): Do not check allocation for LRA.
20053 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
20055 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
20056 (nvptx_output_simt_exit): Declare.
20057 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
20058 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
20059 (init_softstack_frame): Move initialization of crtl->is_leaf to...
20060 (nvptx_declare_function_name): ...here. Emit declaration of local
20061 memory space buffer for omp_simt_enter insn.
20062 (nvptx_output_unisimt_switch): New.
20063 (nvptx_output_softstack_switch): New.
20064 (nvptx_output_simt_enter): New.
20065 (nvptx_output_simt_exit): New.
20066 * config/nvptx/nvptx.h (struct machine_function): New fields
20067 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
20068 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
20069 (UNSPECV_SIMT_EXIT): Ditto.
20070 (omp_simt_enter_insn): New insn.
20071 (omp_simt_enter): New expansion.
20072 (omp_simt_exit): New insn.
20073 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
20075 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
20076 (expand_GOMP_SIMT_ENTER_ALLOC): New.
20077 (expand_GOMP_SIMT_EXIT): New.
20078 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
20079 (GOMP_SIMT_ENTER_ALLOC): Ditto.
20080 (GOMP_SIMT_EXIT): Ditto.
20081 * target-insns.def (omp_simt_enter): New insn.
20082 (omp_simt_exit): Ditto.
20083 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
20085 (lower_rec_simd_input_clauses): Implement SIMT privatization.
20086 (lower_rec_input_clauses): Likewise.
20087 (lower_lastprivate_clauses): Handle SIMT privatization.
20089 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
20090 (ompdevlow_adjust_simt_enter): New.
20091 (find_simtpriv_var_op): New.
20092 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
20093 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
20095 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
20096 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
20097 (copy_decl_for_dup_finish): Ditto.
20099 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
20101 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
20104 * config/i386/i386.c (ix86_option_override_internal): Always
20105 allow -mpreferred-stack-boundary=3 for 64-bit targets.
20107 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20109 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
20111 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20113 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
20114 mark new edge's irreducible flag accordign to it.
20115 (vect_do_peeling): Check loop preheader edge's irreducible flag
20116 and pass it to function slpeel_add_loop_guard.
20118 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
20120 PR tree-optimization/80218
20121 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
20122 Update block frequencies and counts.
20124 2017-03-28 Richard Biener <rguenther@suse.de>
20126 PR tree-optimization/78644
20127 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
20128 of a simplification result we may not use it at all.
20130 2017-03-28 Richard Biener <rguenther@suse.de>
20133 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
20134 without arguments, generate default definition of a SSA name.
20136 2017-03-28 Richard Biener <rguenther@suse.de>
20138 PR middle-end/80222
20139 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
20140 TYPE_REF_CAN_ALIAS_ALL references.
20141 * fold-const.c (fold_indirect_ref_1): Likewise.
20143 2017-03-28 Martin Liska <mliska@suse.cz>
20146 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
20147 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
20149 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20150 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
20152 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
20153 (EXTRA_SPECS): Define.
20154 (SUBTARGET_EXTRA_SPECS): Likewise.
20155 (SUBTARGET_CPP_SPEC): Likewise.
20156 * config/arc/elf.h (EXTRA_SPECS): Renamed to
20157 SUBTARGET_EXTRA_SPECS.
20158 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
20160 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20162 * config/arc/simdext.md (vst64_insn): Update pattern.
20163 (vld32wh_insn): Likewise.
20164 (vld32wl_insn): Likewise.
20165 (vld64_insn): Likewise.
20166 (vld32_insn): Likewise.
20168 2017-03-28 Marek Polacek <polacek@redhat.com>
20171 * fold-const.c (fold_comparison): Use protected_set_expr_location
20172 instead of SET_EXPR_LOCATION.
20174 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
20176 * tree.c (add_expr): Avoid name lookup warning.
20178 2017-03-27 Jeff Law <law@redhat.com>
20180 PR tree-optimization/80216
20181 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
20182 function name. Limit recursion depth.
20183 (record_temporary_equivalences): Corresponding changes.
20185 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
20187 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
20190 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20193 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
20195 * cfgcleanup.c (reg_note_cfa_p): New array.
20196 (insns_have_identical_cfa_notes): New function.
20197 (old_insns_match_p): Don't cross-jump in between /f
20198 and non-/f instructions. If both i1 and i2 are frame related,
20199 verify all CFA notes, their order and content.
20201 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20204 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
20205 HImode and SImode with zero extend to DImode to one insn.
20206 (bswap<mode>2_extenddi): Likewise.
20207 (bswapsi2_extenddi): Likewise.
20208 (bswaphi2_extendsi): Likewise.
20209 (bswaphi2): Combine bswap HImode and SImode into one insn.
20210 Separate memory insns from swapping register.
20211 (bswapsi2): Likewise.
20212 (bswap<mode>2): Likewise.
20213 (bswaphi2_internal): Delete, no longer used.
20214 (bswapsi2_internal): Likewise.
20215 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
20216 store, and gpr<-gpr swap insns.
20217 (bswap<mode>2_store): Likewise.
20218 (bswaphi2_reg): Register only splitter, combine with the splitter.
20219 (bswaphi2 splitter): Likewise.
20220 (bswapsi2_reg): Likewise.
20221 (bswapsi2 splitter): Likewise.
20222 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
20223 the insns into load, store, and register/register insns.
20224 (bswapdi2_ldbrx): Likewise.
20225 (bswapdi2_load): Likewise.
20226 (bswapdi2_store): Likewise.
20227 (bswapdi2_reg): Likewise.
20229 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
20231 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
20232 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
20234 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
20237 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
20239 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20240 special handling for target option conflicts between dform
20241 options (-mpower9-dform, -mpower9-dform-vector,
20242 -mpower9-dform-scalar) and -mno-direct-move.
20244 2017-03-27 Richard Biener <rguenther@suse.de>
20246 PR tree-optimization/80181
20247 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
20249 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20251 * config/arc/predicates.md (move_double_src_operand): Replace the
20252 call to move_double_src_operand with a call to address_operand.
20254 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20256 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
20257 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
20258 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
20260 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20262 * config/arc/predicates.md (long_immediate_loadstore_operand):
20263 Consider scaled addresses cases.
20265 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20267 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
20268 restored when in interrupt.
20269 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
20270 doesn't have delay slot.
20272 2017-03-27 Richard Biener <rguenther@suse.de>
20275 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
20276 inlined thunk clones.
20278 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20281 * asan.c (instrument_derefs): Copy over last operand from
20282 original COMPONENT_REF to the new COMPONENT_REF with
20283 DECL_BIT_FIELD_REPRESENTATIVE.
20284 * ubsan.c (instrument_object_size): Likewise.
20286 2017-03-27 Richard Biener <rguenther@suse.de>
20288 PR tree-optimization/80170
20289 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
20290 sure DR/SCEV didnt fold in constants we do not see when looking
20291 at the reference base alignment.
20293 2017-03-27 Richard Biener <rguenther@suse.de>
20295 PR middle-end/80171
20296 * gimple-fold.c (fold_ctor_reference): Properly guard against
20297 NULL return value from canonicalize_constructor_val.
20299 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
20302 * config/i386/i386.c (ix86_expand_builtin)
20303 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
20304 flags reg setting and flags reg using instructions.
20305 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
20306 clobbering instructions to zero extend op2.
20308 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
20310 * doc/install.texi (Configuration) <--with-aix-soname>:
20311 Update link to AIX ld.
20313 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
20315 PR rtl-optimization/80160
20316 PR rtl-optimization/80159
20317 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
20318 reg_alternate_class into account.
20320 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
20323 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
20324 to consider in curr_insn_transform.
20326 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20328 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
20329 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
20330 and emit_mode_inner.
20332 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20334 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
20335 argument to the overloaded builtin variants. Use the new flag to
20336 deprecate certain builtin variants.
20337 * config/s390/s390-builtin-types.def: Add new builtin types.
20338 * config/s390/s390-builtins.h: Support new flags field for
20339 overloaded builtins.
20340 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
20341 (s390_macro_to_expand): Enable vector float data type.
20342 (s390_cpu_cpp_builtins_internal): Indicate support of the new
20343 builtins by incrementing the __VEC__ version number.
20344 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
20346 (s390_resolve_overloaded_builtin): Emit error messages depending
20347 on the builtin flags.
20348 * config/s390/s390.c (s390_expand_builtin): Support additional
20349 flags argument. Change error message to match the messages
20350 emitted in s390-c.c.
20351 * config/s390/s390.md: New UNSPEC_* constants.
20352 (op_type): Add new instruction types.
20353 * config/s390/vecintrin.h: Add new builtins and test data class
20355 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
20356 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
20357 (VEC_INEXACT, VEC_NOINEXACT): New constants.
20358 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
20359 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
20360 ("vec_mergel<mode>"): V_HW -> VEC_HW.
20362 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
20363 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
20364 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
20365 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
20367 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
20368 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
20369 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
20370 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
20372 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
20373 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
20374 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
20375 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
20376 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
20377 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
20378 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
20380 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
20381 ("vec_scatter_element<V_HW_4:mode>_DI")
20382 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
20383 ("vec_fpint<mode>", "vflls")
20384 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
20385 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
20386 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
20387 ("*vec_cmphe<mode>_cc"): ... these.
20389 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
20390 mode constant instead of magic value.
20392 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20394 * config/s390/s390.c (s390_expand_vec_compare): Support other
20395 vector floating point modes than just V2DF.
20396 (s390_expand_vcond): Likewise.
20397 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20398 (s390_cannot_change_mode_class): Prevent mode changes between TF
20399 and V1TF in vector registers.
20400 * config/s390/s390.md (DF, SF): New mode attributes.
20401 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20402 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20403 SFmode support for VRs.
20404 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20406 (VFT, VF_HW): New mode iterators.
20407 (vw, sdx): New mode attributes.
20408 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20409 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20410 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20411 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20412 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20413 also the new vector floating point modes. Renaming to ...
20415 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20416 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20417 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20418 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20419 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20420 ("vec_unordered<mode>"): ... these.
20422 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20423 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20424 ("*vec_extendv2df"): New insn definitions.
20426 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20428 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20429 ("mulditi3_2", "*muldi3_sign"): New patterns.
20430 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20431 rename the pattern definition.
20433 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20435 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20437 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20439 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20441 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20442 instruction if possible.
20443 * config/s390/vector.md (vec_halfnumelts): New mode
20445 ("*vec_vllezlf<mode>"): New pattern.
20447 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20449 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20450 ("popcountv4si2", "popcountv2di2"): Rename to ...
20451 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20452 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20454 ("popcount<mode>2_vxe"): New pattern.
20456 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20458 * common/config/s390/s390-common.c (processor_flags_table): Add
20460 * config.gcc: Add arch12.
20461 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20462 Default to arch12 for unknown CPU model numbers.
20463 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20464 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20465 PROCESSOR_max sanity check.
20466 * config/s390/s390-opts.h (enum processor_type): Add
20468 * config/s390/s390.c (processor_table): Add arch12.
20469 (s390_expand_builtin): Add check for B_VXE flag.
20470 (s390_issue_rate): Add PROCESSOR_ARCH12.
20471 (s390_get_sched_attrmask): Likewise.
20472 (s390_get_unit_mask): Likewise.
20473 (s390_sched_score): Enable z13 scheduling for arch12.
20474 (s390_sched_reorder): Likewise.
20475 (s390_sched_variable_issue): Likewise.
20476 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20478 (s390_tune_attr): Use z13 scheduling also for arch12.
20479 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20480 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20481 (TARGET_VXE_P): New macros.
20482 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20483 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20484 * config/s390/s390.opt: Add arch12 as processor_type.
20486 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20488 * config/s390/s390.md
20489 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20490 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20491 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20493 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20494 Rename expanders to ...
20496 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20497 ("fixuns_truncdddi2_emu"): ... these.
20499 ("fixuns_trunc<mode>si2_emu"): New expander.
20501 ("*fixuns_truncdfdi2_z13"): Rename to ...
20502 ("*fixuns_truncdfdi2_vx"): ... this.
20504 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20506 * config/s390/2964.md: Remove the single element vector compare
20507 instructions which are no longer used.
20508 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20510 (s390_canonicalize_comparison): Remove handling of DFmode
20512 (s390_expand_vec_compare_scalar): Remove function.
20513 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20514 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20516 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20518 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20520 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20522 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20523 will anyway by matched by mov<mode>_64dfp.
20525 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20527 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20528 vlef/vstef. Add missing operand to vleif.
20530 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20532 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20533 pair for all vector types with 64 bit elements.
20534 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20535 * config/s390/vector.md (V_HW_64): ... here.
20536 (V_128_NOSINGLE): New mode iterator.
20537 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20538 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20539 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20540 ("*vec_load_pairv2di"): Change to ...
20541 ("*vec_load_pair<mode>"): ... this one.
20543 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20545 * config/s390/constraints.md: Add comments.
20546 (jKK): Reject element sizes > 8 bytes.
20547 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20549 * config/s390/s390.md: Add the s_operand checks formerly in
20550 s390_split_ok_p to various splitters where they are still
20552 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20553 for 128 bit vectors. Plus two splitters.
20555 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20557 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20560 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20563 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20564 error if the boundary argument is not constant.
20566 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20568 PR rtl-optimization/80112
20569 * loop-doloop.c (doloop_condition_get): Don't check condition
20570 if cmp isn't SET with IF_THEN_ELSE src.
20572 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20574 PR tree-optimization/80158
20575 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20576 replacing a candidate statement, also replace it for the
20577 candidate's alternate interpretation.
20578 (replace_rhs_if_not_dup): Likewise.
20579 (replace_one_candidate): Likewise.
20581 2017-03-24 Richard Biener <rguenther@suse.de>
20583 PR tree-optimization/80167
20584 * graphite-isl-ast-to-gimple.c
20585 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20587 (translate_isl_ast_to_gimple::get_rename): Likewise.
20589 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20591 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20592 handling of certain combinations of target options, including the
20593 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20594 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20596 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20599 * config/arm/arm.md (*load_multiple): Add reload_completed to
20600 matching condition.
20602 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20603 Richard Biener <rguenth@suse.de>
20605 PR tree-optimization/79908
20606 PR tree-optimization/80136
20607 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20608 been cast away, gimplify_and_add suffices.
20610 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20612 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20614 2017-03-23 Richard Biener <rguenther@suse.de>
20616 PR tree-optimization/80032
20617 * gimplify.c (gimple_push_cleanup): Forced unconditional
20618 cleanups still have to go to the conditional_cleanups
20621 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20623 PR tree-optimization/80072
20624 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20626 (next_operand_entry_id): Change type to unsigned int.
20627 (sort_by_operand_rank): Make sure to return the right return value
20628 even if unsigned fields are bigger than INT_MAX.
20629 (struct oecount): Change cnt and id type to unsigned int.
20630 (oecount_hasher::equal): Formatting fix.
20631 (oecount_cmp): Make sure to return the right return value
20632 even if unsigned fields are bigger than INT_MAX.
20633 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
20636 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
20637 TREE_READONLY on result if writing it more than once.
20640 * doc/invoke.texi (-fsanitize=thread): Document that with
20641 -fnon-call-exceptions atomics are not able to throw
20645 * tsan.c: Include tree-eh.h.
20646 (instrument_builtin_call): Call maybe_clean_eh_stmt or
20647 maybe_clean_or_replace_eh_stmt where needed.
20648 (instrument_memory_accesses): Add cfg_changed argument.
20649 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
20650 if it returned true.
20651 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20653 PR rtl-optimization/63191
20654 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20655 wrapper function, moved the whole old content into ...
20656 (ix86_delegitimize_address_1): ... this. New inline function.
20657 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20658 true as last argument instead of ix86_delegitimize_address.
20660 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20662 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20663 cortexa57_branch_cost.
20665 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20667 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20669 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20672 * doc/md.texi (Constraints): Document wA constraint.
20673 * config/rs6000/constraints.md (wA): New.
20674 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20675 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20676 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20677 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20679 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20682 * gimplify.c (is_oacc_declared): New function.
20683 (oacc_default_clause): Use it to set default flags for acc declared
20684 variables inside parallel regions.
20685 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20686 declared variables.
20687 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20688 declare attribute to any decl as necessary.
20690 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20693 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20694 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20695 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20696 (arm_arch_lpae): This.
20697 * config/arm/arm.c (arm_arch7ve): Rename into ...
20698 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20699 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20702 2017-03-22 Martin Liska <mliska@suse.cz>
20705 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20706 error message instead of an ICE.
20708 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20710 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20712 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20714 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20716 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20719 2017-03-21 Martin Sebor <msebor@redhat.com>
20721 * doc/extend.texi: Use "cannot" instead of "can't."
20722 * doc/hostconfig.texi: Same.
20723 * doc/install.texi: Same.
20724 * doc/invoke.texi: Same.
20725 * doc/loop.texi: Same.
20726 * doc/md.texi: Same.
20727 * doc/objc.texi: Same.
20728 * doc/rtl.texi: Same.
20729 * doc/tm.texi: Same.
20730 * doc/tm.texi.in: Same.
20731 * doc/trouble.texi: Same.
20733 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20736 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20737 (collect_checksum_attributes): Set it.
20738 (die_checksum_ordered): Use it.
20740 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20742 PR tree-optimization/79908
20743 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
20744 change: For a VA_ARG whose LHS has been cast away, use
20745 force_gimple_operand to construct the side effects.
20747 2017-03-21 David Malcolm <dmalcolm@redhat.com>
20749 PR translation/80001
20750 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
20751 more amenable to translation.
20752 (oacc_loop_auto_partitions): Likewise.
20754 2017-03-21 Marek Polacek <polacek@redhat.com>
20755 Martin Sebor <msebor@redhat.com>
20757 PR tree-optimization/80109
20758 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
20759 on INTEGRAL_TYPE_P.
20761 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20762 Segher Boessenkool <segher@kernel.crashing.org>
20765 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
20766 check reg_used_between_p between insn and one of succ or succ2
20767 depending on if succ is artificial insn not inserted into insn
20770 2017-03-21 Martin Liska <mliska@suse.cz>
20772 PR gcov-profile/80081
20773 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
20774 * doc/gcc.texi: Include gcov-dump stuff.
20775 * doc/gcov-dump.texi: New file.
20777 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
20779 PR rtl-optimization/79150
20780 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
20781 conditional jump, if the jump is the last insn of the loop.
20783 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20784 Richard Biener <rguenth@suse.de>
20786 PR tree-optimization/79908
20787 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20788 been cast away, use force_gimple_operand to construct the side
20791 2017-03-21 Martin Liska <mliska@suse.cz>
20793 PR libfortran/79956
20794 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
20797 2017-03-21 Brad Spengler <spender@grsecurity.net>
20800 * plugin.c (htab_hash_plugin): New function.
20801 (add_new_plugin): Use it and adjust.
20802 (parse_plugin_arg_opt): Adjust.
20803 (init_one_plugin): Likewise.
20805 2017-03-21 Richard Biener <rguenther@suse.de>
20807 PR tree-optimization/80032
20808 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
20809 if set force the cleanup to happen unconditionally.
20810 (gimplify_target_expr): Push inserted clobbers with force_uncond
20811 to avoid them being removed by control-dependent DCE.
20813 2017-03-21 Richard Biener <rguenther@suse.de>
20815 PR tree-optimization/80122
20816 * tree-inline.c (copy_bb): Do not expans va-arg packs or
20817 va_arg_pack_len when the inlined call stmt requires pack
20819 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
20821 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20824 * tsan.c (instrument_builtin_call): If the memory model argument
20825 is not a constant, assume it is valid.
20828 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
20831 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20833 PR rtl-optimization/79910
20834 * combine.c (can_combine_p): Do not allow combining an I0 or I1
20835 if its dest is used by an insn before I2 (other than the combined
20836 insns themselves, which are properly handled already).
20838 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20841 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20843 * combine.c (record_used_regs): New static function.
20844 (try_combine): Handle situations where there is an additional
20845 instruction between I2 and I3 which needs to have a LOG_LINK
20849 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20851 * combine.c (try_combine): Delete redundant i1 test. Call
20852 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20854 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20857 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
20858 alternatives 13/14.
20860 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20862 PR tree-optimization/80054
20863 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
20864 the optimization if a PHI or any of its arguments is not dominated
20865 by the candidate's basis. Use gphi* rather than gimple* as
20867 (replace_profitable_candidates): Clean up a gimple* variable that
20868 should be a gphi* variable.
20870 2017-03-20 Martin Sebor <msebor@redhat.com>
20873 * doc/extend.texi (attribute constructor): Document present limitation.
20875 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
20878 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
20879 __POWER9_VECTOR__ #ifdef control, change template definition to
20880 use Power9-specific built-in function.
20881 (vec_any_eq): Likewise.
20882 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
20883 to control outcomes from this test.
20884 (vector_ae_<mode>p): For VEC_F modes, likewise.
20886 2017-03-20 Ian Lance Taylor <iant@google.com>
20888 * config/i386/i386.c (ix86_function_regparm): Save an extra
20889 register for -fsplit-stack with DECL_STATIC_CHAIN.
20891 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20894 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
20895 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
20897 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20899 * config/riscv/riscv.c (riscv_print_operand): Use "fence
20901 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
20904 2017-03-20 Marek Polacek <polacek@redhat.com>
20907 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
20909 2017-03-20 Richard Biener <rguenther@suse.de>
20911 PR tree-optimization/80113
20912 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
20913 allocate extra SSA name for PHI def.
20914 (add_close_phis_to_outer_loops): Likewise.
20915 (add_close_phis_to_merge_points): Likewise.
20916 (copy_loop_close_phi_args): Likewise.
20917 (copy_cond_phi_nodes): Likewise.
20919 2017-03-20 Martin Liska <mliska@suse.cz>
20921 PR middle-end/79753
20922 * tree-chkp.c (chkp_build_returned_bound): Do not build
20923 returned bounds for a LHS that's not a BOUNDED_P type.
20925 2017-03-20 Martin Liska <mliska@suse.cz>
20929 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
20930 COMPLEX_CST and VECTOR_CST.
20932 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20935 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
20936 target operand. A new splitter adds the clobber statement in case
20937 the target operand is dead anyway.
20939 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
20941 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
20942 to age-old versions of binutils and glibc.
20944 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
20946 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
20948 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20950 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
20952 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20954 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
20955 requirement for binutils 2.13.
20957 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20959 * combine.c (try_combine): Delete redundant i1 test. Call
20960 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20962 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
20964 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
20965 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
20967 <riscv64-*-elf>: Re-arrange section
20968 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
20969 <riscv32-*-linux>: Likewise.
20970 <riscv64-*-elf>: Likewise
20971 <riscv64-*-linux>: Likewise.
20973 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
20976 * aarch64.opt(verbose-cost-dump): Fix typo.
20978 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20981 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20982 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20984 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20986 * reload.c (find_reloads): When reloading a nonoffsettable address,
20987 use RELOAD_OTHER for it and its address reloads.
20989 PR rtl-optimization/79910
20990 * combine.c (record_used_regs): New static function.
20991 (try_combine): Handle situations where there is an additional
20992 instruction between I2 and I3 which needs to have a LOG_LINK
20995 2017-03-17 Jeff Law <law@redhat.com>
20997 PR tree-optimization/71437
20998 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
20999 conditional in the hash table first.
21000 (vrp_dom_walker::before_dom_children): Extract condition from
21001 ASSERT_EXPR. Record condition, its inverion and any implied
21002 conditions as well.
21004 2017-03-17 Marek Polacek <polacek@redhat.com>
21005 Markus Trippelsdorf <markus@trippelsdorf.de>
21007 PR tree-optimization/80079
21008 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
21011 2017-03-17 Richard Biener <rguenther@suse.de>
21013 PR middle-end/80075
21014 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
21015 Properly verify the LHS before the RHS possibly claims to be
21017 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
21020 2017-03-17 Martin Jambor <mjambor@suse.cz>
21022 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
21023 (List of -O2 options): Likewise.
21024 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
21027 2017-03-17 Tom de Vries <tom@codesourcery.com>
21029 * gcov-dump.c (print_usage): Print bug_report_url.
21031 2017-03-17 Richard Biener <rguenther@suse.de>
21033 PR middle-end/80050
21034 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
21035 (parser::peek): Likewise.
21037 2017-03-17 Richard Biener <rguenther@suse.de>
21039 PR tree-optimization/80048
21040 * sese.c (free_sese_info): Properly release rename_map and
21041 copied_bb_map elements.
21043 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
21045 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
21046 Add linked-list forward and backlinks. Insert on
21047 construction, remove on destruction.
21048 (class pass_store_merging): Add m_stores_head field.
21049 (pass_store_merging::terminate_and_process_all_chains):
21050 Iterate over m_stores_head list.
21051 (pass_store_merging::terminate_all_aliasing_chains):
21053 (pass_store_merging::execute): Check for debug stmts first.
21054 Push new chains onto the m_stores_head stack.
21056 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21059 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
21060 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
21061 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
21063 2017-03-16 Jeff Law <law@redhat.com>
21065 PR tree-optimization/71437
21066 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
21067 member function. Implementation moved into after_dom_children
21068 member function and into the threader's thread_outgoing_edges
21070 (dom_opt_dom_walker::after_dom_children): Simplify by moving
21071 some code into new thread_outgoing_edges.
21072 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
21073 definition. Simplify marker handling (do it here). Assume we always
21074 have the available expression and the const/copies tables.
21075 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
21077 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
21078 * tree-vrp.c (equiv_stack): No longer file scoped.
21079 (vrp_dom_walker): New class.
21080 (vrp_dom_walker::before_dom_children): New member function.
21081 (vrp_dom_walker::after_dom_children): Likewise.
21082 (identify_jump_threads): Setup domwalker. Use it rather than
21083 walking edges in a random order by hand. Simplify setup/finalization.
21084 (finalize_jump_threads): Remove.
21085 (vrp_finalize): Do not call identify_jump_threads here.
21086 (execute_vrp): Do it here instead and call thread_through_all_blocks
21089 PR tree-optimization/71437
21090 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
21092 (simplify_stmt_for_jump_threading): Add basic_block argument. All
21094 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
21095 (dom_opt_dom_walker::thread_across_edge): Remove
21096 handle_dominating_asserts argument. All callers changed.
21097 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
21098 changes. Remove calls to lhs_of_dominating_assert. Other
21099 uses of handle_dominating_asserts turn into unconditional code
21100 (simplify_control_stmt_condition_1): Likewise.
21101 (simplify_control_stmt_condition): Likewise.
21102 (thread_through_normal_block, thread_across_edge): Likewise.
21103 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
21104 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
21105 object if it is not an SSA_NAME.
21106 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
21107 before calling into the VRP specific simplifiers.
21108 (identify_jump_threads): Remove handle_dominating_asserts
21111 2017-03-16 Jakub Jelinek <jakub@redhat.com>
21114 * tree-diagnostic.c (default_tree_printer): No longer static.
21115 * tree-diagnostic.h (default_tree_printer): New prototype.
21117 2017-03-16 Tamar Christina <tamar.christina@arm.com>
21119 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
21120 Change ins into fmov.
21122 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21124 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
21125 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
21126 Use h_con constraint for operand 1.
21127 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
21128 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
21130 2017-03-15 Jeff Law <law@redhat.com>
21132 PR tree-optimization/71437
21133 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
21134 (record_temporary_equivalences): Use it.
21136 PR tree-optimization/71437
21137 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
21138 tree-ssa-scopedtables.
21139 (lookup_avail_expr, build_and_record_new_cond): Likewise.
21140 (record_conditions, record_cond, vuse_eq): Likewise.
21141 (record_edge_info): Adjust to API tweak of record_conditions.
21142 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
21143 (record_temporary_equivalences, optimize_stmt): Likewise.
21144 (eliminate_redundant_computations): Likewise.
21145 (record_equivalences_from_stmt): Likewise.
21146 * tree-ssa-scopedtables.c: Include options.h and params.h.
21147 (vuse_eq): New function, moved from tree-ssa-dom.c
21148 (build_and_record_new_cond): Likewise.
21149 (record_conditions): Likewise. Accept vector of conditions rather
21150 than edge_equivalence structure for first argument.
21151 for the first argument.
21152 (avail_exprs_stack::lookup_avail_expr): New member function, moved
21153 from tree-ssa-dom.c.
21154 (avail_exprs_stack::record_cond): Likewise.
21155 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
21156 from tree-ssa-dom.c.
21157 (avail_exprs_stack): Add new member functions lookup_avail_expr
21159 (record_conditions): Declare.
21161 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
21164 * lra-constraints.c (process_alt_operands): Increase reject for
21165 reloading an input/output operand.
21167 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
21170 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
21171 insns to convert from signed/unsigned char/short to IEEE 128-bit
21173 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
21175 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
21178 * config/i386/i386.c (ix86_vector_duplicate_value): Create
21179 subreg of inner mode for values already in registers.
21181 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
21183 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
21184 iteration reg is used after the loop.
21186 2017-03-14 Martin Sebor <msebor@redhat.com>
21188 PR tree-optimization/79800
21189 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
21190 precision in negative-positive range.
21191 (format_floating): Call non-const overload with adjusted precision.
21193 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21196 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
21199 2017-03-14 Martin Sebor <msebor@redhat.com>
21201 PR middle-end/80020
21202 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
21203 * builtins.def (aligned_alloc): Use it.
21206 * Makefile.in (GTFILES): Add calls.c.
21207 * calls.c: Include "gt-calls.h".
21209 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
21211 PR rtl-optimization/79728
21212 * regs.h (struct target_regs): New field
21213 x_contains_allocatable_regs_of_mode.
21214 (contains_allocatable_regs_of_mode): New macro.
21215 * reginfo.c (init_reg_sets_1): Initialize it, and change
21216 contains_reg_of_mode so it includes global regs as well.
21217 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
21218 rather than contains_regs_of_mode.
21220 2017-03-14 Martin Liska <mliska@suse.cz>
21222 * doc/invoke.texi: Document options that can't be combined with
21223 -fcheck-pointer-bounds.
21225 2017-03-14 Martin Liska <mliska@suse.cz>
21227 PR middle-end/79831
21228 * doc/invoke.texi (-Wchkp): Document the option.
21230 2017-03-14 Martin Liska <mliska@suse.cz>
21232 * Makefile.in: Install gcov-dump.
21234 2017-03-14 Martin Liska <mliska@suse.cz>
21236 * multiple_target.c (expand_target_clones): Bail out for
21237 an invalid attribute.
21239 2017-03-14 Richard Biener <rguenther@suse.de>
21241 * alias.c (struct alias_set_entry): Pack properly.
21242 * cfgloop.h (struct loop): Likewise.
21243 * cse.c (struct set): Likewise.
21244 * ipa-utils.c (struct searchc_env): Likewise.
21245 * loop-invariant.c (struct invariant): Likewise.
21246 * lra-remat.c (struct cand): Likewise.
21247 * recog.c (struct change_t): Likewise.
21248 * rtl.h (struct address_info): Likewise.
21249 * symbol-summary.h (function_summary): Likewise.
21250 * tree-loop-distribution.c (struct partition): Likewise.
21251 * tree-object-size.c (struct object_size_info): Likewise.
21252 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
21253 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
21254 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
21255 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
21256 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
21257 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
21258 (struct _stmt_vec_info): Likewise.
21260 2017-03-14 Martin Liska <mliska@suse.cz>
21263 * multiple_target.c (create_dispatcher_calls): Check that
21264 a target can create a function dispatcher.
21266 2017-03-14 Martin Liska <mliska@suse.cz>
21269 * multiple_target.c (expand_target_clones): Drop local.local
21270 flag for default implementation.
21272 2017-03-14 Richard Biener <rguenther@suse.de>
21274 PR tree-optimization/80030
21275 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
21277 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
21279 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
21280 gcc_fallthrough() instead of __attribute__((fallthrough));
21282 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21284 * doc/gcc.texi: Remove "up" link to (DIR).
21285 * doc/gccint.texi: Ditto.
21287 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21289 * doc/install.texi (Specific) <avr>: Remove reference to
21292 2017-03-13 Jeff Law <law@redhat.com>
21294 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
21295 attribute rather than comments.
21297 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
21298 match_scratch operand is highest.
21300 2017-03-13 Martin Liska <mliska@suse.cz>
21302 PR middle-end/78339
21303 * ipa-pure-const.c (warn_function_noreturn): If the declarations
21304 is a CHKP clone, use original declaration.
21306 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21308 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
21309 (arc_conditional_register_usage): Use a different allocation order
21310 when optimizing for size.
21311 * common/config/arc/arc-common.c (arc_option_optimization_table):
21312 Section anchors default on when optimizing for size.
21314 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21316 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
21318 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21320 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
21321 * config/arc/arc.md (cpu_facility): Add cd variant.
21322 (*movqi_insn): Add code density variant.
21323 (*movhi_insn): Likewise.
21324 (*movqi_insn): Likewise.
21325 (*addsi3_mixed): Likewise.
21326 (subsi3_insn): Likewise.
21328 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21330 * config/arc/arc.md (movsi_cond_exec): Update constraint.
21332 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21334 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
21335 expressions with MINUS and UNARY ops.
21337 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21340 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
21342 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
21343 between vec_select and vector argument.
21344 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
21345 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
21346 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
21347 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
21348 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
21349 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
21351 2017-03-13 Richard Biener <rguenther@suse.de>
21354 * params.def (vect-max-peeling-for-alignment): Fix typo.
21356 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21358 * doc/install.texi (Specific) <mips-*-*>: Remove description of
21359 issue that only occurred with binutils below 2.18.
21361 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21363 * doc/install.texi (Specific) <cris-axis-elf>: No longer
21364 refer to binutils 2.11/2.12 minimum.
21366 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21368 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
21369 ftp.kernel.org and simplify binutils requirement.
21371 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
21373 * doc/invoke.texi (Warning Options): Fix spelling of link-time
21375 (Optimize Options): Ditto. Also remove redundancy.
21377 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21379 PR translation/79848
21380 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
21382 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
21383 to G_ to avoid double translation.
21385 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21387 PR translation/79923
21388 * auto-profile.c (get_combined_location): Convert leading
21389 character of diagnostics to lower case and remove trailing period.
21390 (read_profile): Likewise for various diagnostics.
21391 * config/arm/arm.c (arm_option_override): Remove trailing period
21392 from various diagnostics.
21393 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
21394 (msp430_expand_delay_cycles): Likewise.
21396 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21399 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21400 full command-line argument, rather than just "str".
21401 (aarch64_validate_march): Likewise.
21402 (aarch64_validate_mtune): Likewise.
21404 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21406 PR rtl-optimization/78911
21407 * lra-assigns.c (must_not_spill_p): New function.
21408 (spill_for): Use it.
21410 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21412 PR tree-optimization/79981
21413 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21414 ATOMIC_COMPARE_EXCHANGE ifn result.
21415 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21416 IFN_ATOMIC_COMPARE_EXCHANGE.
21418 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21421 * opts.c (parse_sanitizer_options): Add missing question mark to
21422 "did you mean" message.
21424 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21426 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21428 (VMULEUH_UNS): Likewise.
21429 (VMULOUB_UNS): Likewise.
21430 (VMULOUH_UNS): Likewise.
21431 * config/rs6000/rs6000.c (builtin_function_type): Remove
21432 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21434 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21437 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21438 x with result of extra_parsing_for_operand_code_0.
21439 (function_reader::extra_parsing_for_operand_code_0): Convert
21440 return type from void to rtx, returning x. When reading
21441 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21442 larger size containing struct block_symbol.
21444 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21446 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21447 -mfloat128-hardware without -m64.
21449 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21452 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21453 entries to the case statement that marks unsigned arguments to
21454 overloaded functions.
21456 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21458 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21459 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21461 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21464 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21465 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21467 2017-03-10 Martin Liska <mliska@suse.cz>
21471 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21472 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21475 2017-03-10 Olivier Hainque <hainque@adacore.com>
21477 * tree-switch-conversion (array_value_type): Start by resetting
21478 candidate type to it's main variant.
21480 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21482 PR rtl-optimization/79909
21483 * combine.c (try_combine): Use simplify_replace_rtx on individual
21484 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21485 of the whole CALL_INSN_FUNCTION_USAGE.
21487 PR tree-optimization/79972
21488 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21489 get_range_info on SSA_NAMEs. Formatting fixes.
21491 2017-03-10 Richard Biener <rguenther@suse.de>
21492 Jakub Jelinek <jakub@redhat.com>
21494 PR tree-optimization/77975
21495 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21496 edge to be constant.
21497 (get_val_for): For constant x return it. Formatting fix.
21498 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21499 would use the same bases as the current one.
21501 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21503 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21504 instead of vec_select for V1TImode.
21505 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21507 (VSX_LE_128): Add V1TI to this mode iterator.
21508 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21509 (*vsx_le_perm_store_<mode>): Likewise.
21510 (pre-reload splitter for VSX stores): Likewise.
21511 (post-reload splitter for VSX stores): Likewise.
21512 (*vsx_xxpermdi2_le_<mode>): Likewise.
21513 (*vsx_lxvd2x2_le_<mode>): Likewise.
21514 (*vsx_stxvd2x2_le_<mode>): Likewise.
21516 2017-03-09 Michael Eager <eager@eagercon.com>
21518 Correct failures with --enable-checking=yes,rtl.
21520 * config/microblaze/microblaze.c (microblaze_expand_shift):
21521 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21522 test for const0_rtx.
21523 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21524 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21526 2017-03-09 Richard Biener <rguenther@suse.de>
21528 PR tree-optimization/79977
21529 * graphite-scop-detection.c (scop_detection::merge_sese):
21530 Handle the case of extra exits to blocks dominating the entry.
21532 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21534 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21537 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21539 PR rtl-optimization/79949
21540 * lra-constraints.c (process_alt_operands): Check memory when
21541 trying to predict a cycle. Print about the overall increase.
21543 2017-03-09 Richard Biener <rguenther@suse.de>
21545 PR middle-end/79971
21546 * gimple-expr.c (useless_type_conversion_p): Preserve
21547 TYPE_SATURATING for fixed-point types.
21549 2017-03-09 Richard Biener <rguenther@suse.de>
21552 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21553 alignment of BLKmode params.
21555 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21558 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21559 (VALL_NO_V2Q): Likewise.
21561 * config/aarch64/aarch64-simd.md
21562 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21564 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21565 VALL_NO_V2Q mode iterator.
21566 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21568 2017-03-09 Martin Liska <mliska@suse.cz>
21570 PR tree-optimization/79631
21571 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21572 tree_int_cst_sign_bit just for INTEGER constants.
21574 2017-03-09 Martin Liska <mliska@suse.cz>
21578 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21581 2017-03-09 Marek Polacek <polacek@redhat.com>
21584 * tree.c (inchash::add_expr): Handle TREE_VEC.
21586 2017-03-09 Martin Liska <mliska@suse.cz>
21589 (chkp_narrow_size_and_offset): New function.
21590 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21591 (void chkp_parse_bit_field_ref): New function.
21592 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21593 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21595 2017-03-09 Martin Liska <mliska@suse.cz>
21598 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21599 (chkp_find_bounds_1): Remove gcc_unreachable.
21601 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21604 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21605 BUILT_IN_SYNC*, determine the access type from the size suffix and
21606 always build a MEM_REF with that type. Handle forgotten
21607 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21610 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21611 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21612 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21613 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21614 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21615 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21616 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21617 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21618 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21619 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21620 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21621 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21622 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21623 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21624 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21625 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21626 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21627 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21628 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21629 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21630 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21631 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21632 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21633 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21634 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21635 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
21636 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
21637 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
21638 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
21639 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
21640 definitions outside of __OPTIMIZE__ guarded section.
21643 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
21644 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
21645 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
21646 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
21649 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21651 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21652 ("vfenez<mode>"): Add missing constraints.
21654 2017-03-08 Martin Sebor <msebor@redhat.com>
21657 * config/nds32/nds32.c (nds32_option_override):
21658 Fix misspelled diagnostic.
21660 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21663 * gimplify.c (gimplify_omp_for): Replace index var in outer
21664 taskloop statement with an artificial variable and add
21665 OMP_CLAUSE_PRIVATE clause for it.
21667 2017-03-08 Richard Biener <rguenther@suse.de>
21669 PR tree-optimization/79955
21670 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21671 for accesses that are completely outside of the variable.
21673 2017-03-08 Andrew Haley <aph@redhat.com>
21675 PR tree-optimization/79943
21676 * tree-ssa-loop-split.c (compute_new_first_bound): When
21677 calculating the new upper bound, (END-BEG) should be added, not
21680 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21682 * config/avr/avr.md (setmemhi): Make sure match_dup
21683 operand number comes before match_scratch.
21685 2017-03-08 Richard Biener <rguenther@suse.de>
21687 PR tree-optimization/79920
21688 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21689 with ncopies == 1 to ...
21690 (vect_transform_slp_perm_load): ... here. Properly compute
21691 all element loads by iterating VF times over the group. Do
21692 not handle ncopies (computed in a broken way) in
21693 vect_create_mask_and_perm.
21695 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21698 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21699 is a uniform vector, use uniform_vector_p return value instead of
21700 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21702 2017-03-07 Marek Polacek <polacek@redhat.com>
21704 PR middle-end/79809
21705 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21706 (alloca_call_type): Likewise.
21708 2017-03-07 Martin Liska <mliska@suse.cz>
21710 * gcov.c (process_args): Put comment to correct location.
21712 2017-03-07 Martin Liska <mliska@suse.cz>
21714 PR middle-end/68270
21715 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21716 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21717 (chkp_narrow_bounds_for_field): Likewise.
21718 (chkp_parse_array_and_component_ref): Pass one more argument to
21721 2017-03-07 Richard Biener <rguenther@suse.de>
21723 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21726 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21728 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21729 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21731 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21734 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21735 to end of description.
21736 (PARAM_MAX_STORES_TO_MERGE): Likewise.
21738 2017-03-07 Jakub Jelinek <jakub@redhat.com>
21740 PR rtl-optimization/79901
21741 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
21743 (*avx512f_<code><mode>3<mask_name>): ... this.
21744 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
21745 iterator instead of VI8_AVX2_AVX512BW.
21747 PR rtl-optimization/79901
21748 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
21749 min/max expander, expand it using expand_vec_cond_expr.
21752 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
21755 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21758 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
21759 to void * for PCH reasons.
21760 * dwarf2out.c (output_loc_operands, output_die): Cast
21761 v.val_vec.array to unsigned char *.
21763 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
21766 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
21769 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
21771 PR rtl-optimization/79571
21772 * lra-constraints.c (process_alt_operands): Calculate static
21773 reject and subtract it from overall when only addresses will be
21776 2017-03-06 Julia Koval <julia.koval@intel.com>
21779 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
21780 incoming stack boundary to 128 for 64-bit targets.
21782 2017-03-06 Richard Biener <rguenther@suse.de>
21784 PR tree-optimization/79894
21785 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
21786 to NULL after folding it.
21788 2017-03-06 Richard Biener <rguenther@suse.de>
21790 PR tree-optimization/79824
21791 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
21792 check disabling peeling for gaps.
21794 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
21796 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21797 attributes): Document gettimeofday.
21799 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21801 * config/s390/s390.c (s390_option_override_internal): Set
21802 PARAM_MIN_VECT_LOOP_BOUND
21804 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21806 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
21807 * config/s390/s390.md: Likewise.
21809 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21812 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
21813 (<avx2_avx512>_perm<mode>): Rename to ...
21814 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
21816 (<avx512>_perm<mode>_mask): Rename to ...
21817 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
21819 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
21820 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
21821 instead of VI8F_256_512.
21822 (avx512f_perm<mode>): New define_expand.
21823 (avx512f_perm<mode>_mask): Likewise.
21824 (avx512f_perm<mode>_1<mask_name>): New define_insn.
21825 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
21827 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21829 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
21830 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
21832 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
21834 2017-03-06 Martin Liska <mliska@suse.cz>
21837 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
21838 when having a SSA NAME w/o VAR_DECL assigned to it.
21840 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21842 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
21843 msa_dpsub_<su>_d): Fix MODE for vec_select.
21845 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21847 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
21849 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
21851 2017-03-06 Richard Biener <rguenther@suse.de>
21853 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
21854 * plugin.c (register_plugin_info): Likewise.
21855 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
21857 2017-03-05 Jakub Jelinek <jakub@redhat.com>
21859 * config/i386/sse.md (sse_storehps, sse_storelps,
21860 avx_<castmode><avxsizesuffix>_<castmode>,
21861 avx512f_<castmode><avxsizesuffix>_<castmode>,
21862 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
21863 in condition that at least one operand is not a MEM.
21865 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21867 PR middle-end/79805
21868 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
21869 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
21871 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
21872 gimple_call_nothrow_p flag based on whether original builtin can throw.
21873 If it can, emit following stmts on the fallthrough edge.
21874 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
21875 don't create new bb if inserting just debug stmts on the edge, try to
21876 insert them on the fallthru bb or just reset debug stmts.
21878 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
21881 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
21882 restore recog_data (including the operand rtxes inside it) around
21883 the call to get_insn_template.
21885 2017-03-03 Martin Sebor <msebor@redhat.com>
21887 PR tree-optimization/79699
21888 * context.c (context::~context): Free MPFR caches to avoid
21889 a memory leak on program exit.
21891 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21893 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
21894 Use wide_int::ulow () instead of .elt (0).
21896 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21898 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
21899 (*pushxf): Limit oF constraint to 32bit targets and add oC
21900 constraint for 64bit targets.
21901 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
21902 (*pushdf): Change rmF constraint to rmC.
21904 2017-03-03 Martin Liska <mliska@suse.cz>
21906 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
21907 Remove unused variable.
21909 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21912 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
21913 is a memory operand, increase num_memory.
21914 (ix86_expand_args_builtin): Likewise.
21916 2017-03-03 Jan Hubicka <jh@suse.cz>
21919 * ipa-devirt.c (maybe_record_node): Properly handle
21920 __cxa_pure_virtual visibility.
21922 2017-03-03 Martin Liska <mliska@suse.cz>
21924 PR tree-optimization/79803
21925 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
21927 (pass_loop_prefetch::execute): Disabled optimization if an
21928 assumption about L1 cache size is not met.
21930 2017-03-03 Martin Liska <mliska@suse.cz>
21932 PR rtl-optimization/79574
21933 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
21934 (hash_scan_set): Likewise.
21935 (dump_hash_table): Likewise.
21936 (hoist_code): Likewise.
21938 2017-03-03 Richard Biener <rguenther@suse.de>
21940 * fixed-value.c (fixed_from_string): Restore use of elt (1)
21941 in place of uhigh ().
21942 (fixed_convert_from_real): Likewise.
21944 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21947 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
21949 2017-03-03 Richard Biener <rguenther@suse.de>
21951 PR middle-end/79818
21952 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
21953 TYPE_OVERFLOW_UNDEFINED check.
21955 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21957 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
21959 (vector_ae_<mode>_p): Likewise.
21960 (vector_nez_<mode>_p): Likewise.
21961 (vector_ne_v2di_p): Likewise.
21962 (vector_ae_v2di_p): Likewise.
21963 (vector_ne_<mode>_p): Likewise.
21964 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
21966 (vsx_tsqrt<mode>2_fe): Likewise.
21968 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
21971 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
21973 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21975 PR rtl-optimization/79780
21976 * cprop.c (one_cprop_pass): When second and further conditional trap
21977 in a single basic block is turned into an unconditional trap, turn it
21978 into a deleted note to avoid RTL verification failures.
21980 2017-03-02 Richard Biener <rguenther@suse.de>
21982 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21984 2017-03-02 Richard Biener <rguenther@suse.de>
21986 PR tree-optimization/79345
21988 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21989 param and abort the walk, returning -1 if it is hit.
21990 (walk_aliased_vdefs): Take a limit param and pass it on.
21991 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21992 defaulting to 0 and return a signed int.
21993 * tree-ssa-uninit.c (struct check_defs_data): New struct.
21994 (check_defs): New helper.
21995 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
21996 about uninitialized memory.
21997 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
21998 bogus uninitialized warning.
21999 (fixed_convert_from_real): Likewise.
22001 2017-03-02 Bin Cheng <bin.cheng@arm.com>
22003 PR tree-optimization/66768
22004 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
22005 iv_use if base object can't be determined.
22007 2017-03-02 Jakub Jelinek <jakub@redhat.com>
22009 PR tree-optimization/79345
22010 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
22011 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
22012 (get_pattern_stats): Initialize it.
22013 * genemit.c (gen_expand): Verify match_scratch numbers come after
22014 match_operand/match_dup numbers.
22015 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
22016 match_scratch numbers.
22017 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
22019 * config/s390/s390.md (trunctdsd2): Likewise.
22021 2017-03-02 Richard Biener <rguenther@suse.de>
22023 * wide-int.h (wide_int_storage::operator=): Implement in terms
22026 2017-03-02 Richard Biener <rguenther@suse.de>
22028 PR tree-optimization/79777
22029 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
22030 the to insert expression to sth existing.
22032 2017-03-01 Martin Sebor <msebor@redhat.com>
22034 PR middle-end/79692
22035 * gimple-ssa-sprintf.c
22036 (directive::known_width_and_precision): New function.
22037 (format_integer): Use it.
22038 (get_mpfr_format_length): Consider the full range of precision
22039 when computing %g output with the # flag. Set the likely byte
22040 count to 3 rather than 1 when precision is indeterminate.
22041 (format_floating): Correct the lower bound of precision.
22043 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22045 * doc/invoke.texi: Document default code model for 64-bit Linux.
22047 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22050 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
22051 udiv rather than div since input pattern is unsigned.
22053 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
22055 * config/i386/i386.c (print_reg): Warn for values of
22056 unsupported size in integer register.
22058 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
22061 * config/rs6000/predicates.md (current_file_function_operand): Do
22062 not allow self calls to be local if the function is replaceable.
22064 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
22067 * config/rs6000/altivec.h (vec_ctz and others): Change the
22068 preprocessor macro that controls conditional compilation from
22069 _ARCH_PWR9 to __POWER9_VECTOR__.
22070 (vec_all_ne): Change parameterization of __altivec_scalar_pred
22071 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
22072 control (instead of _ARCH_PWR9 control) so that template
22073 definition uses power9-specific function.
22074 (vec_any_eq): Likewise.
22075 (vec_all_ne): Change macro definition to use a power9-specific
22076 expansion under #ifdef __POWER9_VECTOR__ control (instead of
22077 _ARCH_PWR9 control).
22078 (vec_any_eq) Likewise.
22079 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
22080 expansion for CMPNEF to remove support for xvcmpnesp instruction.
22081 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
22082 support for xvcmpnedp instruction.
22083 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
22084 macro expansion so that Power9 implementation of vec_all_ne does
22085 not use the AltiVec predicate framework.
22086 (VCMPNEH_P): Likewise.
22087 (VCMPNEW_P): Likewise.
22088 (VCMPNED_P): Likewise.
22089 (VCMPNEFP_P): Likewise.
22090 (VCMPNEDP_P): Likewise.
22091 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
22092 implementation of vec_any_eq to not use AltiVec predicate
22094 (VCMPAEH_P): Likewise.
22095 (VCMPAEW_P): Likewise.
22096 (VCMPAED_P): Likewise.
22097 (VCMPAEFP_P): Likewise.
22098 (VCMPAEDP_P): Likewise.
22099 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
22100 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
22101 not use the AltiVec predicate framework.
22102 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
22103 of vec_any_eq to not use AltiVec predicate framework.
22104 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22105 support for predefined __POWER9_VECTOR__ macro to indicate that
22106 Power9 instruction selection is enabled.
22107 (altivec_overloaded_builtins): Remove extraneous
22108 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
22109 function argument types RS6000_BTI_bool_V16QI and
22110 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
22111 entry for overloaded function argument types RS6000_BTI_bool_V4SI
22112 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
22113 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
22114 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
22115 Power9 for implementations of vec_cmpne. Change the signature for
22116 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
22117 (representing vec_all_ne) to remove the previously described first
22118 argument of type RS6000_BTI_INTSI, as this was an artifact of
22119 reliance on the AltiVec predicate framework, which is no longer
22120 used in the implementation of these functions. Add
22121 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
22122 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
22123 since, unlike the AltiVec predicate framework implementation, we
22124 do not share function descriptors between vec_alle and vec_anyeq.
22125 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
22126 set of modes that receive special treatment even when
22127 TARGET_P9_VECTOR is true. The special treatment emits code that
22128 does not depend on Power9 instructions.
22129 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
22130 define_expand to not rely on AltiVec predicate framework.
22131 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22133 (vector_ne_v2di_p): Change this define_expand to not rely on
22134 AltiVec predicate framework.
22135 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
22137 (vector_ne_<mode>_p): Change this define_expand to not rely on
22138 AltiVec predicate framework.
22139 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22141 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
22142 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
22143 define_insn pattern.
22144 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
22145 define_insn pattern because the xvcmpne<VSs>. instruction is not
22147 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
22148 instruction is not supported.
22150 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22152 * config/nvptx/nvptx.c: Include intl.h.
22154 2017-03-01 Martin Jambor <mjambor@suse.cz>
22157 * ipa-prop.h (ipa_bits): Removed field known.
22158 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
22159 to pointers. Adjusted their comments to warn about their sharing.
22160 (ipcp_transformation_summary): Change bits to a vector of pointers.
22161 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
22162 (ipa_get_ipa_bits_for_value): Declare.
22163 * tree-vrp.h (value_range): Mark as GTY((for_user)).
22164 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
22165 (ipa_bits_hash_table): Likewise.
22166 (ipa_vr_ggc_hash_traits): Likewise.
22167 (ipa_vr_hash_table): Likewise.
22168 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
22169 being pointers and vr_known being removed.
22170 (ipa_set_jf_unknown): Likewise.
22171 (ipa_get_ipa_bits_for_value): New function.
22172 (ipa_set_jfunc_bits): Likewise.
22173 (ipa_get_value_range): New overloaded functions.
22174 (ipa_set_jfunc_vr): Likewise.
22175 (ipa_compute_jump_functions_for_edge): Use the above functions to
22176 construct bits and vr parts of jump functions.
22177 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
22178 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22180 (ipcp_grow_transformations_if_necessary): Also allocate
22181 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22183 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
22184 them. Fix too long lines.
22185 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
22186 vr_known being removed.
22187 (ipa_read_jump_function): Use new setter functions to construct bits
22188 and vr parts of jump functions or set them to NULL.
22189 (write_ipcp_transformation_info): Adjust for bits being pointers.
22190 (read_ipcp_transformation_info): Likewise.
22191 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
22193 Include gt-ipa-prop.h.
22194 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
22196 (ipcp_store_bits_results): Likewise.
22197 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
22198 Do not write to existing jump functions but use a temporary instead.
22200 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22203 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
22204 attempt to use its first operand as BIT_FIELD_REF base.
22206 2017-03-01 Richard Biener <rguenther@suse.de>
22208 PR middle-end/79721
22209 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
22210 interpolating formula in wrapping arithmetic.
22211 (chrec_apply): Convert chrec_evaluate return value to wanted type.
22213 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22215 PR tree-optimization/79734
22216 * tree-vect-generic.c (expand_vector_condition): Optimize
22217 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
22218 Handle VEC_COND_EXPR where comparison has different inner width from
22219 type's inner width.
22221 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
22223 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
22224 markup, and similar issues. Remove @opindex entries for things
22225 that aren't options. Add missing -mmpy-option entries.
22227 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22229 PR tree-optimization/79737
22230 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
22231 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
22232 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
22233 instead of byte_size. Formatting fix.
22234 (shift_bytes_in_array_right): Formatting fix.
22236 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
22239 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
22240 condition on optimize for the leaf function test.
22242 2017-02-28 Martin Liska <mliska@suse.cz>
22245 * read-rtl-function.c (function_reader::handle_unknown_directive):
22246 Bail out when one uses -flto.
22248 2017-02-28 Martin Liska <mliska@suse.cz>
22250 * common.opt: Replace space with tabular for options of <number>
22252 * config/i386/i386.opt: Show <number> value for
22253 -mlarge-data-threshold.
22254 * opts.c (print_filtered_help): Do not display number in hexadecimal
22257 2017-02-28 Martin Liska <mliska@suse.cz>
22259 * common.opt: Fix --help=option -Q for options which are of
22262 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
22264 * config/i386/i386.c (print_reg): Error out for values
22265 of 8-bit size in invalid integer register.
22267 2017-02-28 Martin Sebor <msebor@redhat.com>
22269 PR tree-optimization/79691
22270 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
22272 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22275 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
22276 gcc_unreachable with output_operand_lossage.
22278 2017-02-28 Richard Biener <rguenther@suse.de>
22280 PR tree-optimization/79740
22281 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
22283 (visit_nary_op): Insert the nary into the hashtable if we
22284 pattern-matched sth.
22285 * tree-ssa-pre.c (eliminate_insert): Robustify.
22287 2017-02-28 Richard Biener <rguenther@suse.de>
22289 PR middle-end/79731
22290 * fold-const.c (decode_field_reference): Reject out-of-bound
22293 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22295 * config/i386/i386.c: Include intl.h.
22296 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
22297 instead of just cond ? "..." : "...".
22298 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
22299 * coverage.c (read_counts_file): Likewise.
22300 * omp-offload.c: Include intl.h.
22301 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
22302 of just cond ? "..." : "...".
22303 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
22304 of just cond ? "..." : "...".
22306 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
22309 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
22311 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
22312 'tune for' CPU name.
22313 * config/arm/arm-cpu-data.h: Regenerated.
22315 2017-02-28 Richard Biener <rguenther@suse.de>
22317 PR tree-optimization/79732
22318 * tree-inline.c (expand_call_inline): Do not shadow var.
22320 2017-02-28 Richard Biener <rguenther@suse.de>
22322 PR tree-optimization/79723
22323 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
22324 address-space properly.
22326 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
22328 * doc/optinfo.texi (Optimization groups): Fix option used for
22330 * doc/invoke.texi (-fopt-info): Document "omp".
22331 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
22332 (OPTGROUP_ALL): Add OPTGROUP_OMP.
22333 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
22334 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
22335 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
22337 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
22339 * dumpfile.c (optgroup_options): Instead of "openmp", associate
22340 OPTGROUP_OMP with "omp".
22342 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
22345 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
22346 for arithmetic shift of unsigned V2DI.
22348 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
22350 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
22351 arc/linux.h headers.
22352 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
22353 (LINK_SPEC): Likewise.
22354 (ARC_TLS_EXTRA_START_SPEC): Likewise.
22355 (EXTRA_SPECS): Likewise.
22356 (STARTFILE_SPEC): Likewise.
22357 (ENDFILE_SPEC): Likewise.
22358 (LIB_SPEC): Likewise.
22359 (TARGET_SDATA_DEFAULT): Likewise.
22360 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
22361 (MULTILIB_DEFAULTS): Likewise.
22362 (DWARF2_UNWIND_INFO): Likewise.
22363 * config/arc/big.h: New file.
22364 * config/arc/elf.h: Likewise.
22365 * config/arc/linux.h: Likewise.
22366 * config/arc/t-uClibc: Remove.
22368 2017-02-27 Bin Cheng <bin.cheng@arm.com>
22370 PR tree-optimization/77536
22371 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
22372 (tree_transform_and_unroll_loop): Use above function to compute the
22373 estimated niter of unrolled loop and use it when scaling profile.
22374 Also use count info rather than frequency if it's non-zero.
22375 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
22376 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
22377 (vect_transform_loop): Call above function.
22379 2017-02-27 Richard Biener <rguenther@suse.de>
22381 PR tree-optimization/45397
22382 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
22383 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
22384 (visit_nary_op): Add pattern matching for CSEing sign-changed
22385 or truncated operations with wider ones.
22387 2017-02-27 Richard Biener <rguenther@suse.de>
22389 PR tree-optimization/79690
22390 * tree-vect-stmts.c (vectorizable_store): Use vector type
22391 built from the DR with address-space.
22393 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22395 * doc/invoke.texi (Optimize Options): Refine the description
22396 of asan-use-after-return.
22398 2017-02-25 Alan Modra <amodra@gmail.com>
22400 PR rtl-optimization/79584
22401 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22402 base, not ad->base_term, the reg within base. Remove assertion
22403 that ad->base == ad->base_term. Replace gen_int_mode using
22404 bogus mode with const0_rtx.
22406 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22408 PR middle-end/79396
22409 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22410 FMA_EXPR like tcc_binary or tcc_unary.
22412 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22415 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22417 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22418 (output_loc_operands): Handle DW_OP_call_ref and
22419 DW_OP_GNU_variable_value.
22420 (struct variable_value_struct): New type.
22421 (struct variable_value_hasher): Likewise.
22422 (variable_value_hash): New variable.
22423 (string_types): Remove.
22424 (copy_loc_descr): New function.
22425 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22426 (prepend_loc_descr_to_each): New function.
22427 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22428 instead of add_loc_descr_to_each if the first argument is single
22429 location list and the second has multiple.
22430 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22431 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22432 when looking for variable value which doesn't have other location info.
22433 (loc_list_from_tree): Formatting fix.
22434 (gen_array_type_die): Simplify DW_AT_string_length handling.
22435 (adjust_string_types): Remove.
22436 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22437 string_types. Call resolve_variable_values.
22438 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22439 (resolve_addr_in_expr): Likewise. Add A argument.
22440 (copy_deref_exprloc): Remove deref argument. Adjust for the
22441 original expression being DW_OP_GNU_variable_value with optionally
22442 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22443 optionally after it.
22444 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22445 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22446 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22447 (variable_value_hasher::hash, variable_value_hasher::equal): New
22449 (resolve_variable_value_in_expr, resolve_variable_value,
22450 resolve_variable_values, note_variable_value_in_expr,
22451 note_variable_value): New functions.
22452 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22455 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22458 * opts.h (handle_generated_option): Add GENERATED_P argument.
22459 * opts-common.c (handle_option): Adjust function comment.
22460 (handle_generated_option): Add GENERATED_P argument, pass it to
22462 (control_warning_option): Pass false to handle_generated_option
22464 * opts.c (maybe_default_option): Pass true to handle_generated_option
22466 * optc-gen.awk: Likewise.
22468 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22470 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22471 a REG, look at the REG it is a SUBREG of.
22472 (splitter for cmpeqsi_t): Ditto.
22474 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22476 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22477 the special USEs with the pattern of the insn, not the insn itself.
22479 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22482 * doc/invoke.texi: Document -mload-store-pairs.
22484 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22485 Sandra Loosemore <sandra@codesourcery.com>
22487 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22488 argument isn't a CONST_INT.
22489 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22490 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22491 (nios2_validate_compare): Bypass alternate compare logic if *op2
22492 is not a CONST_INT.
22493 (ldstwm_operation_p): Return false if first_base is not a REG or
22494 if first_offset is not a CONST_INT.
22496 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22498 * config/cris/cris.md: Use correct operand in a define_peephole2.
22500 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22502 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22504 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22506 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22507 this_insn if it is an INSN or JUMP_INSN.
22508 (force_offsettable): Look at base, not at addr.
22509 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22510 on things that aren't necessarily CONST_INTs.
22512 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22514 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22515 -mfpmath=sse is the default also for x86-32 targets with SSE2
22516 instruction set when @option{-ffast-math} is enabled
22518 2017-02-24 Jeff Law <law@redhat.com>
22520 PR rtl-optimizatoin/79286
22521 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22524 2017-02-24 Richard Biener <rguenther@suse.de>
22526 PR tree-optimization/79389
22527 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22530 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22532 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22533 function comment to reflect reality.
22534 (loop_exits_before_overflow): Fix typo in function description.
22536 2017-02-24 Richard Biener <rguenther@suse.de>
22538 PR tree-optimization/79389
22539 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22540 properly that a threading opportunity exists. Detect conditional
22541 copy/constant propagation opportunities.
22543 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22545 * config/visium/visium.md (type): Add trap.
22546 (b): New mode attribute.
22547 (*btst): Rename into...
22548 (*btst<mode>): ...this and adjust.
22549 (*cbranchsi4_btst_insn): Rename into...
22550 (*cbranch<mode>4_btst_insn): ...this and adjust.
22551 (trap): New define_insn.
22553 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22555 PR tree-optimization/79389
22556 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22557 (noce_reversed_cond_code): New function.
22558 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22559 reversed_comparison_code. Formatting fix.
22560 (noce_try_store_flag): Test rev_cond != NULL in addition to
22561 reversed_comparison_code.
22562 (noce_try_store_flag_constants): Likewise.
22563 (noce_try_store_flag_mask): Likewise.
22564 (noce_try_addcc): Use rev_cond if non-NULL instead of
22565 reversed_comparison_code.
22566 (noce_try_cmove_arith): Likewise. Formatting fixes.
22567 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22568 (noce_find_if_block): Initialize rev_cond.
22569 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22570 instead of false as last argument never attempt to reverse it
22573 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22575 PR tree-optimization/79663
22576 * tree-predcom.c (combine_chains): Process refs in reverse order
22577 only for ZERO length chains, and add explaining comment.
22579 2017-02-23 Jeff Law <law@redhat.com>
22581 PR tree-optimization/79578
22582 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22583 in call to operand_equal_p.
22585 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22588 * config/i386/cpuid.h: Fix another undefined behavior.
22590 2017-02-23 Richard Biener <rguenther@suse.de>
22592 PR tree-optimization/79683
22593 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22594 vector types for data-refs.
22596 2017-02-23 Martin Liska <mliska@suse.cz>
22598 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22600 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22602 PR middle-end/79665
22603 * internal-fn.c (get_range_pos_neg): Moved to ...
22604 * tree.c (get_range_pos_neg): ... here. No longer static.
22605 * tree.h (get_range_pos_neg): New prototype.
22606 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22607 are known to be in between 0 and signed maximum inclusive, try to
22608 expand both unsigned and signed divmod and use the cheaper one from
22611 2017-02-22 Jeff Law <law@redhat.com>
22613 PR tree-optimization/79578
22614 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22615 to compare base operands.
22617 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22620 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22621 gpc_reg_operand instead of fpr_reg_operand.
22623 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22625 * config/mips/mips.c (mips_return_in_memory): Force FP
22626 vector types to be returned in memory for o32 ABI.
22628 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22630 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22631 instead of DW_TAG_member for static data member declarations and don't
22632 set no_linkage_name for static inline data members.
22633 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
22636 2017-02-22 Martin Liska <mliska@suse.cz>
22638 * doc/invoke.texi: Replace inequality signs with square brackets
22641 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22643 PR tree-optimization/68644
22644 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
22646 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
22649 * lra-constraints.c (simplify_operand_subreg): Handle
22650 WORD_REGISTER_OPERATIONS targets.
22652 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22655 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22656 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22657 elimination by swapping fld*.
22659 2017-02-22 Richard Biener <rguenther@suse.de>
22661 PR tree-optimization/79673
22662 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22663 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22664 irrelevant address-space qualifiers and avoiding a
22665 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22667 2017-02-22 Richard Biener <rguenther@suse.de>
22669 PR tree-optimization/79666
22670 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22671 to not symbolically negate if that may introduce undefined
22674 2017-02-22 Martin Liska <mliska@suse.cz>
22677 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22678 * data-streamer-out.c (streamer_write_gcov_count_stream):
22680 * value-prof.c (stream_out_histogram_value): Make assert more
22681 precise based on type of counter.
22683 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22686 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22687 Use nonimmediate_operand instead of memory_operand for operand 1.
22688 (float-extend standard_x87sse_constant_load splitter): Ditto.
22690 2017-02-21 Jeff Law <law@redhat.com>
22692 PR tree-optimization/79621
22693 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22694 blocks with edges to themselves.
22696 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22699 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22700 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22701 Use gimple_call_builtin_p.
22704 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22705 on temporarily removed DEBUG_INSNs.
22707 PR tree-optimization/79649
22708 * tree-loop-distribution.c (classify_partition): Give up on
22709 non-generic address space loads/stores.
22711 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22713 * doc/loop.texi (Loop manipulation): Remove nonexistent
22714 tree_ssa_loop_version from the documentation.
22715 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22717 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22720 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22721 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22722 * config/rs6000/rs6000.c: Include except.h.
22723 (rs6000_expand_split_stack_prologue): Call
22724 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22726 2017-02-21 Martin Jambor <mjambor@suse.cz>
22729 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22730 have been analyzed.
22732 2017-02-21 Martin Jambor <mjambor@suse.cz>
22734 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22735 for backward compatibility only.
22736 * doc/invoke.texi (Option Summary): Remove all references to
22737 -fipa-cp-alignment.
22739 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
22743 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22745 * lra-constraints.c (curr_insn_transform): Handle
22746 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22748 2017-02-21 Martin Liska <mliska@suse.cz>
22750 * config/i386/i386.opt: Replace -masm-dialect with -masm.
22752 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
22754 PR translation/79638
22755 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
22757 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
22760 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
22761 (arm_function_ok_for_sibcall): Return false for an indirect call by
22762 descriptor if all the argument registers are used.
22763 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
22764 alignment of the function.
22766 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22768 PR tree-optimization/61441
22769 * simplify-rtx.c (simplify_const_unary_operation): For
22770 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
22771 the sNaN unmodified.
22773 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22775 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22776 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22777 instead of SYSTEM_HEADER_DIR.
22779 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
22780 Martin Liška <mliska@suse.cz>
22782 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
22783 Fix typos and grammar, use active voice, and clarify.
22785 2017-02-20 Marek Polacek <polacek@redhat.com>
22787 PR middle-end/79537
22788 * gimplify.c (gimplify_expr): Handle unused *&&L;.
22791 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
22793 2017-02-20 Jakub Jelinek <jakub@redhat.com>
22796 * config/i386/i386.c (ix86_expand_builtin): Handle
22797 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
22798 ix86_builtins_isa[fcode].isa as a requirement of those
22799 flags and any other flag in the bitmask.
22800 (ix86_init_mmx_sse_builtins): Use 0 instead of
22801 ~OPTION_MASK_ISA_64BIT as mask.
22802 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
22803 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
22804 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
22805 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
22807 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22810 * lra-constraints.c (split_reg): Check requested split mode
22811 is supported by the register.
22813 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22815 * lra-constraints.c (simplify_operand_subreg): Remove early
22818 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22821 * lra-constraints.c (curr_insn_transform): Tighten condition
22822 for converting SUBREG reloads from OP_OUT to OP_INOUT.
22824 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22827 * lra-constraints.c (curr_insn_transform): Handle
22828 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22830 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
22833 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
22835 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
22837 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
22840 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
22843 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22845 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
22848 2017-02-18 Jakub Jelinek <jakub@redhat.com>
22850 * final.c (last_columnnum, override_columnnum): New variables.
22851 (final_start_function): Set last_columnnum, pass it to begin_prologue
22852 hook and pass 0 to dwarf2out_begin_prologue.
22853 (final_scan_insn): Update override_columnnum. Pass last_columnnum
22854 to source_line debug hook.
22855 (notice_source_line): Compute last_columnnum and for debug_column_info
22856 return true on column changes.
22857 * debug.h (struct gcc_debug_hooks): Add column argument to
22858 source_line and begin_prologue hooks.
22859 (debug_nothing_int_charstar_int_bool): Remove prototype.
22860 (debug_nothing_int_int_charstar,
22861 debug_nothing_int_int_charstar_int_bool): New prototypes.
22862 (dwarf2out_begin_prologue): Add column argument.
22863 * debug.c (do_nothing_debug_hooks): Adjust source_line and
22864 begin_prologue hooks.
22865 (debug_nothing_int_charstar_int_bool): Remove.
22866 (debug_nothing_int_int_charstar,
22867 debug_nothing_int_int_charstar_int_bool): New functions.
22868 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
22869 through to dwarf2out_source_line.
22870 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
22871 (dwarf2out_source_line): Add column argument, emit it if requested.
22872 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
22874 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22875 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22876 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
22877 through to dwarf2out_begin_prologue.
22878 (vmsdbgout_source_line): Add column argument, pass it through to
22879 dwarf2out_source_line.
22880 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
22881 dbxout_source_line caller.
22882 (dbxout_source_line): Add column argument.
22884 * common.opt (gno-column-info, gcolumn-info): New options.
22885 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
22886 (check_die): Also test for multiple DW_AT_decl_column attributes.
22887 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
22888 DW_AT_decl_column if requested.
22889 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
22891 (gen_variable_die): Likewise.
22892 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
22893 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
22896 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
22897 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
22898 (ix86_handle_option): Handle OPT_m3dnowa.
22899 * doc/invoke.texi (-m3dnowa): Document.
22900 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
22901 -m3dnowa instead of -m3dnow -march=athlon.
22904 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
22905 instead of gcc_assert for K, r and R code checks. Formatting fixes.
22907 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22910 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
22911 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
22912 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
22913 generator for vsx_xxpermdi_<mode>_be.
22914 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
22915 force big-endian semantics.
22916 (vsx_xxpermdi_<mode>_be): New define_expand with same
22917 implementation as previous version of vsx_xxpermdi_<mode>.
22919 2017-02-17 Jakub Jelinek <jakub@redhat.com>
22921 PR tree-optimization/79327
22922 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
22923 variable, its initialization and use.
22925 2017-02-17 Julia Koval <julia.koval@intel.com>
22927 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
22928 (OPTION_MASK_ISA_PKU_UNSET): New.
22929 (ix86_handle_option): Handle -mrdpid.
22930 * config/i386/cpuid.h (bit_RDPID): New.
22931 * config/i386/driver-i386.c (host_detect_local_cpu):
22932 Detect RDPID feature.
22933 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
22934 * config/i386/i386-c.c (ix86_target_macros_internal):
22936 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
22937 (ix86_valid_target_attribute_inner_p): Add "rdpid".
22938 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22939 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
22940 * config/i386/i386.md (define_insn "rdpid"): New.
22941 * config/i386/i386.opt Add -mrdpid.
22942 * config/i386/immintrin.h (_rdpid_u32): New.
22944 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
22946 PR rtl-optimization/79541
22947 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
22948 instead of transforming it into USE.
22950 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
22952 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
22953 If HONOR_SNANS (SFmode) force the input to a register.
22954 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
22955 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
22956 an frsp or similar insn.
22958 2017-02-17 Martin Liska <mliska@suse.cz>
22960 PR rtl-optimization/79577
22961 * params.def (selsched-max-sched-times): Increase minimum to 1.
22963 2017-02-17 Martin Liska <mliska@suse.cz>
22965 PR rtl-optimization/79574
22966 * gcse.c (want_to_gcse_p): Prevent integer overflow.
22968 2017-02-17 Martin Liska <mliska@suse.cz>
22970 PR tree-optimization/79529
22971 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
22972 ssa_defined_default_def_p to handle cases which are implicitly
22974 * tree-ssa.c (ssa_defined_default_def_p): New function.
22975 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
22976 which are implicitly defined.
22977 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22979 2017-02-17 Richard Biener <rguenther@suse.de>
22981 PR middle-end/79576
22982 * params.def (max-ssa-name-query-depth): Limit to 10.
22984 2017-02-17 Richard Biener <rguenther@suse.de>
22986 PR tree-optimization/79552
22987 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22990 2017-02-17 Richard Biener <rguenther@suse.de>
22993 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
22995 2017-02-17 Marek Polacek <polacek@redhat.com>
22997 PR middle-end/79536
22998 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
22999 (fold_negate_expr): New wrapper.
23001 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
23003 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
23004 Correct terminology and de-emphasize pre-standard behavior.
23006 2017-02-16 Alan Modra <amodra@gmail.com>
23008 PR rtl-optimization/79286
23009 * ira.c (def_dominates_uses): New function.
23010 (update_equiv_regs): Don't create an equivalence for insns that
23011 may trap where the register def does not dominate the use.
23013 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
23015 PR rtl-optimization/78127
23016 * lra.c (lra): Call lra_eliminate before finish the loop after
23019 2017-02-16 Richard Biener <rguenther@suse.de>
23021 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
23023 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
23024 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
23025 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
23026 (isl_val_int_from_wi): New function.
23027 (extract_affine_gmp): Rename to ...
23028 (extract_affine_wi): ... this, take a widest_int.
23029 (extract_affine_int): Just wrap extract_affine_wi.
23030 (add_param_constraints): Use isl_val_int_from_wi.
23031 (add_loop_constraints): Likewise, and extract_affine_wi.
23033 2017-02-15 Jeff Law <law@redhat.com>
23035 PR middle-end/79521
23036 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
23037 ira_init_register_move_cost_if_necessary.
23039 2017-02-15 Martin Sebor <msebor@redhat.com>
23041 PR middle-end/32003
23042 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
23043 removed in a prior commit.
23045 2017-02-15 Bin Cheng <bin.cheng@arm.com>
23047 PR tree-optimization/79347
23048 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
23049 counters during peeling.
23051 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
23053 * Makefile.in (site.exp): Remove "set ISLVER".
23055 2017-02-15 Jakub Jelinek <jakub@redhat.com>
23058 * real.c (real_from_integer): Call real_convert even for decimal.
23060 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
23063 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
23065 2017-02-14 Andrew Pinski <apinski@cavium.com>
23067 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
23068 cores and change the partno/implementer to be correct.
23069 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
23070 the 'B" as the implementer.
23071 * config/aarch64/aarch64-tune.md: Regenerate.
23073 2017-02-14 Carl Love <cel@us.ibm.com>
23075 * config/rs6000/rs6000.c: Add case statement entry to make the
23076 xvcvuxdsp built-in argument unsigned.
23077 * config/rs6000/vsx.md: Fix the source and return operand types so they
23078 match the instruction definitions from the ISA document. Fix typo
23079 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
23082 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
23085 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
23086 member early_clobber_alts.
23087 * lra-lives.c (reg_early_clobber_p): New.
23088 (process_bb_lives): Use it.
23089 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
23090 (debug_operand_data): Initialize early_clobber_alts.
23091 (setup_operand_alternative): Set up early_clobber_alts.
23092 (collect_non_operand_hard_regs): Ditto. Pass early clobber
23093 alternatives to new_insn_reg.
23094 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
23096 (lra_update_insn_regno_info): Pass the new arg.
23098 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23100 PR middle-end/79505
23101 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
23102 (new_oacc_loop_raw): Don't clear already cleared fields.
23105 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
23106 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
23107 _mm512_prefetch_i64gather_ps): New inline functions and macros.
23109 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
23112 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
23114 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
23117 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
23118 the extra instruction to the right place to store 128-bit constant
23121 2017-02-14 Martin Sebor <msebor@redhat.com>
23123 PR middle-end/79448
23124 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
23125 warning for strings of unknown length.
23127 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23129 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
23131 2017-02-14 Jeff Law <law@redhat.com>
23134 * ira-costs.c (scan_one_insn): Initialize register move costs
23135 for pseudos seen in USE/CLOBBER insns.
23137 PR tree-optimization/79095
23138 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
23139 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
23140 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
23141 if the operands are known to be not equal, then the resulting range
23143 (intersect_ranges): If the new range is ~[0,0] and the old range is
23144 wide, then prefer ~[0,0].
23145 * tree-vrp.c (overflow_comparison_p_1): New function.
23146 (overflow_comparison_p): New function.
23147 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
23148 if NAME is used in an overflow test.
23149 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
23150 overflow check that can be expressed as an equality test, then adjust
23151 ops to be that equality test.
23153 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23155 * config/s390/s390-builtin-types.def: Remove flags argument.
23156 * config/s390/s390.c (s390_init_builtins): Likewise.
23158 2017-02-14 Martin Liska <mliska@suse.cz>
23160 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
23161 vector. Fix trailing white spaces.
23163 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
23165 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
23168 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23170 PR rtl-optimization/68664
23171 * config/arm/arm.c (arm_sched_can_speculate_insn):
23172 New function. Declare prototype.
23173 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23175 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23177 PR rtl-optimization/68664
23178 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
23180 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23182 2017-02-14 Amit Pawar <amit.pawar@amd.com>
23184 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
23185 max skip bytes for function, loop and jump.
23187 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23189 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
23190 ABS_EXPR for gimple dump.
23192 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23195 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
23197 PR tree-optimization/79408
23198 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
23199 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
23200 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
23201 also if rhs1 is INTEGER_CST.
23203 2017-02-14 Richard Biener <rguenther@suse.de>
23205 PR middle-end/79432
23206 * tree-into-ssa.c (insert_phi_nodes): When the function can
23207 have abnormal edges rewrite SSA names with broken use-def
23208 dominance out of SSA and register them for PHI insertion.
23210 2017-02-13 Martin Sebor <msebor@redhat.com>
23212 PR middle-end/79496
23213 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
23214 clearing info.nowrite flag when snprintf size argument is a range.
23216 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23218 * cprop.c (cprop_jump): Add missing space in string literal.
23219 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
23220 (get_constraint_for_component_ref): Likewise.
23221 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
23222 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
23223 * lra-constraints.c (process_alt_operands): Likewise.
23224 * ipa-inline.c (inline_small_functions): Likewise.
23225 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
23226 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23227 * trans-mem.c (diagnose_tm_1_op): Likewise.
23228 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
23229 (grid_parallel_clauses_gridifiable): Likewise.
23231 * config/nvptx/mkoffload.c (process): Add space in between
23234 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
23235 "MOD4_SSE_REGS" and "ALL_REGS".
23237 * spellcheck.c (test_data): Add , in between "foo" and "food".
23239 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23242 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
23243 boundary crossing check and subsequent code generation agree.
23245 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23247 * config/aarch64/aarch64.c (has_memory_op): Delete.
23248 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
23251 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23253 PR rtl-optimization/79388
23254 PR rtl-optimization/79450
23255 * combine.c (distribute_notes): When removing TEM_INSN for which
23256 corresponding dest has last value recorded, invalidate that last
23259 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23261 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
23262 of explicit '@'. Add missing assembly comment marker on branch costs
23265 2017-02-13 Nathan Sidwell <nathan@acm.org>
23267 * gengtype-lex.l (<in_struct>): Add '/'.
23269 2017-02-13 Martin Liska <mliska@suse.cz>
23272 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
23274 2017-02-13 Richard Biener <rguenther@suse.de>
23276 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
23278 * configure: Re-generate.
23279 * config.in: Likewise.
23280 * graphite-dependences.c: Simplify as if
23281 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
23282 * graphite-isl-ast-to-gimple.c: Likewise.
23283 * graphite-optimize-isl.c: Likewise.
23284 * graphite-poly.c: Likewise.
23285 * graphite-sese-to-poly.c: Likewise.
23286 * graphite.h: Likewise.
23287 * toplev.c: Include isl/version.h and use isl_version () for
23288 printing the ISL version.
23289 * doc/install.texi: Update ISL requirement.
23291 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23293 * doc/standards.texi (Standards): Update reference to
23296 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23298 * doc/extend.texi (Named Address Spaces): sourceware.org now
23300 * doc/install.texi (Binaries): Ditto.
23303 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23305 * doc/cpp.texi: Replace "stringify"/"stringification" with C
23306 standard terminology "stringize"/"stringizing" throughout.
23307 * doc/cppinternals.texi: Likewise.
23309 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23311 * doc/extend.texi: Fix some spelling mistakes and typos.
23312 * doc/invoke.texi: Likewise.
23314 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23317 * params.def (inline-min-speedup) Change from 10 to 8.
23319 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23321 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
23324 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23327 * ipa-inline-analysis.c (get_minimal_bb): New function.
23328 (record_modified): Use it.
23329 (remap_edge_change_prob): Handle also ancestor functions.
23331 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
23333 * doc/contrib.texi (Contributors): Remove broken link into
23334 the Mauve CVS repository.
23336 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23338 PR middle-end/79454
23339 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
23340 result computation whenever lhs doesn't have vector mode, not
23341 just when it has BLKmode.
23343 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23345 * doc/makefile.texi (profiledbootstrap): Refer to the
23346 installation instructions only in textual form.
23348 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23351 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
23353 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23355 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
23356 (Specific): Update mingw-w64 reference.
23358 (Specific): Remove broken link to Renesas RX processor.
23360 2017-02-10 Richard Biener <rguenther@suse.de>
23362 * toplev.c (process_options): Do not mention obsolete graphite
23363 options when printing sorry message about missing graphite support.
23364 Mention -floop-nest-optimize.
23366 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
23368 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
23369 (vtst_p16): Likewise.
23370 (vtstq_p8): Likewise.
23371 (vtstq_p16): Likewise.
23373 (vtstq_p64): Likewise.
23374 * config/arm/arm_neon.h (vgetq_lane_p64): New.
23375 (vset_lane_p64): New.
23376 (vsetq_lane_p64): New.
23378 2017-02-10 Jakub Jelinek <jakub@redhat.com>
23380 PR tree-optimization/79411
23381 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
23382 stmt operands are SSA_NAMEs used in abnormal phis.
23383 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
23386 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23389 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
23392 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23394 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23396 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23398 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23402 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23403 not arbitrary TREE_CONSTANT.
23406 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23407 "omp declare target link" attribute unless is_global_var.
23408 * omp-offload.c (find_link_var_op): Likewise.
23410 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23411 Chung-Lin Tang <cltang@codesourcery.com>
23413 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23415 (gimplify_adjust_omp_clauses): Don't delete TILE.
23416 (gimplify_omp_for): Deal with TILE.
23417 * internal-fn.c (expand_GOACC_TILE): New function.
23418 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23420 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23421 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23423 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23424 avoid DIV for outermost collapse var.
23425 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23426 Remove out of date comments, fix whitespace.
23427 * omp-general.c (omp_extract_for_data): Deal with tiling.
23428 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23429 adjust OLF_DIM_BASE value.
23430 (struct omp_for_data): Add tiling field.
23431 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23432 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23433 for auto loops. Remove default auto determining, moved to
23434 oacc_loop_fixed_partitions.
23435 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23436 stmts, add e_mask field.
23437 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23438 (oacc_thread_numbers): Use oacc_dim_call.
23439 (oacc_xform_tile): New.
23440 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23441 (finish_oacc_loop): Adjust for ifns vector.
23442 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23443 add case for GOACC_TILE fns.
23444 (oacc_loop_xform_loop): Delete.
23445 (oacc_loop_process): Iterate over call list directly, and add
23446 handling for GOACC_TILE fns.
23447 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23449 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23450 vector partitioning to outer loops. Assign 2 partitions to loops
23451 when available. Add TILE handling.
23452 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23453 (execite_oacc_device_lower): Process GOACC_TILE fns,
23454 ignore unknown specs.
23455 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23456 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23457 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23459 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23461 * configure.ac (ACX_BUGURL): Update.
23462 * configure: Regenerate.
23464 2017-02-09 Richard Biener <rguenther@suse.de>
23466 PR tree-optimization/69823
23467 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23468 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23470 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23472 * config/arc/arc-c.def: Add __NPS400__ definition.
23473 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23474 (TARGET_NPS400): Define.
23476 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23478 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23480 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23481 pointer, arch_info.
23482 (arc_cpu_types): Fill the arch_info field with a pointer into the
23483 arc_arch_types table.
23484 (arc_selected_cpu): Declare.
23485 * config/arc/arc.c (arc_selected_cpu): Make global.
23486 (arc_selected_arch): Delete.
23487 (arc_base_cpu): Delete.
23488 (arc_override_options): Remove references to deleted variables,
23489 update access to arch information.
23490 (ARC_OPT): Update access to arch information.
23491 (ARC_OPTX): Likewise.
23492 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23493 (TARGET_ARC600): Update access to arch information.
23494 (TARGET_ARC601): Likewise.
23495 (TARGET_ARC700): Likewise.
23496 (TARGET_EM): Likewise.
23497 (TARGET_HS): Likewise.
23498 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23501 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23504 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23505 condition/operands for integer GE/LE/GEU/LEU operations.
23507 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23509 PR translation/79397
23510 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23513 2017-02-08 Martin Jambor <mjambor@suse.cz>
23516 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23517 whether allocation happened.
23518 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23519 nothing was allocated.
23521 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23523 PR tree-optimization/79408
23524 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23525 constant, but SSA_NAME with a known integer range, use the minimum
23526 of that range instead of op1 to determine if modulo can be replaced
23527 with its first operand.
23529 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23531 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23533 2017-02-08 Richard Biener <rguenther@suse.de>
23535 PR tree-optimization/71824
23536 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23537 Check all loops contained in the merged region.
23539 2017-02-07 Andrew Pinski <apinski@cavium.com>
23541 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23543 2017-02-07 Andrew Pinski <apinski@cavium.com>
23545 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23546 (thunderxt88): Likewise.
23547 (thunderxt81): Disable LSE and change v8.1 to v8.
23548 (thunderxt83): Likewise.
23550 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23551 Richard Biener <rguenther@suse.de>
23553 PR middle-end/79399
23554 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23555 type from int to size_t.
23556 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23558 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23560 PR rtl-optimization/79386
23561 * cprop.c (bypass_conditional_jumps): Initialize
23562 bypass_last_basic_block already before splitting bbs after
23563 unconditional traps...
23564 (bypass_conditional_jumps): ... rather than here.
23567 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23568 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23569 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23570 fix -masm=intel patterns.
23572 2017-02-07 Richard Biener <rguenther@suse.de>
23574 PR tree-optimization/79256
23575 PR middle-end/79278
23576 * builtins.c (get_object_alignment_2): Use min_align_of_type
23577 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23578 and ADJUST_FIELD_ALIGN.
23580 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23582 * doc/tm.texi: Regenerate.
23583 * stor-layout.c (layout_decl): Adjust.
23584 (update_alignment_for_field): Likewise.
23585 (place_field): Likewise.
23586 (min_align_of_type): Likewise.
23587 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23588 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23589 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23590 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23591 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23592 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23593 * config/i386/i386.c (x86_field_alignment): Likewise.
23594 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23595 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23596 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23597 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23598 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23599 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23603 2017-01-30 Richard Biener <rguenther@suse.de>
23605 PR tree-optimization/79256
23606 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23607 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23610 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23612 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23613 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23614 builtins to SImode and emit a zero-extend, if necessary.
23616 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23618 * docs/invoke.texi (RISC-V Options): Alphabetize.
23620 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23622 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23625 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23627 * config/riscv/riscv.c: New file.
23628 * common/config/riscv/riscv-common.c: Likewise.
23629 * config.gcc: Likewise.
23630 * config/riscv/constraints.md: Likewise.
23631 * config/riscv/elf.h: Likewise.
23632 * config/riscv/generic.md: Likewise.
23633 * config/riscv/linux.h: Likewise.
23634 * config/riscv/multilib-generator: Likewise.
23635 * config/riscv/peephole.md: Likewise.
23636 * config/riscv/pic.md: Likewise.
23637 * config/riscv/predicates.md: Likewise.
23638 * config/riscv/riscv-builtins.c: Likewise.
23639 * config/riscv/riscv-c.c: Likewise.
23640 * config/riscv/riscv-ftypes.def: Likewise.
23641 * config/riscv/riscv-modes.def: Likewise.
23642 * config/riscv/riscv-opts.h: Likewise.
23643 * config/riscv/riscv-protos.h: Likewise.
23644 * config/riscv/riscv.h: Likewise.
23645 * config/riscv/riscv.md: Likewise.
23646 * config/riscv/riscv.opt: Likewise.
23647 * config/riscv/sync.md: Likewise.
23648 * config/riscv/t-elf-multilib: Likewise.
23649 * config/riscv/t-linux: Likewise.
23650 * config/riscv/t-linux-multilib: Likewise.
23651 * config/riscv/t-riscv: Likewise.
23652 * configure.ac: Likewise.
23653 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23654 Waterman as RISC-V maintainers.
23655 * doc/install.texi: Add RISC-V entries.
23656 * doc/invoke.texi: Add RISC-V options section.
23657 * doc/md.texi: Add RISC-V constraints section.
23658 * configure: Regenerated.
23660 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23663 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23664 false values to be constant vectors with all 0 or all 1 bits set.
23665 (vcondu<mode><mode>): Likewise.
23666 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23668 (fpmask_comparison_operator): Update comment.
23669 (vecint_comparison_operator): New predicate.
23670 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23671 vector conditionals when the true and false values are constant
23672 vectors with all 0 bits or all 1 bits set.
23674 2017-02-06 Martin Sebor <msebor@redhat.com>
23676 PR tree-optimization/79376
23677 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23679 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23681 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23682 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23683 to simplify split condition.
23685 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23687 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23688 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23691 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23693 PR rtl-optimization/68664
23694 * target.def (can_speculate_insn): New hook.
23695 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23696 * doc/tm.texi: Regenerate.
23697 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23698 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23699 (rs6000_sched_can_speculate_insn): New function.
23701 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23703 PR tree-optimization/79284
23704 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23705 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23706 vectorizable_mask_load_store, vectorizable_operation,
23707 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23708 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23709 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23710 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23711 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23712 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23713 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23714 is_gimple_assign (stmt). Replace another such test with
23715 is_gimple_assign (stmt).
23717 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23720 * config/avr/avr.c (rtl-iter.h): Include it.
23721 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23722 (avr_legitimate_combined_insn): ...and implementation.
23724 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23726 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23727 * config/s390/s390.c (s390_const_operand_ok)
23728 (s390_canonicalize_comparison, s390_extract_part)
23729 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23730 (s390_contiguous_bitmask_p, s390_rtx_costs)
23731 (legitimize_pic_address): Likewise.
23732 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23733 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23734 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23735 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23736 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
23738 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23740 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
23741 REGNO($0) == REGNO($1).
23743 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23745 * config/s390/linux.h(SIZE_TYPE): Add comment.
23747 2017-02-06 Julian Brown <julian@codesourcery.com>
23748 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23749 Virendra Pathak <virendra.pathak@broadcom.com>
23751 * config/aarch64/aarch64-cores.def: Change the scheduler
23753 * config/aarch64/aarch64.md: Include thunderx2t99.md.
23754 * config/aarch64/thunderx2t99.md: New file.
23756 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23758 * doc/standards.texi (Go Language): Update link to language
23761 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23763 * tree-eh.c (lower_resx): Sanitize profile.
23764 (cleanup_empty_eh_move_lp): Likewise.
23766 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23769 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
23771 * cfgloopmanip.h (loop_version): Update prototype.
23772 * modulo-sched.c (sms_schedule): Update call of loop_version.
23773 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
23774 * tree-parloops.c (gen_parallel_loop): Likewise.
23775 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
23776 * tree-ssa-loop-split.c (split_loop): Likewise.
23777 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
23778 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23780 2017-02-05 Martin Liska <mliska@suse.cz>
23783 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
23785 (print_operand_address): Initialize a struct to zero.
23787 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23789 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
23790 garbage collector only in textual form.
23792 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23794 * doc/extend.texi (x86 specific memory model extensions for
23795 transactional memory): Simplify a phrase.
23797 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
23800 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
23801 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
23802 (atomic_storedi_1): Likewise.
23804 2017-02-04 Jakub Jelinek <jakub@redhat.com>
23806 PR tree-optimization/79338
23807 * tree-parloops.c (gather_scalar_reductions): Don't call
23808 vect_analyze_loop_form for loop->inner before destroying loop's
23811 2017-02-03 Martin Sebor <msebor@redhat.com>
23813 PR tree-optimization/79327
23814 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
23815 when precision has resulted in leading zeros.
23816 (format_integer): Adjust the likely counter to assume an unknown
23817 argument that may be zero is non-zero.
23819 2017-02-03 Jason Merrill <jason@redhat.com>
23822 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
23823 avoid copying non-taken branch.
23825 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23827 PR tree-optimization/79340
23828 * tree-vect-loop.c (vectorizable_reduction): Release
23829 vec_defs elements after safe_splicing them into other vectors.
23832 PR tree-optimization/79327
23833 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
23834 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
23836 (format_integer): Use wide_int_to_tree instead of build_int_cst
23837 + to_?hwi. If argmin is NULL, just set argmin and argmax to
23838 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
23839 of shortest and longest sequence.
23841 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
23843 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
23844 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
23846 2017-02-03 Walter Lee <walt@tilera.com>
23849 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
23850 after initial stackframe link reg save.
23851 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
23853 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23856 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
23857 wu for stxssp alternative.
23859 2017-02-03 Martin Sebor <msebor@redhat.com>
23861 PR tree-optimization/79352
23862 * gimple-fold.c (get_range_strlen): Add argument.
23863 (get_range_strlen): Change return type to bool.
23864 (get_maxval_strlen): Pass in a dummy argument.
23865 * gimple-fold.h (get_range_strlen): Change return type to bool.
23866 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
23867 * tree.h (array_at_struct_end_p): Add argument.
23868 * tree.c (array_at_struct_end_p): Handle it.
23870 2017-02-03 Martin Liska <mliska@suse.cz>
23873 * multiple_target.c (create_dispatcher_calls): Redirect edge
23874 from a caller of a dispatcher.
23875 (expand_target_clones): Make the clones local.
23876 (ipa_target_clone): Do both target clones and resolvers.
23877 (ipa_dispatcher_calls): Remove the pass.
23878 (pass_dispatcher_calls::gate): Likewise.
23879 (make_pass_dispatcher_calls): Likewise.
23880 * passes.def (pass_target_clone): Put as very first IPA early
23883 2017-02-03 Martin Liska <mliska@suse.cz>
23885 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
23886 in case of a function with ifunc attribute.
23888 2017-02-03 Martin Liska <mliska@suse.cz>
23890 * cgraph.c (cgraph_node::dump): Dump function version info.
23891 * symtab.c (symtab_node::dump_base): Add missing new line.
23893 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23895 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
23896 (ifcombine_ifandif): Use it.
23898 2017-02-03 Martin Liska <mliska@suse.cz>
23900 * doc/invoke.texi: Document default value for
23901 use-after-scope-direct-emission-threshold.
23903 2017-02-03 Martin Liska <mliska@suse.cz>
23905 PR tree-optimization/79339
23906 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
23907 (format_floating): Likewise.
23909 2017-02-03 Martin Liska <mliska@suse.cz>
23912 * ipa-prop.c (ipa_node_params_t::insert): Remove current
23914 (ipa_node_params_t::remove): Likewise.
23915 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
23916 initialization from removed ipa_node_params_t::insert.
23917 (ipa_node_params::~ipa_node_params): Move from removed
23918 ipa_node_params_t::release.
23919 * symbol-summary.h (symbol_summary::m_released): New member.
23920 Do not release a summary twice. Do not allow to call finalizer
23921 for types of a summary that live in GGC memory.
23923 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23925 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
23928 2017-02-02 Martin Sebor <msebor@redhat.com>
23930 PR middle-end/79275
23931 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
23932 (format_string): Tighten up the range of output for non-constant
23933 strings and correct the expected range for wide non-constant strings.
23935 2017-02-02 Martin Sebor <msebor@redhat.com>
23937 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
23939 PR middle-end/32003
23940 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
23942 (-fdump-tree-@var): Add to index and document how to come up
23943 with pass-specific option and dump file names.
23944 (-fdump-passes): Clarify where to look for output.
23946 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23948 PR middle-end/77445
23949 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
23950 statistics of the analyzed path; allow threading for speed when
23951 any of BBs along the path are optimized for speed.
23953 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
23955 PR middle-end/78468
23956 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
23957 settings of the virtual registers.
23960 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
23962 * explow.c (get_dynamic_stack_size): Take known alignment of stack
23963 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
23966 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23968 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
23969 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
23971 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23973 * config/s390/s390.md: Add missing comments with the expanded
23975 * config/s390/vector.md: Likewise.
23976 * config/s390/vx-builtins.md: Likewise.
23978 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23981 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23982 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23983 conditions on a single line.
23985 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23987 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23988 __S390_VX__ to __VX__.
23990 2017-02-01 Andrew Pinski <apinski@cavium.com>
23992 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
23993 stmt_info to record_stmt_cost.
23994 (vect_get_known_peeling_cost): Pass stmt_info if known to
23996 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
23997 cpu_vector_cost field into
23998 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
23999 field into vec_int_stmt_cost and vec_fp_stmt_cost.
24000 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
24001 splitting of scalar_stmt_cost and vec_stmt_cost.
24002 (thunderx_vector_cost): Likewise.
24003 (cortexa57_vector_cost): LIkewise.
24004 (exynosm1_vector_cost): Likewise.
24005 (xgene1_vector_cost): Likewise.
24006 (thunderx2t99_vector_cost): Improve after the splitting of the two
24008 (aarch64_builtin_vectorization_cost): Update for the splitting of
24009 scalar_stmt_cost and vec_stmt_cost.
24011 2017-02-01 Torvald Riegel <triegel@redhat.com>
24012 Richard Henderson <rth@redhat.com>
24014 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
24015 conditional on existance of a fast atomic load.
24016 * optabs-query.c (can_atomic_load_p): New function.
24017 * optabs-query.h (can_atomic_load_p): Declare it.
24018 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
24019 no fast atomic load is available for the particular size of access.
24020 (expand_atomic_compare_and_swap): Likewise.
24021 (expand_atomic_load): Likewise.
24022 (expand_atomic_store): Likewise.
24023 (expand_atomic_fetch_op): Likewise.
24024 * testsuite/lib/target-supports.exp
24025 (check_effective_target_sync_int_128): Remove x86 because it provides
24026 no fast atomic load.
24027 (check_effective_target_sync_int_128_runtime): Likewise.
24029 2017-02-01 Richard Biener <rguenther@suse.de>
24031 * graphite.c: Include tree-vectorizer.h for find_loop_location.
24032 (graphite_transform_loops): Provide opt-info for optimized nests.
24033 * tree-parloop.c (parallelize_loops): Provide opt-info for
24034 parallelized loops.
24036 2017-02-01 Richard Biener <rguenther@suse.de>
24038 PR middle-end/79315
24039 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
24040 was not set before.
24042 2017-02-01 Richard Biener <rguenther@suse.de>
24044 PR tree-optimization/71824
24045 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
24046 Verify the loops are valid in the merged SESE region.
24047 (scop_detection::can_represent_loop_1): Check analyzing the
24048 evolution of the number of iterations in the region succeeds.
24050 2017-01-31 Ian Lance Taylor <iant@golang.org>
24052 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
24053 REG_ARGS_SIZE note to 32-bit push insns and call insn.
24055 2017-01-31 David Malcolm <dmalcolm@redhat.com>
24057 PR preprocessor/79210
24058 * input.c (get_substring_ranges_for_loc): Replace line_width
24059 assertion with error-handling.
24061 2017-01-31 Richard Biener <rguenther@suse.de>
24063 PR tree-optimization/77318
24064 * graphite-sese-to-poly.c (extract_affine): Fix assert.
24065 (create_pw_aff_from_tree): Take loop parameter.
24066 (add_condition_to_pbb): Pass loop of the condition to
24067 create_pw_aff_from_tree.
24069 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24071 * config/s390/s390.c (s390_asan_shadow_offset): New function.
24072 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
24074 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
24078 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
24080 (convert_int_to_float128): Likewise.
24081 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
24082 (convert_int_to_float128): Likewise.
24083 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
24084 (UNSPEC_IEEE128_CONVERT): Likewise.
24085 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
24086 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
24087 Use local variables for IBM extended format.
24088 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
24089 (fix_trunc<mode>si2_fprs): Likewise.
24090 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
24091 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
24092 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
24093 to know that we can now have integers of all sizes in vector
24095 (fix<uns>_<mode>di2_hw): Likewise.
24096 (float<uns>_<mode>si2_hw): Likewise.
24097 (fix_<mode>si2_hw): Likewise.
24098 (fixuns_<mode>si2_hw): Likewise.
24099 (float<uns>_<mode>di2_hw): Likewise.
24100 (float_<mode>di2_hw): Likewise.
24101 (float_<mode>si2_hw): Likewise.
24102 (floatuns_<mode>di2_hw): Likewise.
24103 (floatuns_<mode>si2_hw): Likewise.
24104 (xscvqp<su>wz_<mode>): Delete, no longer used.
24105 (xscvqp<su>dz_<mode>): Likewise.
24106 (xscv<su>dqp_<mode>): Likewise.
24107 (ieee128_mfvsrd_64bit): Likewise.
24108 (ieee128_mfvsrd_32bit): Likewise.
24109 (ieee128_mfvsrwz): Likewise.
24110 (ieee128_mtvsrw): Likewise.
24111 (ieee128_mtvsrd_64bit): Likewise.
24112 (ieee128_mtvsrd_32bit): Likewise.
24114 2017-01-31 Martin Liska <mliska@suse.cz>
24117 * ipa-prop.c (ipa_free_all_node_params): Call release method
24118 instead of ~sumbol_summary to not to trigger double times
24121 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
24123 PR tree-optimization/71691
24124 * bitmap.h (class auto_bitmap): New.
24125 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
24126 is_maybe_undefined instead of ssa_undefined_value_p.
24128 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24130 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24131 __S390_ARCH_LEVEL__ to __ARCH__.
24133 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24135 PR tree-optimization/79267
24136 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
24137 if should_remove_lhs_p is true.
24139 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
24142 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
24143 (add_alignment_attribute): New.
24144 (base_type_die): Add alignment attribute.
24145 (subrange_type_die): Likewise.
24146 (modified_type_die): Likewise.
24147 (gen_array_type_die): Likewise.
24148 (gen_descr_array_type_die: Likewise.
24149 (gen_enumeration_type_die): Likewise.
24150 (gen_subprogram_die): Likewise.
24151 (gen_variable_die): Likewise.
24152 (gen_field_die): Likewise.
24153 (gen_ptr_to_mbr_type_die): Likewise.
24154 (gen_struct_or_union_type_die): Likewise.
24155 (gen_subroutine_type_die): Likewise.
24156 (gen_typedef_die): Likewise.
24157 (base_type_cmp): Compare alignment attribute.
24159 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24162 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
24163 (setb_unsigned) New pattern for setb with CCUNS.
24164 * config/rs6000/rs6000.c (expand_block_compare): Use a different
24165 subfc./subfe sequence to avoid overflow problems. Generate a
24166 shorter sequence with cmpld/setb for power9.
24167 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
24168 for generating subfc. instruction.
24169 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
24170 now uses this instruction.
24172 2017-01-30 Ian Lance Taylor <iant@google.com>
24175 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
24176 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
24178 2017-01-30 Martin Sebor <msebor@redhat.com>
24180 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
24181 Move constant to the right of a relational operator.
24182 (get_mpfr_format_length, format_character, format_string): Ditto.
24183 (should_warn_p, maybe_warn): Same.
24185 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
24187 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
24190 * asan.c (get_translation_unit_decl): Remove function.
24191 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
24193 2017-01-30 Martin Liska <mliska@suse.cz>
24195 PR gcov-profile/79259
24196 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
24197 -fprofile-generate.
24199 2017-01-30 Martin Liska <mliska@suse.cz>
24202 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
24203 Initialize variables with NULL value.
24205 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
24208 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
24210 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
24212 2017-01-30 Richard Biener <rguenther@suse.de>
24214 PR tree-optimization/79276
24215 * tree-vrp.c (process_assert_insertions): Properly adjust common
24216 when removing a duplicate.
24218 * gcc.dg/torture/pr79276.c: New testcase.
24220 2017-01-30 Richard Biener <rguenther@suse.de>
24222 PR tree-optimization/79256
24223 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24224 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24226 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
24228 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
24231 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
24232 ("*r<noxa>sbg_<mode>_sll_bitmask")
24233 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
24234 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
24235 Use contiguous_bitmask_nowrap_operand.
24237 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24240 * config/rs6000/altivec.h (vec_xl): Revise #define.
24241 (vec_xst): Likewise.
24243 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
24245 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
24247 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
24249 PR rtl-optimization/79194
24250 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
24251 traps before call to bypass_conditional_jumps.
24253 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24255 PR tree-optimization/71374
24256 * lra-constraints.c (check_conflict_input_operands): New.
24257 (match_reload): Use it.
24259 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24262 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
24263 account to calculate conflict_set.
24265 2017-01-27 Bin Cheng <bin.cheng@arm.com>
24267 PR rtl-optimization/78559
24268 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
24269 other_insn in combine.
24271 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24273 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
24274 uint16_type_node for BT_UINT16.
24276 2017-01-27 David Malcolm <dmalcolm@redhat.com>
24278 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
24279 "RTL Tests" to menu.
24280 (GIMPLE Tests): New node.
24281 (RTL Tests): New node.
24283 2017-01-27 Richard Biener <rguenther@suse.de>
24285 PR tree-optimization/79245
24286 * tree-loop-distribution.c (distribute_loop): Apply cost
24287 modeling also to detected patterns.
24289 2017-01-27 Richard Biener <rguenther@suse.de>
24291 PR tree-optimization/71433
24292 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
24293 (compare_assert_loc): New function.
24294 (process_assert_insertions): Sort and optimize assert locations
24295 to remove duplicates and push down identical assertions on
24296 edges to their destination block.
24298 2017-01-27 Richard Biener <rguenther@suse.de>
24300 PR tree-optimization/79244
24301 * tree-vrp.c (remove_range_assertions): Forcefully propagate
24302 out SSA names even if abnormal.
24304 2017-01-27 Jakub Jelinek <jakub@redhat.com>
24306 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
24307 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
24308 instead of MPFR_RNDN.
24310 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
24313 * arm.c (arm_option_override): Don't call build_target_option_node
24314 until after doing all option overrides.
24315 (arm_valid_target_attribute_tree): Likewise.
24317 2017-01-27 Martin Liska <mliska@suse.cz>
24319 * doc/invoke.texi (-fprofile-arcs): Document profiling support
24320 for {cd}tors and C++ {cd}tors.
24322 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
24324 * config/s390/s390.md ("*setmem_long_and")
24325 ("*setmem_long_and_31z"): Use zero_extend instead of and.
24327 2017-01-26 Martin Sebor <msebor@redhat.com>
24329 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
24332 2017-01-26 Martin Sebor <msebor@redhat.com>
24334 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
24335 HAVE_DFmode before using XFmode or DFmode.
24336 (parse_directive): Avoid using the z length modifier to avoid
24337 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
24339 PR middle-end/78703
24340 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
24341 to accept adjustment as an array.
24342 (get_int_range): New function.
24343 (struct directive): Make width and prec arrays.
24344 (directive::set_width, directive::set_precision): Call get_int_range.
24345 (format_integer, format_floating): Handle width and precision ranges.
24346 (format_string, parse_directive): Same.
24348 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24351 * dwarf2out.c (generate_skeleton_bottom_up): For children with
24352 comdat_type_p set, just clone them, but keep the children in the
24356 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
24357 which have direct callers with -fvar-tracking-assignments enabled
24359 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
24360 inside of type units.
24362 2017-01-26 Martin Sebor <msebor@redhat.com>
24364 PR middle-end/78703
24365 * gimple-ssa-sprintf.c (struct result_range): Add likely and
24367 (struct format_result): Replace number_chars, number_chars_min,
24368 and number_chars_max with a single member of struct result_range.
24370 (format_result::operator+=): Adjust.
24371 (struct fmtresult): Remove bounded. Handle likely and unlikely
24373 (fmtresult::adjust_for_width_or_precision): New function.
24374 (fmtresult:type_max_digits): New function.
24375 (bytes_remaining): Handle likely and unlikely counters.
24376 (min_bytes_remaining): Remove.
24377 (format_percent): Simplify.
24378 (format_integer, format_floating): Set likely and unlikely counters.
24379 (get_string_length, format_character, format_string): Same.
24380 (format_plain, should_warn_p): New function.
24381 (maybe_warn): Call should_warn_p. Update diagnostic messages
24382 and handle those for all directives, including plain strings.
24383 (format_directive): Handle likely and unlikely counters.
24384 Remove unnecessary quoting from diagnostics. Add an informational
24386 (add_bytes): Remove.
24387 (pass_sprintf_length::compute_format_length): Simplify.
24388 (try_substitute_return_value): Handle likely and unlikely counters.
24390 2017-01-26 Carl Love <cel@us.ibm.com>
24392 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
24393 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24395 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24398 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24399 endianess for subregs into account.
24400 * lra-constraints.c (lra_constraints): Do risky transformations
24401 always on the first iteration.
24402 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24403 last_call_used_reg_set.
24404 (process_bb_lives): Define and use last_call_used_reg_set.
24405 * lra.c (lra): Always continue after lra_constraints on the first
24408 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24410 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24412 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24414 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24416 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24417 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24418 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24419 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24420 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24421 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24422 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24423 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24424 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24426 2017-01-26 Marek Polacek <polacek@redhat.com>
24429 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24430 for the third operand.
24432 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24434 PR middle-end/79236
24435 * omp-low.c (struct omp_context): Add simt_stmt field.
24436 (scan_omp_for): Return omp_context *.
24437 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24438 context to the _simt_ SIMD stmt.
24439 (lower_omp_for): For combined SIMD with sibling _simt_
24440 SIMD, make sure to use the same decls in _looptemp_
24441 clauses as in the sibling.
24443 2017-01-26 David Sherwood <david.sherwood@arm.com>
24445 PR middle-end/79212
24446 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24449 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24452 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24453 emit fld b; fld a; if possible.
24455 * brig-builtins.def: Update copyright years.
24456 * config/arm/arm_acle_builtins.def: Update copyright years.
24458 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24461 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24462 constraint instead of o for the stxsd instruction.
24464 2017-01-25 Carl Love <cel@us.ibm.com>
24466 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24467 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24469 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24471 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24473 2017-01-25 Richard Biener <rguenther@suse.de>
24475 PR tree-optimization/69264
24476 * target.def (vector_alignment_reachable): Improve documentation.
24477 * doc/tm.texi: Regenerate.
24478 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24480 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24481 earlier changes with respect to TYPE_USER_ALIGN.
24482 (vector_alignment_reachable_p): Likewise. Improve dumping.
24484 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24487 * config/arm/arm.md (xordi3): Force constant operand into a register
24490 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24492 * doc/invoke.texi (-fstore-merging): Correct default optimization
24493 levels at which it is enabled.
24494 (-O): Move -fstore-merging from list to...
24497 2017-01-25 Richard Biener <rguenther@suse.de>
24500 * omp-expand.c: Include debug.h.
24501 (expand_omp_taskreg): Make sure to generate early debug before
24502 outlining anything from a function.
24503 (expand_omp_target): Likewise.
24504 (grid_expand_target_grid_body): Likewise.
24506 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24509 * asan.c (get_translation_unit_decl): New function.
24510 (asan_add_global): Extract modules file name from globals
24511 TRANSLATION_UNIT_DECL name.
24513 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24516 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24517 for long calls with APCS frame and VFP.
24519 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24521 * cfg.c (original_copy_tables_initialized_p): New function.
24522 * cfg.h (original_copy_tables_initialized_p): New decl.
24523 * cfgrtl.c (relink_block_chain): Guard the call to
24524 free_original_copy_tables with a call to
24525 original_copy_tables_initialized_p.
24526 * cgraph.h (symtab_node::native_rtl_p): New decl.
24527 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24528 (symtab_node::needed_p): Don't assert for early assembly output
24529 for __RTL functions.
24530 (cgraph_node::finalize_function): Set "force_output" for __RTL
24532 (cgraph_node::analyze): Bail out early for __RTL functions.
24533 (analyze_functions): Update assertion to support __RTL functions.
24534 (cgraph_node::expand): Bail out early for __RTL functions.
24535 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24537 * function.h (struct function): Update comment for field
24539 * gimple-expr.c: Include "tree-pass.h".
24540 (gimple_has_body_p): Return false for __RTL functions.
24541 * Makefile.in (OBJS): Add run-rtl-passes.o.
24542 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24544 (gcc::pass_manager::get_clean_slate): New accessor.
24545 * passes.c: Include "insn-addr.h".
24546 (should_skip_pass_p): Add logging. Update logic for running
24547 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24548 property-provider override so it is only done for gimple passes.
24550 (skip_pass): New function.
24551 (execute_one_pass): Call skip_pass when skipping passes.
24552 * read-md.c (md_reader::read_char): Support filtering
24553 the input to a subset of line numbers.
24554 (md_reader::md_reader): Initialize fields
24555 m_first_line and m_last_line.
24556 (md_reader::read_file_fragment): New function.
24557 * read-md.h (md_reader::read_file_fragment): New decl.
24558 (md_reader::m_first_line): New field.
24559 (md_reader::m_last_line): New field.
24560 * read-rtl-function.c (function_reader::create_function): Only
24561 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24562 curr_properties. Set DECL_INITIAL to a dummy block.
24563 (read_rtl_function_body_from_file_range): New function.
24564 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24566 * run-rtl-passes.c: New file.
24567 * run-rtl-passes.h: New file.
24569 2017-01-24 Jeff Law <law@redhat.com>
24571 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24574 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24576 PR tree-optimization/79159
24577 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24578 (record_nonwrapping_iv): Improve boundary using above function if no
24579 value range information.
24581 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24582 Martin Jambor <mjambor@suse.cz>
24584 * brig-builtins.def: New file.
24585 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24586 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24587 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24588 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24589 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24590 * builtin-types.def (BT_INT8): New.
24591 (BT_INT16): Likewise.
24592 (BT_UINT8): Likewise.
24593 (BT_UINT16): Likewise.
24594 (BT_FN_ULONG): Likewise.
24595 (BT_FN_UINT_INT): Likewise.
24596 (BT_FN_UINT_ULONG): Likewise.
24597 (BT_FN_UINT_LONG): Likewise.
24598 (BT_FN_UINT_PTR): Likewise.
24599 (BT_FN_ULONG_PTR): Likewise.
24600 (BT_FN_INT8_FLOAT): Likewise.
24601 (BT_FN_INT16_FLOAT): Likewise.
24602 (BT_FN_UINT32_FLOAT): Likewise.
24603 (BT_FN_UINT16_FLOAT): Likewise.
24604 (BT_FN_UINT8_FLOAT): Likewise.
24605 (BT_FN_UINT64_FLOAT): Likewise.
24606 (BT_FN_UINT16_UINT32): Likewise.
24607 (BT_FN_UINT32_UINT16): Likewise.
24608 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24609 (BT_FN_INT_PTR_INT): Likewise.
24610 (BT_FN_UINT_PTR_UINT): Likewise.
24611 (BT_FN_LONG_PTR_LONG): Likewise.
24612 (BT_FN_ULONG_PTR_ULONG): Likewise.
24613 (BT_FN_VOID_UINT64_UINT64): Likewise.
24614 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24615 (BT_FN_INT8_INT8_INT8): Likewise.
24616 (BT_FN_INT16_INT16_INT16): Likewise.
24617 (BT_FN_INT_INT_INT): Likewise.
24618 (BT_FN_UINT_FLOAT_UINT): Likewise.
24619 (BT_FN_FLOAT_UINT_UINT): Likewise.
24620 (BT_FN_ULONG_UINT_UINT): Likewise.
24621 (BT_FN_ULONG_UINT_PTR): Likewise.
24622 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24623 (BT_FN_UINT_UINT_UINT): Likewise.
24624 (BT_FN_VOID_UINT_PTR): Likewise.
24625 (BT_FN_UINT_UINT_PTR: Likewise.
24626 (BT_FN_UINT32_UINT64_PTR): Likewise.
24627 (BT_FN_INT_INT_UINT_UINT): Likewise.
24628 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24629 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24630 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24631 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24632 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24633 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24634 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24635 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
24636 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
24637 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
24638 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
24639 * doc/frontends.texi: List BRIG FE.
24640 * doc/install.texi (Testing): Add BRIG tesring requirements.
24641 * doc/invoke.texi (Overall Options): Mention BRIG.
24642 * doc/standards.texi (Standards): Doucment BRIG HSA version.
24644 2017-01-24 Richard Biener <rguenther@suse.de>
24646 PR translation/79208
24647 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
24649 2017-01-24 Martin Jambor <mjambor@suse.cz>
24652 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24653 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24654 and known_contexts.
24656 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24658 PR middle-end/79123
24659 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24660 casts from signed to unsigned really don't have a range.
24662 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24664 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24665 GMP_RNDx for compatiblity.
24667 2017-01-24 Martin Liska <mliska@suse.cz>
24670 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24671 that would prevent us to call alloca with -1 as argument.
24673 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24675 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24676 Avoid -Wformat-security warning.
24678 2017-01-23 Andrew Pinski <apinski@cavium.com>
24680 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24683 2017-01-23 Martin Sebor <msebor@redhat.com>
24685 PR middle-end/78703
24686 * gimple-ssa-sprintf.c (warn_level): New global.
24687 (format_integer): Use it here and throughout the rest of the file.
24688 Use the same switch to compute sign as base.
24689 (maybe_warn): New function.
24690 (format_directive): Factor out warnings into maybe_warn.
24691 Add debugging output. Use warn_level.
24692 (add_bytes): Use warn_level.
24693 (pass_sprintf_length::compute_format_length): Add debugging output.
24694 (try_substitute_return_value): Same.
24695 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24697 PR middle-end/78703
24698 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24699 (struct fmtresult, format_integer, format_floating): Adjust.
24700 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24701 (get_string_length, format_string,format_directive): Same.
24702 (pass_sprintf_length::compute_format_length): Same.
24703 (try_substitute_return_value): Simplify slightly.
24705 PR middle-end/78703
24706 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24707 (fmtresult::operator+=): Outlined.
24708 (struct fmtresult): Add ctors.
24709 (struct conversion_spec): Rename...
24710 (struct directive): ...to this. Add and remove data members.
24711 (directive::set_width, directive::set_precision): New functions.
24712 (format_percent): Use fmtresult ctor.
24713 (get_width_and_precision): Remove.
24714 (format_integer): Make naming changes. Avoid computing width and
24716 (format_floating): Same. Adjust indentation.
24717 (format_character, format_none): New functions.
24718 (format_string): Moved character handling to format_character.
24719 (format_directive): Remove arguments, change return type.
24720 (parse_directive): New function.
24721 (pass_sprintf_length::compute_format_length): Move directive
24722 parsing to parse_directive.
24724 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24726 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24727 (assign_assembler_name_if_needed): ... this.
24728 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24729 (assign_assembler_name_if_needed): ... this.
24730 (free_lang_data_in_cgraph): Adjust callers.
24731 * cgraphunit.c (cgraph_node::analyze): Likewise.
24732 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24735 2017-01-23 Richard Biener <rguenther@suse.de>
24737 PR tree-optimization/79088
24738 PR tree-optimization/79188
24739 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
24740 resetting loop bounds after last path deletion. Reset loop
24741 bounds of the target loop, make code match the comments.
24742 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
24743 Make sure loops need no fixups.
24745 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24747 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
24748 exponent support with double type for first argument.
24749 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
24750 type returned by __builtin_vec_extract_sig,
24751 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
24752 functions from "vector int" to "vector unsigned int" or from
24753 "vector long long int" to "vector unsigned long long int".
24754 Changed type returned by __builtin_vec_extract_exp,
24755 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
24756 functions from "vector int" to "vector unsigned int" or from
24757 "vector long long int" to "vector unsigned long long int".
24758 Changed return type of __builtin_vec_test_data_class,
24759 __builtin_vec_test_data_class_sp, and
24760 __builtin_vec_test_data_class_dp from "vector int" to
24761 "vector bool int" or from "vector long long int" to "vector bool
24762 long long int" and changed second argument type from "unsigned
24763 int" to "int". Added new overloaded function forms "vector float
24764 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
24765 "vector float __builtin_vec_insert_exp_sp (vector float, vector
24766 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
24767 double, vector unsigned long long int)" and "vector double
24768 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
24769 long int)". Changed return type of
24770 __builtin_scalar_test_data_class and
24771 __builtin_scalar_test_data_class_sp and
24772 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
24773 int" and changed second argument from "unsigned int" to "int".
24774 Changed type returned by __builtin_scalar_test_neg,
24775 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
24776 from "int" to "bool int". Added new overloaded function form
24777 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
24778 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
24779 exponent double-precision with floating point first argument.
24780 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
24781 documentation of scalar_test_data_class, scalar_test_neg,
24782 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
24783 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
24784 vec_test_data_class built-in functions to reflect refinements in
24785 their type signatures.
24787 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
24789 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
24791 (aarch64_elf_asm_destructor): Likewise.
24793 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
24795 PR rtl-optimization/78634
24796 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
24797 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
24798 * ifcvt.c (noce_try_cmove): Add missing cost check.
24800 PR rtl-optimization/71724
24801 * combine.c (if_then_else_cond): Look for situations where it is
24802 beneficial to undo the work of one of the recursive calls.
24804 2017-01-23 Bin Cheng <bin.cheng@arm.com>
24806 PR tree-optimization/70754
24807 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
24808 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
24809 combined stmt before it if not NULL.
24810 (combine_chains): Process refs reversely and compute dominance point
24813 2017-01-23 Martin Liska <mliska@suse.cz>
24815 PR tree-optimization/79196
24816 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
24817 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
24819 (strlen_optimize_stmt): Call the renamed function.
24821 2017-01-23 Michael Matz <matz@suse.de>
24823 PR tree-optimization/78384
24824 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
24826 2017-01-23 Richard Biener <rguenther@suse.de>
24828 PR tree-optimization/79186
24829 * tree-vrp.c (register_new_assert_for): Make sure we've seen
24830 both incoming edges before moving an assert.
24832 2017-01-23 Martin Jambor <mjambor@suse.cz>
24834 * ipa-prop.c (load_from_param_1): Removed.
24835 (load_from_unmodified_param): Bits from load_from_param_1 put back
24837 (load_from_param): Removed.
24838 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
24839 with stmt. Reverted back to use of load_from_unmodified_param.
24841 2017-01-23 Martin Jambor <mjambor@suse.cz>
24844 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
24845 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
24846 field a pointer to garbage collected vector, mark lattices and
24847 ipcp_orig_node with GTY((skip)).
24848 (ipa_get_param_count): Adjust to descriptors being a pointer.
24849 (ipa_get_param): Likewise.
24850 (ipa_get_type): Likewise.
24851 (ipa_get_param_move_cost): Likewise.
24852 (ipa_set_param_used): Likewise.
24853 (ipa_get_controlled_uses): Likewise.
24854 (ipa_set_controlled_uses): Likewise.
24855 (ipa_is_param_used): Likewise.
24856 (ipa_node_params_t): Move into garbage collector. New methods insert
24858 (ipa_node_params_sum): Annotate wth GTY(()).
24859 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
24861 (ipa_load_from_parm_agg): Adjust declaration.
24862 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
24863 * ipa-profile.c (ipa_profile): Likewise.
24864 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
24865 (ipa_populate_param_decls): Make descriptors parameter garbage
24867 (ipa_dump_param): Adjust to descriptors being a pointer.
24868 (ipa_alloc_node_params): Likewise.
24869 (ipa_initialize_node_params): Likewise.
24870 (load_from_param_1): Make descriptors parameter garbage collected.
24871 (load_from_unmodified_param): Likewise.
24872 (load_from_param): Likewise.
24873 (ipa_load_from_parm_agg): Likewise.
24874 (ipa_node_params::~ipa_node_params): Removed.
24875 (ipa_free_all_node_params): Remove call to delete operator.
24876 (ipa_node_params_t::insert): New.
24877 (ipa_node_params_t::remove): Likewise.
24878 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
24879 copy known_csts and known_contexts vectors.
24880 (ipa_read_node_info): Adjust to descriptors being a pointer.
24881 (ipcp_modif_dom_walker): Make m_descriptors field garbage
24883 (ipcp_transform_function): Make descriptors variable garbage
24886 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
24888 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
24889 * config/i386/avx512dqintrin.h: Ditto.
24890 * config/i386/avx512fintrin.h: Ditto.
24891 * config/i386/i386.c: Handle new builtins.
24892 * config/i386/i386-builtin.def: Add new builtins.
24893 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
24894 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
24896 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24897 Martin Liska <mliska@suse.cz>
24899 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
24900 * asan.c (asan_expand_poison_ifn): Support stores and use
24901 appropriate ASAN report function.
24902 * internal-fn.c (expand_ASAN_POISON_USE): New function.
24903 * internal-fn.def (ASAN_POISON_USE): Declare.
24904 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
24905 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
24906 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
24907 ASAN_POISON calls w/o LHS.
24908 * tree-ssa.c (execute_update_addresses_taken): Create clobber
24909 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
24910 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
24911 * gimplify.c (asan_poison_variables): Add attribute
24912 use_after_scope_memory to variables that really needs to live
24914 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
24915 having the attribute.
24917 2017-01-23 Martin Liska <mliska@suse.cz>
24919 * asan.c (create_asan_shadow_var): New function.
24920 (asan_expand_poison_ifn): Likewise.
24921 * asan.h (asan_expand_poison_ifn): New declaration.
24922 * internal-fn.c (expand_ASAN_POISON): Likewise.
24923 * internal-fn.def (ASAN_POISON): New builtin.
24924 * sanopt.c (pass_sanopt::execute): Expand
24925 asan_expand_poison_ifn.
24926 * tree-inline.c (copy_decl_for_dup_finish): Make function
24928 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
24929 * tree-ssa.c (is_asan_mark_p): New function.
24930 (execute_update_addresses_taken): Rewrite local variables
24931 (identified just by use-after-scope as addressable) into SSA.
24933 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24935 * doc/install.texi (Specific): opensource.apple.com uses https
24936 now. Remove trailing slash.
24938 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24940 * README.Portability: Remove note on an Irix compatibility issue.
24942 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
24944 * gcov.c (INCLUDE_ALGORITHM): Define.
24945 (INCLUDE_VECTOR): Define.
24946 No longer include <vector> and <algorithm> directly.
24948 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
24950 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
24952 * doc/invoke.texi (Code Gen Options): Ditto.
24954 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
24957 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
24959 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
24961 rtl-optimization/79125
24962 * cprop.c (local_cprop_pass): Handle cases where we make an
24963 unconditional trap.
24965 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
24969 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
24970 read from, for big endian.
24972 2017-01-20 Jiong Wang <jiong.wang@arm.com>
24974 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
24975 register pauth builtins for LP64 only.
24977 2017-01-20 Marek Polacek <polacek@redhat.com>
24980 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24983 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24985 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24987 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24988 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24989 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24991 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24994 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
24995 in big-endian mode when they are not a single duplicated value.
24997 2017-01-20 Richard Biener <rguenther@suse.de>
24999 * BASE-VER: Bump to 7.0.1.
25001 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
25003 * omp-low.c (omplow_simd_context): New struct. Use it...
25004 (lower_rec_simd_input_clauses): ...here and...
25005 (lower_rec_input_clauses): ...here to hold common data. Adjust all
25006 references to idx, lane, max_vf, is_simt.
25008 2017-01-20 Graham Markall <graham.markall@embecosm.com>
25010 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
25013 2017-01-20 Martin Jambor <mjambor@suse.cz>
25015 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
25016 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
25018 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
25019 (GTFILES): Rename hsa.c to hsa-common.c.
25020 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
25021 * hsa-dump.c: Likewise.
25022 * hsa-gen.c: Likewise.
25023 * hsa-regalloc.c: Likewise.
25024 * ipa-hsa.c: Likewise.
25025 * omp-expand.c: Likewise.
25026 * omp-low.c: Likewise.
25027 * toplev.c: Likewise.
25029 2017-01-20 Marek Polacek <polacek@redhat.com>
25032 * doc/invoke.texi: Document -Wduplicated-branches.
25033 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
25034 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
25035 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
25036 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
25037 return 0 only when not OEP_LEXICOGRAPHIC.
25038 (fold_build_cleanup_point_expr): Use the expression
25039 location when building CLEANUP_POINT_EXPR.
25040 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
25041 * tree.c (add_expr): Handle error_mark_node.
25043 2017-01-20 Martin Liska <mliska@suse.cz>
25046 * tree-profile.c (init_ic_make_global_vars): Do not call
25048 (gimple_init_gcov_profiler): Likewise.
25050 2017-01-20 Martin Liska <mliska@suse.cz>
25053 * cgraph.h (maybe_create_reference): Remove argument and
25055 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
25057 * ipa-cp.c (create_specialized_node): Likewise.
25058 * symtab.c (symtab_node::maybe_create_reference): Handle
25059 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
25061 2017-01-20 Martin Liska <mliska@suse.cz>
25063 * read-rtl-function.c (function_reader::create_function): Use
25064 build_decl instread of build_decl_stat.
25066 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
25068 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
25069 * config/i386/avx512dqintrin.h: Ditto.
25070 * config/i386/avx512fintrin.h: Ditto.
25071 * config/i386/i386-builtin-types.def: Add new types.
25072 * config/i386/i386.c: Handle new types.
25073 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
25074 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
25075 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
25076 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
25077 (__builtin_ia32_kshiftridi): New.
25078 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
25080 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
25084 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
25085 define to rs6000_init_stack_protect_guard.
25086 (rs6000_init_stack_protect_guard): New function.
25088 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25089 Yunqiang Su <yunqiang.su@imgtec.com>
25091 * config.gcc (supported_defaults): Add madd4.
25092 (with_madd4): Add validation.
25093 (all_defaults): Add madd4.
25094 * config/mips/mips.opt (mmadd4): New option.
25095 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25097 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25099 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
25100 (ISA_HAS_FUSED_MADD4): Likewise.
25101 * doc/invoke.texi (-mmadd4): Document the new option.
25102 * doc/install.texi (--with-madd4): Document the new option.
25104 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25106 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
25107 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
25108 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
25109 (aarch64_init_pauth_hint_builtins): New.
25110 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
25111 (aarch64_expand_builtin): Expand new builtins.
25113 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25115 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
25116 * combine-stack-adj.c (no_unhandled_cfa): Handle
25117 REG_CFA_TOGGLE_RA_MANGLE.
25118 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
25119 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
25120 info for return address signing.
25121 (aarch64_expand_epilogue): Likewise.
25123 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25125 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
25126 * config/aarch64/aarch64-protos.h
25127 (aarch64_return_address_signing_enabled): New declaration.
25128 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
25130 (aarch64_expand_prologue): Sign return address before it's pushed onto
25132 (aarch64_expand_epilogue): Authenticate return address fetched from
25134 (aarch64_override_options): Sanity check for ILP32 and ISA level.
25135 (aarch64_attributes): New function attributes for "sign-return-address".
25136 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
25137 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
25138 ("*do_return"): Generate combined instructions according to key index.
25139 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
25140 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
25142 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
25143 * config/aarch64/aarch64.opt (msign-return-address=): New.
25144 * doc/extend.texi (AArch64 Function Attributes): Documents
25145 "sign-return-address=".
25146 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
25148 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25150 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
25151 overall option summary.
25153 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25155 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
25156 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
25157 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
25158 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
25160 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
25162 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
25163 -mpower9-minmax by default for -mcpu=power9.
25164 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
25165 128-bit floating point.
25167 2017-01-20 Alan Modra <amodra@gmail.com>
25169 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
25170 optimizing for size.
25172 2017-01-20 Alan Modra <amodra@gmail.com>
25175 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
25176 for strcmp and strncmp from corresponding builtin decl.
25178 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25180 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
25181 instead of i386/rtems-64.h.
25182 * config/i386/rtems-64.h: Remove.
25184 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25188 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
25190 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
25192 2017-01-19 Tamar Christina <tamar.christina@arm.com>
25194 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
25195 Change int to HOST_WIDE_INT.
25196 * config/aarch64/aarch64-protos.h
25197 (aarch64_simd_gen_const_vector_dup): Likewise.
25198 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
25200 2017-01-19 David Malcolm <dmalcolm@redhat.com>
25202 * langhooks-def.h (lhd_type_for_size): New decl.
25203 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
25204 * langhooks.c (lhd_type_for_size): New function, taken from
25207 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
25209 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
25210 define_bypass for CR latency.
25211 (power9-cracked-alu): Update bypass latency and remove power9-branch.
25212 (power9-alu2): Add define_bypass for CR latency.
25214 (power9-mul): Update insn latency.
25215 (power9-mul-compare): Update insn latency, bypass latency and remove
25218 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25220 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
25222 * config/aarch64/aarch64.md
25223 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
25224 aarch64_nopcrelative_literal_loads.
25225 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
25227 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
25229 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
25230 TARGET_LOONGSON_3A.
25231 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
25233 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
25236 * config.gcc (supported_defaults): Add lxc1-sxc1.
25237 (with_lxc1_sxc1): Add validation.
25238 (all_defaults): Add lxc1-sxc1.
25239 * config/mips/mips.opt (mlxc1-sxc1): New option.
25240 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25242 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25243 __mips_no_lxc1_sxc1.
25244 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
25245 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
25246 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
25248 2017-01-19 Richard Biener <rguenther@suse.de>
25250 PR tree-optimization/72488
25251 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
25252 sure to restore SSA info.
25253 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
25255 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
25257 PR rtl-optimization/79121
25258 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
25259 of the inner type when shifting an extended value.
25261 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25264 * symtab.c (symtab_node::equal_address_to): Fix comparing of
25265 interposable aliases.
25267 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
25270 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
25271 Use the evmergelohi instruction.
25272 (mov_si<mode>_e500_subreg4_2_le): Likewise.
25273 (mov_sitf_e500_subreg8_2_be): Likewise.
25274 (mov_sitf_e500_subreg12_2_le): Likewise.
25275 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
25276 (mov_si<mode>_e500_subreg4_2_be): Likewise.
25277 (mov_sitf_e500_subreg8_2_le): Likewise.
25278 (mov_sitf_e500_subreg12_2_be): Likewise.
25280 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25282 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
25283 attribute from vecsimple to vecperm.
25284 (altivec_vbpermq2): Likewise.
25286 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25289 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
25291 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25292 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
25293 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
25294 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
25295 case where N arg is SIZE_MAX.
25296 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
25297 (cmpstrsi): Add pattern.
25299 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25301 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25302 __builtin_vec_revb builtins.
25303 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
25304 built-in functions to support generation of the ISA 3.0 XXBR<x>
25305 vector byte reverse instructions.
25306 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
25307 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
25308 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
25309 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
25310 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
25311 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
25312 (P9V_BUILTIN_VEC_REVB): Likewise.
25313 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
25314 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
25315 (p9_xxbrq_v16qi): Likewise.
25316 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
25317 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
25318 (p9_xxbrh_v8hi): Likewise.
25319 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
25320 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
25321 vec_revb built-in functions.
25323 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
25325 PR rtl-optimization/78952
25326 * config/i386/i386.md (any_extract): New code iterator.
25327 (*insvqi_2): Use any_extract for source operand.
25328 (*insvqi_3): Use any_shiftrt for source operand.
25330 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25332 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
25334 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
25336 2017-01-18 Matthias Klose <doko@ubuntu.com>
25338 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
25340 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25342 * config/rs6000/altivec.h (vec_bperm): Change #define.
25343 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
25344 (altivec_vbpermq2): New define_insn.
25345 (altivec_vbpermd): Likewise.
25346 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
25347 function interface.
25348 (VBPERMD): Likewise.
25349 (VBPERM): New polymorphic function interface.
25350 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
25351 Add entries for P9V_BUILTIN_VEC_VBPERM.
25352 * doc/extend.texi: Add interfaces for vec_bperm.
25354 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25356 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
25357 first letter of error messages.
25358 (s390_resolve_overloaded_builtin): Likewise.
25359 * config/s390/s390.c (s390_expand_builtin): Likewise.
25360 (s390_invalid_arg_for_unprototyped_fn): Likewise.
25361 (s390_valid_target_attribute_inner_p): Likewise.
25362 * config/s390/s390.md ("tabort"): Likewise.
25364 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
25366 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
25367 (ISA_AVOID_DIV_HILO): New macro.
25368 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
25369 (ISA_HAS_DDIV): Likewise.
25371 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25373 * doc/invoke.texi (fabi-version): Correct number of occurrences.
25375 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25377 * doc/invoke.texi (fabi-version): Spelling fix.
25379 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25382 * doc/invoke.texi (fabi-version): Mention mangling fix for
25385 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25388 * doc/invoke.texi (fabi-version): Document discriminator mangling.
25390 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
25393 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
25394 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25396 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25398 (stack_protect_test): Ditto.
25399 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25400 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25402 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25403 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25404 -mstack-protector-guard-offset=.
25405 (RS/6000 and PowerPC Options): Ditto.
25407 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25409 * config/i386/i386.h (MASK_CLASS_P): New define.
25410 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25411 there are no registers from different register sets also when
25412 mask registers are used. Update function comment.
25413 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25414 to (*k/*r) and (*k/*km) alternatives.
25416 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25418 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25419 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25420 (EH_RETURN_HANDLER_RTX): New define.
25421 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25422 Force frame pointer in EH return functions.
25423 (aarch64_expand_epilogue): Add barrier for eh_return.
25424 (aarch64_final_eh_return_addr): Remove.
25425 (aarch64_eh_return_handler_rtx): New function.
25426 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25428 (aarch64_eh_return_handler_rtx): New prototype.
25430 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25432 * config/rs6000/altivec.h (vec_rlmi): New #define.
25433 (vec_vrlnm): Likewise.
25434 (vec_rlnm): Likewise.
25435 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25436 (UNSPEC_VRLNM): Likewise.
25437 (VIlong): New mode iterator.
25438 (altivec_vrl<VI_char>mi): New define_insn.
25439 (altivec_vrl<VI_char>nm): Likewise.
25440 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25442 (VRLDNM): Likewise.
25443 (RLNM): New polymorphic function entry.
25444 (VRLWMI): New monomorphic function entry.
25445 (VRLDMI): Likewise.
25446 (RLMI): New polymorphic function entry.
25447 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25448 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25449 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25452 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25455 * dwarf2out.c (field_byte_offset): Restore the
25456 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25457 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25458 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25461 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25464 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25466 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25469 * dwarf2out.c (add_data_member_location_attribute): For constant
25470 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25471 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25472 DW_AT_byte_size attributes.
25474 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25476 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25477 after forcing to constant memory when the code model is medium.
25479 2017-01-17 Julia Koval <julia.koval@intel.com>
25482 * config/i386/avx512fintrin.h
25483 (_mm512_i32gather_ps): Change __addr type to void const*.
25484 (_mm512_mask_i32gather_ps): Ditto.
25485 (_mm512_i32gather_pd): Ditto.
25486 (_mm512_mask_i32gather_pd): Ditto.
25487 (_mm512_i64gather_ps): Ditto.
25488 (_mm512_mask_i64gather_ps): Ditto.
25489 (_mm512_i64gather_pd): Ditto.
25490 (_mm512_mask_i64gather_pd): Ditto.
25491 (_mm512_i32gather_epi32): Ditto.
25492 (_mm512_mask_i32gather_epi32): Ditto.
25493 (_mm512_i32gather_epi64): Ditto.
25494 (_mm512_mask_i32gather_epi64): Ditto.
25495 (_mm512_i64gather_epi32): Ditto.
25496 (_mm512_mask_i64gather_epi32): Ditto.
25497 (_mm512_i64gather_epi64): Ditto.
25498 (_mm512_mask_i64gather_epi64): Ditto.
25499 (_mm512_i32scatter_ps): Change __addr type to void*.
25500 (_mm512_mask_i32scatter_ps): Ditto.
25501 (_mm512_i32scatter_pd): Ditto.
25502 (_mm512_mask_i32scatter_pd): Ditto.
25503 (_mm512_i64scatter_ps): Ditto.
25504 (_mm512_mask_i64scatter_ps): Ditto.
25505 (_mm512_i64scatter_pd): Ditto.
25506 (_mm512_mask_i64scatter_pd): Ditto.
25507 (_mm512_i32scatter_epi32): Ditto.
25508 (_mm512_mask_i32scatter_epi32): Ditto.
25509 (_mm512_i32scatter_epi64): Ditto.
25510 (_mm512_mask_i32scatter_epi64): Ditto.
25511 (_mm512_i64scatter_epi32): Ditto.
25512 (_mm512_mask_i64scatter_epi32): Ditto.
25513 (_mm512_i64scatter_epi64): Ditto.
25514 (_mm512_mask_i64scatter_epi64): Ditto.
25515 * config/i386/avx512pfintrin.h
25516 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25517 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25518 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25519 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25520 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25521 (_mm512_prefetch_i32scatter_ps): Ditto.
25522 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25523 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25524 (_mm512_prefetch_i64scatter_pd): Ditto.
25525 (_mm512_prefetch_i64scatter_ps): Ditto.
25526 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25527 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25528 * config/i386/avx512vlintrin.h
25529 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25530 (_mm_mmask_i32gather_ps): Ditto.
25531 (_mm256_mmask_i32gather_pd): Ditto.
25532 (_mm_mmask_i32gather_pd): Ditto.
25533 (_mm256_mmask_i64gather_ps): Ditto.
25534 (_mm_mmask_i64gather_ps): Ditto.
25535 (_mm256_mmask_i64gather_pd): Ditto.
25536 (_mm_mmask_i64gather_pd): Ditto.
25537 (_mm256_mmask_i32gather_epi32): Ditto.
25538 (_mm_mmask_i32gather_epi32): Ditto.
25539 (_mm256_mmask_i32gather_epi64): Ditto.
25540 (_mm_mmask_i32gather_epi64): Ditto.
25541 (_mm256_mmask_i64gather_epi32): Ditto.
25542 (_mm_mmask_i64gather_epi32): Ditto.
25543 (_mm256_mmask_i64gather_epi64): Ditto.
25544 (_mm_mmask_i64gather_epi64): Ditto.
25545 (_mm256_i32scatter_ps): Change __addr type to void*.
25546 (_mm256_mask_i32scatter_ps): Ditto.
25547 (_mm_i32scatter_ps): Ditto.
25548 (_mm_mask_i32scatter_ps): Ditto.
25549 (_mm256_i32scatter_pd): Ditto.
25550 (_mm256_mask_i32scatter_pd): Ditto.
25551 (_mm_i32scatter_pd): Ditto.
25552 (_mm_mask_i32scatter_pd): Ditto.
25553 (_mm256_i64scatter_ps): Ditto.
25554 (_mm256_mask_i64scatter_ps): Ditto.
25555 (_mm_i64scatter_ps): Ditto.
25556 (_mm_mask_i64scatter_ps): Ditto.
25557 (_mm256_i64scatter_pd): Ditto.
25558 (_mm256_mask_i64scatter_pd): Ditto.
25559 (_mm_i64scatter_pd): Ditto.
25560 (_mm_mask_i64scatter_pd): Ditto.
25561 (_mm256_i32scatter_epi32): Ditto.
25562 (_mm256_mask_i32scatter_epi32): Ditto.
25563 (_mm_i32scatter_epi32): Ditto.
25564 (_mm_mask_i32scatter_epi32): Ditto.
25565 (_mm256_i32scatter_epi64): Ditto.
25566 (_mm256_mask_i32scatter_epi64): Ditto.
25567 (_mm_i32scatter_epi64): Ditto.
25568 (_mm_mask_i32scatter_epi64): Ditto.
25569 (_mm256_i64scatter_epi32): Ditto.
25570 (_mm256_mask_i64scatter_epi32): Ditto.
25571 (_mm_i64scatter_epi32): Ditto.
25572 (_mm_mask_i64scatter_epi32): Ditto.
25573 (_mm256_i64scatter_epi64): Ditto.
25574 (_mm256_mask_i64scatter_epi64): Ditto.
25575 (_mm_i64scatter_epi64): Ditto.
25576 (_mm_mask_i64scatter_epi64): Ditto.
25577 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25578 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25579 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25580 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25581 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25582 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25583 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25584 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25585 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25586 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25587 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25588 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25589 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25590 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25591 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25592 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25593 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25594 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25595 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25596 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25597 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25598 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25599 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25600 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25601 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25602 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25603 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25604 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25605 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25606 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25607 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25608 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25609 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25610 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25611 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25612 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25613 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25614 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25615 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25616 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25617 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25618 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25619 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25620 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25621 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25622 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25623 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25624 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25625 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25626 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25627 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25628 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25629 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25630 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25631 definitions accordingly.
25633 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25634 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25637 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
25640 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
25643 * ira-conflicts.c (ira_build_conflicts): Update total conflict
25644 hard regs for inner regno.
25646 2017-01-17 Martin Liska <mliska@suse.cz>
25649 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
25650 assumption and add comment.
25652 2017-01-17 Nathan Sidwell <nathan@acm.org>
25654 * ipa-visibility.c (localize_node): New function, broken out of ...
25655 (function_and_variable_visibility): ... here. Call it.
25657 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25659 PR middle-end/77445
25660 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25661 correctly set frequency of oudgoing edge.
25662 (duplicate_thread_path): Fix profile updating.
25664 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25667 * configure.ac: Add GCC_BASE_VER.
25668 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25669 version from BASE-VER file.
25670 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25671 (gcc.o): Depend on $(BASEVER).
25672 * common.opt (dumpfullversion): New option.
25673 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25674 * doc/invoke.texi: Document -dumpfullversion.
25675 * doc/install.texi: Document --with-gcc-major-version-only.
25676 * configure: Regenerated.
25678 2017-01-17 Richard Biener <rguenther@suse.de>
25680 PR tree-optimization/71433
25681 * tree-vrp.c (register_new_assert_for): Merge same asserts
25682 on all incoming edges.
25683 (process_assert_insertions_for): Handle insertions at the
25686 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25688 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25689 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25691 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25694 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25697 2017-01-17 Alan Modra <amodra@gmail.com>
25700 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25701 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25702 symbolic stack limit when pic.
25704 2017-01-16 Martin Sebor <msebor@redhat.com>
25706 PR tree-optimization/78608
25707 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25709 2017-01-16 Jeff Law <law@redhat.com>
25712 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25713 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25714 for several include directories that may be relative to sysroot.
25715 * config/i386/x-mingw32 (gplus_includedir): Define.
25716 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25717 (native_system_includedir): Likewise.
25718 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25719 override if TARGET_SYSTEM_ROOT is defined.
25720 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25722 PR tree-optimization/79090
25723 PR tree-optimization/33562
25724 PR tree-optimization/61912
25725 PR tree-optimization/77485
25726 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25727 and computed trims into the dump file.
25729 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25731 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25733 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25736 * gimplify.c (gimplify_init_constructor): If want_value and
25737 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
25741 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
25742 sequence. Formatting fixes.
25743 (doloop_optimize): Formatting fixes.
25746 * gcc.c (debug_level_greater_than_spec_func): New function.
25747 (static_spec_functions): Add debug-level-gt spec function.
25748 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
25750 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25751 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25752 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
25753 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
25754 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
25755 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
25757 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
25759 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
25760 QImode fixups to general and mask registers only.
25762 2017-01-16 Carl Love <cel@us.ibm.com>
25764 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
25765 for built-in functions
25766 vector signed char vec_nabs (vector signed char)
25767 vector signed short vec_nabs (vector signed short)
25768 vector signed int vec_nabs (vector signed int)
25769 vector signed long long vec_nabs (vector signed long long)
25770 vector float vec_nabs (vector float)
25771 vector double vec_nabs (vector double)
25772 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
25774 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
25775 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
25776 * doc/extend.texi: Update the documentation file for the new built-in
25779 2017-01-16 Martin Sebor <msebor@redhat.com>
25781 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
25784 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25786 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
25787 UNSPEC_VSX__XXSPLTD to require special splat handling.
25789 2017-01-16 David Malcolm <dmalcolm@redhat.com>
25792 * system.h: Poison strndup.
25794 2017-01-16 Alan Modra <amodra@gmail.com>
25797 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
25800 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
25802 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
25804 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
25806 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
25807 call recog here. Assert that INSN_CODE (insn) is non-negative.
25809 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25812 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
25814 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
25815 in the currently scheduled RTL fragment.
25817 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25819 PR rtl-optimization/78751
25820 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
25823 2017-01-14 Jeff Law <law@redhat.com>
25825 PR tree-optimization/79090
25826 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
25827 variable length stores.
25828 (compute_trims): Delete dead assignment to *trim_tail.
25829 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
25832 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
25834 PR rtl-optimization/78626
25835 PR rtl-optimization/78727
25836 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
25837 of a block, and split such blocks after everything else is finished.
25839 2017-01-14 Alan Modra <amodra@gmail.com>
25842 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
25843 target legitimate_combined_insn.
25844 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
25845 (rs6000_legitimate_combined_insn): New function.
25846 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
25848 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
25849 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
25850 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
25852 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
25854 * doc/frontends.texi (G++ and GCC): Remove references to Java.
25856 2017-01-13 Jeff Law <law@redhat.com>
25858 PR tree-optimization/33562
25859 PR tree-optimization/61912
25860 PR tree-optimization/77485
25861 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
25863 (delete_dead_assignment): Likewise.
25864 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
25865 statement to delete_dead_call and delete_dead_assignment.
25867 2017-01-13 David Malcolm <dmalcolm@redhat.com>
25870 * substring-locations.c (format_warning_va): Strengthen case 1 so
25871 that both endpoints of the substring must be within the format
25872 range for just the substring to be printed.
25874 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
25876 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
25877 * config/i386/i386.c (ix86_target_string): Add missing options
25878 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
25879 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
25880 flags_other and ix86_target_other to flags2_other. Display unknown
25882 (ix86_valid_target_attribute_inner_p): Add missing options and
25883 reorder options by implied ISAs, as in ix86_target_string.
25885 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25887 * hash-table.h (hash_table::too_empty_p): New function.
25888 (hash_table::expand): Use it.
25889 (hash_table::traverse): Likewise.
25890 (hash_table::empty_slot): Use sizeof (value_type) instead of
25891 sizeof (PTR) to convert bytes to elements. Shrink the table
25892 if the current size is excessive for the current number of
25895 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25897 * ira-costs.c (record_reg_classes): Break from the inner loop
25898 early once alt_fail is known to be true. Update outer loop
25899 handling accordingly.
25901 2017-01-13 Jeff Law <law@redhat.com>
25903 * tree-ssa-dse.c (decrement_count): New function.
25904 (increment_start_addr, maybe_trim_memstar_call): Likewise.
25905 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
25906 when we know the partially dead statement is a mem* function.
25908 PR tree-optimization/61912
25909 PR tree-optimization/77485
25910 * tree-ssa-dse.c: Include expr.h.
25911 (maybe_trim_constructor_store): New function.
25912 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
25914 PR tree-optimization/33562
25915 PR tree-optimization/61912
25916 PR tree-optimization/77485
25917 * doc/invoke.texi: Document new dse-max-object-size param.
25918 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
25919 * tree-ssa-dse.c: Include params.h.
25920 (dse_store_status): New enum.
25921 (initialize_ao_ref_for_dse): New, partially extracted from
25923 (valid_ao_ref_for_dse, normalize_ref): New.
25924 (setup_live_bytes_from_ref, compute_trims): Likewise.
25925 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
25926 (maybe_trim_partially_dead_store): Likewise.
25927 (maybe_trim_complex_store): Likewise.
25928 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
25929 Track what bytes live from the original store. Return tri-state
25930 for dead, partially dead or live.
25931 (dse_dom_walker): Add constructor, destructor and new private members.
25932 (delete_dead_call, delete_dead_assignment): New extracted from
25934 (dse_optimize_stmt): Make a member of dse_dom_walker.
25935 Use initialize_ao_ref_for_dse.
25937 PR tree-optimization/33562
25938 PR tree-optimization/61912
25939 PR tree-optimization/77485
25940 * sbitmap.h (bitmap_count_bits): Prototype.
25941 (bitmap_clear_range, bitmap_set_range): Likewise.
25942 * sbitmap.c (bitmap_clear_range): New function.
25943 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
25945 2017-01-13 Martin Liska <mliska@suse.cz>
25948 * function.c (set_cfun): Add new argument force.
25949 * function.h (set_cfun): Likewise.
25950 * ipa-inline-transform.c (inline_call): Use the function when
25951 strict alising from is dropped for function we inline to.
25953 2017-01-13 Richard Biener <rguenther@suse.de>
25955 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
25956 for dumping GIMPLE INTEGER_CSTs.
25958 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25960 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
25961 to 201112L since C++17.
25963 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
25966 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
25967 if -fsanitize=kernel-address is present.
25969 2017-01-13 Richard Biener <rguenther@suse.de>
25971 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
25972 as _Literal ( type ) number in case usual suffixes do not
25973 preserve all information.
25975 2017-01-13 Richard Biener <rguenther@suse.de>
25977 PR tree-optimization/77283
25978 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25979 and ssa-iterators.h.
25980 (is_feasible_trace): Implement a cost model based on joiner
25983 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25986 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25987 char or short to __float128/_Float128 directly.
25989 2017-01-12 Martin Sebor <msebor@redhat.com>
25991 to -Wformat-overflow.
25992 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
25993 (min_bytes_remaining): Same.
25994 (get_string_length): Same.
25995 (format_string): Same.
25996 (format_directive): Same.
25998 (pass_sprintf_length::handle_gimple_call): Same.
26000 2017-01-12 Jakub Jelinek <jakub@redhat.com>
26002 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
26003 info.nowrite calls with no lhs that can't throw. Return bool
26004 whether gsi_remove has been called or not.
26005 (pass_sprintf_length::handle_gimple_call): Return bool whether
26006 try_substitute_return_value called gsi_remove. Formatting fix.
26007 (pass_sprintf_length::execute): Don't use gsi_remove if
26008 handle_gimple_call returned true.
26011 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
26012 be removed due to side-effects, don't remove following barrier nor
26013 turn the successor edge into fallthru edge.
26015 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26018 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
26019 element-reversing loads and stores as not swappable.
26021 2017-01-12 Nathan Sidwell <nathan@acm.org>
26022 Nicolai Stange <nicstange@gmail.com>
26024 * combine.c (try_combine): Don't ignore result of overlap checking
26025 loop. Combine overlap & asm check into single loop.
26027 2017-01-12 Richard Biener <rguenther@suse.de>
26029 * tree-pretty-print.c (dump_generic_node): Provide -gimple
26030 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
26032 2017-01-12 Richard Biener <rguenther@suse.de>
26034 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
26035 and TS_TARGET_OPTION directly derive from TS_BASE.
26036 * tree-core.h (tree_optimization_option): Derive from tree_base.
26037 (tree_target_option): Likewise.
26039 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
26041 * config/i386/i386.c (memory_address_length): Increase len
26042 only when rip_relative_addr_p returns false.
26044 2017-01-11 Julia Koval <julia.koval@intel.com>
26046 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
26047 (OPTION_MASK_ISA_SGX_SET): New.
26048 (ix86_handle_option): Handle OPT_msgx.
26049 * config.gcc: Added sgxintrin.h.
26050 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
26051 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
26052 * config/i386/i386.c (ix86_target_string): Add -msgx.
26054 (ix86_option_override_internal): Handle new options.
26055 (ix86_valid_target_attribute_inner_p): Add sgx.
26056 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
26057 * config/i386/i386.opt: Add msgx.
26058 * config/i386/sgxintrin.h: New file.
26059 * config/i386/x86intrin.h: Add sgxintrin.h.
26061 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26064 * fold-const.c (maybe_nonzero_address): Return 1 for function
26066 (tree_single_nonzero_warnv_p): Don't handle function local objects
26070 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
26073 2017-01-11 David Malcolm <dmalcolm@redhat.com>
26076 * opts.c: Include "spellcheck.h"
26077 (struct string_fragment): New struct.
26078 (struct edit_distance_traits<const string_fragment &>): New
26080 (get_closest_sanitizer_option): New function.
26081 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
26083 2017-01-11 Jakub Jelinek <jakub@redhat.com>
26085 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
26087 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
26088 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
26089 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
26090 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
26091 for initial die_offset if dwarf_split_debug_info.
26092 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
26093 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
26095 (output_skeleton_debug_sections): Formatting fix. Use
26096 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
26097 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
26099 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
26101 * config/arm/cortex-a53.md: Add bypasses for
26102 cortex_a53_r2f_cvt.
26103 (cortex_a53_r2f): Only use for transfers.
26104 (cortex_a53_f2r): Likewise.
26105 (cortex_a53_r2f_cvt): Add reservation for conversions.
26106 (cortex_a53_f2r_cvt): Likewise.
26108 2017-01-11 Tamar Christina <tamar.christina@arm.com>
26110 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
26111 to all inlined functions, change static to extern.
26113 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
26116 * config/arm/arm.c (legitimize_pic_address): Handle reference to
26118 (arm_assemble_integer): Likewise.
26120 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
26122 * config.gcc: Use new awk script to check CPU, FPU and architecture
26123 parameters for --with-... options.
26124 * config/arm/parsecpu.awk: New file
26125 * config/arm/arm-cpus.in: New file.
26126 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
26128 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
26130 * config/arm/t-arm: Update dependency rules.
26131 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
26132 of processing .def files.
26133 * config/arm/genopt.sh: Deleted.
26134 * config/arm/gentune.sh: Deleted.
26135 * config/arm/arm-cores.def: Deleted.
26136 * config/arm/arm-arches.def: Deleted.
26137 * config/arm/arm-fpus.def: Deleted.
26138 * config/arm/arm-tune.md: Regenerated.
26139 * config/arm/arm-tables.opt: Regenerated.
26140 * config/arm/arm-cpu.h: New generated file.
26141 * config/arm/arm-cpu-data.h: New generated file.
26142 * config/arm/arm-cpu-cdata.h: New generated file.
26144 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
26147 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
26149 (input_varpool_node): Unpack dynamically_initialized bit.
26151 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
26153 PR rtl-optimization/79032
26154 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
26155 the alignment of the adjusted memory reference against that of MODE,
26156 instead of the alignment of the original memory reference.
26158 2017-01-11 Martin Jambor <mjambor@suse.cz>
26160 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
26162 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
26163 decorated functions.
26165 2017-01-11 Richard Biener <rguenther@suse.de>
26167 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
26168 set range/nonnull info for PHI results. Do not set it on
26169 stmts marked for removal.
26171 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
26173 * expr.c (store_field): In the bitfield case, fetch the return value
26174 from the registers before applying a single big-endian adjustment.
26175 Always do a final load for a BLKmode value not larger than a word.
26177 2017-01-10 David Malcolm <dmalcolm@redhat.com>
26180 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26181 that we correctly handle column numbers greater than
26182 LINE_MAP_MAX_COLUMN_NUMBER.
26184 2017-01-10 Martin Sebor <msebor@redhat.com>
26186 PR middle-end/78245
26187 * gimple-ssa-sprintf.c (get_destination_size): Call
26188 {init,fini}object_sizes.
26189 * tree-object-size.c (addr_object_size): Adjust.
26190 (pass_through_call): Adjust.
26191 (pass_object_sizes::execute): Adjust.
26192 * tree-object-size.h (fini_object_sizes): Declare.
26194 2017-01-10 Martin Sebor <msebor@redhat.com>
26196 PR tree-optimization/78775
26197 * builtins.c (get_size_range): Move...
26198 * calls.c: ...to here.
26199 (alloc_max_size): Accept zero argument.
26200 (operand_signed_p): Remove.
26201 (maybe_warn_alloc_args_overflow): Call get_size_range.
26202 * calls.h (get_size_range): Declare.
26204 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
26206 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
26207 from TI's devices.csv file as of September 2016.
26208 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
26210 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
26212 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
26213 * doc/invoke.texi: Likewise.
26214 * doc/md.texi: Likewise.
26215 * doc/objc.texi: Likewise.
26217 2017-01-10 Joshua Conner <joshconner@google.com>
26219 * config/arm/fuchsia-elf.h: New file.
26220 * config/fuchsia.h: New file.
26221 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
26222 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
26224 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
26226 2016-01-10 Richard Biener <rguenther@suse.de>
26228 PR tree-optimization/79034
26229 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26230 Propagate out degenerate PHIs in the joiner.
26232 2017-01-10 Martin Liska <mliska@suse.cz>
26234 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
26235 (sort_congruence_classes_by_decl_uid): Likewise.
26236 (sort_congruence_class_groups_by_decl_uid): Likewise.
26237 (sem_item_optimizer::merge_classes): Sort class, groups in these
26238 classes and members in the groups by DECL_UID of declarations.
26239 This would make merge operations stable.
26241 2017-01-10 Martin Liska <mliska@suse.cz>
26243 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
26244 usage of m_classes_vec.
26245 (sem_item_optimizer::~sem_item_optimizer): Likewise.
26246 (sem_item_optimizer::get_group_by_hash): Likewise.
26247 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
26248 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
26249 (sem_item_optimizer::verify_classes): Likewise.
26250 (sem_item_optimizer::process_cong_reduction): Likewise.
26251 (sem_item_optimizer::dump_cong_classes): Likewise.
26252 (sem_item_optimizer::merge_classes): Likewise.
26253 * ipa-icf.h (congruence_class_hash): Rename from
26254 congruence_class_group_hash. Remove declaration of m_classes_vec.
26256 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
26258 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
26259 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
26260 * config.gcc: Add avx512vpopcntdqintrin.h.
26261 * config/i386/avx512vpopcntdqintrin.h: New.
26262 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
26263 * config/i386/i386-builtin-types.def: Add new types.
26264 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
26265 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
26266 __builtin_ia32_vpopcountq_v8di_mask): New.
26267 * config/i386/i386-c.c (ix86_target_macros_internal): Define
26268 __AVX512VPOPCNTDQ__.
26269 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
26270 (PTA_AVX512VPOPCNTDQ): Define.
26271 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
26272 TARGET_AVX512VPOPCNTDQ_P): Define.
26273 * config/i386/i386.opt: Add mavx512vpopcntdq.
26274 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
26275 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
26277 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26279 PR middle-end/77484
26280 * predict.def (PRED_CALL): Set to 67.
26282 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
26284 * expr.c (store_field): In the bitfield case, if the value comes from
26285 a function call and is of an aggregate type returned in registers, do
26286 not modify the field mode; extract the value in all cases if the mode
26287 is BLKmode and the size is not larger than a word.
26289 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
26292 * config/i386/cpuid.h: Fix undefined behavior.
26294 2017-01-04 Jeff Law <law@redhat.com>
26296 PR tree-optimization/79007
26297 PR tree-optimization/67955
26298 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
26299 conservative for pt.null when flag_non_call_exceptions is on.
26301 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26303 PR translation/79019
26304 PR translation/79020
26305 * params.def (PARAM_INLINE_MIN_SPEEDUP,
26306 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
26307 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
26309 * config/avr/avr.opt (maccumulate-args): Likewise.
26310 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
26311 * common.opt (freport-bug): Likewise.
26312 * cif-code.def (CIF_FINAL_ERROR): Likewise.
26313 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
26314 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
26315 translatable string.
26316 * config/i386/i386.c (function_value_32): Likewise.
26317 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
26318 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
26320 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
26321 * common/config/msp430/msp430-common.c (msp430_handle_option):
26323 * symtab.c (symtab_node::verify_base): Likewise.
26324 * opts.c (set_debug_level): Likewise.
26325 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
26326 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
26327 missing whitespace to translatable strings.
26328 * config/avr/avr.md (bswapsi2): Fix typo in comment.
26329 * config/sh/superh.h: Likewise.
26330 * config/i386/xopintrin.h: Likewise.
26331 * config/i386/znver1.md: Likewise.
26332 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
26333 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
26334 * double-int.h (struct double_int): Likewise.
26335 * double-int.c (div_and_round_double): Likewise.
26336 * wide-int.cc: Likewise.
26337 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
26338 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
26339 * cfgcleanup.c (crossjumps_occured): Renamed to ...
26340 (crossjumps_occurred): ... this.
26341 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
26344 PR tree-optimization/78899
26345 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
26346 returning bool return struct loop *, NULL for failure and the new
26348 (versionable_outer_loop_p): Don't version outer loop if it has
26349 dont_vectorized bit set.
26350 (tree_if_conversion): When versioning outer loop, ensure
26351 tree_if_conversion is performed also on the inner loop of the
26352 non-vectorizable outer loop copy.
26353 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
26354 LOOP_VECTORIZED in inner loop of the scalar outer loop and
26355 prevent vectorization of it.
26356 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
26357 the outer loop vectorization of the non-scalar version is attempted
26358 before vectorization of the inner loop in scalar version. If
26359 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
26360 vectorization of its inner loop.
26361 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
26362 has 2 inner loops, rename also on edges from bb whose single pred
26363 is outer_loop->header. Fix typo in function comment.
26365 2017-01-09 Martin Sebor <msebor@redhat.com>
26368 * asan.c (asan_emit_stack_protection): Increase local buffer size
26369 to avoid snprintf truncation warning.
26371 2017-01-09 Andrew Pinski <apinski@cavium.com>
26373 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
26374 to reference thunderx2t99 for the tuning structure
26375 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
26377 (thunderx2t99_extra_costs): This.
26378 * config/aarch64/aarch64-tune.md: Regenerate.
26379 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
26380 (vulcan_addrcost_table): This.
26381 (vulcan_regmove_cost): Rename to ...
26382 (thunderx2t99_regmove_cost): This.
26383 (vulcan_vector_cost): Rename to ...
26384 (thunderx2t99_vector_cost): this.
26385 (vulcan_branch_cost): Rename to ...
26386 (thunderx2t99_branch_cost): This.
26387 (vulcan_tunings): Rename to ...
26388 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
26389 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
26391 2017-01-09 Martin Jambor <mjambor@suse.cz>
26395 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26396 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26397 (propagate_vr_accross_jump_function): Use the above function for all
26398 value range computations for pass-through jump functions and type
26399 converasion from explicit value range values.
26400 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26401 parameters from TYPE_ARG_TYPES.
26402 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26403 (ipa_write_node_info): Stream type of the actual argument.
26404 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26406 2017-01-09 Martin Liska <mliska@suse.cz>
26409 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26410 (lookup_compiler): Do not show error message with have_E.
26412 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26414 PR tree-optimization/78938
26415 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26416 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26417 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26418 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26421 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26423 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26426 2017-01-09 Richard Biener <rguenther@suse.de>
26428 PR tree-optimization/78997
26429 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26430 name condition properly.
26432 2017-01-09 Richard Biener <rguenther@suse.de>
26435 * dwarf2out.c (is_cxx): New overload with context.
26436 (is_naming_typedef_decl): Use it.
26438 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26440 * invoke.texi (Option Summary): Correct spacing in option lists
26441 and add line breaks to fix over-long lines.
26443 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26445 PR middle-end/17660
26447 * extend.texi (Common Variable Attributes): Add xref to GCC
26448 Internals manual to explain mode attribute keywords.
26450 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26453 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26454 and Preprocessor Options.
26455 (Options for Linking): Document -pthread here....
26456 (RS/6000 and PowerPC Options): ...not here.
26457 (Solaris 2 Options): ...or here.
26458 * doc/cppopts.texi: Document -pthread.
26460 2017-01-08 Martin Sebor <msebor@redhat.com>
26462 PR middle-end/77708
26463 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26464 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26465 New member functions.
26466 (format_directive): Used them.
26468 (pass_sprintf_length::handle_gimple_call): Same.
26469 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26470 to avoid truncation for any argument.
26471 (extract_affine_mul): Same.
26472 * tree.c (get_file_function_name): Same.
26474 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26476 PR middle-end/77484
26477 * predict.def (PRED_INDIR_CALL): Set to 86.
26479 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26481 PR preprocessor/54124
26482 * doc/cppopts.texi: Reformat -d subtable to list the full name
26483 of the options. Add cross-reference to the docs for the general
26484 compiler -d options.
26485 * doc/invoke.texi (Developer Options): Add cross-reference to the
26486 preprocessor-specific -d option documentation.
26488 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26490 PR preprocessor/13498
26491 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26492 redudant material, and reflect new command-line options.
26493 (System Headers): Likewise.
26495 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26497 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26498 -isystem, and -idirafter. Copy-edit.
26499 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26500 default for -ftrack-macro-expansion. Delete obsolete and
26501 badly-formatted implementation details about -fdebug-cpp output.
26502 * doc/cppwarnopts.texi: Copy-edit.
26504 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26507 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26508 that the transition from a max line width >= 1<<10 to narrower
26509 lines works correctly.
26511 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26513 * doc/options.texi (PerFunction): New.
26514 * opt-functions.awk (switch_flags): Map both Optimization and
26515 PerFunction to CL_OPTIMIZATION.
26516 * opth-gen.awk: Test for PerFunction flag along with
26518 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26519 it only when the latter is present. Skip those that don't in
26520 the hash function generator.
26521 * common.opt (fvar-tracking): Mark as PerFunction instead of
26523 (fvar-tracking-assignments): Likewise.
26524 (fvar-tracking-assignments-toggle): Likewise.
26525 (fvar-tracking-uninit): Likewise.
26527 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26529 PR translation/79018
26530 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26533 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26536 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26537 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26538 TARGET_LONG_JUMP_TABLE_OFFSETS.
26539 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26540 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26541 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26542 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26543 * config/m68k/m68k.md (tablejump expander): Likewise.
26544 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26545 TARGET_LONG_JUMP_TABLE_OFFSETS.
26546 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26547 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26549 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26550 David Holsgrove <david.holsgrove@xilinx.com>
26552 * common/config/microblaze/microblaze-common.c
26553 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26554 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26556 * config/microblaze/microblaze.c (microblaze_must_save_register)
26557 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26559 (microblaze_eh_return): New function.
26560 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26561 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26562 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26563 * config/microblaze/microblaze.md (eh_return): New pattern.
26565 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26567 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26568 GCC_DIAGNOSTIC_STRINGIFY): Define.
26570 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26572 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26574 * config/arm/arm.md (<mcrr>): New.
26576 * config/arm/arm.c (arm_arch5te): New.
26577 (arm_option_override): Set arm_arch5te.
26578 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26580 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26581 (arm_mcrr_qualifiers): ... this. New.
26582 (MRRC_QUALIFIERS): Define to...
26583 (arm_mrrc_qualifiers): ... this. New.
26584 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26586 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26587 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26588 (MRRCI, mrrc, MRRC): New.
26589 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26590 VUNSPEC_MRRC2): New.
26592 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26594 * config/arm/arm.md (<mcr>): New.
26596 * config/arm/arm.c (arm_coproc_builtin_available): Add
26597 support for mcr, mrc, mcr2 and mrc2.
26598 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26599 (arm_mcr_qualifiers): ... this. New.
26600 (MRC_QUALIFIERS): Define to ...
26601 (arm_mrc_qualifiers): ... this. New.
26602 (MCR_QUALIFIERS): Define to ...
26603 (arm_mcr_qualifiers): ... this. New.
26604 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26606 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26607 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26608 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26609 VUNSPEC_MRC2): New.
26611 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26613 * config/arm/arm.md (*ldc): New.
26617 * config/arm/arm.c (arm_coproc_builtin_available): Add
26618 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26619 (arm_coproc_ldc_stc_legitimate_address): New.
26620 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26621 'qualifier_const_pointer'.
26622 (LDC_QUALIFIERS): Define to...
26623 (arm_ldc_qualifiers): ... this. New.
26624 (STC_QUALIFIERS): Define to...
26625 (arm_stc_qualifiers): ... this. New.
26626 * config/arm/arm-protos.h
26627 (arm_coproc_ldc_stc_legitimate_address): New.
26628 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26629 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26630 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26631 stc2, stcl, stc2l): New.
26632 * config/arm/constraints.md (Uz): New.
26633 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26634 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26635 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
26636 VUNSPEC_STC2L): New.
26638 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26640 * config/arm/arm.md (<cdp>): New.
26641 * config/arm/arm.c (neon_const_bounds): Rename this ...
26642 (arm_const_bounds): ... this.
26643 (arm_coproc_builtin_available): New.
26644 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
26645 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
26646 (CDP_QUALIFIERS): Define to...
26647 (arm_cdp_qualifiers): ... this. New.
26649 (arm_expand_builtin_args): Add case for 6 arguments.
26650 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26651 (arm_const_bounds): ... this.
26652 (arm_coproc_builtin_available): New.
26653 * config/arm/arm_acle.h (__arm_cdp): New.
26655 * config/arm/arm_acle_builtins.def (cdp): New.
26657 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26658 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26659 'arm_const_bounds'.
26660 * config/arm/types.md (coproc): New.
26661 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26662 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26663 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26664 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26666 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26668 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26669 (UBINOP_QUALIFIERS): New.
26671 (acle_builtin_data): New. Change comment.
26672 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26673 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26674 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26675 arm_acle_builtins.def.
26676 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26677 (arm_init_acle_builtins): New.
26678 (CRC32_BUILTIN): Remove.
26679 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26680 crc32cb, crc32ch and crc32cw.
26681 (arm_init_crc32_builtins): Remove.
26682 (arm_init_builtins): Use arm_init_acle_builtins rather
26683 than arm_init_crc32_builtins.
26684 (arm_expand_acle_builtin): New.
26685 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26686 * config/arm/arm_acle_builtins.def: New.
26688 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26690 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26691 (arm_builtin_datum): ... this.
26692 (arm_init_neon_builtin): Rename to ...
26693 (arm_init_builtin): ... this. Add a new parameters PREFIX
26694 and USE_SIG_IN_NAME.
26695 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26696 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26697 'arm_builtin_datum'.
26698 (arm_init_vfp_builtins): Likewise.
26699 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26700 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26701 (arm_expand_neon_args): Rename to ...
26702 (arm_expand_builtin_args): ... this. Rename builtin_arg
26703 enum values and differentiate between ARG_BUILTIN_MEMORY
26704 and ARG_BUILTIN_NEON_MEMORY.
26705 (arm_expand_neon_builtin_1): Rename to ...
26706 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26707 values, arm_expand_builtin_args and add bool parameter NEON.
26708 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26709 (arm_expand_vfp_builtin): Likewise.
26710 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26712 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26714 PR middle-end/77484
26715 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26716 * predict.c (tree_estimate_probability_bb): Reverse direction of
26717 polymorphic call predictor.
26719 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26721 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26722 (determine_pass_name_match): ...this new function and...
26723 (should_skip_pass_p): ...this new function.
26725 2017-01-06 Nathan Sidwell <nathan@acm.org>
26727 * ipa-visibility.c (function_and_variable_visibility): Reformat
26728 comments and long lines. Remove extrneous if.
26729 * symtab.c (symtab_node::make_decl_local): Fix code format.
26730 (symtab_node::set_section_for_node): Fix comment typo.
26732 2017-01-06 Martin Liska <mliska@suse.cz>
26735 * lra-constraints.c: Rename invariant to lra_invariant.
26736 * predict.c (set_even_probabilities): Initialize e to NULL.
26738 2017-01-05 Martin Sebor <msebor@redhat.com>
26740 PR tree-optimization/78910
26741 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
26742 (format_integer): Correct off-by-one error in the handling
26743 of precision with negative numbers in signed conversions..
26745 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
26747 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
26749 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26751 PR tree-optimization/71016
26752 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
26753 factor_out_conditional_conversion. Formatting fix.
26754 (factor_out_conditional_conversion): Add cond_stmt argument.
26755 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
26756 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
26759 2017-01-05 David Malcolm <dmalcolm@redhat.com>
26761 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
26762 read-rtl-function.o, and selftest-rtl.o.
26763 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
26764 (selftest::aarch64_test_loading_full_dump): New function.
26765 (selftest::aarch64_run_selftests): New function.
26766 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
26767 selftest::aarch64_run_selftests.
26768 * config/i386/i386.c
26769 (selftest::ix86_test_loading_dump_fragment_1): New function.
26770 (selftest::ix86_test_loading_call_insn): New function.
26771 (selftest::ix86_test_loading_full_dump): New function.
26772 (selftest::ix86_test_loading_unspec): New function.
26773 (selftest::ix86_run_selftests): Call the new functions.
26774 * emit-rtl.c (maybe_set_max_label_num): New function.
26775 * emit-rtl.h (maybe_set_max_label_num): New decl.
26776 * function.c (instantiate_decls): Guard call to
26777 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
26778 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
26780 * gensupport.c (gen_reader::gen_reader): Pass "false"
26781 for new "compact" param of rtx_reader.
26782 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
26783 rather than an empty string for NULL strings.
26784 * read-md.c: Potentially include config.h rather than bconfig.h.
26785 Wrap include of errors.h with #ifdef GENERATOR_FILE.
26786 (have_error): New global, copied from errors.c.
26787 (md_reader::read_name): Rename to...
26788 (md_reader::read_name_1): ...this, adding "out_loc" param,
26789 and converting "missing name or number" to returning false, rather
26791 (md_reader::read_name): Reimplement in terms of read_name_1.
26792 (md_reader::read_name_or_nil): New function.
26793 (md_reader::read_string): Handle "(nil)" by returning NULL.
26794 (md_reader::md_reader): Add new param "compact".
26795 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
26796 (md_reader::read_file): New method.
26797 * read-md.h (md_reader::md_reader): Add new param "compact".
26798 (md_reader::read_file): New method.
26799 (md_reader::is_compact): New accessor.
26800 (md_reader::read_name): Convert return type from void to file_location.
26801 (md_reader::read_name_or_nil): New decl.
26802 (md_reader::read_name_1): New decl.
26803 (md_reader::m_compact): New field.
26804 (noop_reader::noop_reader): Pass "false" for new "compact" param
26806 (rtx_reader::rtx_reader): Add new "compact" param.
26807 (rtx_reader::read_rtx_operand): Make virtual and convert return
26808 type from void to rtx.
26809 (rtx_reader::read_until): New decl.
26810 (rtx_reader::handle_any_trailing_information): New virtual function.
26811 (rtx_reader::postprocess): New virtual function.
26812 (rtx_reader::finalize_string): New virtual function.
26813 (rtx_reader::m_in_call_function_usage): New field.
26814 (rtx_reader::m_reuse_rtx_by_id): New field.
26815 * read-rtl-function.c: New file.
26816 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
26817 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
26818 (selftest::verify_three_block_rtl_cfg): New decl.
26819 * read-rtl-function.h: New file.
26820 * read-rtl.c: Potentially include config.h rather than bconfig.h.
26821 For host, include function.h, memmodel.h, and emit-rtl.h.
26822 (one_time_initialization): New function.
26823 (struct compact_insn_name): New struct.
26824 (compact_insn_names): New array.
26825 (find_code): Handle insn codes in compact dumps.
26826 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
26827 (bind_subst_iter_and_attr): Likewise.
26828 (add_condition_to_string): Likewise.
26829 (add_condition_to_rtx): Likewise.
26830 (apply_attribute_uses): Likewise.
26831 (add_current_iterators): Likewise.
26832 (apply_iterators): Likewise.
26833 (initialize_iterators): Guard usage of apply_subst_iterator with
26834 #ifdef GENERATOR_FILE.
26835 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
26836 (md_reader::read_mapping): Likewise.
26837 (add_define_attr_for_define_subst): Likewise.
26838 (add_define_subst_attr): Likewise.
26839 (read_subst_mapping): Likewise.
26840 (check_code_iterator): Likewise.
26841 (rtx_reader::read_rtx): Likewise. Move one-time initialization
26843 (one_time_initialization): New function.
26844 (rtx_reader::read_until): New method.
26845 (read_flags): New function.
26846 (parse_reg_note_name): New function.
26847 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
26848 Handle reuse_rtx ids.
26849 Wrap iterator lookup within #ifdef GENERATOR_FILE.
26850 Add parsing support for RTL dumps, mirroring the special-cases in
26851 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
26852 values, and calling handle_any_trailing_information.
26853 (rtx_reader::read_rtx_operand): Convert return type from void
26854 to rtx, returning return_rtx. Handle case 'e'. Call
26855 finalize_string on XSTR and XTMPL fields.
26856 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
26857 "(nil)" values were omitted. Call the postprocess vfunc on the
26859 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
26860 class ctor. Initialize m_in_call_function_usage. Call
26861 one_time_initialization.
26862 * rtl-tests.c (selftest::test_uncond_jump): Call
26863 set_new_first_and_last_insn.
26864 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
26865 * selftest-rtl.c: New file.
26866 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
26867 (selftest::get_insn_by_uid): New decl.
26868 * selftest-run-tests.c (selftest::run_tests): Call
26869 read_rtl_function_c_tests.
26870 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
26871 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
26874 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
26876 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
26877 operands in a special way. Assert that pos+len <= mode precision.
26879 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26881 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
26882 3 argument Alias with unlimited for the negative form.
26883 (fno-vect-cost-model): Removed.
26885 2017-01-05 Martin Liska <mliska@suse.cz>
26887 * hsa-gen.c (gen_hsa_divmod): New function.
26888 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
26890 2017-01-05 Martin Liska <mliska@suse.cz>
26893 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
26896 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26898 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
26899 small constant length operands.
26901 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26903 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
26904 between loop iterations.
26906 2017-01-05 Martin Liska <mliska@suse.cz>
26909 * gimplify.c (gimplify_decl_expr): Compare to
26910 asan_poisoned_variables instread of checking flags.
26911 (gimplify_target_expr): Likewise.
26912 (gimplify_expr): Likewise.
26913 (gimplify_function_tree): Conditionally initialize
26914 asan_poisoned_variables.
26916 2017-01-04 Jeff Law <law@redhat.com>
26918 PR tree-optimizatin/78812
26919 * rtl.h (contains_mem_rtx_p): Prototype.
26920 * ifcvt.c (containts_mem_rtx_p): Move from here to...
26921 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
26922 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
26923 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
26924 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
26926 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26928 * input.c (assert_char_at_range): Default-initialize actual_range.
26930 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26932 * df-scan.c (df_ref_create_structure): Make regno unsigned,
26933 to match the caller.
26935 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26937 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
26938 insns after final jump in test to emit dummy move.
26940 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26942 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
26943 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
26945 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26947 * multiple_target.c (create_dispatcher_calls): Init e_next.
26948 * tree-ssa-loop-split.c (split_loop): Init border.
26949 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
26952 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
26957 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
26958 (altivec_register_operand): Do not return true if the operand
26959 contains a SUBREG mixing SImode and SFmode.
26960 (vsx_register_operand): Likewise.
26961 (vsx_reg_sfsubreg_ok): New predicate.
26962 (vfloat_operand): Do not return true if the operand contains a
26963 SUBREG mixing SImode and SFmode.
26964 (vint_operand): Likewise.
26965 (vlogical_operand): Likewise.
26966 (gpc_reg_operand): Likewise.
26967 (int_reg_operand): Likewise.
26968 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
26969 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
26970 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
26972 (rs6000_emit_move_si_sf_subreg): New helper function.
26973 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
26974 fixup SUBREGs involving SImode and SFmode.
26975 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
26976 numbers for the new peephole2 optimization.
26977 (peephole2 for SFmode unions): New peephole2 to optimize cases in
26978 the GLIBC math library that do AND/IOR/XOR operations on single
26979 precision floating point.
26980 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26981 target macros to say whether we need to avoid SUBREGs mixing
26983 (TARGET_ALLOW_SF_SUBREG): Likewise.
26984 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26985 (UNSPEC_SI_FROM_SF): Likewise.
26986 (iorxor): Change spacing.
26987 (and_ior_xor): New iterator for AND, IOR, and XOR.
26988 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26989 (movdi_from_sf_zero_ext): Likewise.
26990 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26991 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26992 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
26993 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
26994 (fms<mode>4): Likewise.
26995 (fnma<mode>4): Likewise.
26996 (fnms<mode>4): Likewise.
26997 (nfma<mode>4): Likewise.
26998 (nfms<mode>4): Likewise.
27000 2017-01-04 Marek Polacek <polacek@redhat.com>
27003 * doc/invoke.texi: Document -Wpointer-compare.
27005 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27007 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
27010 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
27011 descriptions for -gdwarf-5 and emit them as uleb128 instead of
27014 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
27017 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
27018 documentation of the powerpc_popcntb_ok attribute.
27019 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27020 code to issue warning messages if a requested CPU configuration is
27021 not supported by the binary (assembler and loader) toolchain.
27022 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
27023 made to define a built-in function that has been disabled.
27024 (paired_init_builtins): Add assertion to prevent ICE if attempt is
27025 made to define a built-in function that has been disabled.
27026 (altivec_init_builtins): Add comment explaining why definition
27027 of the DST built-in functions is not preceded by an assertion
27028 check. Add assertions to prevent ICE if attempts are made to
27029 define an altivec predicate or an abs* built-in function that has
27031 (htm_init_builtins): Add comment explaining why definition of the
27032 htm built-in functions is not preceded by an assertion check.
27034 2017-01-04 Jeff Law <law@redhat.com>
27036 PR tree-optimizatin/67955
27037 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
27038 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
27039 the points-to solution does not include pt_null. Use DECL_PT_UID
27042 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
27044 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
27045 Use gen_int_mode instead of gen_lopwart for const_int operands.
27047 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27049 PR tree-optimization/71563
27050 * match.pd: Simplify X << Y into X if Y is known to be 0 or
27051 out of range value - has low bits known to be zero.
27053 2017-01-04 Alan Modra <amodra@gmail.com>
27055 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
27056 * configure: Regenerate.
27057 * config.in: Regenerate.
27059 2017-01-04 Jakub Jelinek <jakub@redhat.com>
27062 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
27063 a substring of the message, but strcmp with the whole message. Ifdef
27064 ENABLE_NLS, translate the message first using dgettext.
27066 2017-01-03 Jeff Law <law@redhat.com>
27068 PR tree-optimizatin/78856
27069 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
27070 (mark_threaded_blocks): Remove code to truncate thread paths that
27071 cross multiple loop headers. Instead invalidate the cached loop
27072 iteration information and handle case of a thread path walking
27073 into an irreducible region.
27075 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
27078 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
27079 assertions. Add support for doing the signbit if the IEEE 128-bit
27080 floating point value is in a GPR.
27081 * config/rs6000/rs6000.md (Fsignbit): Delete.
27082 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
27083 Update the length attribute if the value is in a GPR.
27084 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
27085 the sign or zero extension instruction, since the value is always 0/1.
27086 (signbit<mode>2_dm2): Delete using <Fsignbit>.
27089 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
27090 extracting SImode to a GPR register so that we can generate a
27091 store, limit the vector to be in a traditional Altivec register
27092 for the vextuwrx instruction.
27094 2017-01-03 Ian Lance Taylor <iant@google.com>
27096 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
27098 2017-01-03 Martin Sebor <msebor@redhat.com>
27100 PR tree-optimization/78696
27101 * gimple-ssa-sprintf.c (format_floating): Correct handling of
27102 precision. Use MPFR for %f for greater fidelity. Correct handling
27104 (pass_sprintf_length::compute_format_length): Set width and precision
27105 specified by asrerisk to void_node for vararg functions.
27106 (try_substitute_return_value): Adjust dump output.
27108 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
27110 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
27112 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
27114 * doc/invoke.texi (SPARC options): Document -mlra as the default.
27115 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
27116 -mlra/-mno-lra was passed to the compiler.
27118 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
27120 PR rtl-optimization/65618
27121 * emit-rtl.c (try_split): Move initialization of "before" and
27122 "after" to just before the call to emit_insn_after_setloc.
27124 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
27126 * doc/md.texi (Standard Names): Remove reference to Java frontend.
27128 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
27130 * dwarf2out.c (gen_enumeration_type_die): When
27131 -gno-strict-dwarf, add a DW_AT_encoding attribute.
27133 2017-01-03 Jakub Jelinek <jakub@redhat.com>
27135 PR tree-optimization/78965
27136 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
27137 Change first argument from const call_info & to call_info &. For %n
27138 set info.nowrite to false.
27140 PR middle-end/78901
27141 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
27142 possibly throwing calls.
27144 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
27145 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
27146 and fns handling, rather than in a separate case SSA_NAME.
27148 2017-01-02 Jeff Law <law@redhat.com>
27150 * config/darwin-driver.c (darwin_driver_init): Const-correctness
27151 fixes for first_period and second_period variables.
27153 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
27156 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
27157 (*insvqi_1): New insn pattern.
27158 (*insvqi_1_mem_rex64): Ditto.
27159 (*insvqi_2): Ditto.
27160 (*insvqi_3): Rename from *insvqi.
27162 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
27164 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
27166 * doc/cfg.texi (Edges): Remove reference to Java.
27167 (Maintaining the CFG): Ditto.
27169 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27171 PR middle-end/77674
27172 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
27173 transparent aliases.
27175 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27177 PR middle-end/77484
27178 * predict.def (PRED_CALL): Update hitrate.
27179 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
27180 * predict.c (tree_estimate_probability_bb): Split CALL predictor
27181 into direct/indirect/polymorphic variants.
27183 2017-01-01 Jakub Jelinek <jakub@redhat.com>
27185 Update copyright years.
27187 * gcc.c (process_command): Update copyright notice dates.
27188 * gcov-dump.c (print_version): Ditto.
27189 * gcov.c (print_version): Ditto.
27190 * gcov-tool.c (print_version): Ditto.
27191 * gengtype.c (create_file): Ditto.
27192 * doc/cpp.texi: Bump @copying's copyright year.
27193 * doc/cppinternals.texi: Ditto.
27194 * doc/gcc.texi: Ditto.
27195 * doc/gccint.texi: Ditto.
27196 * doc/gcov.texi: Ditto.
27197 * doc/install.texi: Ditto.
27198 * doc/invoke.texi: Ditto.
27200 Copyright (C) 2017 Free Software Foundation, Inc.
27202 Copying and distribution of this file, with or without modification,
27203 are permitted in any medium without royalty provided the copyright
27204 notice and this notice are preserved.