1 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
3 PR tree-optimization/71026
4 * match.pd: Move RDIV patterns from fold-const.c
5 * fold-const.c (distribute_real_division): Removed.
6 (fold_binary_loc): Remove calls to distribute_real_divison.
8 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11 c++1z and gnu++1z as deprecated. Change other references to
12 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
13 Change -Wc++1z-compat to -Wc++17-compat.
14 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15 * dwarf2out.c (highest_c_language): Handle C++17.
16 (gen_compile_unit_die): Likewise.
18 2017-09-15 Jakub Jelinek <jakub@redhat.com>
20 PR rtl-optimization/82192
21 * combine.c (make_extraction): Don't look through non-paradoxical
22 SUBREGs or TRUNCATE if pos + len is or might be bigger than
25 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
26 Alan Hayward <alan.hayward@arm.com>
27 David Sherwood <david.sherwood@arm.com>
29 * target.def (function_arg_offset): New hook.
30 * targhooks.h (default_function_arg_offset): Declare.
31 * targhooks.c (default_function_arg_offset): New function.
32 * function.c (locate_and_pad_parm): Use
33 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
34 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
35 (TARGET_FUNCTION_ARG_OFFSET): ...this.
36 * doc/tm.texi: Regenerate.
37 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
38 * config/spu/spu.c (spu_function_arg_offset): New function.
39 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
40 * system.h (FUNCTION_ARG_OFFSET): Poison.
42 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
43 Alan Hayard <alan.hayward@arm.com>
44 David Sherwood <david.sherwood@arm.com>
46 * target.def (truly_noop_truncation): New hook.
47 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
48 than TRULY_NOOP_TRUNCATION.
49 * hooks.h (hook_bool_uint_uint_true): Declare.
50 * hooks.c (hook_bool_uint_uint_true): New function.
51 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
52 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
53 * doc/tm.texi: Regenerate.
54 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
55 rather than TRULY_NOOP_TRUNCATION in comments.
56 (simplify_comparison): Likewise.
57 (record_truncated_value): Likewise.
58 * expmed.c (extract_bit_field_1): Likewise.
59 (extract_split_bit_field): Likewise.
60 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
61 instead of TRULY_NOOP_TRUNCATION.
62 * function.c (assign_parm_setup_block): Likewise.
63 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
64 * rtlhooks.c: Include target.h.
65 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
66 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
67 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
68 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
69 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
70 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
71 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
72 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
73 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
74 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
75 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
76 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
77 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
78 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
79 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
80 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
81 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
82 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
83 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
84 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
85 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
86 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
87 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
88 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
89 * config/mips/mips.c (mips_truly_noop_truncation): New function.
90 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
91 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
92 rather than TRULY_NOOP_TRUNCATION in comments.
93 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
94 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
95 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
96 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
97 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
98 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
99 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
100 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
101 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
102 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
103 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
104 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
105 rather than TRULY_NOOP_TRUNCATION in comments.
106 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
107 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
108 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
109 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
110 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
111 TRULY_NOOP_TRUNCATION condition.
112 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
113 (TRULY_NOOP_TRUNCATION): Delete.
114 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
115 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
116 * config/spu/spu.c (spu_truly_noop_truncation): New function.
117 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
118 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
119 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
120 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
121 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
122 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
123 rather than TRULY_NOOP_TRUNCATION in comments.
124 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
125 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
126 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
127 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
128 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
129 * system.h (TRULY_NOOP_TRUNCATION): Poison.
131 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
134 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
135 (*cmp_ior): Likewise.
136 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
137 (*ior_scc_scc_cmp): Likewise.
138 (*and_scc_scc): Likewise.
139 (*and_scc_scc_cmp): Likewise.
141 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
142 Alan Hayard <alan.hayward@arm.com>
143 David Sherwood <david.sherwood@arm.com>
145 * target.def (can_change_mode_class): New hook.
146 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
147 (hard_regno_nregs): Likewise.
148 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
149 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
150 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
151 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
152 (LOAD_EXTEND_OP): Update accordingly.
153 * doc/tm.texi: Regenerate.
154 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
155 CANNOT_CHANGE_MODE_CLASS.
156 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
157 (REG_CAN_CHANGE_MODE_P): ...this new macro.
158 * combine.c (simplify_set): Update accordingly.
159 * emit-rtl.c (validate_subreg): Likewise.
160 * recog.c (general_operand): Likewise.
161 * regcprop.c (mode_change_ok): Likewise.
162 * reload1.c (choose_reload_regs): Likewise.
163 (inherit_piecemeal_p): Likewise.
164 * rtlanal.c (simplify_subreg_regno): Likewise.
165 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
166 instead of CANNOT_CHANGE_MODE_CLASS.
167 (reload_cse_simplify_operands): Likewise.
168 * reload.c (push_reload): Use targetm.can_change_mode_class
169 instead of CANNOT_CHANGE_MODE_CLASS.
170 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
171 REG_CANNOT_CHANGE_MODE_P.
172 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
173 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
174 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
175 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
176 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
177 (arm_can_change_mode_class): New function.
178 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
179 than CANNOT_CHANGE_MODE_CLASS in comments.
180 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
181 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
182 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
183 (ix86_can_change_mode_class): ...this new function, inverting the
184 sense of the return value.
185 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
186 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
187 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
188 (ia64_can_change_mode_class): New function.
189 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
190 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
191 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
192 (m32c_can_change_mode_class): ...this new function, inverting the
193 sense of the return value.
194 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
195 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
196 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
197 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
198 (mips_can_change_mode_class): ...this new function, inverting the
199 sense of the return value.
200 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
201 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
202 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
203 (msp430_can_change_mode_class): New function.
204 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
205 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
206 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
207 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
208 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
209 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
210 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
211 (pa_cannot_change_mode_class): Replace with...
212 (pa_can_change_mode_class): ...this new function, inverting the
213 sense of the return value.
214 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
215 than CANNOT_CHANGE_MODE_CLASS in comments.
216 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
217 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
218 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
219 (pdp11_cannot_change_mode_class): Replace with...
220 (pdp11_can_change_mode_class): ...this new function, inverting the
221 sense of the return value.
222 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
223 * config/powerpcspe/powerpcspe-protos.h
224 (rs6000_cannot_change_mode_class_ptr): Delete.
225 * config/powerpcspe/powerpcspe.c
226 (rs6000_cannot_change_mode_class_ptr): Delete.
227 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
228 (rs6000_option_override_internal): Assign to
229 targetm.can_change_mode_class instead of
230 rs6000_cannot_change_mode_class_ptr.
231 (rs6000_cannot_change_mode_class): Replace with...
232 (rs6000_can_change_mode_class): ...this new function, inverting the
233 sense of the return value.
234 (rs6000_debug_cannot_change_mode_class): Replace with...
235 (rs6000_debug_can_change_mode_class): ...this new function.
236 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
237 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
238 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
239 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
240 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
242 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
243 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
244 (rs6000_option_override_internal): Assign to
245 targetm.can_change_mode_class instead of
246 rs6000_cannot_change_mode_class_ptr.
247 (rs6000_cannot_change_mode_class): Replace with...
248 (rs6000_can_change_mode_class): ...this new function, inverting the
249 sense of the return value.
250 (rs6000_debug_cannot_change_mode_class): Replace with...
251 (rs6000_debug_can_change_mode_class): ...this new function.
252 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
253 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
254 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
255 (s390_can_change_mode_class): ...this new function, inverting the
256 sense of the return value.
257 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
258 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
259 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
260 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
261 (sh_cannot_change_mode_class): Replace with...
262 (sh_can_change_mode_class): ...this new function, inverting the
263 sense of the return value.
264 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
265 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
266 (sparc_can_change_mode_class): New function.
267 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
268 * config/spu/spu.c (spu_can_change_mode_class): New function.
269 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
270 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
271 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
272 (visium_can_change_mode_class): New function.
273 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
275 2017-09-15 Richard Biener <rguenther@suse.de>
277 PR tree-optimization/82217
278 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
279 but not undefined case.
281 2017-09-15 Jakub Jelinek <jakub@redhat.com>
284 * postreload.c (reload_cse_simplify_operands): Skip
285 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
287 2017-09-15 Richard Biener <rguenther@suse.de>
289 PR tree-optimization/68823
290 * graphite-scop-detection.c (build_alias_set): If we have a
291 possible dependence check whether we can handle them by just
292 looking at the DRs DR_ACCESS_FNs.
293 (build_scops): If build_alias_set fails, fail the SCOP.
295 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
297 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
298 to support float128 built-in functions that require the ISA 3.0
300 (BU_FLOAT128_3_HW): Likewise.
301 (SQRTF128): Add support for the IEEE 128-bit square root and fma
305 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
306 support for built-in functions that need the ISA 3.0 IEEE 128-bit
307 floating point instructions.
308 (rs6000_invalid_builtin): Likewise.
309 (rs6000_builtin_mask_names): Likewise.
310 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
311 (RS6000_BTM_FLOAT128_HW): Likewise.
312 (RS6000_BTM_COMMON): Likewise.
313 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
315 * doc/extend.texi (RS/6000 built-in functions): Document the
316 IEEE 128-bit floating point square root and fused multiply-add
319 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
321 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
322 reg (r2) isn't in the set of registers defined in the prologue.
324 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
325 Alan Hayward <alan.hayward@arm.com>
326 David Sherwood <david.sherwood@arm.com>
328 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
329 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
330 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
331 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
332 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
334 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
335 max_vectorization_factor.
336 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
338 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
339 Alan Hayward <alan.hayward@arm.com>
340 David Sherwood <david.sherwood@arm.com>
342 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
343 (vect_worthwhile_without_simd_p): Declare.
344 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
345 (vectorizable_reduction): Use it.
346 * tree-vect-stmts.c (vectorizable_shift): Likewise.
347 (vectorizable_operation): Likewise.
349 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
350 Alan Hayward <alan.hayward@arm.com>
351 David Sherwood <david.sherwood@arm.com>
353 * tree-vectorizer.h (vect_get_num_copies): New function.
354 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
355 * tree-vect-loop.c (vectorizable_reduction): Likewise.
356 (vectorizable_induction): Likewise.
357 (vectorizable_live_operation): Likewise.
358 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
359 (vectorizable_bswap): Likewise.
360 (vectorizable_call): Likewise.
361 (vectorizable_conversion): Likewise.
362 (vectorizable_assignment): Likewise.
363 (vectorizable_shift): Likewise.
364 (vectorizable_operation): Likewise.
365 (vectorizable_store): Likewise.
366 (vectorizable_load): Likewise.
367 (vectorizable_condition): Likewise.
368 (vectorizable_comparison): Likewise.
369 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
371 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
372 Alan Hayward <alan.hayward@arm.com>
373 David Sherwood <david.sherwood@arm.com>
375 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
378 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
379 Alan Hayward <alan.hayward@arm.com>
380 David Sherwood <david.sherwood@arm.com>
382 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
383 an inline wrapper that provides a location.
384 (gimple_build_vector): Likewise.
385 * gimple-fold.c (gimple_build_vector_from_val): New function.
386 (gimple_build_vector): Likewise.
387 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
388 functions to build the initial value. Always return a gimple value.
389 (get_initial_defs_for_reduction): Likewise. Only compute
391 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
392 vect_init_vector on the results from get_initial_def(s)_for_reduction.
393 (vectorizable_induction): Use gimple_build_vector rather than
396 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
397 Alan Hayward <alan.hayward@arm.com>
398 David Sherwood <david.sherwood@arm.com>
400 * target.h (vec_perm_indices): New typedef.
401 (auto_vec_perm_indices): Likewise.
402 * optabs-query.h: Include target.h
403 (can_vec_perm_p): Take a vec_perm_indices *.
404 * optabs-query.c (can_vec_perm_p): Likewise.
405 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
406 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
407 * tree-vect-generic.c (lower_vec_perm): Likewise.
408 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
409 (vect_grouped_load_supported): Likewise.
410 (vect_shift_permute_load_chain): Likewise.
411 (vect_permute_store_chain): Use auto_vec_perm_indices.
412 (vect_permute_load_chain): Likewise.
413 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
414 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
415 Update uses of can_vec_perm_p.
416 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
417 mode with a number of elements. Take a vec_perm_indices *.
418 (vect_create_epilog_for_reduction): Update accordingly.
419 Use auto_vec_perm_indices.
420 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
421 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
422 (vect_transform_slp_perm_load): Likewise.
423 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
424 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
425 (vect_gen_perm_mask_checked): Likewise.
426 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
427 (vect_gen_perm_mask_checked): Likewise.
428 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
429 (vectorizable_store): Likewise.
430 (vectorizable_load): Likewise.
431 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
432 (vectorizable_bswap): Likewise.
434 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
435 Alan Hayward <alan.hayward@arm.com>
436 David Sherwood <david.sherwood@arm.com>
438 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
439 * tree.c (build_vector): Likewise.
440 (build_vector_from_ctor): Update accordingly.
441 (build_vector_from_val): Likewise.
442 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
443 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
444 * tree-vect-generic.c (add_rshift): Likewise.
445 (expand_vector_divmod): Likewise.
446 (optimize_vector_constructor): Likewise.
447 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
448 (vect_transform_slp_perm_load): Likewise.
449 (vect_schedule_slp_instance): Likewise.
450 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
451 (vectorizable_call): Likewise.
452 (vect_gen_perm_mask_any): Likewise. Add elements in order.
453 * expmed.c (make_tree): Likewise.
454 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
455 a vector passed to build_vector.
456 (fold_convert_const): Likewise.
457 (exact_inverse): Likewise.
458 (fold_ternary_loc): Likewise.
459 (fold_relational_const): Likewise.
460 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
461 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
462 (const_unop): Likewise. Store the reduction accumulator in a
463 variable rather than an array.
464 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
465 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
466 the new vector, rather than constructing it after the input arrays.
467 (native_interpret_vector): Use auto_vec<tree> when building
468 a vector passed to build_vector. Add elements in order.
469 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
470 auto_vec<tree> when building a vector passed to build_vector.
471 (vect_create_epilog_for_reduction): Likewise.
472 (vectorizable_induction): Likewise.
473 (get_initial_def_for_reduction): Likewise. Fix indentation of
475 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
477 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
478 passed to build_vector.
480 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
481 Alan Hayward <alan.hayward@arm.com>
482 David Sherwood <david.sherwood@arm.com>
484 * tree-core.h (tree_base::u): Add an "nelts" field.
485 (tree_vector): Use VECTOR_CST_NELTS as the length.
486 * tree.c (tree_size): Likewise.
487 (make_vector): Initialize VECTOR_CST_NELTS.
488 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
489 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
490 TYPE_VECTOR_SUBPARTS.
491 * expr.c (const_vector_mask_from_tree): Consistently use "units"
492 as the number of units, setting it from VECTOR_CST_NELTS.
493 (const_vector_from_tree): Likewise.
494 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
495 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
496 (fold_negate_expr_1): Likewise.
497 (fold_convert_const): Likewise.
498 (const_binop): Likewise. Differentiate the number of output and
500 (const_unop): Likewise.
501 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
502 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
503 in cases that did the opposite.
505 2017-09-14 Richard Biener <rguenther@suse.de>
507 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
510 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
512 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
514 2017-09-14 Jakub Jelinek <jakub@redhat.com>
517 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
518 if and where to split a bb, except for splitting before debug insn
519 sequences followed by non-label real insn. Delete debug insns
520 in between basic blocks.
522 * combine.c (make_compound_operation_int): Formatting fixes.
524 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
525 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
526 * config/netbsd.h (LINK_EH_SPEC): Likewise.
527 * config/sol2.h (LINK_EH_SPEC): Likewise.
528 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
529 * config/s390/linux.h (LINK_SPEC): Likewise.
530 * config/freebsd.h (LINK_EH_SPEC): Likewise.
531 * config/openbsd.h (LINK_EH_SPEC): Likewise.
532 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
533 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
534 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
535 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
536 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
537 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
539 2017-09-13 Jakub Jelinek <jakub@redhat.com>
541 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
543 (ENDFILE_LINUX_SPEC): Likewise.
544 (LINK_EH_SPEC): Likewise.
545 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
546 (LINK_OS_LINUX_SPEC32): Likewise.
547 (LINK_OS_LINUX_SPEC64): Likewise.
548 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
549 (LINK_OS_LINUX_SPEC): Likewise.
551 2017-09-13 Martin Liska <mliska@suse.cz>
554 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
555 CASE_HIGH is NULL_TREE.
557 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
558 Alan Hayward <alan.hayward@arm.com>
559 David Sherwood <david.sherwood@arm.com>
561 * target.def (secondary_memory_needed): New hook.
562 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
563 instead of SECONDARY_MEMORY_NEEDED.
564 (secondary_memory_needed_mode): Likewise.
565 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
566 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
567 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
568 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
569 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
570 * doc/tm.texi: Regenerate.
571 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
572 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
573 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
574 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
575 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
576 * config/i386/i386.c (inline_secondary_memory_needed): Put the
577 mode argument first and change the reg_class arguments to reg_class_t.
578 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
579 Make static. Update the call to inline_secondary_memory_needed.
580 (ix86_register_move_cost): Update the call to
581 inline_secondary_memory_needed.
582 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
583 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
585 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
586 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
588 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
589 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
590 * config/mips/mips.c (mips_secondary_memory_needed): Make static
591 and match hook interface. Add comment from mips.h.
592 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
593 * config/mmix/mmix.md (truncdfsf2): Refer to
594 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
596 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
597 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
598 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
599 (pa_secondary_memory_needed): New function.
600 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
601 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
602 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
603 (pdp11_secondary_memory_needed): Make static and match hook interface.
604 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
605 * config/powerpcspe/powerpcspe-protos.h
606 (rs6000_secondary_memory_needed_ptr): Delete.
607 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
609 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
610 (rs6000_option_override_internal): Assign to
611 targetm.secondary_memory_needed rather than
612 rs6000_secondary_memory_needed_ptr.
613 (rs6000_secondary_memory_needed): Match hook interface.
614 (rs6000_debug_secondary_memory_needed): Likewise.
615 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
616 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
617 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
618 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
619 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
620 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
622 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
623 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
624 (rs6000_option_override_internal): Assign to
625 targetm.secondary_memory_needed rather than
626 rs6000_secondary_memory_needed_ptr.
627 (rs6000_secondary_memory_needed): Match hook interface.
628 (rs6000_debug_secondary_memory_needed): Likewise.
629 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
630 * config/s390/s390.c (s390_secondary_memory_needed): New function.
631 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
632 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
633 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
634 (sparc_secondary_memory_needed): New function.
635 * lra-constraints.c (check_and_process_move): Refer to
636 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
638 (curr_insn_transform): Likewise.
639 (process_alt_operands): Use targetm.secondary_memory_needed
640 instead of TARGET_SECONDARY_MEMORY_NEEDED.
641 (check_secondary_memory_needed_p): Likewise.
642 (choose_split_class): Likewise.
643 * reload.c: Unconditionally include code that was previously
644 conditional on SECONDARY_MEMORY_NEEDED.
645 (push_secondary_reload): Use targetm.secondary_memory_needed
646 instead of TARGET_SECONDARY_MEMORY_NEEDED.
647 (push_reload): Likewise.
648 * reload1.c: Unconditionally include code that was previously
649 conditional on SECONDARY_MEMORY_NEEDED.
650 (choose_reload_regs): Use targetm.secondary_memory_needed
651 instead of TARGET_SECONDARY_MEMORY_NEEDED.
652 (gen_reload): Likewise.
653 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
655 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
656 Alan Hayward <alan.hayward@arm.com>
657 David Sherwood <david.sherwood@arm.com>
659 * target.def (secondary_memory_needed_mode): New hook:
660 * targhooks.c (default_secondary_memory_needed_mode): Declare.
661 * targhooks.h (default_secondary_memory_needed_mode): New function.
662 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
663 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
664 * doc/tm.texi: Regenerate.
665 * lra-constraints.c (check_and_process_move): Use
666 targetm.secondary_memory_needed_mode instead of
667 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
668 (curr_insn_transform): Likewise.
669 * reload.c (get_secondary_mem): Likewise.
670 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
671 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
673 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
674 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
675 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
676 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
677 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
679 * config/powerpcspe/powerpcspe-protos.h
680 (rs6000_secondary_memory_needed_mode): Delete.
681 * config/powerpcspe/powerpcspe.c
682 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
683 (rs6000_secondary_memory_needed_mode): Make static.
684 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
685 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
687 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
689 (rs6000_secondary_memory_needed_mode): Make static.
690 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
691 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
692 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
693 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
694 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
696 (sparc_secondary_memory_needed_mode): New function.
697 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
699 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
701 * config/aarch64/constraints.md (Umq): New constraint.
702 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
704 (mov<mode>): Update condition.
706 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
708 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
709 when bitposition is the same.
711 2017-09-13 Richard Biener <rguenther@suse.de>
713 * dwarf2out.c (output_die_symbol): Remove.
714 (output_die): Do not output a DIEs symbol.
716 2017-09-13 Richard Biener <rguenther@suse.de>
719 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
720 default-def to avoid breaking iterator update with the weird
721 interaction with cgraph_update_edges_for_call_stmt_node.
723 2017-09-13 Richard Biener <rguenther@suse.de>
725 * tree-cfg.c (verify_gimple_assign_binary): Add verification
726 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
727 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
728 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
730 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
732 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
733 Disable pc relative literal load irrespective of
734 TARGET_FIX_ERR_A53_84341 for default.
736 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
738 * config/sparc/sparc.c (output_return): Output the source location of
739 the insn in the delay slot, if any.
740 (output_sibcall): Likewise.
742 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
745 * common.opt (-static-pie): New alias.
746 (shared): Negate static-pie.
747 (-no-pie): Update help text.
749 (static-pie): New option.
750 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
752 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
753 (LINK_EH_SPEC): Likewise.
754 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
755 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
756 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
757 * gcc.c (LINK_COMMAND_SPEC): Likewise.
758 (init_gcc_specs): Likewise.
759 (init_spec): Likewise.
760 (display_help): Update help message for -pie.
761 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
764 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
766 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
767 (movdi_aarch64): Likewise.
768 (movti_aarch64): Likewise.
770 2017-09-12 Simon Wright <simon@pushface.org>
773 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
774 calculation of the minor version, always output as 0.
776 2017-09-12 Jakub Jelinek <jakub@redhat.com>
779 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
780 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
781 on it early, rather than manual conversion late. For
782 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
783 instead of performing manual conversion.
785 2017-09-12 Carl Love <cel@us.ibm.com>
787 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
788 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
790 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
791 VMULOSW): Add definitions.
792 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
793 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
794 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
795 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
796 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
798 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
800 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
802 (movti_aarch64): Likewise.
803 (movdf_aarch64): Likewise.
804 (movtf_aarch64): Likewise.
805 (load_pairdi): Likewise.
806 (store_pairdi): Likewise.
807 (load_pairdf): Likewise.
808 (store_pairdf): Likewise.
809 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
810 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
811 (ldr_got_small_<mode>): Likewise.
812 (ldr_got_small_28k_<mode>): Likewise.
813 (ldr_got_tiny): Likewise.
814 * config/aarch64/iterators.md (ldst_sz): New.
815 (ldpstp_sz): Likewise.
816 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
818 (thunderx_load): Split load_8 to load_16.
819 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
821 (thunderx2t99_storepair_basic): Split store_8 to store_16.
822 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
823 (xgene1_store_pair): Split store_8 to store_16.
824 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
825 (falkor_st_0_st_sd): Split store_8 to store_16.
827 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
829 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
830 and store1/2/3/4 to store_4/8/12/16.
831 * config/aarch64/aarch64.md: Update for rename.
832 * config/arm/arm.md: Likewise.: Likewise.
833 * config/arm/arm.c: Likewise.
834 * config/arm/thumb1.md: Likewise.
835 * config/arm/thumb2.md: Likewise.
836 * config/arm/vfp.md: Likewise.
837 * config/arm/arm-generic.md: Likewise.
838 * config/arm/arm1020e.md: Likewise.
839 * config/arm/arm1026ejs.md: Likewise.
840 * config/arm/arm1136jfs.md: Likewise.
841 * config/arm/arm926ejs.md: Likewise.
842 * config/arm/cortex-a15.md: Likewise.
843 * config/arm/cortex-a17.md: Likewise.
844 * config/arm/cortex-a5.md: Likewise.
845 * config/arm/cortex-a53.md: Likewise.
846 * config/arm/cortex-a57.md: Likewise.
847 * config/arm/cortex-a7.md: Likewise.
848 * config/arm/cortex-a8.md: Likewise.
849 * config/arm/cortex-a9.md: Likewise.
850 * config/arm/cortex-m4.md: Likewise.
851 * config/arm/cortex-m7.md: Likewise.
852 * config/arm/cortex-r4.md: Likewise.
853 * config/arm/exynos-m1.md: Likewise.
854 * config/arm/fa526.md: Likewise.
855 * config/arm/fa606te.md: Likewise.
856 * config/arm/fa626te.md: Likewise.
857 * config/arm/fa726te.md: Likewise.
858 * config/arm/fmp626.md: Likewise.
859 * config/arm/iwmmxt.md: Likewise.
860 * config/arm/ldmstm.md: Likewise.
861 * config/arm/marvell-pj4.md: Likewise.
862 * config/arm/xgene1.md: Likewise.
863 * config/aarch64/thunderx.md: Likewise.
864 * config/aarch64/thunderx2t99.md: Likewise.
865 * config/aarch64/falkor.md: Likewise.
867 2017-09-12 Martin Liska <mliska@suse.cz>
869 * attribs.c (private_lookup_attribute): New function.
870 * attribs.h (private_lookup_attribute): Declared here.
871 (lookup_attribute): Called from this place.
873 2017-09-12 Richard Biener <rguenther@suse.de>
875 PR tree-optimization/82157
876 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
877 stmts with side-effects.
879 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
880 Alan Hayward <alan.hayward@arm.com>
881 David Sherwood <david.sherwood@arm.com>
883 * target.def (hard_regno_nregs): New hook.
884 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
885 * targhooks.h (default_hard_regno_nregs): Declare.
886 * targhooks.c (default_hard_regno_nregs): New function.
887 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
888 (TARGET_HARD_REGNO_NREGS): ...this hook.
889 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
890 (CLASS_MAX_NREGS): Likewise.
891 * doc/tm.texi: Regenerate.
892 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
893 instead of HARD_REGNO_NREGS.
894 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
895 HARD_REGNO_NREGS in the comment.
896 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
897 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
898 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
899 Return an unsigned int.
900 (TARGET_HARD_REGNO_NREGS): Redefine.
901 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
902 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
903 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
904 (arc_hard_regno_nregs): New function.
905 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
906 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
907 (arm_hard_regno_nregs): New function.
908 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
909 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
910 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
911 (TARGET_HARD_REGNO_NREGS): Redefine.
912 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
913 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
914 (HARD_REGNO_NREGS): Delete.
915 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
916 (cr16_hard_regno_nregs): New function.
917 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
918 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
919 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
920 (cris_hard_regno_nregs): New function.
921 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
922 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
923 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
924 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
925 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
926 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
927 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
928 (frv_hard_regno_nregs): Make static. Take and return an
930 (frv_class_max_nregs): Remove outdated copy of documentation.
931 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
932 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
933 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
934 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
935 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
936 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
937 (TARGET_HARD_REGNO_NREGS): Redefine.
938 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
939 (CLASS_MAX_NREGS): Update comment.
940 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
941 (ia64_hard_regno_nregs): New function.
942 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
943 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
944 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
945 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
946 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
948 (m32c_hard_regno_nregs): Likewise. Make static.
949 (TARGET_HARD_REGNO_NREGS): Redefine.
950 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
951 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
952 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
953 (m68k_hard_regno_nregs): New function.
954 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
955 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
956 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
957 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
958 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
959 Take and return an unsigned int.
960 (TARGET_HARD_REGNO_NREGS): Redefine.
961 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
962 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
963 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
964 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
965 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
966 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
967 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
968 (msp430_hard_regno_nregs): Make static. Take and return an
970 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
971 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
972 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
973 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
974 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
975 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
976 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
977 (TARGET_HARD_REGNO_NREGS): Redefine.
978 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
979 (PA_HARD_REGNO_NREGS): ...this.
980 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
981 (PA_HARD_REGNO_NREGS): ...this.
982 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
983 (pa_hard_regno_nregs): New function.
984 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
985 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
986 (pdp11_hard_regno_nregs): New function.
987 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
988 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
989 (rs6000_hard_regno_nregs_hook): New function.
990 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
991 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
992 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
993 Take and return an unsigned int. Move earlier in file.
994 (TARGET_HARD_REGNO_NREGS): Redefine.
995 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
996 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
997 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
998 (rl78_hard_regno_nregs): Make static. Take and return an
1000 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
1001 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
1002 (rs6000_hard_regno_nregs_hook): New function.
1003 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
1004 * config/rx/rx.c (rx_hard_regno_nregs): New function.
1005 (TARGET_HARD_REGNO_NREGS): Redefine.
1006 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
1007 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
1008 instead of HARD_REGNO_NREGS.
1009 (s390_hard_regno_nregs): New function.
1010 (s390_hard_regno_mode_ok): Add comment from s390.h.
1011 (TARGET_HARD_REGNO_NREGS): Redefine.
1012 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
1013 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
1014 (sh_hard_regno_nregs): New function.
1015 (sh_pass_in_reg_p): Use it.
1016 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
1017 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
1018 (sparc_hard_regno_nregs): New function.
1019 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
1020 * config/spu/spu.c (spu_hard_regno_nregs): New function.
1021 (spu_function_arg_advance): Use it, supplying a valid register number.
1022 (TARGET_HARD_REGNO_NREGS): Redefine.
1023 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
1024 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
1025 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
1026 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
1027 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
1028 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
1029 (CLASS_MAX_NREGS): Remove copy of old documentation.
1030 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
1031 (visium_hard_regno_nregs): New function.
1032 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
1033 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
1034 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
1035 xtensa_hard_regno_nregs): New function.
1036 * system.h (HARD_REGNO_NREGS): Poison.
1038 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1040 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
1041 hard_regno_nregs instead of HARD_REGNO_NREGS.
1042 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
1043 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
1044 (c6x_expand_epilogue): Likewise.
1045 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
1046 (frv_read_iacc_argument): Likewise.
1047 * config/sh/sh.c: Include regs.h.
1048 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1049 (regs_used): Likewise.
1050 (output_stack_adjust): Likewise.
1051 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
1052 * expmed.c: Include regs.h.
1053 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
1054 * ree.c: Include regs.h.
1055 (combine_reaching_defs): Use hard_regno_nregs instead of
1057 (add_removable_extension): Likewise.
1059 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1061 * regs.h (hard_regno_nregs): Turn into a function.
1062 (end_hard_regno): Update accordingly.
1063 * caller-save.c (setup_save_areas): Likewise.
1064 (save_call_clobbered_regs): Likewise.
1065 (replace_reg_with_saved_mem): Likewise.
1066 (insert_restore): Likewise.
1067 (insert_save): Likewise.
1068 * combine.c (can_change_dest_mode): Likewise.
1069 (move_deaths): Likewise.
1070 (distribute_notes): Likewise.
1071 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
1072 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
1073 (rs6000_split_multireg_move): Likewise.
1074 (rs6000_register_move_cost): Likewise.
1075 (rs6000_memory_move_cost): Likewise.
1076 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
1077 (rs6000_split_multireg_move): Likewise.
1078 (rs6000_register_move_cost): Likewise.
1079 (rs6000_memory_move_cost): Likewise.
1080 * cselib.c (cselib_reset_table): Likewise.
1081 (cselib_lookup_1): Likewise.
1082 * emit-rtl.c (set_mode_and_regno): Likewise.
1083 * function.c (aggregate_value_p): Likewise.
1084 * ira-color.c (setup_profitable_hard_regs): Likewise.
1085 (check_hard_reg_p): Likewise.
1086 (calculate_saved_nregs): Likewise.
1087 (assign_hard_reg): Likewise.
1088 (improve_allocation): Likewise.
1089 (calculate_spill_cost): Likewise.
1090 * ira-emit.c (modify_move_list): Likewise.
1091 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
1092 (ira_hard_reg_in_set_p): Likewise.
1093 * ira.c (setup_reg_mode_hard_regset): Likewise.
1094 (clarify_prohibited_class_mode_regs): Likewise.
1095 (check_allocation): Likewise.
1096 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1097 (lra_setup_reg_renumber): Likewise.
1098 (setup_try_hard_regno_pseudos): Likewise.
1099 (spill_for): Likewise.
1100 (assign_hard_regno): Likewise.
1101 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1102 * lra-constraints.c (in_class_p): Likewise.
1103 (lra_constraint_offset): Likewise.
1104 (simplify_operand_subreg): Likewise.
1105 (lra_constraints): Likewise.
1106 (split_reg): Likewise.
1107 (split_if_necessary): Likewise.
1108 (invariant_p): Likewise.
1109 (inherit_in_ebb): Likewise.
1110 * lra-lives.c (process_bb_lives): Likewise.
1111 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
1112 (get_hard_regs): Likewise.
1113 (do_remat): Likewise.
1114 * lra-spills.c (assign_spill_hard_regs): Likewise.
1115 * mode-switching.c (create_pre_exit): Likewise.
1116 * postreload.c (reload_combine_recognize_pattern): Likewise.
1117 * recog.c (peep2_find_free_register): Likewise.
1118 * regcprop.c (kill_value_regno): Likewise.
1119 (set_value_regno): Likewise.
1120 (copy_value): Likewise.
1121 (maybe_mode_change): Likewise.
1122 (find_oldest_value_reg): Likewise.
1123 (copyprop_hardreg_forward_1): Likewise.
1124 * regrename.c (check_new_reg_p): Likewise.
1125 (regrename_do_replace): Likewise.
1126 * reload.c (push_reload): Likewise.
1127 (combine_reloads): Likewise.
1128 (find_dummy_reload): Likewise.
1129 (operands_match_p): Likewise.
1130 (find_reloads): Likewise.
1131 (find_equiv_reg): Likewise.
1132 (reload_adjust_reg_for_mode): Likewise.
1133 * reload1.c (count_pseudo): Likewise.
1134 (count_spilled_pseudo): Likewise.
1135 (find_reg): Likewise.
1136 (clear_reload_reg_in_use): Likewise.
1137 (free_for_value_p): Likewise.
1138 (allocate_reload_reg): Likewise.
1139 (choose_reload_regs): Likewise.
1140 (reload_adjust_reg_for_temp): Likewise.
1141 (emit_reload_insns): Likewise.
1142 (delete_output_reload): Likewise.
1143 * rtlanal.c (subreg_get_info): Likewise.
1144 * sched-deps.c (sched_analyze_reg): Likewise.
1145 * sel-sched.c (init_regs_for_mode): Likewise.
1146 (mark_unavailable_hard_regs): Likewise.
1147 (choose_best_reg_1): Likewise.
1148 (verify_target_availability): Likewise.
1149 * valtrack.c (dead_debug_insert_temp): Likewise.
1150 * var-tracking.c (track_loc_p): Likewise.
1151 (emit_note_insn_var_location): Likewise.
1152 * varasm.c (make_decl_rtl): Likewise.
1153 * reginfo.c (choose_hard_reg_mode): Likewise.
1154 (init_reg_modes_target): Refer directly to
1155 this_target_regs->x_hard_regno_nregs.
1157 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1159 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
1160 instead of hard_regno_nregs.
1162 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1164 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
1165 end_hard_regno instead of hard_regno_nregs.
1166 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
1167 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
1168 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
1169 * ira-color.c (improve_allocation): Likewise.
1170 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1171 * lra-lives.c (mark_regno_live): Likewise.
1172 (mark_regno_dead): Likewise.
1173 * lra-remat.c (operand_to_remat): Likewise.
1174 * lra.c (collect_non_operand_hard_regs): Likewise.
1175 * postreload.c (reload_combine_note_store): Likewise.
1176 (move2add_valid_value_p): Likewise.
1177 * reload.c (regno_clobbered_p): Likewise.
1179 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1181 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
1183 * config/v850/v850.c (v850_reorg): Likewise.
1184 * reload.c (refers_to_regno_for_reload_p): Likewise.
1185 (find_equiv_reg): Likewise.
1186 * reload1.c (reload_reg_reaches_end_p): Likewise.
1188 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1190 * caller-save.c (add_used_regs): Use REG_NREGS instead of
1192 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
1193 * config/arm/arm.c (output_move_neon): Likewise.
1194 (arm_attr_length_move_neon): Likewise.
1195 (neon_split_vcombine): Likewise.
1196 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
1197 (c6x_mark_reg_written): Likewise.
1198 (c6x_dwarf_register_span): Likewise.
1199 * config/i386/i386.c (ix86_save_reg): Likewise.
1200 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
1201 (rws_access_reg): Likewise.
1202 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
1203 * mode-switching.c (create_pre_exit): Likewise.
1204 * ree.c (combine_reaching_defs): Likewise.
1205 (add_removable_extension): Likewise.
1206 * regcprop.c (find_oldest_value_reg): Likewise.
1207 (copyprop_hardreg_forward_1): Likewise.
1208 * reload.c (reload_inner_reg_of_subreg): Likewise.
1209 (push_reload): Likewise.
1210 (combine_reloads): Likewise.
1211 (find_dummy_reload): Likewise.
1212 (reload_adjust_reg_for_mode): Likewise.
1213 * reload1.c (find_reload_regs): Likewise.
1214 (forget_old_reloads_1): Likewise.
1215 (reload_reg_free_for_value_p): Likewise.
1216 (reload_adjust_reg_for_temp): Likewise.
1217 (emit_reload_insns): Likewise.
1218 (delete_output_reload): Likewise.
1219 * sel-sched.c (choose_best_reg_1): Likewise.
1220 (choose_best_pseudo_reg): Likewise.
1222 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1223 Alan Hayward <alan.hayward@arm.com>
1224 David Sherwood <david.sherwood@arm.com>
1226 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
1227 * target.def (slow_unaligned_access): New hook.
1228 * targhooks.h (default_slow_unaligned_access): Declare.
1229 * targhooks.c (default_slow_unaligned_access): New function.
1230 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
1231 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
1232 * doc/tm.texi: Regenerate.
1233 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
1234 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
1235 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
1237 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
1238 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
1240 (rs6000_slow_unaligned_access): New function.
1241 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1242 (expand_block_compare): Likewise.
1243 (expand_strn_compare): Likewise.
1244 (rs6000_rtx_costs): Likewise.
1245 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
1246 (riscv_slow_unaligned_access): Likewise.
1247 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
1248 (riscv_slow_unaligned_access_p): ...this and make static.
1249 (riscv_option_override): Update accordingly.
1250 (riscv_slow_unaligned_access): New function.
1251 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1252 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
1253 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
1254 (rs6000_slow_unaligned_access): New function.
1255 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
1256 (rs6000_rtx_costs): Likewise.
1257 * config/rs6000/rs6000-string.c (expand_block_compare)
1258 (expand_strn_compare): Use targetm.slow_unaligned_access instead
1259 of SLOW_UNALIGNED_ACCESS.
1260 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
1261 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
1262 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
1263 of SLOW_UNALIGNED_ACCESS.
1264 * expmed.c (simple_mem_bitfield_p): Likewise.
1265 * expr.c (alignment_for_piecewise_move): Likewise.
1266 (emit_group_load_1): Likewise.
1267 (emit_group_store): Likewise.
1268 (copy_blkmode_from_reg): Likewise.
1269 (emit_push_insn): Likewise.
1270 (expand_assignment): Likewise.
1271 (store_field): Likewise.
1272 (expand_expr_real_1): Likewise.
1273 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1274 * lra-constraints.c (simplify_operand_subreg): Likewise.
1275 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1276 * gimple-ssa-store-merging.c: Likewise in block comment at start
1278 * tree-ssa-strlen.c: Include target.h.
1279 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
1280 of SLOW_UNALIGNED_ACCESS.
1281 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
1283 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
1285 PR rtl-optimization/82185
1286 * expmed.c (emit_store_flag_int): Only test tem if it has been
1289 2017-09-12 Richard Biener <rguenther@suse.de>
1292 * match.pd ((FTYPE) N CMP CST): Fix typo.
1294 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
1296 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
1298 (mips_near_type_p): Add 'short_call' attribute as a synonym
1300 * doc/extend.texi (short_call): Document new function attribute.
1302 2017-09-12 Jakub Jelinek <jakub@redhat.com>
1305 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
1306 assertion check that in the condition.
1307 (get_atomic_generic_size): Likewise. Before testing if parameter
1308 has pointer type, if it has array type, call for C++
1309 default_conversion to perform array-to-pointer conversion.
1311 2017-09-12 Richard Biener <rguenther@suse.de>
1313 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
1314 for operations we cannot scalarize.
1316 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
1318 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
1319 vectors heap vectors. Clean up comments.
1320 Make visited_bbs a reference.
1321 (profitable_jump_thread_path): Make GC
1322 vectors heap vectors. Clean up comments.
1324 (convert_and_register_jump_thread_path): Make GC vectors heap
1326 (check_subpath_and_update_thread_path): Same. Clean up comments.
1327 Make visited_bbs a reference.
1328 (handle_phi): Abstract common code to to
1329 register_jump_thread_path_if_profitable.
1330 Rename VAR_BB to DEF_BB.
1332 Make GC vectors heap vectors.
1333 Make visited_bbs a reference.
1334 (handle_assignment): Same.
1335 (register_jump_thread_path_if_profitable): New.
1336 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
1338 Make GC vectors heap vectors. Clean up comments.
1339 Make visited_bbs a reference.
1340 (find_jump_threads_backwards): Make visited_bbs live in the stack.
1341 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
1344 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
1347 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
1348 words of E_DImode object are reachable by xtensa_uimm8x4 access.
1350 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
1352 Revert r251800 and r251799.
1354 2017-09-11 Martin Jambor <mjambor@suse.cz>
1357 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
1358 arguments in advance.
1359 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
1360 use it to find predecessor edges.
1361 (naive_outof_ssa): Collect vector of predecessors.
1363 2017-09-08 Jason Merrill <jason@redhat.com>
1365 PR c++/70029 - ICE with ref-qualifier and -flto
1366 * langhooks.h (struct lang_hooks_for_types): Add
1367 copy_lang_qualifiers.
1368 * attribs.c (build_type_attribute_qual_variant): Use it.
1369 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
1371 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
1372 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
1374 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
1377 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
1378 (*mulsi3_sp64): New instruction.
1379 (mulsi3): New expander.
1381 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
1383 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
1385 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1387 * sancov.c: Include memmodel.h.
1389 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
1392 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
1395 2017-09-07 Carl Love <cel@us.ibm.com>
1397 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
1398 the sldi instruction.
1400 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
1402 * sancov.c: Include tm_p.h.
1404 2017-09-07 Jakub Jelinek <jakub@redhat.com>
1407 * output.h (switch_to_other_text_partition): New declaration.
1408 * varasm.c (switch_to_other_text_partition): New function.
1409 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
1410 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
1411 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
1412 to the other text partition before emitting LCL label and switch back
1413 after emitting the word after it.
1415 2017-09-07 Richard Biener <rguenther@suse.de>
1417 * passes.def (pass_split_crit_edges): Remove instance before PRE.
1418 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
1419 critical edges here, after loop init.
1420 (pass_data_pre): Remove PROP_no_crit_edges flags.
1421 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
1422 for valueization of call args to avoid leaking VN_TOP.
1423 (visit_use): Assert we do not visit default defs.
1424 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
1425 Use error_mark_node to more easily detect leaking VN_TOP.
1426 All default-defs are varying, not VN_TOP. Mark them visited.
1427 (run_scc_vn): Make code match comment.
1429 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1431 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
1432 OPTION_MASK_FLOAT128_KEYWORD.
1433 (POWERPC_MASKS): Likewise.
1434 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
1435 support for the -mfloat128-type option, and make -mfloat128
1436 default on PowerPC Linux systems. Define or undefine
1437 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
1438 Define __float128 to be __ieee128 if IEEE 128-bit support is
1439 enabled, or undefine it.
1440 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
1441 Delete defining __FLOAT128_TYPE__.
1442 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
1443 -mfloat128-type option and make -mfloat128 default on PowerPC
1445 (TARGET_FLOAT128_TYPE): Likewise.
1446 (-mfloat128-type): Likewise.
1447 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1448 Delete the -mfloat128-type option and make -mfloat128 default on
1449 PowerPC Linux systems. Always use __ieee128 to be the keyword for
1450 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
1451 128-bit floating point is enabled. Change tests from using
1452 -mfloat128-type to -mfloat128.
1453 (rs6000_mangle_type): Use the correct mangling for the __float128
1454 type even if normal long double is restricted to 64-bits.
1455 (floatn_mode): Enable the _Float128 type by default on VSX Linux
1457 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
1458 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
1459 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
1461 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
1462 documentation for -mfloat128.
1464 2017-09-06 Olivier Hainque <hainque@adacore.com>
1466 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
1468 2017-09-06 Wish Wu <wishwu007@gmail.com>
1469 Jakub Jelinek <jakub@redhat.com>
1471 * asan.c (initialize_sanitizer_builtins): Add
1472 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
1473 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
1474 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
1475 BT_FN_VOID_UINT64_PTR variables.
1476 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
1477 (BT_FN_VOID_UINT16_UINT16): Likewise.
1478 (BT_FN_VOID_UINT32_UINT32): Likewise.
1479 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
1480 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
1481 (BT_FN_VOID_UINT64_PTR): Likewise.
1482 * common.opt (flag_sanitize_coverage): New variable.
1483 (fsanitize-coverage=trace-pc): Remove.
1484 (fsanitize-coverage=): Add.
1485 * flag-types.h (enum sanitize_coverage_code): New enum.
1486 * fold-const.c (fold_range_test): Disable non-short-circuit
1487 optimization if flag_sanitize_coverage.
1488 (fold_truth_andor): Likewise.
1489 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
1490 * opts.c (COVERAGE_SANITIZER_OPT): Define.
1491 (coverage_sanitizer_opts): New array.
1492 (get_closest_sanitizer_option): Add OPTS argument, handle also
1493 OPT_fsanitize_coverage_.
1494 (parse_sanitizer_options): Adjusted to also handle
1495 OPT_fsanitize_coverage_.
1496 (common_handle_option): Add OPT_fsanitize_coverage_.
1497 * sancov.c (instrument_comparison, instrument_switch): New function.
1498 (sancov_pass): Add trace-cmp support.
1499 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
1500 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
1501 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
1502 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
1503 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
1504 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
1505 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
1506 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
1507 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
1508 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
1510 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1512 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1513 error. Only quit immediately if parsing is complete.
1514 (BEGIN): Initialize fatal_err and parse_done.
1515 (begin fpu, end fpu): Check number of arguments.
1516 (begin arch, end arch): Likewise.
1517 (begin cpu, end cpu): Likewise.
1518 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1519 (optalias): Likewise.
1521 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
1523 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1524 * config/arm/arm-isa.h: Delete. Move definitions to ...
1525 * arm-cpus.in: ... here. Use new feature and fgroup values.
1526 * config/arm/arm.c (arm_option_override): Use lower case for feature
1528 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1529 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1530 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1532 (print_isa_bits_for): New function.
1533 (gen_isa): New function.
1534 (gen_comm_data): Use print_isa_bits_for.
1535 (define feature): New keyword.
1536 (define fgroup): New keyword.
1537 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
1538 (arm-isa.h): Add rule to generate file.
1539 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1540 case for feature bit names.
1542 2017-09-06 Richard Biener <rguenther@suse.de>
1544 * tree-ssa-pre.c (NECESSARY): Remove.
1545 (create_expression_by_pieces): Do not touch pass-local flags.
1546 (insert_into_preds_of_block): Likewise.
1547 (do_pre_regular_insertion): Likewise.
1548 (eliminate_insert): Likewise.
1549 (eliminate_dom_walker::before_dom_children): Likewise.
1550 (fini_eliminate): Do not look at inserted_exprs.
1551 (mark_operand_necessary): Remove.
1552 (remove_dead_inserted_code): Replace with simple work-list
1553 algorithm based on inserted_exprs and SSA uses.
1554 (pass_pre::execute): Re-order fini_eliminate and
1555 remove_dead_inserted_code.
1557 2017-09-06 Olivier Hainque <hainque@adacore.com>
1559 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
1560 for VxWorks 7. Adjust surrounding comments.
1562 2017-09-06 Richard Biener <rguenther@suse.de>
1564 * gimple-ssa-strength-reduction.c
1565 (find_candidates_dom_walker::before_dom_children): Also allow
1568 2017-09-06 Richard Biener <rguenther@suse.de>
1570 PR tree-optimization/82108
1571 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
1572 for gap in the non-permutation SLP case.
1574 2017-09-06 Martin Jambor <mjambor@suse.cz>
1576 PR tree-optimization/82078
1577 * tree-sra.c (sort_and_splice_var_accesses): Move call to
1578 add_access_to_work_queue...
1579 (build_accesses_from_assign): ...here.
1580 (propagate_all_subaccesses): Make sure racc is the group
1581 representative, if there is one.
1583 2017-09-06 Jakub Jelinek <jakub@redhat.com>
1586 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
1589 2017-09-06 Richard Biener <rguenther@suse.de>
1591 * gimple-ssa-strength-reduction.c
1592 (find_candidates_dom_walker::before_doom_children): Use a
1593 type and not a mode check.
1595 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1598 * config/arm/predicates.md (arm_general_adddi_operand): Create new
1600 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
1602 2017-09-05 Jeff Law <law@redhat.com>
1604 PR tree-optimization/64910
1605 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
1606 cases where we have 3 or more operands.
1608 2017-09-05 Jakub Jelinek <jakub@redhat.com>
1611 * omp-low.c (lower_omp_for): Recompute tree invariant if
1612 gimple_omp_for_initial/final is ADDR_EXPR.
1615 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
1616 into gimple val before gimplification fo the COND_EXPR.
1618 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
1620 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
1621 REGION_COPY argument.
1622 (thread_through_all_blocks): Remove unused argument to
1623 duplicate_thread_path.
1625 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1626 Alan Hayward <alan.hayward@arm.com>
1627 David Sherwood <david.sherwood@arm.com>
1629 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
1630 Take a scalar_mode rather than a machine_mode.
1631 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1632 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
1633 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
1634 (aarch64_gen_adjusted_ldpstp): Likewise.
1635 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
1637 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1638 Alan Hayward <alan.hayward@arm.com>
1639 David Sherwood <david.sherwood@arm.com>
1641 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
1642 Take a scalar_int_mode instead of a machine_mode.
1643 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1644 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1645 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1646 (aarch64_simd_attr_length_rglist): Delete.
1647 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
1648 a scalar_int_mode instead of a machine_mode.
1649 (aarch64_add_offset): Likewise.
1650 (aarch64_internal_mov_immediate): Likewise
1651 (aarch64_add_constant_internal): Likewise.
1652 (aarch64_add_constant): Likewise.
1653 (aarch64_movw_imm): Likewise.
1654 (aarch64_rtx_arith_op_extract_p): Likewise.
1655 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
1656 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
1657 Remove assert that the mode isn't a vector.
1658 (aarch64_output_scalar_simd_mov_immediate): Likewise.
1659 (aarch64_expand_mov_immediate): Update calls after above changes.
1660 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
1661 (aarch64_and_bitmask_imm): Check for scalar integer modes.
1662 (aarch64_move_imm): Likewise.
1663 (aarch64_can_const_movi_rtx_p): Likewise.
1664 (aarch64_strip_extend): Likewise.
1665 (aarch64_extr_rtx_p): Likewise.
1666 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
1667 a CONST_INT when the mode parameter is VOIDmode.
1668 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
1670 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1672 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
1673 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1674 (bitwise_type_for_mode): Update accordingly.
1676 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1678 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
1679 * stor-layout.c (mode_for_size_tree): Likewise.
1680 (mode_for_array): Update accordingly.
1681 (layout_decl): Likewise.
1682 (compute_record_mode): Likewise. Only set the mode once.
1684 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1686 * target.def (get_mask_mode): Change return type to opt_mode.
1688 * doc/tm.texi: Regenerate.
1689 * targhooks.h (default_get_mask_mode): Return an opt_mode.
1690 * targhooks.c (default_get_mask_mode): Likewise.
1691 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1692 * optabs-query.c (can_vec_mask_load_store_p): Update use of
1693 targetm.get_mask_mode.
1694 * tree.c (build_truth_vector_type): Likewise.
1696 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1698 * machmode.h (mode_for_vector): Return an opt_mode.
1699 * stor-layout.c (mode_for_vector): Likewise.
1700 (mode_for_int_vector): Update accordingly.
1701 (layout_type): Likewise.
1702 * config/i386/i386.c (emit_memmov): Likewise.
1703 (ix86_expand_set_or_movmem): Likewise.
1704 (ix86_expand_vector_init): Likewise.
1705 (ix86_get_mask_mode): Likewise.
1706 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
1708 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
1709 * expmed.c (extract_bit_field_1): Likewise.
1710 * expr.c (expand_expr_real_2): Likewise.
1711 * optabs-query.c (can_vec_perm_p): Likewise.
1712 (can_vec_mask_load_store_p): Likewise.
1713 * optabs.c (expand_vec_perm): Likewise.
1714 * targhooks.c (default_get_mask_mode): Likewise.
1715 * tree-vect-stmts.c (vectorizable_store): Likewise.
1716 (vectorizable_load): Likewise.
1717 (get_vectype_for_scalar_type_and_size): Likewise.
1719 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1721 * machmode.h (mode_for_int_vector): New function.
1722 * stor-layout.c (mode_for_int_vector): Likewise.
1723 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
1724 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
1725 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1726 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1727 (s390_expand_vcond): Likewise.
1729 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1731 * machmode.h (opt_machine_mode): New type.
1732 (opt_mode<T>): Allow construction from anything that can be
1734 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
1735 (mode_for_size): Return an opt_machine_mode.
1736 * stor-layout.c (mode_for_size): Likewise.
1737 (mode_for_size_tree): Update call accordingly.
1738 (bitwise_mode_for_mode): Likewise.
1739 (make_fract_type): Likewise.
1740 (make_accum_type): Likewise.
1741 * caller-save.c (replace_reg_with_saved_mem): Update call
1743 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1744 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1745 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1746 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
1747 * expmed.c (extract_bit_field_1): Likewise.
1748 * reload.c (get_secondary_mem): Likewise.
1749 * varasm.c (assemble_integer): Likewise.
1750 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
1753 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1755 * machmode.h (decimal_float_mode_for_size): New function.
1756 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
1757 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
1758 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
1759 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
1760 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
1761 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
1763 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1765 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
1766 (get_builtin_sync_mode): Likewise.
1767 (expand_ifn_atomic_compare_exchange): Likewise.
1768 (expand_builtin_atomic_clear): Likewise.
1769 (expand_builtin_atomic_test_and_set): Likewise.
1770 (fold_builtin_atomic_always_lock_free): Likewise.
1771 * calls.c (compute_argument_addresses): Likewise.
1772 (emit_library_call_value_1): Likewise.
1773 (store_one_arg): Likewise.
1774 * combine.c (combine_instructions): Likewise.
1775 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
1776 * config/arm/arm.c (arm_function_value): Likewise.
1777 (aapcs_allocate_return_reg): Likewise.
1778 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
1779 * config/i386/i386.c (construct_container): Likewise.
1780 (ix86_gimplify_va_arg): Likewise.
1781 (ix86_expand_sse_cmp): Likewise.
1782 (emit_memmov): Likewise.
1783 (emit_memset): Likewise.
1784 (expand_small_movmem_or_setmem): Likewise.
1785 (ix86_expand_pextr): Likewise.
1786 (ix86_expand_pinsr): Likewise.
1787 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
1788 * config/microblaze/microblaze.c (microblaze_block_move_straight):
1790 * config/mips/mips.c (mips_function_value_1) Likewise.
1791 (mips_block_move_straight): Likewise.
1792 (mips_expand_ins_as_unaligned_store): Likewise.
1793 * config/powerpcspe/powerpcspe.c
1794 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1795 (rs6000_darwin64_record_arg_flush): Likewise.
1796 * config/rs6000/rs6000.c
1797 (rs6000_darwin64_record_arg_advance_flush): Likewise.
1798 (rs6000_darwin64_record_arg_flush): Likewise.
1799 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
1800 (sparc_function_value_1): Likewise.
1801 * config/spu/spu.c (adjust_operand): Likewise.
1802 (spu_emit_branch_or_set): Likewise.
1803 (arith_immediate_p): Likewise.
1804 * emit-rtl.c (gen_lowpart_common): Likewise.
1805 * expr.c (expand_expr_real_1): Likewise.
1806 * function.c (assign_parm_setup_block): Likewise.
1807 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1808 * reload1.c (alter_reg): Likewise.
1809 * stor-layout.c (mode_for_vector): Likewise.
1810 (layout_type): Likewise.
1812 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
1814 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
1815 (spu_convert_move): Likewise.
1816 * lower-subreg.c (resolve_simple_move): Likewise.
1818 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1821 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
1822 define_insn to a define_expand.
1823 (altivec_vsum2sws_direct): New define_insn.
1824 (altivec_vsumsws): Convert from a define_insn to a define_expand.
1826 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1828 * config/arm/arm.c (arm_option_params_internal): Improve setting of
1831 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
1836 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
1837 realign stack if stack alignment needed is less than incoming
1840 2017-09-05 Marek Polacek <polacek@redhat.com>
1843 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
1846 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
1848 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
1850 2017-09-05 Richard Biener <rguenther@suse.de>
1852 PR tree-optimization/82084
1853 * fold-const.c (can_native_encode_string_p): Handle wide characters.
1855 2017-09-05 Richard Biener <rguenther@suse.de>
1857 PR tree-optimization/82102
1858 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
1860 2017-09-05 Martin Liska <mliska@suse.cz>
1862 PR tree-optimization/82032
1863 * tree-cfg.c (generate_range_test): New function.
1864 * tree-cfg.h (generate_range_test): Declared here.
1865 * tree-cfgcleanup.c (convert_single_case_switch): New function.
1866 (cleanup_control_expr_graph): Use it.
1867 * tree-switch-conversion.c (try_switch_expansion): Remove
1869 (emit_case_nodes): Use generate_range_test.
1871 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1874 * config/i386/i386.md (*<btsc><mode>_mask): Add
1875 TARGET_USE_BT to insn constraint.
1876 (*btr<mode>_mask): Ditto.
1878 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
1880 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
1881 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
1883 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1886 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
1887 TARGET_NEON and TARGET_IWMMXT.
1888 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
1889 TARGET_NEON and TARGET_IWMMXT.
1890 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
1892 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
1894 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
1895 (ix86_rewrite_tls_address): Ditto.
1896 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
1897 (ix86_rewrite_tls_address_1): Ditto.
1898 (ix86_rewrite_tls_address): Ditto.
1899 * config/i386/predicates.md (tls_address_pattern): New predicate.
1900 * config/i386/i386.md (TLS address splitter): New splitter.
1902 2017-09-04 Richard Biener <rguenther@suse.de>
1904 PR tree-optimization/82084
1905 * fold-const.h (can_native_encode_string_p): Declare.
1906 * fold-const.c (can_native_encode_string_p): Factor out from ...
1907 (native_encode_string): ... here.
1908 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
1909 vectorizing stores from constants we later cannot handle.
1911 2017-09-04 Marek Polacek <polacek@redhat.com>
1914 * doc/invoke.texi: Update -Wtautological-compare documentation.
1916 2017-09-04 Jeff Law <law@redhat.com>
1918 PR tree-optimization/64910
1919 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
1920 swap the first and last operand if the last is a constant.
1922 2017-09-04 Marek Polacek <polacek@redhat.com>
1925 * convert.c (do_narrow): When sanitizing signed integer overflows,
1926 bail out for signed types.
1927 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
1929 2017-09-04 Richard Biener <rguenther@suse.de>
1931 PR tree-optimization/82060
1932 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
1933 Move devirtualization after stmt folding and before EH/AB/noreturn
1934 cleanup to get the stmt refs canonicalized. Use a bool instead
1935 of gimple_modified_p since that doesn't work for NOPs. Schedule
1936 NOPs generated by folding for removal.
1938 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1939 Alan Hayward <alan.hayward@arm.com>
1940 David Sherwood <david.sherwood@arm.com>
1942 * coretypes.h (pad_direction): New enum.
1943 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
1944 (FUNCTION_ARG_PADDING): Likewise.
1945 * target.def (function_arg_padding): New hook.
1946 * targhooks.h (default_function_arg_padding): Declare.
1947 * targhooks.c (default_function_arg_padding): New function.
1948 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
1949 (TARGET_FUNCTION_ARG_PADDING): ...this.
1950 * doc/tm.texi: Regenerate.
1951 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
1952 instead of direction.
1953 (compute_argument_addresses): Likewise.
1954 (load_register_parameters): Likewise.
1955 (emit_library_call_value_1): Likewise.
1956 (store_one_arg): Use targetm.calls.function_arg_padding instead
1957 of FUNCTION_ARG_PADDING.
1958 (must_pass_in_stack_var_size_or_pad): Likewise.
1959 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
1960 (emit_group_store): Likewise.
1961 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
1962 instead of FUNCTION_ARG_PADDING.
1963 (emit_push_insn): Likewise, and propagate enum change throughout
1965 * function.h (direction): Delete.
1966 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
1968 * function.c (assign_parm_find_stack_rtl): Likewise.
1969 (assign_parm_setup_block_p): Likewise.
1970 (assign_parm_setup_block): Likewise.
1971 (gimplify_parameters): Likewise.
1972 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
1973 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
1975 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
1976 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1977 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
1978 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
1979 (aarch64_function_arg_padding): ...this new function.
1980 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
1981 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1982 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
1983 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
1984 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
1985 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1986 (arm_pad_arg_upward): Replace with...
1987 (arm_function_arg_padding): ...this new function.
1988 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
1990 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
1991 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
1992 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1993 (ia64_hpux_function_arg_padding): Replace with...
1994 (ia64_function_arg_padding): ...this new function. Use pad_direction
1995 instead of direction. Check for TARGET_HPUX.
1996 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
1997 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1998 (iq2000_function_arg_padding): New function.
1999 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
2000 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
2001 (mips_function_arg_padding): ...this new function.
2002 (mips_pad_reg_upward): Update accordingly.
2003 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2004 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
2005 targetm.calls.function_arg_padding.
2006 (FUNCTION_ARG_PADDING): Delete.
2007 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
2008 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
2009 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2010 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
2011 (nios2_block_reg_padding): Return pad_direction instead of direction.
2012 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
2013 instead of direction.
2014 (nios2_function_arg_padding): Likewise. Make static.
2015 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2016 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
2017 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
2018 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
2019 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2020 (pa_function_arg_padding): Make static. Return pad_direction instead
2022 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
2023 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2024 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
2025 instead of direction. Use targetm.calls.function_arg_padding.
2026 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
2027 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2028 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
2029 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
2030 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
2032 (function_arg_padding): Rename to...
2033 (rs6000_function_arg_padding): ...this. Make static. Return
2034 pad_direction instead of direction.
2035 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2036 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
2037 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
2038 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
2039 instead of direction. Use targetm.calls.function_arg_padding.
2040 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
2041 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
2042 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
2043 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
2044 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2045 (function_arg_padding): Rename to...
2046 (rs6000_function_arg_padding): ...this. Make static. Return
2047 pad_direction instead of direction.
2048 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
2049 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
2050 * config/s390/s390.c (s390_function_arg_padding): New function.
2051 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2052 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
2053 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
2054 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
2055 (function_arg_padding): Rename to...
2056 (sparc_function_arg_padding): ...this. Make static. Return
2057 pad_direction instead of direction.
2058 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
2059 * config/spu/spu.c (spu_function_arg_padding): New function.
2060 (TARGET_FUNCTION_ARG_PADDING): Redefine.
2061 * system.h (FUNCTION_ARG_PADDING): Poison.
2063 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2064 Alan Hayward <alan.hayward@arm.com>
2065 David Sherwood <david.sherwood@arm.com>
2067 * target.def (modes_tieable_p): New hook.
2068 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
2069 (TARGET_MODES_TIEABLE_P): ...this.
2070 * doc/tm.texi.in: Regenerate.
2071 * hooks.h (hook_bool_mode_mode_true): Declare.
2072 * hooks.c (hook_bool_mode_mode_true): New function.
2073 * combine.c (subst): Use targetm.modes_tieable_p instead of
2075 * dse.c (find_shift_sequence): Likewise.
2076 * expmed.c (extract_low_bits): Likewise.
2077 * lower-subreg.c: Include target.h.
2078 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
2080 * rtlanal.c (rtx_cost): Likewise.
2081 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
2082 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
2083 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
2084 (TARGET_MODES_TIEABLE_P): Redefine.
2085 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
2086 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
2087 (TARGET_MODES_TIEABLE_P): Redefine.
2088 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
2089 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
2090 (arc_modes_tieable_p): New function.
2091 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
2092 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
2093 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
2094 (arm_modes_tieable_p): Make static.
2095 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
2096 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
2097 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
2098 (TARGET_MODES_TIEABLE_P): Redefine.
2099 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
2100 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
2101 (TARGET_MODES_TIEABLE_P): Redefine.
2102 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
2103 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
2104 (cr16_modes_tieable_p): New function.
2105 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
2106 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
2107 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
2108 (TRULY_NOOP_TRUNCATION): Update comment.
2109 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
2110 (TRULY_NOOP_TRUNCATION): Update comment.
2111 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
2112 (frv_modes_tieable_p): New function.
2113 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
2114 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
2115 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
2116 (TARGET_MODES_TIEABLE_P): Redefine.
2117 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
2118 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
2119 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
2120 (TARGET_MODES_TIEABLE_P): Redefine.
2121 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
2122 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
2123 (ia64_modes_tieable_p): New function.
2124 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
2125 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
2126 (iq2000_modes_tieable_p): New function.
2127 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
2128 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
2129 (lm32_modes_tieable_p): New function.
2130 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
2131 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
2132 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
2133 (TARGET_MODES_TIEABLE_P): Redefine.
2134 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
2135 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
2136 (m32r_modes_tieable_p): New function.
2137 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
2138 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
2139 (m68k_modes_tieable_p): New function.
2140 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
2141 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
2142 (mcore_modes_tieable_p): New function.
2143 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
2144 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
2146 (TARGET_MODES_TIEABLE_P): Redefine.
2147 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
2148 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
2149 * config/mips/mips.c (mips_modes_tieable_p): Make static.
2150 (TARGET_MODES_TIEABLE_P): Redefine.
2151 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
2152 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
2153 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
2154 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
2155 (mn10300_modes_tieable_p): ...this and make static.
2156 (TARGET_MODES_TIEABLE_P): Redefine.
2157 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
2158 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
2159 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
2160 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
2161 (msp430_modes_tieable_p): Make static.
2162 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
2163 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
2164 (TARGET_MODES_TIEABLE_P): Redefine.
2165 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
2166 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
2167 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
2168 (TARGET_MODES_TIEABLE_P): Redefine.
2169 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
2170 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
2171 * config/pa/pa.c (pa_modes_tieable_p): Make static.
2172 (TARGET_MODES_TIEABLE_P): Redefine.
2173 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
2174 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
2175 (pdp11_modes_tieable_p): New function.
2176 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
2177 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
2178 (rs6000_modes_tieable_p): New function.
2179 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2180 * config/powerpcspe/powerpcspe.md: Update comment.
2181 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
2182 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
2183 (TARGET_MODES_TIEABLE_P): Redefine.
2184 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
2185 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
2186 (rl78_modes_tieable_p): New function.
2187 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
2188 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
2189 (rs6000_modes_tieable_p): New function.
2190 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
2191 * config/rs6000/rs6000.md: Update comment.
2192 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
2193 * config/rx/rx.c (rx_modes_tieable_p): New function.
2194 (TARGET_MODES_TIEABLE_P): Redefine.
2195 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
2196 * config/s390/s390.c (s390_modes_tieable_p): New function.
2197 (TARGET_MODES_TIEABLE_P): Redefine.
2198 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
2199 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
2200 (sh_modes_tieable_p): New function.
2201 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
2202 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
2203 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
2204 (sparc_modes_tieable_p): Make static.
2205 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
2206 * config/spu/spu.c (spu_modes_tieable_p): New function.
2207 (TARGET_MODES_TIEABLE_P): Redefine.
2208 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
2209 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
2210 (TARGET_MODES_TIEABLE_P): Redefine.
2211 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
2212 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
2213 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
2214 * config/v850/v850.c (v850_modes_tieable_p): New function.
2215 (TARGET_MODES_TIEABLE_P): Redefine.
2216 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
2217 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
2218 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
2219 (visium_modes_tieable_p): New function.
2220 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
2221 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
2222 (xtensa_modes_tieable_p): New function.
2223 * system.h (MODES_TIEABLE_P): Poison.
2225 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2226 Alan Hayward <alan.hayward@arm.com>
2227 David Sherwood <david.sherwood@arm.com>
2229 * target.def (hard_regno_mode_ok): New hook.
2230 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
2231 (TARGET_HARD_REGNO_MODE_OK): ...this.
2232 * doc/tm.texi.in: Regenerate.
2233 * hooks.h (hook_bool_uint_mode_true): Declare.
2234 * hooks.c (hook_bool_uint_mode_true): New function.
2235 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
2237 * genpreds.c (write_insn_preds_c): Add an include of target.h.
2238 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
2239 instead of HARD_REGNO_MODE_OK.
2240 * caller-save.c: Include target.h.
2241 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
2243 * combine.c (can_combine_p): Likewise.
2244 (combinable_i3pat): Likewise.
2245 (can_change_dest_mode): Likewise.
2246 * expr.c (init_expr_target): Likewise.
2247 (convert_move): Likewise.
2248 (convert_modes): Likewise.
2249 * ira.c (setup_prohibited_class_mode_regs): Likewise.
2250 (setup_prohibited_mode_move_regs): Likewise.
2251 * ira.h (target_ira): Likewise.
2252 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2253 * lra-constraints.c (process_alt_operands): Likewise.
2254 (split_reg): Likewise.
2255 * recog.c (peep2_find_free_register): Likewise.
2256 * ree.c (combine_reaching_defs): Likewise.
2257 * regcprop.c (maybe_mode_change): Likewise.
2258 * reginfo.c (init_reg_sets_1): Likewise.
2259 (choose_hard_reg_mode): Likewise.
2260 (simplifiable_subregs): Likewise.
2261 * regrename.c (check_new_reg_p): Likewise.
2262 * reload.c (find_valid_class): Likewise.
2263 (find_valid_class_1): Likewise.
2264 (reload_inner_reg_of_subreg): Likewise.
2265 (push_reload): Likewise.
2266 (combine_reloads): Likewise.
2267 (find_dummy_reload): Likewise.
2268 (find_reloads): Likewise.
2269 * reload1.c (find_reg): Likewise.
2270 (set_reload_reg): Likewise.
2271 (allocate_reload_reg): Likewise.
2272 (choose_reload_regs): Likewise.
2273 (reload_adjust_reg_for_temp): Likewise.
2274 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
2275 (simplify_subreg_regno): Likewise.
2276 * sel-sched.c (init_regs_for_mode): Likewise.
2277 * varasm.c (make_decl_rtl): Likewise.
2278 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
2279 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
2281 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
2282 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
2283 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2284 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
2285 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
2286 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2287 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
2288 (arc_mode_class): Delete.
2289 (HARD_REGNO_MODE_OK): Delete.
2290 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2291 (arc_hard_regno_mode_ok): Rename old array to...
2292 (arc_hard_regno_mode_ok_modes): ...this.
2293 (arc_conditional_register_usage): Update accordingly.
2294 (arc_mode_class): Make static.
2295 (arc_hard_regno_mode_ok): New function.
2296 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
2297 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
2298 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2299 (arm_hard_regno_mode_ok): Make static.
2300 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
2302 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
2303 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
2304 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
2306 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2307 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
2308 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
2309 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
2310 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
2311 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2312 * config/bfin/predicates.md (valid_reg_operand): Use
2313 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
2314 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
2315 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
2316 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2317 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
2318 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
2319 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2320 (cr16_hard_regno_mode_ok): Make static and return a bool.
2321 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
2322 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2323 (cris_hard_regno_mode_ok): New function.
2324 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
2325 (epiphany_mode_class): Delete.
2326 (HARD_REGNO_MODE_OK): Delete.
2327 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
2328 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2329 (hard_regno_mode_ok): Rename to...
2330 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
2331 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
2332 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
2334 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
2335 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
2336 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2337 (frv_hard_regno_mode_ok): Make static and return a bool.
2338 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
2340 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
2341 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
2342 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
2343 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
2345 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2346 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
2347 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
2348 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
2350 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2351 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
2352 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2353 (ia64_hard_regno_mode_ok): New function.
2354 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
2355 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2356 (iq2000_hard_regno_mode_ok): New function.
2357 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
2358 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2359 (lm32_hard_regno_mode_ok): New function.
2360 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
2361 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
2362 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
2363 instead of HARD_REGNO_MODE_OK.
2364 (m32c_hard_regno_ok): Rename to...
2365 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
2366 (m32c_cannot_change_mode_class): Update accordingly.
2367 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2368 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
2369 (m32r_mode_class): Delete.
2370 (HARD_REGNO_MODE_OK): Delete.
2371 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2372 (m32r_hard_regno_mode_ok): Rename to...
2373 (m32r_hard_regno_modes): ...this.
2374 (m32r_mode_class): Make static.
2375 (m32r_hard_regno_mode_ok): New function.
2376 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
2377 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
2378 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2379 (m68k_hard_regno_mode_ok): Make static.
2380 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
2381 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2382 (mcore_hard_regno_mode_ok): New function.
2383 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
2384 (HARD_REGNO_MODE_OK): Delete.
2385 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
2387 (microblaze_hard_regno_mode_ok_p): ...this and make static.
2388 (microblaze_hard_regno_mode_ok): New function.
2389 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2390 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
2391 (mips_hard_regno_mode_ok): Delete.
2392 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
2393 (mips_hard_regno_mode_ok_p): ...this and make static.
2394 (mips_hard_regno_mode_ok_p): Rename to...
2395 (mips_hard_regno_mode_ok_uncached): ...this.
2396 (mips_hard_regno_mode_ok): New function.
2397 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
2398 of HARD_REGNO_MODE_OK.
2399 (mips_option_override): Update after above name changes.
2400 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2401 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
2402 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
2403 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
2404 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
2405 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2406 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
2407 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
2408 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
2409 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2410 (msp430_hard_regno_mode_ok): Make static and return a bool.
2411 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
2412 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
2413 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
2415 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2416 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
2417 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
2418 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
2419 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
2420 (PA_HARD_REGNO_MODE_OK): ...this
2421 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
2422 (PA_HARD_REGNO_MODE_OK): ...this.
2423 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2424 (pa_hard_regno_mode_ok): New function.
2425 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
2426 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2427 (pdp11_hard_regno_mode_ok): New function.
2428 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
2429 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
2431 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
2433 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2434 (rs6000_hard_regno_mode_ok): Rename to...
2435 (rs6000_hard_regno_mode_ok_uncached): ...this.
2436 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2437 (rs6000_hard_regno_mode_ok): New function.
2438 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
2439 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
2440 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
2441 (riscv_hard_regno_mode_ok): ...this and make static.
2442 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2443 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
2444 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
2445 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2446 (rl78_hard_regno_mode_ok): Make static and return bool.
2447 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
2448 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
2450 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
2451 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2452 (rs6000_hard_regno_mode_ok): Rename to...
2453 (rs6000_hard_regno_mode_ok_uncached): ...this.
2454 (rs6000_init_hard_regno_mode_ok): Update accordingly.
2455 (rs6000_hard_regno_mode_ok): New function.
2456 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
2457 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
2458 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2459 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
2460 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
2461 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
2462 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2463 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
2464 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
2465 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2466 (sh_hard_regno_mode_ok): Make static.
2467 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
2468 instead of HARD_REGNO_MODE_OK.
2469 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
2470 (sparc_mode_class): Delete.
2471 (HARD_REGNO_MODE_OK): Delete.
2472 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2473 (hard_regno_mode_classes): Make static.
2474 (sparc_mode_class): Likewise.
2475 (sparc_hard_regno_mode_ok): New function.
2476 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
2477 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
2478 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
2480 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2481 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
2482 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
2483 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
2484 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
2485 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2486 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
2487 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
2488 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
2489 (visium_hard_regno_mode_ok): New function.
2490 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
2491 instead of HARD_REGNO_MODE_OK.
2492 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
2493 (HARD_REGNO_MODE_OK): Delete.
2494 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
2495 (xtensa_hard_regno_mode_ok_p): ...this and make static.
2496 (xtensa_option_override): Update accordingly.
2497 (TARGET_HARD_REGNO_MODE_OK): Redefine.
2498 (xtensa_hard_regno_mode_ok): New function.
2499 * system.h (HARD_REGNO_MODE_OK): Poison.
2501 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2502 Alan Hayward <alan.hayward@arm.com>
2503 David Sherwood <david.sherwood@arm.com>
2505 * target.def (hard_regno_call_part_clobbered): New hook.
2506 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
2507 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
2508 * doc/tm.texi: Regenerate.
2509 * hooks.h (hook_bool_uint_mode_false): Declare.
2510 * hooks.c (hook_bool_uint_mode_false): New function.
2511 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2512 * cselib.c (cselib_process_insn): Use
2513 targetm.hard_regno_call_part_clobbered instead of
2514 HARD_REGNO_CALL_PART_CLOBBERED.
2515 * ira-conflicts.c (ira_build_conflicts): Likewise.
2516 * ira-costs.c (ira_tune_allocno_costs): Likewise.
2517 * lra-constraints.c (need_for_call_save_p): Likewise.
2518 * lra-lives.c: Include target.h.
2519 (check_pseudos_live_through_calls): Use
2520 targetm.hard_regno_call_part_clobbered instead of
2521 HARD_REGNO_CALL_PART_CLOBBERED.
2522 * regcprop.c: Include target.h.
2523 (copyprop_hardreg_forward_1): Use
2524 targetm.hard_regno_call_part_clobbered instead of
2525 HARD_REGNO_CALL_PART_CLOBBERED.
2526 * reginfo.c (choose_hard_reg_mode): Likewise.
2527 * regrename.c (check_new_reg_p): Likewise.
2528 * reload.c (find_equiv_reg): Likewise.
2529 * reload1.c (emit_reload_insns): Likewise.
2530 * sched-deps.c (deps_analyze_insn): Likewise.
2531 * sel-sched.c (init_regs_for_mode): Likewise.
2532 (mark_unavailable_hard_regs): Likewise.
2533 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
2534 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2535 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
2537 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2538 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2539 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
2541 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
2543 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2544 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2545 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
2547 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2548 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2549 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
2551 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2552 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
2554 * config/powerpcspe/powerpcspe.c
2555 (rs6000_hard_regno_call_part_clobbered): New function.
2556 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2557 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2558 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
2560 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2561 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2562 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
2564 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
2565 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
2566 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
2568 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2569 Alan Hayward <alan.hayward@arm.com>
2570 David Sherwood <david.sherwood@arm.com>
2572 * rtl.h (subreg_memory_offset): Declare.
2573 * emit-rtl.c (subreg_memory_offset): New function.
2574 * expmed.c (store_bit_field_1): Use it.
2575 * expr.c (undefined_operand_subword_p): Likewise.
2576 * simplify-rtx.c (simplify_subreg): Likewise.
2578 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
2580 PR rtl-optimization/57448
2583 * optabs.c (expand_atomic_load): Place compiler memory barriers if
2584 using atomic_load pattern.
2585 (expand_atomic_store): Likewise.
2587 2017-09-04 Jakub Jelinek <jakub@redhat.com>
2590 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
2591 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
2592 handling. Use replace_call_with_value with NULL instead of
2593 gsi_replace, unlink_stmt_vdef and release_defs.
2595 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
2598 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
2600 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2603 * rtl.h (emit_library_call_value_1): Declare.
2604 (emit_library_call): Replace declaration with a series of overloads.
2605 Remove the parameter count argument.
2606 (emit_library_call_value): Likewise.
2607 * calls.c (emit_library_call_value_1): Make global. Replace varargs
2608 with an "rtx_mode_t *".
2609 (emit_library_call_value): Delete.
2610 (emit_library_call): Likewise.
2611 * asan.c (asan_emit_stack_protection): Update calls accordingly.
2612 (asan_emit_allocas_unpoison): Likewise.
2613 * builtins.c (expand_builtin_powi): Likewise.
2614 (expand_asan_emit_allocas_unpoison): Likewise.
2615 * cfgexpand.c (expand_main_function): Likewise.
2616 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
2617 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
2618 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
2619 * config/arm/arm.c (arm_trampoline_init): Likewise.
2620 (arm_call_tls_get_addr): Likewise.
2621 (arm_expand_divmod_libfunc): Likewise.
2622 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
2623 (smulsi3_highpart): Likewise.
2624 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
2625 (c6x_expand_compare): Likewise.
2626 (c6x_expand_movmem): Likewise.
2627 * config/frv/frv.c (frv_trampoline_init): Likewise.
2628 * config/i386/i386.c (ix86_trampoline_init): Likewise.
2629 (ix86_expand_divmod_libfunc): Likewise.
2630 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
2631 (ia64_expand_compare): Likewise.
2632 (ia64_profile_hook): Likewise.
2633 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
2634 (nonlocal_goto): Likewise.
2635 (restore_stack_nonlocal): Likewise.
2636 * config/m32r/m32r.c (block_move_call): Likewise.
2637 (m32r_trampoline_init): Likewise.
2638 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
2639 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
2640 (m68k_call_m68k_read_tp): Likewise.
2641 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
2642 (microblaze_expand_divide): Likewise.
2643 * config/mips/mips.h (mips_args): Likewise.
2644 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
2645 (MIPS_ICACHE_SYNC): Likewise.
2646 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
2647 (nios2_trampoline_init): Likewise.
2648 * config/pa/pa.c (hppa_tls_call): Likewise.
2649 (pa_trampoline_init): Likewise.
2650 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
2651 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
2652 (expand_strn_compare): Likewise.
2653 (rs6000_generate_compare): Likewise.
2654 (rs6000_expand_float128_convert): Likewise.
2655 (output_profile_hook): Likewise.
2656 (rs6000_trampoline_init): Likewise.
2657 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
2658 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
2659 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
2660 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
2661 (rs6000_generate_compare): Likewise.
2662 (rs6000_expand_float128_convert): Likewise.
2663 (output_profile_hook): Likewise.
2664 (rs6000_trampoline_init): Likewise.
2665 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
2666 * config/sh/sh.c (sh_trampoline_init): Likewise.
2667 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
2668 (sparc_emit_float_lib_cmp): Likewise.
2669 (sparc32_initialize_trampoline): Likewise.
2670 (sparc64_initialize_trampoline): Likewise.
2671 (sparc_profile_hook): Likewise.
2672 * config/spu/spu.c (ea_load_store): Likewise.
2673 * config/spu/spu.md (floatunssidf2): Likewise.
2674 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
2675 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
2676 * config/visium/visium.c (expand_block_move_4): Likewise.
2677 (expand_block_move_2): Likewise.
2678 (expand_block_move_1): Likewise.
2679 (expand_block_set_4): Likewise.
2680 (expand_block_set_2): Likewise.
2681 (expand_block_set_1): Likewise.
2682 (visium_trampoline_init): Likewise.
2683 (visium_profile_hook): Likewise.
2684 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
2685 (xtensa_setup_frame_addresses): Likewise.
2686 (xtensa_trampoline_init): Likewise.
2687 * except.c (sjlj_emit_function_enter): Likewise.
2688 (sjlj_emit_function_exit): Likewise.
2689 * explow.c (allocate_dynamic_stack_space): Likewise.
2690 (probe_stack_range): Likewise.
2691 * expr.c (convert_mode_scalar): Likewise.
2692 * optabs.c (expand_binop): Likewise.
2693 (expand_twoval_binop_libfunc): Likewise.
2694 (expand_unop): Likewise.
2695 (prepare_cmp_insn): Likewise.
2696 (prepare_float_lib_cmp): Likewise.
2697 (expand_float): Likewise.
2698 (expand_fix): Likewise.
2699 (expand_fixed_convert): Likewise.
2700 (maybe_emit_sync_lock_test_and_set): Likewise.
2701 (expand_atomic_compare_and_swap): Likewise.
2702 (expand_mem_thread_fence): Likewise.
2703 (expand_atomic_fetch_op): Likewise.
2705 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
2707 * doc/generic.texi (OpenACC): Adjust URL.
2708 * doc/invoke.texi (C Dialect Options): Ditto.
2710 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
2712 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
2713 predicate for operand 1. Add (m,<S>) constraint.
2714 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
2715 Prevent memory operand 1 with register operand 2.
2717 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
2719 PR rtl-optimization/82024
2720 * combine.c (try_combine): If the combination result is a PARALLEL,
2721 and we only need to retain the SET in there that would be placed
2722 at I2, check that we can place that at I3 instead, before doing so.
2724 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2727 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
2729 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
2730 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
2733 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
2734 Jeff Law <law@redhat.com>
2736 * varasm.c (bss_initializer_p): Do not put constants into .bss
2737 (categorize_decl_for_section): Handle bss_initializer_p returning
2738 false when DECL_INITIAL is NULL.
2740 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2743 * config/s390/s390.c (s390_can_inline_p): New function.
2745 2017-09-01 Jeff Law <law@redhat.com>
2747 PR tree-optimization/82052
2748 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
2749 Always initialize the returned slot after a hash table miss
2750 when INSERT is true.
2752 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
2754 * config/s390/s390.md (mem_signal_fence): Remove.
2755 * doc/md.texi (mem_signal_fence): Remove.
2756 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
2758 * target-insns.def (mem_signal_fence): Remove.
2760 2017-09-01 Jakub Jelinek <jakub@redhat.com>
2763 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
2766 * asan.c (create_odr_indicator): Strip name encoding from assembler
2767 name before appending it after __odr_asan_.
2769 2017-09-01 Martin Liska <mliska@suse.cz>
2771 PR tree-optimization/82059
2772 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
2773 frequency only when an edge is redirected.
2775 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2777 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
2778 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
2779 (arc_conditional_register_usage): Remove ARC600 lp_count
2781 (arc_file_start): Emit Tag_ARC_CPU_variation.
2782 (arc_can_use_doloop_p): New conditions to use ZOLs.
2783 (hwloop_fail): New function.
2784 (hwloop_optimize): Likewise.
2785 (hwloop_pattern_reg): Likewise.
2786 (arc_doloop_hooks): New struct, to be used with reorg_loops.
2787 (arc_reorg_loops): New function, calls reorg_loops.
2788 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
2789 (arc600_corereg_hazard): Remove ZOL checking, case handled by
2791 (arc_loop_hazard): Remove function, functionality moved into
2793 (arc_hazard): Remove arc_loop_hazard call.
2794 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
2795 into hwloop_optimize.
2796 (arc_label_align): Remove ZOL handling.
2797 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
2798 * config/arc/arc.md (doloop_begin): Remove pattern.
2799 (doloop_begin_i): Likewise.
2800 (doloop_end_i): Likewise.
2801 (doloop_fallback): Likewise.
2802 (doloop_fallback_m): Likewise.
2803 (doloop_end): Reimplement expand.
2804 (arc_lp): New pattern for LP instruction.
2805 (loop_end): New pattern.
2806 (loop_fail): Likewise.
2807 (decrement_and_branch_until_zero): Likewise.
2808 * config/arc/arc.opt (mlpc-width): New option.
2809 * doc/invoke.texi (mlpc-width): Document option.
2811 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2813 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
2814 (arc_ccfsm_advance): Fix checking for delay slots.
2815 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
2817 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2819 * config/arc/arc.md (movqi_insn): Add stores to save constant long
2821 (movhi_insn): Update store instruction constraint which are saving
2822 6-bit short immediates.
2823 (movsi_insn): Consider also short scaled load operations.
2824 (zero_extendhisi2_i): Use Usd constraint instead of T.
2825 (extendhisi2_i): Add q constraint.
2826 (arc_clzsi2): Add type and length attributes.
2827 (arc_ctzsi2): Likewise.
2828 * config/arc/constraints.md (Usc): Update constraint, the
2829 assembler can parse two relocations for a single instruction.
2831 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
2833 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
2834 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
2836 2017-08-31 Olivier Hainque <hainque@adacore.com>
2838 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
2839 match as powerpc-wrs-vxworks*.
2841 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
2843 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
2844 register constraint for by-element operand.
2845 (aarch64_mls_elt_merge<mode>): Likewise.
2847 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2849 * config/arc/arc.c (arc_can_follow_jump): Check for short
2852 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2854 * config.gcc: Use g.opt for arc.
2855 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
2856 functionality moved to ...
2857 (legitimate_scaled_address_p): New function, ...here.
2858 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
2859 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
2860 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
2862 (arc_override_options): Handle G option.
2863 (arc_output_pic_addr_const): Correct function definition.
2864 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
2865 (arc_decl_anon_ns_mem_p): Delete.
2866 (arc_in_small_data_p): Overhaul this function to take into
2867 consideration the value given via G option.
2868 (arc_rewrite_small_data_1): Renamed and corrected old
2869 arc_rewrite_small_data function.
2870 (arc_rewrite_small_data): New function.
2871 (small_data_pattern): Don't use pic_offset_table_rtx.
2872 * config/arc/arc.h (CC1_SPEC): Recognize G option.
2873 * config/arc/simdext.md (movmisalignv2hi): Use
2874 prepare_move_operands function.
2876 (movmisalign*): Likewise.
2877 * doc/invoke.texi (ARC options): Document -G option.
2879 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
2881 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
2883 * config/arc/arc.c (arc_print_operand): Output scalled address for
2884 sdata whenever is possible.
2885 (arc_in_small_data_p): Allow sdata for 64bit datum when double
2886 load/stores are available.
2887 (compact_sda_memory_operand): Check for the alignment required by
2888 code density instructions.
2889 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
2891 * config/arc/constraints.md (Usd): Update constraint.
2892 (Us0): New constraint.
2893 (Usc): Update constraint.
2895 2017-08-31 Richard Biener <rguenther@suse.de>
2898 * dwarf2out.c (dwarf2out_early_global_decl): Process each
2901 2017-08-31 Tamar Christina <tamar.christina@arm.com>
2903 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
2904 Resize type_signature.
2906 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2907 Alan Hayward <alan.hayward@arm.com>
2908 David Sherwood <david.sherwood@arm.com>
2910 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
2911 subregs whose inner modes can be stored in GPRs.
2912 (aarch64_classify_index): Likewise.
2914 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2915 Alan Hayward <alan.hayward@arm.com>
2916 David Sherwood <david.sherwood@arm.com>
2918 * config/aarch64/iterators.md (V_cmp_result): Rename to...
2919 (V_INT_EQUIV): ...this.
2920 (v_cmp_result): Rename to...
2921 (v_int_equiv): ...this.
2922 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
2923 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
2924 (copysign<mode>3): Likewise.
2925 (aarch64_simd_bsl<mode>_internal): Likewise.
2926 (aarch64_simd_bsl<mode>): Likewise.
2927 (vec_cmp<mode><mode>): Likewise.
2928 (vcond<mode><mode>): Likewise.
2929 (vcond<v_cmp_mixed><mode>): Likewise.
2930 (vcondu<mode><v_cmp_mixed>): Likewise.
2931 (aarch64_cm<optab><mode>): Likewise.
2932 (aarch64_cmtst<mode>): Likewise.
2933 (aarch64_fac<optab><mode>): Likewise.
2934 (vec_perm_const<mode>): Likewise.
2935 (vcond_mask_<mode><v_cmp_result>): Rename to...
2936 (vcond_mask_<mode><v_int_equiv>): ...this.
2937 (vec_cmp<mode><v_cmp_result>): Rename to...
2938 (vec_cmp<mode><v_int_equiv>): ...this.
2940 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
2941 Alan Hayward <alan.hayward@arm.com>
2942 David Sherwood <david.sherwood@arm.com>
2944 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
2946 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
2947 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
2948 (UNSPEC_LD4_DREG): New unspecs.
2949 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
2950 (aarch64_ld2<mode>_dreg_be): Replace with...
2951 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
2953 (aarch64_ld3<mode>_dreg_le)
2954 (aarch64_ld3<mode>_dreg_be): Replace with...
2955 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
2957 (aarch64_ld4<mode>_dreg_le)
2958 (aarch64_ld4<mode>_dreg_be): Replace with...
2959 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
2962 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2964 PR tree-optimization/81987
2965 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
2966 insert an initializer in a location not dominated by the stride
2969 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2971 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
2972 on the entire header of the finally block in the fallthru case.
2974 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
2976 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
2978 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
2980 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
2981 rs6000_emit_move_from_cr and call renamed function.
2982 (rs6000_emit_prologue): Call renamed functions.
2983 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
2984 movesi_from_cr, remove volatile CRs.
2986 2017-08-30 Jon Beniston <jon@beniston.com>
2987 Richard Biener <rguenther@suse.de>
2989 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
2990 of VECTOR_MODE_P check.
2991 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
2992 element vector types.
2994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2996 * df.h (df_read_modify_subreg_p): Remove in favor of...
2997 * rtl.h (read_modify_subreg_p): ...this new function. Take a
2998 const_rtx instead of an rtx.
2999 * cprop.c (local_cprop_find_used_regs): Update accordingly.
3000 * df-problems.c (df_word_lr_mark_ref): Likewise.
3001 * ira-lives.c (mark_pseudo_reg_live): Likewise.
3002 (mark_pseudo_reg_dead): Likewise.
3003 (mark_ref_dead): Likewise.
3004 * reginfo.c (init_subregs_of_mode): Likewise.
3005 * sched-deps.c (sched_analyze_1): Likewise.
3006 * df-scan.c (df_def_record_1): Likewise.
3007 (df_uses_record): Likewise.
3008 (df_read_modify_subreg_p): Remove in favor of...
3009 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
3010 const_rtx instead of an rtx.
3012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3013 Alan Hayward <alan.hayward@arm.com>
3014 David Sherwood <david.sherwood@arm.com>
3016 * rtl.h (partial_subreg_p): New function.
3017 * caller-save.c (save_call_clobbered_regs): Use it.
3018 * calls.c (expand_call): Likewise.
3019 * combine.c (combinable_i3pat): Likewise.
3020 (simplify_set): Likewise.
3021 (make_extraction): Likewise.
3022 (make_compound_operation_int): Likewise.
3023 (gen_lowpart_or_truncate): Likewise.
3024 (force_to_mode): Likewise.
3025 (make_field_assignment): Likewise.
3026 (reg_truncated_to_mode): Likewise.
3027 (record_truncated_value): Likewise.
3028 (move_deaths): Likewise.
3029 * cse.c (record_jump_cond): Likewise.
3030 (cse_insn): Likewise.
3031 * cselib.c (cselib_lookup_1): Likewise.
3032 * expmed.c (extract_bit_field_using_extv): Likewise.
3033 * function.c (assign_parm_setup_reg): Likewise.
3034 * ifcvt.c (noce_convert_multiple_sets): Likewise.
3035 * ira-build.c (create_insn_allocnos): Likewise.
3036 * lra-coalesce.c (merge_pseudos): Likewise.
3037 * lra-constraints.c (match_reload): Likewise.
3038 (simplify_operand_subreg): Likewise.
3039 (curr_insn_transform): Likewise.
3040 * lra-lives.c (process_bb_lives): Likewise.
3041 * lra.c (new_insn_reg): Likewise.
3042 (lra_substitute_pseudo): Likewise.
3043 * regcprop.c (mode_change_ok): Likewise.
3044 (maybe_mode_change): Likewise.
3045 (copyprop_hardreg_forward_1): Likewise.
3046 * reload.c (push_reload): Likewise.
3047 (find_reloads): Likewise.
3048 (find_reloads_subreg_address): Likewise.
3049 * reload1.c (alter_reg): Likewise.
3050 (eliminate_regs_1): Likewise.
3051 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3053 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
3055 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
3056 back to if statements, including unpack.
3058 2017-08-30 Martin Liska <mliska@suse.cz>
3061 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
3063 (func_checker::compare_asm_inputs_outputs): ... this function.
3064 (func_checker::compare_gimple_asm): Use the function to compare
3065 also ASM constrains.
3066 * ipa-icf-gimple.h: Rename the function.
3068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3069 Alan Hayward <alan.hayward@arm.com>
3070 David Sherwood <david.sherwood@arm.com>
3072 * coretypes.h (complex_mode): New type.
3073 * gdbhooks.py (build_pretty_printer): Handle it.
3074 * machmode.h (complex_mode): New class.
3075 (complex_mode::includes_p): New function.
3076 (is_complex_int_mode): Likewise.
3077 (is_complex_float_mode): Likewise.
3078 * genmodes.c (get_mode_class): Handle complex mode classes.
3079 * function.c (expand_function_end): Use is_complex_int_mode.
3081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3082 Alan Hayward <alan.hayward@arm.com>
3083 David Sherwood <david.sherwood@arm.com>
3085 * coretypes.h (scalar_mode_pod): New typedef.
3086 * gdbhooks.py (build_pretty_printer): Handle it.
3087 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
3088 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
3089 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
3090 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
3093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3094 Alan Hayward <alan.hayward@arm.com>
3095 David Sherwood <david.sherwood@arm.com>
3097 * machmode.h (mode_for_vector): Take a scalar_mode instead
3099 * stor-layout.c (mode_for_vector): Likewise.
3100 * explow.c (promote_mode): Use as_a <scalar_mode>.
3101 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
3103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3104 Alan Hayward <alan.hayward@arm.com>
3105 David Sherwood <david.sherwood@arm.com>
3107 * target.def (preferred_simd_mode): Take a scalar_mode
3108 instead of a machine_mode.
3109 * targhooks.h (default_preferred_simd_mode): Likewise.
3110 * targhooks.c (default_preferred_simd_mode): Likewise.
3111 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
3112 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
3113 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
3114 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
3115 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
3116 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
3117 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
3118 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
3120 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
3121 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
3122 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
3123 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
3124 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
3125 * doc/tm.texi: Regenerate.
3126 * optabs-query.c (can_vec_mask_load_store_p): Return false for
3129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3130 Alan Hayward <alan.hayward@arm.com>
3131 David Sherwood <david.sherwood@arm.com>
3133 * target.def (scalar_mode_supported_p): Take a scalar_mode
3134 instead of a machine_mode.
3135 * targhooks.h (default_scalar_mode_supported_p): Likewise.
3136 * targhooks.c (default_scalar_mode_supported_p): Likewise.
3137 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
3138 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3139 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
3140 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
3141 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
3142 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
3143 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
3144 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
3145 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
3146 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
3147 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
3148 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
3150 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
3151 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
3152 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3153 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3154 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
3156 * doc/tm.texi: Regenerate.
3158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3159 Alan Hayward <alan.hayward@arm.com>
3160 David Sherwood <david.sherwood@arm.com>
3162 * coretypes.h (opt_scalar_mode): New typedef.
3163 * gdbhooks.py (build_pretty_printers): Handle it.
3164 * machmode.h (mode_iterator::get_2xwider): Add overload for
3166 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
3168 * expr.c (convert_mode_scalar): Likewise.
3169 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3170 * optabs.c (expand_float): Likewise.
3171 (expand_fix): Likewise.
3172 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3175 Alan Hayward <alan.hayward@arm.com>
3176 David Sherwood <david.sherwood@arm.com>
3178 * optabs.c (expand_float): Explicitly check for scalars before
3179 using a branching expansion.
3180 (expand_fix): Likewise.
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 * expr.c (convert_mode): Split scalar handling out into...
3187 (convert_mode_scalar): ...this new function. Treat the modes
3190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3191 Alan Hayward <alan.hayward@arm.com>
3192 David Sherwood <david.sherwood@arm.com>
3194 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
3196 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
3198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3199 Alan Hayward <alan.hayward@arm.com>
3200 David Sherwood <david.sherwood@arm.com>
3202 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
3203 rather than a machine_mode.
3204 (fixed_from_string): Likewise.
3205 (fixed_convert): Likewise.
3206 (fixed_convert_from_int): Likewise.
3207 (fixed_convert_from_real): Likewise.
3208 (real_convert_from_fixed): Likewise.
3209 * fixed-value.c (fixed_from_double_int): Likewise.
3210 (fixed_from_string): Likewise.
3211 (fixed_convert): Likewise.
3212 (fixed_convert_from_int): Likewise.
3213 (fixed_convert_from_real): Likewise.
3214 (real_convert_from_fixed): Likewise.
3215 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
3217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3218 Alan Hayward <alan.hayward@arm.com>
3219 David Sherwood <david.sherwood@arm.com>
3221 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
3222 of separate mode class checks. Do not allow vector modes here.
3223 (immed_wide_int_const): Use as_a <scalar_mode>.
3224 * explow.c (trunc_int_for_mode): Likewise.
3225 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
3226 (wi::shwi): Likewise.
3227 (wi::min_value): Likewise.
3228 (wi::max_value): Likewise.
3229 * dwarf2out.c (loc_descriptor): Likewise.
3230 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
3233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3234 Alan Hayward <alan.hayward@arm.com>
3235 David Sherwood <david.sherwood@arm.com>
3237 * tree.h (SCALAR_TYPE_MODE): New macro.
3238 * expr.c (expand_expr_addr_expr_1): Use it.
3239 (expand_expr_real_2): Likewise.
3240 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
3241 (fold_convert_const_fixed_from_int): Likewise.
3242 (fold_convert_const_fixed_from_real): Likewise.
3243 (native_encode_fixed): Likewise
3244 (native_encode_complex): Likewise
3245 (native_encode_vector): Likewise.
3246 (native_interpret_fixed): Likewise.
3247 (native_interpret_real): Likewise.
3248 (native_interpret_complex): Likewise.
3249 (native_interpret_vector): Likewise.
3250 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
3251 (simd_clone_adjust_argument_types): Likewise.
3252 (simd_clone_init_simd_arrays): Likewise.
3253 (simd_clone_adjust): Likewise.
3254 * stor-layout.c (layout_type): Likewise.
3255 * tree.c (build_minus_one_cst): Likewise.
3256 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3257 * tree-inline.c (estimate_move_cost): Likewise.
3258 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
3259 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
3260 (vectorizable_reduction): Likewise.
3261 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
3262 (vect_recog_mixed_size_cond_pattern): Likewise.
3263 (check_bool_pattern): Likewise.
3264 (adjust_bool_pattern): Likewise.
3265 (search_type_for_mask_1): Likewise.
3266 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3267 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3268 (vectorizable_load): Likewise.
3269 (vectorizable_store): Likewise.
3270 * ubsan.c (ubsan_encode_value): Likewise.
3271 * varasm.c (output_constant): Likewise.
3273 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3274 Alan Hayward <alan.hayward@arm.com>
3275 David Sherwood <david.sherwood@arm.com>
3277 * coretypes.h (scalar_mode): New class.
3278 * machmode.h (scalar_mode): Likewise.
3279 (scalar_mode::includes_p): New function.
3280 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
3281 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
3282 * genmodes.c (get_mode_class): Handle remaining scalar modes.
3283 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
3284 * expmed.c (store_bit_field_1): Likewise.
3285 (extract_bit_field_1): Likewise.
3286 * expr.c (write_complex_part): Likewise.
3287 (read_complex_part): Likewise.
3288 (emit_move_complex_push): Likewise.
3289 (expand_expr_real_2): Likewise.
3290 * function.c (assign_parm_setup_reg): Likewise.
3291 (assign_parms_unsplit_complex): Likewise.
3292 * optabs.c (expand_binop): Likewise.
3293 * rtlanal.c (subreg_get_info): Likewise.
3294 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3295 * varasm.c (output_constant_pool_2): Likewise.
3297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3298 Alan Hayward <alan.hayward@arm.com>
3299 David Sherwood <david.sherwood@arm.com>
3301 * expmed.c (extract_high_half): Use scalar_int_mode and remove
3303 (expmed_mult_highpart_optab): Likewise.
3304 (expmed_mult_highpart): Likewise.
3306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3307 Alan Hayward <alan.hayward@arm.com>
3308 David Sherwood <david.sherwood@arm.com>
3310 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
3311 instead of a machine_mode.
3312 (builtin_memset_read_str): Likewise.
3313 * builtins.c (c_readstr): Likewise.
3314 (builtin_memcpy_read_str): Likewise.
3315 (builtin_strncpy_read_str): Likewise.
3316 (builtin_memset_read_str): Likewise.
3317 (builtin_memset_gen_str): Likewise.
3318 (expand_builtin_signbit): Use scalar_int_mode for local variables.
3319 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
3320 instead of a machine_mode.
3321 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
3323 (make_extraction): Likewise.
3324 (try_widen_shift_mode): Take and return scalar_int_modes instead
3326 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
3327 a scalar_int_mode instead of a machine_mode.
3328 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
3329 (avr_addr_space_pointer_mode): Likewise.
3330 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
3331 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
3332 (msp430_unwind_word_mode): Likewise.
3333 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
3334 (spu_addr_space_pointer_mode): Likewise.
3335 (spu_addr_space_address_mode): Likewise.
3336 (spu_libgcc_cmp_return_mode): Likewise.
3337 (spu_libgcc_shift_count_mode): Likewise.
3338 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
3339 (rl78_addr_space_pointer_mode): Likewise.
3340 (fl78_unwind_word_mode): Likewise.
3341 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
3343 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
3344 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
3345 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
3346 (mips_valid_pointer_mode): Likewise.
3347 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
3348 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
3349 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
3351 (ft32_addr_space_address_mode): Likewise.
3352 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
3353 scalar_int_mode instead of a machine_mode.
3354 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
3356 (m32c_addr_space_address_mode): Likewise.
3357 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
3358 (rs6000_eh_return_filter_mode): Likewise.
3359 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
3360 (rs6000_eh_return_filter_mode): Likewise.
3361 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
3362 (s390_libgcc_shift_count_mode): Likewise.
3363 (s390_unwind_word_mode): Likewise.
3364 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
3366 * target.def (mode_rep_extended): Likewise.
3367 (valid_pointer_mode): Likewise.
3368 (addr_space.valid_pointer_mode): Likewise.
3369 (eh_return_filter_mode): Return a scalar_int_mode rather than
3371 (libgcc_cmp_return_mode): Likewise.
3372 (libgcc_shift_count_mode): Likewise.
3373 (unwind_word_mode): Likewise.
3374 (addr_space.pointer_mode): Likewise.
3375 (addr_space.address_mode): Likewise.
3376 * doc/tm.texi: Regenerate.
3377 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
3379 (do_jump): Use scalar_int_mode for local variables.
3380 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
3381 rather than a machine_mode.
3382 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
3383 (scompare_loc_descriptor_wide): Likewise.
3384 (scompare_loc_descriptor_narrow): Likewise.
3385 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
3387 * except.c (sjlj_emit_dispatch_table): Likewise.
3388 (expand_builtin_eh_copy_values): Likewise.
3389 * explow.c (convert_memory_address_addr_space_1): Likewise.
3390 Take a scalar_int_mode rather than a machine_mode.
3391 (convert_memory_address_addr_space): Take a scalar_int_mode rather
3392 than a machine_mode.
3393 (memory_address_addr_space): Use scalar_int_mode for local variables.
3394 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
3395 rather than a machine_mode.
3396 * expmed.c (mask_rtx): Likewise.
3397 (init_expmed_one_conv): Likewise.
3398 (expand_mult_highpart_adjust): Likewise.
3399 (extract_high_half): Likewise.
3400 (expmed_mult_highpart_optab): Likewise.
3401 (expmed_mult_highpart): Likewise.
3402 (expand_smod_pow2): Likewise.
3403 (expand_sdiv_pow2): Likewise.
3404 (emit_store_flag_int): Likewise.
3405 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
3407 (extract_low_bits): Likewise.
3408 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
3410 * expr.c (pieces_addr::adjust): Likewise.
3411 (can_store_by_pieces): Likewise.
3412 (store_by_pieces): Likewise.
3413 (clear_by_pieces_1): Likewise.
3414 (expand_expr_addr_expr_1): Likewise.
3415 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
3416 (expand_expr_real_1): Likewise.
3417 (try_casesi): Likewise.
3418 * final.c (shorten_branches): Likewise.
3419 * fold-const.c (fold_convert_const_int_from_fixed): Change the
3420 type of "mode" to machine_mode.
3421 * internal-fn.c (expand_arith_overflow_result_store): Take a
3422 scalar_int_mode rather than a machine_mode.
3423 (expand_mul_overflow): Use scalar_int_mode for local variables.
3424 * loop-doloop.c (doloop_modify): Likewise.
3425 (doloop_optimize): Likewise.
3426 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
3427 than a machine_mode.
3428 (expand_doubleword_shift_condmove): Likewise.
3429 (expand_doubleword_shift): Likewise.
3430 (expand_doubleword_clz): Likewise.
3431 (expand_doubleword_popcount): Likewise.
3432 (expand_doubleword_parity): Likewise.
3433 (expand_absneg_bit): Use scalar_int_mode for local variables.
3434 (prepare_float_lib_cmp): Likewise.
3435 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
3436 rather than a machine_mode.
3437 (convert_memory_address_addr_space): Likewise.
3438 (get_mode_bounds): Likewise.
3439 (get_address_mode): Return a scalar_int_mode rather than a
3441 * rtlanal.c (get_address_mode): Likewise.
3442 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
3443 than a machine_mode.
3444 * targhooks.c (default_mode_rep_extended): Likewise.
3445 (default_valid_pointer_mode): Likewise.
3446 (default_addr_space_valid_pointer_mode): Likewise.
3447 (default_eh_return_filter_mode): Return a scalar_int_mode rather
3448 than a machine_mode.
3449 (default_libgcc_cmp_return_mode): Likewise.
3450 (default_libgcc_shift_count_mode): Likewise.
3451 (default_unwind_word_mode): Likewise.
3452 (default_addr_space_pointer_mode): Likewise.
3453 (default_addr_space_address_mode): Likewise.
3454 * targhooks.h (default_eh_return_filter_mode): Likewise.
3455 (default_libgcc_cmp_return_mode): Likewise.
3456 (default_libgcc_shift_count_mode): Likewise.
3457 (default_unwind_word_mode): Likewise.
3458 (default_addr_space_pointer_mode): Likewise.
3459 (default_addr_space_address_mode): Likewise.
3460 (default_mode_rep_extended): Take a scalar_int_mode rather than
3462 (default_valid_pointer_mode): Likewise.
3463 (default_addr_space_valid_pointer_mode): Likewise.
3464 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
3466 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
3467 rather than a machine_mode.
3468 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
3469 for local variables.
3470 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3471 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
3472 than a machine_mode.
3474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3475 Alan Hayward <alan.hayward@arm.com>
3476 David Sherwood <david.sherwood@arm.com>
3478 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
3479 the mode argument to scalar_int_mode.
3480 (do_jump_by_parts_zero_rtx): Likewise.
3481 (do_jump_by_parts_equality_rtx): Likewise.
3482 (do_jump_by_parts_greater): Take a mode argument.
3483 (do_jump_by_parts_equality): Likewise.
3484 (do_jump_1): Update calls accordingly.
3486 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3487 Alan Hayward <alan.hayward@arm.com>
3488 David Sherwood <david.sherwood@arm.com>
3490 * is-a.h (safe_dyn_cast): New function.
3491 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
3492 (jump_table_for_label): Likewise.
3493 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
3494 instead of an rtx_insn *.
3495 (shorten_branches): Use dyn_cast instead of LABEL_P and
3496 JUMP_TABLE_DATA_P. Use jump_table_for_label and
3497 rtx_jump_table_data::get_data_mode.
3498 (final_scan_insn): Likewise.
3500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3501 Alan Hayward <alan.hayward@arm.com>
3502 David Sherwood <david.sherwood@arm.com>
3504 * combine.c (try_combine): Use is_a <scalar_int_mode> when
3505 trying to combine a full-register integer set with a subreg
3508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3509 Alan Hayward <alan.hayward@arm.com>
3510 David Sherwood <david.sherwood@arm.com>
3512 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
3513 that is always either address_mode or pointer_mode.
3515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3516 Alan Hayward <alan.hayward@arm.com>
3517 David Sherwood <david.sherwood@arm.com>
3519 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
3520 when the two are known to be equal.
3522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3523 Alan Hayward <alan.hayward@arm.com>
3524 David Sherwood <david.sherwood@arm.com>
3526 * simplify-rtx.c (simplify_const_unary_operation): Use
3527 is_a <scalar_int_mode> instead of checking for a nonzero
3528 precision. Forcibly convert op_mode to a scalar_int_mode
3529 in that case. More clearly differentiate the operand and
3530 result modes and use the former when deciding what the value
3531 of a count-bits operation should be. Use is_int_mode instead
3532 of checking for a MODE_INT. Remove redundant check for whether
3533 this mode has a zero precision.
3535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3536 Alan Hayward <alan.hayward@arm.com>
3537 David Sherwood <david.sherwood@arm.com>
3539 * optabs.c (widen_leading): Change the type of the mode argument
3540 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
3541 (widen_bswap): Likewise.
3542 (expand_parity): Likewise.
3543 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
3544 (expand_ffs): Likewise.
3545 (epand_unop): Check for scalar integer modes before calling the
3548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3549 Alan Hayward <alan.hayward@arm.com>
3550 David Sherwood <david.sherwood@arm.com>
3552 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
3554 (expand_expr_real_1): Update call accordingly.
3556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3557 Alan Hayward <alan.hayward@arm.com>
3558 David Sherwood <david.sherwood@arm.com>
3560 * expmed.c (store_bit_field_using_insv): Add op0_mode and
3561 value_mode arguments. Use scalar_int_mode internally.
3562 (store_bit_field_1): Rename the new integer mode from imode
3563 to op0_mode and use it instead of GET_MODE (op0). Update calls
3564 to store_split_bit_field, store_bit_field_using_insv and
3565 store_fixed_bit_field.
3566 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
3567 Use scalar_int_mode internally. Use a bit count rather than a mode
3568 when calculating the largest bit size for get_best_mode.
3569 Update calls to store_split_bit_field and store_fixed_bit_field_1.
3570 (store_fixed_bit_field_1): Add mode and value_mode arguments.
3571 Remove assertion that OP0 has a scalar integer mode.
3572 (store_split_bit_field): Add op0_mode and value_mode arguments.
3573 Update calls to extract_fixed_bit_field.
3574 (extract_bit_field_using_extv): Add an op0_mode argument.
3575 Use scalar_int_mode internally.
3576 (extract_bit_field_1): Rename the new integer mode from imode to
3577 op0_mode and use it instead of GET_MODE (op0). Update calls to
3578 extract_split_bit_field, extract_bit_field_using_extv and
3579 extract_fixed_bit_field.
3580 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
3581 to extract_split_bit_field and extract_fixed_bit_field_1.
3582 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
3583 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
3585 (extract_split_bit_field): Add an op0_mode argument. Update call
3586 to extract_fixed_bit_field.
3588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3589 Alan Hayward <alan.hayward@arm.com>
3590 David Sherwood <david.sherwood@arm.com>
3592 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
3593 * explow.c (hard_function_value): Likewise.
3594 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
3595 convert_to_mode call outside the loop.
3596 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
3597 for the mode iterator. Require the mode specified by max_pieces
3599 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
3601 (copy_blkmode_to_reg): Likewise.
3602 (set_storage_via_setmem): Likewise.
3603 * optabs.c (prepare_cmp_insn): Likewise.
3604 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3605 * stor-layout.c (finish_bitfield_representative): Likewise.
3607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3608 Alan Hayward <alan.hayward@arm.com>
3609 David Sherwood <david.sherwood@arm.com>
3611 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
3612 * expr.c (convert_move): Use them.
3613 (convert_modes): Likewise.
3614 (store_expr_with_bounds): Likewise.
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 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
3621 parameter for the mode of "x". Remove the "known_x", "known_mode"
3622 and "known_ret" arguments. Change the type of the mode argument
3624 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
3625 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
3626 (reg_num_sign_bit_copies_for_combine): Likewise.
3627 * rtlanal.c (nonzero_bits1): Likewise.
3628 (num_sign_bit_copies1): Likewise.
3629 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
3630 (reg_num_sign_bit_copies_general): Likewise.
3631 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
3632 (reg_nonzero_bits_general): Likewise.
3634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3635 Alan Hayward <alan.hayward@arm.com>
3636 David Sherwood <david.sherwood@arm.com>
3638 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
3639 than in subroutines. Return 1 for non-integer modes.
3640 (cached_num_sign_bit_copies): Change the type of the mode parameter
3642 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
3643 classes. Handle CONST_INT_P first and then check whether X also
3644 has a scalar integer mode. Check the same thing for inner registers
3645 of a SUBREG and for values that are being extended or truncated.
3647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3648 Alan Hayward <alan.hayward@arm.com>
3649 David Sherwood <david.sherwood@arm.com>
3651 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
3652 in subroutines. Return the mode mask for non-integer modes.
3653 (cached_nonzero_bits): Change the type of the mode parameter
3655 (nonzero_bits1): Likewise. Remove early exit for other mode
3656 classes. Handle CONST_INT_P first and then check whether X
3657 also has a scalar integer mode.
3659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3660 Alan Hayward <alan.hayward@arm.com>
3661 David Sherwood <david.sherwood@arm.com>
3663 * expr.c (widest_int_mode_for_size): Make the comment match the code.
3664 Return a scalar_int_mode and assert that the size is greater than
3666 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
3667 (op_by_pieces_d::op_by_pieces_d): Likewise.
3668 (op_by_pieces_d::run): Likewise.
3669 (can_store_by_pieces): Likewise.
3671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3672 Alan Hayward <alan.hayward@arm.com>
3673 David Sherwood <david.sherwood@arm.com>
3675 * combine.c (extract_left_shift): Add a mode argument and update
3677 (make_compound_operation_int): Change the type of the mode parameter
3678 to scalar_int_mode and update the call to extract_left_shift.
3680 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3681 Alan Hayward <alan.hayward@arm.com>
3682 David Sherwood <david.sherwood@arm.com>
3684 * combine.c (simplify_and_const_int): Change the type of the mode
3685 parameter to scalar_int_mode.
3686 (simplify_and_const_int_1): Likewise. Update recursive call.
3688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3689 Alan Hayward <alan.hayward@arm.com>
3690 David Sherwood <david.sherwood@arm.com>
3692 * combine.c (simplify_compare_const): Check that the mode is a
3693 scalar_int_mode (rather than VOIDmode) before testing its
3695 (simplify_comparison): Move COMPARISON_P handling out of the
3696 loop and restrict the latter part of the loop to scalar_int_modes.
3697 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
3698 and when considering SUBREG_REGs. Use is_int_mode instead of
3699 checking GET_MODE_CLASS against MODE_INT.
3701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3702 Alan Hayward <alan.hayward@arm.com>
3703 David Sherwood <david.sherwood@arm.com>
3705 * combine.c (try_widen_shift_mode): Move check for equal modes to...
3706 (simplify_shift_const_1): ...here. Use scalar_int_mode for
3707 shift_unit_mode and for modes involved in scalar shifts.
3709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3710 Alan Hayward <alan.hayward@arm.com>
3711 David Sherwood <david.sherwood@arm.com>
3713 * combine.c (force_int_to_mode): New function, split out from...
3714 (force_to_mode): ...here. Keep xmode up-to-date and use it
3715 instead of GET_MODE (x).
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 * optabs-query.h (extraction_insn::struct_mode): Change type to
3722 opt_scalar_int_mode and update comment.
3723 (extraction_insn::field_mode): Change type to scalar_int_mode.
3724 (extraction_insn::pos_mode): Likewise.
3725 * combine.c (make_extraction): Update accordingly.
3726 * optabs-query.c (get_traditional_extraction_insn): Likewise.
3727 (get_optab_extraction_insn): Likewise.
3728 * recog.c (simplify_while_replacing): Likewise.
3729 * expmed.c (narrow_bit_field_mem): Change the type of the mode
3730 parameter to opt_scalar_int_mode.
3732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3733 Alan Hayward <alan.hayward@arm.com>
3734 David Sherwood <david.sherwood@arm.com>
3736 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
3737 to a scalar_int_mode instead of a machine_mode.
3738 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
3739 (get_best_mode): Return a boolean and use a pointer argument to store
3740 the selected mode. Replace the limit mode parameter with a bit limit.
3741 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
3742 for the values returned by bit_field_mode_iterator::next_mode.
3743 (store_bit_field): Update call to get_best_mode.
3744 (store_fixed_bit_field): Likewise.
3745 (extract_fixed_bit_field): Likewise.
3746 * expr.c (optimize_bitfield_assignment_op): Likewise.
3747 * fold-const.c (optimize_bit_field_compare): Likewise.
3748 (fold_truth_andor_1): Likewise.
3749 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
3750 Update for new type of m_mode.
3751 (get_best_mode): As above.
3753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3754 Alan Hayward <alan.hayward@arm.com>
3755 David Sherwood <david.sherwood@arm.com>
3757 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
3758 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
3759 (store_bit_field): Check is_a <scalar_int_mode> before calling
3760 strict_volatile_bitfield_p.
3761 (extract_bit_field): Likewise.
3763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3764 Alan Hayward <alan.hayward@arm.com>
3765 David Sherwood <david.sherwood@arm.com>
3767 * target.def (cstore_mode): Return a scalar_int_mode.
3768 * doc/tm.texi: Regenerate.
3769 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
3770 * targhooks.h (default_cstore_mode): Likewise.
3771 * targhooks.c (default_cstore_mode): Likewise, using a forced
3773 * expmed.c (emit_cstore): Expect the target of the cstore to be
3776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3777 Alan Hayward <alan.hayward@arm.com>
3778 David Sherwood <david.sherwood@arm.com>
3780 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
3782 (niter_desc): Likewise mode.
3783 (iv_analyze): Add a mode parameter.
3785 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
3786 and change its type to scalar_int_mode.
3787 * loop-iv.c: Update commentary at head of file.
3788 (iv_constant): Pass the mode paraeter before the rtx it describes
3789 and change its type to scalar_int_mode. Remove VOIDmode handling.
3790 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
3791 (iv_extend): Likewise.
3792 (shorten_into_mode): Likewise.
3793 (iv_add): Use scalar_int_mode.
3794 (iv_mult): Likewise.
3795 (iv_shift): Likewise.
3796 (canonicalize_iv_subregs): Likewise.
3797 (get_biv_step_1): Pass the outer_mode parameter before the rtx
3798 it describes and change its mode to scalar_int_mode. Also change
3799 the type of the returned inner_mode to scalar_int_mode.
3800 (get_biv_step): Likewise, turning outer_mode from a pointer
3801 into a direct parameter. Update call to get_biv_step_1.
3802 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
3803 iv_constant and get_biv_step.
3804 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
3805 and change its type to scalar_int_mode. Don't initialise iv->mode
3806 to VOIDmode and remove later checks for its still being VOIDmode.
3807 Update calls to iv_analyze_op and iv_analyze_expr. Check
3808 is_a <scalar_int_mode> when changing the mode under consideration.
3809 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
3810 Update call to iv_analyze_expr.
3811 (iv_analyze_op): Add a mode parameter. Reject subregs whose
3812 inner register is not also a scalar_int_mode. Update call to
3814 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
3815 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
3816 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
3817 separate mode class checks. Update calls to iv_analyze. Remove
3818 fix-up of VOIDmodes after iv_analyze_biv.
3819 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
3820 don't have a scalar_int_mode. Update call to biv_p.
3822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3823 Alan Hayward <alan.hayward@arm.com>
3824 David Sherwood <david.sherwood@arm.com>
3826 * cfgexpand.c (convert_debug_memory_address): Use
3827 as_a <scalar_int_mode>.
3828 * combine.c (expand_compound_operation): Likewise.
3829 (make_extraction): Likewise.
3830 (change_zero_ext): Likewise.
3831 (simplify_comparison): Likewise.
3832 * cse.c (cse_insn): Likewise.
3833 * dwarf2out.c (minmax_loc_descriptor): Likewise.
3834 (mem_loc_descriptor): Likewise.
3835 (loc_descriptor): Likewise.
3836 * expmed.c (init_expmed_one_mode): Likewise.
3837 (synth_mult): Likewise.
3838 (emit_store_flag_1): Likewise.
3839 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
3840 of a comparison with size.
3841 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
3842 (reduce_to_bit_field_precision): Likewise.
3843 * function.c (expand_function_end): Likewise.
3844 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
3845 * loop-doloop.c (doloop_modify): Likewise.
3846 * optabs.c (expand_binop): Likewise.
3847 (expand_unop): Likewise.
3848 (expand_copysign_absneg): Likewise.
3849 (prepare_cmp_insn): Likewise.
3850 (maybe_legitimize_operand): Likewise.
3851 * recog.c (const_scalar_int_operand): Likewise.
3852 * rtlanal.c (get_address_mode): Likewise.
3853 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3854 (simplify_cond_clz_ctz): Likewise.
3855 * tree-nested.c (get_nl_goto_field): Likewise.
3856 * tree.c (build_vector_type_for_mode): Likewise.
3857 * var-tracking.c (use_narrower_mode): Likewise.
3859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3860 Alan Hayward <alan.hayward@arm.com>
3861 David Sherwood <david.sherwood@arm.com>
3863 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
3864 * builtins.c (expand_builtin_signbit): Use it.
3865 * cfgexpand.c (expand_debug_expr): Likewise.
3866 * dojump.c (do_jump): Likewise.
3867 (do_compare_and_jump): Likewise.
3868 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
3869 * expmed.c (make_tree): Likewise.
3870 * expr.c (expand_expr_real_2): Likewise.
3871 (expand_expr_real_1): Likewise.
3872 (try_casesi): Likewise.
3873 * fold-const-call.c (fold_const_call_ss): Likewise.
3874 * fold-const.c (unextend): Likewise.
3875 (extract_muldiv_1): Likewise.
3876 (fold_single_bit_test): Likewise.
3877 (native_encode_int): Likewise.
3878 (native_encode_string): Likewise.
3879 (native_interpret_int): Likewise.
3880 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
3881 * internal-fn.c (expand_addsub_overflow): Likewise.
3882 (expand_neg_overflow): Likewise.
3883 (expand_mul_overflow): Likewise.
3884 (expand_arith_overflow): Likewise.
3885 * match.pd: Likewise.
3886 * stor-layout.c (layout_type): Likewise.
3887 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
3888 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
3889 * tree-ssanames.c (get_range_info): Likewise.
3890 * tree-switch-conversion.c (array_value_type) Likewise.
3891 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
3892 (vect_recog_divmod_pattern): Likewise.
3893 (vect_recog_mixed_size_cond_pattern): Likewise.
3894 * tree-vrp.c (extract_range_basic): Likewise.
3895 (simplify_float_conversion_using_ranges): Likewise.
3896 * tree.c (int_fits_type_p): Likewise.
3897 * ubsan.c (instrument_bool_enum_load): Likewise.
3898 * varasm.c (mergeable_string_section): Likewise.
3899 (narrowing_initializer_constant_valid_p): Likewise.
3900 (output_constant): Likewise.
3902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3903 Alan Hayward <alan.hayward@arm.com>
3904 David Sherwood <david.sherwood@arm.com>
3906 * machmode.h (NARROWEST_INT_MODE): New macro.
3907 * expr.c (alignment_for_piecewise_move): Use it instead of
3908 GET_CLASS_NARROWEST_MODE (MODE_INT).
3909 (push_block): Likewise.
3910 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3912 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3915 Alan Hayward <alan.hayward@arm.com>
3916 David Sherwood <david.sherwood@arm.com>
3918 * postreload.c (move2add_valid_value_p): Change the type of the
3919 mode parameter to scalar_int_mode.
3920 (move2add_use_add2_insn): Add a mode parameter and use it instead
3922 (move2add_use_add3_insn): Likewise.
3923 (reload_cse_move2add): Update accordingly.
3925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3926 Alan Hayward <alan.hayward@arm.com>
3927 David Sherwood <david.sherwood@arm.com>
3929 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
3931 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
3932 * optabs.c (expand_unop): Likewise.
3934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3935 Alan Hayward <alan.hayward@arm.com>
3936 David Sherwood <david.sherwood@arm.com>
3938 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
3939 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
3940 (popcount_loc_descriptor): Likewise.
3941 (bswap_loc_descriptor): Likewise.
3942 (rotate_loc_descriptor): Likewise.
3943 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
3944 calling the functions above.
3946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3947 Alan Hayward <alan.hayward@arm.com>
3948 David Sherwood <david.sherwood@arm.com>
3950 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
3952 (try_combine): Likewise.
3953 (simplify_if_then_else): Likewise.
3954 * cse.c (cse_insn): Likewise.
3955 * dwarf2out.c (mem_loc_descriptor): Likewise.
3956 * emit-rtl.c (gen_lowpart_common): Likewise.
3957 * simplify-rtx.c (simplify_truncation): Likewise.
3958 (simplify_binary_operation_1): Likewise.
3959 (simplify_const_relational_operation): Likewise.
3960 (simplify_ternary_operation): Likewise.
3961 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
3963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3964 Alan Hayward <alan.hayward@arm.com>
3965 David Sherwood <david.sherwood@arm.com>
3967 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
3968 * reload.c (push_reload): Likewise.
3969 (find_reloads): Likewise.
3971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3972 Alan Hayward <alan.hayward@arm.com>
3973 David Sherwood <david.sherwood@arm.com>
3975 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
3976 (make_compound_operation_int): Likewise.
3977 (change_zero_ext): Likewise.
3978 * expr.c (convert_move): Likewise.
3979 (convert_modes): Likewise.
3980 * fwprop.c (forward_propagate_subreg): Likewise.
3981 * loop-iv.c (get_biv_step_1): Likewise.
3982 * optabs.c (widen_operand): Likewise.
3983 * postreload.c (move2add_valid_value_p): Likewise.
3984 * recog.c (simplify_while_replacing): Likewise.
3985 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3986 (simplify_binary_operation_1): Likewise. Remove redundant
3987 mode equality check.
3989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3990 Alan Hayward <alan.hayward@arm.com>
3991 David Sherwood <david.sherwood@arm.com>
3993 * combine.c (combine_simplify_rtx): Add checks for
3994 is_a <scalar_int_mode>.
3995 (simplify_if_then_else): Likewise.
3996 (make_field_assignment): Likewise.
3997 (simplify_comparison): Likewise.
3998 * ifcvt.c (noce_try_bitop): Likewise.
3999 * loop-invariant.c (canonicalize_address_mult): Likewise.
4000 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4003 Alan Hayward <alan.hayward@arm.com>
4004 David Sherwood <david.sherwood@arm.com>
4006 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
4007 is_a <scalar_int_mode> instead of != BLKmode.
4009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4010 Alan Hayward <alan.hayward@arm.com>
4011 David Sherwood <david.sherwood@arm.com>
4013 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
4014 instead of != VOIDmode.
4015 * combine.c (if_then_else_cond): Likewise.
4016 (change_zero_ext): Likewise.
4017 * dwarf2out.c (mem_loc_descriptor): Likewise.
4018 (loc_descriptor): Likewise.
4019 * rtlanal.c (canonicalize_condition): Likewise.
4020 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
4022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4023 Alan Hayward <alan.hayward@arm.com>
4024 David Sherwood <david.sherwood@arm.com>
4026 * simplify-rtx.c (simplify_binary_operation_1): Use
4027 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
4029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4030 Alan Hayward <alan.hayward@arm.com>
4031 David Sherwood <david.sherwood@arm.com>
4033 * wide-int.h (int_traits<unsigned char>) New class.
4034 (int_traits<unsigned short>) Likewise.
4035 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
4036 Use GET_MODE_UNIT_PRECISION and remove redundant test for
4038 * combine.c (set_nonzero_bits_and_sign_copies): Use
4039 is_a <scalar_int_mode>.
4040 (find_split_point): Likewise.
4041 (combine_simplify_rtx): Likewise.
4042 (simplify_logical): Likewise.
4043 (expand_compound_operation): Likewise.
4044 (expand_field_assignment): Likewise.
4045 (make_compound_operation): Likewise.
4046 (extended_count): Likewise.
4047 (change_zero_ext): Likewise.
4048 (simplify_comparison): Likewise.
4049 * dwarf2out.c (scompare_loc_descriptor): Likewise.
4050 (ucompare_loc_descriptor): Likewise.
4051 (minmax_loc_descriptor): Likewise.
4052 (mem_loc_descriptor): Likewise.
4053 (loc_descriptor): Likewise.
4054 * expmed.c (init_expmed_one_mode): Likewise.
4055 * lra-constraints.c (lra_constraint_offset): Likewise.
4056 * optabs.c (prepare_libcall_arg): Likewise.
4057 * postreload.c (move2add_note_store): Likewise.
4058 * reload.c (operands_match_p): Likewise.
4059 * rtl.h (load_extend_op): Likewise.
4060 * rtlhooks.c (gen_lowpart_general): Likewise.
4061 * simplify-rtx.c (simplify_truncation): Likewise.
4062 (simplify_unary_operation_1): Likewise.
4063 (simplify_binary_operation_1): Likewise.
4064 (simplify_const_binary_operation): Likewise.
4065 (simplify_const_relational_operation): Likewise.
4066 (simplify_subreg): Likewise.
4067 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4068 * var-tracking.c (adjust_mems): Likewise.
4069 (prepare_call_arguments): Likewise.
4071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4072 Alan Hayward <alan.hayward@arm.com>
4073 David Sherwood <david.sherwood@arm.com>
4075 * machmode.h (is_int_mode): New fuction.
4076 * combine.c (find_split_point): Use it.
4077 (combine_simplify_rtx): Likewise.
4078 (simplify_if_then_else): Likewise.
4079 (simplify_set): Likewise.
4080 (simplify_shift_const_1): Likewise.
4081 (simplify_comparison): Likewise.
4082 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
4083 * cse.c (notreg_cost): Likewise.
4084 (cse_insn): Likewise.
4085 * cselib.c (cselib_lookup_1): Likewise.
4086 * dojump.c (do_jump_1): Likewise.
4087 (do_compare_rtx_and_jump): Likewise.
4088 * dse.c (get_call_args): Likewise.
4089 * dwarf2out.c (rtl_for_decl_init): Likewise.
4090 (native_encode_initializer): Likewise.
4091 * expmed.c (emit_store_flag_1): Likewise.
4092 (emit_store_flag): Likewise.
4093 * expr.c (convert_modes): Likewise.
4094 (store_field): Likewise.
4095 (expand_expr_real_1): Likewise.
4096 * fold-const.c (fold_read_from_constant_string): Likewise.
4097 * gimple-ssa-sprintf.c (get_format_string): Likewise.
4098 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
4099 * optabs.c (expand_binop): Likewise.
4100 (expand_unop): Likewise.
4101 (expand_abs_nojump): Likewise.
4102 (expand_one_cmpl_abs_nojump): Likewise.
4103 * simplify-rtx.c (mode_signbit_p): Likewise.
4104 (val_signbit_p): Likewise.
4105 (val_signbit_known_set_p): Likewise.
4106 (val_signbit_known_clear_p): Likewise.
4107 (simplify_relational_operation_1): Likewise.
4108 * tree.c (vector_type_mode): Likewise.
4110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4111 Alan Hayward <alan.hayward@arm.com>
4112 David Sherwood <david.sherwood@arm.com>
4114 * machmode.h (smallest_mode_for_size): Fix formatting.
4115 (smallest_int_mode_for_size): New function.
4116 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
4117 instead of smallest_mode_for_size.
4118 * combine.c (make_extraction): Likewise.
4119 * config/arc/arc.c (arc_expand_movmem): Likewise.
4120 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
4121 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4122 * config/s390/s390.c (s390_expand_insv): Likewise.
4123 * config/sparc/sparc.c (assign_int_registers): Likewise.
4124 * config/spu/spu.c (spu_function_value): Likewise.
4125 (spu_function_arg): Likewise.
4126 * coverage.c (get_gcov_type): Likewise.
4127 (get_gcov_unsigned_t): Likewise.
4128 * dse.c (find_shift_sequence): Likewise.
4129 * expmed.c (store_bit_field_1): Likewise.
4130 * expr.c (convert_move): Likewise.
4131 (store_field): Likewise.
4132 * internal-fn.c (expand_arith_overflow): Likewise.
4133 * optabs-query.c (get_best_extraction_insn): Likewise.
4134 * optabs.c (expand_twoval_binop_libfunc): Likewise.
4135 * stor-layout.c (layout_type): Likewise.
4136 (initialize_sizetypes): Likewise.
4137 * targhooks.c (default_get_mask_mode): Likewise.
4138 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
4140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4141 Alan Hayward <alan.hayward@arm.com>
4142 David Sherwood <david.sherwood@arm.com>
4144 * machmode.h (opt_mode::else_blk): New function.
4145 (int_mode_for_mode): Declare.
4146 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
4147 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
4149 * cfgexpand.c (expand_debug_expr): Likewise.
4150 * combine.c (gen_lowpart_or_truncate): Likewise.
4151 (gen_lowpart_for_combine): Likewise.
4152 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
4153 * config/avr/avr.c (avr_to_int_mode): Likewise.
4154 (avr_out_plus_1): Likewise.
4155 (avr_out_plus): Likewise.
4156 (avr_out_round): Likewise.
4157 * config/i386/i386.c (ix86_split_to_parts): Likewise.
4158 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4159 (s390_expand_vcond): Likewise.
4160 * config/spu/spu.c (spu_split_immediate): Likewise.
4161 (spu_expand_mov): Likewise.
4162 * dse.c (get_stored_val): Likewise.
4163 * expmed.c (store_bit_field_1): Likewise.
4164 (convert_extracted_bit_field): Use int_mode_for_mode instead of
4166 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
4167 (extract_low_bits): Likewise.
4168 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
4169 handling rather than repeating the check.
4170 (emit_group_store): Likewise.
4171 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
4172 * optabs.c (expand_absneg_bit): Likewise.
4173 (expand_copysign_absneg): Likewise.
4174 (expand_copysign_bit): Likewise.
4175 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
4176 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
4177 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
4178 * var-tracking.c (prepare_call_arguments): Likewise.
4179 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
4180 int_mode_for_mode instead of mode_for_size.
4181 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4184 Alan Hayward <alan.hayward@arm.com>
4185 David Sherwood <david.sherwood@arm.com>
4187 * machmode.h (int_mode_for_size): New function.
4188 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
4189 instead of mode_for_size.
4190 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
4192 * combine.c (expand_field_assignment): Use int_mode_for_size
4193 instead of mode_for_size.
4194 (make_extraction): Likewise.
4195 (simplify_shift_const_1): Likewise.
4196 (simplify_comparison): Likewise.
4197 * dojump.c (do_jump): Likewise.
4198 * dwarf2out.c (mem_loc_descriptor): Likewise.
4199 * emit-rtl.c (init_derived_machine_modes): Likewise.
4200 * expmed.c (flip_storage_order): Likewise.
4201 (convert_extracted_bit_field): Likewise.
4202 * expr.c (copy_blkmode_from_reg): Likewise.
4203 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
4204 * internal-fn.c (expand_mul_overflow): Likewise.
4205 * lower-subreg.c (simple_move): Likewise.
4206 * optabs-libfuncs.c (init_optabs): Likewise.
4207 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4208 * tree.c (vector_type_mode): Likewise.
4209 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
4210 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
4211 * tree-vect-generic.c (expand_vector_parallel): Likewise.
4212 * tree-vect-stmts.c (vectorizable_load): Likewise.
4213 (vectorizable_store): Likewise.
4215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4216 Alan Hayward <alan.hayward@arm.com>
4217 David Sherwood <david.sherwood@arm.com>
4219 * coretypes.h (pod_mode): New type.
4220 (scalar_int_mode_pod): New typedef.
4221 * machmode.h (pod_mode): New class.
4222 (int_n_data_t::m): Change type to scalar_int_mode_pod.
4223 * genmodes.c (emit_mode_int_n): Update accordingly.
4224 * lower-subreg.h (target_lower_subreg): Change type to
4225 scalar_int_mode_pod.
4226 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
4227 scalar_int_mode_pod.
4229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4230 Alan Hayward <alan.hayward@arm.com>
4231 David Sherwood <david.sherwood@arm.com>
4233 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
4234 machine_mode to scalar_int_mode.
4235 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
4236 (rs6000_option_override_internal): Remove cast to int.
4237 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
4238 machine_mode to scalar_int_mode.
4239 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
4240 (rs6000_option_override_internal): Remove cast to int.
4241 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
4242 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
4244 * config/s390/s390.c (s390_expand_builtin): Likewise.
4245 * coretypes.h (scalar_int_mode): New type.
4246 (opt_scalar_int_mode): New typedef.
4247 * machmode.h (scalar_int_mode): New class.
4248 (scalar_int_mode::includes_p): New function.
4249 (byte_mode): Change type to scalar_int_mode.
4250 (word_mode): Likewise.
4251 (ptr_mode): Likewise.
4252 * emit-rtl.c (byte_mode): Likewise.
4253 (word_mode): Likewise.
4254 (ptr_mode): Likewise.
4255 (init_derived_machine_modes): Update accordingly.
4256 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
4257 and MODE_PARTIAL_INT.
4258 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
4259 opt_scalar_int_mode.
4261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4262 Alan Hayward <alan.hayward@arm.com>
4263 David Sherwood <david.sherwood@arm.com>
4265 * target.def (libgcc_floating_mode_supported_p): Take a
4267 * doc/tm.texi: Regenerate.
4268 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
4270 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4271 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
4274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4275 Alan Hayward <alan.hayward@arm.com>
4276 David Sherwood <david.sherwood@arm.com>
4278 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
4279 * doc/tm.texi: Regenerate.
4280 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
4281 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
4282 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
4283 * targhooks.h (default_floatn_mode): Likewise.
4284 * targhooks.c (default_floatn_mode): Likewise.
4285 * tree.c (build_common_tree_nodes): Update accordingly.
4287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4288 Alan Hayward <alan.hayward@arm.com>
4289 David Sherwood <david.sherwood@arm.com>
4291 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
4292 (mode_iterator::iterate_p): Likewise.
4293 (mode_iterator::get_wider): Likewise.
4294 * expr.c (init_expr_target): Use opt_scalar_float_mode.
4296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4297 Alan Hayward <alan.hayward@arm.com>
4298 David Sherwood <david.sherwood@arm.com>
4300 * coretypes.h (opt_scalar_float_mode): New typedef.
4301 * machmode.h (float_mode_for_size): New function.
4302 * emit-rtl.c (double_mode): Delete.
4303 (init_emit_once): Use float_mode_for_size.
4304 * stor-layout.c (layout_type): Likewise.
4305 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
4307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4308 Alan Hayward <alan.hayward@arm.com>
4309 David Sherwood <david.sherwood@arm.com>
4311 * output.h (assemble_real): Take a scalar_float_mode.
4312 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
4313 * config/arm/arm.md (consttable_4): Likewise.
4314 (consttable_8): Likewise.
4315 (consttable_16): Likewise.
4316 * config/mips/mips.md (consttable_float): Likewise.
4317 * config/s390/s390.c (s390_output_pool_entry): Likewise.
4318 * varasm.c (assemble_real): Take a scalar_float_mode.
4319 (output_constant_pool_2): Update accordingly.
4320 (output_constant): Likewise.
4322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4323 Alan Hayward <alan.hayward@arm.com>
4324 David Sherwood <david.sherwood@arm.com>
4326 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
4327 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
4328 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
4329 (native_encode_real): Likewise.
4330 (native_interpret_real): Likewise.
4331 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
4332 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4335 Alan Hayward <alan.hayward@arm.com>
4336 David Sherwood <david.sherwood@arm.com>
4338 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
4339 <scalar_float_mode>. Simplify.
4340 (gen_extend_conv_libfunc): Likewise.
4342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4343 Alan Hayward <alan.hayward@arm.com>
4344 David Sherwood <david.sherwood@arm.com>
4346 * coretypes.h (scalar_float_mode): New type.
4347 * machmode.h (mode_traits::from_int): Use machine_mode if
4348 USE_ENUM_MODES is defined.
4349 (is_a): New function.
4351 (dyn_cast): Likewise.
4352 (scalar_float_mode): New class.
4353 (scalar_float_mode::includes_p): New function.
4354 (is_float_mode): Likewise.
4355 * gdbhooks.py (MachineModePrinter): New class.
4356 (build_pretty_printer): Use it for scalar_float_mode.
4357 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
4358 (format_helper::format_helper): Turn into a template.
4359 * genmodes.c (get_mode_class): New function.
4360 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
4361 or machine_mode if none.
4362 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
4363 as_a <scalar_float_mode>.
4364 * dwarf2out.c (mem_loc_descriptor): Likewise.
4365 (insert_float): Likewise.
4366 (add_const_value_attribute): Likewise.
4367 * simplify-rtx.c (simplify_immed_subreg): Likewise.
4368 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
4369 (expand_unop): Update accordingly.
4370 (expand_abs_nojump): Likewise.
4371 (expand_copysign_absneg): Take a scalar_float_mode.
4372 (expand_copysign_bit): Likewise.
4373 (expand_copysign): Update accordingly.
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 * coretypes.h (opt_mode): New class.
4380 * machmode.h (opt_mode): Likewise.
4381 (opt_mode::else_void): New function.
4382 (opt_mode::require): Likewise.
4383 (opt_mode::exists): Likewise.
4384 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
4385 (GET_MODE_2XWIDER_MODE): Likewise.
4386 (mode_iterator::get_wider): Update accordingly.
4387 (mode_iterator::get_2xwider): Likewise.
4388 (mode_iterator::get_known_wider): Likewise, turning into a template.
4389 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
4390 forcing a wider mode to exist.
4391 * config/cr16/cr16.h (LONG_REG_P): Likewise.
4392 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4393 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
4394 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
4395 * lower-subreg.c (init_lower_subreg): Likewise.
4396 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
4397 on the final iteration.
4398 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
4399 a wider mode exists before asking for a move pattern.
4400 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
4401 forcing a wider mode to exist.
4402 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
4403 returning false if no such mode exists.
4404 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
4405 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
4406 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
4407 Avoid checking for a MODE_INT if we already know the mode is not a
4409 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
4410 forcing a wider mode to exist.
4411 (expmed_mult_highpart_optab): Likewise.
4412 (expmed_mult_highpart): Likewise.
4413 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
4415 * lto-streamer-in.c (lto_input_mode_table): Likewise.
4416 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
4417 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4418 * internal-fn.c (expand_mul_overflow): Update use of
4419 GET_MODE_2XWIDER_MODE.
4420 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4421 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
4422 GET_MODE_WIDER_MODE.
4423 (convert_plusminus_to_widen): Likewise.
4424 * tree-switch-conversion.c (array_value_type): Likewise.
4425 * var-tracking.c (emit_note_insn_var_location): Likewise.
4426 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
4427 Return false inside rather than outside the loop if no wider mode
4429 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
4430 and GET_MODE_2XWIDER_MODE
4431 (can_compare_p): Use else_void.
4432 * gdbhooks.py (OptMachineModePrinter): New class.
4433 (build_pretty_printer): Use it for opt_mode.
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 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
4440 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
4442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4443 Alan Hayward <alan.hayward@arm.com>
4444 David Sherwood <david.sherwood@arm.com>
4446 * machmode.h (mode_traits): New structure.
4447 (get_narrowest_mode): New function.
4448 (mode_iterator::start): Likewise.
4449 (mode_iterator::iterate_p): Likewise.
4450 (mode_iterator::get_wider): Likewise.
4451 (mode_iterator::get_known_wider): Likewise.
4452 (mode_iterator::get_2xwider): Likewise.
4453 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
4454 (FOR_EACH_MODE): Likewise.
4455 (FOR_EACH_MODE_FROM): Likewise.
4456 (FOR_EACH_MODE_UNTIL): Likewise.
4457 (FOR_EACH_WIDER_MODE): Likewise.
4458 (FOR_EACH_2XWIDER_MODE): Likewise.
4459 * builtins.c (expand_builtin_strlen): Use new mode iterators.
4460 * combine.c (simplify_comparison): Likewise
4461 * config/i386/i386.c (type_natural_mode): Likewise.
4462 * cse.c (cse_insn): Likewise.
4463 * dse.c (find_shift_sequence): Likewise.
4464 * emit-rtl.c (init_derived_machine_modes): Likewise.
4465 (init_emit_once): Likewise.
4466 * explow.c (hard_function_value): Likewise.
4467 * expmed.c (extract_fixed_bit_field_1): Likewise.
4468 (extract_bit_field_1): Likewise.
4469 (expand_divmod): Likewise.
4470 (emit_store_flag_1): Likewise.
4471 * expr.c (init_expr_target): Likewise.
4472 (convert_move): Likewise.
4473 (alignment_for_piecewise_move): Likewise.
4474 (widest_int_mode_for_size): Likewise.
4475 (emit_block_move_via_movmem): Likewise.
4476 (copy_blkmode_to_reg): Likewise.
4477 (set_storage_via_setmem): Likewise.
4478 (compress_float_constant): Likewise.
4479 * omp-low.c (omp_clause_aligned_alignment): Likewise.
4480 * optabs-query.c (get_best_extraction_insn): Likewise.
4481 * optabs.c (expand_binop): Likewise.
4482 (expand_twoval_unop): Likewise.
4483 (expand_twoval_binop): Likewise.
4484 (widen_leading): Likewise.
4485 (widen_bswap): Likewise.
4486 (expand_parity): Likewise.
4487 (expand_unop): Likewise.
4488 (prepare_cmp_insn): Likewise.
4489 (prepare_float_lib_cmp): Likewise.
4490 (expand_float): Likewise.
4491 (expand_fix): Likewise.
4492 (expand_sfix_optab): Likewise.
4493 * postreload.c (move2add_use_add2_insn): Likewise.
4494 * reg-stack.c (reg_to_stack): Likewise.
4495 * reginfo.c (choose_hard_reg_mode): Likewise.
4496 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
4497 * stor-layout.c (mode_for_size): Likewise.
4498 (smallest_mode_for_size): Likewise.
4499 (mode_for_vector): Likewise.
4500 (finish_bitfield_representative): Likewise.
4501 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
4502 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
4503 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4504 * var-tracking.c (prepare_call_arguments): Likewise.
4506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4507 Alan Hayward <alan.hayward@arm.com>
4508 David Sherwood <david.sherwood@arm.com>
4510 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
4511 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
4512 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
4513 * machmode.h (mode_size): Move earlier in file.
4514 (mode_precision): Likewise.
4515 (mode_inner): Likewise.
4516 (mode_nunits): Likewise.
4517 (mode_unit_size): Likewise.
4518 (unit_unit_precision): Likewise.
4519 (mode_wider): Likewise.
4520 (mode_2xwider): Likewise.
4521 (machine_mode): New class.
4522 (mode_to_bytes): New function.
4523 (mode_to_bits): Likewise.
4524 (mode_to_precision): Likewise.
4525 (mode_to_inner): Likewise.
4526 (mode_to_unit_size): Likewise.
4527 (mode_to_unit_precision): Likewise.
4528 (mode_to_nunits): Likewise.
4529 (GET_MODE_SIZE): Use mode_to_bytes.
4530 (GET_MODE_BITSIZE): Use mode_to_bits.
4531 (GET_MODE_PRECISION): Use mode_to_precision.
4532 (GET_MODE_INNER): Use mode_to_inner.
4533 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
4534 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
4535 (GET_MODE_NUNITS): Use mode_to_nunits.
4536 * system.h (ALWAYS_INLINE): New macro.
4537 * config/powerpcspe/powerpcspe-c.c
4538 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
4539 int for arg1_mode and arg2_mode.
4541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4542 Alan Hayward <alan.hayward@arm.com>
4543 David Sherwood <david.sherwood@arm.com>
4545 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
4546 Prefix mode names with E_ in case statements.
4547 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4548 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
4549 (aarch64_split_simd_move): Likewise.
4550 (aarch64_gen_storewb_pair): Likewise.
4551 (aarch64_gen_loadwb_pair): Likewise.
4552 (aarch64_gen_store_pair): Likewise.
4553 (aarch64_gen_load_pair): Likewise.
4554 (aarch64_get_condition_code_1): Likewise.
4555 (aarch64_constant_pool_reload_icode): Likewise.
4556 (get_rsqrte_type): Likewise.
4557 (get_rsqrts_type): Likewise.
4558 (get_recpe_type): Likewise.
4559 (get_recps_type): Likewise.
4560 (aarch64_gimplify_va_arg_expr): Likewise.
4561 (aarch64_simd_container_mode): Likewise.
4562 (aarch64_emit_load_exclusive): Likewise.
4563 (aarch64_emit_store_exclusive): Likewise.
4564 (aarch64_expand_compare_and_swap): Likewise.
4565 (aarch64_gen_atomic_cas): Likewise.
4566 (aarch64_emit_bic): Likewise.
4567 (aarch64_emit_atomic_swap): Likewise.
4568 (aarch64_emit_atomic_load_op): Likewise.
4569 (aarch64_evpc_trn): Likewise.
4570 (aarch64_evpc_uzp): Likewise.
4571 (aarch64_evpc_zip): Likewise.
4572 (aarch64_evpc_ext): Likewise.
4573 (aarch64_evpc_rev): Likewise.
4574 (aarch64_evpc_dup): Likewise.
4575 (aarch64_gen_ccmp_first): Likewise.
4576 (aarch64_gen_ccmp_next): Likewise.
4577 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
4578 (alpha_emit_xfloating_libcall): Likewise.
4579 (emit_insxl): Likewise.
4580 (alpha_arg_type): Likewise.
4581 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
4582 (arc_preferred_simd_mode): Likewise.
4583 (arc_secondary_reload): Likewise.
4584 (get_arc_condition_code): Likewise.
4585 (arc_print_operand): Likewise.
4586 (arc_legitimate_constant_p): Likewise.
4587 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4588 * config/arc/arc.md (casesi_load): Likewise.
4589 (casesi_compact_jump): Likewise.
4590 * config/arc/predicates.md (proper_comparison_operator): Likewise.
4591 (cc_use_register): Likewise.
4592 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4593 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
4594 (arm_init_iwmmxt_builtins): Likewise.
4595 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
4596 (neon_expand_vector_init): Likewise.
4597 (arm_attr_length_move_neon): Likewise.
4598 (maybe_get_arm_condition_code): Likewise.
4599 (arm_emit_vector_const): Likewise.
4600 (arm_preferred_simd_mode): Likewise.
4601 (arm_output_iwmmxt_tinsr): Likewise.
4602 (thumb1_output_casesi): Likewise.
4603 (thumb2_output_casesi): Likewise.
4604 (arm_emit_load_exclusive): Likewise.
4605 (arm_emit_store_exclusive): Likewise.
4606 (arm_expand_compare_and_swap): Likewise.
4607 (arm_evpc_neon_vuzp): Likewise.
4608 (arm_evpc_neon_vzip): Likewise.
4609 (arm_evpc_neon_vrev): Likewise.
4610 (arm_evpc_neon_vtrn): Likewise.
4611 (arm_evpc_neon_vext): Likewise.
4612 (arm_validize_comparison): Likewise.
4613 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
4614 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
4615 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
4616 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
4617 (c6x_preferred_simd_mode): Likewise.
4618 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
4619 (epiphany_rtx_costs): Likewise.
4620 * config/epiphany/predicates.md (proper_comparison_operator):
4622 * config/frv/frv.c (condexec_memory_operand): Likewise.
4623 (frv_emit_move): Likewise.
4624 (output_move_single): Likewise.
4625 (output_condmove_single): Likewise.
4626 (frv_hard_regno_mode_ok): Likewise.
4627 (frv_matching_accg_mode): Likewise.
4628 * config/h8300/h8300.c (split_adds_subs): Likewise.
4629 (h8300_rtx_costs): Likewise.
4630 (h8300_print_operand): Likewise.
4631 (compute_mov_length): Likewise.
4632 (output_logical_op): Likewise.
4633 (compute_logical_op_length): Likewise.
4634 (compute_logical_op_cc): Likewise.
4635 (h8300_shift_needs_scratch_p): Likewise.
4636 (output_a_shift): Likewise.
4637 (compute_a_shift_length): Likewise.
4638 (compute_a_shift_cc): Likewise.
4639 (expand_a_rotate): Likewise.
4640 (output_a_rotate): Likewise.
4641 * config/i386/i386.c (classify_argument): Likewise.
4642 (function_arg_advance_32): Likewise.
4643 (function_arg_32): Likewise.
4644 (function_arg_64): Likewise.
4645 (function_value_64): Likewise.
4646 (ix86_gimplify_va_arg): Likewise.
4647 (ix86_legitimate_constant_p): Likewise.
4648 (put_condition_code): Likewise.
4649 (split_double_mode): Likewise.
4650 (ix86_avx256_split_vector_move_misalign): Likewise.
4651 (ix86_expand_vector_logical_operator): Likewise.
4652 (ix86_split_idivmod): Likewise.
4653 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
4654 (ix86_build_const_vector): Likewise.
4655 (ix86_build_signbit_mask): Likewise.
4656 (ix86_match_ccmode): Likewise.
4657 (ix86_cc_modes_compatible): Likewise.
4658 (ix86_expand_branch): Likewise.
4659 (ix86_expand_sse_cmp): Likewise.
4660 (ix86_expand_sse_movcc): Likewise.
4661 (ix86_expand_int_sse_cmp): Likewise.
4662 (ix86_expand_vec_perm_vpermi2): Likewise.
4663 (ix86_expand_vec_perm): Likewise.
4664 (ix86_expand_sse_unpack): Likewise.
4665 (ix86_expand_int_addcc): Likewise.
4666 (ix86_split_to_parts): Likewise.
4667 (ix86_vectorize_builtin_gather): Likewise.
4668 (ix86_vectorize_builtin_scatter): Likewise.
4669 (avx_vpermilp_parallel): Likewise.
4670 (inline_memory_move_cost): Likewise.
4671 (ix86_tieable_integer_mode_p): Likewise.
4672 (x86_maybe_negate_const_int): Likewise.
4673 (ix86_expand_vector_init_duplicate): Likewise.
4674 (ix86_expand_vector_init_one_nonzero): Likewise.
4675 (ix86_expand_vector_init_one_var): Likewise.
4676 (ix86_expand_vector_init_concat): Likewise.
4677 (ix86_expand_vector_init_interleave): Likewise.
4678 (ix86_expand_vector_init_general): Likewise.
4679 (ix86_expand_vector_set): Likewise.
4680 (ix86_expand_vector_extract): Likewise.
4681 (emit_reduc_half): Likewise.
4682 (ix86_emit_i387_round): Likewise.
4683 (ix86_mangle_type): Likewise.
4684 (ix86_expand_round_sse4): Likewise.
4685 (expand_vec_perm_blend): Likewise.
4686 (canonicalize_vector_int_perm): Likewise.
4687 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
4688 (expand_vec_perm_1): Likewise.
4689 (expand_vec_perm_interleave3): Likewise.
4690 (expand_vec_perm_even_odd_pack): Likewise.
4691 (expand_vec_perm_even_odd_1): Likewise.
4692 (expand_vec_perm_broadcast_1): Likewise.
4693 (ix86_vectorize_vec_perm_const_ok): Likewise.
4694 (ix86_expand_vecop_qihi): Likewise.
4695 (ix86_expand_mul_widen_hilo): Likewise.
4696 (ix86_expand_sse2_abs): Likewise.
4697 (ix86_expand_pextr): Likewise.
4698 (ix86_expand_pinsr): Likewise.
4699 (ix86_preferred_simd_mode): Likewise.
4700 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
4701 * config/i386/sse.md (*andnot<mode>3): Likewise.
4702 (<mask_codefor><code><mode>3<mask_name>): Likewise.
4703 (*<code><mode>3): Likewise.
4704 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
4705 (ia64_expand_atomic_op): Likewise.
4706 (ia64_arg_type): Likewise.
4707 (ia64_mode_to_int): Likewise.
4708 (ia64_scalar_mode_supported_p): Likewise.
4709 (ia64_vector_mode_supported_p): Likewise.
4710 (expand_vec_perm_broadcast): Likewise.
4711 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
4712 (iq2000_function_arg_advance): Likewise.
4713 (iq2000_function_arg): Likewise.
4714 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
4715 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
4716 (m68k_libcall_value): Likewise.
4717 (m68k_function_value): Likewise.
4718 (sched_attr_op_type): Likewise.
4719 * config/mcore/mcore.c (mcore_output_move): Likewise.
4720 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
4722 (microblaze_function_arg): Likewise.
4723 * config/mips/mips.c (mips16_build_call_stub): Likewise.
4724 (mips_print_operand): Likewise.
4725 (mips_mode_ok_for_mov_fmt_p): Likewise.
4726 (mips_vector_mode_supported_p): Likewise.
4727 (mips_preferred_simd_mode): Likewise.
4728 (mips_expand_vpc_loongson_even_odd): Likewise.
4729 (mips_expand_vec_unpack): Likewise.
4730 (mips_expand_vi_broadcast): Likewise.
4731 (mips_expand_vector_init): Likewise.
4732 (mips_expand_vec_reduc): Likewise.
4733 (mips_expand_msa_cmp): Likewise.
4734 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
4735 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
4736 (cc_flags_for_mode): Likewise.
4737 * config/msp430/msp430.c (msp430_print_operand): Likewise.
4738 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
4739 (nds32_output_casesi_pc_relative): Likewise.
4740 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4741 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
4742 (nvptx_gen_unpack): Likewise.
4743 (nvptx_gen_pack): Likewise.
4744 (nvptx_gen_shuffle): Likewise.
4745 (nvptx_gen_wcast): Likewise.
4746 (nvptx_preferred_simd_mode): Likewise.
4747 * config/pa/pa.c (pa_secondary_reload): Likewise.
4748 * config/pa/predicates.md (base14_operand): Likewise.
4749 * config/powerpcspe/powerpcspe-c.c
4750 (altivec_resolve_overloaded_builtin): Likewise.
4751 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
4753 (rs6000_preferred_simd_mode): Likewise.
4754 (output_vec_const_move): Likewise.
4755 (rs6000_expand_vector_extract): Likewise.
4756 (rs6000_split_vec_extract_var): Likewise.
4757 (reg_offset_addressing_ok_p): Likewise.
4758 (rs6000_legitimate_offset_address_p): Likewise.
4759 (rs6000_legitimize_address): Likewise.
4760 (rs6000_emit_set_const): Likewise.
4761 (rs6000_const_vec): Likewise.
4762 (rs6000_emit_move): Likewise.
4763 (spe_build_register_parallel): Likewise.
4764 (rs6000_darwin64_record_arg_recurse): Likewise.
4765 (swap_selector_for_mode): Likewise.
4766 (spe_init_builtins): Likewise.
4767 (paired_init_builtins): Likewise.
4768 (altivec_init_builtins): Likewise.
4769 (do_load_for_compare): Likewise.
4770 (rs6000_generate_compare): Likewise.
4771 (rs6000_expand_float128_convert): Likewise.
4772 (emit_load_locked): Likewise.
4773 (emit_store_conditional): Likewise.
4774 (rs6000_output_function_epilogue): Likewise.
4775 (rs6000_handle_altivec_attribute): Likewise.
4776 (rs6000_function_value): Likewise.
4777 (emit_fusion_gpr_load): Likewise.
4778 (emit_fusion_p9_load): Likewise.
4779 (emit_fusion_p9_store): Likewise.
4780 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
4781 (fusion_gpr_mem_load): Likewise.
4782 (fusion_addis_mem_combo_load): Likewise.
4783 (fusion_addis_mem_combo_store): Likewise.
4784 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
4785 (fusion_gpr_mem_load): Likewise.
4786 (fusion_addis_mem_combo_load): Likewise.
4787 (fusion_addis_mem_combo_store): Likewise.
4788 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4790 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
4791 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4792 (rs6000_preferred_simd_mode): Likewise.
4793 (output_vec_const_move): Likewise.
4794 (rs6000_expand_vector_extract): Likewise.
4795 (rs6000_split_vec_extract_var): Likewise.
4796 (reg_offset_addressing_ok_p): Likewise.
4797 (rs6000_legitimate_offset_address_p): Likewise.
4798 (rs6000_legitimize_address): Likewise.
4799 (rs6000_emit_set_const): Likewise.
4800 (rs6000_const_vec): Likewise.
4801 (rs6000_emit_move): Likewise.
4802 (rs6000_darwin64_record_arg_recurse): Likewise.
4803 (swap_selector_for_mode): Likewise.
4804 (paired_init_builtins): Likewise.
4805 (altivec_init_builtins): Likewise.
4806 (rs6000_expand_float128_convert): Likewise.
4807 (emit_load_locked): Likewise.
4808 (emit_store_conditional): Likewise.
4809 (rs6000_output_function_epilogue): Likewise.
4810 (rs6000_handle_altivec_attribute): Likewise.
4811 (rs6000_function_value): Likewise.
4812 (emit_fusion_gpr_load): Likewise.
4813 (emit_fusion_p9_load): Likewise.
4814 (emit_fusion_p9_store): Likewise.
4815 * config/rx/rx.c (rx_gen_move_template): Likewise.
4816 (flags_from_mode): Likewise.
4817 * config/s390/predicates.md (s390_alc_comparison): Likewise.
4818 (s390_slb_comparison): Likewise.
4819 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
4820 (s390_vector_mode_supported_p): Likewise.
4821 (s390_cc_modes_compatible): Likewise.
4822 (s390_match_ccmode_set): Likewise.
4823 (s390_canonicalize_comparison): Likewise.
4824 (s390_emit_compare_and_swap): Likewise.
4825 (s390_branch_condition_mask): Likewise.
4826 (s390_rtx_costs): Likewise.
4827 (s390_secondary_reload): Likewise.
4828 (__SECONDARY_RELOAD_CASE): Likewise.
4829 (s390_expand_cs): Likewise.
4830 (s390_preferred_simd_mode): Likewise.
4831 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
4832 * config/sh/sh.c (sh_print_operand): Likewise.
4833 (dump_table): Likewise.
4834 (sh_secondary_reload): Likewise.
4835 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4836 * config/sh/sh.md (casesi_worker_1): Likewise.
4837 (casesi_worker_2): Likewise.
4838 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
4839 (fcc_comparison_operator): Likewise.
4840 * config/sparc/sparc.c (sparc_expand_move): Likewise.
4841 (emit_soft_tfmode_cvt): Likewise.
4842 (sparc_preferred_simd_mode): Likewise.
4843 (output_cbranch): Likewise.
4844 (sparc_print_operand): Likewise.
4845 (sparc_expand_vec_perm_bmask): Likewise.
4846 (vector_init_bshuffle): Likewise.
4847 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
4848 (spu_vector_mode_supported_p): Likewise.
4849 (spu_expand_insv): Likewise.
4850 (spu_emit_branch_or_set): Likewise.
4851 (spu_handle_vector_attribute): Likewise.
4852 (spu_builtin_splats): Likewise.
4853 (spu_builtin_extract): Likewise.
4854 (spu_builtin_promote): Likewise.
4855 (spu_expand_sign_extend): Likewise.
4856 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
4857 (tilegx_simd_int): Likewise.
4858 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
4859 (tilepro_simd_int): Likewise.
4860 * config/v850/v850.c (const_double_split): Likewise.
4861 (v850_print_operand): Likewise.
4862 (ep_memory_offset): Likewise.
4863 * config/vax/vax.c (vax_rtx_costs): Likewise.
4864 (vax_output_int_move): Likewise.
4865 (vax_output_int_add): Likewise.
4866 (vax_output_int_subtract): Likewise.
4867 * config/visium/predicates.md (visium_branch_operator): Likewise.
4868 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
4869 (visium_print_operand_address): Likewise.
4870 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
4871 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
4872 (xtensa_expand_conditional_branch): Likewise.
4873 (xtensa_copy_incoming_a7): Likewise.
4874 (xtensa_output_literal): Likewise.
4875 * dfp.c (decimal_real_maxval): Likewise.
4876 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
4878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4879 Alan Hayward <alan.hayward@arm.com>
4880 David Sherwood <david.sherwood@arm.com>
4882 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
4883 (mode_nunits_inline): Likewise.
4884 (mode_inner_inline): Likewise.
4885 (mode_unit_size_inline): Likewise.
4886 (mode_unit_precision_inline): Likewise.
4887 (emit_insn_modes_h): Likewise. Also emit a #define of the
4889 (emit_mode_wider): Add an E_ prefix to mode names.
4890 (emit_mode_complex): Likewise.
4891 (emit_mode_inner): Likewise.
4892 (emit_mode_adjustments): Likewise.
4893 (emit_mode_int_n): Likewise.
4894 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
4895 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
4896 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
4897 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
4898 (CRC32_BUILTIN, ENTRY): Likewise.
4899 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
4900 (aarch64_pop_regs): Likewise.
4901 (aarch64_process_components): Likewise.
4902 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
4903 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
4904 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
4905 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
4906 * config/arm/arm.c (arm_init_libfuncs): Likewise.
4907 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
4909 * config/i386/i386-builtin.def (pcmpestr): Likewise.
4910 (pcmpistr): Likewise.
4911 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
4912 * config/mmix/mmix.c (mmix_output_condition): Likewise.
4913 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
4915 * config/rl78/rl78.c (mduc_regs): Likewise.
4916 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4917 (htm_expand_builtin): Likewise.
4918 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
4919 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
4920 * config/xtensa/xtensa.c (print_operand): Likewise.
4921 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
4922 (NUM_MODE_VECTOR_INT): Likewise.
4923 * genoutput.c (null_operand): Likewise.
4924 (output_operand_data): Likewise.
4925 * genrecog.c (print_parameter_value): Likewise.
4926 * lra.c (debug_operand_data): Likewise.
4928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4930 * dwarf2out.c (scompare_loc_descriptor_wide)
4931 (scompare_loc_descriptor_narrow): New functions, split out from...
4932 (scompare_loc_descriptor): ...here.
4933 * expmed.c (emit_store_flag_int): New function, split out from...
4934 (emit_store_flag): ...here.
4936 2017-08-30 Richard Biener <rguenther@suse.de>
4938 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
4939 (dwarf2out_early_finish): Move setting of AT_pubnames from
4940 early debug output to early finish.
4942 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4944 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
4945 and -mdata-region to the assembler.
4947 2017-08-30 Richard Biener <rguenther@suse.de>
4949 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
4951 (gen_subprogram_die): Add DW_AT_object_pointer only early.
4952 (dwarf2out_early_global_decl): Only generate a DIE for the
4953 abstract origin if it doesn't already exist or is a declaration DIE.
4954 (resolve_addr): Do not add the linkage name twice when
4955 generating a stub DIE for the DW_TAG_GNU_call_site target.
4957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
4959 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4960 Use machine_mode rather than int for arg1_mode.
4962 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
4965 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
4966 that the second argument of the built-in functions to unpack
4967 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
4968 switch statement instead a lot of if statements.
4969 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
4970 Allow 64-bit values to be in Altivec registers as well as
4971 traditional floating point registers.
4972 (pack<mode>, FMOVE128_VSX iterator): Likewise.
4974 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
4976 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
4977 MAX_REGS_PER_ADDRESS == 1.
4979 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
4981 * config/i386/i386.opt (flag_fentry): Do not init to -1.
4982 * config/i386/i386.c (ix86_option_override_internal): Simplify
4983 setting of opts->x_flag_entry.
4985 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4986 Jakub Jelinek <jakub@redhat.com>
4987 Richard Biener <rguenther@suse.de>
4989 PR tree-optimization/81503
4990 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
4991 folded constant fits in the target type; reorder tests for clarity.
4993 2017-08-29 Martin Liska <mliska@suse.cz>
4995 * passes.def: Include pass_lower_switch.
4996 * stmt.c (dump_case_nodes): Remove and move to
4997 tree-switch-conversion.
4998 (case_values_threshold): Likewise.
4999 (expand_switch_as_decision_tree_p): Likewise.
5000 (emit_case_decision_tree): Likewise.
5001 (expand_case): Likewise.
5002 (balance_case_nodes): Likewise.
5003 (node_has_low_bound): Likewise.
5004 (node_has_high_bound): Likewise.
5005 (node_is_bounded): Likewise.
5006 (emit_case_nodes): Likewise.
5007 (struct simple_case_node): New struct.
5008 (add_case_node): Remove.
5009 (emit_case_dispatch_table): Use vector instead of case_list.
5010 (reset_out_edges_aux): Remove.
5011 (compute_cases_per_edge): Likewise.
5012 (expand_case): Build list of simple_case_node.
5013 (expand_sjlj_dispatch_table): Use it.
5014 * tree-switch-conversion.c (struct case_node): Moved from
5015 stmt.c and adjusted.
5016 (emit_case_nodes): Likewise.
5017 (node_has_low_bound): Likewise.
5018 (node_has_high_bound): Likewise.
5019 (node_is_bounded): Likewise.
5020 (case_values_threshold): Likewise.
5021 (reset_out_edges_aux): Likewise.
5022 (compute_cases_per_edge): Likewise.
5023 (add_case_node): Likewise.
5024 (dump_case_nodes): Likewise.
5025 (balance_case_nodes): Likewise.
5026 (expand_switch_as_decision_tree_p): Likewise.
5027 (emit_jump): Likewise.
5028 (emit_case_decision_tree): Likewise.
5029 (try_switch_expansion): Likewise.
5030 (do_jump_if_equal): Likewise.
5031 (emit_cmp_and_jump_insns): Likewise.
5032 (fix_phi_operands_for_edge): New function.
5033 (record_phi_operand_mapping): Likewise.
5034 (class pass_lower_switch): New pass.
5035 (pass_lower_switch::execute): New function.
5036 (make_pass_lower_switch): Likewise.
5037 (conditional_probability):
5038 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
5039 * tree-pass.h: Add make_pass_lower_switch.
5041 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5044 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
5047 2017-08-29 Richard Biener <rguenther@suse.de>
5049 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
5050 we do not add a DW_AT_inline attribute twice.
5051 (gen_subprogram_die): Remove code setting DW_AT_inline on
5052 DECL_ABSTRACT_P nodes.
5054 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
5056 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
5057 calls to internal functions.
5058 (gimplify_modify_expr): Likewise.
5059 * tree-call-cdce.c (use_internal_fn): Likewise.
5060 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5061 (convert_to_divmod): Set the nothrow flag.
5062 * tree-if-conv.c (predicate_mem_writes): Likewise.
5063 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5064 (vectorizable_call): Likewise.
5065 (vectorizable_store): Likewise.
5066 (vectorizable_load): Likewise.
5067 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
5068 (vect_recog_mask_conversion_pattern): Likewise.
5070 2017-08-29 Martin Liska <mliska@suse.cz>
5073 * gcc.c (driver_handle_option): Add new argument.
5074 * opts-common.c (handle_option): Pass
5075 target_option_override_hook.
5076 * opts-global.c (lang_handle_option): Add new option.
5077 (set_default_handlers): Add new argument.
5078 (decode_options): Likewise.
5079 * opts.c (target_handle_option): Likewise.
5080 (common_handle_option): Call target_option_override_hook.
5081 * opts.h (struct cl_option_handler_func): Add hook for
5082 target option override.
5083 (struct cl_option_handlers): Likewise.
5084 (set_default_handlers): Add new argument.
5085 (decode_options): Likewise.
5086 (common_handle_option): Likewise.
5087 (target_handle_option): Likewise.
5088 * toplev.c (toplev::main): Pass targetm.target_option.override
5091 2017-08-29 Richard Biener <rguenther@suse.de>
5092 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5094 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
5095 life to the active subtree.
5097 2017-08-28 Jeff Law <law@redhat.com>
5099 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
5100 derive_equivalences.
5101 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
5103 (edge_info::derive_equivalences): New private member function
5105 * tree-ssa-dom.c (class edge_info): Changed from a struct
5106 to a class. Add ctor/dtor, methods and data members.
5107 (edge_info::edge_info): Renamed from allocate_edge_info.
5108 Initialize additional members.
5109 (edge_info::~edge_info): New.
5110 (free_dom_edge_info): Delete the edge info.
5111 (record_edge_info): Use new class & associated member functions.
5112 Tighten forms for testing for edge equivalences.
5113 (record_temporary_equivalences): Iterate over the simple
5114 equivalences rather than assuming there's only one per edge.
5115 (cprop_into_successor_phis): Iterate over the simple
5116 equivalences rather than assuming there's only one per edge.
5117 (optimize_stmt): Use operand_equal_p rather than pointer
5118 equality for mini-DSE code.
5120 2017-08-28 Nathan Sidwell <nathan@acm.org>
5122 * gcc.c (execute): Fold SIGPIPE handling into switch
5123 statement. Adjust internal error message.
5125 2017-08-28 Richard Biener <rguenther@suse.de>
5128 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
5129 Do nothing for removed DIEs.
5131 2017-08-28 Richard Biener <rguenther@suse.de>
5133 PR tree-optimization/81977
5134 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
5137 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
5140 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
5141 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
5142 using targetm.gen_mem_thread_fence.
5144 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
5147 * config/i386/i386.md (*<btsc><mode>): Change operand 2
5148 predicate to register_operand. Reorder operands.
5149 (*btr<mode>): Ditto.
5150 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
5151 (*btr<mode>_mask): Ditto.
5153 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
5155 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
5156 * config/rs6000/xmmintrin.h: New file.
5157 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
5159 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5162 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
5163 parameter and_insn and return it.
5164 (recombine_lvx_pattern): Insert a copy to ensure availability of
5165 the base register of the copied masking operation at the point of
5166 the instruction replacement.
5167 (recombine_stvx_pattern): Likewise.
5169 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5171 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
5172 undocumented switches.
5173 (-mpower9-dform-vector): Likewise.
5174 (-mpower9-dform): Likewise.
5175 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
5176 comments to delete references to -mpower9-dform* switches.
5177 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
5178 Delete reference to -mpower9-dform* switches, test for
5179 -mpower9-vector instead.
5180 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
5181 (OTHER_P9_VECTOR_MASKS): Likewise.
5182 (POWERPC_MASKS): Likewise.
5183 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
5184 tests against -mpower9-dform* to -mpower9-vector. Delete code
5185 that checked for -mpower9-dform* consistancy with other options.
5186 Add test for -mpower9-misc to enable other power9 switches.
5187 (rs6000_init_hard_regno_mode_ok): Likewise.
5188 (rs6000_option_override_internal): Likewise.
5189 (rs6000_emit_prologue): Likewise.
5190 (rs6000_emit_epilogue): Likewise.
5191 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
5192 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
5193 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
5195 (emit_fusion_p9_store): Likewise.
5196 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
5197 resetting these macros if the assembler does not support ISA 3.0
5199 (TARGET_P9_DFORM_VECTOR): Likewise.
5200 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
5201 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
5203 2017-08-25 Alan Modra <amodra@gmail.com>
5205 PR rtl-optimization/81747
5206 * cse.c (cse_extended_basic_block): Don't attempt to record
5207 equivalences for degenerate conditional branches that branch
5208 to their fall-through.
5210 2017-08-24 Martin Sebor <msebor@redhat.com>
5213 * gimple-fold.c (size_must_be_zero_p): New function.
5214 (gimple_fold_builtin_memory_op): Call it.
5216 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
5218 * config/rs6000/mm_malloc.h: New file.
5220 2017-08-24 Bin Cheng <bin.cheng@arm.com>
5222 PR tree-optimization/81913
5223 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
5224 analysis when either IVs in condition can wrap.
5226 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
5228 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
5229 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
5231 2017-08-24 Richard Biener <rguenther@suse.de>
5234 * targhooks.c (default_target_can_inline_p): Properly
5235 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
5236 is present and always compare.
5237 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
5238 infer -mfpmath=sse from TARGET_SSE_P.
5239 (ix86_can_inline_p): Properly use target_option_default_node when
5240 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
5242 2017-08-24 Richard Biener <rguenther@suse.de>
5245 * dwarf2out.c (output_die): Handle flag_generate_offload like
5247 (output_comp_unit): Likewise.
5248 (gen_array_type_die): Likewise.
5249 (dwarf2out_early_finish): Likewise.
5250 (note_variable_value_in_expr): Likewise.
5251 (dwarf2out_finish): Likewise. Adjust assert.
5252 * cgraphunit.c (symbol_table::compile): Move setting of
5253 flag_generate_offload earlier ...
5254 (symbol_table::finalize_compilation_unit): ... here, before
5255 early debug finalization.
5257 2017-08-24 Richard Biener <rguenther@suse.de>
5259 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
5260 and ipa-fnsummary.h.
5261 (ix86_can_inline_p): When ix86_fpmath flags do not match
5262 check whether the callee uses FP math at all.
5264 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
5267 * tree-ssanames.c (get_nonzero_bits): Use element_precision
5268 instead of TYPE_PRECISION.
5270 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
5271 Alan Hayward <alan.hayward@arm.com>
5272 David Sherwood <david.sherwood@arm.com>
5274 * combine.c (make_extraction): Use subreg_offset_from_lsb.
5276 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
5278 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
5280 (ix86_frame::stack_realign_allocate): New field.
5281 (struct machine_frame_state): Modify comments.
5282 (machine_frame_state::sp_realigned_fp_end): New field.
5283 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
5285 (sp_valid_at): Add assertion to assure no attempt to access invalid
5286 offset of a realigned stack.
5287 (fp_valid_at): Likewise.
5288 (choose_baseaddr): Modify comments.
5289 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
5290 ix86_expand_prologue.
5291 (ix86_expand_prologue): Modify stack realignment and allocation.
5292 (ix86_expand_epilogue): Modify comments.
5293 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
5294 avx2_runtime, avx512f, and avx512f_runtime.
5296 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
5298 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
5299 (mstackrealign): Do not init to -1.
5300 * config/i386/i386.c (ix86_option_override_internal):
5301 Check opts_set, not opts when setting default value of
5302 opts->x_ix86_force_align_arg_pointer.
5304 2017-08-23 Richard Biener <rguenther@suse.de>
5306 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
5307 lang_hooks.decl_printable_name.
5308 * print-rtl-function.c (print_rtx_function): Likewise.
5309 * tree-pretty-print.c (dump_function_header): Likewise.
5311 2017-08-23 Richard Biener <rguenther@suse.de>
5314 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
5315 -g0 at compile-time.
5317 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5320 * doc/sourcebuild.texi (Other hardware attributes):
5323 2017-08-23 Tamar Christina <tamar.christina@arm.com>
5326 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
5327 Fix single-use check.
5329 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
5331 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
5333 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5335 * doc/install.texi: Modify to add more details on running selected
5338 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
5340 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
5341 is combined with -mabi=ms.
5342 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
5345 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5347 PR tree-optimization/81488
5348 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
5349 and cached_basis fields.
5350 (MAX_SPREAD): New constant.
5351 (alloc_cand_and_find_basis): Initialize new fields.
5352 (clear_visited): New function.
5353 (create_phi_basis_1): Rename from create_phi_basis, set visited
5354 and cached_basis fields.
5355 (create_phi_basis): New wrapper function.
5356 (phi_add_costs_1): Rename from phi_add_costs, add spread
5357 parameter, set visited field, short-circuit when limits reached.
5358 (phi_add_costs): New wrapper function.
5359 (record_phi_increments_1): Rename from record_phi_increments, set
5361 (record_phi_increments): New wrapper function.
5362 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
5363 (phi_incr_cost): New wrapper function.
5364 (all_phi_incrs_profitable_1): Rename from
5365 all_phi_incrs_profitable, set visited field.
5366 (all_phi_incrs_profitable): New wrapper function.
5368 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5369 Alan Hayward <alan.hayward@arm.com>
5370 David Sherwood <david.sherwood@arm.com>
5372 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
5373 that takes the outer and inner modes.
5374 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
5375 comparison as the canonical test for a paradoxical subreg.
5376 * combine.c (simplify_set): Use paradoxical_subreg_p.
5377 (make_extraction): Likewise.
5378 (force_to_mode): Likewise.
5379 (rtx_equal_for_field_assignment_p): Likewise.
5380 (gen_lowpart_for_combine): Likewise.
5381 (simplify_comparison): Likewise.
5382 * cse.c (equiv_constant): Likewise.
5383 * expmed.c (store_bit_field_1): Likewise.
5384 * final.c (alter_subreg): Likewise.
5385 * fwprop.c (propagate_rtx): Likewise.
5386 (forward_propagate_subreg): Likewise.
5387 * ira-conflicts.c (ira_build_conflicts): Likewise.
5388 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
5389 * lra-constraints.c (curr_insn_transform): Likewise.
5390 (split_reg): Likewise.
5391 * lra-eliminations.c (move_plus_up): Likewise.
5392 (lra_eliminate_regs_1): Likewise.
5393 * recog.c (general_operand): Likewise.
5394 * ree.c (combine_reaching_defs): Likewise.
5395 * reload.c (push_reload): Likewise.
5396 (find_reloads): Likewise.
5397 * reload1.c (elimination_effects): Likewise.
5398 (compute_reload_subreg_offset): Likewise.
5399 (choose_reload_regs): Likewise.
5400 * rtlanal.c (subreg_lsb_1): Likewise.
5401 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5402 (simplify_subreg): Likewise.
5403 * var-tracking.c (track_loc_p): Likewise.
5404 * emit-rtl.c (byte_lowpart_offset): Likewise.
5405 (paradoxical_subreg_p): Delete out-of-line definition.
5407 2017-08-22 Jeff Law <law@redhat.com>
5409 PR tree-optimization/81741
5410 PR tree-optimization/71947
5411 * tree-ssa-dom.c: Include tree-inline.h.
5412 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
5413 equivalences if one is more expensive to compute than the other.
5414 * tree-ssa-scopedtables.h (class const_or_copies): Make
5415 record_const_or_copy_raw method private.
5416 (class avail_exprs_stack): New method simplify_binary_operation.
5417 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
5418 avail_exprs_stack::simplify_binary_operation as needed.
5419 (avail_exprs_stack::simplify_binary_operation): New function.
5421 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
5423 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
5424 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
5425 (DOT_SYMBOLS): Likewise.
5426 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
5427 (RELOCATABLE_NEEDS_FIXUP): Likewise.
5428 (RS6000_ABI_NAME): Likewise.
5429 (TARGET_CMODEL): Likewise.
5430 (TOC_SECTION_ASM_OP): Likewise.
5431 (SET_CMODEL): New macro.
5432 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
5434 2017-08-22 Richard Biener <rguenther@suse.de>
5436 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
5437 to allow for free-lang-data replacements similar to verify_type_variant.
5439 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
5440 Alan Hayward <alan.hayward@arm.com>
5441 David Sherwood <david.sherwood@arm.com>
5443 * config/aarch64/aarch64.md (casesi): Use DImode rather than
5444 VOIDmode for the LABEL_REF.
5446 2017-08-22 Richard Biener <rguenther@suse.de>
5448 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
5450 2017-08-22 Richard Biener <rguenther@suse.de>
5452 * common.opt (feliminate-dwarf2-dups): Ignore.
5453 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
5454 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
5455 same_die_p_wrap, compute_section_prefix,
5456 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
5457 (comdat_symbol_id, comdat_symbol_number): Likewise.
5458 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
5460 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
5461 (output_die): Mark unreachable path unreachable.
5462 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
5463 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
5464 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
5465 (dwarf2out_early_finish): Likewise.
5467 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
5469 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
5471 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
5474 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
5475 not VAR_P. Filter attribute warnings with OPT_Wattributes.
5476 (avr_attribute_table) <io, io_low, address>: Initialize
5477 .decl_required with true.
5479 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
5481 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
5482 undocumented debugging options.
5483 (-mvsx-scalar-double): Likewise.
5484 (-mallow-df-permute): Likewise.
5485 (-mvectorize-builtins): Likewise.
5486 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
5487 (rs6000_builtin_vectorized_function): Likewise.
5488 (rs6000_builtin_md_vectorized_function): Likewise.
5489 (rs6000_opt_vars): Likewise.
5491 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
5494 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
5495 (*btrq_imm): Rename from *btrq.
5496 (*btcq_imm): Rename from *btcq.
5497 (btsc): New code attribute.
5498 (*<btsc><mode>): New insn pattern.
5499 (*btr<mode>): Ditto.
5500 (*<btsc><mode>_mask): New insn_and_split pattern.
5501 (*btr<mode>_mask): Ditto.
5503 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5504 Alan Hayward <alan.hayward@arm.com>
5505 David Sherwood <david.sherwood@arm.com>
5507 * function.c (pad_below): Simplify padding calculation.
5509 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5510 Alan Hayward <alan.hayward@arm.com>
5511 David Sherwood <david.sherwood@arm.com>
5513 * target.def (function_prologue): Remove frame size argument.
5514 (function_epilogue): Likewise.
5515 * doc/tm.texi: Regenerate.
5516 * final.c (final_start_function): Update call to function_prologue.
5517 (final_end_function): Update call to function_epilogue.
5518 (default_function_pro_epilogue): Remove frame size argument.
5519 * output.h (default_function_pro_epilogue): Likewise.
5520 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
5521 (arm_output_function_prologue): Likewise.
5522 * config/frv/frv.c (frv_function_prologue): Likewise.
5523 (frv_function_epilogue): Likewise.
5524 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
5525 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
5526 (ia64_output_function_epilogue): Likewise.
5527 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
5528 (m32r_output_function_epilogue): Likewise.
5529 * config/microblaze/microblaze.c (microblaze_function_prologue)
5530 (microblaze_function_epilogue): Likewise.
5531 * config/mips/mips.c (mips_output_function_prologue): Likewise.
5532 (mips_output_function_epilogue): Likewise.
5533 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
5534 (mmix_target_asm_function_epilogue): Likewise.
5535 * config/msp430/msp430.c (msp430_start_function): Likewise.
5536 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
5537 (nds32_asm_function_epilogue): Likewise.
5538 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
5539 * config/pa/pa.c (pa_output_function_prologue): Likewise.
5540 (pa_output_function_epilogue): Likewise.
5541 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
5542 (rs6000_output_function_epilogue): Likewise.
5543 * config/rl78/rl78.c (rl78_start_function): Likewise.
5544 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
5545 (rs6000_output_function_epilogue): Likewise.
5546 * config/rx/rx.c (rx_output_function_prologue): Likewise.
5547 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
5548 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
5549 (sparc_asm_function_epilogue): Likewise.
5551 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5553 * tree.h (type_has_mode_precision_p): New function.
5554 * convert.c (convert_to_integer_1): Use it.
5555 * expr.c (expand_expr_real_2): Likewise.
5556 (expand_expr_real_1): Likewise.
5557 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
5558 * match.pd: Likewise.
5559 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
5560 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
5561 * tree-tailcall.c (process_assignment): Likewise.
5562 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5563 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
5564 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
5565 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5566 (vectorizable_assignment): Likewise.
5567 (vectorizable_shift): Likewise.
5568 (vectorizable_operation): Likewise.
5569 * tree-vrp.c (register_edge_assert_for_2): Likewise.
5571 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
5573 * match.pd: Add pow (C, x) simplification.
5575 2017-08-21 Richard Biener <rguenther@suse.de>
5577 PR tree-optimization/81900
5578 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
5579 for blocks with abnormal predecessors.
5580 (compute_antic): Do not set visited flag prematurely.
5582 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
5585 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
5587 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5589 * stor-layout.h (vector_type_mode): Move to...
5590 * tree.h (vector_type_mode): ...here.
5591 * stor-layout.c (vector_type_mode): Move to...
5592 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
5594 2017-08-21 Richard Biener <rguenther@suse.de>
5596 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
5597 register_external_die hooks.
5598 (debug_false_tree_charstarstar_uhwistar): Declare.
5599 (debug_nothing_tree_charstar_uhwi): Likewise.
5600 * debug.c (do_nothing_debug_hooks): Adjust.
5601 (debug_false_tree_charstarstar_uhwistar): New do nothing.
5602 (debug_nothing_tree_charstar_uhwi): Likewise.
5603 * dbxout.c (dbx_debug_hooks): Adjust.
5604 (xcoff_debug_hooks): Likewise.
5605 * sdbout.c (sdb_debug_hooks): Likewise.
5606 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5607 * dwarf2out.c (macinfo_label_base): New global.
5608 (dwarf2out_register_external_die): New function for the
5609 register_external_die hook.
5610 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
5611 (dwarf2_debug_hooks): Use them.
5612 (dwarf2_lineno_debug_hooks): Adjust.
5613 (struct die_struct): Add with_offset flag.
5614 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
5615 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
5616 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
5617 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
5618 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
5619 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
5620 defining section names for the early LTO debug variants.
5621 (reset_indirect_string): New helper.
5622 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
5623 (print_dw_val): Add support for offsetted symbol references.
5624 (get_ultimate_context): Split out from is_cxx.
5625 (is_cxx): Use get_ultimate_context.
5626 (is_fortran): Add decl overload.
5627 (compute_comp_unit_symbol): Split out worker from
5628 compute_section_prefix.
5629 (compute_section_prefix): Call compute_comp_unit_symbol and
5630 set comdat_type_p here.
5631 (output_die): Skip DIE symbol output for the LTO added one.
5632 Handle DIE symbol references with offset.
5633 (output_comp_unit): Guard section name mangling properly.
5634 For LTO debug sections emit a symbol at the section beginning
5635 which we use to refer to its DIEs.
5636 (add_abstract_origin_attribute): For DIEs registered via
5637 dwarf2out_register_external_die directly refer to the early
5638 DIE rather than indirectly through the shadow one we created.
5639 Remove obsolete call to dwarf2out_abstract_function for
5640 non-function/block origins.
5641 (gen_array_type_die): When generating early LTO debug do
5642 not emit DW_AT_string_length.
5643 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
5644 late when in LTO. As suggested place a gcc_unreachable for
5645 the DECL_ABSTRACT_P case.
5646 (gen_subprogram_die): Avoid another specification DIE
5647 for early built declarations/definitions for the late LTO case.
5648 (gen_variable_die): Add type references for late duplicated VLA dies
5650 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
5651 we have the abstract instance already.
5652 (process_scope_var): Adjust decl DIE contexts in LTO which
5653 first puts them in limbo.
5654 (gen_decl_die): Do not generate type DIEs late apart from
5655 types for VLAs or for decls we do not yet have a DIE. Do not
5656 call dwarf2out_abstract_function late.
5657 (dwarf2out_early_global_decl): Make sure to create DIEs
5658 for abstract instances of a decl first.
5659 (dwarf2out_late_global_decl): Adjust comment.
5660 (output_macinfo_op): With multiple macro sections use
5661 macinfo_label_base to distinguish labels.
5662 (output_macinfo): Likewise. Update macinfo_label_base.
5663 Pass in the line info label.
5664 (note_variable_value_in_expr): When generating LTO resolve
5665 all variable values here by generating DIEs as needed.
5666 (init_sections_and_labels): Add early LTO debug flag parameter
5667 and generate different sections and names if set. Add generation
5668 counter for the labels so we can have multiple of them.
5669 (reset_dies): Helper to allow DIEs to be output multiple times.
5670 (dwarf2out_finish): When outputting DIEs to the fat part of an
5671 LTO object first reset DIEs.
5672 (dwarf2out_early_finish): Output early DIEs when generating LTO.
5673 (modified_type_die): Check for decl_ultimate_origin being self
5675 (gen_type_die_with_usage): Likewise.
5676 (gen_typedef_die): Allow decl_ultimate_origin being self.
5677 (set_decl_abstract_flags): Remove.
5678 (set_block_abstract_flags): Likewise.
5679 (dwarf2out_abstract_function): Treat the early generated DIEs
5680 as the abstract copy and only add DW_AT_inline and
5681 DW_AT_artificial here and call set_decl_origin_self.
5682 If the DIE has an abstract origin don't do anything.
5683 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
5684 if we have none yet (Go fails to build one, PR78628).
5685 (variably_modified_type_p): Prevent endless recursion for Ada
5686 cyclic pointer types.
5687 * lto-streamer-in.c: Include debug.h.
5688 (dref_queue): New global.
5689 (lto_read_tree_1): Stream in DIE references.
5690 (lto_input_tree): Register DIE references.
5691 (input_function): Stream DECL_DEBUG_ARGS.
5692 * lto-streamer-out.c: Include debug.h.
5693 (lto_write_tree_1): Output DIE references.
5694 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
5695 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
5696 (output_function): Stream DECL_DEBUG_ARGS.
5697 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
5698 Stream DECL_ABSTRACT_ORIGIN.
5699 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
5700 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
5701 DECL_CONTEXT for file-scope decls.
5702 * lto-streamer.h (struct dref_entry): Declare.
5703 (dref_queue): Likewise.
5704 * cfgexpand.c (pass_expand::execute): Do not call the
5705 outlining_inline_function hook here.
5706 * lto-wrapper.c (debug_obj): New global.
5707 (tool_cleanup): Unlink it if required.
5708 (debug_objcopy): New function.
5709 (run_gcc): Handle early debug sections in the IL files by
5710 extracting them to separate files, partially linkin them and
5711 feeding the result back as result to the linker.
5712 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
5713 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
5714 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
5715 sections into a separate segment.
5716 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
5718 (darwin_asm_dwarf_section): Likewise.
5719 (darwin_asm_output_dwarf_offset): Likewise.
5720 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
5722 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5723 Alan Hayward <alan.hayward@arm.com>
5724 David Sherwood <david.sherwood@arm.com>
5726 * read-md.h (md_reader::record_potential_iterator_use): Replace
5727 pointer argument with an rtx and an index.
5728 * read-rtl.c (iterator_group::apply_iterator): Likewise.
5729 (apply_mode_iterator): Likewise.
5730 (apply_code_iterator): Likewise.
5731 (apply_int_iterator): Likewise.
5732 (apply_subst_iterator): Likewise.
5733 (record_iterator_use): Likewise.
5734 (record_attribute_use): Likewise.
5735 (md_reader::record_potential_iterator_use): Likewise. Update calls
5736 to record_iterator_use and apply_iterator.
5737 (iterator_use): Replace ptr with x and index.
5738 (attribute_use): Likewise.
5739 (apply_attribute_uses): Update calls to apply_iterator.
5740 (apply_iterators): Likewise. Update initialization of iterator_use.
5741 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
5742 and record_potential_iterator_use.
5743 (rtx_reader::read_rtx_operand): Likewise.
5745 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
5746 Alan Hayward <alan.hayward@arm.com>
5747 David Sherwood <david.sherwood@arm.com>
5749 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
5752 2017-08-21 Richard Biener <rguenther@suse.de>
5755 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
5756 at struct end conservatively when comparing common bases.
5758 2017-08-21 Richard Biener <rguenther@suse.de>
5760 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
5761 (mem_ref_in_stmt): Remove.
5762 (determine_max_movement): Use ref index to get at the reference.
5763 (invariantness_dom_walker::before_dom_children): Deal with
5764 lim data already initialized.
5765 (gather_mem_refs_stmt): Initialize lim data and record ref index.
5767 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5769 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
5770 (TARGET_ISA_ROUND): Ditto.
5771 (TARGET_ROUND): Ditto.
5772 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
5773 * config/i386/i386.md: Ditto.
5774 * config/i386/sse.md: Ditto.
5775 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
5776 with OPTION_MASK_ISA_SSE4_1.
5778 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
5781 * doc/extend.texi (x86 Built-in Functions): Correct the name of
5782 __builtin_ia32_lzcnt_u16.
5784 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5787 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
5788 (rs6000_set_current_function): Rewrite function to use it.
5790 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
5793 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
5794 and TYPE_WARN_IF_NOT_ALIGN.
5795 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
5796 (handle_warn_if_not_align): New.
5797 (place_union_field): Call handle_warn_if_not_align.
5798 (place_field): Call handle_warn_if_not_align.
5799 Copy TYPE_WARN_IF_NOT_ALIGN.
5800 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
5801 (layout_type): Likewise.
5802 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
5804 (tree_decl_common): Add warn_if_not_align.
5805 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
5806 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
5807 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
5808 (DECL_WARN_IF_NOT_ALIGN): Likewise.
5809 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
5810 * doc/extend.texi: Document warn_if_not_aligned attribute.
5811 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
5813 2017-08-17 Martin Liska <mliska@suse.cz>
5816 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
5817 (get_data_dependence): Use it as pointer type.
5818 (distribute_loop): Likewise.
5820 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5822 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
5823 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
5824 (p8_vmrgow_<mode>_direct): New define_insn.
5825 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
5826 handle endianness for vmrgew and vmrgow permute patterns.
5828 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5830 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
5831 * config/rs6000/rs6000-cpus.def: Remove comment.
5832 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
5833 (POWERPC_MASKS): Likewise.
5834 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
5835 use of TARGET_VSX_TIMODE.
5836 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
5837 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
5838 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
5839 (rs6000_option_override_internal): Remove dead code.
5840 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
5841 (rs6000_legitimize_reload_address): Likewise.
5842 (rs6000_legitimate_address_p): Likewise.
5843 (rs6000_opt_masks): Delete "vsx-timode".
5844 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
5845 from function comment.
5846 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
5847 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
5848 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
5850 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
5851 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
5852 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
5855 (define_peephole2): Likewise.
5857 2017-08-17 Martin Sebor <msebor@redhat.com>
5860 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
5861 past the end of an array.
5862 (test_pp_format): Add test cases.
5864 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
5866 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
5867 missing ECF_NOTHROW flags.
5869 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
5872 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
5873 operands residing in integer registers.
5874 (*vsx_le_perm_load_<mode>): Likewise.
5875 (*vsx_le_perm_store_<mode>): Likewise.
5876 (define_peephole2): Add peepholes to optimize the above.
5878 2017-08-17 Marek Polacek <polacek@redhat.com>
5881 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
5882 to mimic what shorten_compare did. Change the return type to bool.
5883 (fold_cond_expr_with_comparison): Update call to
5884 operand_equal_for_comparison_p.
5885 (fold_ternary_loc): Likewise.
5887 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
5889 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
5891 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
5893 2017-08-17 Richard Biener <rguenther@suse.de>
5895 * tree-ssa-structalias.c (solve_graph): When propagating
5896 to successors update the graphs succ edges and avoid duplicate work.
5898 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
5901 * config/i386/i386.c (ix86_option_override_internal): Save target
5902 specific options after ix86_stack_protector_guard_reg was changed.
5904 2017-08-17 Richard Biener <rguenther@suse.de>
5906 PR tree-optimization/81827
5907 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
5908 (new_var_info): Initialize it conservatively.
5909 (get_call_vi): Mark register vars.
5910 (new_scalar_tmp_constraint_exp): Likewise.
5911 (handle_rhs_call): Likewise.
5912 (handle_const_call): Likewise.
5913 (create_function_info_for): Likewise.
5914 (solve_constraints): Sort varinfos to separate register from
5915 non-register vars to pack points-to solution bitmaps during
5918 2017-08-17 Marek Polacek <polacek@redhat.com>
5920 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
5922 2017-08-17 Richard Biener <rguenther@suse.de>
5924 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
5925 to true when overflow is undefined and we saturated the result.
5927 2017-08-17 Alan Modra <amodra@gmail.com>
5930 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
5931 Don't use store multiple if only one reg needs saving.
5932 (interesting_frame_related_regno): New function.
5933 (rs6000_frame_related): Don't emit frame info for regs that
5935 (rs6000_emit_epilogue): Likewise.
5937 2017-08-16 Nathan Sidwell <nathan@acm.org>
5939 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
5940 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
5941 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
5942 * tree.c (free_lang_data_in_type): Use else-if chain. Always
5943 clear TYPE_LANG_1. Remove obsolete member-function stripping.
5944 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
5945 (verify_type): Adjust for TYPE_BINFO move.
5946 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
5947 process TYPE_BINFO directly.
5948 (hash_tree): Likewise.
5949 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5951 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5954 2017-08-16 David Malcolm <dmalcolm@redhat.com>
5956 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
5958 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
5961 * config/i386/i386.md (*anddi_1_btr): Change predicates of
5962 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
5963 Add ix86_binary_operator_ok to insn constraint.
5964 (*iordi_1_bts): Ditto.
5965 (*xordi_1_btc): Ditto.
5966 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
5967 Update corresponding peephole2 pattern.
5971 2017-08-16 Bin Cheng <bin.cheng@arm.com>
5973 PR tree-optimization/81832
5974 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
5975 copy loop header which has IFN_LOOP_DIST_ALIAS call.
5977 2017-08-16 Marek Polacek <polacek@redhat.com>
5980 * fold-const.c (fold_indirect_ref_1): Restore original behavior
5981 regarding size_zero_node.
5983 2017-08-16 Martin Liska <mliska@suse.cz>
5986 * config.gcc: Respect previously set extra_objs in case
5989 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
5991 PR tree-optimization/81835
5992 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
5993 the phi SSA_NAME. Check that the condition in a COND_EXPR does
5994 not depend on the phi.
5996 2017-08-16 Alan Modra <amodra@gmail.com>
5998 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
6001 2017-08-16 Alan Modra <amodra@gmail.com>
6003 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
6004 (save_reg_p): ..into this. Update all callers.
6005 (first_reg_to_save): Simplify.
6007 2017-08-16 Alan Modra <amodra@gmail.com>
6009 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
6012 2017-08-15 Joseph Myers <joseph@codesourcery.com>
6016 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
6017 constant count if that count is less than 32.
6019 2017-08-15 Nathan Sidwell <nathan@acm.org>
6021 * gcc.c (execute): Emit friendlier message if inferior is killed
6022 by an external cause.
6024 2017-08-15 Richard Biener <rguenther@suse.de>
6026 PR tree-optimization/81790
6027 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
6028 CONSTRUCTORs from simplifying and VN.
6030 2017-08-14 Martin Sebor <msebor@redhat.com>
6032 * builtin-attrs.def: Add comments.
6034 2017-08-14 Martin Sebor <msebor@redhat.com>
6037 * doc/extend.texi (attribute nonstring): Document new attribute.
6039 2017-08-14 Martin Sebor <msebor@redhat.com>
6042 * tree-diagnostic.c (default_tree_printer): Handle %G.
6043 * gimple-pretty-print.h (percent_G_format): Declare new function.
6044 * gimple-pretty-print.c (percent_G_format): Define.
6045 * tree-pretty-print.c (percent_K_format): Add argument.
6047 2017-08-14 Martin Sebor <msebor@redhat.com>
6049 PR translation/79998
6050 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
6051 Remove a stray space.
6053 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
6056 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
6057 (*iordi_1_bts): Ditto.
6058 (*xordi_1_btc): Ditto.
6060 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6063 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
6064 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6066 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
6067 quoted strings, and make more translator-friendly.
6068 (darwin_rs6000_override_options): Likewise.
6069 (rs6000_option_override_internal): Likewise.
6070 (rs6000_return_in_memory): Fix overlong line.
6071 (init_cmulative_args): Use quoted strings, and make more
6072 translator-friendly.
6073 (rs6000_pass_by_reference): Fix overlong line.
6074 (def_builtin): Use quoted strings.
6075 (altivec_expand_predicate_builtin): Use quoted strings, and make
6076 more translator-friendly.
6077 (htm_expand_builtin): Use quoted strings.
6078 (cpu_expand_builtin): Use quoted strings, and make more
6079 translator-friendly.
6080 (altivec_expand_builtin): Likewise.
6081 (paired_expand_predicate_builtin): Likewise.
6082 (rs6000_invalid_builtin): Likewise.
6083 (builtin_function_type): Use quoted strings.
6084 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
6085 more translator-friendly.
6086 (rs6000_trampoline_init): Likewise.
6087 (rs6000_handle_altivec_attribute): Likewise.
6088 (rs6000_inner_target_options): Use quoted strings.
6089 (rs6000_disable_incompatible_switches): Likewise.
6090 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
6091 strings, and make more translator-friendly.
6092 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
6094 2017-08-14 Bin Cheng <bin.cheng@arm.com>
6096 PR tree-optimization/81799
6097 * tree-loop-distribution.c (version_loop_by_alias_check): Force
6098 cond_expr to simple gimple operand.
6100 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
6103 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
6105 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
6109 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
6110 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
6111 TARGET_GASISR_PROLOGUES.
6112 * config/avr/avr.c (avr_option_override): Same.
6113 (avr_pass_pre_proep::execute): Same.
6115 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
6118 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
6119 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
6121 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
6123 * config/i386/i386.md (*load_tp_<mode>): Redefine as
6124 define_insn_and_split. Split to a memory load from 0 in
6125 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
6126 using PTR mode iterator.
6127 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
6128 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
6129 (*add_tp_<mode>): Redefine as define_insn_and_split.
6130 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
6131 address space. Merge with *add_tp_x32 using PTR mode iterator.
6132 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
6133 Split to an add with a memory load from 0 in
6134 DEFAULT_TLS_SEG_REG address space.
6136 2017-08-12 Andrew Pinski <apinski@cavium.com>
6138 * config/aarch64/aarch64-option-extensions.def (rdma):
6139 Fix feature string to what Linux prints out in /proc/cpuinfo.
6141 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
6144 * dwarf2out.c (modified_type_die): For C typedef types that have
6145 an ultimate origin, process the ultimate origin instead of the
6147 (gen_typedef_die): Assert that input DECLs have no ultimate
6149 (gen_type_die_with_usage): For typedef variants that have an
6150 ultimate origin, just call gen_decl_die on the original DECL.
6151 (process_scope_var): Avoid creating DIEs for local typedefs and
6152 concrete static variables.
6154 2017-08-12 Alan Modra <amodra@gmail.com>
6158 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
6159 match gnu-user.h startfile.
6160 (ENDFILE_LINUX_SPEC): Similarly.
6162 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
6165 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
6167 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
6169 2017-08-11 Tamar Christina <tamar.christina@arm.com>
6170 * config/aarch64/aarch64.md (mov<mode>): Change.
6171 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6172 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
6173 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
6175 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
6177 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
6178 for storage order barriers.
6180 2017-08-11 Martin Liska <mliska@suse.cz>
6183 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
6184 variables of void type.
6186 2017-08-11 Martin Liska <mliska@suse.cz>
6188 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
6189 TARGET_SUPPORTS_ALIASES.
6190 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
6191 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
6192 (optimize_weakref): Likewise.
6193 * symtab.c (symtab_node::noninterposable_alias): Likewise.
6194 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
6195 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
6197 2017-08-11 Martin Liska <mliska@suse.cz>
6200 * config/i386/i386.c (make_resolver_func): Do complete
6201 refactoring of the function.
6203 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6206 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
6207 * config/i386/i386.c (ix86_stack_protect_guard): Use
6208 ix86_stack_protect_guard_symbol_str to generate varible declaration.
6209 * doc/invoke.texi (x86 Options): Document
6210 -mstack-protector-guard-symbol= option.
6212 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
6214 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
6215 * config/i386/i386.c (ix86_split_stack_guard): New function.
6216 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
6217 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
6218 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
6219 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
6220 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
6221 (split_stack_space_check): Call ix86_split_stack_guard.
6223 2017-08-10 Martin Sebor <msebor@redhat.com>
6225 * print-tree.c (print_node): Print location using the established
6227 Replace spaces with colons.
6228 (debug_raw, debug): Ditto.
6230 2017-08-10 Martin Sebor <msebor@redhat.com>
6233 * pretty-print.c (pp_format): Correct the handling of %s precision.
6235 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
6238 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6240 (ix86_finalize_stack_frame_flags): This. Also clear
6241 frame_pointer_needed if -fno-omit-frame-pointer is used without
6243 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6244 with ix86_finalize_stack_frame_flags.
6245 (ix86_expand_epilogue): Likewise.
6246 (ix86_expand_split_stack_prologue): Likewise.
6247 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
6249 2017-08-10 Martin Liska <mliska@suse.cz>
6252 * c-attribs.c (handle_target_attribute):
6253 Report warning for an empty string argument of target attribute.
6255 2017-08-09 Jakub Jelinek <jakub@redhat.com>
6258 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
6260 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
6261 or DECL_NONLOCAL labels.
6262 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
6263 or DECL_NONLOCAL labels here.
6265 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
6267 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
6268 to indicate when early gimple folding has been disabled.
6269 (rs6000_gimple_fold_builtin): Add debug content.
6270 (rs6000_invalid_builtin): Fix whitespace.
6271 (rs6000_expand_builtin): Fix whitespace.
6272 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
6274 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
6277 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
6278 SAVE_MULTIPLE if not all the registers that saves, should be saved.
6280 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
6282 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
6283 (qdf24xx): Likewise.
6284 * config/aarch64/aarch64.md: Include falkor.md.
6285 * config/aarch64/falkor.md: New.
6287 2017-08-09 Marek Polacek <polacek@redhat.com>
6290 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
6291 * diagnostic.c (emit_diagnostic): Add a comment.
6292 (emit_diagnostic_valist): New function.
6294 2017-08-09 Marek Polacek <polacek@redhat.com>
6297 * input.c (make_location): New overload.
6298 * input.h (make_location): Declare.
6300 2017-08-08 Alan Modra <amodra@gmail.com>
6301 H.J. Lu <hongjiu.lu@intel.com>
6304 * gcc.c (NO_PIE_SPEC): Delete.
6305 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
6307 (LINK_PIE_SPEC): ..to here.
6308 (LINK_COMMAND_SPEC): Support -no-pie.
6309 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
6310 chain of crtbegin*.o selection, update for PIE_SPEC changes and
6312 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
6313 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
6314 (ENDFILE_CRTEND_SPEC): Similarly.
6316 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
6319 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
6320 (mstack-protector-guard-offset=): Ditto.
6321 * config/i386/i386.c (ix86_option_override): Handle
6322 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
6324 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
6325 ix86_stack_protect_guard_offset variables.
6326 (TARGET_STACK_PROTECT_GUARD): Always define.
6327 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
6328 and -mstack-protector-guard-offset= options.
6330 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6332 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
6333 boundary case for the last candidate.
6335 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6337 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
6338 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
6340 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6343 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
6344 * config/aarch64/aarch64-builtins.c
6345 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
6346 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
6347 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
6349 2017-08-08 Tamar Christina <tamar.christina@arm.com>
6350 Andrew Pinski <pinskia@gmail.com>
6353 * internal-fn.def (XORSIGN): New.
6354 * optabs.def (xorsign_optab): New.
6355 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
6356 (convert_expand_mult_copysign): New.
6357 (pass_optimize_widening_mul::execute): Call
6358 convert_expand_mult_copysign.
6360 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6362 PR tree-optimization/81354
6363 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
6364 Insert on edges rather than explicitly creating landing pads.
6365 (analyze_candidates_and_replace): Commit edge inserts.
6367 2017-08-08 Richard Biener <rguenther@suse.de>
6370 * tree-ssa-loop-niter.c: Include tree-dfa.h.
6371 (expand_simple_operations): Also look through ADDR_EXPRs with
6372 MEM_REF bases treating them as POINTER_PLUS_EXPR.
6374 2017-08-08 Richard Biener <rguenther@suse.de>
6376 PR tree-optimization/81723
6377 * tree-vect-slp.c (struct bst_traits): New hash traits.
6378 (bst_fail): New global.
6379 (vect_build_slp_tree_2): New worker, split out from ...
6380 (vect_build_slp_tree): ... this now wrapping it with using
6381 bst_fail set to cache SLP tree build fails. Properly handle
6383 (vect_analyze_slp_instance): Allocate and free bst_fail.
6385 2017-08-08 Martin Liska <mliska@suse.cz>
6388 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
6389 LABEL_DECLs that can be from a different function.
6391 2017-08-08 Bin Cheng <bin.cheng@arm.com>
6393 PR tree-optimization/81744
6394 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
6395 loop's number of iterations.
6397 2017-08-08 Martin Liska <mliska@suse.cz>
6399 * asan.c: Include header files.
6400 * attribs.c (build_decl_attribute_variant): New function moved
6402 (build_type_attribute_qual_variant): Likewise.
6403 (cmp_attrib_identifiers): Likewise.
6404 (simple_cst_list_equal): Likewise.
6405 (omp_declare_simd_clauses_equal): Likewise.
6406 (attribute_value_equal): Likewise.
6407 (comp_type_attributes): Likewise.
6408 (build_type_attribute_variant): Likewise.
6409 (lookup_ident_attribute): Likewise.
6410 (remove_attribute): Likewise.
6411 (merge_attributes): Likewise.
6412 (merge_type_attributes): Likewise.
6413 (merge_decl_attributes): Likewise.
6414 (merge_dllimport_decl_attributes): Likewise.
6415 (handle_dll_attribute): Likewise.
6416 (attribute_list_equal): Likewise.
6417 (attribute_list_contained): Likewise.
6418 * attribs.h (lookup_attribute): New function moved from tree.[ch].
6419 (lookup_attribute_by_prefix): Likewise.
6420 * bb-reorder.c: Include header files.
6421 * builtins.c: Likewise.
6422 * calls.c: Likewise.
6423 * cfgexpand.c: Likewise.
6424 * cgraph.c: Likewise.
6425 * cgraphunit.c: Likewise.
6426 * convert.c: Likewise.
6427 * dwarf2out.c: Likewise.
6428 * final.c: Likewise.
6429 * fold-const.c: Likewise.
6430 * function.c: Likewise.
6431 * gimple-expr.c: Likewise.
6432 * gimple-fold.c: Likewise.
6433 * gimple-pretty-print.c: Likewise.
6434 * gimple.c: Likewise.
6435 * gimplify.c: Likewise.
6436 * hsa-common.c: Likewise.
6437 * hsa-gen.c: Likewise.
6438 * internal-fn.c: Likewise.
6439 * ipa-chkp.c: Likewise.
6440 * ipa-cp.c: Likewise.
6441 * ipa-devirt.c: Likewise.
6442 * ipa-fnsummary.c: Likewise.
6443 * ipa-inline.c: Likewise.
6444 * ipa-visibility.c: Likewise.
6446 * lto-cgraph.c: Likewise.
6447 * omp-expand.c: Likewise.
6448 * omp-general.c: Likewise.
6449 * omp-low.c: Likewise.
6450 * omp-offload.c: Likewise.
6451 * omp-simd-clone.c: Likewise.
6452 * opts-global.c: Likewise.
6453 * passes.c: Likewise.
6454 * predict.c: Likewise.
6455 * sancov.c: Likewise.
6456 * sanopt.c: Likewise.
6457 * symtab.c: Likewise.
6458 * toplev.c: Likewise.
6459 * trans-mem.c: Likewise.
6460 * tree-chkp.c: Likewise.
6461 * tree-eh.c: Likewise.
6462 * tree-into-ssa.c: Likewise.
6463 * tree-object-size.c: Likewise.
6464 * tree-parloops.c: Likewise.
6465 * tree-profile.c: Likewise.
6466 * tree-ssa-ccp.c: Likewise.
6467 * tree-ssa-live.c: Likewise.
6468 * tree-ssa-loop.c: Likewise.
6469 * tree-ssa-sccvn.c: Likewise.
6470 * tree-ssa-structalias.c: Likewise.
6471 * tree-ssa.c: Likewise.
6472 * tree-streamer-in.c: Likewise.
6473 * tree-vectorizer.c: Likewise.
6474 * tree-vrp.c: Likewise.
6476 * ubsan.c: Likewise.
6477 * varasm.c: Likewise.
6478 * varpool.c: Likewise.
6479 * tree.c: Remove functions moved to attribs.[ch].
6481 * config/aarch64/aarch64.c: Add attrs.h header file.
6482 * config/alpha/alpha.c: Likewise.
6483 * config/arc/arc.c: Likewise.
6484 * config/arm/arm.c: Likewise.
6485 * config/avr/avr.c: Likewise.
6486 * config/bfin/bfin.c: Likewise.
6487 * config/c6x/c6x.c: Likewise.
6488 * config/cr16/cr16.c: Likewise.
6489 * config/cris/cris.c: Likewise.
6490 * config/darwin.c: Likewise.
6491 * config/epiphany/epiphany.c: Likewise.
6492 * config/fr30/fr30.c: Likewise.
6493 * config/frv/frv.c: Likewise.
6494 * config/ft32/ft32.c: Likewise.
6495 * config/h8300/h8300.c: Likewise.
6496 * config/i386/winnt.c: Likewise.
6497 * config/ia64/ia64.c: Likewise.
6498 * config/iq2000/iq2000.c: Likewise.
6499 * config/lm32/lm32.c: Likewise.
6500 * config/m32c/m32c.c: Likewise.
6501 * config/m32r/m32r.c: Likewise.
6502 * config/m68k/m68k.c: Likewise.
6503 * config/mcore/mcore.c: Likewise.
6504 * config/microblaze/microblaze.c: Likewise.
6505 * config/mips/mips.c: Likewise.
6506 * config/mmix/mmix.c: Likewise.
6507 * config/mn10300/mn10300.c: Likewise.
6508 * config/moxie/moxie.c: Likewise.
6509 * config/msp430/msp430.c: Likewise.
6510 * config/nds32/nds32-isr.c: Likewise.
6511 * config/nds32/nds32.c: Likewise.
6512 * config/nios2/nios2.c: Likewise.
6513 * config/nvptx/nvptx.c: Likewise.
6514 * config/pa/pa.c: Likewise.
6515 * config/pdp11/pdp11.c: Likewise.
6516 * config/powerpcspe/powerpcspe.c: Likewise.
6517 * config/riscv/riscv.c: Likewise.
6518 * config/rl78/rl78.c: Likewise.
6519 * config/rx/rx.c: Likewise.
6520 * config/s390/s390.c: Likewise.
6521 * config/sh/sh.c: Likewise.
6522 * config/sol2.c: Likewise.
6523 * config/sparc/sparc.c: Likewise.
6524 * config/spu/spu.c: Likewise.
6525 * config/stormy16/stormy16.c: Likewise.
6526 * config/tilegx/tilegx.c: Likewise.
6527 * config/tilepro/tilepro.c: Likewise.
6528 * config/v850/v850.c: Likewise.
6529 * config/vax/vax.c: Likewise.
6530 * config/visium/visium.c: Likewise.
6531 * config/xtensa/xtensa.c: Likewise.
6533 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6536 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
6537 constraints since the -mupper-regs-* switches have been
6539 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
6540 into a vector from a double word element that was extracted from
6541 another vector, and eliminate extra XXPERMDI instructions.
6542 (vsx_concat_<mode>_2): Likewise.
6543 (vsx_concat_<mode>_3): Likewise.
6544 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
6545 concat to allow optimizing inserts from previous extracts.
6547 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
6549 * config/i386/i386.c (ix86_stack_protect_guard): Generate
6550 memory reference to a SSP offset in TLS address space.
6551 (ix86_print_operand) <case '@'>: Remove.
6552 (ix86_print_operand_punct_valid_p): Remove '@' code.
6553 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
6555 (stack_tls_protect_set_<mode>): Remove.
6556 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
6557 (stack_tls_protect_test_<mode>): Remove.
6558 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
6560 2017-08-07 Olivier Hainque <hainque@adacore.com>
6563 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
6565 2017-08-07 Douglas Rupp <rupp@adacore.com>
6567 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
6568 variable was referenced as multidir in command.
6570 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6573 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
6576 2017-08-07 Martin Liska <mliska@suse.cz>
6578 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
6579 * config/rl78/rl78.c: Add include of attribs.h.
6580 * config/sh/sh.c: Likewise.
6581 * config/v850/v850.c: Likewise.
6583 2017-08-07 Tom de Vries <tom@codesourcery.com>
6586 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
6588 2017-08-07 Martin Liska <mliska@suse.cz>
6590 * config/mips/mips.c: Include attribs.h.
6592 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
6595 * doc/invoke.texi: Document change in behvaior for -Ofast for
6598 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
6600 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
6602 (aarch64_pop_regs): Likewise.
6603 (aarch64_gen_load_pair): Likewise.
6604 (aarch64_save_callee_saves): Likewise.
6605 (aarch64_restore_callee_saves): Likewise.
6607 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6609 * config/i386/i386.c: Revert the last change.
6611 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6614 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
6616 (ix86_finalize_stack_frame_flags): This. Also clear
6617 frame_pointer_needed if -fno-omit-frame-pointer is used without
6619 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
6620 with ix86_finalize_stack_frame_flags.
6621 (ix86_expand_epilogue): Likewise.
6622 (ix86_expand_split_stack_prologue): Likewise.
6624 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
6627 * config/i386/i386.c (get_builtin_code_for_version): Set priority
6628 to P_AES for Westmere.
6630 2017-08-07 Jonathan Yong <10walls@gmail.com>
6632 * config/i386/mingw.opt (fset-stack-executable): Removed.
6633 * config/i386/cygming.opt (fset-stack-executable): Moved
6635 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
6637 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
6639 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
6641 2017-08-07 Marek Polacek <polacek@redhat.com>
6644 * fold-const.c (fold_indirect_ref_1): Check type_domain.
6646 2017-08-07 Martin Liska <mliska@suse.cz>
6648 * attribs.h (canonicalize_attr_name): New function.
6649 (cmp_attribs): Move from c-format.c and adjusted.
6650 (is_attribute_p): Moved from tree.h.
6651 * tree-inline.c: Add new includes.
6652 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
6653 (private_is_attribute_p): Remove.
6654 (private_lookup_attribute): Likewise.
6655 (private_lookup_attribute_by_prefix): Simplify.
6656 (remove_attribute): Use is_attribute_p.
6657 * tree.h: Remove removed declarations.
6659 2017-08-07 Jakub Jelinek <jakub@redhat.com>
6662 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
6663 instead of computing it in the function. Formatting fix.
6664 (expand_case): Don't rely on default_edge being the first edge,
6665 clear it if removing it, pass default_edge to
6666 emit_case_dispatch_table.
6667 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
6670 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
6672 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
6673 insn in the function, emit NOP after the insn.
6675 2017-08-06 Tom de Vries <tom@codesourcery.com>
6677 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
6680 2017-08-06 Tom de Vries <tom@codesourcery.com>
6682 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
6685 2017-08-04 Yury Gribov <tetra2005@gmail.com>
6687 PR tree-optimization/57371
6688 * match.pd: New pattern.
6690 2017-08-04 Marek Polacek <polacek@redhat.com>
6693 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
6694 perform the computation in offset_int.
6696 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6698 PR tree-optimization/81136
6699 * tree-vectorizer.h: Include tree-hash-traits.h.
6700 (vec_base_alignments): New typedef.
6701 (vec_info): Add a base_alignments field.
6702 (vect_record_base_alignments): Declare.
6703 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
6705 (DR_IS_CONDITIONAL_IN_STMT): New macro.
6706 (create_data_ref): Add an is_conditional_in_stmt argument.
6707 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
6708 the is_conditional_in_stmt field.
6709 (data_ref_loc): Add an is_conditional_in_stmt field.
6710 (get_references_in_stmt): Set the is_conditional_in_stmt field.
6711 (find_data_references_in_stmt): Update call to create_data_ref.
6712 (graphite_find_data_references_in_stmt): Likewise.
6713 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
6714 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
6715 (vect_record_base_alignment): New function.
6716 (vect_record_base_alignments): Likewise.
6717 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
6718 for nested statements even if we fail to compute a misalignment.
6719 Use pooled base alignments for unconditional references.
6720 (vect_find_same_alignment_drs): Compare base addresses instead
6722 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
6723 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
6725 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6727 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
6728 Add an explicit name for the enum. Use auto_vec for slp_instances
6730 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
6732 (_bb_vec_info): Add a constructor and destructor.
6733 (vinfo_for_stmt): Return NULL for uids of -1 as well.
6734 (destroy_loop_vec_info): Delete.
6735 (vect_destroy_datarefs): Likewise.
6736 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
6737 (vec_info::vec_info): New function.
6738 (vec_info::~vec_info): Likewise.
6739 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
6740 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
6741 destroy_loop_vec_info.
6742 * tree-vect-loop.c (new_loop_vec_info): Replace with...
6743 (_loop_vec_info::_loop_vec_info): ...this.
6744 (destroy_loop_vec_info): Replace with...
6745 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
6746 the stmt_vec_infos. Leave handling of vec_info information to its
6747 destructor. Remove explicit vector releases.
6748 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
6749 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
6750 * tree-vect-slp.c (new_bb_vec_info): Replace with...
6751 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
6752 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
6753 (destroy_bb_vec_info): Replace with...
6754 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
6755 information to its destructor.
6756 (vect_slp_analyze_bb_1): Use new and delete instead of
6757 new_bb_vec_info and destroy_bb_vec_info.
6758 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
6761 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
6763 * tree-data-ref.h (subscript): Add access_fn field.
6764 (data_dependence_relation): Add could_be_independent_p.
6765 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
6766 (same_access_functions): Move to tree-data-ref.c.
6767 * tree-data-ref.c (ref_contains_union_access_p): New function.
6768 (access_fn_component_p): Likewise.
6769 (access_fn_components_comparable_p): Likewise.
6770 (dr_analyze_indices): Add a reference to access_fn_component_p.
6771 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
6773 (constant_access_functions): Likewise.
6774 (add_other_self_distances): Likewise.
6775 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
6776 (initialize_data_dependence_relation): Use XCNEW and remove
6777 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
6778 of access functions that have the same type. Allow the
6779 subsequence to end with different bases in some circumstances.
6780 Record the chosen access functions in SUB_ACCESS_FN.
6781 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
6782 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
6783 (subscript_dependence_tester_1): Likewise dra and drb.
6784 (build_classic_dist_vector): Update calls accordingly.
6785 (subscript_dependence_tester): Likewise.
6786 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
6787 DDR_COULD_BE_INDEPENDENT_P.
6788 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
6789 comp_alias_ddrs instead of may_alias_ddrs.
6790 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
6792 (vect_analyze_data_ref_dependence): Use it if
6793 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
6794 distance vectors if that fails.
6795 (dependence_distance_ge_vf): New function.
6796 (vect_prune_runtime_alias_test_list): Use it. Don't clear
6797 LOOP_VINFO_MAY_ALIAS_DDRS.
6799 2017-08-04 Richard Biener <rguenther@suse.de>
6802 * fold-const.c (fold_binary_loc): Properly restrict
6803 minus_var0 && minus_var1 case when associating undefined overflow
6806 2017-08-04 Tom de Vries <tom@codesourcery.com>
6808 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
6810 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6812 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6813 Don't start diagnostic messages with a capital letter.
6814 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6816 (rs6000_invalid_builtin): Likewise.
6817 (rs6000_trampoline_init): Likewise.
6819 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6822 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
6823 after setting changeable df flags.
6825 2017-08-03 Richard Biener <rguenther@suse.de>
6827 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
6828 up if the use is in USE - X.
6830 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
6832 * toplev.c (dumpfile.h): New include.
6833 (internal_error_reentered): New static function. Use it...
6834 (internal_error_function): ...here to handle reentered internal_error.
6836 2017-08-03 Richard Biener <rguenther@suse.de>
6839 * fold-const.c (split_tree): Add minus_var and minus_con
6840 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
6841 here but always use minus_*.
6842 (associate_trees): Assert we never associate with MINUS_EXPR
6843 and NULL first operand. Do not recurse for PLUS_EXPR operands
6844 when associating as MINUS_EXPR either.
6845 (fold_binary_loc): Track minus_var and minus_con.
6847 2017-08-03 Tom de Vries <tom@codesourcery.com>
6850 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
6851 ACCEL_COMPILER, apply finish_options on
6852 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
6854 2017-08-03 Tom de Vries <tom@codesourcery.com>
6857 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
6858 function_entry_patch_area_size > 0.
6860 2017-08-03 Jakub Jelinek <jakub@redhat.com>
6863 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
6864 instead of 10??LU, perform unit multiplication in wide_int,
6865 don't change alloc_object_size_limit if the limit is larger
6868 PR tree-optimization/81655
6869 PR tree-optimization/81588
6870 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
6871 the case when ranges[i].low and high are 1 for unsigned type with
6875 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
6876 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
6878 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6880 * tree-vrp.h: Add include guard.
6882 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
6885 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
6886 (ud2): New insn pattern.
6887 * config/i386/i386.c (ix86_expand_epilogue):
6888 For naked functions, generate ud2 instead of trap insn.
6890 2017-08-02 Marek Polacek <polacek@redhat.com>
6893 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
6895 2017-08-02 Tom de Vries <tom@codesourcery.com>
6896 Cesar Philippidis <cesar@codesourcery.com>
6898 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
6899 Add missing edge probabilities.
6901 2017-08-02 Tamar Christina <tamar.christina@arm.com>
6903 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
6906 2017-08-02 Jakub Jelinek <jakub@redhat.com>
6909 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
6910 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
6911 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
6912 prologue_seq sequences - if any.
6914 2017-08-02 Richard Biener <rguenther@suse.de>
6916 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
6917 via vectors if supported, integer extracts via punning if supported
6918 or otherwise vector extracts.
6920 2017-08-02 Richard Biener <rguenther@suse.de>
6922 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
6924 (bitmap_insert_into_set): ... this.
6926 2017-08-02 Richard Biener <rguenther@suse.de>
6928 PR tree-optimization/81633
6930 2015-08-17 Alan Hayward <alan.hayward@arm.com>
6932 PR tree-optimization/71752
6933 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
6935 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
6937 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
6938 (machine_function::call_ms2sysv_pad_out): Remove field.
6939 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
6940 (ix86_compute_frame_layout): Likewise.
6942 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
6945 * config/i386/i386.c (ix86_set_func_type): Disallow naked
6946 attribute with interrupt attribute.
6948 2017-08-01 Andrew Pinski <apinski@cavium.com>
6950 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
6951 BIT_INSERT_EXPR's operand 1
6952 to see if the types precision matches.
6954 2017-08-01 Martin Liska <mliska@suse.cz>
6957 * builtins.c (expand_builtin_memcpy_args): Remove.
6958 (expand_builtin_memcpy): Call newly added function
6959 expand_builtin_memory_copy_args.
6960 (expand_builtin_memcpy_with_bounds): Likewise.
6961 (expand_builtin_mempcpy): Remove last argument.
6962 (expand_builtin_mempcpy_with_bounds): Likewise.
6963 (expand_builtin_memory_copy_args): New function created from
6964 expand_builtin_mempcpy_args with small modifications.
6965 (expand_builtin_mempcpy_args): Remove.
6966 (expand_builtin_stpcpy): Remove unused argument.
6967 (expand_builtin): Likewise.
6968 (expand_builtin_with_bounds): Likewise.
6970 2017-08-01 Martin Liska <mliska@suse.cz>
6973 Make mempcpy more optimal (PR middle-end/70140).
6975 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6978 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6979 __builtin_vec_cmpne verify both arguments are compatible vectors
6980 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
6981 verify arg1_type is a pointer or array type. For __builtin_vec_st,
6982 move computation of aligned to after checking the argument types.
6986 * config/rs6000/vsx.md (vextract_fp_from_shorth,
6987 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
6990 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
6991 Doug Rupp <rupp@adacore.com>
6992 Olivier Hainque <hainque@adacore.com>
6994 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
6995 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
6997 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
6998 for TARGET_OS_CPP_BUILTIN.
6999 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
7000 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
7002 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
7003 passing required abi options to the assembler for EABI configurations.
7004 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
7005 of .text.hot and .text.unlikely sections for kernel modules when
7006 using ARM style exceptions.
7007 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
7008 options. Add EXTRA_CC1_SPEC.
7009 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
7011 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
7013 (ARM_TARGET2_DWARF_FORMAT): Define.
7014 * config/arm/t-vxworks: Adjust multilib control to removal of the
7015 Diab command line options.
7017 2017-08-01 Martin Liska <mliska@suse.cz>
7019 PR gcov-profile/81561
7020 * gcov.c (unblock): Make unblocking safe as we need to preserve
7021 index correspondence of blocks and block_lists.
7023 2017-08-01 Richard Biener <rguenther@suse.de>
7025 PR tree-optimization/81181
7026 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7027 (compute_antic): ... end of iteration here.
7029 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
7031 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
7032 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
7033 (ftree-slp-vectorize): Likewise.
7034 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
7035 can no longer be set independent of flag_tree_loop_vectorize.
7036 * omp-general.c (emp_max_vf): Likewise.
7037 * opts.c (enable_fdo_optimizations): Remove references to
7038 flag_tree_vectorize, these are now implicit.
7039 (common_handle_option): Remove handling for OPT_ftree_vectorize,
7040 and leave it for the options machinery.
7042 2017-08-01 Martin Liska <mliska@suse.cz>
7045 * builtins.c (expand_builtin_memcpy_args): Remove.
7046 (expand_builtin_memcpy): Call newly added function
7047 expand_builtin_memory_copy_args.
7048 (expand_builtin_memcpy_with_bounds): Likewise.
7049 (expand_builtin_mempcpy): Remove last argument.
7050 (expand_builtin_mempcpy_with_bounds): Likewise.
7051 (expand_builtin_memory_copy_args): New function created from
7052 expand_builtin_mempcpy_args with small modifications.
7053 (expand_builtin_mempcpy_args): Remove.
7054 (expand_builtin_stpcpy): Remove unused argument.
7055 (expand_builtin): Likewise.
7056 (expand_builtin_with_bounds): Likewise.
7058 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7061 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
7062 print "ds:" only for immediates in generic address space.
7064 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
7067 * config/i386/i386.c (ix86_funciton_naked): New prototype.
7068 (ix86_function_ok_for_sibcall): Return false for naked functions.
7070 2017-08-01 Richard Biener <rguenther@suse.de>
7072 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
7073 (compute_antic): Seed worklist with exit block predecessors.
7074 * cfganal.c (dfs_find_deadend): For a cycle return the source
7075 of the edge closing it.
7077 2017-08-01 Tamar Christina <tamar.christina@arm.com>
7079 * config/aarch64/aarch64.c
7080 (aarch64_can_const_movi_rtx_p): Move 0 check.
7082 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7084 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
7085 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
7087 * match.pd: Ditto in address comparison pattern.
7089 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7091 PR tree-optimization/81627
7092 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
7093 closed ssa form for store-store chain.
7095 2017-08-01 Bin Cheng <bin.cheng@arm.com>
7097 PR tree-optimization/81620
7098 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
7099 for store-store chain.
7101 2017-08-01 Jakub Jelinek <jakub@redhat.com>
7103 PR tree-optimization/81588
7104 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
7105 ranges[i].in_p, invert comparison code ccode. For >/>=,
7106 swap rhs1 and rhs2 and comparison code unconditionally,
7107 for </<= don't do that. Don't swap rhs1/rhs2 again if
7108 ranges[i].in_p, instead invert comparison code ccode if
7109 opcode or oe->rank is BIT_IOR_EXPR.
7112 * optabs.def (vec_extract_optab, vec_init_optab): Change from
7113 a direct optab to conversion optab.
7114 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7115 with GET_MODE_INNER as last argument instead of optab_handler.
7116 * expmed.c (extract_bit_field_1): Likewise. Use vector from
7117 vector extraction if possible and optab is available.
7118 * expr.c (store_constructor): Use convert_optab_handler instead
7119 of optab_handler. Use vector initialization from smaller
7120 vectors if possible and optab is available.
7121 * tree-vect-stmts.c (vectorizable_load): Likewise.
7122 * doc/md.texi (vec_extract, vec_init): Document that the optabs
7124 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
7125 of vec_init from half-sized vectors with the same element mode.
7126 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
7127 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
7128 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
7129 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
7130 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
7131 after mode in gen_vec_extract* calls.
7132 (vec_extract<mode>): Renamed to ...
7133 (vec_extract<mode><ssescalarmodelower>): ... this.
7134 (vec_extract<mode><ssehalfvecmodelower>): New expander.
7135 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
7136 element mode after mode in gen_vec_init* calls.
7137 (VEC_INIT_HALF_MODE): New mode iterator.
7138 (vec_init<mode>): Renamed to ...
7139 (vec_init<mode><ssescalarmodelower>): ... this.
7140 (vec_init<mode><ssehalfvecmodelower>): New expander.
7141 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
7142 (vec_extractv2sfsf): ... this.
7143 (vec_initv2sf): Renamed to ...
7144 (vec_initv2sfsf): ... this.
7145 (vec_extractv2si): Renamed to ...
7146 (vec_extractv2sisi): ... this.
7147 (vec_initv2si): Renamed to ...
7148 (vec_initv2sisi): ... this.
7149 (vec_extractv4hi): Renamed to ...
7150 (vec_extractv4hihi): ... this.
7151 (vec_initv4hi): Renamed to ...
7152 (vec_initv4hihi): ... this.
7153 (vec_extractv8qi): Renamed to ...
7154 (vec_extractv8qiqi): ... this.
7155 (vec_initv8qi): Renamed to ...
7156 (vec_initv8qiqi): ... this.
7157 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
7158 (vec_init<mode>): Renamed to ...
7159 (vec_init<mode><VEC_base_l>): ... this.
7160 (vec_extract<mode>): Renamed to ...
7161 (vec_extract<mode><VEC_base_l>): ... this.
7162 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
7163 (vec_initv2sfsf): ... this.
7164 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
7165 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7166 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
7167 element mode after mode in gen_vec_init* calls.
7168 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
7169 (vec_init<mode><Vel>): ... this.
7170 (vec_extract<mode>): Renamed to ...
7171 (vec_extract<mode><Vel>): ... this.
7172 * config/aarch64/iterators.md (Vel): New mode attribute.
7173 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
7174 Add element mode after mode in gen_vec_extract* calls.
7175 * config/s390/vector.md (non_vec_l): New mode attribute.
7176 (vec_extract<mode>): Renamed to ...
7177 (vec_extract<mode><non_vec_l>): ... this.
7178 (vec_init<mode>): Renamed to ...
7179 (vec_init<mode><non_vec_l>): ... this.
7180 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
7181 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
7183 * config/arm/iterators.md (V_elem_l): New mode attribute.
7184 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
7185 (vec_extract<mode><V_elem_l>): ... this.
7186 (vec_extractv2di): Renamed to ...
7187 (vec_extractv2didi): ... this.
7188 (vec_init<mode>): Renamed to ...
7189 (vec_init<mode><V_elem_l>): ... this.
7190 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
7191 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
7192 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
7193 Add element mode after gen_vec_extract* calls.
7194 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
7195 (vec_init<mode><unitmode>): ... this.
7196 (vec_extract<mode>): Renamed to ...
7197 (vec_extract<mode><unitmode>): ... this.
7198 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
7199 (vec_init<mode><unitmode>): ... this.
7200 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
7201 (vec_initv2sfsf): ... this.
7202 (vec_extractv2sf): Renamed to ...
7203 (vec_extractv2sfsf): ... this.
7204 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
7205 Add element mode after gen_vec_extract* calls.
7206 * config/mips/mips.md (unitmode): New mode iterator.
7207 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
7208 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
7209 * config/spu/spu.md (inner_l): New mode attribute.
7210 (vec_init<mode>): Renamed to ...
7211 (vec_init<mode><inner_l>): ... this.
7212 (vec_extract<mode>): Renamed to ...
7213 (vec_extract<mode><inner_l>): ... this.
7214 * config/sparc/sparc.md (veltmode): New mode iterator.
7215 (vec_init<VMALL:mode>): Renamed to ...
7216 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
7217 * config/ia64/vect.md (vec_initv2si): Renamed to ...
7218 (vec_initv2sisi): ... this.
7219 (vec_initv2sf): Renamed to ...
7220 (vec_initv2sfsf): ... this.
7221 (vec_extractv2sf): Renamed to ...
7222 (vec_extractv2sfsf): ... this.
7223 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
7224 (vec_init<mode>): Renamed to ...
7225 (vec_init<mode><VEC_base_l>): ... this.
7226 (vec_extract<mode>): Renamed to ...
7227 (vec_extract<mode><VEC_base_l>): ... this.
7228 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
7229 (vec_initv2sfsf): ... this.
7230 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
7231 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
7232 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
7233 gen_vec_init* calls.
7235 2017-08-01 Richard Biener <rguenther@suse.de>
7237 PR tree-optimization/81297
7238 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
7239 TREE_OVERFLOW from INTEGER_CSTs.
7241 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
7243 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
7245 2017-07-31 Carl Love <cel@us.ibm.com>
7247 * config/rs6000/rs6000-c: Add support for built-in functions
7248 vector signed char vec_xl_be (signed long long, signed char *);
7249 vector unsigned char vec_xl_be (signed long long, unsigned char *);
7250 vector signed int vec_xl_be (signed long long, signed int *);
7251 vector unsigned int vec_xl_be (signed long long, unsigned int *);
7252 vector signed long long vec_xl_be (signed long long, signed long long *);
7253 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
7254 vector signed short vec_xl_be (signed long long, signed short *);
7255 vector unsigned short vec_xl_be (signed long long, unsigned short *);
7256 vector double vec_xl_be (signed long long, double *);
7257 vector float vec_xl_be (signed long long, float *);
7258 * config/rs6000/altivec.h (vec_xl_be): Add #define.
7259 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
7260 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
7262 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
7263 (altivec_expand_builtin): Add switch statement to call
7264 altivec_expand_xl_be for each builtin.
7265 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
7266 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
7267 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
7268 __builtin_vsx_le_be_v16qi.
7269 * doc/extend.texi: Update the built-in documentation file for the
7270 new built-in functions.
7272 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7275 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
7277 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7279 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7281 * config.gcc: Add z14.
7282 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7283 CPU model numbers for z13s and z14.
7284 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
7286 * config/s390/s390-opts.h (enum processor_type): Rename
7287 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7288 * config/s390/s390.c (processor_table): Add field for CPU name to
7289 be passed to Binutils.
7290 (s390_asm_output_machine_for_arch): Use the new field in
7291 processor_table for Binutils.
7292 (s390_expand_builtin): Replace arch12 with z14.
7293 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
7294 (s390_get_sched_attrmask): Likewise.
7295 (s390_get_unit_mask): Likewise.
7296 * config/s390/s390.opt: Add z14 to processor_type enum.
7298 2017-07-31 Martin Jambor <mjambor@suse.cz>
7301 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
7302 regardless of optimization level.
7304 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
7305 Martin Liska <mliska@suse.cz>
7307 * predict.def: Remove old comment and adjust probability.
7308 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
7311 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
7314 * config/i386/i386.c (ix86_function_naked): New function.
7315 (ix86_can_use_return_insn_p): Return false for naked functions.
7316 (ix86_expand_prologue): Skip prologue for naked functions.
7317 (ix86_expand_epilogue): Skip epilogue for naked functions
7318 and emit trap instruction.
7319 (ix86_warn_func_return): New function.
7320 (ix86_attribute_table): Add "naked" attribute specification.
7321 (TARGET_WARN_FUNC_RETURN): Define.
7322 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
7324 2017-07-31 Martin Liska <mliska@suse.cz>
7326 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
7327 (dump_gimple_bb_header): Always dump BB info.
7328 (pp_cfg_jump): Do not append info about BB when dumping a jump.
7330 2017-07-31 Martin Liska <mliska@suse.cz>
7333 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
7334 also with current_function_decl non-null equality.
7336 2017-07-31 Jakub Jelinek <jakub@redhat.com>
7339 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
7340 change type to the element type, instead add eltype variable and
7341 use it where we are interested in the element type.
7343 PR tree-optimization/81603
7344 * ipa-polymorphic-call.c
7345 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
7346 offset arithmetic in offset_int, bail out if the resulting bit offset
7347 doesn't fit into shwi.
7349 2017-07-31 Martin Liska <mliska@suse.cz>
7351 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
7352 (gimplify_save_expr): Fix comment.
7354 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
7357 * config/i386/i386.c (ix86_function_arg): Update arguments for
7359 (ix86_compute_frame_layout): Set the initial stack offset to
7360 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
7361 INCOMING_FRAME_SP_OFFSET.
7362 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
7363 stack before exception handler returns.
7364 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
7365 the 'ERROR_CODE' for exception handler.
7367 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
7369 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
7370 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
7371 (ASM_OUTPUT_REG_POP): Ditto.
7372 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
7373 instead of asm_fprintf to output pure string.
7375 2017-07-29 Jakub Jelinek <jakub@redhat.com>
7377 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
7378 to imported_module_or_decl hook.
7379 (debug_nothing_tree_tree_tree_bool): Remove.
7380 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
7381 * debug.c (do_nothing_debug_hooks): Use
7382 debug_nothing_tree_tree_tree_bool_bool instead of
7383 debug_nothing_tree_tree_tree_bool.
7384 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7385 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
7386 * sdbout.c (sdb_debug_hooks): Likewise.
7387 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
7388 (gen_namespace_die): Add DW_AT_export_symbols attribute if
7390 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
7391 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
7392 attribute, don't add anything.
7394 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7396 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
7397 (fold_build2_stat_loc): Likewise.
7398 (fold_build3_stat_loc): Likewise.
7399 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
7400 (fold_build1_loc): Remove macro.
7401 (fold_build2_loc): Likewise.
7402 (fold_build3_loc): Likewise.
7404 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7406 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
7407 (gimple_build_debug_bind_source_stat): Likewise.
7408 * gimple.h (gimple_build_debug_bind): Remove macro.
7409 (gimple_build_debug_bind_source): Likewise.
7411 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7413 * bitmap.c (bitmap_alloc): Adjust.
7414 (bitmap_gc_alloc): Likewise.
7415 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
7417 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7419 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
7420 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
7421 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
7422 (bitmap_gc_alloc_stat): Likewise.
7423 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
7425 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7427 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
7428 * rtl.h (shallow_copy_rtx): Remove macro.
7430 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7432 * emit-rtl.c (gen_raw_REG): Adjust.
7433 * gengenrtl.c (gendef): Likewise.
7434 * rtl.c (rtx_alloc_stat): Remove _stat from name.
7435 * rtl.h (rtx_alloc): Remove macro.
7437 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7439 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
7440 (build_tree_list_stat): Likewise.
7441 * tree.h (build_tree_list): Remove macro.
7442 (build_tree_list_vec): Likewise.
7444 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7446 * tree.c (make_vector_stat): Remove _stat from name.
7447 (build_vector_stat): Likewise.
7448 * tree.h (make_vector_stat): Remove macro.
7449 (build_vector_stat): Likewise.
7451 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7453 * tree.h (build_var_debug_value): Remove prototype.
7455 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7457 * tree.c (tree_cons_stat): Remove _stat from name.
7458 * tree.h (tree_cons): Remove macro.
7460 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7462 * tree.c (build_vl_exp_stat): Remove _stat from name.
7463 * tree.h (build_vl_exp): Remove macro.
7465 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7467 * tree.c (build_decl_stat): Remove _stat from name.
7468 * tree.h (build_decl): Remove macro.
7470 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7472 * gimple.c (gimple_build_with_ops_stat): Adjust.
7473 (gimple_alloc_stat): Remove _stat from name.
7474 * gimple.h (gimple_alloc): Remove macro.
7476 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7478 * tree.c (make_tree_vec_stat): Remove _stat from name.
7479 (grow_tree_vec_stat): Likewise.
7480 * tree.h (make_tree_vec_stat): Adjust prototype.
7481 (grow_tree_vec_stat): Likewise.
7482 (make_tree_vec): Remove macro.
7483 (grow_tree_vec): Likewise.
7485 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7487 * fold-const.c (fold_build1_stat_loc): Adjust.
7488 (fold_build2_stat_loc): Likewise.
7489 (fold_build3_stat_loc): Likewise.
7490 * tree.c (build0_stat): Remove _stat from name.
7491 (build1_stat): Likewise.
7492 (build2_stat): Likewise.
7493 (build3_stat): Likewise.
7494 (build4_stat): Likewise.
7495 (build5_stat): Likewise.
7496 * tree.h (build1_loc): Remove macro, and rename _stat function
7498 (build2_loc): Likewise.
7499 (build3_loc): Likewise.
7500 (build4_loc): Likewise.
7501 (build5_loc): Likewise.
7503 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7505 * tree.c (make_int_cst_stat): Remove _stat from name.
7506 * tree.h (make_int_cst_stat): Adjust prototype.
7507 (make_int_cst): Remove macro.
7509 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7511 * tree.c (make_tre_binfo_stat): Remove _stat from name.
7512 * tree.h (make_tree_binfo_stat): Adjust prototype.
7513 (make_tree_binfo): Remove.
7515 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7517 * tree.c (copy_node_stat): Rename to copy_node.
7518 (build_distinct_type_copy): Adjust.
7519 * tree.h (copy_node_stat): Adjust prototype.
7520 (copy_node): Remove macro.
7522 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7524 * tree.c (make_node_stat): rename to make_node.
7525 (build_tree_list_stat): Adjust.
7526 (build0_stat): Likewise.
7527 (build2_stat): Likewise.
7528 (build3_stat): Likewise.
7529 (build4_stat): Likewise.
7530 (build5_stat): Likewise.
7531 (build_decl_stat): Likewise.
7532 * tree.h (make_node_stat): Adjust prototype.
7533 (make_node): remove macro.
7535 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
7537 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
7538 (PPC_FEATURE2_SCV): Likewise.
7539 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
7541 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7543 * config/aarch64/aarch64.c
7544 (aarch64_internal_mov_immediate): Add new special pattern.
7545 * config/aarch64/aarch64.md (*movdi_aarch64):
7546 Add reg/32bit const mov case.
7548 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7549 Richard Sandiford <richard.sandiford@linaro.org>
7551 * config/aarch64/aarch64.md (mov<mode>): Generalize.
7552 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
7553 Add integer and movi cases.
7554 (movi-split-hf-df-sf split, fp16): New.
7555 (enabled): Added TARGET_FP_F16INST.
7556 * config/aarch64/iterators.md (GPF_HF): New.
7557 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
7559 2017-07-28 Tamar Christina <tamar.christina@arm.com>
7561 * config/aarch64/aarch64.c
7562 (aarch64_simd_container_mode): Add prototype.
7563 (aarch64_expand_mov_immediate): Add HI support.
7564 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7565 (aarch64_can_const_movi_rtx_p): New.
7566 (aarch64_preferred_reload_class):
7567 Remove restrictions of using FP registers for certain SIMD operations.
7568 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
7569 (aarch64_valid_floating_const): Add integer move validation.
7570 (aarch64_simd_imm_scalar_p): Remove.
7571 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
7572 (aarch64_legitimate_constant_p): Expand list of supported cases.
7573 * config/aarch64/aarch64-protos.h
7574 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
7575 (aarch64_reinterpret_float_as_int): New.
7576 (aarch64_simd_imm_scalar_p): Remove.
7577 * config/aarch64/constraints.md (Uvi): New.
7578 (Dd): Split into Ds and new Dd.
7579 * config/aarch64/aarch64.md (*movsi_aarch64):
7581 (*movdi_aarch64): Add SIMD mov case.
7583 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7585 * tree-predcom.c: (struct chain): Handle store-store chain in which
7586 stores for elimination only store loop invariant values.
7587 (execute_pred_commoning_chain): Ditto.
7588 (prepare_initializers_chain_store_elim): Ditto.
7589 (prepare_finalizers): Ditto.
7590 (is_inv_store_elimination_chain): New function.
7591 (initialize_root_vars_store_elim_1): New function.
7593 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7595 * tree-predcom.c: Revise general description of the pass.
7596 (enum chain_type): New enum type for store elimination.
7597 (struct chain): New field supporting store elimination.
7598 (struct component): Ditto.
7599 (dump_chain): Dump store-stores chain.
7600 (release_chain): Release resources.
7601 (split_data_refs_to_components): Compute and create component
7602 contains only stores for elimination.
7603 (get_chain_last_ref_at): New function.
7604 (make_invariant_chain): Initialization.
7605 (make_rooted_chain): Specify chain type in parameter and record it.
7606 (add_looparound_copies): Skip for store-stores chain.
7607 (determine_roots_comp): Compute type of chain and pass it to
7609 (initialize_root_vars_store_elim_2): New function.
7610 (finalize_eliminated_stores): New function.
7611 (remove_stmt): Handle store for elimination.
7612 (execute_pred_commoning_chain): Execute predictive commoning on
7614 (determine_unroll_factor): Skip unroll for store-stores chain.
7615 (prepare_initializers_chain_store_elim): New function.
7616 (prepare_initializers_chain): Hanlde store-store chain.
7617 (prepare_finalizers_chain, prepare_finalizers): New function.
7618 (tree_predictive_commoning_loop): Return integer value indicating
7619 if loop is unrolled or lcssa form is corrupted.
7620 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
7622 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7624 * tree-predcom.c (initialize_root): Delete.
7625 (execute_pred_commoning_chain): Initialize root vars and replace
7626 reference of non-combined chain directly, rather than call above
7629 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7631 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
7632 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
7634 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7636 * tree-predcom.c (struct chain): New field init_seq.
7637 (release_chain): Release init_seq.
7638 (prepare_initializers_chain): Record intialization stmts in above
7640 (insert_init_seqs): New function.
7641 (tree_predictive_commoning_loop): Call insert_init_seqs.
7643 2017-07-28 Bin Cheng <bin.cheng@arm.com>
7645 * tree-predcom.c (determine_roots_comp): Skip trivial components.
7647 2017-07-28 Richard Biener <rguenther@suse.de>
7649 * match.pd: Remove superfluous :c.
7650 * genmatch.c (simplify::id): Add member.
7651 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
7653 (current_id): New global.
7654 (dt_node::parent): Move from ...
7655 (dt_operand::parent): ... here. Add for_id member.
7656 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
7657 (decision_tree::find_node): Relax order requirement when
7658 merging DT_TRUE nodes to ones inbetween the current simplify
7659 and the one we try to merge with. Add diagnostic whenever
7660 we need to enforce pattern order by not merging.
7661 (decision_tree::insert): Set current_id.
7662 (decision_tree::print_node): Dump parent node and for_id.
7663 (parser::last_id): Add member.
7664 (parser::push_simplify): Assign unique id.
7665 (parser::parser): Initialize last_id.
7667 2017-07-28 Martin Liska <mliska@suse.cz>
7670 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
7671 gimple_build_debug_bind.
7673 2017-07-28 Richard Biener <rguenther@suse.de>
7675 PR tree-optimization/81502
7676 * match.pd: Add pattern combining BIT_INSERT_EXPR with
7678 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
7680 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
7681 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
7682 for BIT_FIELD_REF args.
7683 * fold-const.c (make_bit_field_ref): Likewise.
7684 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7686 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7689 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
7690 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
7691 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
7692 Or it into SANITIZER_UNDEFINED.
7693 * ubsan.c: Include gimple-fold.h and varasm.h.
7694 (ubsan_expand_ptr_ifn): New function.
7695 (instrument_pointer_overflow): New function.
7696 (maybe_instrument_pointer_overflow): New function.
7697 (instrument_object_size): Formatting fix.
7698 (pass_ubsan::execute): Call instrument_pointer_overflow
7699 and maybe_instrument_pointer_overflow.
7700 * internal-fn.c (expand_UBSAN_PTR): New function.
7701 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
7702 * sanitizer.def (__ubsan_handle_pointer_overflow,
7703 __ubsan_handle_pointer_overflow_abort): New builtins.
7704 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
7705 * internal-fn.def (UBSAN_PTR): New internal function.
7706 * opts.c (sanitizer_opts): Add pointer-overflow.
7707 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
7708 * fold-const.c (build_range_check): Compute pointer range check in
7709 integral type if pointer arithmetics would be needed. Formatting
7712 2017-07-28 Martin Liska <mliska@suse.cz>
7715 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
7716 parameters that are of a variable-length.
7718 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
7720 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
7722 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
7723 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
7724 (CRT_CALL_STATIC_FUNCTION): Likewise.
7725 (ASM_DEFAULT_SPEC): New define.
7726 (ASM_SPEC32): Likewise.
7727 (ASM_SPEC64): Likewise.
7728 (ASM_SPEC_COMMON): Likewise.
7729 (ASM_SPEC): Likewise.
7730 (INVALID_64BIT): Likewise.
7731 (LINK_OS_DEFAULT_SPEC): Likewise.
7732 (LINK_OS_SPEC32): Likewise.
7733 (LINK_OS_SPEC64): Likewise.
7734 (POWERPC_LINUX): Likewise.
7735 (PTRDIFF_TYPE): Likewise.
7736 (RESTORE_FP_PREFIX): Likewise.
7737 (RESTORE_FP_SUFFIX): Likewise.
7738 (SAVE_FP_PREFIX): Likewise.
7739 (SAVE_FP_SUFFIX): Likewise.
7740 (SIZE_TYPE): Likewise.
7741 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
7742 (TARGET_64BIT): Likewise.
7743 (TARGET_64BIT): Likewise.
7744 (TARGET_AIX): Likewise.
7745 (WCHAR_TYPE_SIZE): Likewise.
7746 (WCHAR_TYPE): Undefine.
7747 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
7748 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
7749 (CPP_OS_RTEMS_SPEC): Delete.
7750 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
7751 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
7753 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
7755 2017-07-28 Jakub Jelinek <jakub@redhat.com>
7757 PR tree-optimization/81578
7758 * tree-parloops.c (build_new_reduction): Bail out if
7759 reduction_code isn't one of the standard OpenMP reductions.
7760 Move the details printing after that decision.
7762 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7764 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
7765 related to reload_in_progress.
7766 (splat_input_operand): Likewise.
7767 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
7769 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
7771 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
7772 (TARGET_INSTANTIATE_DECLS): Likewise.
7773 (legitimate_indexed_address_p): Delete reload_in_progress code.
7774 (rs6000_debug_legitimate_address_p): Likewise.
7775 (rs6000_eliminate_indexed_memrefs): Likewise.
7776 (rs6000_emit_le_vsx_store): Likewise.
7777 (rs6000_emit_move_si_sf_subreg): Likewise.
7778 (rs6000_emit_move): Likewise.
7779 (register_to_reg_type): Likewise.
7780 (rs6000_pre_atomic_barrier): Likewise.
7781 (rs6000_machopic_legitimize_pic_address): Likewise.
7782 (rs6000_allocate_stack_temp): Likewise.
7783 (rs6000_address_for_fpconvert): Likewise.
7784 (rs6000_address_for_altivec): Likewise.
7785 (rs6000_secondary_memory_needed_rtx): Delete function.
7786 (rs6000_check_sdmode): Likewise.
7787 (rs6000_alloc_sdmode_stack_slot): Likewise.
7788 (rs6000_instantiate_decls): Likewise.
7789 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
7790 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
7791 Delete reload_in_progress.
7792 (*vec_reload_and_plus_<mptrsize>): Likewise.
7793 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
7794 (vsx_div_v2di): Likewise.
7795 (vsx_udiv_v2di): Likewise.
7797 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7799 * config/rs6000/rs6000.opt (mlra): Replace with stub.
7800 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
7801 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
7802 (rs6000_debug_reg_global): Delete print of LRA status.
7803 (rs6000_option_override_internal): Delete dead LRA related code.
7804 (rs6000_lra_p): Delete function.
7805 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
7807 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
7809 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
7810 * config/riscv/rtems.h: New file.
7812 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7813 Sudakshina Das <sudi.das@arm.com>
7815 * config/aarch64/aarch64.md
7816 (define_split for and<mode>3nr_compare): Move
7817 non aarch64_logical_operand to a register.
7818 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
7819 register immediate operand to a register.
7820 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
7822 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
7825 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
7827 2017-07-27 Richard Biener <rguenther@suse.de>
7829 PR tree-optimization/81573
7830 PR tree-optimization/81494
7831 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
7832 multi defuse cycle case.
7834 2017-07-27 Richard Biener <rguenther@suse.de>
7836 PR tree-optimization/81571
7837 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
7840 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
7842 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
7843 earlier and only if MASK_FPU is set. Adjust formatting.
7845 2017-07-27 Martin Liska <mliska@suse.cz>
7847 * opt-functions.awk: Add validation of value of Init.
7848 * optc-gen.awk: Pass new argument.
7850 2017-07-27 Martin Liska <mliska@suse.cz>
7852 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
7853 Fix wrong condition.
7855 2017-07-27 Martin Liska <mliska@suse.cz>
7857 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
7858 BBs and edges seen by autoFDO.
7860 2017-07-27 Richard Biener <rguenther@suse.de>
7862 PR tree-optimization/81502
7863 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
7864 with incompatible but same sized type.
7865 (execute_update_addresses_taken): Likewise.
7867 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
7869 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
7870 flag_tree_loop_vectorize rather than flag_tree_vectorize.
7872 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7875 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
7876 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
7877 Change s_operand to memory_operand.
7879 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
7881 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
7882 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
7883 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
7884 Emit instructions rather than returning an expression. Handle TFmode
7885 and KFmode by casting to TImode.
7886 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
7887 (rs6000_emit_le_vsx_store): Likewise.
7888 * config/rs6000/vsx.md (VSX_TI): New iterator.
7889 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
7890 (*vsx_le_undo_permute_<mode>): Likewise.
7891 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
7892 emit the split sequence.
7893 (*vsx_le_perm_store_<mode>): Likewise.
7895 2017-07-27 Jakub Jelinek <jakub@redhat.com>
7897 PR tree-optimization/81555
7898 PR tree-optimization/81556
7899 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
7900 if true, force CHANGED for the recursive invocation.
7901 (reassociate_bb): Remember original length of ops array, pass
7902 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
7904 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
7905 attributes for noipa attribute. For naked attribute use
7906 lookup_attribute first before lookup_attribute_spec.
7907 * final.c (rest_of_handle_final): Disable IPA RA for functions with
7909 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
7910 for functions with noipa attribute.
7911 (cgraph_externally_visible_p): Return true for functions with noipa
7913 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
7914 for functions with noipa attribute.
7915 * doc/extend.texi: Document noipa function attribute.
7916 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
7917 also for functions with noipa attribute.
7918 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
7920 2017-07-26 Andrew Pinski <apinski@cavium.com>
7922 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
7923 vec_unalign_load_cost and vec_unalign_store_cost.
7925 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7927 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
7928 -mvsx-small-integer option.
7929 (ISA_3_0_MASKS_IEEE): Likewise.
7930 (OTHER_VSX_VECTOR_MASKS): Likewise.
7931 (POWERPC_MASKS): Likewise.
7932 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
7933 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
7934 code, only testing for DImode being allowed in non-VSX floating
7936 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
7937 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
7939 (rs6000_option_override_internal): Delete -mvsx-small-integer.
7940 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
7941 TARGET_P8_VECTOR test.
7942 (rs6000_secondary_reload_simple_move): Likewise.
7943 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
7944 since TARGET_P9_VECTOR was already tested.
7945 (rs6000_opt_masks): Remove -mvsx-small-integer.
7946 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
7947 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7949 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
7950 test for TARGET_VEXTRACTUB was used, and that uses
7952 (p9 extract splitter): Likewise.
7953 (vsx_extract_<mode>_di_p9): Likewise.
7954 (vsx_extract_<mode>_store_p9): Likewise.
7955 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
7956 for TARGET_P9_VECTOR was used. Delete code that is now dead with
7957 the elimination of TARGET_VSX_SMALL_INTEGER.
7958 (vsx_extract_<mode>_p8): Likewise.
7959 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
7960 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
7961 (vsx_set_<mode>_p9): Likewise.
7962 (vsx_set_v4sf_p9): Likewise.
7963 (vsx_set_v4sf_p9_zero): Likewise.
7964 (vsx_insert_extract_v4sf_p9): Likewise.
7965 (vsx_insert_extract_v4sf_p9_2): Likewise.
7966 * config/rs6000/rs6000.md (sign extend splitter): Change
7967 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
7968 (floatsi<mode>2_lfiwax_mem): Likewise.
7969 (floatunssi<mode>2_lfiwzx_mem): Likewise.
7970 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
7971 since a test for TARGET_P9_VECTOR was used.
7972 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7973 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
7974 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7975 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7976 TARGET_P8_VECTOR test.
7977 (fix_trunc<mode>si2_stfiwx): Likewise.
7978 (fix_trunc<mode>si2_internal): Likewise.
7979 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
7980 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7982 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7983 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
7984 TARGET_P8_VECTOR test.
7985 (fixuns_trunc<mode>si2_stfiwx): Likewise.
7986 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
7987 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
7989 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
7990 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
7991 since a test for TARGET_P9_VECTOR was used.
7992 (splitter for loading small constants): Likewise.
7994 2017-07-26 Andrew Pinski <apinski@cavium.com>
7996 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
7999 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
8002 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
8003 (fp_valid_at): Likewise.
8005 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8007 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
8008 (qdf24xx_addrcost_table): Likewise.
8009 (cortexa57_tunings): Update to use generic_branch_cost.
8010 (cortexa72_tunings): Likewise.
8011 (cortexa73_tunings): Likewise.
8012 (qdf24xx_tunings): Likewise.
8014 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
8016 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
8017 (thunderx2t99_branch_cost): Likewise.
8018 (cortexa35_tunings): Update to use generic_branch_cost.
8019 (cortexa53_tunings): Likewise.
8020 (cortexa57_tunings): Likewise.
8021 (cortexa72_tunings): Likewise.
8022 (cortexa73_tunings): Likewise.
8023 (thunderx2t99_tunings): Likewise.
8025 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8027 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
8028 (sparc_option_override): Honour MASK_FSMULD.
8029 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
8030 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
8031 * config/sparc/sparc.opt (mfsmuld): New option.
8032 * doc/invoke.texi (mfsmuld): Document option.
8034 2017-07-26 Marek Polacek <polacek@redhat.com>
8037 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
8039 2017-07-26 Richard Biener <rguenther@suse.de>
8041 * gimple-match-head.c (do_valueize): Return OP if valueize
8043 (get_def): New helper to get at the def stmt of a SSA name
8045 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
8046 do_valueize to get at the def stmt.
8047 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
8049 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
8052 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
8054 2017-07-26 Martin Liska <mliska@suse.cz>
8057 * function.c (expand_function_start): Make expansion of
8058 nonlocal_goto_save_area after parm_birth_insn.
8060 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
8062 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
8063 from all CPU target flags enable members.
8065 2017-07-26 Richard Biener <rguenther@suse.de>
8067 * genmatch.c (dt_simplify::gen): Make iterator vars const.
8068 (decision_tree::gen): Make 'type' const.
8069 (write_predicate): Likewise.
8071 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8073 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
8074 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
8075 (rs6000_option_override_internal): Likewise.
8076 (rs6000_expand_vector_set): Likewise.
8077 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
8078 (TARGET_UPPER_REGS_SF): Likewise.
8079 (TARGET_UPPER_REGS_DI): Likewise.
8080 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
8081 (TARGET_DIRECT_MOVE_64BIT): Likewise.
8082 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
8083 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
8084 (Splitters for DI constants in Altivec registers): Likewise.
8085 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
8086 (vsx_set_v4sf_p9): Likewise.
8087 (vsx_set_v4sf_p9_zero): Likewise.
8088 (vsx_insert_extract_v4sf_p9): Likewise.
8089 (vsx_insert_extract_v4sf_p9_2): Likewise.
8091 2017-07-25 Carl Love <cel@us.ibm.com>
8093 * doc/extend.texi: Update the built-in documentation file for the
8094 existing built-in functions
8095 vector signed char vec_cnttz (vector signed char);
8096 vector unsigned char vec_cnttz (vector unsigned char);
8097 vector signed short vec_cnttz (vector signed short);
8098 vector unsigned short vec_cnttz (vector unsigned short);
8099 vector signed int vec_cnttz (vector signed int);
8100 vector unsigned int vec_cnttz (vector unsigned int);
8101 vector signed long long vec_cnttz (vector signed long long);
8102 vector unsigned long long vec_cnttz (vector unsigned long long);
8104 2017-07-25 Andrew Pinski <apinski@cavium.com>
8106 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
8107 accesses where the use is for the first operand of a BIT_INSERT.
8109 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8112 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8113 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8115 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
8117 * config/i386/gstabs.h: Delete.
8118 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
8120 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
8122 * config/i386/i386.c (ix86_decompose_address): Do not check for
8123 register RTX when looking at index_reg or base_reg.
8124 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
8126 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
8128 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
8129 to update EH info here.
8131 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8133 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
8135 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8137 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
8139 2017-07-25 Torsten Duwe <duwe@suse.de>
8141 * common.opt: Introduce -fpatchable-function-entry
8142 command line option, and its variables function_entry_patch_area_size
8143 and function_entry_patch_area_start.
8144 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
8145 including a two-value parser.
8146 * target.def (print_patchable_function_entry): New target hook.
8147 * targhooks.h (default_print_patchable_function_entry): New function.
8148 * targhooks.c (default_print_patchable_function_entry): Likewise.
8149 * toplev.c (process_options): Switch off IPA-RA if
8150 patchable function entries are being generated.
8151 * varasm.c (assemble_start_function): Look at the
8152 patchable-function-entry command line switch and current
8153 function attributes and maybe generate NOP instructions by
8154 calling the print_patchable_function_entry hook.
8155 * doc/extend.texi: Document patchable_function_entry attribute.
8156 * doc/invoke.texi: Document -fpatchable_function_entry
8157 command line option.
8158 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
8160 * doc/tm.texi: Re-generate.
8162 2017-07-25 Jakub Jelinek <jakub@redhat.com>
8165 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
8166 TARGET_AVX512DQ rather than TARGET_AVX512BW.
8168 2017-07-25 Tamar Christina <tamar.christina@arm.com>
8170 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
8172 2017-07-25 Richard Biener <rguenther@suse.de>
8174 PR tree-optimization/81455
8175 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
8176 not walk in cycles when looking for guards.
8178 2017-07-25 Richard Biener <rguenther@suse.de>
8180 PR tree-optimization/81529
8181 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
8182 when optimizing backedge uses.
8184 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
8186 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
8188 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
8189 to dl_section_ref. On AIX, append an expression to subtract
8190 the size of the section length to dl_section_ref.
8192 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
8194 * configure.ac: If any of the config.* scripts fail, exit 1.
8195 * configure: Regenerate.
8197 2017-07-25 Richard Biener <rguenther@suse.de>
8200 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
8201 of immediate uses, be more verbose on errors.
8203 2017-07-25 Richard Biener <rguenther@suse.de>
8205 PR tree-optimization/81510
8206 * tree-vect-loop.c (vect_is_simple_reduction): When the
8207 reduction stmt is not inside the loop bail out.
8209 2017-07-25 Richard Biener <rguenther@suse.de>
8211 PR tree-optimization/81303
8212 * tree-vect-loop-manip.c (vect_loop_versioning): Build
8213 profitability check against LOOP_VINFO_NITERSM1.
8215 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
8217 * domwalk.c (cmp_bb_postorder): Simplify.
8218 (sort_bbs_postorder): New function. Use it...
8219 (dom_walker::walk): ...here to optimize common cases.
8221 2017-07-25 Martin Liska <mliska@suse.cz>
8224 * ipa-visibility.c (function_and_variable_visibility): Make the
8225 redirection just on target that supports aliasing.
8226 Fix GNU coding style.
8228 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
8231 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
8233 * config/arm/rtems.h: Likewise.
8234 * config/bfin/rtems.h: Likewise.
8235 * config/i386/rtemself.h: Likewise.
8236 * config/lm32/rtems.h: Likewise.
8237 * config/m32c/rtems.h: Likewise.
8238 * config/m68k/rtemself.h: Likewise.
8239 * config/microblaze/rtems.h: Likewise.
8240 * config/mips/rtems.h: Likewise.
8241 * config/moxie/rtems.h: Likewise.
8242 * config/nios2/rtems.h: Likewise.
8243 * config/powerpcspe/rtems.h: Likewise.
8244 * config/rs6000/rtems.h: Likewise.
8245 * config/rtems.h: Likewise.
8246 * config/sh/rtems.h: Likewise.
8247 * config/sh/rtemself.h: Likewise.
8248 * config/sparc/rtemself.h: Likewise.
8250 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8253 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
8254 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
8255 * tree-ssa-structalias.c (alias_get_name): Same.
8257 2017-07-25 Bin Cheng <bin.cheng@arm.com>
8260 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
8261 instructions if no du chain is found.
8263 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
8265 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
8267 2017-07-25 Richard Biener <rguenther@suse.de>
8270 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
8273 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
8275 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
8277 (ISA_2_7_MASKS_SERVER): Likewise.
8278 (ISA_3_0_MASKS_IEEE): Likewise.
8279 (OTHER_P8_VECTOR_MASKS): Likewise.
8280 (OTHER_VSX_VECTOR_MASKS): Likewise.
8281 (POWERPC_MASKS): Likewise.
8282 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
8283 duplicate list of options.
8284 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
8285 explicit -mupper-regs options.
8286 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
8287 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
8288 alias for -mupper-regs-df.
8289 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8290 (rs6000_init_hard_regno_mode_ok): Likewise.
8291 (rs6000_option_override_internal): Likewise.
8292 (rs6000_opt_masks): Likewise.
8293 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
8294 options in terms of whether -mvsx or -mpower8-vector was used.
8295 (TARGET_UPPER_REGS_DI): Likewise.
8296 (TARGET_UPPER_REGS_SF): Likewise.
8297 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
8298 -mupper-regs-* options.
8300 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8302 * passes.c (emergency_dump_function): Print some empty lines and a
8303 header before the RTL dump.
8305 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
8307 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
8309 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
8312 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8313 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
8315 2017-07-24 Carl Love <cel@us.ibm.com>
8317 * config/rs6000/rs6000-c.c: Add support for built-in functions
8318 vector float vec_extract_fp32_from_shorth (vector unsigned short);
8319 vector float vec_extract_fp32_from_shortl (vector unsigned short);
8320 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
8321 vec_extract_fp_from_shortl): Add defines for the two builtins.
8322 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
8323 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
8325 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
8326 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
8327 * doc/extend.texi: Update the built-in documentation file for the
8328 new built-in function.
8330 2017-07-24 Jakub Jelinek <jakub@redhat.com>
8333 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
8335 * doc/generic.texi: Likewise.
8336 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
8337 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
8339 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
8341 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
8342 (aarch64_mls_elt_merge<mode>): Likewise.
8344 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
8346 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8347 having __cxa_atexit.
8349 2017-07-23 Michael Collison <michael.collison@arm.com>
8351 * config/arm/arm.c (arm_option_override): Deprecate
8352 use of -mstructure-size-boundary.
8353 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
8354 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
8356 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8359 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
8360 Reduce cost estimate for direct moves.
8362 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
8365 * config/i386/i386.c (ix86_option_override_internal): Disable
8366 BMI, BMI2 and TBM instructions for -m16.
8368 2017-07-21 Carl Love <cel@us.ibm.com>
8370 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8371 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8372 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8373 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8374 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8375 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8376 VMULOSW): New enum "unspec" values.
8377 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8378 altivec_vmulosw): New patterns.
8379 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8380 VMULOSW): Add definitions.
8382 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
8384 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
8385 (qdf24xx): Likewise.
8386 * config/aarch64/aarch64-options-extensions.def (rdma); New.
8387 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
8388 (AARCH64_FL_V8_1): Renumber.
8389 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
8390 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
8391 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
8392 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
8393 rdma to feature modifiers list.
8395 2017-07-21 Yury Gribov <tetra2005@gmail.com>
8398 * ipa-visibility (function_and_variable_visibility): Convert
8399 recursive PLT call to direct call if appropriate.
8401 2017-07-21 Andrew Pinski <apinski@cavium.com>
8403 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
8404 operand 1 to see if the types precision matches.
8405 * fold-const.c (operand_equal_p): Likewise.
8407 2017-07-21 Richard Biener <rguenther@suse.de>
8409 PR tree-optimization/81303
8410 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
8411 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
8412 (vect_peeling_hash_get_lowest_cost): Adjust.
8413 (vect_enhance_data_refs_alignment): Likewise. Use
8414 vect_get_peeling_costs_all_drs to compute the penalty for no
8415 peeling to match up costs.
8417 2017-07-21 Richard Biener <rguenther@suse.de>
8419 PR tree-optimization/81500
8420 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
8421 we didn't identify a reduction path.
8423 2017-07-21 Tom de Vries <tom@codesourcery.com>
8424 Cesar Philippidis <cesar@codesourcery.com>
8426 PR gcov-profile/81442
8427 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
8430 2017-07-21 Tom de Vries <tom@codesourcery.com>
8433 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
8435 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
8436 nvptx_override_options_after_change.
8438 2017-07-21 Ulrich Drepper <drepper@redhat.com>
8440 * dwarf2out.c (output_file_names): Avoid double testing for
8443 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
8445 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
8447 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8449 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
8451 (try_crossjump_to_edge): Do not punt on partitioned functions.
8453 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
8455 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
8456 Put all BBs reachable only via paths crossing cold region to cold
8458 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
8460 2016-07-21 Richard Biener <rguenther@suse.de>
8462 PR tree-optimization/81303
8463 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
8464 into account prologue and epilogue iterations when raising
8465 min_profitable_iters to sth at least covering one vector iteration.
8467 2017-07-21 Tamar Christina <tamar.christina@arm.com>
8469 * config/arm/arm.c (arm_test_cpu_arch_dat):
8472 2017-07-20 Nathan Sidwell <nathan@acm.org>
8474 Remove TYPE_METHODS.
8475 * tree.h (TYPE_METHODS): Delete.
8476 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
8477 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
8478 (dbxout_type_methods): Scan TYPE_FIELDS.
8479 (dbxout_type): Don't check TYPE_METHODS here.
8480 * function.c (use_register_for_decl): Always ignore register for
8481 class types when not optimizing.
8482 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
8483 * tree.c (free_lang_data_in_type): Stitch out member functions and
8484 templates from TYPE_FIELDS.
8485 (build_distinct_type_copy, verify_type_variant,
8486 verify_type): Member fns are on TYPE_FIELDS.
8487 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
8488 * tree-pretty-print.c (dump_generic_node): Likewise.
8490 2017-07-20 Jakub Jelinek <jakub@redhat.com>
8493 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
8494 V2TImode and V4TImode.
8495 (ix86_expand_vector_extract): Likewise.
8496 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
8497 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
8498 (ssescalarmode): Handle V4TImode and V2TImode.
8499 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
8500 (*vec_extractv2ti, *vec_extractv4ti): New insns.
8501 (VEXTRACTI128_MODE): New mode iterator.
8502 (splitter for *vec_extractv?ti first element): New.
8503 (VEC_INIT_MODE): New mode iterator.
8504 (vec_init<mode>): Consolidate 3 expanders into one using
8505 VEC_INIT_MODE mode iterator.
8507 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8509 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
8510 non_spilled_static_chain_regno_p.
8512 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
8514 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
8516 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
8518 * bb-reorder.c (connect_traces): Allow copying of blocks within
8521 2017-07-20 Richard Biener <rguenther@suse.de>
8523 * gimple.h (gimple_phi_result): Add gphi * overload.
8524 (gimple_phi_result_ptr): Likewise.
8525 (gimple_phi_arg): Likewise. Adjust index assert to only
8526 allow actual argument accesses rather than all slots available
8528 (gimple_phi_arg_def): Add gphi * overload.
8529 * tree-phinodes.c (make_phi_node): Initialize only actual
8531 (resize_phi_node): Clear memory not covered by old node,
8532 do not initialize excess argument slots.
8533 (reserve_phi_args_for_new_edge): Initialize new argument slot
8536 2017-07-20 Bin Cheng <bin.cheng@arm.com>
8538 PR tree-optimization/81388
8540 2016-07-21 Bin Cheng <bin.cheng@arm.com>
8542 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
8543 by removing computation of may_be_zero.
8545 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8546 Tom de Vries <tom@codesourcery.com>
8549 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
8550 when gimple level profile disagrees with what RTL expander did.
8552 2017-07-20 Richard Biener <rguenther@suse.de>
8554 PR tree-optimization/61171
8555 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
8556 (vect_analyze_stmt): Add slp instance parameter.
8557 (vectorizable_reduction): Likewise.
8558 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
8559 (vect_is_simple_reduction): Deal with chains not detected
8560 as SLP reduction chain, specifically not properly associated
8561 chains containing a mix of plus/minus.
8562 (get_reduction_op): Remove.
8563 (get_initial_defs_for_reduction): Simplify, pass in whether
8564 this is a reduction chain, pass in the SLP node for the PHIs.
8565 (vect_create_epilog_for_reduction): Get the SLP instance as
8567 (vectorizable_reduction): Get the SLP instance as arg.
8568 During analysis remember the SLP node with the PHIs in the
8569 instance. Simplify getting at the vectorized reduction PHIs.
8570 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
8571 through SLP instance.
8572 (vect_slp_analyze_operations): Likewise.
8573 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
8574 (vect_transform_stmt): Likewise.
8576 2017-07-20 Tom de Vries <tom@codesourcery.com>
8578 PR tree-optimization/81489
8579 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
8580 read of phi arg location to before loop that modifies phi.
8582 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8584 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
8587 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8590 * except.c (execute): Fix ordering issue.
8592 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8594 PR rtl-optimization/81423
8595 * combine.c (make_compound_operation_int): Don't try to optimize
8596 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
8598 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
8600 PR rtl-optimization/81423
8601 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
8602 with a constant that is -1 in the truncated to mode.
8604 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8606 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
8607 (determine_unlikely_bbs): ... here.
8608 * predict.h (propagate_unlikely_bbs_forward): Declare.
8609 * cfgexpand.c (pass_expand::execute): Use it.
8610 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
8612 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
8613 propagate_unlikely_bbs_forward.
8615 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
8618 * except.c (maybe_add_nop_after_section_switch): New function.
8621 2017-07-19 Tom de Vries <tom@codesourcery.com>
8623 * gimple.h (gimple_phi_set_arg): Make assert more strict.
8625 2017-07-19 Tom de Vries <tom@codesourcery.com>
8627 * gimple.h (gimple_phi_arg): Make assert more strict.
8629 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
8631 * config.gcc (powerpc*-*-*): Add mmintrin.h.
8632 * config/rs6000/mmintrin.h: New file.
8633 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
8635 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8637 PR tree-optimization/81346
8638 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
8640 2017-07-19 Tom de Vries <tom@codesourcery.com>
8642 * config/nvptx/nvptx.md (VECIM): Add V2DI.
8644 2017-07-19 Tom de Vries <tom@codesourcery.com>
8646 * config/nvptx/nvptx-modes.def: Add V2DImode.
8647 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
8648 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
8649 (nvptx_output_mov_insn): Handle lack of mov.b128.
8650 (nvptx_print_operand): Handle 'H' and 'L' codes.
8651 (nvptx_vector_mode_supported): Allow V2DImode.
8652 (nvptx_preferred_simd_mode): New function.
8653 (nvptx_data_alignment): New function.
8654 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
8655 nvptx_preferred_simd_mode.
8656 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
8658 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
8660 2017-07-19 Tom de Vries <tom@codesourcery.com>
8662 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
8663 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
8664 (nvptx_vector_mode_supported): New function. Allow V2SImode.
8665 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
8666 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
8667 (mov<VECIM>_insn): New define_insn.
8668 (define_expand "mov<VECIM>): New define_expand.
8670 2017-07-19 Tom de Vries <tom@codesourcery.com>
8672 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
8674 2017-07-19 Jakub Jelinek <jakub@redhat.com>
8676 PR tree-optimization/81346
8677 * fold-const.h (fold_div_compare, range_check_type): Declare.
8678 * fold-const.c (range_check_type): New function.
8679 (build_range_check): Use range_check_type.
8680 (fold_div_compare): No longer static, rewritten into
8681 a match.pd helper function.
8682 (fold_comparison): Don't call fold_div_compare here.
8683 * match.pd (X / C1 op C2): New optimization using fold_div_compare
8686 2017-07-19 Nathan Sidwell <nathan@acm.org>
8688 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
8689 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
8690 * tree.c (find_decls_types_r, verify_type): Use
8691 TYPE_{MIN,MAX}_VALUE_RAW.
8692 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
8693 (hash_tree): Likewise.
8694 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8696 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8699 2017-07-18 Tom de Vries <tom@codesourcery.com>
8702 * omp-expand.c (expand_omp_for_static_chunk): Handle
8703 equal-argument loop exit phi.
8705 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
8708 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
8709 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
8710 operand 2 predicate.
8711 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
8712 operand 2 predicate.
8713 (ror,rol -> rorx splitters): Use const_int_operand as
8714 operand 2 predicate.
8716 2017-06-18 Richard Biener <rguenther@suse.de>
8718 PR tree-optimization/81410
8719 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
8720 the gap in the ! slp_perm SLP case after each group.
8722 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8725 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
8728 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
8731 * predict.c (set_even_probabilities): Cleanup; do not affect
8732 probabilities that are already known.
8733 (combine_predictions_for_bb): Call even when count is set.
8735 2017-07-18 Nathan Sidwell <nathan@acm.org>
8737 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
8740 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8743 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
8744 optimization for loop niter analysis.
8746 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
8749 * config/avr/avr.c (avr_optimize_casesi): Don't use
8750 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
8752 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8754 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
8755 body_cost_vec from _vect_peel_extended_info.
8756 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
8757 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
8760 2017-07-18 Bin Cheng <bin.cheng@arm.com>
8762 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
8764 2017-07-18 Richard Biener <rguenther@suse.de>
8766 PR tree-optimization/80620
8767 PR tree-optimization/81403
8768 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
8769 info when re-using a VN table entry.
8771 2017-07-18 Richard Biener <rguenther@suse.de>
8773 PR tree-optimization/81418
8774 * tree-vect-loop.c (vectorizable_reduction): Properly compute
8775 vectype_in. Verify that with lane-reducing reduction operations
8776 we have a single def-use cycle.
8778 2017-07-17 Carl Love <cel@us.ibm.com>
8780 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
8782 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8783 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8784 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8785 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8786 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8787 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8788 VMULOSW): New enum "unspec" values.
8789 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8790 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8791 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8792 altivec_vmulosw): New patterns.
8793 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8794 VMULOSW): Add definitions.
8795 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
8797 * config/alpha/alpha.c: Include predict.h.
8799 2017-07-17 Yury Gribov <tetra2005@gmail.com>
8801 * tree-vrp.c (compare_assert_loc): Fix comparison function
8802 to return predictable results.
8804 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8806 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
8809 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
8810 * doc/invoke.texi (mexpand-adddi): Update text.
8812 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8814 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
8815 that also clobbers the CC register. The old expand code is moved
8817 (*arc_clzsi2): ... here.
8818 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
8819 the CC register. The old expand code is moved to ...
8820 (arc_ctzsi2): ... here.
8822 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8824 * config/arc/arc.opt (mindexed-loads): Use initial value
8825 TARGET_INDEXED_LOADS_DEFAULT.
8826 (mauto-modify-reg): Use initial value
8827 TARGET_AUTO_MODIFY_REG_DEFAULT.
8828 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8829 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8830 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
8831 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
8833 2017-07-17 Martin Liska <mliska@suse.cz>
8836 * opts.c (finish_options): Do not allow -fgnu-tm
8837 w/ -fsanitize={kernel-,}address. Say sorry.
8839 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8842 * tree-loop-distribution.c (classify_partition): Only assert on
8843 numer of iterations.
8844 (merge_dep_scc_partitions): Delete prameter. Update function call.
8845 (distribute_loop): Remove code handling loop with unknown niters.
8846 (pass_loop_distribution::execute): Skip loop with unknown niters.
8848 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8851 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
8852 function sort_partitions_by_post_order.
8854 2017-07-17 Bin Cheng <bin.cheng@arm.com>
8856 PR tree-optimization/81374
8857 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
8858 the max index of basic blocks, rather than number of basic blocks.
8860 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8862 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
8864 (arc_legitimate_pic_operand_p): Likewise.
8865 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
8867 (arc_needs_pcl_p): Likewise.
8868 (arc_legitimate_pc_offset_p): Likewise.
8869 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
8870 function is also used in constrains.md.
8871 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
8872 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
8873 PLUS. Only return true/false in known cases, otherwise assert.
8874 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
8875 is already called in arc_legitimate_constant_p.
8876 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
8878 (LEGITIMATE_PIC_OPERAND_P): Use
8879 arc_raw_symbolic_reference_mentioned_p function.
8880 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
8885 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8886 Andrew Burgess <andrew.burgess@embecosm.com>
8888 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
8889 (arc_return_address_register): New function.
8890 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
8891 (arc_handle_fndecl_attribute): Add naked attribute.
8892 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
8893 (TARGET_WARN_FUNC_RETURN): Likewise.
8894 (arc_allocate_stack_slots_for_args): New function.
8895 (arc_warn_func_return): Likewise.
8896 (machine_function): Change type fn_type.
8897 (arc_compute_function_type): Consider new naked function type,
8898 change function return type.
8899 (arc_must_save_register): Adapt to handle new
8900 arc_compute_function_type's return type.
8901 (arc_expand_prologue): Likewise.
8902 (arc_expand_epilogue): Likewise.
8903 (arc_return_address_regs): Delete.
8904 (arc_return_address_register): New function.
8905 (arc_epilogue_uses): Use above function.
8906 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
8907 (arc_function_type): Change encoding, add naked type.
8908 (ARC_INTERRUPT_P): Change to handle the new encoding.
8909 (ARC_FAST_INTERRUPT_P): Likewise.
8910 (ARC_NORMAL_P): Define.
8911 (ARC_NAKED_P): Likewise.
8912 (arc_compute_function_type): Delete prototype.
8913 * config/arc/arc.md (in_ret_delay_slot): Use
8914 arc_return_address_register function.
8915 (simple_return): Likewise.
8916 (p_return_i): Likewise.
8918 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8920 PR tree-optimization/81428
8921 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
8922 can't be built for those types.
8924 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8926 Remove stuff dead since r239246.
8928 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
8929 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
8930 (avr_inform_devices): Remove dead stuff.
8932 2017-07-17 Tamar Christina <tamar.christina@arm.com>
8934 * config/arm/arm_neon.h: Fix softp typo.
8936 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8938 PR tree-optimization/81365
8939 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
8940 aggregate moves onto bb predecessor edges, make sure there are no
8941 loads that could alias the lhs in between the start of bb and the
8944 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
8947 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
8948 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
8949 [LSHIFTRT, outer_code = TRUNCATE]: Same.
8951 2017-07-17 Jakub Jelinek <jakub@redhat.com>
8953 PR tree-optimization/81396
8954 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
8955 (init_symbolic_number): Initialize it to 1.
8956 (perform_symbolic_merge): Add n_ops from both operands into the new
8958 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
8959 without base_addr as useless if they need more than one operation.
8960 (bswap_replace): Handle !bswap case for NULL base_addr.
8962 2017-07-17 Tom de Vries <tom@codesourcery.com>
8965 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
8968 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
8970 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
8971 conditional builtin define __FIX_LEON3FT_B2BST.
8973 2017-07-17 Daniel Cederman <cederman@gaisler.com>
8975 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
8976 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
8979 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8981 PR rtl-optimization/81424
8982 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
8983 to remove potential trapping from operands if -fnon-call-exceptions.
8985 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8987 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
8988 profile_proability for scalling.
8989 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
8991 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8993 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
8995 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
8997 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
8998 fixpoint arithmetics.
9000 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9002 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
9003 fixpoint arithmetics.
9005 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9007 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
9008 fixpoint arithmetics.
9010 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
9012 * profile-count.h (profile_probability::from_reg_br_prob_note,
9013 profile_probability::to_reg_br_prob_note): New functions.
9014 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
9015 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
9016 * predict.c (probability_reliable_p): Update.
9017 (edge_probability_reliable_p): Update.
9018 (br_prob_note_reliable_p): Update.
9019 (invert_br_probabilities): Update.
9020 (add_reg_br_prob_note): New function.
9021 (combine_predictions_for_insn): Update.
9022 * asan.c (asan_clear_shadow): Update.
9023 * cfgbuild.c (compute_outgoing_frequencies): Update.
9024 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
9025 (update_br_prob_note): Update.
9026 (rtl_verify_edges): Update.
9027 (purge_dead_edges): Update.
9028 (fixup_reorder_chain): Update.
9029 * emit-rtl.c (try_split): Update.
9030 * ifcvt.c (cond_exec_process_insns): Update.
9031 (cond_exec_process_if_block): Update.
9032 (dead_or_predicable): Update.
9033 * internal-fn.c (expand_addsub_overflow): Update.
9034 (expand_neg_overflow): Update.
9035 (expand_mul_overflow): Update.
9036 * loop-doloop.c (doloop_modify): Update.
9037 * loop-unroll.c (compare_and_jump_seq): Update.
9038 * optabs.c (emit_cmp_and_jump_insn_1): Update.
9039 * predict.h: Update.
9040 * reorg.c (mostly_true_jump): Update.
9042 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
9043 * config/alpha/alpha.c (emit_unlikely_jump): Update.
9044 * config/arc/arc.c: (emit_unlikely_jump): Update.
9045 * config/arm/arm.c: (emit_unlikely_jump): Update.
9046 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
9047 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
9048 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
9049 (ix86_print_operand): Update.
9050 (ix86_split_fp_branch): Update.
9051 (predict_jump): Update.
9052 * config/ia64/ia64.c (ia64_print_operand): Update.
9053 * config/mmix/mmix.c (mmix_print_operand): Update.
9054 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
9055 (rs6000_expand_split_stack_prologue): Update.
9056 * config/rs6000/rs6000.c: Update.
9057 * config/s390/s390.c (s390_expand_vec_strlen): Update.
9058 (s390_expand_vec_movstr): Update.
9059 (s390_expand_cs_tdsi): Update.
9060 (s390_expand_split_stack_prologue): Update.
9061 * config/sh/sh.c (sh_print_operand): Update.
9062 (expand_cbranchsi4): Update.
9063 (expand_cbranchdi4): Update.
9064 * config/sparc/sparc.c (output_v9branch): Update.
9065 * config/spu/spu.c (get_branch_target): Update.
9066 (ea_load_store_inline): Update.
9067 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
9068 * config/tilepro/tilepro.c: Update.
9070 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
9072 * gimplify.c (mostly_copy_tree_r): Revert latest change.
9073 (gimplify_save_expr): Likewise.
9075 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9077 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
9079 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9081 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
9083 * timevar.def (TV_IPA_FNSUMMARY): Define.
9085 2017-07-16 Daniel Cederman <cederman@gaisler.com>
9087 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
9088 to back store errata sensitive sequence from being generated.
9089 (sqrtdf2_fix): Likewise.
9091 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9093 * tree-ssa-threadupdate.c (compute_path_counts,
9094 update_joiner_offpath_counts): Use profile_probability.
9096 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
9099 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9101 * config/arm/arm-c.c (arm_cpu_builtins): Define
9102 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9104 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
9106 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9107 array entries to represent __ieee128 versions of the
9108 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
9109 scalar_extract_sig, and scalar_insert_exp built-in functions.
9110 (altivec_resolve_overloaded_builtin): Add special case handling
9111 for the __builtin_scalar_insert_exp function, as represented by
9112 the P9V_BUILTIN_VEC_VSIEDP constant.
9113 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
9114 exponent support for __ieee128 argument.
9115 (VSESQP): Add scalar extract signature support for __ieee128
9117 (VSTDCNQP): Add scalar test negative support for __ieee128
9119 (VSIEQP): Add scalar insert exponent support for __int128 argument
9120 with __ieee128 result.
9121 (VSIEQPF): Add scalar insert exponent support for __ieee128
9122 argument with __ieee128 result.
9123 (VSTDCQP): Add scalar test data class support for __ieee128
9125 (VSTDCNQP): Add overload support for scalar test negative with
9127 (VSTDCQP): Add overload support for scalar test data class
9129 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
9131 (UNSPEC_VSX_SIEXPQP): New constant.
9132 (xsxexpqp): New insn for VSX scalar extract exponent quad
9134 (xsxsigqp): New insn for VSX scalar extract significand quad
9136 (xsiexpqpf): New insn for VSX scalar insert exponent quad
9137 precision with floating point argument.
9138 (xststdcqp): New expand for VSX scalar test data class quad
9140 (xststdcnegqp): New expand for VSX scalar test negative quad
9142 (xststdcqp): New insn to match expansions for VSX scalar test data
9143 class quad precision and VSX scalar test negative quad precision.
9144 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
9145 special case operand checking to enforce that second operand of
9146 VSX scalar test data class with quad precision argument is a 7-bit
9148 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9149 prototypes and descriptions of __ieee128 versions of
9150 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
9151 scalar_test_data_class, and scalar_test_neg built-in functions.
9153 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9155 PR tree-optimization/81162
9156 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
9157 replace a negate with an add.
9159 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
9161 * doc/invoke.texi (arm/-mcpu): Document +crypto.
9163 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9165 * config/arm/arm-c.c (arm_cpu_builtins): Define
9166 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
9168 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9170 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
9171 (armv8-r): Set ARM Cortex-R52 as default CPU.
9172 * config/arm/arm-tables.opt: Regenerate.
9173 * config/arm/arm-tune.md: Regenerate.
9174 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9176 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
9177 extension for -mcpu=cortex-r52.
9179 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
9181 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
9182 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
9183 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
9184 (fp-armv8): Define it as FP_ARMv8 only.
9185 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
9186 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
9188 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
9189 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
9190 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
9191 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
9192 than TARGET_FPU_ARMV8.
9193 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
9194 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
9195 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
9197 * config/arm/neon.md (neon_vrint): Likewise.
9198 (neon_vcvt): Likewise.
9199 (neon_<fmaxmin_op><mode>): Likewise.
9200 (<fmaxmin><mode>3): Likewise.
9201 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
9202 * config/arm/predicates.md (arm_cond_move_operator): Check against
9203 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
9205 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
9207 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
9210 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9212 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
9213 loop in comment with memset.
9215 2017-07-14 Martin Liska <mliska@suse.cz>
9217 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
9218 * dwarf2out.c (is_java): Remove the function.
9219 (output_pubname): Remove usage of the function.
9220 (lower_bound_default): Remove usage of DW_LANG_Java.
9221 (gen_compile_unit_die): Likewise.
9222 * gcc.c: Remove compiler defaults for .java and .zip files.
9223 * gimple-expr.c (remove_suffix): Change as there's no longer
9224 extension than 4-letter one.
9225 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
9226 (gimplify_save_expr): Likewise.
9227 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
9228 as it's possible even for other languages than Java.
9229 * langhooks.h (struct lang_hooks): Remove Java from a comment.
9230 * lto-opts.c (lto_write_options): Remove reference to Java.
9231 * opts.c (strip_off_ending): Update file extension handling.
9232 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
9233 * tree-eh.c (lower_resx): Likewise.
9234 * tree.c (free_lang_data_in_type): Remove dead code.
9235 (find_decls_types_r): Likewise.
9236 (build_common_builtin_nodes): Remove Java from a comment.
9237 (verify_type): Remove dead code.
9238 * varasm.c (assemble_external): Remove Java from a comment.
9240 2017-07-14 Martin Liska <mliska@suse.cz>
9242 * opts.c (finish_options): Add quotes.
9243 (common_handle_option): Likewise.
9245 2017-07-14 Martin Liska <mliska@suse.cz>
9247 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
9248 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
9250 * dwarf2out.c (lower_bound_default): Do not handle
9252 (gen_compile_unit_die): Likewise.
9253 * gcc.c: Remove default extension binding for GNU Pascal.
9254 * stmt.c: Remove Pascal language from a comment.
9255 * xcoffout.c: Likewise.
9257 2017-07-13 David Malcolm <dmalcolm@redhat.com>
9260 * diagnostic-show-locus.c (fixit_cmp): New function.
9261 (layout::layout): Sort m_fixit_hints.
9262 (column_range::column_range): Assert that the values are valid.
9263 (struct char_span): New struct.
9264 (correction::overwrite): New method.
9265 (struct source_line): New struct.
9266 (line_corrections::add_hint): Add assertions. Reimplement memcpy
9267 calls in terms of classes source_line and char_span, and
9268 correction::overwrite.
9269 (selftest::test_overlapped_fixit_printing_2): New function.
9270 (selftest::diagnostic_show_locus_c_tests): Call it.
9272 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9274 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
9275 early if there is no lhs.
9277 2017-07-13 Martin Liska <mliska@suse.cz>
9279 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
9280 (gen_reference_type_die): Likewise.
9281 * stor-layout.c: Remove Pascal-related comment.
9283 2017-07-13 Martin Liska <mliska@suse.cz>
9285 * opts.c (finish_options): Add quotes to error messages.
9286 (parse_sanitizer_options): Likewise.
9288 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
9290 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
9292 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
9294 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
9296 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
9298 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
9300 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
9302 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9305 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
9306 provides the hardware capability bits, define the macro
9307 __BUILTIN_CPU_SUPPORTS__.
9308 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
9309 if GLIBC does not provide the hardware capability bits. Add a
9310 gcc_unreachable call if the built-in cpu function is neither
9311 __builtin_cpu_is nor __builtin_cpu_supports.
9312 (rs6000_get_function_versions_dispatcher): Change the warning
9313 that an old GLIBC is used which does not export the capability
9314 bits to be an error.
9315 * doc/extend.texi (target_clones attribute): Document the
9316 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
9317 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
9318 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
9319 the macros defined by GCC if the newer GLIBC is available.
9321 2017-07-12 Jeff Law <law@redhat.com>
9323 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
9324 remaining includes slightly.
9325 * config/riscv/riscv-builtins.c: Include profile-count.h.
9327 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9330 * config/avr/avr.c (avr_set_current_function): In diagnostic
9331 messages: Quote keywords and (parts of) identifiers.
9332 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
9335 2017-07-12 Carl Love <cel@us.ibm.com>
9337 * config/rs6000/rs6000-c.c: Add support for built-in functions
9338 vector bool char vec_revb (vector bool char);
9339 vector bool short vec_revb (vector short char);
9340 vector bool int vec_revb (vector bool int);
9341 vector bool long long vec_revb (vector bool long long);
9342 * doc/extend.texi: Update the built-in documentation file for the
9343 new built-in functions.
9345 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9347 * config/s390/s390.md: Remove movcc splitter.
9349 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9351 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
9352 load/store on condition.
9354 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
9357 * config/avr/avr.c (avr_encode_section_info)
9358 [progmem && !TREE_READONLY]: Error if progmem object needs
9361 2017-07-11 Michael Collison <michael.collison@arm.com>
9363 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
9366 2017-07-11 Carl Love <cel@us.ibm.com>
9368 * config/rs6000/rs6000-c.c: Add support for builtins
9369 vector unsigned int vec_parity_lsbb (vector signed int);
9370 vector unsigned int vec_parity_lsbb (vector unsigned int);
9371 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
9372 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
9373 vector unsigned long long vec_parity_lsbb (vector signed long long);
9374 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
9375 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
9376 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
9377 * doc/extend.texi: Update the built-in documentation file for the
9378 new built-in functions.
9380 2017-07-11 David Malcolm <dmalcolm@redhat.com>
9382 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
9383 (layout::m_primary_loc): New field.
9384 (layout::layout): Initialize new field. Move location filtering
9385 logic from here to...
9386 (layout::maybe_add_location_range): ...this new method. Add
9387 support for filtering to just the lines already specified by other
9389 (layout::will_show_line_p): New method.
9390 (gcc_rich_location::add_location_if_nearby): New method.
9391 (selftest::test_add_location_if_nearby): New test function.
9392 (selftest::diagnostic_show_locus_c_tests): Call it.
9393 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
9396 2017-07-11 Tom de Vries <tom@codesourcery.com>
9398 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
9399 (bb_first_real_insn): New function.
9400 (nvptx_single): Add extra initialization of broadcasted condition
9403 2017-07-11 Nathan Sidwell <nathan@acm.org>
9405 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
9407 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
9409 * doc/extend.texi (AVR Function Attributes): Remove weblink to
9410 Binutils doc as TEXI will mess them up.
9411 * doc/invoke.texi (AVR Options): Same here.
9413 2017-07-11 Daniel Cederman <cederman@gaisler.com>
9415 * config/sparc/sparc.opt (mfix-ut700): New option.
9416 (mfix-gr712rc): Likewise.
9417 (sparc_fix_b2bst): New variable.
9418 * doc/invoke.texi (SPARC options): Document them.
9419 (ARM options): Fix warnings.
9420 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
9421 instructions to prevent sequences that can trigger the store-store
9422 errata for certain LEON3FT processors.
9423 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
9424 (sparc_option_override): Set sparc_fix_b2bst appropriately.
9425 * config/sparc/sparc.md (fix_b2bst): New attribute.
9426 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
9428 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
9431 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
9433 (*rsqrtsf2_sse): Ditto.
9435 (div<mode>3): Macroize insn from divdf3 and divsf3
9436 using MODEF mode iterator.
9438 2017-07-10 Martin Sebor <msebor@redhat.com>
9440 PR tree-optimization/80397
9441 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
9442 instead of testing for equality to INTEGER_TYPE.
9444 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
9446 * config.gcc: Remove uclibc from arc target spec.
9448 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
9450 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
9452 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
9455 * lto-wrapper.c (remove_option): New function.
9456 (merge_and_complain): Merge PIC/PIE options more realistically.
9458 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9460 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
9464 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
9465 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
9466 * config.in: Regenerate.
9467 * configure: Regenerate.
9468 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
9469 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
9470 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
9471 (TARGET_GASISR_PROLOGUES): ...target mask.
9472 * common/config/avr/avr-common.c
9473 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
9474 Set -mgas-isr-prologues.
9475 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
9476 INSERT_PASS_BEFORE for it.
9477 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
9478 * config/avr/avr.c (avr_option_override)
9479 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
9480 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
9481 (avr_attribute_table) <no_gccisr>: Add new function attribute.
9482 (avr_set_current_function) <is_no_gccisr>: Init machine field.
9483 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
9485 (make_avr_pass_pre_proep): New function.
9486 (emit_push_sfr) <treg>: Add argument to function and use it
9488 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
9489 and set machine->gasisr.yes.
9490 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
9491 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
9492 __gcc_isr.n_pushed to .L__stack_usage.
9493 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
9494 (avr_asm_final_postscan_insn): ...this new static function.
9495 * config/avr/avr.h (machine_function)
9496 <is_no_gccisr, use_L__stack_usage>: New fields.
9497 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
9498 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
9499 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
9500 (gasisr, *gasisr): New expander and insn.
9501 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9502 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
9503 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
9505 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
9507 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
9510 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
9512 Move jump-tables out of .text again.
9515 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
9516 (ASM_OUTPUT_ADDR_VEC): New function.
9517 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
9518 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
9519 INSN_ADDRESSes as asm comment.
9520 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
9521 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
9522 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
9523 * config/avr/avr.md (*tablejump): Adjust comment.
9524 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
9525 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
9527 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
9528 (avr_output_addr_vec): New proto.
9529 (avr_log_t) <insn_addresses>: New field.
9531 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
9534 * config/i386/i386.c (ix86_function_arg_advance): Set
9535 outgoing_args_on_stack to true if there are outgoing arguments
9537 (ix86_function_arg): Likewise.
9538 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
9539 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
9540 * config/i386/i386.h (machine_function): Add
9541 outgoing_args_on_stack.
9543 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
9545 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
9547 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
9549 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
9551 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
9552 (REAL_H): Remove $(MACHMODE_H).
9553 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
9555 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
9556 $(MACHMODE_H) and double-int.h.
9557 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
9559 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
9562 2017-07-07 Andrew Pinski <apinski@cavium.com>
9564 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
9565 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
9567 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9569 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
9570 Add warning if GCC was not configured to link against a GLIBC that
9571 exports the hardware capability bits.
9572 (make_resolver_func): Make resolver function private and not a
9573 COMDAT function. Create the name with clone_function_name instead
9574 of make_unique_name.
9577 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
9578 correct operand in doing the split.
9580 2017-07-07 Carl Love <cel@us.ibm.com>
9582 * config/rs6000/rs6000-c: Add support for built-in function
9583 vector unsigned short vec_pack_to_short_fp32 (vector float,
9585 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
9586 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
9587 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
9588 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
9589 (convert_4f32_8i16): Add define_expand.
9590 * doc/extend.texi: Update the built-in documentation file for the
9591 new built-in function.
9593 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9595 * config/sparc/m8.md: New file.
9596 * config/sparc/sparc.md: Include m8.md.
9598 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9600 * config/sparc/sparc.opt: New option -mvis4b.
9601 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
9602 (sparc_option_override): Handle VIS4B.
9603 (enum sparc_builtins): Define
9604 SPARC_BUILTIN_DICTUNPACK{8,16,32},
9605 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
9606 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
9607 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
9608 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
9609 (check_constant_argument): New function.
9610 (sparc_vis_init_builtins): Define builtins
9611 __builtin_vis_dictunpack{8,16,32},
9612 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
9613 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
9614 __builtin_vis_fpcmpde{8,16,32}shl and
9615 __builtin_vis_fpcmpur{8,16,32}shl.
9616 (sparc_expand_builtin): Check that the constant operands to
9617 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
9618 constant and in range.
9619 * config/sparc/sparc-c.c (sparc_target_macros): Handle
9621 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
9622 (SPARC_IMM5_P): Likewise.
9623 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
9624 (enabled): Handle vis4b.
9625 (UNSPEC_DICTUNPACK): New unspec.
9626 (UNSPEC_FPCMPSHL): Likewise.
9627 (UNSPEC_FPUCMPSHL): Likewise.
9628 (UNSPEC_FPCMPDESHL): Likewise.
9629 (UNSPEC_FPCMPURSHL): Likewise.
9630 (cpu_feature): New CPU feature `vis4b'.
9631 (dictunpack{8,16,32}): New insns.
9632 (FPCSMODE): New mode iterator.
9633 (fpcscond): New code iterator.
9634 (fpcsucond): Likewise.
9635 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
9636 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
9637 (fpcmpde{8,16,32}{si,di}shl): Likewise.
9638 (fpcmpur{8,16,32}{si,di}shl): Likewise.
9639 * config/sparc/constraints.md: Define constraints `q' for unsigned
9640 2-bit integer constants and `t' for unsigned 5-bit integer
9642 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
9644 (imm5_operand_dictunpack16): Likewise.
9645 (imm5_operand_dictunpack32): Likewise.
9646 (imm2_operand): Likewise.
9647 * doc/invoke.texi (SPARC Options): Document -mvis4b.
9648 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
9649 ditunpack* and fpcmp*shl builtins.
9651 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9653 * config.gcc: Handle m8 in --with-{cpu,tune} options.
9654 * config.in: Add HAVE_AS_SPARC6 define.
9655 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
9657 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
9659 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
9660 (CPP_CPU_SPEC): Handle m8.
9661 (ASM_CPU_SPEC): Likewise.
9662 * config/sparc/sparc-opts.h (enum processor_type): Add
9664 * config/sparc/sparc.c (m8_costs): New struct.
9665 (sparc_option_override): Handle TARGET_CPU_m8.
9666 (sparc32_initialize_trampoline): Likewise.
9667 (sparc64_initialize_trampoline): Likewise.
9668 (sparc_issue_rate): Likewise.
9669 (sparc_register_move_cost): Likewise.
9670 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
9671 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
9672 (ASM_CPU64_DEFAULT_SPEC): Likewise.
9673 (CPP_CPU_SPEC): Handle M8.
9674 (ASM_CPU_SPEC): Likewise.
9675 (AS_M8_FLAG): Define.
9676 * config/sparc/sparc.md: Add m8 to the cpu attribute.
9677 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
9678 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
9680 * configure: Regenerate.
9681 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
9684 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9686 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
9688 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
9689 ("*movdi_insn_sp32"): Do not set v3pipe.
9690 ("*movsi_insn"): Likewise.
9691 ("*movdi_insn_sp64"): Likewise.
9692 ("*movsf_insn"): Likewise.
9693 ("*movdf_insn_sp32"): Likewise.
9694 ("*movdf_insn_sp64"): Likewise.
9695 ("*zero_extendsidi2_insn_sp64"): Likewise.
9696 ("*sign_extendsidi2_insn"): Likewise.
9697 ("*mov<VM32:mode>_insn"): Likewise.
9698 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9699 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9700 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9701 ("<vlop:code><VL:mode>3"): Likewise.
9702 ("*not_<vlop:code><VL:mode>3"): Likewise.
9703 ("*nand<VL:mode>_vis"): Likewise.
9704 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
9705 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
9706 ("one_cmpl<VL:mode>2"): Likewise.
9707 ("faligndata<VM64:mode>_vis"): Likewise.
9708 ("alignaddrsi_vis"): Likewise.
9709 ("alignaddrdi_vis"): Likweise.
9710 ("alignaddrlsi_vis"): Likewise.
9711 ("alignaddrldi_vis"): Likewise.
9712 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9713 ("bmaskdi_vis"): Likewise.
9714 ("bmasksi_vis"): Likewise.
9715 ("bshuffle<VM64:mode>_vis"): Likewise.
9716 ("cmask8<P:mode>_vis"): Likewise.
9717 ("cmask16<P:mode>_vis"): Likewise.
9718 ("cmask32<P:mode>_vis"): Likewise.
9719 ("pdistn<P:mode>_vis"): Likewise.
9720 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9722 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9724 * config/sparc/sparc.md ("subtype"): New insn attribute.
9725 ("*wrgsr_sp64"): Set insn subtype.
9726 ("*rdgsr_sp64"): Likewise.
9727 ("alignaddrsi_vis"): Likewise.
9728 ("alignaddrdi_vis"): Likewise.
9729 ("alignaddrlsi_vis"): Likewise.
9730 ("alignaddrldi_vis"): Likewise.
9731 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
9732 ("fexpand_vis"): Likewise.
9733 ("fpmerge_vis"): Likewise.
9734 ("faligndata<VM64:mode>_vis"): Likewise.
9735 ("bshuffle<VM64:mode>_vis"): Likewise.
9736 ("cmask8<P:mode>_vis"): Likewise.
9737 ("cmask16<P:mode>_vis"): Likewise.
9738 ("cmask32<P:mode>_vis"): Likewise.
9739 ("fchksm16_vis"): Likewise.
9740 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
9741 ("fmean16_vis"): Likewise.
9742 ("fp<plusminus_insn>64_vis"): Likewise.
9743 ("<plusminus_insn>v8qi3"): Likewise.
9744 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
9745 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
9746 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
9747 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
9748 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
9749 ("*movqi_insn"): Likewise.
9750 ("*movhi_insn"): Likewise.
9751 ("*movsi_insn"): Likewise.
9752 ("movsi_pic_gotdata_op"): Likewise.
9753 ("*movdi_insn_sp32"): Likewise.
9754 ("*movdi_insn_sp64"): Likewise.
9755 ("movdi_pic_gotdata_op"): Likewise.
9756 ("*movsf_insn"): Likewise.
9757 ("*movdf_insn_sp32"): Likewise.
9758 ("*movdf_insn_sp64"): Likewise.
9759 ("*zero_extendhisi2_insn"): Likewise.
9760 ("*zero_extendqihi2_insn"): Likewise.
9761 ("*zero_extendqisi2_insn"): Likewise.
9762 ("*zero_extendqidi2_insn"): Likewise.
9763 ("*zero_extendhidi2_insn"): Likewise.
9764 ("*zero_extendsidi2_insn_sp64"): Likewise.
9765 ("ldfsr"): Likewise.
9766 ("prefetch_64"): Likewise.
9767 ("prefetch_32"): Likewise.
9768 ("tie_ld32"): Likewise.
9769 ("tie_ld64"): Likewise.
9770 ("*tldo_ldub_sp32"): Likewise.
9771 ("*tldo_ldub1_sp32"): Likewise.
9772 ("*tldo_ldub2_sp32"): Likewise.
9773 ("*tldo_ldub_sp64"): Likewise.
9774 ("*tldo_ldub1_sp64"): Likewise.
9775 ("*tldo_ldub2_sp64"): Likewise.
9776 ("*tldo_ldub3_sp64"): Likewise.
9777 ("*tldo_lduh_sp32"): Likewise.
9778 ("*tldo_lduh1_sp32"): Likewise.
9779 ("*tldo_lduh_sp64"): Likewise.
9780 ("*tldo_lduh1_sp64"): Likewise.
9781 ("*tldo_lduh2_sp64"): Likewise.
9782 ("*tldo_lduw_sp32"): Likewise.
9783 ("*tldo_lduw_sp64"): Likewise.
9784 ("*tldo_lduw1_sp64"): Likewise.
9785 ("*tldo_ldx_sp64"): Likewise.
9786 ("*mov<VM32:mode>_insn"): Likewise.
9787 ("*mov<VM64:mode>_insn_sp64"): Likewise.
9788 ("*mov<VM64:mode>_insn_sp32"): Likewise.
9790 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9792 * config/sparc/sparc.md ("type"): New insn type viscmp.
9793 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
9795 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
9796 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
9797 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
9798 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
9800 ("n7_vis_logical_11cycle"): Likewise.
9801 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
9802 * config/sparc/niagara2.md ("niag3_vis": Likewise.
9803 * config/sparc/niagara.md ("niag_vis"): Likewise.
9804 * config/sparc/ultra3.md ("us3_fga"): Likewise.
9805 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
9807 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
9809 * config/sparc/sparc.md: New instruction type `bmask'.
9810 (bmaskdi_vis): Use the `bmask' type.
9811 (bmasksi_vis): Likewise.
9812 * config/sparc/ultra3.md (us3_array): Likewise.
9813 * config/sparc/niagara7.md (n7_array): Likewise.
9814 * config/sparc/niagara4.md (n4_array): Likewise.
9815 * config/sparc/niagara2.md (niag2_vis): Likewise.
9816 (niag3_vis): Likewise.
9817 * config/sparc/niagara.md (niag_vis): Likewise.
9819 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9821 * ipa-comdats.c: Remove optimize check from gate.
9822 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
9823 for functions not optimized.
9824 (ipa_fn_summary_read): Skip optimize check.
9825 (ipa_fn_summary_write): Likewise.
9826 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
9828 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
9830 (can_inline_edge_p): Check flag_pcc_struct_return for match.
9831 (check_callers): Give up on caller which is not optimized.
9832 (inline_small_functions): Likewise.
9833 (ipa_inline): Do not give up when not optimizing.
9834 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
9835 away unoptimizes cdtors.
9836 (whole_program_function_and_variable_visibility): Do
9837 ipa_discover_readonly_nonaddressable_vars in LTO mode.
9838 * ipa.c (process_references): Do not check optimize.
9839 (symbol_table::remove_unreachable_nodes): Update optimize check.
9840 (set_writeonly_bit): Update optimize check.
9841 (pass_ipa_cdtor_merge::gate): Do not check optimize.
9842 (pass_ipa_single_use::gate): Remove.
9844 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9846 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
9847 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
9848 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
9849 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
9850 permute_load, permute_store, adjust_extract, adjust_splat,
9851 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
9852 replace_swap_with_copy, dump_swap_insn_table,
9853 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
9854 recombine_lvx_pattern, recombine_stvx_pattern,
9855 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
9856 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
9857 to file rs6000-p8swap.c.
9858 * config/rs6000/rs6000-p8swap.c: New file.
9859 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
9860 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
9861 and rs6000*-*-* targets.
9863 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9865 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
9867 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9869 * lto-wrapper.c (merge_and_complain): Do not merge
9870 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
9871 fsigned_zeros, ftrapping_math, fwrapv.
9872 (append_compiler_options): Do not track these options.
9873 (append_linker_options): Likewie
9875 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9877 * cgraphunit.c (cgraph_node::finalize_function): When
9878 !flag_toplevel_reorde set no_reorder flag.
9879 (varpool_node::finalize_decl): Likewise.
9880 (symbol_table::compile): Drop no toplevel reorder path.
9882 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9884 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
9885 edges; zero probability is not better than uninitialized.
9887 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9889 * asan.h (asan_sanitize_allocas_p): Declare.
9890 * asan.c (asan_sanitize_allocas_p): New function.
9891 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
9892 (handle_builtin_alloca): Likewise.
9893 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
9894 if !asan_sanitize_allocas_p.
9895 * params.def (asan-instrument-allocas): Add new option.
9896 * params.h (ASAN_PROTECT_ALLOCAS): Define.
9897 * opts.c (common_handle_option): Disable allocas sanitization for
9900 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
9902 * asan.c: Include gimple-fold.h.
9903 (get_last_alloca_addr): New function.
9904 (handle_builtin_stackrestore): Likewise.
9905 (handle_builtin_alloca): Likewise.
9906 (asan_emit_allocas_unpoison): Likewise.
9907 (get_mem_refs_of_builtin_call): Add new parameter, remove const
9908 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
9909 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
9910 (instrument_builtin_call): Pass gimple iterator to
9911 get_mem_refs_of_builtin_call.
9912 (last_alloca_addr): New global.
9913 * asan.h (asan_emit_allocas_unpoison): Declare.
9914 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
9915 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
9916 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
9917 if function calls alloca.
9918 * gimple-fold.c (replace_call_with_value): Remove static keyword.
9919 * gimple-fold.h (replace_call_with_value): Declare.
9920 * internal-fn.c: Include asan.h.
9921 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
9922 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
9924 2017-07-06 David Malcolm <dmalcolm@redhat.com>
9926 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
9927 (C_SELFTEST_FLAGS): New.
9928 (CPP_SELFTEST_FLAGS): New.
9929 (SELFTEST_DEPS): New, from deps of s-selftest.
9930 (C_SELFTEST_DEPS): New, from deps of s-selftest.
9931 (CPP_SELFTEST_DEPS): New.
9932 (selftest): Add dependency on s-selftest-c++.
9933 (s-selftest): Rename to...
9934 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
9935 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
9936 than SELFTEST_FLAGS.
9937 (selftest-gdb): Rename to...
9938 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
9940 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
9941 (selftest-valgrind): Rename to...
9942 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
9944 (selftest-valgrind): Reintroduce as an alias for
9945 selftest-c-valgrind.
9946 (s-selftest-c++): New.
9947 (selftest-c++-gdb): New.
9948 (selftest-c++-valgrind): New.
9950 2017-07-06 Olivier Hainque <hainque@adacore.com>
9952 * gcc.c (process_command): When deciding if undefined variables
9953 should be ignored when processing specs, accept "gcc -v" as well.
9955 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
9957 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
9958 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
9960 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
9962 * config/arm/arm-cpus.in (armv8-r): Add new entry.
9963 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
9964 * config/arm/arm-tables.opt: Regenerate.
9965 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
9967 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
9969 2017-07-06 Carl Love <cel@us.ibm.com>
9971 * ChangeLog: Clean up from mid air collision
9973 2017-07-06 Carl Love <cel@us.ibm.com>
9975 * config/rs6000/rs6000-c.c: Add support for built-in functions
9976 vector signed int vec_subc (vector signed int, vector signed int);
9977 vector signed __int128 vec_subc (vector signed __int128,
9978 vector signed __int128);
9979 vector unsigned __int128 vec_subc (vector unsigned __int128,
9980 vector unsigned __int128);
9981 vector signed int vec_sube (vector signed int, vector signed int,
9983 vector unsigned int vec_sube (vector unsigned int,
9984 vector unsigned int,
9985 vector unsigned int);
9986 vector signed __int128 vec_sube (vector signed __int128,
9987 vector signed __int128,
9988 vector signed__int128);
9989 vector unsigned __int128 vec_sube (vector unsigned __int128,
9990 vector unsigned __int128,
9991 vector unsigned __int128);
9992 vector signed int vec_subec (vector signed int, vector signed int,
9994 vector unsigned int vec_subec (vector unsigned int,
9995 vector unsigned int,
9996 vector unsigned int);
9997 vector signed __int128 vec_subec (vector signed __int128,
9998 vector signed __int128,
9999 vector signed__int128);
10000 vector unsigned __int128 vec_subec (vector unsigned __int128,
10001 vector unsigned __int128,
10002 vector unsigned __int128);
10003 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
10004 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
10005 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
10006 BU_ALTIVEC_OVERLOAD_X definitions.
10007 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
10008 * doc/extend.texi: Update the built-in documentation file for the new
10009 built-in functions.
10011 2017-07-06 David Malcolm <dmalcolm@redhat.com>
10014 * diagnostic-show-locus.c (layout::layout): Use start and finish
10015 spelling location for the start and finish of each range.
10016 * genmatch.c (linemap_client_expand_location_to_spelling_point):
10017 Add unused aspect param.
10018 * input.c (expand_location_1): Add "aspect" param, and use it
10019 to access the correct part of the location.
10020 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
10022 (expand_location_to_spelling_point): Likewise.
10023 (linemap_client_expand_location_to_spelling_point): Add "aspect"
10024 param, and pass it to expand_location_1.
10026 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
10028 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
10029 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
10030 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
10031 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
10032 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
10033 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
10034 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
10035 _mm_maskz_getmant_ss): New intrinsics.
10036 (__builtin_ia32_getexpss128_mask): Changed to ...
10037 __builtin_ia32_getexpss128_round ... this.
10038 (__builtin_ia32_getexpsd128_mask): Changed to ...
10039 __builtin_ia32_getexpsd128_round ... this.
10040 * config/i386/i386-builtin-types.def
10041 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
10042 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
10043 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
10044 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
10045 __builtin_ia32_getmantss_mask_round): New builtins.
10046 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
10047 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
10048 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
10049 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
10050 * config/i386/sse.md
10051 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
10052 avx512f_sgetexp<mode><mask_scalar_name>
10053 <round_saeonly_scalar_name> ... this.
10054 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
10055 %0, %1, %2<round_saeonly_op3>}): Changed to ...
10056 vgetexp<ssescalarmodesuffix>
10057 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10058 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
10059 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
10060 avx512f_vgetmant<mode><mask_scalar_name>
10061 <round_saeonly_scalar_name> ... this.
10062 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
10063 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
10064 vgetmant<ssescalarmodesuffix>
10065 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
10066 %0<mask_scalar_operand4>, %1, %2
10067 <round_saeonly_scalar_mask_op4>, %3} ... this.
10068 * config/i386/subst.md (mask_scalar_operand4,
10069 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
10070 round_saeonly_scalar_nimm_predicate): New subst attributes.
10072 2017-07-06 Julia Koval <julia.koval@intel.com>
10074 * config/i386/i386.c (ix86_erase_embedded_rounding):
10075 Remove code for old rounding pattern.
10077 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
10079 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
10081 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
10083 * doc/sourcebuild.texi (Test Directives, Variants of
10084 dg-require-support): Add documentation for dg-require-stack-check.
10086 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
10088 * config/i386/subst.md (mask_scalar, round_scalar,
10089 round_saeonly_scalar): New meta-templates.
10090 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
10091 round_scalar_mask_operand3, round_scalar_mask_op3,
10092 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
10093 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
10094 round_saeonly_scalar_constraint,
10095 round_saeonly_scalar_prefix): New subst attribute.
10096 * config/i386/sse.md
10097 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
10098 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
10099 <round_scalar_name> ... this.
10100 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
10101 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
10102 <round_scalar_name> ... this.
10103 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
10104 <sse>_vm<code><mode>3<mask_scalar_name>
10105 <round_saeonly_scalar_name> ... this.
10106 (v<plusminus_mnemonic><ssescalarmodesuffix>
10107 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10108 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10109 v<plusminus_mnemonic><ssescalarmodesuffix>
10110 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10111 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10112 (v<multdiv_mnemonic><ssescalarmodesuffix>
10113 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
10114 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
10115 v<multdiv_mnemonic><ssescalarmodesuffix>
10116 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10117 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
10118 (v<maxmin_float><ssescalarmodesuffix>
10119 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
10120 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
10121 v<maxmin_float><ssescalarmodesuffix>
10122 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
10123 %0<mask_scalar_operand3>, %1, %<iptr>2
10124 <round_saeonly_scalar_mask_op3>} ... this.
10126 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
10128 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
10129 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
10131 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
10132 Alan Hayward <alan.hayward@arm.com>
10133 David Sherwood <david.sherwood@arm.com>
10135 * combine.c (simplify_if_then_else): Remove "enum" before
10137 * compare-elim.c (can_eliminate_compare): Likewise.
10138 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10140 (aarch64_lookup_simd_builtin_type): Likewise.
10141 (aarch64_simd_builtin_type): Likewise.
10142 (aarch64_init_simd_builtin_types): Likewise.
10143 (aarch64_simd_expand_args): Likewise.
10144 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
10146 (aarch64_reverse_mask): Likewise.
10147 (aarch64_simd_emit_reg_reg_move): Likewise.
10148 (aarch64_gen_adjusted_ldpstp): Likewise.
10149 (aarch64_ccmp_mode_to_code): Likewise.
10150 (aarch64_operands_ok_for_ldpstp): Likewise.
10151 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10152 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10154 (aarch64_min_divisions_for_recip_mul): Likewise.
10155 (aarch64_reassociation_width): Likewise.
10156 (aarch64_get_condition_code_1): Likewise.
10157 (aarch64_simd_emit_reg_reg_move): Likewise.
10158 (aarch64_simd_attr_length_rglist): Likewise.
10159 (aarch64_reverse_mask): Likewise.
10160 (aarch64_operands_ok_for_ldpstp): Likewise.
10161 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10162 (aarch64_gen_adjusted_ldpstp): Likewise.
10163 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
10165 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
10166 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10167 (arm_lookup_simd_builtin_type): Likewise.
10168 (arm_simd_builtin_type): Likewise.
10169 (arm_init_simd_builtin_types): Likewise.
10170 (arm_expand_builtin_args): Likewise.
10171 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
10172 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
10173 (ft32_setup_incoming_varargs): Likewise.
10174 (ft32_function_arg): Likewise.
10175 (ft32_function_arg_advance): Likewise.
10176 (ft32_pass_by_reference): Likewise.
10177 (ft32_arg_partial_bytes): Likewise.
10178 (ft32_valid_pointer_mode): Likewise.
10179 (ft32_addr_space_pointer_mode): Likewise.
10180 (ft32_addr_space_legitimate_address_p): Likewise.
10181 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
10183 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
10184 (ix86_emit_outlined_ms2sysv_restore): Likewise.
10185 (iamcu_alignment): Likewise.
10186 (canonicalize_vector_int_perm): Likewise.
10187 (ix86_noce_conversion_profitable_p): Likewise.
10188 (ix86_mpx_bound_mode): Likewise.
10189 (ix86_operands_ok_for_move_multiple): Likewise.
10190 * config/microblaze/microblaze-protos.h
10191 (microblaze_expand_conditional_branch_reg): Likewise.
10192 * config/microblaze/microblaze.c
10193 (microblaze_expand_conditional_branch_reg): Likewise.
10194 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10196 (rs6000_reassociation_width): Likewise.
10197 (rs6000_invalid_binary_op): Likewise.
10198 (fusion_p9_p): Likewise.
10199 (emit_fusion_p9_load): Likewise.
10200 (emit_fusion_p9_store): Likewise.
10201 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
10203 (riscv_hard_regno_mode_ok_p): Likewise.
10204 (riscv_address_insns): Likewise.
10205 (riscv_split_symbol): Likewise.
10206 (riscv_legitimize_move): Likewise.
10207 (riscv_function_value): Likewise.
10208 (riscv_hard_regno_nregs): Likewise.
10209 (riscv_expand_builtin): Likewise.
10210 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
10211 (riscv_build_integer): Likewise.
10212 (riscv_split_integer): Likewise.
10213 (riscv_legitimate_constant_p): Likewise.
10214 (riscv_cannot_force_const_mem): Likewise.
10215 (riscv_regno_mode_ok_for_base_p): Likewise.
10216 (riscv_valid_base_register_p): Likewise.
10217 (riscv_valid_offset_p): Likewise.
10218 (riscv_valid_lo_sum_p): Likewise.
10219 (riscv_classify_address): Likewise.
10220 (riscv_legitimate_address_p): Likewise.
10221 (riscv_address_insns): Likewise.
10222 (riscv_load_store_insns): Likewise.
10223 (riscv_force_binary): Likewise.
10224 (riscv_split_symbol): Likewise.
10225 (riscv_force_address): Likewise.
10226 (riscv_legitimize_address): Likewise.
10227 (riscv_move_integer): Likewise.
10228 (riscv_legitimize_const_move): Likewise.
10229 (riscv_legitimize_move): Likewise.
10230 (riscv_address_cost): Likewise.
10231 (riscv_subword): Likewise.
10232 (riscv_output_move): Likewise.
10233 (riscv_canonicalize_int_order_test): Likewise.
10234 (riscv_emit_int_order_test): Likewise.
10235 (riscv_function_arg_boundary): Likewise.
10236 (riscv_pass_mode_in_fpr_p): Likewise.
10237 (riscv_pass_fpr_single): Likewise.
10238 (riscv_pass_fpr_pair): Likewise.
10239 (riscv_get_arg_info): Likewise.
10240 (riscv_function_arg): Likewise.
10241 (riscv_function_arg_advance): Likewise.
10242 (riscv_arg_partial_bytes): Likewise.
10243 (riscv_function_value): Likewise.
10244 (riscv_pass_by_reference): Likewise.
10245 (riscv_setup_incoming_varargs): Likewise.
10246 (riscv_print_operand): Likewise.
10247 (riscv_elf_select_rtx_section): Likewise.
10248 (riscv_save_restore_reg): Likewise.
10249 (riscv_for_each_saved_reg): Likewise.
10250 (riscv_register_move_cost): Likewise.
10251 (riscv_hard_regno_mode_ok_p): Likewise.
10252 (riscv_hard_regno_nregs): Likewise.
10253 (riscv_class_max_nregs): Likewise.
10254 (riscv_memory_move_cost): Likewise.
10255 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
10256 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
10257 (rl78_addr_space_address_mode): Likewise.
10258 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10260 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10261 (rs6000_reassociation_width): Likewise.
10262 (rs6000_invalid_binary_op): Likewise.
10263 (fusion_p9_p): Likewise.
10264 (emit_fusion_p9_load): Likewise.
10265 (emit_fusion_p9_store): Likewise.
10266 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
10267 (ok_for_simple_move_operands): Likewise.
10268 (ok_for_simple_move_strict_operands): Likewise.
10269 (ok_for_simple_arith_logic_operands): Likewise.
10270 (visium_legitimize_reload_address): Likewise.
10271 (visium_select_cc_mode): Likewise.
10272 (output_cbranch): Likewise.
10273 (visium_split_double_move): Likewise.
10274 (visium_expand_copysign): Likewise.
10275 (visium_expand_int_cstore): Likewise.
10276 (visium_expand_fp_cstore): Likewise.
10277 * config/visium/visium.c (visium_pass_by_reference): Likewise.
10278 (visium_function_arg): Likewise.
10279 (visium_function_arg_advance): Likewise.
10280 (visium_libcall_value): Likewise.
10281 (visium_setup_incoming_varargs): Likewise.
10282 (visium_legitimate_constant_p): Likewise.
10283 (visium_legitimate_address_p): Likewise.
10284 (visium_legitimize_address): Likewise.
10285 (visium_secondary_reload): Likewise.
10286 (visium_register_move_cost): Likewise.
10287 (visium_memory_move_cost): Likewise.
10288 (prepare_move_operands): Likewise.
10289 (ok_for_simple_move_operands): Likewise.
10290 (ok_for_simple_move_strict_operands): Likewise.
10291 (ok_for_simple_arith_logic_operands): Likewise.
10292 (visium_function_value_1): Likewise.
10293 (rtx_ok_for_offset_p): Likewise.
10294 (visium_legitimize_reload_address): Likewise.
10295 (visium_split_double_move): Likewise.
10296 (visium_expand_copysign): Likewise.
10297 (visium_expand_int_cstore): Likewise.
10298 (visium_expand_fp_cstore): Likewise.
10299 (visium_split_cstore): Likewise.
10300 (visium_select_cc_mode): Likewise.
10301 (visium_split_cbranch): Likewise.
10302 (output_cbranch): Likewise.
10303 (visium_print_operand_address): Likewise.
10304 * expmed.c (flip_storage_order): Likewise.
10305 * expmed.h (emit_cstore): Likewise.
10306 (flip_storage_order): Likewise.
10307 * genrecog.c (validate_pattern): Likewise.
10308 * hsa-gen.c (gen_hsa_addr): Likewise.
10309 * internal-fn.c (expand_arith_overflow): Likewise.
10310 * ira-color.c (allocno_copy_cost_saving): Likewise.
10311 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10312 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
10313 (process_invariant_for_inheritance): Likewise.
10314 * lra-eliminations.c (move_plus_up): Likewise.
10315 * omp-low.c (lower_oacc_reductions): Likewise.
10316 * simplify-rtx.c (simplify_subreg): Likewise.
10317 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
10318 (TARGET_CHKP_BOUND_MODE): Likewise..
10319 * targhooks.c (default_chkp_bound_mode): Likewise.
10320 (default_setup_incoming_vararg_bounds): Likewise.
10321 * targhooks.h (default_chkp_bound_mode): Likewise.
10322 (default_setup_incoming_vararg_bounds): Likewise.
10323 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
10324 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
10325 (have_whole_vector_shift): Likewise.
10326 * tree-vect-stmts.c (vectorizable_load): Likewise.
10327 * doc/tm.texi: Regenerate.
10329 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10331 Graceful degrade if Binutils PR21472 is not available.
10334 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
10335 .rodata in flash test fails.
10336 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
10337 * confgure: Regenerate.
10338 * config.in: Regenerate.
10339 * config/avr/avr.c (avr_asm_named_section)
10340 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
10341 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
10342 (avr_asm_init_sections): Same.
10344 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10346 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
10347 (fma<VH:mode>4_intrinsic): Likewise.
10348 (*fmsub<VCVTF:mode>4): Likewise.
10349 (*fmsub<VH:mode>4_intrinsic): Likewise.
10351 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
10354 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
10355 Don't depend on "optimize > 0".
10356 (out_movhi_r_mr, out_movqi_mr_r): Same.
10357 (out_movhi_mr_r, out_movqi_r_mr): Same.
10358 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
10359 io_address_operand on "optimize > 0".
10361 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10363 * tree-loop-distribution.c: Add general explanantion on the pass.
10364 (generate_loops_for_partition): Mark distributed loop.
10365 (pg_add_dependence_edges): New parameter. Handle alias data
10366 dependence specially and record it in the parameter if asked.
10367 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
10368 (init_partition_graph_vertices, add_partition_graph_edge): New.
10369 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
10370 (free_partition_graph_vdata, build_partition_graph): New.
10371 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
10372 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
10373 (data_ref_segment_size, latch_dominated_by_data_ref): New.
10374 (compute_alias_check_pairs, version_loop_by_alias_check): New.
10375 (version_for_distribution_p, finalize_partitions): New.
10376 (distribute_loop): Handle alias data dependence specially. Factor
10377 out loop fusion code as functions and call these functions.
10379 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10381 * tree-loop-distribution.c (classify_partition): New parameter and
10382 better handle reduction statement.
10383 (rdg_build_partitions): Revise comment.
10384 (distribute_loop): Compute statements in all partitions and pass it
10385 to classify_partition.
10387 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10389 * tree-loop-distribution.c (enum partition_type): New.
10390 (struct partition): New field type.
10391 (partition_merge_into): Add parameter. Update partition type.
10392 (data_dep_in_cycle_p, update_type_for_merge): New functions.
10393 (build_rdg_partition_for_vertex): Compute partition type.
10394 (rdg_build_partitions): Dump partition type.
10395 (distribute_loop): Update calls to partition_merge_into.
10397 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10399 * tree-loop-distribution.c (struct ddr_hasher): New.
10400 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
10402 (classify_partition): Call get_data_dependence.
10403 (pg_add_dependence_edges): Ditto.
10404 (distribute_loop): Release data dependence hash table.
10406 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10408 * tree-loop-distribution.c (ref_base_address): Delete.
10409 (similar_memory_accesses): Rename ...
10410 (share_memory_accesses): ... to this. Check if partitions access
10411 the same memory reference.
10412 (distribute_loop): Call share_memory_accesses.
10414 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10416 * tree-loop-distribution.c (struct partition): New field recording
10417 its data reference.
10418 (partition_alloc, partition_free): Init and release data refs.
10419 (partition_merge_into): Merge data refs.
10420 (build_rdg_partition_for_vertex): Collect data refs for partition.
10421 (pg_add_dependence_edges): Change parameters from vector to bitmap.
10423 (distribute_loop): Remve data refs from vertice data of partition
10426 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10428 * tree-loop-distribution.c (params.h): Include header file.
10429 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
10430 (datarefs_vec): New global var.
10431 (create_rdg_vertices): Use datarefs_vec directly.
10432 (free_rdg): Don't free data references.
10433 (build_rdg): Update use. Don't free data references.
10434 (distribute_loop): Compute global variable for data references.
10435 Bail out if there are too many data references.
10437 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10439 * tree-loop-distribution.c (loop_nest): New global var.
10440 (build_rdg): Use loop directly, rather than loop nest.
10441 (pg_add_dependence_edges): Remove loop nest parameter. Use global
10443 (distribute_loop): Compute global variable loop nest. Update use.
10445 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10447 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
10448 (partition_merge_into): New parameter. Dump reason for fusion.
10449 (distribute_loop): Update use of partition_merge_into.
10451 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10453 * tree-loop-distribution.c (bb_top_order_index): New.
10454 (bb_top_order_index_size, bb_top_order_cmp): New.
10455 (stmts_from_loop): Use topological order.
10456 (pass_loop_distribution::execute): Compute and release topological
10457 order for basic blocks.
10459 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10461 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
10464 2017-07-05 Bin Cheng <bin.cheng@arm.com>
10466 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
10467 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
10468 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
10469 * internal-fn.def (LOOP_DIST_ALIAS): New.
10470 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
10471 (fold_loop_internal_call): ... this.
10472 (vect_loop_dist_alias_call): New function.
10473 (set_uid_loop_bbs): Call fold_loop_internal_call.
10474 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
10477 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10480 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
10481 Require dead FLAGS_REG at the beginning of a peephole.
10483 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
10486 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
10487 arguments in the call to __builtin_ia32_sbb_u32.
10488 (_subborrow_u64): Swap _X and _Y arguments in the call to
10489 __builtin_ia32_sbb_u64.
10491 2017-07-04 Jakub Jelinek <jakub@redhat.com>
10494 * tree-vrp.c (compare_assert_loc): Turn into a function template
10495 with stable template parameter. Only test if a->e is NULL,
10496 !a->e == !b->e has been verified already. Use e == NULL or
10497 e != NULL instead of e or ! e tests. If stable is true, don't use
10498 iterative_hash_expr, on the other side allow a or b or both NULL
10499 and sort the NULLs last.
10500 (process_assert_insertions): Sort using compare_assert_loc<false>
10501 instead of compare_assert_loc, later sort using
10502 compare_assert_loc<true> before calling process_assert_insertions_for
10503 in a loop. Use break instead of continue once seen NULL pointer.
10505 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10507 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
10508 Cortex-R7 and Cortex-R8 processors.
10510 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10512 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
10513 uninitialized while src is not.
10515 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
10517 * common/config/arm/arm-common.c: Adjust include path for
10519 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
10520 (arm-cpu.h): Create in build directory. Adjust dependency rules.
10521 (arm-cpu-data.h): Likewise.
10522 (arm-cpu-cdata.h): Likewise.
10523 * config/arm/arm-cpu.h: Delete.
10524 * config/arm/arm-cpu-cdata.h: Delete.
10525 * config/arm/arm-cpu-data.h: Delete.
10527 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
10529 * config/arm/arm-cpus.in (cortex-a55): New.
10530 (cortex-a75): Likewise.
10531 (cortex-a75.cortex-a55): Likewise.
10532 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
10534 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
10535 * config/arm/arm-cpu-cdata.h: Regenerate.
10536 * config/arm/arm-cpu-data.h: Regenerate.
10537 * config/arm/arm-cpu.h: Regenerate.
10538 * config/arm/arm-tables.opt: Regenerate.
10539 * config/arm/arm-tune.md: Regenerate.
10541 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10543 * haifa-sched.c (sched_create_recovery_edges): Update profile.
10545 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
10547 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
10550 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
10552 PR tree-optimization/81292
10553 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
10554 full_string_p, also call adjust_related_strinfos if the adjustment
10555 is simple, otherwise invalidate related strinfos.
10557 2017-07-04 Martin Liska <mliska@suse.cz>
10560 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
10561 newly created variable as DECL_IGNORED_P.
10563 2017-07-04 Martin Liska <mliska@suse.cz>
10566 * ipa-inline.c (inline_small_functions):
10567 Use xstrdup_for_dump.
10569 2017-07-04 Tom de Vries <tom@codesourcery.com>
10571 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
10573 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10576 * config/darwin.c (darwin_function_switched_text_sections):
10579 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
10581 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
10583 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10585 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
10587 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10589 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
10590 min_profitable_iters, and th as inclusive lower bounds.
10591 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
10592 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
10593 for min_profitable_iters and min_profitable_estimate.
10594 (vect_transform_loop): Treat th as an inclusive lower bound.
10595 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
10597 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
10600 * config/darwin.c (darwin_function_switched_text_sections):
10601 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
10602 in two pieces, and suppress the use of buf.
10604 2017-07-03 Nathan Sidwell <nathan@acm.org>
10606 * hash-table.h (hash_table_mod1): Fix indentation.
10608 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10610 PR middle-end/81290
10611 * predict.c (force_edge_cold): Be more careful about propagation
10613 * profile-count.h (profile_probability::guessed,
10614 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
10616 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
10618 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
10620 * doc/invoke.texi (rcpc architecture extension): Document it.
10622 2017-07-03 Richard Biener <rguenther@suse.de>
10624 PR tree-optimization/60510
10625 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
10626 the scalar reduction PHI and use it.
10627 (vectorizable_reduction): Properly guard the single_defuse_cycle
10628 path for non-SLP reduction chains where we cannot use it.
10629 Rework reduc_def/index and vector type deduction. Rework
10630 vector operand gathering during reduction op code-gen.
10631 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
10632 chains dissolve the chain and leave it to non-SLP reduction
10635 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10637 * tree-data-ref.h (dr_alignment): Declare.
10638 * tree-data-ref.c (dr_alignment): New function.
10639 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
10640 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
10642 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
10644 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10646 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
10647 and base_misalignment fields.
10648 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
10649 * tree-data-ref.c: Include builtins.h.
10650 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
10651 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
10652 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
10653 * tree-vect-data-refs.c: Include tree-cfg.h.
10654 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
10655 fields instead of calculating an alignment here.
10656 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
10657 innermost_loop_behavior fields.
10659 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10661 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
10663 (DR_STEP_ALIGNMENT): New macro.
10664 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
10665 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
10666 (create_data_ref): Print it.
10667 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
10668 to tell whether the step preserves vector (mis)alignment.
10669 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10670 Move the check for an integer step and generalise to all INTEGER_CST.
10671 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
10672 Print the outer step alignment.
10674 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10676 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
10677 with offset_alignment.
10678 (DR_ALIGNED_TO): Delete.
10679 (DR_OFFSET_ALIGNMENT): New macro.
10680 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
10681 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
10682 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
10683 (create_data_ref): Likewise.
10684 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
10685 (vect_analyze_data_refs): Likewise.
10686 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
10687 creating dummy innermost behavior.
10689 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10691 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
10692 with a "innermost_loop_behavior *" and refeence tree.
10693 * tree-data-ref.c (dr_analyze_innermost): Likewise.
10694 (create_data_ref): Update call accordingly.
10695 * tree-predcom.c (find_looparound_phi): Likewise.
10697 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10699 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
10700 fields with dr_wrt_vec_loop.
10701 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
10702 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
10703 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
10704 (vect_dr_behavior): New function.
10705 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10706 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
10707 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
10708 track whether the step preserves the misalignment.
10709 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
10710 Use vect_dr_behavior.
10711 (vect_setup_realignment): Update call accordingly.
10712 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
10713 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
10714 call to vect_create_addr_base_for_vector_ref.
10715 (vect_create_cond_for_align_checks): Likewise.
10716 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
10717 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
10718 (vect_recog_mask_conversion_pattern): Likewise.
10719 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
10720 (new_stmt_vec_info): Remove redundant zeroing.
10722 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
10724 * common/config/arm/arm-common.c (arm_be8_option): New function.
10725 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
10726 (ISA_ARMv6): Add isa_bit_be8.
10727 * config/arm/arm.h (arm_be8_option): Add prototype.
10728 (BE8_SPEC_FUNCTION): New define.
10729 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
10730 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
10731 (mlittle-endian): Similarly.
10732 (mbe8, mbe32): New options.
10733 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
10734 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
10736 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10738 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
10740 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10742 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
10743 (cleanup_tree_cfg_bb): Use it.
10744 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
10746 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
10748 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10751 * loop-doloop.c (add_test): Update profile.
10753 2017-07-03 Martin Liska <mliska@suse.cz>
10756 * sanopt.c (rewrite_usage_of_param): New function.
10757 (sanitize_rewrite_addressable_params): Likewise.
10758 (pass_sanopt::execute): Call rewrite_usage_of_param.
10760 2017-07-03 Richard Biener <rguenther@suse.de>
10762 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
10763 back to using VIEW_CONVERT_EXPR.
10765 2017-07-03 Martin Liska <mliska@suse.cz>
10768 * doc/extend.texi: Document when a resolver function is
10769 generated for target_clones.
10771 2017-07-03 Martin Liska <mliska@suse.cz>
10773 * asan.c (asan_emit_stack_protection): Unpoison just red zones
10774 and shadow memory of auto variables which are subject of
10775 use-after-scope sanitization.
10776 (asan_expand_mark_ifn): Add do set only when is_poison.
10778 2016-07-03 Richard Biener <rguenther@suse.de>
10780 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
10782 (vect_force_simple_reduction): Record reduction def -> phi mapping.
10783 (vectorizable_reduction): Perform reduction PHI creation when
10784 visiting a reduction PHI and adjust and simplify code generation
10785 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
10786 (vect_transform_loop): Visit reduction PHIs.
10787 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
10788 defs into the SLP tree.
10789 (vect_build_slp_tree): Reduction defs terminate the recursion.
10790 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
10792 (vect_get_vec_defs_for_stmt_copy): Export.
10793 (vect_get_vec_defs): Likewise.
10794 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
10796 (vect_get_vec_defs_for_stmt_copy): Declare.
10797 (vect_get_vec_defs): Likewise.
10799 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
10801 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
10802 parameter with a "loop" parameter and use it instead of the
10803 loop containing DR_STMT. Don't check simple_iv when doing
10804 BB analysis. Describe the two analysis modes in the comment.
10806 2017-07-03 Tom de Vries <tom@codesourcery.com>
10808 PR tree-optimization/69468
10809 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
10810 (find_same_succ_bb): Handle ignore_edge_flags.
10812 2017-07-03 Tom de Vries <tom@codesourcery.com>
10814 PR tree-optimization/81192
10815 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
10817 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
10819 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
10821 2017-07-03 Tom de Vries <tom@codesourcery.com>
10823 PR tree-optimization/81192
10824 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
10825 BB_SAME_SUCC (bb) == NULL.
10827 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10829 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
10832 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10834 * dumpfile.c: Include profile-count.h
10835 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
10837 (insert_cond_bb): Update profile.
10838 * tree-cfg.h (insert_cond_bb): Update prototype.
10839 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
10840 * tree-dump.c: Do not include tree-cfg.
10842 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10844 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
10846 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10848 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
10851 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
10853 * tree-complex.c (expand_complex_div_wide): update profile.
10855 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10856 Alan Hayward <alan.hayward@arm.com>
10857 David Sherwood <david.sherwood@arm.com>
10859 * Makefile.in (MACHMODE_H): Remove insn-modes.h
10860 (CORETYPES_H): New define.
10861 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
10862 (insn-modes-inline.h, s-modes-inline-h): New rules.
10863 (generated_files): Add insn-modes-inline.h.
10864 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
10865 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
10866 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
10867 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
10868 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
10869 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
10870 (build/gencodes.o, build/genconditions.o): Likewise.
10871 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
10872 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
10873 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
10874 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
10875 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
10876 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
10877 * coretypes.h: Include everything up to real.h for generators.
10878 Include insn-modes.h first. Include wide-int-print.h after
10879 wide-int.h. Include insn-modes-inline.h and then machmode.h.
10880 * machmode.h: Don't include insn-modes.h here.
10881 * function-tests.c: Remove includes of signop.h, machmode.h,
10882 double-int.h and wide-int.h.
10884 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
10886 * optc-save-gen.awk: Likewise.
10887 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
10888 * godump.c: Remove include of wide-int-print.h.
10889 * pretty-print.h: Likewise.
10890 * wide-int-print.cc: Likewise.
10891 * wide-int.cc: Likewise.
10892 * hash-map-tests.c: Remove include of signop.h.
10893 * hash-set-tests.c: Likewise.
10894 * rtl-tests.c: Likewise.
10895 * mkconfig.sh: Remove include of machmode.h.
10896 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
10898 (emit_insn_modes_inline_h): ...this new function. Emit the code
10899 into an insn-modes-inline.h header file, adding appropriate
10900 include guards and end comments.
10901 (emit_insn_modes_c_header): Remove include of machmode.h.
10902 (emit_min_insn_modes_c_header): Include coretypes.h rather than
10904 (main): Handle -i flag and call emit_insn_modes_inline_h when
10907 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10909 * tree-ssa-strlen.c (strinfo): Rename the length field to
10910 nonzero_chars. Add a full_string_p field.
10911 (compare_nonzero_chars, zero_length_string_p): New functions.
10912 (get_addr_stridx): Add an offset_out parameter.
10913 Use compare_nonzero_chars.
10914 (get_stridx): Update accordingly. Use compare_nonzero_chars.
10915 (new_strinfo): Update after above changes to strinfo.
10916 (set_endptr_and_length): Set full_string_p.
10917 (get_string_length): Update after above changes to strinfo.
10918 (unshare_strinfo): Update call to new_strinfo.
10919 (maybe_invalidate): Likewise.
10920 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
10921 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
10922 as a uhwi instead of an shwi. Update after above changes to
10923 strinfo and new_strinfo.
10924 (zero_length_string): Assert that chainsi contains full strings.
10925 Use zero_length_string_p. Update call to new_strinfo.
10926 (adjust_related_strinfos): Update after above changes to strinfo.
10927 Copy full_string_p from origsi.
10928 (adjust_last_stmt): Use zero_length_string_p.
10929 (handle_builtin_strlen): Update after above changes to strinfo and
10930 new_strinfo. Install the lhs as the string length if the previous
10931 entry didn't describe a full string.
10932 (handle_builtin_strchr): Update after above changes to strinfo
10934 (handle_builtin_strcpy): Likewise.
10935 (handle_builtin_strcat): Likewise.
10936 (handle_builtin_malloc): Likewise.
10937 (handle_pointer_plus): Likewise.
10938 (handle_builtin_memcpy): Likewise. Track nonzero characters
10939 that aren't necessarily followed by a nul terminator.
10940 (handle_char_store): Likewise.
10942 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10944 PR tree-optimization/80769
10945 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
10946 for malloc and calloc. Document the new invariant that all related
10947 strinfos have delayed lengths or none do.
10948 (verify_related_strinfos): Move earlier in file.
10949 (set_endptr_and_length): New function, split out from...
10950 (get_string_length): ...here. Also set the lengths of related
10952 (zero_length_string): Assert that chainsi has known (rather than
10954 (adjust_related_strinfos): Likewise.
10956 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
10958 PR tree-optimization/81136
10959 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
10960 assert that two references with the same misalignment have the same
10961 compile-time misalignment if those compile-time misalignments
10964 2017-07-01 Andi Kleen <ak@linux.intel.com>
10966 * print-tree.c (print_node): Print all attributes.
10968 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10970 * cfg.c (scale_bbs_frequencies): New function.
10971 * cfg.h (scale_bbs_frequencies): Declare it.
10972 * cfgloopanal.c (single_likely_exit): Cleanup.
10973 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
10975 (scale_loop_profile): Likewise.
10976 (loop_version): Likewise.
10977 (create_empty_loop_on_edge): Update.
10978 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
10979 scale_loop_frequencies, scale_loop_profile, loopify,
10980 loop_version): Update prototypes.
10981 * modulo-sched.c (sms_schedule): Update.
10982 * predict.c (unlikely_executed_edge_p): Also check probability.
10983 (probably_never_executed_edge_p): Fix typo.
10984 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10985 * tree-parloops.c (gen_parallel_loop): Update.
10986 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
10987 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10988 * tree-ssa-loop-split.c (split_loop): Update.
10989 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10990 * tree-vect-loop-manip.c (vect_do_peeling): Update.
10991 (vect_loop_versioning): Update.
10992 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10994 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
10996 * trans-mem.c (split_bb_make_tm_edge): Update profile.
10998 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11000 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
11001 to keep profile consistent.
11003 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11005 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
11006 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
11007 * profile-count.h (max_safe_multiplier): Make unsigned.
11008 (profile_count::guessed_zero): New.
11010 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11012 * bb-reorder.c (fix_up_crossing_landing_pad,
11013 fix_crossing_conditional_branches): Use make_single_succ_edge
11014 to keep profile consistent.
11016 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
11018 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
11021 2017-07-01 Jakub Jelinek <jakub@redhat.com>
11024 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
11025 the right scopes, make sure cond_jump isn't preserved between multiple
11026 iterations. Search for fallthru edge whenever there are 3+ edges and
11027 use find_fallthru_edge for it.
11029 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11031 Patch by Alexander Monakov <amonakov@ispras.ru>
11032 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
11033 probabilities consistently.
11035 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11037 * pa.c (pa_expand_compare_and_swap_loop): Update call of
11038 emit_cmp_and_jump_insns.
11040 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11043 * tree-inline.c (expand_call_inline): Combine profile statuses.
11045 2017-06-30 Andrew Pinski <apinski@cavium.com>
11047 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
11048 fold_stmt returned true.
11050 2017-06-30 Nathan Sidwell <nathan@acm.org>
11052 * ggc.h (empty_string): Delete.
11053 * cfgexpand.c (expand_asm_stmt): Use plain "".
11054 * optabs.c (expand_asm_memory_barrier): Likewise.
11055 * stringpool.c (empty_string): Delete.
11056 (digit_vector, digit_string): Delete.
11057 (ggc_alloc_string): Use plain "", don't optimize single digit
11058 strings. Use ggc_alloc_atomic.
11060 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
11062 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
11063 comparison set and one other set, use the cost of the non-comparison
11066 2017-06-30 Nathan Sidwell <nathan@acm.org>
11068 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
11071 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
11073 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
11074 loops. Remove now unneeded calls to gimple_switch_set_label() that
11075 just set removed labels to NULL_TREE.
11077 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
11079 * tree-ssanames.c (set_range_info_raw): Abstract from ...
11080 (set_range_info): ...here. Only call set_range_info_raw if domain
11082 (set_nonzero_bits): Call set_range_info_raw.
11083 * tree-ssanames.h (set_range_info_raw): New.
11085 2017-06-30 Jakub Jelinek <jakub@redhat.com>
11088 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
11089 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
11090 of nonimmediate_operand and <store_mask_constraint> instead of m
11091 for the input operand. For V8FI iterator, always split if input
11092 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
11093 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
11094 <store_mask_predicate> instead of register_operand and
11095 <store_mask_constraint> instead of v for the input operand. Make
11096 sure both operands aren't MEMs for if not <mask_applied>.
11098 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
11100 * lto-wrapper.c (copy_file) Close both file descriptors before
11103 2017-06-30 Martin Liska <mliska@suse.cz>
11106 * multiple_target.c (create_dispatcher_calls): Make ifunc
11107 also for function that don't have calls or are not referenced.
11109 2017-06-30 Richard Biener <rguenther@suse.de>
11111 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
11112 analyze the first scalar stmt. Move vector type computation
11113 for the BB case here from ...
11114 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
11115 live operation processing in the SLP case properly.
11117 2017-06-30 Richard Biener <rguenther@suse.de>
11119 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
11121 2017-06-30 Martin Liska <mliska@suse.cz>
11124 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
11125 before BUILT_IN_UNWIND_RESUME when ASAN is used.
11127 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
11129 * doc/invoke.texi (AArch64): Add missing options and remove redundant
11132 2017-06-30 Richard Biener <rguenther@suse.de>
11134 PR tree-optimization/81249
11135 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
11136 condition reduction result to original scalar type.
11138 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11140 * profile-count.h (enum profile_quality): Fix typos and whitespace
11143 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11145 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
11146 type for branch probabilities.
11148 2017-06-29 Julian Brown <julian@codesourcery.com>
11149 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11151 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
11152 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
11153 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
11154 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
11156 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11158 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
11159 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
11160 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
11161 CC usage from generic code to here.
11162 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
11163 CC usage into the target macros.
11165 2017-06-29 Maya Rashish <coypu@sdf.org>
11167 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
11170 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11172 * arm/arm-builtins.c: Include profile-count.h
11173 * except.c (sjlj_emit_function_enter): Use
11174 profile_probability::unlikely.
11176 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11178 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
11179 and tocrel_offset be pointer args rather than implicitly using
11181 (legitimate_constant_pool_address_p, rs6000_emit_move,
11182 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
11183 tocrel_offset and use in toc_relative_expr_p call.
11184 (print_operand, print_operand_address): Use static tocrel_base_oac
11185 and tocrel_offset_oac.
11186 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
11189 2017-06-29 Maya Rashish <coypu@sdf.org>
11191 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
11193 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
11195 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
11196 objects, take into account only the alignment of 'op0' and 'mode1' if
11199 2017-06-29 Steve Ellcey <sellcey@cavium.com>
11201 * ccmp.c (ccmp_tree_comparison_p): New function.
11202 (ccmp_candidate_p): Update to use above function.
11203 (get_compare_parts): New function.
11204 (expand_ccmp_next): Update to use new functions.
11205 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
11207 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
11208 take mode as argument.
11209 * ccmp.h (expand_ccmp_expr): Add mode as argument.
11210 * expr.c (expand_expr_real_1): Pass mode as argument.
11212 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
11214 * combine.c (combine_instructions): Print insns to dump_file, together
11217 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
11219 * asan.c (asan_emit_stack_protection): Update.
11220 (create_cond_insert_point): Update.
11221 * auto-profile.c (afdo_propagate_circuit): Update.
11222 * basic-block.h (struct edge_def): Turn probability to
11223 profile_probability.
11224 (EDGE_FREQUENCY): Update.
11225 * bb-reorder.c (find_traces_1_round): Update.
11226 (better_edge_p): Update.
11227 (sanitize_hot_paths): Update.
11228 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
11229 (make_single_succ_edge): Update.
11230 (check_bb_profile): Update.
11231 (dump_edge_info): Update.
11232 (update_bb_profile_for_threading): Update.
11233 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
11234 probabilitycount to 0.
11235 * cfgbuild.c (compute_outgoing_frequencies): Update.
11236 * cfgcleanup.c (try_forward_edges): Update.
11237 (outgoing_edges_match): Update.
11238 (try_crossjump_to_edge): Update.
11239 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
11240 (expand_gimple_tailcall): Update.
11241 (construct_init_block): Use make_single_succ_edge.
11242 (construct_exit_block): Use make_single_succ_edge.
11243 * cfghooks.c (verify_flow_info): Update.
11244 (redirect_edge_succ_nodup): Update.
11245 (split_edge): Update.
11246 (account_profile_record): Update.
11247 * cfgloopanal.c (single_likely_exit): Update.
11248 * cfgloopmanip.c (scale_loop_profile): Update.
11249 (set_zero_probability): Remove.
11250 (duplicate_loop_to_header_edge): Update.
11251 * cfgloopmanip.h (loop_version): Update prototype.
11252 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11253 (force_nonfallthru_and_redirect): Update.
11254 (update_br_prob_note): Update.
11255 (rtl_verify_edges): Update.
11256 (purge_dead_edges): Update.
11257 (rtl_lv_add_condition_to_bb): Update.
11258 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
11259 * cgraphunit.c (init_lowered_empty_function): Update.
11260 (cgraph_node::expand_thunk): Update.
11261 * cilk-common.c: Include profile-count.h
11262 * dojump.c (inv): Remove.
11263 (jumpifnot): Update.
11264 (jumpifnot_1): Update.
11265 (do_jump_1): Update.
11267 (do_jump_by_parts_greater_rtx): Update.
11268 (do_compare_rtx_and_jump): Update.
11269 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
11270 do_jump_1. do_compare_rtx_and_jump): Update prototype.
11271 * dwarf2cfi.c: Include profile-count.h
11272 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
11273 (sjlj_emit_dispatch_table): Likewise.
11274 * explow.c: Include profile-count.h
11275 * expmed.c (emit_store_flag_force): Update.
11276 (do_cmp_and_jump): Update.
11277 * expr.c (compare_by_pieces_d::generate): Update.
11278 (compare_by_pieces_d::finish_mode): Update.
11279 (emit_block_move_via_loop): Update.
11280 (store_expr_with_bounds): Update.
11281 (store_constructor): Update.
11282 (expand_expr_real_2): Update.
11283 (expand_expr_real_1): Update.
11284 * expr.h (try_casesi, try_tablejump): Update prototypes.
11285 * gimple-pretty-print.c (dump_probability): Update.
11286 (dump_profile): New.
11287 (dump_gimple_label): Update.
11288 (dump_gimple_bb_header): Update.
11289 * graph.c (draw_cfg_node_succ_edges): Update.
11290 * hsa-gen.c (convert_switch_statements): Update.
11291 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
11292 (find_if_case_1): Update.
11293 (find_if_case_2): Update.
11294 * internal-fn.c (expand_arith_overflow_result_store): Update.
11295 (expand_addsub_overflow): Update.
11296 (expand_neg_overflow): Update.
11297 (expand_mul_overflow): Update.
11298 (expand_vector_ubsan_overflow): Update.
11299 * ipa-cp.c (good_cloning_opportunity_p): Update.
11300 * ipa-split.c (split_function): Use make_single_succ_edge.
11301 * ipa-utils.c (ipa_merge_profiles): Update.
11302 * loop-doloop.c (add_test): Update.
11303 (doloop_modify): Update.
11304 * loop-unroll.c (compare_and_jump_seq): Update.
11305 (unroll_loop_runtime_iterations): Update.
11306 * lra-constraints.c (lra_inheritance): Update.
11307 * lto-streamer-in.c (input_cfg): Update.
11308 * lto-streamer-out.c (output_cfg): Update.
11309 * mcf.c (adjust_cfg_counts): Update.
11310 * modulo-sched.c (sms_schedule): Update.
11311 * omp-expand.c (expand_omp_for_init_counts): Update.
11312 (extract_omp_for_update_vars): Update.
11313 (expand_omp_ordered_sink): Update.
11314 (expand_omp_for_ordered_loops): Update.
11315 (expand_omp_for_generic): Update.
11316 (expand_omp_for_static_nochunk): Update.
11317 (expand_omp_for_static_chunk): Update.
11318 (expand_cilk_for): Update.
11319 (expand_omp_simd): Update.
11320 (expand_omp_taskloop_for_outer): Update.
11321 (expand_omp_taskloop_for_inner): Update.
11322 * omp-simd-clone.c (simd_clone_adjust): Update.
11323 * optabs.c (expand_doubleword_shift): Update.
11324 (expand_abs): Update.
11325 (emit_cmp_and_jump_insn_1): Update.
11326 (expand_compare_and_swap_loop): Update.
11327 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
11328 * predict.c (predictable_edge_p): Update.
11329 (edge_probability_reliable_p): Update.
11330 (set_even_probabilities): Update.
11331 (combine_predictions_for_insn): Update.
11332 (combine_predictions_for_bb): Update.
11333 (propagate_freq): Update.
11334 (estimate_bb_frequencies): Update.
11335 (force_edge_cold): Update.
11336 * profile-count.c (profile_count::dump): Add missing space into dump.
11337 (profile_count::debug): Add newline.
11338 (profile_count::differs_from_p): Explicitly convert to unsigned.
11339 (profile_count::stream_in): Update.
11340 (profile_probability::dump): New member function.
11341 (profile_probability::debug): New member function.
11342 (profile_probability::differs_from_p): New member function.
11343 (profile_probability::differs_lot_from_p): New member function.
11344 (profile_probability::stream_in): New member function.
11345 (profile_probability::stream_out): New member function.
11346 * profile-count.h (profile_count_quality): Rename to ...
11347 (profile_quality): ... this one.
11348 (profile_probability): New.
11349 (profile_count): Update.
11350 * profile.c (compute_branch_probabilities): Update.
11351 * recog.c (peep2_attempt): Update.
11352 * sched-ebb.c (schedule_ebbs): Update.
11353 * sched-rgn.c (find_single_block_region): Update.
11354 (compute_dom_prob_ps): Update.
11355 (schedule_region): Update.
11356 * sel-sched-ir.c (compute_succs_info): Update.
11357 * stmt.c (struct case_node): Update.
11358 (do_jump_if_equal): Update.
11359 (get_outgoing_edge_probs): Update.
11360 (conditional_probability): Update.
11361 (emit_case_dispatch_table): Update.
11362 (expand_case): Update.
11363 (expand_sjlj_dispatch_table): Update.
11364 (emit_case_nodes): Update.
11365 * targhooks.c: Update.
11366 * tracer.c (better_p): Update.
11367 (find_best_successor): Update.
11368 * trans-mem.c (expand_transaction): Update.
11369 * tree-call-cdce.c: Update.
11370 * tree-cfg.c (gimple_split_edge): Upate.
11371 (move_sese_region_to_fn): Upate.
11372 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
11373 * tree-eh.c (lower_resx): Upate.
11374 (cleanup_empty_eh_move_lp): Upate.
11375 * tree-if-conv.c (version_loop_for_if_conversion): Update.
11376 * tree-inline.c (copy_edges_for_bb): Update.
11377 (copy_cfg_body): Update.
11378 * tree-parloops.c (gen_parallel_loop): Update.
11379 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
11380 (gimple_gen_time_profiler): Update.
11381 * tree-ssa-dce.c (remove_dead_stmt): Update.
11382 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11383 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11384 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11385 (unloop_loops): Update.
11386 (try_peel_loop): Update.
11387 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11388 * tree-ssa-loop-split.c (connect_loops): Update.
11389 (split_loop): Update.
11390 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
11391 (hoist_guard): Update.
11392 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11393 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11394 (value_replacement): Update.
11395 * tree-ssa-reassoc.c (branch_fixup): Update.
11396 * tree-ssa-tail-merge.c (replace_block_by): Update.
11397 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11398 (create_edge_and_update_destination_phis): Update.
11399 (compute_path_counts): Update.
11400 (recompute_probabilities): Update.
11401 (update_joiner_offpath_counts): Update.
11402 (freqs_to_counts_path): Update.
11403 (duplicate_thread_path): Update.
11404 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11405 (struct switch_conv_info): Update.
11406 (gen_inbound_check): Update.
11407 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11408 (vect_do_peeling): Update.
11409 (vect_loop_versioning): Update.
11410 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11411 (optimize_mask_stores): Update.
11412 * ubsan.c (ubsan_expand_null_ifn): Update.
11413 * value-prof.c (gimple_divmod_fixed_value): Update.
11414 (gimple_divmod_fixed_value_transform): Update.
11415 (gimple_mod_pow2): Update.
11416 (gimple_mod_pow2_value_transform): Update.
11417 (gimple_mod_subtract): Update.
11418 (gimple_mod_subtract_transform): Update.
11419 (gimple_ic): Update.
11420 (gimple_stringop_fixed_value): Update.
11421 (gimple_stringops_transform): Update.
11422 * value-prof.h: Update.
11424 2017-06-29 Carl Love <cel@us.ibm.com>
11426 * config/rs6000/rs6000-c.c: Add support for built-in functions
11427 vector signed int vec_signed (vector float);
11428 vector signed long long vec_signed (vector double);
11429 vector signed int vec_signed2 (vector double, vector double);
11430 vector signed int vec_signede (vector double);
11431 vector signed int vec_signedo (vector double);
11432 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
11433 instruction generator.
11434 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11435 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
11436 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
11438 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
11439 vunsignede_v2df): Add define_expands.
11440 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
11441 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
11442 VEC_UNSIGNEDO): Add definitions.
11443 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
11444 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
11445 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
11446 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
11447 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
11448 * config/rs6000/altivec.h (vec_signed, vec_signed2,
11449 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
11450 vec_unsignede, vec_unsignedo): Add builtin defines.
11451 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
11453 * doc/extend.texi: Update the built-in documentation file for the
11454 new built-in functions.
11456 2017-06-29 Richard Biener <rguenther@suse.de>
11458 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
11459 reduction chains to LOOP_VINFO_REDUCTIONS.
11460 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
11461 SLP reductions after processing reduction chains.
11463 2017-06-29 Nathan Sidwell <nathan@acm.org>
11465 * builtins.c (fold_builtin_FUNCTION): Use
11466 lang_hooks.decl_printable_name.
11468 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
11470 PR middle-end/81194
11471 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
11472 with only one label.
11473 * stmt.c (expand_case): Assert NCASES is greater than one.
11475 2017-06-29 Richard Biener <rguenther@suse.de>
11477 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
11479 (group_case_labels): Likewise.
11480 (find_taken_edge): Push sanity checking on val to workers...
11481 (find_taken_edge_cond_expr): ... here
11482 (find_taken_edge_switch_expr): ... and here, handle cases
11483 with just a default label.
11484 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
11485 (group_case_labels): Likewise.
11486 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
11487 group_case_labels does anything cleanup the CFG again.
11489 2017-06-29 Bin Cheng <bin.cheng@arm.com>
11491 PR tree-optimization/81196
11492 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
11493 exit condition comparing two IVs.
11495 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
11497 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
11498 profile to the dummy entry at the end of the list of architectures.
11499 * config/arm/arm-cpu-cdata.h: Regenerated.
11501 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11502 Michael Collison <michael.collison@arm.com>
11505 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
11507 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
11508 (*aarch64_reg_<mode>3_minus_mask): New pattern.
11509 (*aarch64_<optab>_reg_di3_mask2): New pattern.
11510 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
11511 of shift when the shift amount is masked with constant equal to
11512 the size of the mode.
11513 * config/aarch64/predicates.md (subreg_lowpart_operator): New
11516 2017-06-29 Martin Liska <mliska@suse.cz>
11518 * config/i386/i386.opt: Change range from [1,5] to [0,5].
11520 2017-06-29 Yury Gribov <tetra2005@gmail.com>
11523 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
11526 (edge_growth_cache_entry::edge_growth_cache_entry): New
11528 (reset_edge_growth_cache): Update to use constructor.
11530 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11532 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
11533 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11534 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
11536 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
11538 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
11539 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
11541 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
11543 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
11544 (*-linux-uclibc*): Add t-uclibc tmake_file.
11545 * config/t-musl: New.
11546 * config/t-uclibc: New.
11548 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
11550 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
11552 (gen_comm_data): Emit architectural setting of arch_prof.
11553 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
11555 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
11556 (armv8-m.base, armv8-m.main): Likewise.
11557 * arm-protos.h (arm_build_target): Add profile field.
11558 (arch_option): Likewise.
11559 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
11561 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
11562 arm_active_target.profile.
11564 2017-06-28 Richard Biener <rguenther@suse.de>
11566 PR middle-end/81227
11567 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
11568 TYPE_OVERFLOW_WRAPS.
11569 * match.pd (negate_expr_p): Likewise.
11570 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
11571 fold_build2, not fold_binary.
11573 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11575 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
11576 Convert memory address to Pmode.
11577 (aarch64_print_operand): Assert MEM operands are always Pmode.
11579 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
11582 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
11583 Remove redundant if.
11584 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
11585 * config/arm/aarch-common-protos.h
11586 (aarch_forward_to_shift_is_not_shifted_re): Remove.
11587 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
11589 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
11592 * multiple_target.c (create_dispatcher_calls): Set the default
11593 clone to be static, not public.
11595 2017-06-28 Richard Biener <rguenther@suse.de>
11597 * tree-vect-loop.c (vectorizable_reduction): Move special
11598 cond reduction IV var creation ...
11599 (vect_create_epilog_for_reduction): ... here. Remove induction_index
11600 parameter. Use STMT_VINFO_VECTYPE.
11601 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
11604 2017-06-28 Martin Liska <mliska@suse.cz>
11607 * ipa-visibility.c (non_local_p): Handle visibility.
11609 2017-06-28 Martin Liska <mliska@suse.cz>
11612 * common.opt: Add IntegerRange to various options.
11613 * opt-functions.awk (integer_range_info): New function.
11614 * optc-gen.awk: Add integer_range_info to cl_options struct.
11615 * opts-common.c (decode_cmdline_option): Handle
11616 CL_ERR_INT_RANGE_ARG.
11617 (cmdline_handle_error): Likewise.
11618 * opts.c (print_filtered_help): Show valid interval in
11619 when --help is provided.
11620 * opts.h (struct cl_option): Add range_min and range_max fields.
11621 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
11623 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
11625 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
11626 (x * C EQ/NE y * C): New transformation.
11628 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
11630 * genmultilib (combination_space): Accept '+' in option names.
11632 2017-06-28 Martin Liska <mliska@suse.cz>
11635 * asan.c (instrument_derefs): Bail out inner references
11636 that are hard register variables.
11638 2017-06-28 Jakub Jelinek <jakub@redhat.com>
11641 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
11642 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
11644 2017-06-28 Richard Biener <rguenther@suse.de>
11646 * tree-vectorizer.h (vect_get_vec_defs): Remove.
11647 (vect_get_slp_defs): Adjust.
11648 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
11650 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
11652 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11653 get_initial_defs_for_reduction instead of vect_get_vec_defs.
11654 (vectorizable_reduction): Adjust.
11655 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
11657 (vect_get_slp_defs): Likewise.
11658 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
11659 (vectorizable_bswap): Adjust.
11660 (vectorizable_call): Likewise.
11661 (vectorizable_conversion): Likewise.
11662 (vectorizable_assignment): Likewise.
11663 (vectorizable_shift): Likewise.
11664 (vectorizable_operation): Likewise.
11665 (vectorizable_store): Likewise.
11666 (vectorizable_condition): Likewise.
11667 (vectorizable_comparison): Likewise.
11669 2017-06-28 Michael Collison <michael.collison@arm.com>
11672 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
11674 (arm_gen_movmemqi): Removed unused variable 'i'.
11675 Convert 'for' loop into 'while' loop.
11676 (arm_expand_prologue): Remove last unnecessary set of insn.
11677 (thumb_pop): Remove unused variable 'pushed_words'.
11678 (thumb_exit): Remove last unnecessary set of regs_to_pop.
11680 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11682 * config/s390/predicates.md: Use s390_rel_address_ok_p.
11683 * config/s390/s390-protos.h: Add prototype of
11684 s390_rel_address_ok_p.
11685 * config/s390/s390.c (s390_got_symbol): New function.
11686 (s390_rel_address_ok_p): New function.
11687 (legitimize_pic_address): Use s390_rel_address_ok_p.
11688 (s390_load_got): Use s390_got_symbol.
11689 (s390_option_override): Issue error if
11690 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
11691 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
11693 * config/s390/s390.opt: New option mpic-data-is-text-relative.
11695 2017-06-27 Andrew Pinski <apinski@cavium.com>
11697 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
11698 (X * copysign (1.0, X)): New pattern.
11699 (X * copysign (1.0, -X)): New pattern.
11700 (copysign (-1.0, CST)): New pattern.
11702 2017-06-27 Joseph Myers <joseph@codesourcery.com>
11704 * genmultilib (combination_space): Remove variable.
11705 Validate reuse rules against regular expression for any sequence
11706 of multilib options in any order.
11708 2017-06-27 Michael Collison <michael.collison@arm.com>
11710 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
11711 call aarch64_split_simd_combine.
11712 * (aarch64_combine_internal<mode>): Delete pattern.
11713 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
11714 Allow register and subreg operands.
11716 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11718 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
11719 specific need, just fallback on defaults.
11720 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
11722 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11723 Olivier Hainque <hainque@adacore.com>
11725 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
11727 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
11728 targets. Pick a default if no particular attempt applied.
11729 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
11732 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11734 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
11735 (x86_64-wrs-vxworks7): Likewise.
11737 2017-06-27 Marek Polacek <polacek@redhat.com>
11740 * ubsan.c (instrument_null): Check get_base_address's result for null.
11742 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11744 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
11746 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
11748 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
11749 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
11750 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
11751 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
11752 New function types.
11753 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
11754 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
11755 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
11756 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
11757 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
11758 BUILT_IN_FEUPDATEENV): New builtins.
11759 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
11760 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
11761 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
11762 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
11764 (builtin_structptr_types): Adjust size.
11765 * tree.c (builtin_structptr_types): Add four entries.
11767 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11768 Olivier Hainque <hainque@adacore.com>
11770 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
11771 (TLS_SYM): New local macro, forcing reference to __tls__ on
11772 link command lines for VxWorks 7 RTPs, triggering initialization
11774 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
11775 OS features TLS support, true for RTPs on VxWorks 7.
11776 * config/vxworks.c (vxworks_override_options): Setup emutls
11779 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11781 * predict.c (test_prediction_value_range): Use -1U instead of -1
11782 to avoid narrowing conversion warning.
11783 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
11784 to avoid narrowing conversion warning.
11785 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
11787 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
11789 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
11791 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
11792 64bit configurations.
11793 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
11794 (SIZE_TYPE): Likewise.
11795 * config/vxworks.c (vxworks_emutls_var_fields): Use
11796 long_unsigned_type_node instead of unsigned_type_node as the offset
11797 field type, which is "pointer" mode in emutls.c.
11799 2017-06-27 Jakub Jelinek <jakub@redhat.com>
11802 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
11804 PR middle-end/81207
11805 * gimple-fold.c (replace_call_with_call_and_fold): Handle
11806 gimple_vuse copying separately from gimple_vdef copying.
11808 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11810 * value-prof.c (free_hist): Remove call to memset and the enclosing if
11813 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11814 Olivier Hainque <hainque@adacore.com>
11816 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
11817 for all vxworks7 targets.
11818 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
11819 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
11820 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
11821 variations for VX6/VX7 and 32/64bits later on in ...
11822 (VXWORKS_LIB_SPEC): Leverage new macros.
11823 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
11824 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
11826 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
11828 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
11829 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
11831 2017-06-26 Carl Love <cel@us.ibm.com>
11833 * config/rs6000/rs6000-c.c: Add support for built-in functions
11834 vector bool char vec_reve (vector bool char);
11835 vector signed char vec_reve (vector signed char);
11836 vector unsigned char vec_reve (vector unsigned char);
11837 vector bool int vec_reve (vector bool int);
11838 vector signed int vec_reve (vector signed int);
11839 vector unsigned int vec_reve (vector unsigned int);
11840 vector bool long long vec_reve (vector bool long long);
11841 vector signed long long vec_reve (vector signed long long);
11842 vector unsigned long long vec_reve (vector unsigned long long);
11843 vector bool short vec_reve (vector bool short);
11844 vector signed short vec_reve (vector signed short);
11845 vector double vec_reve (vector double);
11846 vector float vec_reve (vector float);
11847 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
11848 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
11849 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
11850 (altivec_vreve): New pattern.
11851 * config/rs6000/altivec.h (vec_reve): New define.
11852 * doc/extend.texi (vec_rev): Update the built-in documentation file
11853 for the new built-in functions.
11855 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11857 PR tree-optimization/71815
11858 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
11860 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
11862 (slsr_process_phi): Likewise.
11863 (replace_uncond_cands_and_profitable_phis): Don't replace a
11864 multiply candidate with a stride of 1 (copy or cast).
11865 (phi_incr_cost): Call uses_consumed_by_stmt rather than
11867 (lowest_cost_path): Likewise.
11868 (total_savings): Likewise.
11870 2017-06-26 Richard Biener <rguenther@suse.de>
11873 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
11874 Use def_builtin_pure for all gather builtins.
11876 2017-06-26 Richard Biener <rguenther@suse.de>
11878 PR tree-optimization/81203
11879 * tree-tailcall.c (find_tail_calls): Do not move stmts into
11880 non-dominating BBs.
11882 2017-06-26 Marek Polacek <polacek@redhat.com>
11885 * doc/invoke.texi: Document -Wmultistatement-macros.
11887 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
11889 * doc/sourcebuild.texi (ARM-specific attributes): Document new
11890 arm_neon_ok_no_float_abi effective target.
11892 2017-06-26 Richard Biener <rguenther@suse.de>
11894 PR tree-optimization/80928
11895 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
11896 (copy_bbs): Set BB_DUPLICATED flag early.
11897 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
11899 (execute_on_shrinking_pred): Likewise.
11900 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
11901 BB_DUPLICATED blocks.
11902 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
11903 iterate over all PHIs considering removal of *gsi.
11905 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
11907 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
11910 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11912 * config/rs6000/rs6000-string.c: (expand_block_clear,
11913 do_load_for_compare, select_block_compare_mode,
11914 compute_current_alignment, expand_block_compare,
11915 expand_strncmp_align_check, expand_strn_compare,
11916 expand_block_move, rs6000_output_load_multiple)
11917 Move functions related to string/block move/compare
11918 to a separate file.
11919 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
11920 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
11921 for this function which is now used in two files.
11922 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
11923 * config.gcc: Add rs6000-string.o to extra_objs for
11924 targets powerpc*-*-* and rs6000*-*-*.
11926 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
11929 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
11930 32-bit, since indexed is not valid for DImode.
11931 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
11932 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
11933 (define_peephole2 for Altivec d-form load): Add 32-bit support.
11934 (define_peephole2 for Altivec d-form store): Likewise.
11937 * multiple_target.c (create_dispatcher_calls): Only create the
11938 dispatcher call if the function is the default clone of a
11939 versioned function.
11941 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
11943 PR middle-end/80902
11944 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
11945 a call, force the call to not be a tail call.
11947 2017-06-23 Jeff Law <law@redhat.com>
11949 * doc/contrib.texi: Add entry for Steven Pemberton's work on
11952 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
11954 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
11955 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
11956 handling for early expansion of vector shifts (sl,sr,sra,rl).
11957 (builtin_function_type): Add vector shift right instructions
11958 to the unsigned argument list.
11960 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11962 rtl-optimizatoin/79286
11963 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
11964 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
11965 trap. PIC register plus a const unspec without offset can never trap.
11967 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
11969 * tree.h (builtin_structptr_type): New type.
11970 (builtin_structptr_types): Declare new array.
11971 * tree.c (builtin_structptr_types): New array.
11972 (free_lang_data, build_common_tree_nodes): Use it.
11974 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
11977 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
11980 2017-06-22 Matt Turner <mattst88@gmail.com>
11982 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
11983 Lake models to skylake case. Assume skylake for unknown
11984 models with clflushopt.
11986 2017-06-22 Jeff Law <law@redhat.com>
11988 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
11989 frame sizes that do not satisfy aarch64_uimm12_shift.
11991 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
11993 * profile-count.h (apply_probability,
11994 apply_scale, probability_in): Fix checks for zero.
11996 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11998 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
11999 * doc/cppdiropts.texi (-I @var{dir}): Document it.
12001 2016-06-22 Richard Biener <rguenther@suse.de>
12003 * tree-vect-loop.c (vect_model_reduction_cost): Handle
12004 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
12005 REDUC_MAX_EXPR support.
12006 (vectorizable_reduction): Likewise.
12007 (vect_create_epilog_for_reduction): Likewise.
12009 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
12011 * match.pd (A / (1 << B) -> A >> B): New.
12012 * generic-match-head.c: Include optabs-tree.h.
12013 * gimple-match-head.c: Likewise.
12014 * optabs-tree.h (target_supports_op_p): New.
12015 * optabs-tree.c (target_supports_op_p): New.
12017 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12019 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
12020 $gcc_cv_ld --help output.
12021 (gcc_cv_ld_demangle): Likewise.
12022 (gcc_cv_ld_eh_frame_hdr): Likewise.
12023 (gcc_cv_ld_pie): Likewise.
12024 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
12025 (gcc_cv_ld_buildid): Likewise.
12026 (gcc_cv_ld_sysroot): Likewise.
12027 (ld_bndplt_support): Likewise.
12028 (ld_pushpopstate_support): Likewise.
12029 * configure: Regenerate.
12030 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
12032 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12035 * config/i386/sse.md (round<mode>2): Renumber match_dup and
12036 operands indexes to avoid gap between operands and match_dups.
12038 2017-06-21 Andrew Pinski <apinski@cavium.com>
12040 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
12041 Increment Arith_shift and Arith_shift_reg by 1.
12042 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
12044 * config/aarch64/aarch64.c (thunderx_tunings): Enable
12045 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
12046 (aarch64_strip_extend): Add new argument and test for it.
12047 (aarch64_cheap_mult_shift_p): New function.
12048 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
12049 add a cost if it is true.
12050 Update calls to aarch64_strip_extend.
12051 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
12053 2017-06-21 Andrew Pinski <apinski@cavium.com>
12055 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
12057 (thunderxt88): Likewise.
12058 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
12059 (thunderx_prefetch_tune): New variable.
12060 (thunderx2t99_prefetch_tune): Update for the correct values.
12061 (thunderxt88_tunings): New variable.
12062 (thunderx_tunings): Use thunderx_prefetch_tune instead of
12063 generic_prefetch_tune.
12064 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
12066 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12068 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
12069 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
12070 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
12071 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
12072 (aarch64_atomic_cas<mode>, GPI): Likewise.
12074 2017-06-21 Martin Liska <mliska@suse.cz>
12076 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
12077 statements on cold and hot labels.
12078 * predict.c (tree_estimate_probability_bb): Remove the
12079 prediction from this place.
12081 2017-06-21 Martin Liska <mliska@suse.cz>
12083 PR tree-optimization/79489
12084 * gimplify.c (maybe_add_early_return_predict_stmt): New
12086 (gimplify_return_expr): Call the function.
12087 * predict.c (tree_estimate_probability_bb): Remove handling
12089 * predict.def: Update comment about early return predictor.
12090 * gimple-predict.h (is_gimple_predict): New function.
12091 * predict.def: Change default value of early return to 66.
12092 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
12094 * passes.def: Put pass_strip_predict_hints to the beginning of
12097 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
12099 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
12100 FUNCTION_DECL declarations.
12101 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
12103 (dwaf2out_decl): Likewise.
12104 * godump.c (go_early_global_decl): Skip call to the real debug hook
12105 for FUNCTION_DECL declarations.
12106 * passes.c (rest_of_decl_compilation): Skip call to the
12107 early_global_decl debug hook for FUNCTION_DECL declarations, unless
12108 -fdump-go-spec is passed.
12110 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12112 * config/i386/i386.c (struct builtin_isa): New field pure_p.
12113 Reorder for compactness.
12114 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
12115 (def_builtin_pure, def_builtin_pure2): New functions.
12116 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
12118 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
12120 * match.pd (nop_convert): New predicate.
12121 ((A +- CST1) +- CST2): Allow some NOP conversions.
12123 2017-06-21 Jakub Jelinek <jakub@redhat.com>
12126 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
12127 with ctors/dtors if GOVD_SHARED is set.
12129 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12131 * config/aarch64/aarch64.md (movti_aarch64):
12132 Emit mov rather than orr.
12133 (movtf_aarch64): Likewise.
12134 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
12135 Emit mov rather than orr.
12137 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12139 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
12140 Swap alternatives, make integer dup more expensive.
12142 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
12144 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
12145 Return true for non-tls symbols.
12147 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
12149 * config/aarch64/aarch64-cores.def (cortex-a55): New.
12150 (cortex-a75): Likewise.
12151 (cortex-a75.cortex-a55): Likewise.
12152 * config/aarch64/aarch64-tune.md: Regenerate.
12153 * doc/invoke.texi (-mtune): Document new values for -mtune.
12155 2017-06-21 Tom de Vries <tom@codesourcery.com>
12157 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
12158 stack_size feature.
12159 (Effective-Target Keywords, Other attributes): Suggest using
12160 dg-add-options stack_size feature to get stack limit in stack_size
12161 effective target documentation.
12163 2017-06-21 Julian Brown <julian@codesourcery.com>
12164 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12166 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
12167 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
12168 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
12170 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
12171 attribute type list for neon_multiply.
12172 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
12173 attribute type list for neon_multiply.
12174 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
12175 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
12176 attribute type list for neon_multiply.
12177 * config/arm/types.md (crypto_pmull): Add.
12178 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
12179 attribute type list.
12181 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
12183 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
12186 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12188 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
12189 to make sure not to dereference a NULL cost_classes_ptr pointer.
12191 2017-06-20 Carl Love <cel@us.ibm.com>
12193 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12194 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12195 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12196 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12197 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12198 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12199 VMULOSW): New enum "unspec" values.
12200 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12201 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12202 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12203 altivec_vmulosw): New patterns.
12204 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12205 VMULOSW): Add definitions.
12207 2017-06-20 Julia Koval <julia.koval@intel.com>
12209 * config/i386/i386.c: Fix rounding expand for new pattern.
12210 * config/i386/subst.md: Fix pattern (parallel -> unspec).
12212 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12214 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
12215 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
12217 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12219 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
12222 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
12224 * config/aarch64/aarch64-cores.def: Rearrange to sort by
12225 architecture, then by implementer ID.
12226 * config/aarch64/aarch64-tune.md: Regenerate.
12228 2017-06-20 Richard Biener <rguenther@suse.de>
12230 PR middle-end/81097
12231 * fold-const.c (split_tree): Fold to type before negating.
12233 2017-06-20 David Malcolm <dmalcolm@redhat.com>
12235 * diagnostic-show-locus.c
12236 (selftest::test_fixit_deletion_affecting_newline): New function.
12237 (selftest::diagnostic_show_locus_c_tests): Call it.
12239 2017-06-20 Andreas Schwab <schwab@suse.de>
12242 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
12245 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
12247 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
12248 __ARM_FEATURE_COPROC according to support.
12250 2017-06-20 Jakub Jelinek <jakub@redhat.com>
12252 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12253 Rewritten to avoid overflow for > 32-bit pointers.
12256 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
12257 by removing enum keyword.
12258 (ubsan_type_descriptor): Likewise. Formatting fix.
12261 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
12262 splitter): Require TARGET_SSE2 in the condition.
12264 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
12267 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
12268 for doing vector set of SFmode on ISA 3.0.
12269 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
12270 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
12272 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
12273 SFmode value into a V4SF variable that was extracted from another
12274 V4SF variable without converting the element to double precision
12275 and back to single precision vector format.
12276 (vsx_insert_extract_v4sf_p9_2): Likewise.
12278 2017-06-19 Jakub Jelinek <jakub@redhat.com>
12280 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
12281 in UWHI to avoid undefined overflow.
12284 * ubsan.h (enum ubsan_encode_value_phase): New.
12285 (ubsan_encode_value): Change second argument to
12286 enum ubsan_encode_value_phase with default value of
12287 UBSAN_ENCODE_VALUE_GENERIC.
12288 * ubsan.c (ubsan_encode_value): Change second argument to
12289 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
12290 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
12291 create_tmp_var_raw instead of create_tmp_var and use a
12293 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
12294 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
12295 ubsan_encode_value callers.
12298 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
12299 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
12300 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
12302 2017-06-19 Richard Biener <rguenther@suse.de>
12304 PR middle-end/81118
12305 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
12306 estimates if we changed anything.
12308 2017-06-19 Richard Biener <rguenther@suse.de>
12310 PR tree-optimization/80887
12311 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
12312 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
12313 simplified lookups, then reset mprts_hook.
12314 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
12316 (try_to_simplify): Likewise.
12318 2017-06-19 Martin Liska <mliska@suse.cz>
12321 * gimplify.c (gimplify_switch_expr):
12322 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
12324 2017-06-19 Martin Liska <mliska@suse.cz>
12326 * doc/install.texi: Document that PGO runs in 4 stages.
12328 2017-06-19 Martin Liska <mliska@suse.cz>
12331 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
12332 to dispatcher function name.
12333 * multiple_target.c (replace_function_decl): New function.
12334 (create_dispatcher_calls): Redirect both edges and references.
12336 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
12338 * profile-count.c (profile_count::dump): Dump quality.
12339 (profile_count::differs_from_p): Update for unsigned val.
12340 * profile-count.h (profile_count_quality): New enum.
12341 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
12343 2017-06-19 Richard Biener <rguenther@suse.de>
12345 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
12346 struct function as arg.
12347 (estimate_numbers_of_iterations): Export overload with loop arg.
12348 (free_numbers_of_iterations_estimates_loop): Use an overload of
12349 free_numbers_of_iterations_estimates instead.
12350 * tree-cfg.c (remove_bb): Adjust.
12351 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
12352 * tree-parloops.c (gen_parallel_loop): Likewise.
12353 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
12355 (tree_unroll_loops_completely): Likewise.
12356 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
12357 Use an overload instead and export.
12358 (estimated_loop_iterations): Adjust.
12359 (max_loop_iterations): Likewise.
12360 (likely_max_loop_iterations): Likewise.
12361 (estimate_numbers_of_iterations): Take struct function as arg
12363 (loop_exits_before_overflow): Adjust.
12364 (free_numbers_of_iterations_estimates_loop): Use an overload.
12365 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
12366 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
12368 2017-06-19 Richard Biener <rguenther@suse.de>
12371 * ipa-prop.c (find_constructor_constant_at_offset): Handle
12372 RANGE_EXPR conservatively.
12374 2017-06-16 Carl Love <cel@us.ibm.com>
12376 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12377 definitions for vec_float, vec_float2, vec_floato,
12378 vec_floate built-ins.
12379 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
12380 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
12382 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
12383 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
12384 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
12385 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
12386 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
12387 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
12388 vec_floato): Add builtin defines.
12389 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
12390 Update the built-in documentation file for the new built-in
12393 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12395 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
12396 (mthumb): Mark as the negative of -marm.
12398 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12400 * doc/invoke.texi (ARM Options, -mcpu): Document supported
12402 (ARM Options, -mtune): Document that this accepts the same
12403 extension options as -mcpu.
12404 (ARM Options, -mfpu): Document addition of -mfpu=auto.
12406 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12408 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
12409 permitted extensions.
12411 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12413 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
12414 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
12415 (armv8-m.main): Add option +nodsp.
12416 * config/arm/arm-cpu-cdata.h: Regenerated.
12418 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12420 * config/arm/t-fuchsia: New file.
12421 * config.gcc (arm*-*-fuchsia*): Use it.
12423 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12425 * config/arm/t-symbian: Rewrite for new option infrastructure.
12427 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12429 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
12430 (MULTILIB_REQUIRED): Likewise.
12432 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12434 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
12435 (MULTILIB_RESUE): Likewise.
12436 (MULTILIB_MATCHES): Likewise.
12437 (MULTLIB_REQUIRED): Likewise.
12439 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12441 * config/arm/t-rtems: Rewrite for new option framework.
12443 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12445 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
12446 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
12447 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
12448 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
12449 * config/arm/t-multilib: ... here.
12450 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
12451 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
12452 armv7-a and armv8*-a when A-profile libraries have not been built.
12453 * config/arm/t-rmprofile: Rewrite.
12455 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12457 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
12458 with a backslash. Remove the backslash after substituting unescaped
12460 * doc/fragments.texi (MULTILIB_REUSE): Document it.
12462 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12464 * config.gcc: (arm*-*-*): When building a-profile libraries, force
12465 the driver to pass through the default setting of -mfloat-abi.
12466 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
12468 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
12469 (all_feat_combs): New rule.
12470 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
12472 * config/arm/t-aprofile: Rewrite.
12474 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12476 * config/arm/arm.h (FPUTYPE_AUTO): Define.
12477 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
12478 fpu is not specified by the user/command-line.
12479 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
12480 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
12481 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
12482 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
12483 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
12484 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
12486 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12488 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
12489 * config/arm/t-arm-elf: Rewritten.
12491 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12493 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
12494 have some floating-point instructions.
12495 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
12496 (TARGET_MAYBE_HARD_FLOAT): New macro.
12497 * config/arm/arm-builtins.c (arm_init_builtins): Use
12498 TARGET_MAYBE_HARD_FLOAT.
12499 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
12501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12503 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
12504 (configargs.h): Include it.
12505 (arm_print_hint_for_fpu_option): New function.
12506 (arm_parse_fpu_option): New function.
12507 (candidate_extension): New class.
12508 (arm_canon_for_multilib): New function.
12509 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
12510 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
12511 (ARCH_CANONICAL_SPECS): New macro.
12512 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
12514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12516 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
12517 are set after handling multilib fragments. Set target_cpu_default2
12520 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12522 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
12524 (arm*-*-*): Set target_cpu_default2 to a quoted string.
12525 * config/arm/parsecpu.awk (check_cpu): Validate any extension
12527 (check_arch): Likewise.
12528 * config/arm/arm.c (arm_configure_build_target): Handle
12529 TARGET_CPU_DEFAULT being a string constant. Scan any feature
12530 options in the default.
12532 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12534 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
12535 when an option is an alias of another.
12536 * config/arm/parsecpu.awk (optalias): New parser token.
12537 (gen_comm_data): Mark non-alias options as such. Emit entries
12538 for extension aliases.
12539 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
12540 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
12541 (armv6kz, armv6zk, armv6t2): Likewise.
12542 (armv7): Make vfpv3-d16 an alias.
12543 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
12545 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
12546 Sort in canonical order.
12547 (armv8-a): Sort in canonical order.
12548 (armv8.1-a, armv8.2-a): Likewise.
12549 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
12551 (cortex-a9): Sort in canonical order.
12552 * config/arm/arm.c (selftests.h): Include it.
12553 (arm_test_cpu_arch_data): New function.
12554 (arm_run_self_tests): New function.
12555 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
12556 (targetm): Move declaration to the end of the file.
12557 * arm-cpu-cdata.h: Regenerated.
12559 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12561 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
12562 call to target_mode_check describing the type of option passed.
12563 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
12564 (arm_target_thumb_only): Use arm_parse_arch_option_name or
12565 arm_parse_cpu_option_name to match parameters against list of
12567 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
12568 arm_arch_core_flags data structure.
12569 * config/arm/arm-cpu_cdata.h: Regenerated.
12571 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12573 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
12575 (arm_print_hint_for_cpu_option): Likewise.
12576 (arm_print_hint_for_arch_option): Likewise.
12577 (arm_parse_cpu_option_name): Likewise.
12578 (arm_parse_arch_option_name): Likewise.
12579 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
12580 of entries in the all_fpus list.
12581 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
12582 (arm_parse_cpu_option_name): Declare.
12583 (arm_parse_arch_option_name): Declare.
12584 (arm_parse_option_features): Declare.
12585 (arm_intialize_isa): Declare.
12586 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
12588 (gen_comm_data): ... here. Make definitions non-static.
12589 * config/arm/arm-cpu-data.h: Regenerated.
12590 * config/arm/arm-cpu-cdata.h: Regenerated.
12592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12594 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
12595 (cpu_arch_extension): New structure.
12596 (cpu_arch_option, arch_option, cpu_option): New structures.
12597 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
12598 architecture types.
12599 (gen_data): Generate new format data tables.
12600 * config/arm/arm.c (cpu_tune): New structure.
12601 (cpu_option, processors): Delete.
12602 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
12603 (arm_print_hint_for_cpu_option): ... this and ...
12604 (arm_print_hint_for_arch_option): ... this.
12605 (arm_parse_arch_cpu_name): Delete. Replace with ...
12606 (arm_parse_cpu_option_name): ... this and ...
12607 (arm_parse_arch_option_name): ... this.
12608 (arm_unrecognized_feature): Change type of target parameter to
12610 (arm_parse_arch_cpu_features): Delete. Replace with ...
12611 (arm_parse_option_features): ... this.
12612 (arm_configure_build_target): Rework to use new configuration data
12614 (arm_print_tune_info): Rework for new configuration data tables.
12615 * config/arm/arm-cpu-data.h: Regenerated.
12616 * config/arm/arm-cpu.h: Regenerated.
12618 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12620 * Makefile.in (OBJS): Move sbitmap.o from here ...
12621 (OBJS-libcommon): ... to here.
12623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12625 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
12626 (ISA_ALL_CRYPTO): New macro.
12627 (ISA_ALL_SIMD): New macro
12628 (ISA_ALL_FP): New macro.
12629 * config/arm/arm.c (fpu_bitlist): Update initializer.
12630 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
12632 (arm9e): Add fpu. Add option for nofp
12633 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
12634 (arm926ej-s, arm1026ej-s): Likewise.
12635 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
12636 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
12637 neon-fp16, neon-vfpv4, nofp and nosimd.
12638 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
12639 (cortex-a8): Add fpu. Add option for nofp.
12640 (cortex-a9): Add fpu. Add options for nosimd and nofp.
12641 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
12642 (cortex-r4f): Add fpu.
12643 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
12644 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
12646 (cortex-r8): Likewise.
12647 (cortex-m4): Add fpu. Add option for nofp.
12648 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
12649 (cortex-a17.cortex-a7): Likewise.
12650 (cortex-a32): Add fpu. Add options for crypto and nofp.
12651 (cortex-a35, cortex-a53): Likewise.
12652 (cortex-a57): Add fpu. Add option for crypto.
12653 (cortex-a72, cortex-a73): Likewise.
12654 (exynos-m1): Likewise.
12655 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
12656 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
12657 (cortex-m33): Add fpu. Add option for nofp.
12658 * config/arm/arm-cpu-cdata.h: Regenerated
12659 * config/arm/arm-cpu-data.h: Regenerated.
12661 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12663 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
12664 (armv5te, armv5tej): Likewise.
12665 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
12666 (armv7): Add options fp and vfpv3-d16.
12667 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
12668 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
12670 (armv7ve): Likewise.
12671 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
12672 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
12673 (armv8-a): Add nocrypto option.
12674 (armv8.1-a, armv8.2-a): Likewise.
12675 (armv8-m.main): add options fp, fp.dp and nofp.
12677 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12679 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
12681 (armv8-a+crc): Delete.
12682 (armv8.1-a): Add options simd, crypto and nofp.
12683 (armv8.2-a): Add options fp16, simd, crypto and nofp.
12684 (armv8.2-a+fp16): Delete.
12685 (armv8-m.main): Add option dsp.
12686 (armv8-m.main+dsp): Delete.
12687 (cortex-a8): Add fpu. Add nofp option.
12688 (cortex-a9): Add fpu. Add nofp and nosimd options.
12689 * config/arm/parsecpu.awk (gen_data): Generate option tables and
12690 link to main cpu and architecture data structures.
12691 (gen_comm_data): Only put isa attributes from the main architecture
12693 (option): New statement for architecture and CPU entries.
12694 * arm.c (struct cpu_option): New structure.
12695 (struct processors): Add entry for options.
12696 (arm_unrecognized_feature): New function.
12697 (arm_parse_arch_cpu_name): Ignore any characters after the first
12699 (arm_parse_arch_cpu_feature): New function.
12700 (arm_configure_build_target): Separate out any CPU and architecture
12701 features and parse separately. Don't error out if -mfpu=auto is
12702 used with only an architecture string.
12703 (arm_print_asm_arch_directives): New function.
12704 (arm_file_start): Call it.
12705 * config/arm/arm-cpu-cdata.h: Regenerated.
12706 * config/arm/arm-cpu-data.h: Likewise.
12707 * config/arm/arm-tables.opt: Likewise.
12709 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12711 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
12712 assembler when it is not -mfpu=auto.
12714 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12716 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
12717 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
12718 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
12719 (ASM_CPU_SPEC): Rewrite.
12720 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
12721 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
12722 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
12723 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
12724 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
12725 copied string is NUL-terminated. Also strip any characters prefixed
12727 (arm_rewrite_selected_arch): New function.
12728 (arm_rewrite_march): New function.
12730 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
12732 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
12733 (x_arm_cpu_string, x_arm_tune_string): Likewise.
12734 (march, mcpu, mtune): Convert to string-based options.
12735 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
12736 (arm_parse_arch_cpu_name): New function.
12737 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
12738 identify selected architecture or CPU.
12739 (arm_option_save): New function.
12740 (TARGET_OPTION_SAVE): Redefine.
12741 (arm_option_restore): Restore string options.
12742 (arm_option_print): Print string options.
12744 2017-06-16 Martin Sebor <msebor@redhat.com>
12746 PR tree-optimization/80933
12747 PR tree-optimization/80934
12748 * builtins.c (fold_builtin_3): Do not handle bcmp here.
12749 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
12750 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
12751 (gimple_fold_builtin): Call them.
12753 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12755 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
12756 as unlikely; update profile.
12758 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12760 * predict.c (force_edge_cold): Handle declaring edges impossible
12763 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12765 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
12767 (try_unroll_loop_completely): Fix reporting.
12769 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12771 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
12773 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
12776 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
12777 if given a non-constant argument for an intrinsic which requires a
12780 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12782 * profile.c (compare_freqs): New function.
12783 (branch_prob): Sort edge list.
12784 (find_spanning_tree): Assume that the list is priority sorted.
12786 2017-06-16 Richard Biener <rguenther@suse.de>
12788 PR tree-optimization/81090
12789 * passes.def (pass_record_bounds): Remove.
12790 * tree-pass.h (make_pass_record_bounds): Likewise.
12791 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
12792 make_pass_record_bounds): Likewise.
12793 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
12794 not free niter estimates at the beginning but at the end.
12795 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
12797 2017-06-16 Richard Biener <rguenther@suse.de>
12799 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
12800 initializer to workaround ICE in host GCC 4.8.
12802 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
12804 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
12806 (clone_inlined_nodes): Update.
12808 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12810 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
12811 prefetch settings, and enable prefetching by default at -O3.
12813 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12815 * config/aarch64/aarch64.c (aarch64_override_options_internal):
12816 Set flag_prefetch_loop_arrays according to tuning data.
12818 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12820 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
12821 New tune structure.
12822 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
12823 [Unrelated to main purpose of the patch] Place the pointer field last
12824 to enable type checking errors when tune structure are wrongly merged.
12825 * config/aarch64/aarch64.c (generic_prefetch_tune,)
12826 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
12827 (thunderx2t99_prefetch_tune): New tune constants.
12828 (tune_params *_tunings): Update all tunings (no functional change).
12829 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
12830 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
12831 from tunings structures.
12833 2017-06-16 Jakub Jelinek <jakub@redhat.com>
12836 * ubsan.c (instrument_null): Add T argument, use it instead
12837 of computing it based on IS_LHS.
12838 (instrument_object_size): Likewise.
12839 (pass_ubsan::execute): Adjust instrument_null and
12840 instrument_object_size callers to pass gimple_get_lhs or
12841 gimple_assign_rhs1 result to it. Use instrument_null instead of
12842 calling get_base_address and instrument_mem_ref. Handle
12843 aggregate call arguments for object-size sanitization.
12845 2017-06-16 Yury Gribov <tetra2005@gmail.com>
12847 PR tree-optimization/81089
12848 * tree-vrp.c (is_masked_range_test): Validate operands of
12851 2017-06-15 Martin Sebor <msebor@redhat.com>
12854 * dumpfile.c (dump_register): Avoid calling memset to initialize
12855 a class with a default ctor.
12856 * gcc.c (struct compiler): Remove const qualification.
12857 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
12858 * hash-table.h: Ditto.
12859 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
12861 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
12862 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
12864 * params.h (struct param_info): Make struct members non-const.
12865 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
12866 with default initialization.
12867 * vec.h (vec_copy_construct, vec_default_construct): New helper
12869 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
12870 with vec_copy_construct.
12871 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
12872 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
12873 * doc/invoke.texi (-Wclass-memaccess): Document.
12875 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12877 * emit-rtl.h (is_leaf): Update comment about local
12878 register allocator.
12880 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12883 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
12884 for a variable to have a section before checking if the section has a
12886 Set section to.persistent if persistent attribute is set.
12887 Warn if .persistent attribute is used on an automatic variable.
12889 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
12891 PR rtl-optimization/80474
12892 * reorg.c (update_block): Do not ignore instructions in a delay slot.
12894 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
12896 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
12899 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
12901 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
12902 (casesi): Emit bounds checking as RTL.
12903 (casesi_internal_mips16_<mode>): Remove bounds checking.
12905 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
12907 * config/xtensa/xtensa.c (xtensa_option_override): Append
12908 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
12909 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
12910 xtensa_doloop_hooks): Define unconditionally.
12911 (xtensa_reorg_loops): Only call reorg_loops in the presence of
12913 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
12914 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
12915 for it in xtensa_option_override.
12916 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
12917 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
12919 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
12921 * doc/cppopts.texi: Document '-' special value to -MF.
12923 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
12925 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
12926 (cortex_a53_fconst): Likewise.
12927 (cortex_a53_fpmul): Likewise.
12928 (cortex_a53_f_load_64): Likewise.
12929 (cortex_a53_f_load_many): Likewise.
12930 (cortex_a53_advsimd_alu): Likewise.
12931 (cortex_a53_advsimd_alu_q): Likewise.
12932 (cortex_a53_advsimd_mul): Likewise.
12933 (cortex_a53_advsimd_mul_q): Likewise.
12934 (fpmac bypass): Add new bypass for fpmac-fpmac case.
12935 Add missing fmul, r2f_cvt and fconst cases.
12937 2017-06-14 Richard Biener <rguenther@suse.de>
12939 PR middle-end/81088
12940 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
12942 (fold_binary_loc): When associating do not treat pre-existing
12943 TREE_OVERFLOW on literal constants as a reason to allow
12944 TREE_OVERFLOW on associated literal constants.
12946 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12948 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
12949 (MASK_FEATURES): New macro.
12950 * config/sparc/sparc.c (sparc_option_override): Remove the special
12951 handling of -mfpu and generalize it to all MASK_FEATURES switches.
12953 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12955 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
12956 a division of 0 if non-call exceptions are enabled.
12958 2017-06-14 Andrew Pinski <apinski@cavium.com>
12959 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12962 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
12963 Improve vector initialization code gen for only variable case.
12965 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
12967 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
12969 2017-06-14 Richard Biener <rguenther@suse.de>
12971 PR tree-optimization/81083
12972 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
12975 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12977 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
12978 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
12979 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
12980 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
12981 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
12982 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
12984 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12986 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
12987 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
12989 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12991 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
12993 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12995 * config/rs6000/t-rtems: Don't handle SPE.
12997 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
12999 * config/rs6000/t-linux: Don't handle SPE.
13001 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13003 * config/rs6000/eabispe.h: Delete file.
13005 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13007 * config/rs6000/t-spe: Delete file.
13009 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13011 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
13012 (rs6000_legitimate_offset_address_p): Return false for anything in
13013 V2SImode or V2SFmode.
13015 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
13017 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
13018 except V2SF and V2SI. Rearrange the vector modes, and add comments.
13019 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
13021 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
13022 (rs6000_legitimate_offset_address_p): Ditto.
13023 (rs6000_emit_move): Ditto.
13024 (rs6000_init_builtins): Remove V4HI_type_node.
13026 2017-06-13 Martin Liska <mliska@suse.cz>
13029 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
13030 (gate_asan): Likewise.
13031 * asan.h (asan_no_sanitize_address_p): Remove the function.
13032 (sanitize_flags_p): New function.
13033 * builtins.def: Fix coding style.
13034 * common.opt: Use renamed enum value.
13035 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
13036 * doc/extend.texi: Document no_sanitize attribute.
13037 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
13038 to SANITIZE_UNDEFINED_NONDEFAULT.
13039 * gcc.c (sanitize_spec_function): Use the renamed enum value.
13040 * gimple-fold.c (optimize_atomic_compare_exchange_p):
13041 Use sanitize_flags_p.
13042 * gimplify.c (gimplify_function_tree): Likewise.
13043 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
13044 * opts.c (parse_no_sanitize_attribute): New function.
13045 (common_handle_option): Use renamed enum value.
13046 * opts.h (parse_no_sanitize_attribute): Declare.
13047 * tree.c (sanitize_flags_p): New function.
13048 * tree.h: Declared here.
13049 * tsan.c: Use sanitize_flags_p.
13050 * ubsan.c (ubsan_expand_null_ifn): Likewise.
13051 (instrument_mem_ref): Likewise.
13052 (instrument_bool_enum_load): Likewise.
13053 (do_ubsan_in_current_function): Remove the function.
13054 (pass_ubsan::execute): Use sanitize_flags_p.
13055 * ubsan.h: Remove do_ubsan_in_current_function
13056 * tree-cfg.c (print_no_sanitize_attr_value): New function.
13057 (dump_function_to_file): Use it here.
13059 2017-06-13 Martin Jambor <mjambor@suse.cz>
13061 PR tree-optimization/80803
13062 PR tree-optimization/81063
13063 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
13064 (propagate_subaccesses_across_link): Enqueue subtree whenever
13065 necessary instead of relying on the caller.
13067 2017-06-13 Martin Jambor <mjambor@suse.cz>
13069 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
13070 that have a first_link.
13071 (sort_and_splice_var_accesses): Do not check first_link before
13073 (subtree_mark_written_and_enqueue): Likewise.
13074 (propagate_all_subaccesses): Likewise and do not stop at first
13075 parent with a first_link.
13077 2017-06-13 Martin Jambor <mjambor@suse.cz>
13079 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
13082 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13084 * match.pd: New pattern.
13086 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13088 * tree-vrp.c (is_masked_range_test): New function.
13089 (register_edge_assert_for): Determine ranges for
13092 2017-06-13 Yury Gribov <tetra2005@gmail.com>
13094 PR tree-optimization/67328
13095 * fold-const.c (maskable_range_p): New function.
13096 (build_range_check): Generate bittests if possible.
13098 2017-06-13 Martin Liska <mliska@suse.cz>
13100 * gimple-pretty-print.c (dump_probability): Add new argument.
13101 (dump_edge_probability): Dump both probability and count.
13102 (dump_gimple_label): Likewise.
13103 (dump_gimple_bb_header): Likewise.
13105 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
13108 * config/avr/avr-devices.c: Fix indentation.
13109 * config/avr/gen-avr-mmcu-specs.c: Dito.
13111 2017-06-13 Richard Biener <rguenther@suse.de>
13113 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
13114 instead get vector type from stmt_info.
13115 (vectorizable_reduction): Adjust. Remove dead code.
13117 2017-06-13 Richard Biener <rguenther@suse.de>
13119 PR middle-end/81065
13120 * fold-const.c (extract_muldiv_1): Remove bogus distribution
13121 case of C * (x * C2 + C3).
13122 (fold_addr_of_array_ref_difference): Properly fold index difference.
13124 2017-06-12 David S. Miller <davem@davemloft.net>
13127 * config/sparc/sparc.md (return expander): Emit frame blockage if
13128 function uses alloca.
13130 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13132 * combine.c (make_field_assignment): Check len rather than the mode
13133 precision when calling force_to_mode.
13135 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
13137 Support multilibs and devices that see flash in RAM address range.
13140 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
13141 (avr_mcu_t) <flash_pm_offset>: New field.
13142 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
13143 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
13144 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
13145 (AVR_TINY_PM_OFFSET): Remove macro.
13146 * config/avr/avr.opt (-mshort-calls): New option.
13147 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13148 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
13149 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
13150 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
13151 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
13152 instead of avr_arch->have_jmp_call.
13153 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
13154 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
13155 avr_arch->flash_pm_offset to define.
13156 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
13157 new field flash_pm_offset. Add entry for avrxmega3.
13158 (avr_texinfo): Add entry for avrxmega3.
13159 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
13160 attiny212, attiny214,
13161 attiny412, attiny414, attiny416, attiny417,
13162 attiny814, attiny816, attiny817,
13163 attiny1614, attiny1616, attiny1617,
13164 attiny3214, attiny3216, attiny3217.
13165 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
13166 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
13167 (avr_print_operand_address) [AVR_TINY]: Same.
13168 (avr_asm_init_sections) <readonly_data_section>: Only patch
13169 callback if avr_arch->flash_pm_offset = 0.
13170 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
13171 for rodata if avr_arch->flash_pm_offset != 0.
13172 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
13173 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
13174 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
13175 (m_options): Append opt_rcall.
13176 (m_dirnames): Append dir_rcall.
13177 * config/avr/t-multilib: Regenerate.
13179 * configure.ac [target=avr]: Check whether avrxmega3 default
13180 linker description file works as needed.
13181 * configure: Regenerate.
13182 * doc/avr-mmcu.texi: Regenerate.
13183 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
13184 <__AVR_ARCH__>: Document avrxmega3 and 103.
13185 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
13186 <__AVR_SHORT_CALLS__>: Document it.
13187 <__AVR_PM_BASE_ADDRESS__>: Document it.
13188 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
13189 (AVR Variable Attributes) <progmem>: Document this is
13190 not needed for avrxmega3.
13191 (AVR Named Address Spaces) <__flash>: Dito.
13193 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
13195 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
13197 2017-06-12 Doug Rupp <rupp@adacore.com>
13199 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
13200 Append vxworks-stdint.h to the tm_file list.
13201 * config/vxworks-stdint.h: New file.
13203 2017-06-12 Martin Liska <mliska@suse.cz>
13205 PR tree-optimization/81041
13206 * tree-profile.c (gimple_gen_ic_func_profiler):
13207 Create an extra BB in profile-generate
13208 (gimple_gen_time_profiler): Likewise.
13210 2017-06-12 Jakub Jelinek <jakub@redhat.com>
13212 PR tree-optimization/81003
13213 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
13214 (update_range_test): Use it instead of force_gimple_operand_gsi.
13216 2017-06-12 Richard Biener <rguenther@suse.de>
13218 PR tree-optimization/81053
13219 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
13220 with backedge value not defined in loop. Simplify def stmt
13223 2017-06-11 Tom de Vries <tom@codesourcery.com>
13226 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
13228 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
13229 nvptx_cannot_force_const_mem.
13231 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13233 * opts.c (finish_options): Move test for flag_split_stack after
13234 it has been initialized.
13236 2017-06-11 Jason Merrill <jason@redhat.com>
13238 * tree.h (id_equal): New.
13239 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
13240 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
13241 instead of strcmp of IDENTIFIER_POINTER.
13243 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13245 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
13246 (mark_all_inlined_calls_cdtor): Fix formating.
13247 (inline_transform): Rescale profile before inlining.
13249 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13251 * cgraph.h (cgraph_edge::clone): Update prototype.
13252 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
13253 (cgraph_node::create_clone): Update.
13254 (cgraph_node::create_version_clone): Update.
13255 * tree-inline.c (copy_bb): Update.
13256 (expand_call_inline): Update.
13258 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13260 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
13261 factored out from ...
13262 (rs6000_emit_prologue): ... here.
13264 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
13266 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
13267 factored out from ...
13268 (rs6000_emit_prologue): ... here.
13270 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13272 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
13274 (handle_missing_profiles): Fix computation of tp_first_run.
13275 (counts_to_freqs): Do not touch freqs when count is 0.
13277 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
13279 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
13282 2017-06-10 Tom de Vries <tom@codesourcery.com>
13284 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
13285 attributes): Document signal effective target.
13287 2017-06-10 Tom de Vries <tom@codesourcery.com>
13289 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13290 Document effective target stack_size.
13292 2017-06-09 David Malcolm <dmalcolm@redhat.com>
13294 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
13295 to the edit_context if they can be auto-applied.
13297 2017-06-9 Ian Lance Taylor <iant@golang.org>
13299 * opts.c (finish_options): If -fsplit-stack, disable implicit
13300 -forder-blocks-and-partition.
13301 * doc/invoke.texi (Optimize Options): Document that when using
13302 -fsplit-stack -forder-blocks-and-partition is not implicitly
13305 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13307 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
13308 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
13309 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
13310 * builtins.def (abort, trap, unreachable): Declare cold.
13311 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
13312 * tree-core.h (ECF_COLD): New.
13313 * tree.c (set_call_expr_flags): Handle ECF_COLD.
13314 (build_common_builtin_nodes): Mark unreachable and abort as cold.
13316 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13318 * predict.c (unlikely_executed_stmt_p): Cleanup.
13320 2017-06-09 Richard Biener <rguenther@suse.de>
13322 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
13323 model if the ref is always written to.
13325 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13327 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
13329 2017-06-09 Tamar Christina <tamar.christina@arm.com>
13331 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
13334 2017-06-09 Tom de Vries <tom@codesourcery.com>
13337 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
13338 "target cannot support label values" when encountering LABEL_REF.
13340 2017-06-09 Martin Liska <mliska@suse.cz>
13342 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
13343 (gimple_gen_ic_func_profiler): Emit direct comparison
13344 of __gcov_indirect_call_callee with NULL.
13345 (gimple_gen_time_profiler): Change probability from
13346 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
13348 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13350 * profile.c (edge_gcov_counts): Turn to pointer.
13351 (compute_branch_probabilities, compute_branch_probabilities): Update.
13352 (branch_prob): Do not clear edge_gcov_count.
13353 * profile.h (edge_gcov_counts): Turn to pointer.
13354 (edge_gcov_count): Update.
13356 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
13358 * gimple.h (gimple_check_failed): Mark cold.
13360 2017-06-09 Richard Biener <rguenther@suse.de>
13362 PR tree-optimization/66623
13363 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
13364 refactor check_reduction into two parts, properly computing
13365 whether we have to check reduction validity for outer loop
13368 2017-06-09 Richard Biener <rguenther@suse.de>
13370 PR tree-optimization/79483
13371 * graphite-scop-detection.c (order): New global.
13372 (get_order): Compute bb to order mapping that satisfies code
13373 generation constraints.
13374 (cmp_pbbs): New helper.
13375 (build_scops): Start domwalk at entry block, sort generated
13378 2017-06-09 Richard Biener <rguenther@suse.de>
13380 PR middle-end/81007
13381 * ipa-polymorphic-call.c
13382 (ipa_polymorphic_call_context::restrict_to_inner_class):
13383 Skip FIELD_DECLs with error_mark_node type.
13384 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
13387 2017-06-09 Martin Liska <mliska@suse.cz>
13389 * predict.c (struct branch_predictor): New struct.
13390 (test_prediction_value_range): New test.
13391 (predict_c_tests): New function.
13392 * selftest-run-tests.c (selftest::run_tests): Run the function.
13393 * selftest.h: Declare new tests.
13395 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
13398 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
13399 gen_add3_insn did not fail.
13400 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
13401 r0, construct that number in a temporary reg and add that reg to r0.
13402 If asked to put the result in r0 as well, fail.
13404 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
13406 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13407 for early expansion of vec_eqv.
13409 2017-06-08 Jakub Jelinek <jakub@redhat.com>
13411 PR middle-end/81005
13412 * ubsan.c (instrument_null): Avoid pointless code temporary.
13413 (pass_ubsan::execute): Instrument aggregate arguments of calls.
13415 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
13419 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
13422 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
13425 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13427 * predict.c (unlikely_executed_edge_p): Move ahead.
13428 (probably_never_executed_edge_p): Use it.
13430 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13432 PR middle-end/79988
13433 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
13434 gimple_call_builtin_p call.
13436 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13438 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
13439 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
13440 rtl_check_failed_type2, rtl_check_failed_code1,
13441 rtl_check_failed_code2, rtl_check_failed_code_mode,
13442 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
13443 rtvec_check_failed_bounds, rtl_check_failed_flag,
13444 _fatal_insn_not_found, _fatal_insn): Likewise.
13445 * tree.h (tree_contains_struct_check_failed,
13446 tree_check_failed, tree_not_check_failed,
13447 tree_class_check_failed, tree_range_check_failed,
13448 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
13449 tree_vec_elt_check_failed, phi_node_elt_check_failed,
13450 tree_operand_check_failed, omp_clause_check_failed,
13451 omp_clause_operand_check_failed, omp_clause_range_check_failed):
13454 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13456 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
13457 flag_branch_probabilities.
13458 * ipa-inline.c (edge_badness): Likewise.
13459 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13460 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
13461 * predict.c (maybe_hot_frequency_p): Likewise.
13462 (probably_never_executed): Likewise.
13463 * sched-ebb.c (schedule_ebbs): Likewise.
13464 * sched-rgn.c (find_single_block_region): Likewise.
13465 * tracer.c (tail_duplicate): Likewise.
13467 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13469 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
13470 longer requires x_flag_profile_use.
13472 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13474 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
13475 instead of flag_reorder_blocks_and_partition.
13476 * dbxout.c (dbxout_function_end): Likewise.
13477 * dwarf2out.c (gen_subprogram_die): Likewise.
13478 * haifa-sched.c (sched_create_recovery_edges): Likewise.
13479 * hw-doloop.c (reorg_loops): Likewise.
13480 * varasm.c (assemble_start_function,
13481 assemble_end_function): Likewise.
13482 (decide_function_section): Do not check for
13483 flag_reorder_blocks_and_partition.
13485 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
13487 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
13489 (chkp_get_hard_register_fake_addr_expr): Ditto.
13490 (chkp_build_addr_expr): Add check for hard reg case.
13491 (chkp_parse_array_and_component_ref): Ditto.
13492 (chkp_find_bounds_1): Ditto.
13493 (chkp_process_stmt): Don't generate bounds store for
13496 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13498 * predict.c (maybe_hot_bb_p): Do not check profile status.
13499 (maybe_hot_edge_p): Likewise.
13500 (probably_never_executed): Check for zero counts even if profile
13502 (unlikely_executed_edge_p): New function.
13503 (unlikely_executed_stmt_p): New function.
13504 (unlikely_executed_bb_p): New function.
13505 (set_even_probabilities): Use unlikely predicates.
13506 (combine_predictions_for_bb): Likewise.
13507 (predict_paths_for_bb): Likewise.
13508 (predict_paths_leading_to_edge): Likewise.
13509 (determine_unlikely_bbs): New function.
13510 (estimate_bb_frequencies): Use it.
13511 (compute_function_frequency): Use zero counts even if profile is
13513 * profile-count.h: Fix typo.
13515 2017-08-08 Julia Koval <julia.koval@intel.com>
13517 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
13518 _mm512_mask_cvtsepi16_storeu_epi8,
13519 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
13520 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
13521 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
13522 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
13523 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
13524 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
13525 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
13526 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
13527 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
13528 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
13529 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
13530 __builtin_ia32_pmovuswb256mem_mask,
13531 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
13532 __builtin_ia32_pmovwb512mem_mask): New builtins.
13534 2017-08-08 Julia Koval <julia.koval@intel.com>
13536 PR target/73350,80862
13537 * config/i386/subst.md (round): Fix round pattern.
13538 * config/i386/i386.c (ix86_erase_embedded_rounding):
13539 Fix erasing rounding for the fixed pattern.
13541 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
13543 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
13545 2017-06-08 Martin Liska <mliska@suse.cz>
13547 PR gcov-profile/80911
13548 * gcov.c (block_info::block_info): New constructor.
13550 2017-06-07 Carl Love <cel@us.ibm.com>
13552 * config/rs6000/rs6000-c: The return type of the following
13553 built-in functions was implemented as int not long long. Fix sign
13554 of return value for the unsigned version of vec_mulo and vec_mule.
13555 vector unsigned long long vec_bperm (vector unsigned long long,
13556 vector unsigned char)
13557 vector signed long long vec_mule (vector signed int,
13559 vector unsigned long long vec_mule (vector unsigned int,
13560 vector unsigned int)
13561 vector signed long long vec_mulo (vector signed int,
13563 vector unsigned long long vec_mulo (vector unsigned int,
13564 vector unsigned int)
13565 * doc/extend.texi: Fix the documentation for the built-in
13568 2017-06-07 Carl Love <cel@us.ibm.com>
13571 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
13574 2017-06-07 Carl Love <cel@us.ibm.com>
13576 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
13577 support, Generate doublehv for signed int/float for BE case only.
13579 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
13581 * doc/invoke.texi (mcx16): Rewrite.
13583 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13585 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
13586 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
13587 *mov<mode>_softfloat, and an anonymous splitter): Use
13588 nonimmediate_operand instead of rs6000_nonimmediate_operand.
13590 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13592 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
13594 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13595 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
13596 (rs6000_debug_reg_global): Adjust.
13597 (rs6000_init_hard_regno_mode_ok): Adjust.
13598 (rs6000_dbx_register_number): Adjust.
13599 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
13600 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13601 Remove SPE_ACC and SPEFSCR.
13602 (REG_ALLOC_ORDER): Ditto.
13603 (FRAME_POINTER_REGNUM): Change to 111.
13604 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
13605 (REG_CLASS_NAMES): Ditto.
13606 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
13607 (REGISTER_NAMES): Ditto.
13608 (ADDITIONAL_REG_NAMES): Ditto.
13609 (rs6000_reg_names): Ditto.
13610 * config/rs6000/rs6000.md: Renumber some register number
13613 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13615 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
13617 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
13618 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
13620 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
13621 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
13622 Delete the SPE high registers.
13623 (REG_ALLOC_ORDER): Ditto.
13624 (enum reg_class): Remove SPE_HIGH_REGS.
13625 (REG_CLASS_NAMES): Ditto.
13626 (REG_CLASS_CONTENTS): Delete the SPE high registers.
13627 (REGISTER_NAMES): Ditto.
13628 (rs6000_reg_names): Ditto.
13629 * doc/tm.texi.in: Remove SPE as example.
13630 * doc/tm.texi: Regenerate.
13632 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13634 * config/rs6000/8540.md (ppc8540_brinc): Delete.
13635 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
13636 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
13637 * config/rs6000/rs6000.md (type): Remove "brinc".
13639 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13641 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
13642 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
13643 * config/rs6000/linuxspe.h: Delete file.
13644 * config/rs6000/rs6000.md: Don't include spe.md.
13645 * config/rs6000/spe.h: Delete file.
13646 * config/rs6000/spe.md: Delete file.
13647 * config/rs6000/t-rs6000: Remove spe.md.
13649 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13651 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
13652 (reg_or_none500mem_operand): Delete.
13653 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
13654 instead of reg_or_none500mem_operand.
13656 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13658 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13659 handling of SPE flags.
13660 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
13662 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13664 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13666 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
13667 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
13668 paired_divv2sf3): Similar.
13669 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
13670 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
13671 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
13673 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
13674 Rename the paired_* instruction patterns.
13675 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
13677 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
13678 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
13679 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
13680 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
13681 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
13682 PAIRED_VECTOR_MODE.
13683 (struct machine_function): Delete field spe_insn_chain_scanned_p.
13684 (spe_func_has_64bit_regs_p): Delete.
13685 (spe_expand_predicate_builtin): Delete.
13686 (spe_expand_evsel_builtin): Delete.
13687 (TARGET_DWARF_REGISTER_SPAN): Do not define.
13688 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
13689 (invalid_e500_subreg): Delete.
13690 (rs6000_legitimize_address): Always force_reg op2 as well, for
13691 paired single memory accesses.
13692 (rs6000_member_type_forces_blk): Delete.
13693 (rs6000_spe_function_arg): Delete.
13694 (rs6000_expand_unop_builtin): Delete SPE handling.
13695 (rs6000_expand_binop_builtin): Ditto.
13696 (spe_expand_stv_builtin): Delete.
13697 (bdesc_2arg_spe): Delete.
13698 (spe_expand_builtin): Delete.
13699 (spe_expand_predicate_builtin): Delete.
13700 (spe_expand_evsel_builtin): Delete.
13701 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
13702 (spe_init_builtins): Delete.
13703 (spe_func_has_64bit_regs_p): Delete.
13704 (savres_routine_name): Delete "info" parameter. Adjust callers.
13705 (rs6000_emit_stack_reset): Ditto.
13706 (rs6000_dwarf_register_span): Delete.
13707 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
13708 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
13709 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
13711 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
13713 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
13714 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
13715 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
13716 mulv2sf3, divv2sf3): Delete expanders.
13718 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13720 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
13722 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13724 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
13725 * config/rs6000/rs6000.c: Ditto.
13727 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13729 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
13730 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
13731 comparison_operator.
13733 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13735 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
13736 * config/rs6000/rs6000.opt: Ditto.
13737 * config/rs6000/t-rtems: Ditto.
13739 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13741 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
13742 TARGET_E500_SINGLE by 0, simplify.
13743 * config/rs6000/rs6000.c: Ditto.
13744 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
13745 (spe_build_register_parallel): Delete.
13746 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
13747 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
13748 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
13749 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
13750 (E500_CONVERT): Delete.
13751 * config/rs6000/spe.md: Remove many patterns and all define_constants.
13753 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
13755 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
13756 * config/rs6000/dfp.md: Ditto.
13757 (negdd2, *negdd2_fpr): Merge.
13758 (absdd2, *absdd2_fpr): Merge.
13759 (negtd2, *negtd2_fpr): Merge.
13760 (abstd2, *abstd2_fpr): Merge.
13761 * config/rs6000/e500.h: Delete file.
13762 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
13763 TARGET_FPRS by 1 and simplify.
13764 * config/rs6000/rs6000-c.c: Ditto.
13765 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
13766 TARGET_DF_SPE by 0.
13767 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
13769 * config/rs6000/rs6000.md: Ditto.
13770 (floatdidf2, *floatdidf2_fpr): Merge.
13771 (move_from_CR_gt_bit): Delete.
13772 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
13773 (E500_CR_IOR_COMPARE): Delete.
13774 (All patterns that require !TARGET_FPRS): Delete.
13775 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
13777 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13779 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
13781 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13783 * graphds.c (add_edge): Intitialize edge's attached data.
13784 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
13785 pointer parameter. Call pointed function on each edge during
13786 graph traversing. Skip traversing the edge when the function
13788 (graphds_dfs, graphds_scc): Ditto.
13789 (for_each_edge): New parameter. Pass the new parameter to callback
13791 * graphds.h (skip_edge_callback): New function pointer type.
13792 (graphds_dfs, graphds_scc): New function pointer parameter.
13793 (graphds_edge_callback, for_each_edge): New parameter.
13795 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13797 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
13798 out code checking if runtime alias check is possible to below ...
13799 Call the new function.
13800 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
13801 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
13803 2017-06-07 Marek Polacek <polacek@redhat.com>
13806 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
13807 TYPE_OVERFLOW_WRAPS check.
13809 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13811 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
13812 if versioning is required.
13813 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
13814 peeling with the check for versioning.
13816 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13818 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
13819 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
13820 Set true to new parameter if new ssa variable is defined.
13821 (vect_gen_vector_loop_niters): Refactor. Set range information
13822 for the new vector loop bound variable.
13823 (vect_do_peeling): Ditto.
13825 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13827 * tree-affine.c (ssa.h): Include header file.
13828 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
13829 has wrapping overflow behavior.
13831 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13833 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
13835 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13837 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
13838 (tree_to_aff_combination): ... here.
13840 2017-06-07 Bin Cheng <bin.cheng@arm.com>
13842 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
13843 reg_pressure model function.
13844 (ivopts_global_cost_for_size): Delete.
13845 (determine_set_costs, iv_ca_recount_cost): Call new model function
13846 ivopts_estimate_reg_pressure.
13848 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13850 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
13851 expensive than udiv. Remove floating point cases from mod.
13853 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13855 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
13856 Increase idiv cost.
13858 2017-06-07 Tamar Christina <tamar.christina@arm.com>
13860 * config/aarch64/aarch64.md
13861 (copysignsf3): Fix mask generation.
13863 2017-06-07 Jakub Jelinek <jakub@redhat.com>
13865 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
13867 (class dump_manager): Add register_dumps method.
13868 * dumpfile.c: Include langhooks.h.
13869 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
13870 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
13871 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
13872 (dump_manager::dump_register): Start with 512 entries instead of 32.
13873 (dump_manager::register_dumps): New method.
13874 * toplev.c (general_init): Instead of invoking register_dumps
13875 langhook, invoke register_dumps method on the dump manager.
13876 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
13879 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
13881 * doc/md.texi: Clarify the restrictions on a define_insn condition.
13882 Say that # requires an associated define_split to exist, and that
13883 the define_split must be suitable for use after register allocation.
13885 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13887 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
13888 (compute_outgoing_frequencies): Also initialize zero counts.
13889 (find_many_sub_basic_blocks): Do not produce uninitialized profile
13890 around loops; preserve more of profile when nothing changes.
13892 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
13894 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
13896 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
13897 * config/arm/arm-cpu-cdata.h: Regenerate.
13898 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
13899 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
13900 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
13901 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
13902 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
13904 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
13906 * config/arm/t-rmprofile: Likewise.
13907 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
13909 2017-06-06 David S. Miller <davem@davemloft.net>
13912 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
13913 blockage if function uses alloca.
13915 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13917 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
13918 New "uid" fields to hold pretty-print IDs of group and ref.
13919 Memory references are now identified as <group_id>:<ref_id>
13920 instead of using [random] addresses.
13921 (dump_mem_details): Simplify, no functional change.
13922 (dump_mem_ref): Simplify and make output more concise.
13923 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
13924 (find_or_create_group): Initialize group uid.
13925 (record_ref): Initialize ref uid. Improve debug output.
13926 (prune_group_by_reuse, should_issue_prefetch_p,)
13927 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
13928 (mark_nontemporal_store, determine_loop_nest_reuse):
13929 Improve debug output.
13931 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
13933 * dbgcnt.def (prefetch): New debug counter.
13934 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
13935 (schedule_prefetches): Stop issueing prefetches if debug counter
13938 2017-06-06 Tom de Vries <tom@codesourcery.com>
13940 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
13941 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
13943 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13945 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
13946 Use aarch64_reg_or_zero predicate for operand 4.
13947 (aarch64_compare_and_swap<mode> define_insn_and_split):
13948 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
13949 (aarch64_store_exclusive<mode>): Likewise for operand 2.
13951 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13953 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
13954 (arm_compute_save_core_reg_mask): This.
13955 (thumb1_compute_save_reg_mask): Rename into ...
13956 (thumb1_compute_save_core_reg_mask): This.
13957 (arm_compute_save_reg0_reg12_mask): Adapt comment.
13958 (arm_compute_frame_layout): Likewise.
13960 2017-06-06 Richard Biener <rguenther@suse.de>
13962 PR tree-optimization/80974
13963 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
13964 keep or clear leaders SSA info.
13966 2017-06-06 Tom de Vries <tom@codesourcery.com>
13968 * config/nvptx/nvptx.c (split_mode_p): New function.
13969 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
13971 2017-06-06 Tom de Vries <tom@codesourcery.com>
13973 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
13975 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13978 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
13981 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13983 * shrink-wrap.c (handle_simple_exit): Update profile.
13984 (try_shrink_wrapping): Upate profile.
13986 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13988 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
13989 (tree_guess_outgoing_edge_probabilities): New.
13990 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
13991 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
13993 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13995 * ipa-split.c (split_function): Initialize return bb profile.
13997 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
13999 * profile.c (compute_branch_probabilities): Also initialize
14000 EXIT_BLOCK profile.
14002 2017-06-06 Richard Biener <rguenther@suse.de>
14004 PR tree-optimization/80928
14005 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
14006 (vect_analyze_loop_operations): Properly guard analysis for
14008 (vect_transform_loop): Likewise.
14009 (vect_analyze_loop_2): Also reset SLP type on PHIs.
14010 (vect_model_induction_cost): Do not cost for pure SLP.
14011 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
14012 of induction in inner loop vectorization.
14013 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
14014 (vect_get_and_check_slp_defs): Handle vect_induction_def.
14015 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
14017 (vect_analyze_slp_cost_1): Cost induction.
14018 (vect_detect_hybrid_slp_stmts): Handle PHIs.
14019 (vect_get_slp_vect_defs): Likewise.
14020 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
14021 (vect_transform_stmt): Handle SLP reductions.
14022 * tree-vectorizer.h (vectorizable_induction): Adjust.
14024 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14026 * config/rs6000/rs6000.c (make_resolver_func): Update
14027 init_lowered_empty_function call.
14029 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
14031 * doc/invoke.texi: Document the -fprofile-abs-path option.
14032 * common.opt (fprofile-abs-path): New option.
14033 * gcov-io.h (gcov_write_filename): Declare.
14034 * gcov-io.c (gcov_write_filename): New function.
14035 * coverage.c (coverage_begin_function): Use gcov_write_filename.
14036 * profile.c (output_location): Likewise.
14038 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14040 * shring-wrap.c: Revert accidental commit.
14042 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
14044 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
14046 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14048 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
14050 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
14051 profile in callgraph edge.
14052 * profile-count.h (apply_probability): If THIS is 0, then result is 0
14053 (apply_scale): Likewise.
14054 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
14055 Also scale profile when inlining function with zero profile.
14056 (initialize_cfun): Update exit block profile even when it is zero.
14057 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
14058 when profile is read.
14060 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14062 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
14063 (CLONE_*): New constants to define the processors we can generate
14064 code for with the target_clone attribute.
14065 (rs6000_clone_map): New array to identify which clone processors
14066 the current program is running on.
14067 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
14068 target_clone attribute.
14069 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
14070 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
14071 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
14072 (cpu_expand_builtin): Add support for target_clone attribute.
14073 (rs6000_valid_attribute_p): Allow "default" attribute.
14074 (get_decl_name): New debug function to simplify printing the
14075 current function name in debugging statements.
14076 (rs6000_clone_priority): New functions to support the target_clone
14077 attribute, and be able to generate code to switch between ISA 2.05
14078 through ISA 3.0 (power6 through power9).
14079 (rs6000_compare_version_priority): Likewise.
14080 (rs6000_get_function_versions_dispatcher): Likewise.
14081 (make_resolver_func): Likewise.
14082 (add_condition_to_bb): Likewise.
14083 (dispatch_function_versions): Likewise.
14084 (rs6000_generate_version_dispatcher_body): Likewise.
14085 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
14086 (fusion_gpr_load_p): Fix a spacing issue.
14087 * doc/extend.texi (Common Function Attributes): Document that the
14088 PowerPC supports the target_clone attribute.
14090 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14092 * config/arm/arm.h: explain F symbol found in description of ARM
14093 register allocation in its legend.
14095 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14097 * config/mips/frame-header-opt.c: Include profile-count.h.
14098 * config/riscv/riscv.c: Include profile-count.h
14100 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
14102 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
14104 (sm_set_flag_if_changed): Add bbs field.
14105 (execute_sm_if_changed_flag_set): Pass BBS.
14106 (execute_sm): Update.
14108 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14110 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
14113 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14115 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
14116 (peephole2): New peephole2 to emit the above.
14117 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
14119 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14121 * config/aarch64/aarch64.c (define_peephole2 above
14122 *sub_<shift>_<mode>): New peephole.
14124 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14126 * config/i386/i386.c (make_resolver_func): Update.
14127 * Makefile.in: Add profile-count.h and profile-count.o
14128 * auto-profile.c (afdo_indirect_call): Update to new API.
14129 (afdo_set_bb_count): Update.
14130 (afdo_propagate_edge): Update.
14131 (afdo_propagate_circuit): Update.
14132 (afdo_calculate_branch_prob): Update.
14133 (afdo_annotate_cfg): Update.
14134 * basic-block.h: Include profile-count.h
14135 (struct edge_def): Turn count to profile_count.
14136 (struct basic_block_def): Likewie.
14137 (REG_BR_PROB_BASE): Move to profile-count.h
14138 (RDIV): Move to profile-count.h
14139 * bb-reorder.c (max_entry_count): Turn to profile_count.
14140 (find_traces): Update.
14141 (rotate_loop):Update.
14142 (connect_traces):Update.
14143 (sanitize_hot_paths):Update.
14144 * bt-load.c (migrate_btr_defs): Update.
14145 * cfg.c (RDIV): Remove.
14146 (init_flow): Use alloc_block.
14147 (alloc_block): Uninitialize count.
14148 (unchecked_make_edge): Uninitialize count.
14149 (check_bb_profile): Update.
14150 (dump_edge_info): Update.
14151 (dump_bb_info): Update.
14152 (update_bb_profile_for_threading): Update.
14153 (scale_bbs_frequencies_int): Update.
14154 (scale_bbs_frequencies_gcov_type): Update.
14155 (scale_bbs_frequencies_profile_count): New.
14156 * cfg.h (update_bb_profile_for_threading): Update.
14157 (scale_bbs_frequencies_profile_count): Declare.
14158 * cfgbuild.c (compute_outgoing_frequencies): Update.
14159 (find_many_sub_basic_blocks): Update.
14160 * cfgcleanup.c (try_forward_edges): Update.
14161 (try_crossjump_to_edge): Update.
14162 * cfgexpand.c (expand_gimple_tailcall): Update.
14163 (construct_exit_block): Update.
14164 * cfghooks.c (verify_flow_info): Update.
14165 (dump_bb_for_graph): Update.
14166 (split_edge): Update.
14167 (make_forwarder_block): Update.
14168 (duplicate_block): Update.
14169 (account_profile_record): Update.
14170 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
14171 (get_estimated_loop_iterations): Update.
14172 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
14173 (single_likely_exit): Update.
14174 * cfgloopmanip.c (scale_loop_profile): Update.
14176 (set_zero_probability): Update.
14177 (lv_adjust_loop_entry_edge): Update.
14178 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14179 (purge_dead_edges): Update.
14180 (rtl_account_profile_record): Update.
14181 * cgraph.c (cgraph_node::create): Uninitialize count.
14182 (symbol_table::create_edge): Uninitialize count.
14183 (cgraph_update_edges_for_call_stmt_node): Update.
14184 (cgraph_edge::dump_edge_flags): Update.
14185 (cgraph_node::dump): Update.
14186 (cgraph_edge::maybe_hot_p): Update.
14187 * cgraph.h: Include profile-count.h
14188 (create_clone), create_edge, create_indirect_edge): Update.
14189 (cgraph_node): Turn count to profile_count.
14190 (cgraph_edge0: Likewise.
14191 (make_speculative, clone): Update.
14192 (create_edge): Update.
14193 (init_lowered_empty_function): Update.
14194 * cgraphclones.c (cgraph_edge::clone): Update.
14195 (duplicate_thunk_for_node): Update.
14196 (cgraph_node::create_clone): Update.
14197 * cgraphunit.c (cgraph_node::analyze): Update.
14198 (cgraph_node::expand_thunk): Update.
14199 * final.c (dump_basic_block_info): Update.
14200 * gimple-streamer-in.c (input_bb): Update.
14201 * gimple-streamer-out.c (output_bb): Update.
14202 * graphite.c (print_global_statistics): Update.
14203 (print_graphite_scop_statistics): Update.
14204 * hsa-brig.c: Include basic-block.h.
14205 * hsa-dump.c: Include basic-block.h.
14206 * hsa-gen.c (T sum_slice): Update.
14207 (convert_switch_statements):Update.
14208 * hsa-regalloc.c: Include basic-block.h.
14209 * ipa-chkp.c (chkp_produce_thunks): Update.
14210 * ipa-cp.c (struct caller_statistics): Update.
14211 (init_caller_stats): Update.
14212 (gather_caller_stats): Update.
14213 (ipcp_cloning_candidate_p): Update.
14214 (good_cloning_opportunity_p): Update.
14215 (get_info_about_necessary_edges): Update.
14216 (dump_profile_updates): Update.
14217 (update_profiling_info): Update.
14218 (update_specialized_profile): Update.
14219 (perhaps_add_new_callers): Update.
14220 (decide_about_value): Update.
14221 (ipa_cp_c_finalize): Update.
14222 * ipa-devirt.c (struct odr_type_warn_count): Update.
14223 (struct decl_warn_count): Update.
14224 (struct final_warning_record): Update.
14225 (possible_polymorphic_call_targets): Update.
14226 (ipa_devirt): Update.
14227 * ipa-fnsummary.c (redirect_to_unreachable): Update.
14228 * ipa-icf.c (sem_function::merge): Update.
14229 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
14230 * ipa-inline.c (compute_uninlined_call_time): Update.
14231 (compute_inlined_call_time): Update.
14232 (want_inline_small_function_p): Update.
14233 (want_inline_self_recursive_call_p): Update.
14234 (edge_badness): Update.
14235 (lookup_recursive_calls): Update.
14236 (recursive_inlining): Update.
14237 (inline_small_functions): Update.
14238 (dump_overall_stats): Update.
14239 (dump_inline_stats): Update.
14240 * ipa-profile.c (ipa_profile_generate_summary): Update.
14241 (ipa_propagate_frequency): Update.
14242 (ipa_profile): Update.
14243 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14244 * ipa-utils.c (ipa_merge_profiles): Update.
14245 * loop-doloop.c (doloop_modify): Update.
14246 * loop-unroll.c (report_unroll): Update.
14247 (unroll_loop_runtime_iterations): Update.
14248 * lto-cgraph.c (lto_output_edge): Update.
14249 (lto_output_node): Update.
14250 (input_node): Update.
14251 (input_edge): Update.
14252 (merge_profile_summaries): Update.
14253 * lto-streamer-in.c (input_cfg): Update.
14254 * lto-streamer-out.c (output_cfg): Update.
14255 * mcf.c (create_fixup_graph): Update.
14256 (adjust_cfg_counts): Update.
14257 (sum_edge_counts): Update.
14258 * modulo-sched.c (sms_schedule): Update.
14259 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
14260 * predict.c (maybe_hot_count_p): Update.
14261 (probably_never_executed): Update.
14262 (dump_prediction): Update.
14263 (combine_predictions_for_bb): Update.
14264 (propagate_freq): Update.
14265 (handle_missing_profiles): Update.
14266 (counts_to_freqs): Update.
14267 (rebuild_frequencies): Update.
14268 (force_edge_cold): Update.
14269 * predict.h: Include profile-count.h
14270 (maybe_hot_count_p, counts_to_freqs): UPdate.
14271 * print-rtl-function.c: Do not include cfg.h
14272 * print-rtl.c: Include basic-block.h
14273 * profile-count.c: New file.
14274 * profile-count.h: New file.
14275 * profile.c (is_edge_inconsistent): Update.
14276 (correct_negative_edge_counts): Update.
14277 (is_inconsistent): Update.
14278 (set_bb_counts): Update.
14279 (read_profile_edge_counts): Update.
14280 (compute_frequency_overlap): Update.
14281 (compute_branch_probabilities): Update; Initialize and deinitialize
14283 (branch_prob): Update.
14284 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
14285 (edge_gcov_count): New.
14286 (bb_gcov_count): New.
14287 * shrink-wrap.c (try_shrink_wrapping): Update.
14288 * tracer.c (better_p): Update.
14289 * trans-mem.c (expand_transaction): Update.
14290 (ipa_tm_insert_irr_call): Update.
14291 (ipa_tm_insert_gettmclone_call): Update.
14292 * tree-call-cdce.c: Update.
14293 * tree-cfg.c (gimple_duplicate_sese_region): Update.
14294 (gimple_duplicate_sese_tail): Update.
14295 (gimple_account_profile_record): Update.
14296 (execute_fixup_cfg): Update.
14297 * tree-inline.c (copy_bb): Update.
14298 (copy_edges_for_bb): Update.
14299 (initialize_cfun): Update.
14300 (freqs_to_counts): Update.
14301 (copy_cfg_body): Update.
14302 (expand_call_inline): Update.
14303 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14304 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
14305 (try_unroll_loop_completely): Update.
14306 (try_peel_loop): Update.
14307 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14308 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
14309 * tree-ssa-loop-split.c (connect_loops): Update.
14310 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
14311 * tree-ssa-reassoc.c (branch_fixup): Update.
14312 * tree-ssa-tail-merge.c (replace_block_by): Update.
14313 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
14314 (compute_path_counts): Update.
14315 (update_profile): Update.
14316 (recompute_probabilities): Update.
14317 (update_joiner_offpath_counts): Update.
14318 (estimated_freqs_path): Update.
14319 (freqs_to_counts_path): Update.
14320 (clear_counts_path): Update.
14321 (ssa_fix_duplicate_block_edges): Update.
14322 (duplicate_thread_path): Update.
14323 * tree-switch-conversion.c (case_bit_test_cmp): Update.
14324 (struct switch_conv_info): Update.
14325 * tree-tailcall.c (decrease_profile): Update.
14326 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14327 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14328 * value-prof.c (check_counter): Update.
14329 (gimple_divmod_fixed_value): Update.
14330 (gimple_mod_pow2): Update.
14331 (gimple_mod_subtract): Update.
14332 (gimple_ic_transform): Update.
14333 (gimple_stringop_fixed_value): Update.
14334 * value-prof.h (gimple_ic): Update.
14336 2017-06-02 Carl Love <cel@us.ibm.com>
14338 * config/rs6000/rs6000-c: Add support for built-in functions
14339 vector double vec_doublee (vector signed int);
14340 vector double vec_doublee (vector unsigned int);
14341 vector double vec_doublee (vector float);
14342 vector double vec_doubleh (vector signed int);
14343 vector double vec_doubleh (vector unsigned int);
14344 vector double vec_doubleh (vector float);
14345 vector double vec_doublel (vector signed int);
14346 vector double vec_doublel (vector unsigned int);
14347 vector double vec_doublel (vector float);
14348 vector double vec_doubleo (vector signed int);
14349 vector double vec_doubleo (vector unsigned int);
14350 vector double vec_doubleo (vector float);.
14351 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
14352 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
14354 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
14355 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
14356 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
14358 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
14359 vec_doublel, vec_doubleh.
14360 * doc/extend.texi: Update the built-in documentation file for the
14361 new built-in functions.
14363 2017-06-02 David Malcolm <dmalcolm@redhat.com>
14366 * ipa-inline-analysis.c (free_growth_caches): Set
14367 edge_removal_hook_holder to NULL after removing it.
14369 2017-06-02 Sudakshina Das <sudi.das@arm.com>
14371 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
14372 comparision with zero.
14374 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14375 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14376 for early expansion of vec_min and vec_max builtins.
14377 (builtin_function_type): Add min/max unsigned variants to those
14378 identified as having unsigned arguments.
14380 2017-06-02 Olivier Hainque <hainque@adacore.com>
14382 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
14384 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14386 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
14387 Use VALL_F16 iterator rather than VALL.
14389 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14391 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
14392 Emit CBNZ inside loop when doing a strong exchange and comparing
14393 against zero. Generate the CC flags after the loop.
14395 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
14397 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
14398 (dl_section_ref): New.
14399 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
14400 On AIX, append an expression to subtract the size of the
14401 section length to dl_section_ref.
14403 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
14405 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14406 for early expansion of vector absolute builtins.
14408 2017-06-02 Richard Biener <rguenther@suse.de>
14410 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
14411 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
14413 2017-06-02 Richard Biener <rguenther@suse.de>
14415 PR tree-optimization/80948
14416 * tree-tailcall.c (find_tail_calls): Track stmts to move in
14417 stmt order as well.
14419 2017-06-02 Richard Biener <rguenther@suse.de>
14421 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
14423 * tree-vect-stmts.c (process_use): Do not mark backedge defs
14424 for inductions as relevant.
14426 2017-06-02 Richard Biener <rguenther@suse.de>
14428 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
14429 (vectorizable_induction): ... this. Remove dead code.
14431 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
14433 * builtins. (expand_builtin_alloca): Remove second parameter and
14434 infer its value from the first parameter instead.
14435 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
14437 2017-06-02 Jakub Jelinek <jakub@redhat.com>
14439 PR rtl-optimization/80903
14440 * loop-doloop.c (add_test): Unshare sequence.
14442 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
14444 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
14446 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
14448 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
14450 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
14451 xlogue_layout::get_instance, logue_layout::xlogue_layout,
14452 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
14453 (xlogue_layout::get_stub_rtx): Make static.
14454 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
14455 (xlogue_layout::compute_stub_managed_regs): Rename to...
14456 (xlogue_layout::count_stub_managed_regs): ...this.
14457 (xlogue_layout::is_stub_managed_reg): New function.
14458 (xlogue_layout::m_stub_names): Rename to...
14459 (xlogue_layout::s_stub_names): ...this, make static.
14460 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
14461 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
14462 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
14463 xlogue_layout::s_stub_names): Instantiate statics.
14464 (stub_managed_regs): Remove.
14465 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
14466 (disable_call_ms2sysv_xlogues): Rename to...
14467 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
14468 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
14470 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
14471 change after reload_completed.
14472 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
14474 (ix86_expand_prologue): Likewise.
14475 (ix86_expand_epilogue): Likewise.
14476 (ix86_expand_split_stack_prologue): Likewise.
14477 (ix86_compute_frame_layout): Remove frame parameter ...
14478 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
14479 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
14481 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
14482 (ix86_frame): Move from here ...
14483 * config/i386/i386.h (ix86_frame): ... to here.
14484 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
14485 complete ix86_frame data structure instead. Remove some_ld_name.
14487 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14489 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
14490 symbols that hold a DECL_VALUE_EXPR.
14492 2017-06-01 Martin Jambor <mjambor@suse.cz>
14494 PR tree-optimization/80898
14495 * tree-sra.c (process_subtree_disqualification): Removed.
14496 (disqualify_candidate): Do not acll
14497 process_subtree_disqualification.
14498 (subtree_mark_written_and_enqueue): New function.
14499 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
14500 RHS has been disqualified and re-queue LHS if necessary. Apart
14501 from that, ignore disqualified RHS.
14503 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14505 * config/s390/s390.c (s390_emit_epilogue): Disable early return
14506 address fetch for z10 or later.
14508 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14510 * config/arc/arc.md (tst_movb): Add guard when splitting.
14512 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14514 * config/arc/arc.c (arc_can_eliminate): Test against
14515 arc_frame_pointer_needed.
14517 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14519 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
14520 to prevent store reordering.
14521 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
14522 (type): Add block type.
14523 (stack_tie): Define special instruction to be used in
14526 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14528 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
14529 constraint. It is not valid for the pattern.
14530 (noncommutative_binary_comparison): Likewise.
14532 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14534 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
14537 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14539 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
14540 be used by the reg-alloc.
14542 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14544 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
14545 reg-alloc when having mul64 or mul32x16 instructions.
14546 (mulsidi3): Likewise.
14547 (umulsidi3): Likewise.
14548 (mulsi32x16): New pattern.
14549 (mulsi64): Likewise.
14550 (mulsidi64): Likewise.
14551 (umulsidi64): Likewise.
14552 (MUL32x16_REG): Define.
14553 (mul64_600): Use MUL32x16_REG.
14554 (mac64_600): Likewise.
14555 (umul64_600): Likewise.
14556 (umac64_600): Likewise.
14558 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
14560 * config/arc/arc.md (mulsi3_700): Make it commutative.
14562 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
14564 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
14566 (*sign_extendsidi2_insn): Likewise for movstosw.
14568 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
14570 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
14571 the type of the input discriminant value. Convert the
14572 discriminant value of signedness vary.
14574 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
14576 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
14577 Add to -Wall section.
14579 2017-06-01 Richard Biener <rguenther@suse.de>
14581 PR middle-end/66313
14582 * fold-const.c (fold_plusminus_mult_expr): If the factored
14583 factor may be zero use a wrapping type for the inner operation.
14584 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
14585 and handle moved defs.
14586 (process_assignment): Properly guard the unary op case. Return a
14587 tri-state indicating that moving the stmt before the call may allow
14588 to continue. Pass through to_move.
14589 (find_tail_calls): Handle moving unrelated defs before
14592 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
14595 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
14596 splitter result in the canonical way.
14598 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
14600 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
14601 also for 32bit target. Update insn attributes.
14602 (zero-extendsidi2 splitter): Allow all registers for operand 1.
14604 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
14606 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
14607 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
14608 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
14609 (_mm_maskz_min_ss): New intrinsics.
14611 2017-05-31 Martin Liska <mliska@suse.cz>
14613 * tree-vect-loop.c (vect_create_epilog_for_reduction):
14614 Change comment style to one we normally use.
14615 (vectorizable_reduction): Likewise.
14616 (vectorizable_induction): Likewise.
14617 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14618 (vectorizable_call): Likewise.
14619 (vectorizable_simd_clone_call): Likewise.
14620 (vectorizable_conversion): Likewise.
14621 (vectorizable_assignment): Likewise.
14622 (vectorizable_shift): Likewise.
14623 (vectorizable_operation): Likewise.
14624 (vectorizable_store): Likewise.
14625 (vectorizable_load): Likewise.
14626 * tree-vectorizer.h: Likewise.
14628 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
14630 * passes.c (emergency_dump_function): New.
14631 * tree-pass.h (emergency_dump_function): Declare.
14632 * plugin.c (plugins_internal_error_function): Remove.
14633 * plugin.h (plugins_internal_error_function): Remove declaration.
14634 * toplev.c (internal_error_function): New static function. Use it...
14635 (general_init): ...here.
14637 2017-05-31 Graham Markall <graham.markall@embecosm.com>
14639 * config/arc/arc.c (arc_print_operand): Handle constant operands.
14640 (arc_rtx_costs): Add costs for new patterns.
14641 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
14642 * config/arc/predicates.md: Add _1_2_3_operand predicate.
14644 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14646 * tree-ssa-strlen.c (get_next_strinfo): New function.
14647 (get_stridx_plus_constant): Use it.
14648 (zero_length_string): Likewise.
14649 (adjust_related_strinfos): Likewise.
14650 (adjust_last_stmt): Likewise.
14652 2017-05-31 Richard Biener <rguenther@suse.de>
14655 * config/i386/i386.c (ix86_expand_builtin): Remove assert
14656 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
14658 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
14660 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
14661 loop_vinfo argument and use of dependence distance vectors.
14662 Check instead whether the two references differ only in their
14663 initial value and assume that they have the same alignment if the
14664 difference is a multiple of the vector alignment.
14665 (vect_analyze_data_refs_alignment): Update call accordingly.
14667 2017-05-31 Martin Liska <mliska@suse.cz>
14670 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
14672 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14674 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
14675 (create_intersect_range_checks): Move from ...
14676 * tree-data-ref.c (create_intersect_range_checks_index)
14677 (create_intersect_range_checks): ... to here.
14678 (create_runtime_alias_checks): New function factored from ...
14679 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
14680 here. Call above function.
14681 * tree-data-ref.h (create_runtime_alias_checks): New function.
14683 2017-05-31 Bin Cheng <bin.cheng@arm.com>
14685 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
14686 segment length for dr_b and compute it in wide_int.
14688 2017-05-31 Richard Biener <rguenther@suse.de>
14690 PR tree-optimization/80906
14691 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
14692 and pass through iv_map.
14693 (copy_bb_and_scalar_dependences): Adjust.
14694 (translate_pending_phi_nodes): Likewise.
14695 (copy_loop_close_phi_args): Handle code-generating IVs instead
14698 2017-05-30 David Malcolm <dmalcolm@redhat.com>
14700 * diagnostic-color.c (color_dict): Add "type-diff".
14701 (parse_gcc_colors): Update comment.
14702 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
14703 -fdiagnostics-show-template-tree and -fno-elide-type.
14704 (GCC_COLORS): Add type-diff to example.
14706 (-fdiagnostics-show-template-tree): New.
14707 (-fno-elide-type): New.
14708 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
14709 the pp_format_decoder callback. Call any m_format_postprocessor's
14711 (pretty_printer::pretty_printer): Initialize
14712 m_format_postprocessor.
14713 (pretty_printer::~pretty_printer): Delete any
14714 m_format_postprocessor.
14715 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
14716 (class format_postprocessor): New class.
14717 (struct pretty_printer::format_decoder): Document the new parameters.
14718 (struct pretty_printer::m_format_postprocessor): New field.
14719 * tree-diagnostic.c (default_tree_printer): Update for new
14720 bool and const char ** params.
14721 * tree-diagnostic.h (default_tree_printer): Likewise.
14723 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
14725 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
14726 (lwa_operand): Delete rs6000_gen_cell_microcode test.
14727 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
14728 rs6000_gen_cell_microcode code.
14729 (rs6000_final_prescan_insn): Delete.
14730 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
14731 "warn-cell-microcode" entries.
14732 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
14733 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
14734 throughout. Change cc_reg_not_micro_cr0_operand to
14735 cc_reg_not_cr0_operand throughout.
14736 (*extendhi<mode>2_noload): Delete.
14737 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
14738 (mwarn-cell-microcode): Delete.
14739 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
14740 -mgen-cell-microcode and -mwarn-cell-microcode.
14742 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
14745 * config/i386/constraints.md (Yd): New constraint.
14747 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
14748 and (?Yd, r) alternatives. Update insn attributes.
14749 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
14750 and (?*Yd, r) alternatives. Update insn attributes.
14751 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
14753 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
14755 * gimplify.c (gimplify_modify_expr): Don't create a
14756 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
14759 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14761 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
14763 2017-05-30 Richard Biener <rguenther@suse.de>
14765 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
14766 and reduc_def fields.
14767 (STMT_VINFO_REDUC_TYPE): New define.
14768 (STMT_VINFO_REDUC_DEF): Likewise.
14769 (vect_force_simple_reduction): Adjust prototype.
14770 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
14771 (vect_is_simple_reduction): Remove check_reduction argument.
14772 (vect_force_simple_reduction): Adjust and set
14773 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14774 (vectorizable_reduction): Do not re-do reduction analysis
14775 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
14776 * tree-parloops.c (gather_scalar_reductions): Adjust.
14778 2017-05-30 Richard Biener <rguenther@suse.de>
14780 PR middle-end/80901
14781 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
14784 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
14786 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
14787 Introduce unknown_misalignment parameter and remove vf.
14788 (vect_peeling_hash_get_lowest_cost):
14789 Pass unknown_misalignment parameter.
14790 (vect_enhance_data_refs_alignment):
14791 Fix unsupportable data ref treatment.
14793 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14795 * tree-vect-data-refs.c (vect_get_data_access_cost):
14796 Workaround for SLP handling.
14797 (vect_enhance_data_refs_alignment):
14798 Compute costs for doing no peeling at all, compare to the best
14799 peeling costs so far and avoid peeling if cheaper.
14801 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14803 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
14804 Return peeling info and set costs to zero for unlimited cost
14806 (vect_enhance_data_refs_alignment): Also inspect all datarefs
14807 with unknown misalignment. Compute and costs for unknown
14808 misalignment, compare them to the costs for known misalignment
14809 and choose the cheapest for peeling.
14811 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14813 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
14814 (vect_get_peeling_costs_all_drs): Create function.
14815 (vect_peeling_hash_get_lowest_cost):
14816 Use vect_get_peeling_costs_all_drs.
14817 (vect_peeling_supportable): Create function.
14818 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
14820 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
14822 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
14823 DR_HAS_NEGATIVE_STEP.
14824 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
14825 (vect_enhance_data_refs_alignment): Use.
14826 (vect_duplicate_ssa_name_ptr_info): Use.
14827 * tree-vectorizer.h (dr_misalignment): Use.
14828 (known_alignment_for_access_p): Use.
14830 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14833 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
14835 (has_section_name): New function.
14837 2017-05-30 Martin Liska <mliska@suse.cz>
14840 * auto-profile.c (get_function_decl_from_block): Fix
14843 2017-05-30 Richard Biener <rguenther@suse.de>
14845 PR middle-end/80876
14846 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
14848 2017-05-30 Martin Liska <mliska@suse.cz>
14850 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
14851 * dumpfile.h (struct dump_file_info): Remove ctors.
14853 2017-05-30 Martin Liska <mliska@suse.cz>
14855 * predict.def: Fix GNU coding style.
14857 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14859 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14860 Mark 'to' argument with ATTRIBUTE_UNUSED.
14862 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
14864 * config/xtensa/xtensa.c (xtensa_emit_call): Use
14865 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
14866 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
14869 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
14871 * doc/install.texi (Options specification): Restore entry of
14872 --enable-sjlj-exceptions.
14874 2017-05-27 Michael Eager <eager@eagercon.com>
14877 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
14879 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
14881 * config/microblaze/microblaze.h
14882 (FIXED_REGISTERS): Update in macro.
14883 (CALL_USED_REGISTERS): Update in macro.
14885 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
14887 * doc/install.texi: Add links to macOS binary distributions.
14889 2017-05-27 Jakub Jelinek <jakub@redhat.com>
14893 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14895 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
14897 2017-05-26 Martin Liska <mliska@suse.cz>
14899 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
14901 2017-05-26 Martin Liska <mliska@suse.cz>
14903 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
14904 always leading ';; '.
14905 (dump_bb_info): Likewise.
14906 (brief_dump_cfg): Likewise.
14907 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
14908 * dumpfile.c: Remove usage of TDF_VERBOSE.
14909 * dumpfile.h (enum dump_kind): Likewise.
14910 (dump_gimple_bb_header): Do not use TDF_COMMENT.
14911 * print-tree.c (debug_verbose): Remove.
14912 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
14913 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
14914 * tree-diagnostic.c (default_tree_printer): Replace
14915 TDF_DIAGNOSTIC with TDF_SLIM.
14917 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14919 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
14920 in parameter loop, rather than loop_vinfo.
14921 (create_intersect_range_checks): Ditto.
14922 (vect_create_cond_for_alias_checks): Update call to above functions.
14924 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14926 PR tree-optimization/80815
14927 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
14928 for merging runtime alias checks. Handle negative DR_STEPs.
14930 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14932 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
14934 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
14935 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
14936 out code pruning runtime alias checks.
14937 * tree-data-ref.c (prune_runtime_alias_test_list): New function
14938 factored out from above.
14939 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14941 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
14943 (prune_runtime_alias_test_list): New decalaration.
14945 2017-05-26 Bin Cheng <bin.cheng@arm.com>
14947 * tree-vect-data-refs.c (compare_tree): Rename and move ...
14948 * tree-data-ref.c (data_ref_compare_tree): ... to here.
14949 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
14950 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
14951 (operator==, comp_dr_with_seg_len_pair): Ditto.
14952 (vect_prune_runtime_alias_test_list): Ditto.
14954 2017-05-26 Martin Liska <mliska@suse.cz>
14957 * params.def: Bound partial-inlining-entry-probability param.
14959 2017-05-26 Marek Polacek <polacek@redhat.com>
14962 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
14965 2017-05-26 Richard Biener <rguenther@suse.de>
14967 PR tree-optimization/80842
14968 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
14971 2017-05-26 Richard Biener <rguenther@suse.de>
14973 PR tree-optimization/80844
14974 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
14976 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
14978 * doc/md.texi (Machine Constraints): Update x86 family
14979 machine constraints section to match 'config/i386/constraints.md'.
14981 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
14983 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
14985 2017-05-25 Nathan Sidwell <nathan@acm.org>
14987 * doc/invoke.texi (--enable-languages): Update documentation.
14989 2017-05-25 Martin Liska <mliska@suse.cz>
14991 * dumpfile.c: Add TDF_FOLDING.
14992 * dumpfile.h (enum dump_kind): Likewise.
14993 * genmatch.c (dt_simplify::gen_1): Use it.
14995 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
14997 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
14999 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15001 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
15002 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
15004 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
15006 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
15007 * match.pd (X == C): Rewrite it here.
15008 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
15009 with_certain_nonzero_bits2): New predicates.
15010 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
15012 2017-05-24 Nathan Sidwell <nathan@acm.org>
15014 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
15017 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
15020 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15022 * config/powerpcspe: New port. Files are copied from the rs6000
15023 port, with "rs6000" in filenames replaced by "powerpcspe".
15025 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
15027 PR rtl-optimization/80754
15028 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
15030 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
15032 * config/sparc/sparc.md (length): Return the correct value for -mflat
15033 sibcalls to match output_sibcall.
15035 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
15039 * config/rs6000/rs6000.c (struct machine_function): Add new field
15041 (rs6000_get_separate_components): Init that field, use it.
15042 (rs6000_components_for_bb): Use the field.
15044 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
15046 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
15048 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
15050 PR middle-end/80823
15051 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
15053 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15056 * config/s390/s390.c (s390_check_qrst_address): Check incoming
15057 address against address_operand predicate.
15058 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
15060 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
15062 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
15063 parameters passed indirectly.
15065 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
15067 * config/i386/i386.md (*movdi_internal): Remove SSE4
15068 alternative 18 (?r, *v). Update insn attributes.
15069 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
15070 Update insn attributes.
15071 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
15072 Update insn attributes.
15073 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
15074 alternative 1 (r, v). Remove isa attribute.
15075 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
15076 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
15077 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
15079 2017-05-23 Tom de Vries <tom@codesourcery.com>
15081 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
15084 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15086 * cgraphunit.c (symbol_table::process_new_functions): Update.
15087 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
15088 (inline_generate_summary): Rename to ...
15089 (ipa_fn_summary_generate): ... this one.
15090 (inline_read_summary): Rename to ...
15091 (ipa_fn_summary_read): ... this one.
15092 (inline_write_summary): Rename to ...
15093 (ipa_fn_summary_write): ... this one.
15094 (inline_free_summary): Rename to ...
15095 (ipa_free_fn_summary): ... this one.
15096 (pass_data_local_fn_summary, pass_local_fn_summary,
15097 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
15098 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
15099 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
15100 make_pass_ipa_fn_summary): New.
15101 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
15102 inline_write_summary, inline_free_summary): Remove.
15103 (ipa_free_fn_summary) : New.
15104 * ipa-inline.c (ipa_inline): Update.
15105 (pass_ipa_inline): Do not generate summaries.
15106 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
15108 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
15109 and add pass_ipa_fn_summary.
15110 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
15112 (make_pass_inline_parameters): Remove.
15114 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
15116 * omp-low.c (struct omp_context): Remove "default_kind" member.
15119 * omp-offload.c (execute_oacc_device_lower): Remove the
15120 parallelism dimensions function attributes for unparallelized
15121 OpenACC kernels constructs.
15123 2017-05-23 Martin Liska <mliska@suse.cz>
15125 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
15127 (cgraph_edge::make_speculative): Likewise.
15128 (cgraph_edge::resolve_speculation): Likewise.
15129 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15130 (cgraph_node::dump): Likewise.
15131 * cgraph.h: Likewise.
15132 * cgraphunit.c (analyze_functions): Likewise.
15133 (symbol_table::compile): Likewise.
15134 * ipa-cp.c (print_all_lattices): Likewise.
15135 (determine_versionability): Likewise.
15136 (initialize_node_lattices): Likewise.
15137 (ipcp_verify_propagated_values): Likewise.
15138 (estimate_local_effects): Likewise.
15139 (update_profiling_info): Likewise.
15140 (create_specialized_node): Likewise.
15141 (perhaps_add_new_callers): Likewise.
15142 (decide_about_value): Likewise.
15143 (decide_whether_version_node): Likewise.
15144 (identify_dead_nodes): Likewise.
15145 (ipcp_store_bits_results): Likewise.
15146 * ipa-devirt.c (dump_targets): Likewise.
15147 (ipa_devirt): Likewise.
15148 * ipa-icf.c (sem_item::dump): Likewise.
15149 (sem_function::equals): Likewise.
15150 (sem_variable::equals): Likewise.
15151 (sem_item_optimizer::read_section): Likewise.
15152 (sem_item_optimizer::execute): Likewise.
15153 (congruence_class::dump): Likewise.
15154 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
15155 (dump_inline_summary): Likewise.
15156 (estimate_node_size_and_time): Likewise.
15157 (inline_analyze_function): Likewise.
15158 * ipa-inline-transform.c (inline_call): Likewise.
15159 * ipa-inline.c (report_inline_failed_reason): Likewise.
15160 (want_early_inline_function_p): Likewise.
15161 (edge_badness): Likewise.
15162 (update_edge_key): Likewise.
15163 (inline_small_functions): Likewise.
15164 * ipa-profile.c (ipa_profile): Likewise.
15165 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
15166 (ipa_make_edge_direct_to_target): Likewise.
15167 (remove_described_reference): Likewise.
15168 (ipa_impossible_devirt_target): Likewise.
15169 (propagate_controlled_uses): Likewise.
15170 (ipa_print_node_params): Likewise.
15171 (ipcp_transform_function): Likewise.
15172 * ipa-pure-const.c (pure_const_read_summary): Likewise.
15173 (propagate_pure_const): Likewise.
15174 * ipa-reference.c (generate_summary): Likewise.
15175 (read_write_all_from_decl): Likewise.
15176 (propagate): Likewise.
15177 (ipa_reference_read_optimization_summary): Likewise.
15178 * ipa-utils.c (ipa_merge_profiles): Likewise.
15179 * ipa.c (walk_polymorphic_call_targets): Likewise.
15180 (symbol_table::remove_unreachable_nodes): Likewise.
15181 (ipa_single_use): Likewise.
15182 * passes.c (execute_todo): Likewise.
15183 * predict.c (drop_profile): Likewise.
15184 * symtab.c (symtab_node::get_dump_name): New function.
15185 (symtab_node::dump_name): Likewise.
15186 (symtab_node::dump_asm_name): Likewise.
15187 (symtab_node::dump_references): Likewise.
15188 (symtab_node::dump_referring): Likewise.
15189 (symtab_node::dump_base): Likewise.
15190 (symtab_node::debug_symtab): Likewise.
15191 * tree-sra.c (convert_callers_for_node): Likewise.
15192 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
15193 * value-prof.c (init_node_map): Likewise.
15195 2017-05-23 Martin Liska <mliska@suse.cz>
15197 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
15198 and symtab_node::debug_symtab to symbol_table::debug.
15199 * cgraphunit.c (analyze_functions): Use the renamed function.
15200 (symbol_table::compile): Likewise.
15201 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
15202 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
15203 * passes.c (execute_todo): Likewise.
15204 * symtab.c (symbol_table::dump): New function.
15205 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
15207 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15209 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
15210 that nonconst implies exec.
15212 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
15214 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15215 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15216 (inline_edge_summary_vec): Turn into ...
15217 (ipa_call_summaries): ... this one.
15218 (redirect_to_unreachable, edge_set_predicate,
15219 evaluate_properties_for_edge, inline_summary_alloc,
15220 reset_ipa_call_summary, reset_inline_summary,
15221 inline_summary_t::duplicate): Update.
15222 (inline_edge_duplication_hook): Turn to ...
15223 (ipa_call_summary_t::duplicate): ... this one.
15224 (inline_edge_removal_hook): Turn to ...
15225 (ipa_call_summary_t::remove): ... this one.
15226 (dump_inline_edge_summary): Turn to ...
15227 (dump_ipa_call_summary): ... this one.
15228 (estimate_function_body_sizes): Update.
15229 (inline_update_callee_summaries): Update.
15230 (remap_edge_change_prob): Update.
15231 (remap_edge_summaries): Update.
15232 (inline_merge_summary): Update.
15233 (do_estimate_edge_time): Update.
15234 (inline_generate_summary): Update.
15235 (inline_read_section): Update.
15236 (inline_read_summary): Update.
15237 (inline_free_summary): Update.
15238 * ipa-inline.c (can_inline_edge_p): Update.
15239 (compute_inlined_call_time): Update.
15240 (want_inline_small_function_p): Update.
15241 (edge_badness): Update.
15242 (early_inliner): Update.
15243 * ipa-inline.h (inline_edge_summary): Turn to ...
15244 (ipa_call_summary): ... this one.
15245 (ipa_call_summary_t): New class.
15246 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15247 (ipa_call_summaries): New.
15248 (inline_edge_summary): Remove.
15249 (estimate_edge_growth): Update.
15250 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15251 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15252 * ipa-split.c (execute_split_functions): Update.
15253 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15255 2017-05-23 Tom de Vries <tom@codesourcery.com>
15257 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15258 attributes): Document rdrand effective target.
15260 2017-05-23 Tom de Vries <tom@codesourcery.com>
15262 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
15263 attributes): Sort alphabetically.
15265 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
15267 * config/avr/genmultilib.awk: Use gsub instead of gensub.
15269 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15272 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
15273 V2DF/V2DI splat into two separate patterns, one that handles
15274 registers, and the other that only handles memory. Drop support
15275 for splatting from a GPR on ISA 2.07 and then splitting the
15276 splat into direct move and splat.
15277 (vsx_splat_<mode>_reg): Likewise.
15278 (vsx_splat_<mode>_mem): Likewise.
15280 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
15282 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
15284 2017-05-22 Jakub Jelinek <jakub@redhat.com>
15286 PR middle-end/80809
15287 * omp-low.c (finish_taskreg_remap): New function.
15288 (finish_taskreg_scan): If unit size of ctx->record_type
15289 is non-constant, unshare the size expression and replace
15290 decls in it with possible outer var refs.
15292 PR middle-end/80809
15293 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
15294 GOVD_SHARED rather than GOVD_PRIVATE with it.
15295 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
15296 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
15298 PR middle-end/80853
15299 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
15300 as last argument to build_outer_var_ref for pointer bases of array
15301 section reductions.
15303 2017-05-19 Martin Sebor <msebor@redhat.com>
15305 * print-tree.c (print_node): Print DECL_READ_P flag.
15307 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15309 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
15310 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
15311 * cgraph.c: Likewise.
15312 * cgraphunit.c: Likewise.
15313 * gengtype.c: Likewise.
15314 * ipa-cp.c: Likewise.
15315 * ipa-devirt.c: Likewise.
15316 * ipa-icf.c: Likewise.
15317 * ipa-predicate.c: Likewise.
15318 * ipa-profile.c: Likewise.
15319 * ipa-prop.c: Likewise.
15320 * ipa-split.c: Likewise.
15322 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
15323 edge_predicate_pool, dump_inline_hints,
15324 inline_summary::account_size_time, redirect_to_unreachable,
15325 edge_set_predicate, set_hint_predicate,
15326 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
15327 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
15328 inline_summary_t::remove, remap_hint_predicate_after_duplication,
15329 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
15330 ipa_call_summary_t::remove, initialize_growth_caches,
15331 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
15332 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
15333 mark_modified, unmodified_parm_1, unmodified_parm,
15334 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
15335 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
15336 compute_bb_predicates, will_be_nonconstant_expr_predicate,
15337 will_be_nonconstant_predicate, record_modified_bb_info,
15338 get_minimal_bb, record_modified, param_change_prob,
15339 phi_result_unknown_predicate, predicate_for_phi_result,
15340 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
15341 estimate_function_body_sizes, compute_inline_parameters,
15342 compute_inline_parameters_for_curren, pass_data_inline_parameters,
15343 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
15344 inline_update_callee_summaries, remap_edge_change_prob,
15345 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
15346 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
15347 inline_analyze_function, inline_summary_t::insert,
15348 inline_generate_summary, read_ipa_call_summary, inline_read_section,
15349 inline_read_summary, write_ipa_call_summary, inline_write_summary,
15350 inline_free_summary): Move to ipa-fnsummary.h
15351 (predicate_t): Remove.
15352 * ipa-fnsummary.c: New file.
15353 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
15354 (enum inline_hints_vals, inline_hints, agg_position_info,
15355 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
15356 inline_summaries, ipa_call_summary, ipa_call_summary_t,
15357 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
15358 dump_inline_summary, dump_inline_hints, inline_generate_summary,
15359 inline_read_summary, inline_write_summary, inline_free_summary,
15360 inline_analyze_function, initialize_inline_failed,
15361 inline_merge_summary, inline_update_overall_summary,
15362 compute_inline_parameters): Move to ipa-fnsummary.h
15363 * ipa-fnsummary.h: New file.
15364 * ipa-inline-transform.h: Include ipa-inline.h.
15365 * ipa-inline.c: LIkewise.
15367 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15369 * ipa-inline.c (edge_badness): Use inlined_time instead of
15370 inline_summaries->get.
15372 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15374 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
15376 2017-05-22 Nathan Sidwell <nathan@acm.org>
15378 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
15379 (fdump-lang): Document 'raw' option.
15380 * dumpfile.h (TDI_tu): Delete.
15381 * dumpfile.c (dump_files): Remove translation-unit.
15382 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
15384 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
15386 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
15387 command option from $(AWK) call.
15388 * config/avr/genmultilib.awk: Simplify and rewrite so that it
15389 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
15390 [FORMAT]: Remove handling of variable.
15391 * config/avr/t-multilib: Regenerate.
15393 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15395 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
15397 (dump_inline_summary): Do not print self_time.
15398 (estimate_function_body_sizes): Do not set self_time.
15399 (compute_inline_parameters): Likewise.
15400 (inline_read_section, inline_write_summary): Do not stream self_time.
15401 * ipa-inline.h (inline_summary): Drop self_time.
15403 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
15405 * ipa-inline-analysis.c (account_size_time): Rename to ...
15406 (inline_summary::account_size_time): ... this one.
15407 (reset_ipa_call_summary): Turn to ...
15408 (ipa_call_summary::reset): ... this one.
15409 (reset_inline_summary): Turn to ...
15410 (inline_summary::reset): ... this one.
15411 (inline_summary_t::remove): Update.
15412 (inline_summary_t::duplicate): Update.
15413 (ipa_call_summary_t::remove): Update.
15414 (dump_inline_summary): Update.
15415 (estimate_function_body_sizes): Update.
15416 (compute_inline_parameters): Update.
15417 (estimate_node_size_and_time): Update.
15418 (inline_merge_summary): Update.
15419 (inline_update_overall_summary): Update.
15420 (inline_read_section): Update.
15421 (inline_write_summary): Update.
15422 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
15423 add account_size_time and reset member functions.
15424 (ipa_call_summary): Add reset function.
15425 * ipa-predicate.h (predicate::operator &): Constify.
15427 2017-05-22 Richard Biener <rguenther@suse.de>
15429 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
15431 2017-05-19 Jason Merrill <jason@redhat.com>
15433 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
15435 2017-05-19 Marek Polacek <polacek@redhat.com>
15438 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
15439 TYPE_OVERFLOW_WRAPS checks.
15441 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
15443 * tree-core.h (enum omp_clause_default_kind): Add
15444 "OMP_CLAUSE_DEFAULT_PRESENT".
15445 * tree-pretty-print.c (dump_omp_clause): Handle it.
15446 * gimplify.c (enum gimplify_omp_var_data): Add
15447 "GOVD_MAP_FORCE_PRESENT".
15448 (gimplify_adjust_omp_clauses_1): Map it to
15449 "GOMP_MAP_FORCE_PRESENT".
15450 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
15452 * gimplify.c (oacc_default_clause): Clarify.
15454 2017-05-19 Nathan Sidwell <nathan@acm.org>
15456 LANG_HOOK_REGISTER_DUMPS
15457 * toplev.c (general_init): Call register dump lang hook.
15458 * doc/invoke.texi: Document -fdump-lang option family.
15459 * dumpfile.c (dump_files): Remove class dump here.
15460 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
15461 * dumpfile.h (tree_dump_index): Remove TDI_class.
15462 * langhooks-def.h (lhd_register_dumps): Declare.
15463 (LANG_HOOKS_REGISTER_DUMPS): Define.
15464 (LANG_HOOKS_INITIALIZER): Add it.
15465 * langhooks.c (lhd_register_dumps): Define.
15466 * langhooks.h (struct lang_hooks): Add register_dumps.
15468 2017-05-19 Nathan Sidwell <nathan@acm.org>
15470 * context.h (context::set_passes): New.
15471 * context.c (context::context): Do not create pass manager.
15472 * toplev.c (general_init): Create pass manager here.
15474 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
15476 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
15477 use this splitter if two add or or instructions would also work for
15478 the constant we want to generate.
15480 2017-05-19 Richard Biener <rguenther@suse.de>
15483 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
15484 predicate evaluation.
15486 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15488 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
15490 * ipa-inline.c (want_inline_small_function_p): Do not cast to
15493 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
15495 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
15496 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
15497 (inline_edge_summary_vec): Turn into ...
15498 (ipa_call_summaries): ... this one.
15499 (redirect_to_unreachable, edge_set_predicate,
15500 evaluate_properties_for_edge, inline_summary_alloc,
15501 reset_ipa_call_summary, reset_inline_summary,
15502 inline_summary_t::duplicate): Update.
15503 (inline_edge_duplication_hook): Turn to ...
15504 (ipa_call_summary_t::duplicate): ... this one.
15505 (inline_edge_removal_hook): Turn to ...
15506 (ipa_call_summary_t::remove): ... this one.
15507 (dump_inline_edge_summary): Turn to ...
15508 (dump_ipa_call_summary): ... this one.
15509 (estimate_function_body_sizes): Update.
15510 (inline_update_callee_summaries): Update.
15511 (remap_edge_change_prob): Update.
15512 (remap_edge_summaries): Update.
15513 (inline_merge_summary): Update.
15514 (do_estimate_edge_time): Update.
15515 (inline_generate_summary): Update.
15516 (inline_read_section): Update.
15517 (inline_read_summary): Update.
15518 (inline_free_summary): Update.
15519 * ipa-inline.c (can_inline_edge_p): Update.
15520 (compute_inlined_call_time): Update.
15521 (want_inline_small_function_p): Update.
15522 (edge_badness): Update.
15523 (early_inliner): Update.
15524 * ipa-inline.h (inline_edge_summary): Turn to ...
15525 (ipa_call_summary): ... this one.
15526 (ipa_call_summary_t): New class.
15527 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
15528 (ipa_call_summaries): New.
15529 (inline_edge_summary): Remove.
15530 (estimate_edge_growth): Update.
15531 * ipa-profile.c (ipa_propagate_frequency_1): Update.
15532 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
15533 * ipa-split.c (execute_split_functions): Update.
15534 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
15536 2017-05-19 Richard Biener <rguenther@suse.de>
15538 PR middle-end/80764
15539 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
15541 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
15543 * config/rs6000/rs6000.c (struct machine_function): Add field
15544 fpr_is_wrapped_separately.
15545 (rs6000_get_separate_components): Use 64 components. Handle the
15546 new FPR components.
15547 (rs6000_components_for_bb): Handle the FPR components.
15548 (rs6000_emit_prologue_components): Handle the FPR components.
15549 (rs6000_emit_epilogue_components): Handle the FPR components.
15550 (rs6000_set_handled_components): Handle the FPR components.
15551 (rs6000_emit_prologue): Don't output prologue code for those FPRs
15552 that are already separately shrink-wrapped.
15553 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
15554 that are already separately shrink-wrapped.
15556 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15559 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
15562 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
15563 (define_peephole2 for Altivec d-form load): Add peepholes to catch
15564 cases where the register allocator uses a move and an offsettable
15565 memory operation to/from a FPR register on ISA 2.06/2.07.
15566 (define_peephole2 for Altivec d-form store): Likewise.
15568 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
15571 * config/i386/mmx.md (*mov<mode>_internal): Enable
15572 alternatives 11, 12, 13 and 14 also for 32bit targets.
15573 Remove alternatives 15, 16, 17 and 18.
15574 * config/i386/sse.md (vec_concatv2di): Change
15575 alternative (!x, *y) to (x, ?!*Yn).
15577 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
15579 * dumpfile.h (enum dump_kind): Remove stray comma.
15581 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15583 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
15584 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
15585 predicate::num_conditions
15586 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
15587 (CHANGED): turn into predicate::changed.
15588 (agg_position_info): Move to ipa-predicate.h
15589 (add_condition, predicate::add_clause, predicate::operator &=,
15590 predicate::or_with, predicate::evaluate, predicate::probability,
15591 dump_condition, dump_clause, predicate::dump,
15592 predicate::remap_after_duplication, predicate::remap_after_inlining,
15593 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
15594 (evaluate_conditions_for_known_args): Update.
15595 (set_cond_stmt_execution_predicate): Update.
15596 * ipa-inline.h: Include ipa-predicate.h
15597 (condition, inline_param_summary, conditions, agg_position_info,
15598 predicate): Move to ipa-predicate.h
15599 * ipa-predicate.c: New file.
15600 * ipa-predicate.h: New file.
15602 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
15604 * final.c (leaf_function_p): Check we are not in a sequence.
15606 2017-05-18 Martin Liska <mliska@suse.cz>
15608 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
15609 * dumpfile.c (dump_register): Use new enum dump_kind.
15610 (get_dump_file_name): Likewise.
15611 (dump_enable_all): Likewise.
15612 (dump_switch_p_1): Likewise.
15613 (enable_rtl_dump_file): Remove usage of TDF_RTL.
15614 * dumpfile.h (enum dump_kind): New enum type.
15615 (struct dump_file_info): Create constructor and
15616 format fields and comments.
15617 * passes.c (pass_manager::register_one_dump_file):
15619 * statistics.c (statistics_early_init): Likewise.
15620 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
15621 TDF_TREE with TDF_SLIM.
15622 (gather_memory_references_ref): Likewise.
15624 2017-05-18 Martin Liska <mliska@suse.cz>
15626 * vec.h (struct vnull): Use it.
15628 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
15630 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
15631 (true_predicate, false_predicate, true_predicate_p,
15632 false_predicate_p): Remove.
15633 (single_cond_predicate, not_inlined_predicate): Turn to member function
15635 (add_condition): Update.
15636 (add_clause): Turn to...
15637 (predicate::add_clause): ... this one; update; allow passing NULL
15639 (and_predicates): Turn to ...
15640 (predicate::operator &=): ... this one.
15641 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
15642 (or_predicates): Turn to ...
15643 (predicate::or_with): ... this one.
15644 (evaluate_predicate): Turn to ...
15645 (predicate::evaluate): ... this one.
15646 (predicate_probability): Turn to ...
15647 (predicate::probability): ... this one.
15648 (dump_condition): Update.
15649 (dump_predicate): Turn to ...
15650 (predicate::dump): ... this one.
15651 (account_size_time): Update.
15652 (edge_set_predicate): Update.
15653 (set_hint_predicate): UPdate.
15654 (evaluate_conditions_for_known_args): Update.
15655 (evaluate_properties_for_edge): Update.
15656 (remap_predicate_after_duplication): Turn to...
15657 (predicate::remap_after_duplication): ... this one.
15658 (remap_hint_predicate_after_duplication): Update.
15659 (inline_summary_t::duplicate): UPdate.
15660 (dump_inline_edge_summary): Update.
15661 (dump_inline_summary): Update.
15662 (set_cond_stmt_execution_predicate): Update.
15663 (set_switch_stmt_execution_predicate): Update.
15664 (compute_bb_predicates): Update.
15665 (will_be_nonconstant_expr_predicate): Update.
15666 (will_be_nonconstant_predicate): Update.
15667 (phi_result_unknown_predicate): Update.
15668 (predicate_for_phi_result): Update.
15669 (array_index_predicate): Update.
15670 (estimate_function_body_sizes): Update.
15671 (estimate_node_size_and_time): Update.
15672 (estimate_ipcp_clone_size_and_time): Update.
15673 (remap_predicate): Rename to ...
15674 (predicate::remap_after_inlining): ... this one.
15675 (remap_hint_predicate): Update.
15676 (inline_merge_summary): Update.
15677 (inline_update_overall_summary): Update.
15678 (estimate_size_after_inlining): Update.
15679 (read_predicate): Rename to ...
15680 (predicate::stream_in): ... this one.
15681 (read_inline_edge_summary): Update.
15682 (write_predicate): Rename to ...
15683 (predicate::stream_out): ... this one.
15684 (write_inline_edge_summary): Update.
15685 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
15686 (clause_t): Turn to uint32_t
15687 (predicate): Turn to class; implement constructor and operators
15689 (size_time_entry): Update.
15690 (inline_summary): Update.
15691 (inline_edge_summary): Update.
15693 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
15695 * fold-const.c (fold_binary_loc): Move transformation...
15696 * match.pd (C - X CMP X): ... here.
15698 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
15700 * config/sparc/sparc.c (sparc_option_override): Set function
15701 alignment for -mcpu=niagara7 to 64 to match the I$ line.
15702 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
15704 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
15706 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
15708 2017-05-18 Marek Polacek <polacek@redhat.com>
15711 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
15712 (pass_ubsan::execute): Call gimple_assign_single_p instead of
15713 gimple_assign_load_p.
15715 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
15717 PR middle-end/80692
15718 * real.c (do_compare): Give decimal_do_compare preference over
15719 comparing just the signs.
15721 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
15723 * doc/md.texi (Canonicalization of Instructions): Describe the
15724 canonical form of instructions that inherently set a condition
15727 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
15729 PR middle-end/80775
15730 * tree-cfg.c: Move deletion of unreachable case statements to after
15731 the merging of consecutive case labels.
15733 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
15735 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
15736 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
15737 restoring of callee-saved registers.
15739 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
15741 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
15742 * config/visium/visium.c (single_set_and_flags): Likewise.
15743 * config/visium/visium.md (Substitutions): Likewise.
15745 2017-05-17 Martin Liska <mliska@suse.cz>
15747 * cfg.c: Introduce dump_flags_t type and
15748 use it instead of int type.
15750 * cfghooks.c: Likewise.
15751 * cfghooks.h (struct cfg_hooks): Likewise.
15752 * cfgrtl.c: Likewise.
15753 * cfgrtl.h: Likewise.
15754 * cgraph.c (cgraph_node::get_body): Likewise.
15755 * coretypes.h: Likewise.
15756 * domwalk.c: Likewise.
15757 * domwalk.h: Likewise.
15758 * dumpfile.c (struct dump_option_value_info): Likewise.
15759 (dump_enable_all): Likewise.
15760 (dump_switch_p_1): Likewise.
15761 (opt_info_switch_p): Likewise.
15762 * dumpfile.h (enum tree_dump_index): Likewise.
15763 (struct dump_file_info): Likewise.
15764 * genemit.c: Likewise.
15765 * generic-match-head.c: Likewise.
15766 * gengtype.c (open_base_files): Likewise.
15767 * gimple-pretty-print.c: Likewise.
15768 * gimple-pretty-print.h: Likewise.
15769 * graph.c (print_graph_cfg): Likewise.
15770 * graphite-scop-detection.c (dot_all_sese): Likewise.
15771 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
15772 * loop-unroll.c (report_unroll): Likewise.
15773 * passes.c (pass_manager::register_one_dump_file): Likewise.
15774 * print-tree.c: Likewise.
15775 * statistics.c: Likewise.
15776 * tree-cfg.c: Likewise.
15777 * tree-cfg.h: Likewise.
15778 * tree-dfa.c: Likewise.
15779 * tree-dfa.h: Likewise.
15780 * tree-dump.c (dump_function): Likewise.
15781 * tree-dump.h (struct dump_info): Likewise.
15782 * tree-pretty-print.c: Likewise.
15783 * tree-pretty-print.h: Likewise.
15784 * tree-ssa-live.c: Likewise.
15785 * tree-ssa-live.h: Likewise.
15786 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
15787 * tree-vect-loop.c: Likewise.
15788 * tree-vect-slp.c: Likewise.
15790 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
15791 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15793 PR tree-optimization/80457
15794 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
15795 of all arguments to a statement as scalar_to_vec operations.
15796 (vectorizable_call): Adjust call to vect_model_simple_cost for
15798 (vectorizable_conversion): Likewise.
15799 (vectorizable_assignment): Likewise.
15800 (vectorizable_shift): Likewise.
15801 (vectorizable_operation): Likewise.
15802 (vectorizable_comparison): Likewise.
15803 (vect_is_simple_cond): Record the def types for operands.
15804 (vectorizable_condition): Likewise, call vect_model_simple_cost.
15805 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
15806 for statement argument count.
15808 2017-05-16 Carl Love <cel@us.ibm.com>
15810 * config/rs6000/rs6000-c: Add support for built-in functions
15811 vector unsigned long long vec_bperm (vector unsigned long long,
15812 vector unsigned char)
15813 vector signed long long vec_mule (vector signed int,
15815 vector unsigned long long vec_mule (vector unsigned int,
15816 vector unsigned int)
15817 vector signed long long vec_mulo (vector signed int,
15819 vector unsigned long long vec_mulo (vector unsigned int,
15820 vector unsigned int)
15821 vector signed char vec_sldw (vector signed char,
15822 vector signed char,
15824 vector unsigned char vec_sldw (vector unsigned char,
15825 vector unsigned char,
15827 vector signed short vec_sldw (vector signed short,
15828 vector signed short,
15830 vector unsigned short vec_sldw (vector unsigned short,
15831 vector unsigned short,
15833 vector signed int vec_sldw (vector signed int,
15836 vector unsigned int vec_sldw (vector unsigned int,
15837 vector unsigned int,
15839 vector signed long long vec_sldw (vector signed long long,
15840 vector signed long long,
15842 vector unsigned long long vec_sldw (vector unsigned long long,
15843 vector unsigned long long,
15845 * config/rs6000/rs6000-c: Add support for built-in functions
15846 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
15847 * config/rs6000/altivec.h: Add defintion for vec_sldw.
15848 * doc/extend.texi: Update the built-in documentation for the
15849 new built-in functions.
15851 2017-05-16 Marek Polacek <polacek@redhat.com>
15855 * tree.c (save_expr): Don't fold the expression.
15857 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
15859 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
15860 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
15861 and (?*y,m). Update insn attributes.
15863 2017-05-16 Martin Liska <mliska@suse.cz>
15865 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
15866 flags argument of print_gimple_stmt, print_gimple_expr,
15867 print_generic_stmt and print_generic_expr.
15868 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
15869 * coretypes.h: Likewise.
15870 * except.c (dump_eh_tree): Likewise.
15871 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15872 * gimple-pretty-print.h: Likewise.
15873 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
15874 (backprop::push_to_worklist): Likewise.
15875 (backprop::pop_from_worklist): Likewise.
15876 (backprop::process_use): Likewise.
15877 (backprop::intersect_uses): Likewise.
15878 (note_replacement): Likewise.
15879 * gimple-ssa-store-merging.c
15880 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
15881 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
15882 (pass_store_merging::execute): Likewise.
15883 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
15884 (ssa_base_cand_dump_callback): Likewise.
15885 (dump_incr_vec): Likewise.
15886 (replace_refs): Likewise.
15887 (replace_mult_candidate): Likewise.
15888 (create_add_on_incoming_edge): Likewise.
15889 (create_phi_basis): Likewise.
15890 (insert_initializers): Likewise.
15891 (all_phi_incrs_profitable): Likewise.
15892 (introduce_cast_before_cand): Likewise.
15893 (replace_one_candidate): Likewise.
15894 * gimplify.c (gimplify_expr): Likewise.
15895 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
15896 (set_rename): Likewise.
15897 (rename_uses): Likewise.
15898 (copy_loop_phi_nodes): Likewise.
15899 (add_close_phis_to_merge_points): Likewise.
15900 (copy_loop_close_phi_args): Likewise.
15901 (copy_cond_phi_args): Likewise.
15902 (graphite_copy_stmts_from_block): Likewise.
15903 (translate_pending_phi_nodes): Likewise.
15904 * graphite-poly.c (print_pdr): Likewise.
15905 (dump_gbb_cases): Likewise.
15906 (dump_gbb_conditions): Likewise.
15907 (print_scop_params): Likewise.
15908 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
15909 (build_cross_bb_scalars_use): Likewise.
15910 (gather_bbs::before_dom_children): Likewise.
15911 * hsa-dump.c (dump_hsa_immed): Likewise.
15912 * ipa-cp.c (print_ipcp_constant_value): Likewise.
15913 (get_replacement_map): Likewise.
15914 * ipa-inline-analysis.c (dump_condition): Likewise.
15915 (estimate_function_body_sizes): Likewise.
15916 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
15917 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
15918 * ipa-prop.c (ipa_dump_param): Likewise.
15919 (ipa_print_node_jump_functions_for_edge): Likewise.
15920 (ipa_modify_call_arguments): Likewise.
15921 (ipa_modify_expr): Likewise.
15922 (ipa_dump_param_adjustments): Likewise.
15923 (ipa_dump_agg_replacement_values): Likewise.
15924 (ipcp_modif_dom_walker::before_dom_children): Likewise.
15925 * ipa-pure-const.c (check_stmt): Likewise.
15926 (pass_nothrow::execute): Likewise.
15927 * ipa-split.c (execute_split_functions): Likewise.
15928 * omp-offload.c (dump_oacc_loop_part): Likewise.
15929 (dump_oacc_loop): Likewise.
15930 * trans-mem.c (tm_log_emit): Likewise.
15931 (tm_memopt_accumulate_memops): Likewise.
15932 (dump_tm_memopt_set): Likewise.
15933 (dump_tm_memopt_transform): Likewise.
15934 * tree-cfg.c (gimple_verify_flow_info): Likewise.
15935 (print_loop): Likewise.
15936 * tree-chkp-opt.c (chkp_print_addr): Likewise.
15937 (chkp_gather_checks_info): Likewise.
15938 (chkp_get_check_result): Likewise.
15939 (chkp_remove_check_if_pass): Likewise.
15940 (chkp_use_outer_bounds_if_possible): Likewise.
15941 (chkp_reduce_bounds_lifetime): Likewise.
15942 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
15943 (chkp_mark_completed_bounds): Likewise.
15944 (chkp_register_incomplete_bounds): Likewise.
15945 (chkp_mark_invalid_bounds): Likewise.
15946 (chkp_maybe_copy_and_register_bounds): Likewise.
15947 (chkp_build_returned_bound): Likewise.
15948 (chkp_get_bound_for_parm): Likewise.
15949 (chkp_build_bndldx): Likewise.
15950 (chkp_get_bounds_by_definition): Likewise.
15951 (chkp_generate_extern_var_bounds): Likewise.
15952 (chkp_get_bounds_for_decl_addr): Likewise.
15953 * tree-chrec.c (chrec_apply): Likewise.
15954 * tree-data-ref.c (dump_data_reference): Likewise.
15955 (dump_subscript): Likewise.
15956 (dump_data_dependence_relation): Likewise.
15957 (analyze_overlapping_iterations): Likewise.
15958 * tree-inline.c (expand_call_inline): Likewise.
15959 (tree_function_versioning): Likewise.
15960 * tree-into-ssa.c (dump_defs_stack): Likewise.
15961 (dump_currdefs): Likewise.
15962 (dump_names_replaced_by): Likewise.
15963 (dump_update_ssa): Likewise.
15964 (update_ssa): Likewise.
15965 * tree-object-size.c (pass_object_sizes::execute): Likewise.
15966 * tree-parloops.c (build_new_reduction): Likewise.
15967 (try_create_reduction_list): Likewise.
15968 (ref_conflicts_with_region): Likewise.
15969 (oacc_entry_exit_ok_1): Likewise.
15970 (oacc_entry_exit_single_gang): Likewise.
15971 * tree-pretty-print.h: Likewise.
15972 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
15973 (get_scalar_evolution): Likewise.
15974 (add_to_evolution): Likewise.
15975 (get_loop_exit_condition): Likewise.
15976 (analyze_evolution_in_loop): Likewise.
15977 (analyze_initial_condition): Likewise.
15978 (analyze_scalar_evolution): Likewise.
15979 (instantiate_scev): Likewise.
15980 (number_of_latch_executions): Likewise.
15981 (gather_chrec_stats): Likewise.
15982 (final_value_replacement_loop): Likewise.
15983 (scev_const_prop): Likewise.
15984 * tree-sra.c (dump_access): Likewise.
15985 (disqualify_candidate): Likewise.
15986 (create_access): Likewise.
15987 (reject): Likewise.
15988 (maybe_add_sra_candidate): Likewise.
15989 (create_access_replacement): Likewise.
15990 (analyze_access_subtree): Likewise.
15991 (analyze_all_variable_accesses): Likewise.
15992 (sra_modify_assign): Likewise.
15993 (initialize_constant_pool_replacements): Likewise.
15994 (find_param_candidates): Likewise.
15995 (decide_one_param_reduction): Likewise.
15996 (replace_removed_params_ssa_names): Likewise.
15997 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
15998 * tree-ssa-copy.c (dump_copy_of): Likewise.
15999 (copy_prop_visit_cond_stmt): Likewise.
16000 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
16001 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
16002 (record_equivalences_from_stmt): Likewise.
16003 * tree-ssa-dse.c (compute_trims): Likewise.
16004 (delete_dead_call): Likewise.
16005 (delete_dead_assignment): Likewise.
16006 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
16007 (forward_propagate_into_cond): Likewise.
16008 (pass_forwprop::execute): Likewise.
16009 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16010 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
16012 (move_computations_worker): Likewise.
16013 (execute_sm): Likewise.
16014 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
16015 (remove_exits_and_undefined_stmts): Likewise.
16016 (remove_redundant_iv_tests): Likewise.
16017 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
16018 (adjust_iv_update_pos): Likewise.
16019 * tree-ssa-math-opts.c (bswap_replace): Likewise.
16020 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
16021 (value_replacement): Likewise.
16022 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
16023 * tree-ssa-pre.c (print_pre_expr): Likewise.
16024 (get_representative_for): Likewise.
16025 (create_expression_by_pieces): Likewise.
16026 (insert_into_preds_of_block): Likewise.
16027 (eliminate_insert): Likewise.
16028 (eliminate_dom_walker::before_dom_children): Likewise.
16029 (eliminate): Likewise.
16030 (remove_dead_inserted_code): Likewise.
16031 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
16032 * tree-ssa-reassoc.c (get_rank): Likewise.
16033 (eliminate_duplicate_pair): Likewise.
16034 (eliminate_plus_minus_pair): Likewise.
16035 (eliminate_not_pairs): Likewise.
16036 (undistribute_ops_list): Likewise.
16037 (eliminate_redundant_comparison): Likewise.
16038 (update_range_test): Likewise.
16039 (optimize_range_tests_var_bound): Likewise.
16040 (optimize_vec_cond_expr): Likewise.
16041 (rewrite_expr_tree): Likewise.
16042 (rewrite_expr_tree_parallel): Likewise.
16043 (linearize_expr): Likewise.
16044 (break_up_subtract): Likewise.
16045 (linearize_expr_tree): Likewise.
16046 (attempt_builtin_powi): Likewise.
16047 (attempt_builtin_copysign): Likewise.
16048 (transform_stmt_to_copy): Likewise.
16049 (transform_stmt_to_multiply): Likewise.
16050 (dump_ops_vector): Likewise.
16051 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
16052 (print_scc): Likewise.
16053 (set_ssa_val_to): Likewise.
16054 (visit_reference_op_store): Likewise.
16055 (visit_use): Likewise.
16056 (sccvn_dom_walker::before_dom_children): Likewise.
16057 (run_scc_vn): Likewise.
16058 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16060 (expr_hash_elt::print): Likewise.
16061 (const_and_copies::pop_to_marker): Likewise.
16062 (const_and_copies::record_const_or_copy_raw): Likewise.
16063 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
16064 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
16065 (dump_predicates): Likewise.
16066 (find_uninit_use): Likewise.
16067 (warn_uninitialized_phi): Likewise.
16068 (pass_late_warn_uninitialized::execute): Likewise.
16069 * tree-ssa.c (verify_vssa): Likewise.
16070 (verify_ssa): Likewise.
16071 (maybe_optimize_var): Likewise.
16072 * tree-vrp.c (dump_value_range): Likewise.
16073 (dump_all_value_ranges): Likewise.
16074 (dump_asserts_for): Likewise.
16075 (register_edge_assert_for_2): Likewise.
16076 (vrp_visit_cond_stmt): Likewise.
16077 (vrp_visit_switch_stmt): Likewise.
16078 (vrp_visit_stmt): Likewise.
16079 (vrp_visit_phi_node): Likewise.
16080 (simplify_cond_using_ranges_1): Likewise.
16081 (fold_predicate_in): Likewise.
16082 (evrp_dom_walker::before_dom_children): Likewise.
16083 (evrp_dom_walker::push_value_range): Likewise.
16084 (evrp_dom_walker::pop_value_range): Likewise.
16085 (execute_early_vrp): Likewise.
16087 2017-05-16 Richard Biener <rguenther@suse.de>
16089 * dwarf2out.c (loc_list_from_tree_1): Do not create
16090 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
16092 2017-05-16 Richard Biener <rguenther@suse.de>
16094 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
16096 (note_variable_value_in_expr): If we resolved the decl ref
16097 do not push to the stack.
16099 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
16101 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
16102 operations in fast-math mode.
16103 (vaddq_f16): Likewise.
16104 (vmul_f16): Likewise.
16105 (vmulq_f16): Likewise.
16106 (vsub_f16): Likewise.
16107 (vsubq_f16): Likewise.
16108 * config/arm/neon.md (add<mode>3): New.
16110 (fma:<VH:mode>3): New. Also remove outdated comment.
16113 2017-05-16 Martin Liska <mliska@suse.cz>
16117 * ipa-devirt.c (warn_types_mismatch): Fix typo.
16118 (odr_types_equivalent_p): Likewise.
16120 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
16122 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
16124 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
16127 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
16128 non-interunit SSE move alternatives with '?'.
16129 (zero-extendsidi peephole2): New peephole to skip intermediate
16130 general register in SSE zero-extend sequence.
16132 2017-05-15 Jeff Law <law@redhat.com>
16134 * reorg.c (relax_delay_slots): Create a new variable to hold
16135 the temporary target rather than clobbering TARGET_LABEL.
16137 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
16138 missing argument to extract_bit_field call.
16139 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
16141 2017-05-15 Martin Liska <mliska@suse.cz>
16144 * gcc.c (process_command): Do not allow empty argument of -o option.
16146 2017-05-15 Renlin Li <renlin.li@arm.com>
16148 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
16149 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
16150 * config/aarch64/constraints.md (Usf): Add long call check.
16151 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
16152 (call_value): Likewise.
16153 (sibcall): Likewise.
16154 (sibcall_value): Likewise.
16156 (call_value_insn): New.
16157 (sibcall_insn): Update rtx pattern.
16158 (sibcall_value_insn): Likewise.
16159 (call_internal): Remove.
16160 (call_value_internal): Likewise.
16161 (sibcall_internal): Likewise.
16162 (sibcall_value_internal): Likewise.
16163 (call_reg): Likewise.
16164 (call_symbol): Likewise.
16165 (call_value_reg): Likewise.
16166 (call_value_symbol): Likewise.
16168 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
16171 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
16173 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
16175 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
16176 compatible with CCGOCmode and with CCZmode.
16178 2017-05-14 Martin Sebor <msebor@redhat.com>
16180 PR middle-end/77671
16181 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
16182 (gimple_fold_builtin_snprintf): Same.
16183 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
16184 (gimple_fold_builtin_snprintf): Same.
16185 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
16186 of character types.
16187 (is_call_safe): New function.
16188 (try_substitute_return_value): Call it.
16189 (try_simplify_call): New function.
16190 (pass_sprintf_length::handle_gimple_call): Call it.
16192 2017-05-14 Martin Sebor <msebor@redhat.com>
16194 PR middle-end/80669
16195 * builtins.c (expand_builtin_stpncpy): Simplify.
16197 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
16199 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
16200 * config/i386/i386.h
16201 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
16202 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
16203 (struct machine_function): Add new members call_ms2sysv,
16204 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
16205 (struct machine_frame_state): New fields sp_realigned and
16206 sp_realigned_offset.
16207 * config/i386/i386.c
16208 (enum xlogue_stub): New enum.
16209 (enum xlogue_stub_sets): New enum.
16210 (class xlogue_layout): New class.
16211 (struct ix86_frame): New fields stack_realign_allocate_offset,
16212 stack_realign_offset and outlined_save_offset. Modify comments to
16213 detail stack layout when using out-of-line stubs.
16214 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
16215 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
16216 -mcall-ms2sysv-xlogues.
16217 (stub_managed_regs): New static variable.
16218 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
16219 registers managed by out-of-line stub.
16220 (disable_call_ms2sysv_xlogues): New function.
16221 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
16222 m->call_ms2sysv when appropriate and compute frame layout for
16224 (sp_valid_at, fp_valid_at): New inline functions.
16225 (choose_basereg): New function.
16226 (choose_baseaddr): Add align parameter, use choose_basereg and modify
16228 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
16229 Use align parameter of choose_baseaddr to generated aligned SSE movs
16231 (pro_epilogue_adjust_stack): Modify to track
16232 machine_frame_state::sp_realigned.
16233 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
16234 (ix86_nsaved_sseregs): Likewise.
16235 (ix86_emit_save_regs): Likewise.
16236 (ix86_emit_save_regs_using_mov): Likewise.
16237 (ix86_emit_save_sse_regs_using_mov): Likewise.
16238 (get_scratch_register_on_entry): Likewise.
16239 (gen_frame_set): New function.
16240 (gen_frame_load): Likewise.
16241 (gen_frame_store): Likewise.
16242 (emit_outlined_ms2sysv_save): Likewise.
16243 (emit_outlined_ms2sysv_restore): Likewise.
16244 (ix86_expand_prologue): Modify stack re-alignment code and call
16245 emit_outlined_ms2sysv_save when appropriate.
16246 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
16247 parameter rtx_insn *insn, which allows the function to be used to only
16248 generate the notes.
16249 (ix86_expand_epilogue): Modify validity checks of frame and stack
16250 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
16251 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
16252 * config/i386/predicates.md
16253 (save_multiple): New predicate.
16254 (restore_multiple): Likewise.
16255 * config/i386/sse.md
16256 (save_multiple<mode>): New pattern.
16257 (save_multiple_realign<mode>): Likewise.
16258 (restore_multiple<mode>): Likewise.
16259 (restore_multiple_and_return<mode>): Likewise.
16260 (restore_multiple_leave_return<mode>): Likewise.
16261 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
16263 2017-05-14 Julia Koval <julia.koval@intel.com>
16265 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
16266 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
16267 (__builtin_ia32_xsetbv): New builtins.
16268 * config/i386/i386.c (ix86_expand_special_args_builtin):
16270 (ix86_expand_builtin): Special expand for new intrinsics.
16271 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
16272 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
16273 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
16275 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16277 * cfganal.c (inverted_post_order_compute): Change argument type
16279 * cfganal.h (inverted_post_order_compute): Adjust prototype.
16280 * df-core.c (rest_of_handle_df_initialize): Adjust.
16281 (rest_of_handle_df_finish): Likewise.
16282 (df_analyze_1): Likewise.
16283 (df_analyze): Likewise.
16284 (loop_inverted_post_order_compute): Change argument to be a vec *.
16285 (df_analyze_loop): Adjust.
16286 (df_get_n_blocks): Likewise.
16287 (df_get_postorder): Likewise.
16288 * df.h (struct df_d): Change field to be a vec.
16289 * lcm.c (compute_laterin): Adjust.
16290 (compute_available): Likewise.
16291 * lra-lives.c (lra_create_live_ranges_1): Likewise.
16292 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
16293 * tree-ssa-pre.c (compute_antic): Likewise.
16295 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16297 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
16298 (depth_first_search::depth_first_search): Change structure init
16299 function to this constructor.
16300 (depth_first_search::add_bb): Rename function to this member.
16301 (depth_first_search::execute): Likewise.
16302 (flow_dfs_compute_reverse_finish): Adjust.
16304 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16306 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
16307 (longest_simple_path): Likewise.
16308 * shrink-wrap.c (spread_components): Likewise.
16309 (disqualify_problematic_components): Likewise.
16310 (emit_common_heads_for_components): Likewise.
16311 (emit_common_tails_for_components): Likewise.
16312 (insert_prologue_epilogue_for_components): Likewise.
16314 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16316 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
16319 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16321 * df-core.c (df_set_blocks): Start using auto_bitmap.
16322 (df_compact_blocks): Likewise.
16323 * df-problems.c (df_rd_confluence_n): Likewise.
16324 * df-scan.c (df_insn_rescan_all): Likewise.
16325 (df_process_deferred_rescans): Likewise.
16326 (df_update_entry_block_defs): Likewise.
16327 (df_update_exit_block_uses): Likewise.
16328 (df_entry_block_bitmap_verify): Likewise.
16329 (df_exit_block_bitmap_verify): Likewise.
16330 (df_scan_verify): Likewise.
16331 * lra-constraints.c (lra_constraints): Likewise.
16332 (undo_optional_reloads): Likewise.
16333 (lra_undo_inheritance): Likewise.
16334 * lra-remat.c (calculate_gen_cands): Likewise.
16335 (do_remat): Likewise.
16336 * lra-spills.c (assign_spill_hard_regs): Likewise.
16337 (spill_pseudos): Likewise.
16338 * tree-ssa-pre.c (bitmap_set_and): Likewise.
16339 (bitmap_set_subtract_values): Likewise.
16341 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16343 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
16344 management with auto_bitmap.
16345 (fix_inter_tick): Likewise.
16346 (fix_recovery_deps): Likewise.
16347 * ira.c (add_store_equivs): Likewise.
16348 (find_moveable_pseudos): Likewise.
16349 (split_live_ranges_for_shrink_wrap): Likewise.
16350 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
16351 (rtx_reuse_manager::seen_def_p): Likewise.
16352 (rtx_reuse_manager::set_seen_def): Likewise.
16353 * print-rtl.h (class rtx_reuse_manager): Likewise.
16355 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16357 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
16359 (migrate_btr_def): Likewise.
16360 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
16361 * df-core.c (loop_post_order_compute): Likewise.
16362 (loop_inverted_post_order_compute): Likewise.
16363 * hsa-common.h: Likewise.
16364 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
16365 * init-regs.c (initialize_uninitialized_regs): Likewise.
16366 * ipa-inline.c (resolve_noninline_speculation): Likewise.
16367 (inline_small_functions): Likewise.
16368 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
16369 * ira.c (combine_and_move_insns): Likewise.
16370 (build_insn_chain): Likewise.
16371 * loop-invariant.c (find_invariants): Likewise.
16372 * lower-subreg.c (propagate_pseudo_copies): Likewise.
16373 * predict.c (tree_predict_by_opcode): Likewise.
16374 (predict_paths_leading_to): Likewise.
16375 (predict_paths_leading_to_edge): Likewise.
16376 (estimate_loops_at_level): Likewise.
16377 (estimate_loops): Likewise.
16378 * shrink-wrap.c (try_shrink_wrapping): Likewise.
16379 (spread_components): Likewise.
16380 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
16381 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
16382 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
16383 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
16384 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
16385 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
16386 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
16387 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
16388 (mark_threaded_blocks): Likewise.
16389 (thread_through_all_blocks): Likewise.
16390 * tree-ssa.c (verify_ssa): Likewise.
16391 (execute_update_addresses_taken): Likewise.
16392 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
16394 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16396 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
16398 (post_order_compute): Likewise.
16399 (inverted_post_order_compute): Likewise.
16400 (pre_and_rev_post_order_compute_fn): Likewise.
16402 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16404 * genrecog.c (int_set::int_set): Explicitly construct our
16405 auto_vec base class.
16406 * vec.h (auto_vec::auto_vec): New constructor.
16408 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16410 * bitmap.h (class auto_bitmap): New constructor taking
16411 bitmap_obstack * argument.
16413 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16415 * bitmap.h (class auto_bitmap): Change type of m_bits to
16416 bitmap_head, and adjust ctor / dtor and member operators.
16418 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
16420 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
16421 when returned register mode doesn't match original mode.
16423 2017-05-12 Jeff Law <law@redhat.com>
16424 Jakub Jelinek <jakub@redhat.com>
16426 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
16427 we look for cc setter after the compare-elim changes.
16428 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
16429 within the vector to match what compare-elim now expects.
16430 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
16431 (xorsi3_flags, one_cmplsi2_flags): Likewise.
16433 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
16434 after the compare-elim changes.
16435 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
16436 the vector to match what compare-elim now expects.
16437 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
16438 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
16439 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
16440 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
16441 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
16443 * config/visium/visium.c (single_set_and_flags): Fix where
16444 we look for cc setter after the compare-elim changes.
16445 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
16446 with the vector to match what compare-elim now expects.
16447 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
16448 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
16449 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
16450 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
16451 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
16452 (neg<mode>2_insn_set_overflow): Likewise.
16454 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
16456 PR middle-end/79794
16457 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
16458 maybe_expand_insn call, set ops[0].target. If still set after call,
16459 set alt_rtl. Add extra arg to recursive calls.
16460 (extract_bit_field): Add alt_rtl argument. Pass to
16462 * expmed.h (extract_bit_field): Fix prototype.
16463 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
16464 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
16465 to extract_bit_field_calls.
16466 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
16467 Pass alt_rtl to extract_bit_field calls.
16468 * calls.c (store_unaligned_arguments_into_psuedos)
16469 load_register_parameters): Pass extra NULL to extract_bit_field calls.
16470 * optabs.c (maybe_legitimize_operand): Clear op->target when call
16472 * optabs.h (struct expand_operand): Add target bitfield.
16474 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16476 * compare-elim.c (try_eliminate_compare): Canonicalize
16477 operation with embedded compare to
16478 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
16479 (set (reg) (operation)].
16481 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
16483 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
16486 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
16487 cost of adding a carry flag for ADC instruction.
16488 [case MINUS]: Ignore the cost of subtracting a carry flag
16489 for SBB instruction.
16491 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
16493 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
16495 * config/rs6000/bmiintrin.h: New file.
16496 * config/rs6000/bmi2intrin.h: New file.
16497 * config/rs6000/x86intrin.h: New file.
16499 2017-05-12 Jeff Law <law@redhat.com>
16501 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
16504 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
16506 PR middle-end/80707
16507 * tree-cfg.c: Remove cfg edges of unreachable case statements.
16509 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16511 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16512 early expansion of vector divide builtins.
16513 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
16514 builtins identified as having unsigned arguments.
16516 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16518 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
16519 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
16520 expansion of vector logical operations (and, andc, or, xor,
16523 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16525 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
16526 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
16528 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16530 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
16531 early GIMPLE expansion of vector multiplies.
16533 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16535 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
16536 TARGET_HAVE_MOVT conditional.
16537 (movt splitter): Likewise.
16539 2017-05-12 Richard Biener <rguenther@suse.de>
16541 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
16542 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16543 Fold all stmts not inplace.
16545 2017-05-12 Richard Biener <rguenther@suse.de>
16547 PR tree-optimization/80713
16548 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
16549 inserted_exprs bit for not removed stmts.
16551 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
16553 PR middle-end/69921
16554 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
16555 parallelized" attribute for parallelized OpenACC kernels.
16556 * omp-offload.c (execute_oacc_device_lower): Use it.
16558 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
16559 Set "oacc kernels" attribute.
16560 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
16561 parameter. Adjust all users.
16562 (oacc_fn_attrib_kernels_p): Remove function.
16563 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
16564 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
16565 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
16566 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
16567 assert "oacc kernels" attribute is set.
16569 2017-05-11 Carl Love <cel@us.ibm.com>
16571 * config/rs6000/rs6000-c: Add support for built-in functions
16572 vector unsigned char vec_popcnt (vector signed char)
16573 vector unsigned char vec_popcnt (vector unsigned char)
16574 vector unsigned short vec_popcnt (vector signed short)
16575 vector unsigned short vec_popcnt (vector unsigned short)
16576 vector unsigned int vec_popcnt (vector signed int)
16577 vector unsigned int vec_popcnt (vector unsigned int)
16578 vector unsigned long long vec_popcnt (vector signed long long)
16579 vector unsigned long long vec_popcnt (vector unsigned long long)
16580 vector signed long long vec_slo (vector signed long long,
16581 vector signed char)
16582 vector signed long long vec_slo (vector signed long long,
16583 vector unsigned char)
16584 vector unsigned long long vec_slo (vector unsigned long long,
16585 vector signed char)
16586 vector unsigned long long vec_slo (vector unsigned long long,
16587 vector unsigned char)
16588 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
16589 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
16590 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
16591 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
16592 * doc/extend.texi: Update the built-in documentation file for the
16593 new built-in functions.
16595 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16597 * attribs.h (sorted_attr_string): Move machine independent
16598 functions for target clone support from the i386 port to common
16599 code. Rename ix86_function_versions to common_function_versions.
16600 Rename make_name to make_unique_name.
16601 (common_function_versions): Likewise.
16602 (make_unique_name): Likewise.
16603 (make_dispatcher_decl): Likewise.
16604 (is_function_default_version): Likewise.
16605 * attribs.c (attr_strcmp): Likewise.
16606 (sorted_attr_string): Likewise.
16607 (common_function_versions): Likewise.
16608 (make_unique_name): Likewise.
16609 (make_dispatcher_decl): Likewise.
16610 (is_function_default_version): Likewise.
16611 * config/i386/i386.c (attr_strcmp): Likewise.
16612 (sorted_attr_string): Likewise.
16613 (ix86_function_versions): Likewise.
16614 (make_name): Likewise.
16615 (make_dispatcher_decl): Likewise.
16616 (is_function_default_version): Likewise.
16617 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
16619 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16622 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16623 Account for direct move costs for vec_construct of integer
16626 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
16629 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
16630 (UNSPEC_STX_ATOMIC): Ditto.
16631 (loaddi_via_sse): New insn.
16632 (storedi_via_sse): Ditto.
16633 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
16634 Update corresponding peephole2 patterns.
16635 (atomic_storedi_fpu): Ditto.
16637 2017-05-11 Julia Koval <julia.koval@intel.com>
16639 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
16640 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
16642 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
16643 (__builtin_ia32_rsqrt14ss_mask): New builtins.
16644 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
16646 2017-05-11 Nathan Sidwell <nathan@acm.org>
16648 * graphite-poly.c: Include dumpfile.h.
16650 * dumpfle.h (dump_function): Declare here ...
16651 * tree-dump.h (dump_function): ... not here.
16652 * dumpfile.c: #include tree-cfg.h.
16653 (dump_function): Move here from ...
16654 * tree-dump.c (dump_function): ... here.
16655 * gimplify.c: #include splay-tree.h, not tree-dump.h.
16656 * graphite-poly.c: Don't include tree-dump.h.
16657 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
16658 * print-tree.c: Likewise.
16659 * stor-layout.c: Likewise.
16660 * tree-nested.c: Likewise.
16662 * dumpfile.c (dump_start): Use TDF_FLAGS.
16663 (dump_enable_all): Fix TDF_KIND check thinko.
16665 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16667 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16668 array entries to represent two legal parameterizations of the
16669 overloaded __builtin_cmpb function, as represented by the
16670 P6_OV_BUILTIN_CMPB constant.
16671 (altivec_resolve_overloaded_builtin): Add special case handling
16672 for the __builtin_cmpb function, as represented by the
16673 P6_OV_BUILTIN_CMPB constant.
16674 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
16675 (BU_P6_64BIT_2): New macro.
16676 (BU_P6_OVERLOAD_2): New macro
16677 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
16678 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
16679 (CMPB): Add overload support to represent both 32-bit and 64-bit
16680 compare-bytes function.
16681 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
16682 support for TARGET_CMPB.
16683 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
16684 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16685 documentation of the __builtin_cmpb overloaded built-in function.
16687 2017-05-11 Richard Biener <rguenther@suse.de>
16689 PR tree-optimization/80705
16690 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
16691 bases are not vectorizable.
16693 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16695 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
16696 when counting register pressure.
16698 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16700 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
16701 (iv_ca_more_deps): Renamed to ...
16702 (iv_ca_compare_deps): ... this.
16703 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
16705 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16707 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
16709 (determine_group_iv_costs): ... here.
16710 (find_inv_vars_cb): Record inv var if it's not recorded before.
16712 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16714 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
16715 (get_shiftadd_cost): Ditto.
16717 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16719 * tree-ssa-address.c: Include header file.
16720 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
16722 (add_to_parts): Refactor.
16723 (addr_to_parts): New parameter. Update use of move_hint_to_base.
16724 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
16727 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16729 PR tree-optimization/53090
16730 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
16732 (extract_cond_operands): Detect condition with IV on both sides
16733 and return COMP_IV_EXPR_2.
16734 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
16735 (rewrite_use_compare): Simplify by removing call to function
16736 extract_cond_operands.
16738 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16740 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
16741 (extract_cond_operands): Detect condition comparing against non-
16742 invariant bound and return appropriate enum value.
16743 (find_interesting_uses_cond): Update use of extract_cond_operands.
16744 Handle its return value accordingly.
16745 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
16747 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16749 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
16750 nonlinear iv_use computation in loop invariant sensitive way.
16752 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16754 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
16755 (find_iv_candidates): Call relate_compare_use_with_all_cands.
16757 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16759 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
16760 (dump_cand): Support iv_cand.inv_exprs.
16761 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
16763 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
16766 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16768 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
16770 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
16771 as local function. Include necessary header files.
16772 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
16774 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16776 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
16778 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16780 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
16781 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
16782 RSHIFT_EXPR and BIT_NOT_EXPR.
16784 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16786 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
16787 (adjust_setup_cost): New parameter supporting round up adjustment.
16788 (struct address_cost_data): Delete.
16789 (force_expr_to_var_cost): Don't bound cost with spill_cost.
16790 (split_address_cost, ptr_difference_cost): Delete.
16791 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
16792 (struct ainc_cost_data): New struct.
16793 (get_address_cost_ainc): New function.
16794 (get_address_cost, get_computation_cost): Reimplement.
16795 (determine_group_iv_cost_address): Record inv_expr for all uses of
16797 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
16798 (iv_ca_has_deps): Reimplemented to ...
16799 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
16801 (iv_ca_extend): Call iv_ca_more_deps.
16803 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16805 * tree-ssa-address.c (struct mem_address): Move to header file.
16806 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
16807 * tree-ssa-address.h (struct mem_address): Move from C file.
16808 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
16810 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16812 * tree-affine.h (aff_combination_type): New interface.
16813 (aff_combination_zero_p): Remove static.
16814 (aff_combination_const_p): New interface.
16815 (aff_combination_singleton_var_p): New interfaces.
16817 2017-05-11 Richard Biener <rguenther@suse.de>
16819 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16820 Skip unreachable blocks and destinations.
16821 (eliminate): Move stmt removal and fixup ...
16822 (fini_eliminate): ... here. Skip inserted exprs.
16823 (pass_pre::execute): Move fini_pre after fini_eliminate.
16824 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
16825 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
16826 PRE to get rid of dead code that has invalid SSA form and
16827 split critical edges again.
16829 2017-05-11 Bin Cheng <bin.cheng@arm.com>
16831 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16833 2017-05-11 Richard Biener <rguenther@suse.de>
16835 * passes.c (execute_function_todo): Verify loops if they are
16836 said to be up-to-date.
16837 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
16838 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
16840 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
16843 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
16844 handle calling assemble_external ourself.
16847 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
16848 modes with zero size. Enhance comment.
16850 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16852 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
16853 built-ins for vec_xl and vec_xst with short and char pointer
16856 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16858 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
16859 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
16860 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
16861 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
16862 (_mm_maskz_min_round_ss): New intrinsics.
16863 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
16864 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
16865 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
16866 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
16867 (__builtin_ia32_minss_mask_round): New builtins.
16868 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16869 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16870 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
16872 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
16873 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
16875 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
16878 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
16880 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
16881 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
16882 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
16883 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
16884 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
16885 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
16886 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
16887 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16888 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16889 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
16890 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
16891 (__builtin_ia32_mulss_mask_round): New builtins.
16892 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16893 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16894 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
16896 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
16897 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16899 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16902 2017-05-10 Julia Koval <julia.koval@intel.com>
16904 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
16905 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
16906 (_mm256_setr_m128i): New intrinsics.
16908 2017-05-10 Julia Koval <julia.koval@intel.com>
16910 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
16911 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
16912 (_mm_maskz_rcp14_ss): New intrinsics.
16913 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
16914 (__builtin_ia32_rcp14ss_mask): New builtins.
16915 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
16917 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
16919 PR tree-optimization/51513
16920 * tree-cfg.c (gimple_seq_unreachable_p): New function.
16921 (assert_unreachable_fallthru_edge_p): Use it.
16922 (group_case_labels_stmt): Likewise.
16923 * tree-cfg.h: Prototype it.
16924 * stmt.c: Include cfghooks.h and tree-cfg.h.
16925 (emit_case_dispatch_table) <gap_label>: New local variable.
16926 Use it to fill dispatch table gaps.
16927 Test for default_label before updating probabilities.
16928 (expand_case) <default_label>: Remove unneeded initialization.
16929 Test for unreachable default case statement and remove its edge.
16930 Set default_label accordingly.
16931 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
16933 2017-05-10 Carl Love <cel@us.ibm.com>
16935 * config/rs6000/rs6000-c: Add support for built-in functions
16936 vector signed char vec_neg (vector signed char)
16937 vector signed short int vec_neg (vector short int)
16938 vector signed int vec_neg (vector signed int)
16939 vector signed long long vec_neg (vector signed long long)
16940 vector float vec_neg (vector float)
16941 vector double vec_neg (vector double)
16942 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
16944 * config/rs6000/altivec.h: Add define for vec_neg
16945 * doc/extend.texi: Update the built-in documentation for the
16946 new built-in functions.
16948 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16950 PR tree-optimization/77644
16951 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
16953 2017-05-10 Nathan Sidwell <nathan@acm.org>
16955 * dumpfile.h (TDI_lang_all): New.
16956 (TDF_KIND): New. Renumber others
16957 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
16959 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
16961 (get_dump_file_name): Adjust suffix generation.
16962 (dump_enable_all): Use TDF_KIND.
16963 * doc/invoke.texi (-fdump-lang-all): Document.
16965 * dumpfile.h: Tabify.
16967 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
16970 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
16971 Move member access before delete.
16973 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
16975 * tree-inline.c (expand_call_inline): Split block at stmt
16978 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
16981 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
16982 are now unused after splitting mov{sf,sd}_hardfloat.
16983 (f32_lr2): Likewise.
16984 (f32_lm): Likewise.
16985 (f32_lm2): Likewise.
16986 (f32_li): Likewise.
16987 (f32_li2): Likewise.
16988 (f32_lv): Likewise.
16989 (f32_sr): Likewise.
16990 (f32_sr2): Likewise.
16991 (f32_sm): Likewise.
16992 (f32_sm2): Likewise.
16993 (f32_si): Likewise.
16994 (f32_si2): Likewise.
16995 (f32_sv): Likewise.
16996 (f32_dm): Likewise.
16997 (f32_vsx): Likewise.
16998 (f32_av): Likewise.
16999 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
17000 For movsf, order stores so the VSX stores occur before the GPR
17001 store which encourages the register allocator to use a traditional
17002 FPR instead of a GPR. For movsd, order the stores so that the GPR
17003 store comes before the VSX stores to allow the power6 to work.
17004 This is due to the power6 not having a 32-bit integer store
17005 instruction from a FPR.
17006 (movsf_hardfloat): Likewise.
17007 (movsd_hardfloat): Likewise.
17009 2017-05-09 Martin Sebor <msebor@redhat.com>
17011 PR translation/80280
17012 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
17015 PR translation/80280
17016 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
17017 data member added in r247778.
17018 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
17020 2017-05-09 Nathan Sidwell <nathan@acm.org>
17022 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
17024 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
17027 2017-05-09 Marek Polacek <polacek@redhat.com>
17029 * doc/invoke.texi: Fix typo.
17031 2017-05-09 Richard Biener <rguenther@suse.de>
17033 * tree-vrp.c (vrp_val_is_max): Adjust comment.
17034 (vrp_val_is_min): Likewise.
17035 (set_value_range_to_value): Likewise.
17036 (set_value_range_to_nonnegative): Likewise.
17037 (gimple_assign_nonzero_p): Likewise.
17038 (gimple_stmt_nonzero_p): Likewise.
17039 (vrp_int_const_binop): Likewise. Remove unreachable case.
17040 (adjust_range_with_scev): Adjust comments.
17041 (compare_range_with_value): Likewise.
17042 (extract_range_from_phi_node): Likewise.
17043 (test_for_singularity): Likewise.
17045 2017-05-09 Richard Biener <rguenther@suse.de>
17047 * tree-vrp.c (get_single_symbol): Add assert that we don't
17048 get overflowed constants as invariant part.
17049 (compare_values_warnv): Add comment before the TREE_NO_WARNING
17050 checks. Use wi::cmp instead of recursing for integer constants.
17051 (compare_values): Just ignore whether we assumed undefined
17052 overflow instead of failing the compare.
17053 (extract_range_for_var_from_comparison_expr): Add comment before the
17054 TREE_NO_WARNING sets.
17055 (test_for_singularity): Likewise.
17056 (extract_range_from_comparison): Do not disable optimization
17057 when we assumed undefined overflow.
17058 (extract_range_basic): Remove init of unused var.
17060 2017-05-09 Richard Biener <rguenther@suse.de>
17062 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
17063 (extract_range_from_multiplicative_op_1): Adjust.
17064 (extract_range_from_binary_expr_1): Use int_const_binop.
17066 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17069 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
17070 rs6000_store_data_bypass_p in seven define_bypass directives and
17071 in several comments.
17072 * config/rs6000/rs6000-protos.h: Add prototype for
17073 rs6000_store_data_bypass_p function.
17074 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
17075 function implements slightly different (rs6000-specific) semantics
17076 than store_data_bypass_p, returning false rather than aborting
17077 with assertion error when arguments do not satisfy the
17078 requirements of store data bypass.
17079 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
17080 rs6000_store_data_bypass_p.
17082 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
17084 * config/xtensa/xtensa-protos.h
17085 (xtensa_initial_elimination_offset): New declaration.
17086 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17087 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
17088 macro definition, add case for FRAME_POINTER_REGNUM when
17089 FRAME_GROWS_DOWNWARD.
17090 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
17091 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
17092 xtensa_initial_elimination_offset.
17094 2017-05-08 Nathan Sidwell <nathan@acm.org>
17096 * doc/invoke.texi: Alphabetize -fdump options.
17098 2017-05-08 Martin Sebor <msebor@redhat.com>
17100 PR translation/80280
17101 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
17103 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
17105 * target.def (compute_frame_layout): New optional target hook.
17106 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
17107 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
17108 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
17110 * reload1.c (verify_initial_elim_offsets): Likewise.
17111 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
17112 (use_simple_return_p): Call arm_compute_frame_layout if needed.
17113 (arm_get_frame_offsets): Split up into this ...
17114 (arm_compute_frame_layout): ... and this function.
17116 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
17118 * config/aarch64/constraints.md (Usa): New constraint.
17119 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
17121 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
17123 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
17124 with_multilib_list after it has been checked.
17126 2017-05-08 Richard Biener <rguenther@suse.de>
17128 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
17129 (bitmap_set_subtract_values): Likewise.
17131 2017-05-08 Richard Biener <rguenther@suse.de>
17133 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
17134 (gimple_assign_nonzero): ... this and remove strict_overflow_p
17136 (gimple_stmt_nonzero_warnv_p): Rename to ...
17137 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
17139 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
17140 (extract_range_basic): Adjust, do not disable propagation on
17141 strict overflow sensitive simplification.
17142 (vrp_visit_cond_stmt): Likewise.
17144 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
17146 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
17147 body size unconditionally.
17149 2017-05-07 Jeff Law <law@redhat.com>
17152 2017-05-06 Jeff Law <law@redhat.com>
17153 PR tree-optimization/78496
17154 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17157 PR tree-optimization/78496
17158 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17159 (simplify_stmt_using_ranges): Call it.
17160 (vrp_dom_walker::before_dom_children): Extract equivalences
17161 from an ASSERT_EXPR with an equality comparison against a
17164 2017-05-06 Jeff Law <law@redhat.com>
17166 PR tree-optimization/78496
17167 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
17170 PR tree-optimization/78496
17171 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
17172 (simplify_stmt_using_ranges): Call it.
17173 (vrp_dom_walker::before_dom_children): Extract equivalences
17174 from an ASSERT_EXPR with an equality comparison against a
17177 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17179 * lra-constraints.c (lra_copy_reg_equiv): New function.
17180 (split_reg): Use it to copy equivalence information from the
17181 original register to the spill register.
17183 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
17185 PR rtl-optimization/75964
17186 * simplify-rtx.c (simplify_const_relational_operation): Remove
17187 invalid handling of comparisons of integer ABS.
17189 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
17191 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
17192 initialize to zero.
17193 (init_regs): Remove declaration.
17194 (function_arg_advance_32): Initialize error_p as boolean variable.
17196 2017-05-05 Nathan Sidwell <nathan@acm.org>
17198 * store-motion.c (remove_reachable_equiv_notes): Reformat long
17199 lines. Use for (;;).
17201 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17203 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
17204 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
17205 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
17206 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
17207 VF=2 that require versioning.
17209 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17211 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
17214 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17216 * diagnostic.h (diagnostic_override_option_index): Convert from
17217 macro to inline function.
17219 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17221 * diagnostic.c (last_module_changed_p): New function.
17222 (set_last_module): New function.
17223 (diagnostic_report_current_module): Convert macro usage to
17224 the above functions.
17225 * diagnostic.h (diagnostic_context::last_module): Strengthen
17226 from const line_map * to const line_map_ordinary *.
17227 (diagnostic_last_module_changed): Delete macro.
17228 (diagnostic_set_last_module): Delete macro.
17230 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17232 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
17233 with diagnostic_report_diagnostic.
17234 (diagnostic_n_impl_richloc): Likewise.
17235 * diagnostic.h (report_diagnostic): Delete macro.
17236 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
17237 with diagnostic_report_diagnostic.
17238 * substring-locations.c (format_warning_va): Likewise.
17240 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17242 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
17243 save/restor of format_spec. Move option-printing code to...
17244 (print_option_information): ...this new function, and
17245 reimplement by simply printing to the pretty_printer,
17246 rather than appending to the format string.
17248 2017-05-05 David Malcolm <dmalcolm@redhat.com>
17250 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
17251 handling logic into...
17252 (update_effective_level_from_pragmas): ...this new function.
17254 2017-05-04 Andrew Waterman <andrew@sifive.com>
17256 * config/riscv/riscv.opt (mstrict-align): New option.
17257 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
17258 (SLOW_UNALIGNED_ACCESS): Define.
17259 (riscv_slow_unaligned_access): Declare.
17260 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
17262 (riscv_slow_unaligned_access): New variable.
17263 (rocket_tune_info): Set slow_unaligned_access to true.
17264 (optimize_size_tune_info): Set slow_unaligned_access to false.
17265 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
17266 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
17267 (riscv_option_override): Set riscv_slow_unaligned_access.
17268 * doc/invoke.texi: Add -mstrict-align to RISC-V.
17270 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
17272 * config/riscv/riscv.md: Unify indentation.
17274 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17279 * config/rs6000/rs6000.md (u code attribute): Add FIX and
17281 (extendsi<mode>2): Add support for doing sign extension via
17282 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
17283 don't have ISA 3.0 instructions.
17284 (extendsi<mode>2 splitter): Likewise.
17285 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
17286 generate the normal insns since SImode can now go in vector
17287 registers. Disallow the special UNSPECs needed for previous
17288 machines to hide SImode being used. Add new insns
17289 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
17290 (fix_trunc<mode>si2_stfiwx): Likewise.
17291 (fix_trunc<mode>si2_internal): Likewise.
17292 (fixuns_trunc<mode>si2): Likewise.
17293 (fixuns_trunc<mode>si2_stfiwx): Likewise.
17294 (fctiw<u>z_<mode>_smallint): Likewise.
17295 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
17296 of floating point to 32-bit integer from doing a direct move to
17297 the GPR registers to do a store.
17298 (fctiwz_<mode>): Break long line.
17300 2017-05-05 Bin Cheng <bin.cheng@arm.com>
17302 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
17303 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
17304 (addr_list, addr_offset_valid_p): New.
17305 (split_address_groups): Check offset validity with above function.
17306 (gt-tree-ssa-loop-ivopts.h): Include header file.
17308 2017-05-05 Nathan Sidwell <nathan@acm.org>
17310 * config.gcc (arm*-*-*): Add missing 'fi'.
17312 2017-05-05 Steve Ellcey <sellcey@cavium.com>
17314 * doc/invoke.texi (-fopt-info): Explicitly say order of options
17315 included in -fopt-info does not matter.
17316 * doc/optinfo.texi (-fopt-info): Fix description of default
17317 behavour. Explicitly say order of options included in -fopt-info
17320 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17322 * config.gcc: Allow combinations of aprofile and rmprofile values for
17323 --with-multilib-list.
17324 * config/arm/t-multilib: New file.
17325 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
17326 variables. Remove setting of ISA and floating-point ABI in
17327 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
17328 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
17329 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
17330 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
17332 * config/arm/t-rmprofile: Likewise except for the matches changes.
17333 * doc/install.texi (--with-multilib-list): Document the combination of
17334 aprofile and rmprofile values and warn about pitfalls in doing that.
17336 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17338 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
17339 (movdi_aarch64): Likewise.
17341 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17343 PR tree-optimization/80632
17344 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
17346 (build_arrays): Initialize it for virtual phis.
17347 (fix_phi_nodes): Use it for virtual phis.
17349 PR tree-optimization/80558
17350 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
17351 [x, y] op z into [x op, y op z] for op & or | if conditions
17354 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17355 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17358 * config/arm/arm.md (use_literal_pool): Remove.
17359 (64-bit immediate split): No longer takes cost into consideration
17360 if arm_disable_literal_pool is enabled.
17361 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
17362 used when arm_disable_literal_pool is enabled.
17363 (arm_max_const_double_inline_cost): Remove use of
17364 arm_disable_literal_pool.
17365 (push_minipool_fix): Add assert.
17366 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
17367 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
17368 (no_literal_pool_sf_immediate): New.
17370 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17372 PR tree-optimization/80613
17373 * tree-ssa-dce.c (propagate_necessity): Remove cases for
17374 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
17376 2017-05-05 Richard Biener <rguenther@suse.de>
17378 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
17380 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
17382 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
17383 of this flag from insn conditions due to removal from r247495.
17385 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
17387 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
17389 (arm_early_store_addr_dep_ptr): Likewise.
17390 * config/arm/aarch-common-protos.h
17391 (arm_early_load_addr_dep_ptr): Add prototype.
17392 (arm_early_store_addr_dep_ptr): Likewise.
17393 * config/arm/cortex-a53.md: Add new bypasses.
17395 2017-05-05 Jakub Jelinek <jakub@redhat.com>
17397 * tree.c (next_type_uid): Change type to unsigned.
17398 (type_hash_canon): Decrement back next_type_uid if
17399 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
17400 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
17403 2017-05-04 Martin Sebor <msebor@redhat.com>
17405 * builtins.c: Fix a trivial typo in a comment.
17407 PR middle-end/79234
17408 * builtins.c (check_sizes): Adjust to handle reading past the end.
17409 Avoid printing excessive upper bound of ranges. Use %E to print
17410 tree nodes instead of converting them to %wu.
17411 (expand_builtin_memchr): New function.
17412 (compute_dest_size): Rename...
17413 (compute_objsize): ...to this.
17414 (expand_builtin_memcpy): Adjust.
17415 (expand_builtin_mempcpy): Adjust.
17416 (expand_builtin_strcat): Adjust.
17417 (expand_builtin_strcpy): Adjust.
17418 (check_strncat_sizes): Adjust.
17419 (expand_builtin_strncat): Adjust.
17420 (expand_builtin_strncpy): Adjust and simplify.
17421 (expand_builtin_memset): Adjust.
17422 (expand_builtin_bzero): Adjust.
17423 (expand_builtin_memcmp): Adjust.
17424 (expand_builtin): Handle memcmp.
17425 (maybe_emit_chk_warning): Check strncat just once.
17427 2017-05-04 Martin Sebor <msebor@redhat.com>
17429 PR preprocessor/79214
17430 PR middle-end/79222
17431 PR middle-end/79223
17432 * builtins.c (check_sizes): Add inlining context and issue
17433 warnings even when -Wno-system-headers is set.
17434 (check_strncat_sizes): Same.
17435 (expand_builtin_strncat): Same.
17436 (expand_builtin_memmove): New function.
17437 (expand_builtin_stpncpy): Same.
17438 (expand_builtin): Handle memmove and stpncpy.
17440 2017-05-04 Bin Cheng <bin.cheng@arm.com>
17442 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
17443 which is not used any more.
17445 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17447 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
17449 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17451 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
17452 (cortexa53_tunings): Likewise.
17453 (cortexa57_tunings): Likewise.
17454 (cortexa72_tunings): Likewise.
17455 (cortexa73_tunings): Likewise.
17457 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17459 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
17460 Set loop alignment to 8.
17462 2017-05-04 Martin Sebor <msebor@redhat.com>
17464 PR translation/80280
17465 * builtins.c (expand_builtin_object_size): Add missing quoting to
17466 %D and like directives.
17467 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
17468 (hsa_type_for_tree_type): Same.
17469 (verify_function_arguments): Same.
17470 * symtab.c (symbol_table::change_decl_assembler_name): Same.
17471 * varasm.c (get_section): Same.
17474 2017-05-04 Martin Sebor <msebor@redhat.com>
17476 PR translation/80280
17477 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
17479 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
17481 * config/aarch64/aarch64.c (generic_addrcost_table):
17482 Change HI/TI mode setting.
17484 2017-05-04 Martin Jambor <mjambor@suse.cz>
17486 PR tree-optimization/80622
17487 * tree-sra.c (comes_initialized_p): New function.
17488 (build_accesses_from_assign): Only set write lazily when
17489 comes_initialized_p is false.
17490 (analyze_access_subtree): Use comes_initialized_p.
17491 (propagate_subaccesses_across_link): Assert !comes_initialized_p
17492 instead of testing for PARM_DECL.
17494 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17496 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
17497 constraint on operand 0 to allow more general addressing modes.
17498 Adjust output template.
17499 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
17501 * config/aarch64/aarch64-protos.h
17502 (aarch64_address_valid_for_prefetch_p): Declare prototype.
17503 * config/aarch64/constraints.md (Dp): New address constraint.
17504 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
17507 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
17509 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
17510 update use of estimate_ipcp_clone_size_and_time.
17511 (estimate_local_effects): Update use of
17512 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
17513 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
17514 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
17515 Return nonspecialized time.
17517 2017-05-04 Richard Biener <rguenther@suse.de>
17519 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
17520 for the last VUSE which def dominates the PHI. Directly call
17522 (get_continuation_for_phi_1): Remove.
17524 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
17526 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
17527 to explain the use of truncating division. Cap the number of
17528 iterations to the maximum given by nb_iterations_upper_bound,
17531 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17533 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
17534 * configure: Regenerate.
17535 * config.in: Regenerate.
17536 * config/i386/driver-mingw32.c: new file.
17537 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
17538 * config.host: Link driver-mingw32.o on MinGW host.
17539 * doc/install.texi: Document new --enable-mingw-wildcard configure
17542 2017-05-04 Marek Polacek <polacek@redhat.com>
17544 PR tree-optimization/80612
17545 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
17547 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17548 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
17550 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
17551 (movt splitter): Likewise.
17552 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
17553 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
17554 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
17555 block for Thumb-1 with MOVT.
17556 (thumb2_legitimate_address_p): Move code block ...
17557 (can_avoid_literal_pool_for_label_p): ... into this new function.
17558 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
17560 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
17561 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
17562 "M-profile targets with the MOVT instruction".
17564 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17566 * config/arm/arm-builtins.c (arm_init_builtins): Rename
17567 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
17568 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
17570 2017-05-04 Martin Liska <mliska@suse.cz>
17572 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
17573 variable cond_code.
17575 2017-05-04 Richard Biener <rguenther@suse.de>
17577 * tree.c (array_at_struct_end_p): Handle arrays at struct
17578 end with flexarrays more conservatively. Refactor and treat
17579 arrays of arrays or aggregates more strict. Fix
17580 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
17581 * tree.c (array_at_struct_end_p): Adjust prototype.
17582 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
17583 * gimple-fold.c (get_range_strlen): Likewise.
17584 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
17586 2017-05-04 Richard Biener <rguenther@suse.de>
17588 PR tree-optimization/31130
17589 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
17591 (supports_overflow_infinity): Likewise.
17592 (is_negative_overflow_infinity): Likewise.
17593 (is_positive_overflow_infinity): Likewise.
17594 (is_overflow_infinity): Likewise.
17595 (stmt_overflow_infinity): Likewise.
17596 (overflow_infinity_range_p): Likewise.
17597 (usable_range_p): Remove as always returning true.
17598 (make_overflow_infinity): Remove.
17599 (negative_overflow_infinity): Likewise.
17600 (positive_overflow_infinity): Likewise.
17601 (avoid_overflow_infinity): Likewise.
17602 (set_value_range): Adjust accordingly.
17603 (set_value_range_to_nonnegative): Likewise, remove now unused
17604 overflow_infinity arg.
17605 (vrp_operand_equal_p): Adjust.
17606 (update_value_range): Likewise.
17607 (range_int_cst_singleton_p): Likewise.
17608 (operand_less_p): Likewise.
17609 (compare_values_warnv): Likewise.
17610 (extract_range_for_var_from_comparison_expr): Likewise.
17611 (vrp_int_const_binop): Likewise.
17612 (zero_nonzero_bits_from_vr): Likewise.
17613 (extract_range_from_multiplicative_op_1): Likewise.
17614 (extract_range_from_binary_expr_1): Likewise.
17615 (extract_range_from_unary_expr): Likewise.
17616 (extract_range_from_comparison): Likewise.
17617 (extract_range_basic): Likewise.
17618 (adjust_range_with_scev): Likewise.
17619 (compare_ranges): Likewise.
17620 (compare_range_with_value): Likewise.
17621 (dump_value_range): Likewise.
17622 (test_for_singularity): Likewise, remove strict_overflow_p parameter
17624 (simplify_cond_using_ranges): Adjust.
17626 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17628 * brig-builtins.def: Added a builtin for class_f64.
17629 * builtin-types.def: Added a builtin type needed by class_f64.
17631 2017-05-03 Jason Merrill <jason@redhat.com>
17633 * timevar.def: Add TV_CONSTEXPR.
17635 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17637 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
17639 2017-05-03 Martin Jambor <mjambor@suse.cz>
17641 * ipa-prop.c (ipa_update_after_lto_read): Removed.
17642 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
17643 * ipa-cp.c (ipcp_propagate_stage): Do not call
17644 ipa_update_after_lto_read.
17645 * ipa-inline.c (ipa_inline): Likewise.
17647 2017-05-03 Martin Jambor <mjambor@suse.cz>
17649 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
17650 tag. Added a default constructor and a destructor.
17651 (ipa_edge_args_sum_t): New class;
17652 (ipa_edge_args_sum): Declare.
17653 (ipa_edge_args_vector): Remove declaration.
17654 (IPA_EDGE_REF): Use ipa_edge_args_sum.
17655 (ipa_free_edge_args_substructures): Remove declaration.
17656 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
17657 (ipa_edge_args_info_available_for_edge_p): Likewise.
17658 * ipa-prop.c (ipa_edge_args_vector): Removed.
17659 (edge_removal_hook_holder): Likewise.
17660 (edge_duplication_hook_holder): Likewise.
17661 (ipa_edge_args_sum): New variable.
17662 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
17663 ipa_edge_args_vector.
17664 (ipa_free_edge_args_substructures): Likewise.
17665 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
17666 ipa_edge_args_vector.
17667 (ipa_edge_removal_hook): Turned into method
17668 ipa_edge_args_sum_t::remove.
17669 (ipa_edge_duplication_hook): Turned into method
17670 ipa_edge_args_sum_t::duplicate.
17671 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
17672 registering edge hooks.
17673 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
17674 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
17675 ipa_edge_args_sum instead of ipa_edge_args_vector.
17676 * ipa-profile.c (ipa_profile): Likewise.
17678 2017-05-03 Martin Jambor <mjambor@suse.cz>
17680 * symbol-summary.h (function_summary): New method exists.
17681 (function_summary::symtab_removal): Deallocate through release.
17682 (call_summary): New class.
17683 (gt_ggc_mx): New overload.
17684 (gt_pch_nx): Likewise.
17685 (gt_pch_nx): Likewise.
17687 2017-05-03 Jeff Law <law@redhat.com>
17689 PR tree-optimization/78496
17690 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
17691 from simplify_cond_using_ranges. Split off code to walk
17692 backwards through casts into ...
17693 (simplify_cond_using_ranges_2): New function.
17694 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
17695 (execute_vrp): After identifying jump threads, call
17696 simplify_cond_using_ranges_2.
17698 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
17701 * ipa-inline.h (inline_summary): Add ctor.
17702 (create_ggc): Do not use ggc_cleared_alloc.
17704 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
17705 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17707 * gcc.c (handle_braces): Support escaping in switch matching
17709 * doc/invoke.texi (Spec Files): Document it.
17710 Remove superfluous @code markup in items.
17712 2017-05-03 David Malcolm <dmalcolm@redhat.com>
17714 * diagnostic-show-locus.c (struct column_range): New struct.
17715 (get_affected_columns): New function.
17716 (get_printed_columns): New function.
17717 (struct correction): New struct.
17718 (correction::ensure_capacity): New function.
17719 (correction::ensure_terminated): New function.
17720 (struct line_corrections): New struct.
17721 (line_corrections::~line_corrections): New dtor.
17722 (line_corrections::add_hint): New function.
17723 (layout::print_trailing_fixits): Reimplement in terms of the new
17725 (selftest::test_overlapped_fixit_printing): New function.
17726 (selftest::diagnostic_show_locus_c_tests): Call it.
17728 2017-05-03 Nathan Sidwell <nathan@acm.org>
17730 Canonicalize canonical type hashing
17731 * tree.h (type_hash_canon_hash): Declare.
17732 * tree.c (type_hash_list, attribute_hash_list): Move into
17733 type_hash_canon_hash.
17734 (build_type_attribute_qual_variant): Break out hash code calc into
17735 type_hash_canon_hash.
17736 (type_hash_canon_hash): New. Generic type hash computation.
17737 (build_range_type_1, build_array_type_1, build_function_type,
17738 build_method_type_directly, build_offset_type, build_complex_type,
17739 make_vector_type): Call it.
17741 2017-05-03 Richard Biener <rguenther@suse.de>
17743 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
17744 When all DRs have unknown misaligned do not always peel
17745 when there is a store but apply the same costing model as if
17746 there were only loads.
17748 2017-05-03 Richard Biener <rguenther@suse.de>
17751 PR tree-optimization/80492
17752 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17753 compare_base_decls returning dont-know properly.
17755 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
17757 * config/arm/iterators.md (CCSI): New mode iterator.
17758 (arch): New mode attribute.
17759 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
17760 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
17761 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
17762 code iterator for success result mode.
17763 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
17764 the corresponding new insn generators.
17766 2017-05-03 Bin Cheng <bin.cheng@arm.com>
17769 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17770 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17772 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
17774 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
17775 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
17776 (DDR_A): Wrap DDR argument in brackets.
17777 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
17778 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
17779 (DDR_REVERSED_P): Likewise.
17781 2017-05-03 Jakub Jelinek <jakub@redhat.com>
17783 PR tree-optimization/79472
17784 * tree-switch-conversion.c (struct switch_conv_info): Add
17785 contiguous_range and default_case_nonstandard fields.
17786 (collect_switch_conv_info): Compute contiguous_range and
17787 default_case_nonstandard fields, don't clear final_bb if
17788 contiguous_range and only the default case doesn't have the required
17790 (check_all_empty_except_final): Set default_case_nonstandard instead
17791 of failing if contiguous_range and the default case doesn't have empty
17793 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
17794 and only the default case doesn't have the required constants. Skip
17796 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
17797 if default_case_nonstandard.
17798 (build_constructors): Build constant 1 just once. Assert that default
17799 values aren't inserted in between cases if contiguous_range. Skip
17801 (build_arrays): Skip virtual phis.
17802 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
17803 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
17804 Handle virtual phis.
17805 (gen_inbound_check): Handle default_case_nonstandard case.
17806 (process_switch): Adjust check_final_bb caller. Call
17807 gather_default_values with the first non-default case instead of
17808 default case if default_case_nonstandard.
17810 2017-05-02 Nathan Sidwell <nathan@acm.org>
17812 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
17813 check. Fix formatting.
17815 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
17817 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
17818 errors when comparing specialized and unspecialized times.
17820 2017-05-02 David Malcolm <dmalcolm@redhat.com>
17822 * diagnostic-show-locus.c
17823 (layout::should_print_annotation_line_p): Make private.
17824 (layout::print_annotation_line): Make private.
17825 (layout::annotation_line_showed_range_p): Make private.
17826 (layout::show_ruler): Make private.
17827 (layout::print_source_line): Make private. Pass in line and
17828 line_width, rather than calling location_get_source_line. Drop
17830 (layout::print_leading_fixits): New method.
17831 (layout::print_any_fixits): Rename to...
17832 (layout::print_trailing_fixits): ...this, and make private.
17833 Don't print newline fixits.
17834 (diagnostic_show_locus): Move logic for printing one row into...
17835 (layout::print_line): ...this new function. Move the
17836 location_get_source_line call and error-handling from
17837 print_source_line to here. Call print_leading_fixits, and rename
17838 print_any_fixits to print_trailing_fixits.
17839 (selftest::test_fixit_insert_containing_newline): Update now that
17840 newlines are partially supported.
17841 (selftest::test_fixit_insert_containing_newline_2): New test.
17842 (selftest::test_fixit_replace_containing_newline): Update comments.
17843 (selftest::diagnostic_show_locus_c_tests): Call the new test.
17844 * edit-context.c (class added_line): New class.
17845 (class edited_line): Describe newline handling in comment.
17846 (edited_line::actually_edited_p): New method.
17847 (edited_line::print_content): Delete redundant decl.
17848 (edited_line::m_predecessors): New field.
17849 (edited_file::print_content): Call edited_line::print_content.
17850 (edited_file::print_diff): Update to support newlines.
17851 (edited_file::print_diff_hunk): Likewise.
17852 (edited_file::print_run_of_changed_lines): New function.
17853 (edited_file::print_diff_line): Convert to...
17854 (print_diff_line): ...this.
17855 (edited_file::get_effective_line_count): New function.
17856 (edited_line::edited_line): Initialize new field m_predecessors.
17857 (edited_line::~edited_line): Clean up m_predecessors.
17858 (edited_line::apply_fixit): Handle newlines.
17859 (edited_line::get_effective_line_count): New function.
17860 (edited_line::print_content): New function.
17861 (edited_line::print_diff_lines): New function.
17862 (selftest::test_applying_fixits_insert_containing_newline): New
17864 (selftest::test_applying_fixits_replace_containing_newline): New
17866 (selftest::insert_line): New function.
17867 (selftest::test_applying_fixits_multiple_lines): Add example of
17869 (selftest::edit_context_c_tests): Call the new tests.
17871 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17873 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
17874 parameter cand. Update dump information.
17875 (get_computation_cost): Update uses.
17877 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17879 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
17880 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
17881 (get_computation_at, rewrite_use_address): Update use of
17882 get_computation_aff.
17884 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17886 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
17887 (get_computation): Delete.
17888 (get_computation_cost): Implement like get_computation_cost_at.
17889 Use get_computation_at.
17890 (get_computation_cost_at): Delete.
17891 (rewrite_use_nonlinear_expr): Use get_computation_at.
17892 (rewrite_use_compare, remove_unused_ivs): Ditto.
17894 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17896 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
17898 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17900 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
17901 (ivopts_global_cost_for_size): Rename parameter and update uses.
17902 (iv_ca_recount_cost): Update uses.
17903 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
17904 candidates seperately in n_invs and n_cands.
17905 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
17907 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17909 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
17910 (find_inv_vars_cb): New.
17911 (find_depends): Renamed to ...
17912 (find_inv_vars): ... this.
17913 (add_candidate_1, force_var_cost): Call find_inv_vars.
17914 (split_address_cost, determine_group_iv_cost_cond): Ditto.
17916 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17918 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
17919 inv_vars. Add inv_exprs.
17920 (struct iv_cand): Rename depends_on to inv_vars.
17921 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
17922 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
17923 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
17924 (dump_cand): Dump inv_vars.
17925 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
17926 (record_invariant, find_depends, add_candidate_1): Ditto.
17927 (set_group_iv_cost, force_var_cost): Ditto.
17928 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
17929 (get_computation_cost_at, get_computation_cost): Ditto.
17930 (determine_group_iv_cost_generic): Ditto.
17931 (determine_group_iv_cost_address): Ditto.
17932 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
17933 (determine_group_iv_costs): Ditto.
17934 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
17935 (iv_ca_set_remove_invariants): Renamed to ...
17936 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
17937 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
17938 (iv_ca_set_add_invariants): Renamed to ...
17939 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
17940 (iv_ca_set_cp): Use iv_ca_set_add_invs.
17941 (iv_ca_has_deps): Support inv_vars and inv_exprs.
17942 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
17943 (create_new_ivs): Remove useless dump.
17945 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17947 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
17949 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
17950 (iv_ca_set_no_cp, create_new_iv): Ditto.
17952 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17954 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
17956 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17958 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
17959 function tree_check2.
17961 2017-05-02 Martin Liska <mliska@suse.cz>
17963 * doc/gcov.texi: Add missing preposition.
17964 * gcov.c (function_info::function_info): Properly fill up
17965 all member variables.
17967 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17969 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
17971 2017-05-02 Tamar Christina <tamar.christina@arm.com>
17973 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
17975 2017-05-02 Martin Liska <mliska@suse.cz>
17978 * lto-streamer-in.c (lto_read_tree_1): Remove
17979 LTO_STREAMER_DEBUG.
17980 * lto-streamer.c (struct tree_hash_entry): Likewise.
17981 (struct tree_entry_hasher): Likewise.
17982 (tree_entry_hasher::hash): Likewise.
17983 (tree_entry_hasher::equal): Likewise.
17984 (lto_streamer_init): Likewise.
17985 (lto_orig_address_map): Likewise.
17986 (lto_orig_address_get): Likewise.
17987 (lto_orig_address_remove): Likewise.
17988 * lto-streamer.h: Likewise.
17989 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
17990 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
17992 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
17994 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
17995 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
17996 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
17997 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
17998 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
17999 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
18000 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
18001 (mm_maskz_sub_ss): New intrinsics.
18002 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18003 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18004 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
18005 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
18006 (__builtin_ia32_subss_mask_round): New builtins.
18007 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18008 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18009 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
18011 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
18012 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18014 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18017 2017-05-02 Martin Jambor <mjambor@suse.cz>
18019 PR tree-optimization/78687
18020 * tree-sra.c (access): New field parent.
18021 (process_subtree_disqualification): New function.
18022 (disqualify_candidate): Call it.
18023 (build_accesses_from_assign): Reset write flag if creating an
18025 (build_access_subtree): Fill in parent field and also prpagate
18026 down grp_write flag.
18027 (create_artificial_child_access): New parameter set_grp_write, set
18028 grp_write to its value.
18029 (propagate_subaccesses_across_link): Also propagate grp_write flag
18031 (propagate_all_subaccesses): Push the closest parent back to work
18032 queue if add_access_to_work_queue returned true.
18034 2017-05-02 Richard Biener <rguenther@suse.de>
18036 * common.opt (fstrict-overflow): Alias negative to fwrapv.
18037 * doc/invoke.texi (fstrict-overflow): Remove all traces of
18038 -fstrict-overflow documentation.
18039 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
18040 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
18041 flag_strict_overflow.
18042 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
18043 * lto-opts.c (lto_write_options): Do not stream it.
18044 * lto-wrapper.c (merge_and_complain): Do not handle it.
18045 * opts.c (default_options_table): Do not set -fstrict-overflow.
18046 (finish_options): Likewise do not clear it when sanitizing.
18047 * simplify-rtx.c (simplify_const_relational_operation): Do not
18048 test flag_strict_overflow.
18050 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18052 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
18053 using enabled attribute.
18054 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
18055 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
18056 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
18057 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
18058 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
18059 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
18060 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
18061 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
18062 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
18063 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
18065 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
18067 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
18069 2017-05-02 Richard Biener <rguenther@suse.de>
18071 PR tree-optimization/80591
18073 2017-04-10 Richard Biener <rguenther@suse.de>
18075 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18078 2017-05-02 Richard Biener <rguenther@suse.de>
18080 PR tree-optimization/80549
18081 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
18082 (cleanup_tree_cfg_noloop): Create forwarders to known loop
18083 headers if they do not have a preheader.
18085 2017-05-02 Martin Liska <mliska@suse.cz>
18088 * common.opt: Fix typo.
18089 * doc/invoke.texi: Likewise.
18091 2017-05-01 Jan Beulich <jbeulich@suse.com>
18093 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
18094 swapping, add (x,x,m,x,n) alternative.
18096 2017-05-01 Nathan Sidwell <nathan@acm.org>
18098 * calls.c (combine_pending_stack_adjustment_and_call): Remove
18099 unnecessary unadjusted_alignment check.
18101 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
18104 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
18106 * gimplify.c (gimplify_cilk_detach): New function.
18107 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
18108 * tree-core.h: Document EXPR_CILK_SPAWN.
18109 * tree.h (EXPR_CILK_SPAWN): Define.
18111 2017-05-01 David Malcolm <dmalcolm@redhat.com>
18113 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
18114 to use new fixit_hint representation, using the "replace" logic.
18115 (get_line_span_for_fixit_hint): Likewise.
18116 (layout::print_any_fixits): Likewise.
18117 (selftest::test_one_liner_many_fixits): Rename to...
18118 (selftest::test_one_liner_many_fixits_1): ...this, and update
18119 comment and expected output to reflect that the multiple fix-it
18120 hints are now consolidated into one insertion.
18121 (selftest::test_one_liner_many_fixits_2): New test.
18122 (selftest::test_diagnostic_show_locus_one_liner): Update for
18124 (selftest::test_fixit_consolidation): Update for fix-it API
18126 * diagnostic.c (print_parseable_fixits): Likewise.
18127 * edit-context.c (edited_line::m_line_events): Convert from
18128 auto_vec <line_event *> to auto_vec <line_event>.
18129 (class line_event): Convert from abstract base class to a concrete
18130 class, taking over the role of replace_event.
18131 (class insert_event): Delete.
18132 (class replace_event): Rename to class line_event. Convert to
18134 (edit_context::add_fixits): Reimplement.
18135 (edit_context::apply_insert): Delete.
18136 (edit_context::apply_replace): Rename to...
18137 (edit_context::apply_fixit): ...this. Convert to half-open range.
18138 (edited_file::apply_insert): Delete.
18139 (edited_file::apply_replace): Rename to...
18140 (edited_file::apply_fixit): ...this.
18141 (edited_line::~edited_line): Drop deletion of events.
18142 (edited_line::apply_insert): Delete.
18143 (edited_line::apply_replace): Rename to...
18144 (edited_line::apply_fixit): ...this. Convert to half-open range.
18145 Update for change to type of m_line_events.
18146 * edit-context.h (edit_context::apply_insert): Delete.
18147 (edit_context::apply_replace): Rename to...
18148 (edit_context::apply_fixit): ...this.
18150 2017-05-01 Martin Sebor <msebor@redhat.com>
18152 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
18155 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
18158 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
18159 __get_cpuid_max returns 0.
18160 (__get_cpuid_count): Ditto.
18162 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
18164 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
18165 replacement expression is another instance of one of its arguments.
18167 2017-05-01 Jakub Jelinek <jakub@redhat.com>
18170 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
18171 check for stack push/pop autoinc.
18172 * config/i386/i386.c (ix86_agi_dependent): Return false
18173 if the only reason why modified_in_p returned true is that
18174 addr is SP based and set_insn is a push or pop.
18176 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18178 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
18181 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
18184 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
18185 (account_size_time): Use two predicates - exec_pred and
18187 (evaluate_conditions_for_known_args): Compute both clause and
18189 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
18190 (inline_summary_t::duplicate): Update.
18191 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
18193 (compute_inline_parameters): Likewise.
18194 (estimate_edge_size_and_time): Update caluclation of time.
18195 (estimate_node_size_and_time): Compute both time and nonspecialized
18197 (estimate_ipcp_clone_size_and_time): Update.
18198 (inline_merge_summary): Update.
18199 (do_estimate_edge_time): Update.
18200 (do_estimate_edge_size): Update.
18201 (do_estimate_edge_hints): Update.
18202 (inline_read_section, inline_write_summary): Stream both new predicates.
18203 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
18205 (compute_inlined_call_time): Cleanup.
18206 (big_speedup_p): Update.
18207 (edge_badness): Update.
18208 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
18209 (size_time_entry): Replace predicate by exec_predicate and
18210 nonconst_predicate.
18211 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
18212 (estimate_edge_time): Return also nonspec_time.
18213 (reset_edge_growth_cache): Update.
18215 2017-04-29 Jakub Jelinek <jakub@redhat.com>
18217 PR rtl-optimization/80491
18218 * ifcvt.c (noce_process_if_block): When looking for x setter
18219 with missing else_bb, don't check only the insn right before
18220 cond_earliest, but look for the last insn that x is modified in
18221 within the same bb.
18223 PR rtl-optimization/80491
18224 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
18226 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
18228 PR tree-optimization/80487
18229 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
18231 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18233 PR tree-optimization/79697
18234 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
18235 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
18236 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18238 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
18239 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
18241 2017-04-28 Martin Sebor <msebor@redhat.com>
18243 PR tree-optimization/80523
18244 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
18245 (init_target_to_host_charmap, target_to_host, target_strtol10): New
18247 (maybe_warn, format_directive, parse_directive): Use new functions.
18248 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
18250 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
18252 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
18254 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18256 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
18257 target_header_dir): Set correctly.
18258 * configure: Regenerated.
18259 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18260 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18261 instead of SYSTEM_HEADER_DIR.
18263 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
18265 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
18266 (estimate_local_effects): Likewise.
18267 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
18268 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
18269 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
18270 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
18271 do_estimate_edge_time, estimate_edge_time): Likewise.
18272 * ipa-inline-analysis.c (estimate_node_size_and_time,
18273 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
18274 (estimate_time_after_inlining): Remove.
18276 2017-04-28 Martin Liska <mliska@suse.cz>
18278 * doc/gcov.texi: Enhance documentation of gcov.
18280 2017-04-28 Martin Liska <mliska@suse.cz>
18282 * doc/gcov.texi: Sort options in alphabetic order.
18283 * doc/gcov-dump.texi: Likewise.
18284 * doc/gcov-tool.texi: Likewise.
18285 * gcov.c (print_usage): Likewise.
18286 * gcov-dump.c (print_usage): Likewise.
18287 * gcov-tool.c (print_merge_usage_message): Likewise.
18288 (print_rewrite_usage_message): Likewise.
18289 (print_overlap_usage_message): Likewise.
18291 2017-04-28 Martin Liska <mliska@suse.cz>
18293 PR gcov-profile/53915
18294 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
18296 2017-04-28 Martin Liska <mliska@suse.cz>
18298 PR gcov-profile/79891
18299 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
18300 is marked by compiler as living on a line.
18301 (get_cycles_count): Remove usage of the union.
18302 (output_intermediate_file): Likewise.
18303 (find_source): Fix GNU coding style.
18304 (accumulate_line_counts): Remove old non-all block mode.
18305 (output_lines): Remove usage of the union.
18306 * profile.c (output_location): Include all BBs, even if
18307 belonging to a same line (and file) as a previous BB.
18309 2017-04-28 Martin Liska <mliska@suse.cz>
18311 * gcov.c (process_args): Handle new argument 'w'.
18312 (read_graph_file): Assign ID to BBs.
18313 (output_branch_count): Display BB # if verbose flag is set.
18314 (output_lines): Likewise for arcs.
18315 (print_usage): Add '--verbose' option help.
18316 * doc/gcov.texi: Document --verbose (-w) option.
18318 2017-04-28 Martin Liska <mliska@suse.cz>
18320 * gcov.c (struct block_location_info): New struct.
18321 (process_file): Fill up the new structure.
18322 (read_graph_file): Replace usage of encoding by the newly added
18324 (add_line_counts): Likewise.
18325 (accumulate_line_counts): Remove usage of the union.
18326 (function_info::function_info): New function.
18327 (function_info::~function_info): Likewise.
18328 (process_file): Call delete instead of release_function.
18329 (release_function): Release the function.
18330 (release_structures): Call delete instead of release_function.
18331 (solve_flow_graph): Replace usage of num_blocks.
18332 (find_exception_blocks): Likewise.
18333 (output_lines): Fix GNU coding style.
18335 2017-04-28 Martin Liska <mliska@suse.cz>
18338 * coverage.c (coverage_remove_note_file): New function.
18339 * coverage.h: Declare the function.
18340 * toplev.c (finalize): Clean if an error has been seen.
18342 2017-04-28 Martin Liska <mliska@suse.cz>
18344 PR gcov-profile/80031
18345 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
18346 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
18347 * gcov.c (read_graph_file): Read just number of blocks.
18348 * profile.c (branch_prob): Do not stream 0 flags per a basic
18351 2017-04-28 Martin Liska <mliska@suse.cz>
18353 * gcov-dump.c (tag_*): Add new argument to declarations.
18354 (dump_gcov_file): Likewise.
18355 (tag_blocks): Add and use new argument depth.
18356 (tag_arcs): Likewise.
18357 (tag_lines): Likewise.
18358 (tag_counters): Likewise.
18359 (tag_summary): Likewise.
18360 (dump_working_sets): Use depth to do a proper indentation.
18362 2017-04-28 Jakub Jelinek <jakub@redhat.com>
18365 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18366 * symtab.c (symtab_node::debug_symtab): Move definition here.
18368 2017-04-28 Richard Biener <rguenther@suse.de>
18370 * lto-streamer.h (LTO_major_version): Bump to 7.
18372 2017-04-28 Richard Biener <rguenther@suse.de>
18374 * tree-vrp.c (assert_info): New struct.
18375 (add_assert_info): New helper.
18376 (register_edge_assert_for_2): Refactor to add asserts to a vector
18378 (register_edge_assert_for_1): Likewise.
18379 (register_edge_assert_for): Likewise.
18380 (finish_register_edge_assert_for): New helper actually registering
18381 asserts where live on edge.
18382 (find_conditional_asserts): Adjust.
18383 (find_switch_asserts): Likewise.
18384 (evrp_dom_walker::try_find_new_range): Generalize.
18385 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
18387 2017-04-27 Marek Polacek <polacek@redhat.com>
18390 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
18391 arg10 and arg11 to itype.
18393 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
18395 * doc/extend.texi (Object Size Checking): Improve grammar.
18397 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
18400 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
18401 that the logic for permitting reciprocal estimates matches that
18404 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18407 * tree.c (type_cache_hasher::equal): Only compare
18408 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
18409 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
18410 non-aggregate element types.
18411 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
18412 about the flag on ARRAY_TYPEs in the comment, formatting fix.
18414 2017-04-27 Richard Biener <rguenther@suse.de>
18416 PR middle-end/80533
18417 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
18418 stripping ARRAY_REFs from MEM_EXPR make sure we're not
18419 keeping a reference to a trailing array.
18421 2017-04-27 Richard Biener <rguenther@suse.de>
18423 PR middle-end/80539
18424 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
18425 being in loop-closed SSA form conservatively.
18426 (chrec_fold_multiply_poly_poly): Likewise.
18428 2017-04-27 Tamar Christina <tamar.christina@arm.com>
18430 PR middle-end/79665
18431 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
18432 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
18434 2017-04-27 Jakub Jelinek <jakub@redhat.com>
18437 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
18438 (aarch64_function_arg_alignment): Return unsigned int again, but still
18439 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
18440 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18441 Don't emit -Wpsabi note.
18442 (aarch64_function_arg_boundary): Likewise.
18443 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18446 2017-04-26 Nathan Sidwell <nathan@acm.org>
18448 * tree.h (crc32_unsigned_n): Declare.
18449 (crc32_unsigned, crc32_unsigned): Make inline.
18450 * tree.c (crc32_unsigned_bits): Replace with ...
18451 (crc32_unsigned_n): ... this.
18452 (crc32_unsigned, crc32_byte): Remove.
18453 (crc32_string): Remove unnecessary braces.
18455 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18457 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
18458 * ipa-inline-analysis.c (MAX_TIME): Remove.
18459 (account_size_time): Use sreal for time.
18460 (dump_inline_summary): Update.
18461 (estimate_function_body_sizes): Update.
18462 (estimate_edge_size_and_time): Update.
18463 (estimate_calls_size_and_time): Update.
18464 (estimate_node_size_and_time): Update.
18465 (inline_merge_summary): Update.
18466 (inline_update_overall_summary): Update.
18467 (estimate_time_after_inlining): Update.
18468 (inline_read_section): Update.
18469 (inline_write_summary): Update.
18470 * ipa-inline.c (compute_uninlined_call_time): Update.
18471 (compute_inlined_call_time): Update.
18472 (recursive_inlining): Update.
18473 (inline_small_functions): Update.
18474 (dump_overall_stats): Update.
18475 * ipa-inline.h: Include sreal.h.
18476 (size_time_entry): Turn time to sreal.
18477 (inline_summary): Turn self_time nad time to sreal.
18479 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
18481 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
18483 (sreal::stream_out, sreal::stream_in): New.
18484 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
18486 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18488 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
18491 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
18494 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
18496 (dimode_scalar_chain::compute_convert_gain): Ditto.
18497 (dimode_scalar_chain::make_vector_copies): Ditto.
18498 (dimode_scalar_chain::convert_reg): Ditto.
18499 (dimode_scalar_chain::convert_insn): Ditto.
18500 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
18501 (VI248_AVX512BW_1): New mode iterator.
18502 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
18503 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
18506 2017-04-25 Martin Sebor <msebor@redhat.com>
18508 PR tree-optimization/80497
18509 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
18510 constants are representable in HOST_WIDE_INT.
18511 (parse_directive): Ditto.
18513 2017-04-25 Martin Sebor <msebor@redhat.com>
18516 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
18517 (new_zero_array): Adjust signature.
18518 (dom_info::dom_init): Used unsigned rather that size_t.
18519 (dom_info::dom_info): Same.
18521 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18522 Jakub Jelinek <jakub@redhat.com>
18525 * config/arm/arm.c: Include gimple.h.
18526 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18527 returns negative, increment ncrn only if it returned positive.
18528 (arm_needs_doubleword_align): Return int instead of bool,
18529 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
18530 members, but if there is any such non-FIELD_DECL
18531 > PARM_BOUNDARY aligned decl, return -1 instead of false.
18532 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
18533 returns negative, increment nregs only if it returned positive.
18534 (arm_setup_incoming_varargs): Likewise.
18535 (arm_function_arg_boundary): Emit -Wpsabi note if
18536 arm_needs_doubleword_align returns negative, return
18537 DOUBLEWORD_ALIGNMENT only if it returned positive.
18539 2017-04-25 Marek Polacek <polacek@redhat.com>
18542 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18543 first argument to type.
18545 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
18548 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
18549 type checks to test for compatibility instead of equality.
18551 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18552 Jakub Jelinek <jakub@redhat.com>
18555 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
18557 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
18558 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
18559 the alignment computation, but return their maximum in warn_alignment.
18560 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
18561 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
18563 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
18564 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
18567 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18569 * config/arc/simdext.md (dmpyh): Fix typo.
18571 2017-04-25 Richard Biener <rguenther@suse.de>
18573 PR tree-optimization/80492
18574 * alias.c (compare_base_decls): Handle registers with asm
18575 specification conservatively.
18576 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
18577 compare_base_decls returning dont-know properly.
18579 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18581 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
18582 (legitimate_offset_address_p): New function.
18583 (arc_legitimate_address_p): Use above function.
18585 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18587 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
18589 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18591 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
18592 ACCH registers whenever they are available.
18594 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18596 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
18597 double regs fix when not used.
18599 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18601 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
18603 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
18604 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
18606 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18608 * config/arc/arc.c (arc_output_addsi): Check for h-register class
18609 when emitting short ADD instructions.
18611 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
18613 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
18615 (cmpsi_cc_c_insn): Likewise.
18616 (cbranchsi4_scratch): Compute proper instruction length using
18617 compact_hreg_operand.
18618 * config/arc/predicates.md (compact_hreg_operand): New predicate.
18620 2017-04-25 Richard Biener <rguenther@suse.de>
18622 PR middle-end/80509
18623 * passes.c (pass_manager::pass_manager): Initialize
18624 m_name_to_pass_map.
18626 2017-04-25 Richard Biener <rguenther@suse.de>
18628 PR tree-optimization/79201
18629 * tree-ssa-sink.c (statement_sink_location): Handle calls.
18631 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18634 * config/s390/vector.md: Split MEM->GPR vector moves for
18635 non-s_operand addresses.
18637 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18640 * config/s390/predicates.md (reload_const_wide_int_operand): New
18642 * config/s390/s390.md ("movti"): Remove d/P alternative.
18643 ("movti_bigconst"): New pattern definition.
18645 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18648 * s390-protos.h (s390_expand_cs_hqi): Removed.
18649 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
18650 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
18651 modes as well as CCZ1mode and CCZmode.
18652 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
18653 signature of s390_emit_compare_and_swap.
18654 (s390_expand_cs_hqi): Likewise, make static.
18655 (s390_expand_cs_tdsi): Generate an explicit compare before trying
18656 compare-and-swap, in some cases.
18657 (s390_expand_cs): Wrapper function.
18658 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
18660 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
18661 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
18662 patterns for small and large integers. Forbid symref memory operands.
18663 Move expander to s390.c. Require cc register.
18664 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
18665 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
18666 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
18667 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
18668 symref memory operands. Remove CC mode and call s390_match_ccmode
18670 ("atomic_exchange<mode>"): Allow and implement all integer modes.
18672 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18674 * config/s390/s390.md (define_peephole2): New peephole to help
18675 combining the load-and-test pattern with volatile memory.
18677 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
18679 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
18680 with CCZmode for TARGET_Z196.
18682 2017-04-25 Jakub Jelinek <jakub@redhat.com>
18684 PR rtl-optimization/80501
18685 * combine.c (make_compound_operation_int): Set subreg_code to SET
18686 even for AND with mask of the sign bit of mode.
18688 PR rtl-optimization/80500
18689 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
18690 sum's initial value.
18692 2017-04-25 Julian Brown <julian@codesourcery.com>
18693 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18695 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
18697 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
18699 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
18701 2017-04-25 Julian Brown <julian@codesourcery.com>
18702 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18704 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
18705 (thunderx2t99_sha): New Reservation.
18707 2017-04-25 Julian Brown <julian@codesourcery.com>
18708 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18710 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
18711 type for 1-element load.
18713 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
18715 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
18717 2017-04-24 Martin Jambor <mjambor@suse.cz>
18719 PR tree-optimization/80293
18720 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
18721 char arrays not totally scalarizable if it is false.
18722 (analyze_all_variable_accesses): Pass correct value in the new
18723 parameter. Add a statistics counter.
18725 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
18727 PR middle-end/79931
18728 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
18730 2017-04-24 Richard Biener <rguenther@suse.de>
18732 PR tree-optimization/80494
18733 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
18734 out for complex types.
18736 2017-04-24 Richard Biener <rguenther@suse.de>
18738 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
18739 * tree-ssa-sccvn.c (print_scc): Print SCC size.
18740 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
18741 (DFS): Adjust and never fail.
18742 (sccvn_dom_walker::fail): Remove.
18743 (sccvn_dom_walker::before_dom_children): Adjust.
18744 (run_scc_vn): Likewise and never fail.
18745 * tree-ssa-pre.c (pass_pre::execute): Adjust.
18746 (pass_fre::execute): Likewise.
18748 2017-04-24 Richard Biener <rguenther@suse.de>
18750 PR tree-optimization/79725
18751 * tree-ssa-sink.c (statement_sink_location): Return whether
18752 failure reason was zero uses. Move that check later.
18753 (sink_code_in_bb): Deal with zero uses by removing the stmt
18756 2017-04-24 Richard Biener <rguenther@suse.de>
18759 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
18760 pointer-based references.
18762 2017-04-24 Richard Biener <rguenther@suse.de>
18765 * pass_manager.h (pass_manager::operator new): Remove.
18766 (pass_manager::operator delete): Likewise.
18767 * passes.c (pass_manager::operator new): Remove.
18768 (pass_manager::operator delete): Likewise.
18769 (pass_manager::pass_manager): Zero individual pass members.
18771 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
18774 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
18775 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
18776 Check "XEXP (src, 1)" operand here.
18777 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
18778 Check "XEXP (src, 1)" operand here.
18779 (dimode_scalar_chain::make_vector_copies): Detect count register
18780 of a shift instruction. Zero extend count register from QImode
18781 to DImode to satisfy vector shift pattern count operand predicate.
18782 Substitute vector shift count operand with a DImode copy.
18783 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
18786 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
18788 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
18789 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18790 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
18791 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
18792 (UNSPEC_NOREX_MEM): Remove definition.
18794 2017-04-21 Richard Biener <rguenther@suse.de>
18796 PR tree-optimization/79547
18797 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18798 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
18799 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
18800 without any constraints.
18802 2017-04-21 Richard Biener <rguenther@suse.de>
18804 PR tree-optimization/78847
18805 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
18807 2017-04-21 Richard Biener <rguenther@suse.de>
18809 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
18810 (build_distinct_type_copy): Likewise.
18811 (build_variant_type_copy): Likewise.
18812 * tree.c (build_qualified_type): Pass down mem-stat info.
18813 (build_distinct_type_copy): Likewise.
18814 (build_variant_type_copy): Likewise.
18816 2017-04-21 Richard Biener <rguenther@suse.de>
18818 PR tree-optimization/80237
18819 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
18821 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
18822 for a simplified result.
18824 2016-04-21 Richard Biener <rguenther@suse.de>
18826 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
18827 sth as strict as a simple_iv but a chrec without symbols and an
18828 operand defined in the loop we are peeling (and not some subloop).
18829 (propagate_constants_for_unrolling): Propagate all constants.
18831 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18834 * config/i386/i386.c (print_reg): Remove assert for disalowed
18835 regno values, call output_operand_lossage instead.
18837 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
18840 * config/i386/constraints.md (Yc): New register constraint.
18841 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
18842 Use Yc constraint for alternative 2 of operand 0. Remove
18843 preferred_for_speed attribute.
18845 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18847 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
18848 lastprivate clauses in SIMT case.
18850 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
18852 * doc/invoke.texi (-Wextra-semi): Document new warning option.
18854 2017-04-20 Richard Biener <rguenther@suse.de>
18856 PR tree-optimization/57796
18857 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
18858 as N scalar stores.
18859 (vect_model_load_cost): Cost gathers as N scalar loads.
18861 2017-04-20 Richard Biener <rguenther@suse.de>
18863 * ggc-page.c (ggc_allocated_p): Rename to ...
18864 (safe_lookup_page_table_entry): ... this and return the lookup
18866 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
18868 2017-04-20 Richard Biener <rguenther@suse.de>
18870 PR tree-optimization/80453
18871 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
18872 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
18873 from the conditions.
18874 (vn_phi_eq): Pass them down.
18875 (vn_phi_lookup): Record them.
18876 (vn_phi_insert): Likewise.
18878 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
18880 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
18881 uninitialized variable warning to avoid buffer overrun.
18883 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
18886 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
18887 is suppressed for '{ 0 }' in C.
18889 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18891 * BASE-VER: Set to 8.0.0.
18893 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
18895 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
18896 priority .init_array and .fini_array section with SECTION_NOTYPE
18899 2017-04-20 Jakub Jelinek <jakub@redhat.com>
18901 PR middle-end/80423
18902 * tree.h (build_array_type): Add typeless_storage default argument.
18903 * tree.c (type_cache_hasher::equal): Also compare
18904 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
18905 (build_array_type): Add typeless_storage argument, set
18906 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
18908 (build_nonshared_array_type): Adjust build_array_type_1 caller.
18909 (build_array_type): Likewise. Add typeless_storage argument.
18911 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18912 Jakub Jelinek <jakub@redhat.com>
18914 PR tree-optimization/80426
18915 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
18916 operation on symbolic operands, also compute the overflow for the
18917 invariant part when the operation degenerates into a negation.
18919 2017-04-19 Jakub Jelinek <jakub@redhat.com>
18922 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
18923 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
18926 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
18928 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
18931 * config/avr/avr.c (tree.h): Include it.
18932 (cgraph.h): Include it.
18933 (avr_encode_section_info): Don't warn for uninitialized progmem
18934 variable if it's just an alias.
18936 2017-04-19 Richard Biener <rguenther@suse.de>
18939 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
18940 when needed by AutoPGO.
18942 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
18945 * doc/lto.texi: Remove an extra 'that'.
18947 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
18949 PR rtl-optimization/80429
18950 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
18951 are only used in debug insns.
18953 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
18954 Vladimir Makarov <vmakarov@redhat.com>
18956 * config/sparc/predicates.md (input_operand): Add comment. Return
18957 true for any memory operand when LRA is in progress.
18958 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
18960 2017-04-18 Jeff Law <law@redhat.com>
18963 * mips.md ({return,simple_return}_internal): Do not overwrite
18966 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18968 PR tree-optimization/80443
18969 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
18970 instead of adding 1, subtract -1 and similarly instead of subtracting
18973 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
18975 PR rtl-optimization/80357
18976 * haifa-sched.c (tmp_bitmap): New variable.
18977 (model_recompute): Handle duplicate use records.
18978 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
18979 (free_global_sched_pressure_data): Free it.
18981 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
18984 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
18985 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
18986 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
18987 instead of SYSTEM_HEADER_DIR.
18989 2017-04-18 Jeff Law <law@redhat.com>
18991 PR middle-end/80422
18992 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
18993 predecessors after walking up the insn chain.
18995 2017-04-18 Jakub Jelinek <jakub@redhat.com>
18998 * dwarf2out.c (modified_type_die): Try harder not to emit internal
18999 sizetype type into debug info.
19001 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19004 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
19005 unneeded test for TARGET_UPPER_REGS_SF.
19006 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
19008 2017-04-18 Jakub Jelinek <jakub@redhat.com>
19011 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
19012 instead of gsi_after_labels.
19014 2017-04-18 Jeff Law <law@redhat.com>
19016 * regcprop.c (maybe_mode_change): Avoid creating copies of the
19020 2017-04-13 Jeff Law <law@redhat.com>
19021 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19022 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19024 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
19027 * config/avr/avr.c (intl.h): Include it.
19028 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
19030 2017-04-18 Martin Liska <mliska@suse.cz>
19032 PR gcov-profile/78783
19033 * gcov-tool.c (gcov_output_files): Validate that destination
19034 file is either removed by the tool or by a user.
19036 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19037 Guy Benyei <guybe@mellanox.com>
19039 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
19040 block, and do not negate it, the stored id is already negative.
19042 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
19044 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
19046 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19049 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
19050 masks of options that should be turned off if the VSX vector
19051 options are turned off.
19052 (OTHER_P8_VECTOR_MASKS): Likewise.
19053 (OTHER_VSX_VECTOR_MASKS): Likewise.
19054 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
19055 rs6000_disable_incompatible_switches to validate no type switches
19057 (rs6000_incompatible_switch): New function to disallow turning on
19058 other vector options if -mno-vsx, -mno-power8-vector, or
19059 -mno-power9-vector are specified.
19061 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19063 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
19065 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19067 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
19068 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
19069 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
19070 (ARG_POINTER_CFA_OFFSET): Likewise.
19072 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
19074 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
19075 conditions to take advantage of various optimizations.
19077 2017-04-13 Jeff Law <law@redhat.com>
19079 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
19080 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
19081 (zero_extendsidi2_dext): Likewise.
19083 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19086 * fold-const.c (fold_ternary_loc): Revert
19087 use op0 instead of fold_convert_loc (loc, type, arg0) part of
19090 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
19092 PR rtl-optimization/80343
19093 * lra-remat.c (update_scratch_ops): Assign original hard reg to
19094 new scratch pseudo.
19096 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
19099 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
19100 to ubsan_encode_value.
19102 2017-04-13 Jeff Law <law@redhat.com>
19104 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
19105 appearing in DEBUG_INSNs.
19107 2017-04-13 Martin Liska <mliska@suse.cz>
19109 PR gcov-profile/80413
19110 * gcov-io.c (gcov_write_string): Copy to buffer just when
19111 allocated size is greater than zero.
19113 2017-04-13 Jakub Jelinek <jakub@redhat.com>
19116 * dwarf2out.c (decls_for_scope): Ignore declarations of
19117 current_function_decl in BLOCK_NONLOCALIZED_VARS.
19119 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
19122 * ipa-visibility.c (non_local_p): Fix typos.
19123 (localize_node): When localizing symbol in same comdat group,
19124 dissolve the group only when we know external symbols are going
19126 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
19128 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19130 PR tree-optimization/79390
19131 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
19132 order does not result in usable sequence, retry with reversed operand
19138 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
19139 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
19140 op0 instead of fold_convert_loc (loc, type, arg0).
19142 2017-04-12 Jeff Law <law@redhat.com>
19144 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
19145 has a delay slot in the generated code.
19147 * config/cris/cris.md (cris_preferred_reload_class): Return
19148 GENNONACR_REGS rather than GENERAL_REGS.
19150 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19153 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
19154 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
19155 signedness of the result type.
19157 2017-04-12 Richard Biener <rguenther@suse.de>
19158 Jeff Law <law@redhat.com>
19160 PR tree-optimization/80359
19161 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
19162 trim stores to TARGET_MEM_REFs.
19164 2017-04-12 Richard Biener <rguenther@suse.de>
19166 PR tree-optimization/79390
19167 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
19168 threading case even more.
19170 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19173 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
19174 for quad_address_p for TImode, instead of just not indexed_address.
19176 2017-04-12 Richard Biener <rguenther@suse.de>
19177 Bernd Edlinger <bernd.edlinger@hotmail.de>
19179 PR middle-end/79671
19180 * alias.c (component_uses_parent_alias_set_from): Handle
19181 TYPE_TYPELESS_STORAGE.
19182 (get_alias_set): Likewise.
19183 * tree-core.h (tree_type_common): Add typeless_storage flag.
19184 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
19185 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
19186 for types containing members with TYPE_TYPELESS_STORAGE.
19187 (place_field): Likewise.
19188 (layout_type): Likewise for ARRAY_TYPE.
19189 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
19190 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
19191 TYPE_TYPELESS_STORAGE.
19192 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
19194 2017-04-12 Jakub Jelinek <jakub@redhat.com>
19197 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
19198 first argument to type.
19200 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19204 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
19205 CONST0_RTX (mode) rather than const0_rtx where appropriate.
19206 (rs6000_expand_binop_builtin): Likewise.
19207 (rs6000_expand_ternop_builtin): Likewise; also add missing
19208 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
19209 vshasigma built-ins.
19210 * doc/extend.texi: Document that vec_xxpermdi's third argument
19211 must be a constant.
19213 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
19215 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
19216 Use shift_const cost parameter when calculating gain of STV shifts.
19218 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
19220 PR rtl-optimization/70478
19221 * lra-constraints.c (process_alt_operands): Check memory for
19222 disfavoring memory insn operand.
19224 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19226 PR middle-end/80100
19227 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
19228 left shift in unsigned HOST_WIDE_INT type.
19230 PR rtl-optimization/80385
19231 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
19232 (not (neg X)) into (plus X -1) for complex or non-integral modes.
19235 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
19236 if they have any depend clauses.
19238 2017-04-11 Martin Liska <mliska@suse.cz>
19241 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
19242 * ipa-split.c (split_function): Create a local comdat symbol
19243 if caller is in a comdat group.
19245 2017-04-11 Martin Liska <mliska@suse.cz>
19248 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
19251 2017-04-11 Martin Sebor <msebor@redhat.com>
19253 PR middle-end/80364
19254 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
19255 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
19257 (directive::set_width, directive::set_precision, format_character):
19259 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
19262 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
19265 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
19266 conflict, set target->arch_name instead of target->cpu_name.
19268 2017-04-11 Richard Biener <rguenther@suse.de>
19270 PR tree-optimization/80374
19271 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
19272 build_zero_cst, remove fold_convertible_p check again.
19274 2017-04-11 Martin Liska <mliska@suse.cz>
19277 * ubsan.c (instrument_object_size): Do not instrument register
19280 2017-04-11 Jakub Jelinek <jakub@redhat.com>
19283 * config/i386/i386-builtin-types.def
19284 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
19285 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
19286 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
19287 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
19288 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
19289 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
19290 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
19291 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
19292 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
19293 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
19294 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
19295 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
19296 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
19297 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
19298 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
19299 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
19300 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
19301 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
19302 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
19303 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
19304 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
19305 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
19306 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
19307 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
19308 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
19309 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
19310 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
19311 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
19312 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
19313 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
19314 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
19315 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
19316 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
19317 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
19318 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
19319 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
19320 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
19321 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
19322 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
19323 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
19324 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
19325 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
19326 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
19327 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
19328 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
19329 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
19331 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
19332 flag to second_arg_count, handle 4 argument function type _COUNT
19333 aliases, handle second_arg_count on second argument rather than last.
19335 2017-04-10 Jeff Law <law@redhat.com>
19337 PR tree-optimization/80374
19338 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
19339 record anything if we can not convert integer_zero_node to the
19342 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19345 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19346 Enhance special handling given to the TARGET_P9_MINMAX option in
19347 relation to certain other options.
19349 2017-04-10 Bin Cheng <bin.cheng@arm.com>
19351 PR tree-optimization/80153
19352 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
19353 remove POINTER_PLUS_EXPR's base part directly, rather than through
19356 2017-04-10 Richard Biener <rguenther@suse.de>
19357 Bin Cheng <bin.cheng@arm.com>
19359 PR tree-optimization/80153
19360 * tree-affine.c (aff_combination_to_tree): Get base pointer from
19361 the first element of pointer type aff_tree. Build result expr in
19363 (add_elt_to_tree): Convert to type unconditionally. Remove other
19364 fold_convert calls.
19365 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
19366 (rewrite_use_nonlinear_expr): Check invariant using iv information.
19368 2017-04-10 Richard Biener <rguenther@suse.de>
19370 * tree-ssa-structalias.c (find_func_aliases): Properly handle
19373 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
19375 PR rtl-optimization/70478
19376 * lra-constraints.c (curr_small_class_check): New.
19377 (update_and_check_small_class_inputs): New.
19378 (process_alt_operands): Update curr_small_class_check. Disfavor
19379 alternative insn memory operands. Check available regs for small
19382 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19385 * config/mips/mips.opt (-mvirt): Update description.
19386 * doc/invoke.texi (-mvirt): Likewise.
19388 2017-04-10 Richard Biener <rguenther@suse.de>
19390 PR middle-end/80362
19391 * fold-const.c (fold_binary_loc): Look at unstripped ops when
19392 looking for NEGATE_EXPR in -A / -B to A / B folding.
19394 2017-04-10 Martin Liska <mliska@suse.cz>
19396 PR gcov-profile/80224
19397 * gcov.c (print_usage): Fix usage string.
19398 (get_gcov_intermediate_filename): Remove.
19399 (output_gcov_file): Use both for normal and intermediate format.
19400 (generate_results): Do not initialize special file for
19401 intermediate format.
19403 2017-04-10 Richard Biener <rguenther@suse.de>
19405 PR tree-optimization/80304
19406 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
19409 2017-04-10 Nathan Sidwell <nathan@acm.org>
19412 * config/rs6000/rs6000.c (rs6000_vector_type): New.
19413 (rs6000_init_builtins): Use it.
19415 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19417 * config/arm/arm.md (<mrc>): Add mode to SET source.
19418 (<mrrc>): Likewise.
19420 2017-04-10 Richard Biener <rguenther@suse.de>
19422 PR middle-end/80344
19423 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
19425 2017-04-10 Jakub Jelinek <jakub@redhat.com>
19428 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
19429 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
19430 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
19431 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
19432 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
19433 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
19434 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
19435 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
19436 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
19437 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
19438 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
19439 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
19440 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
19441 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
19442 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
19443 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
19444 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
19445 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
19446 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
19447 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
19448 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
19449 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
19450 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
19452 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
19454 PR rtl-optimization/70478
19455 * lra-constraints.c: Reverse the last patch.
19457 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
19459 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
19460 Add comment for WCHAR_T.
19462 2017-04-08 Martin Liska <mliska@suse.cz>
19465 2017-04-07 Martin Liska <mliska@suse.cz>
19468 * ipa-split.c (split_function): Add function part to a same comdat
19471 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19474 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
19476 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
19478 * rs6000/rs6000.c (vec_load_pendulum): Rename...
19479 (vec_pairing): ...to this.
19480 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
19481 (rs6000_sched_init): Adjust for name change.
19482 (struct rs6000_sched_context): Likewise.
19483 (rs6000_init_sched_context): Likewise.
19484 (rs6000_set_sched_context): Likewise.
19486 2017-04-07 Jakub Jelinek <jakub@redhat.com>
19492 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
19494 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
19495 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
19496 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
19498 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
19500 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
19502 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19504 PR rtl-optimization/70703
19505 * ira-color.c (update_conflict_hard_regno_costs): Use
19506 int64_t instead of HOST_WIDE_INT.
19508 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
19510 PR rtl-optimization/70478
19511 * lra-constraints.c (process_alt_operands): Disfavor alternative
19512 insn memory operands.
19514 2017-04-07 Jeff Law <law@redhat.com>
19516 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
19517 CALL and NOTE_INSN_CALL_ARG_LOCATION.
19519 2017-04-07 Martin Liska <mliska@suse.cz>
19522 * config/aarch64/aarch64.c (aarch64_process_target_attr):
19523 Show error message instead of an ICE.
19525 2017-04-07 Martin Liska <mliska@suse.cz>
19528 * ipa-split.c (split_function): Add function part to a same comdat
19531 2017-04-07 Richard Biener <rguenther@suse.de>
19533 PR middle-end/80341
19534 * tree.c (get_unwidened): Also handle ! for_type case for
19536 * convert.c (do_narrow): Split out from ...
19537 (convert_to_integer_1): ... here. Do not pass final truncation
19538 type to get_unwidened for TRUNC_DIV_EXPR.
19540 2017-04-07 Richard Biener <rguenther@suse.de>
19542 * tree-affine.c (wide_int_ext_for_comb): Take type rather
19544 (aff_combination_const): Adjust.
19545 (aff_combination_scale): Likewise.
19546 (aff_combination_add_elt): Likewise.
19547 (aff_combination_add_cst): Likewise.
19548 (aff_combination_convert): Likewise.
19549 (add_elt_to_tree): Likewise. Remove unused argument.
19550 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
19552 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
19554 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
19556 * config/arm/arm.c (arm_default_short_enums): Use
19557 ARM_DEFAULT_SHORT_ENUMS.
19558 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
19560 2017-04-06 Jakub Jelinek <jakub@redhat.com>
19563 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
19564 members with redundant out-of-class redeclaration.
19566 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19569 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
19570 * config/i386/i386.md (*zero_extendsidi2):
19571 Add (?*x,*x) and (?*v,*v) alternatives.
19573 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
19576 * config/i386/i386.c (ix86_expand_builtin)
19577 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
19578 mode from insn data. Convert operands to insn operand mode.
19579 Copy operands that don't satisfy insn predicate to a register.
19581 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
19583 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
19586 2017-04-06 Richard Biener <rguenther@suse.de>
19588 PR tree-optimization/80334
19589 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
19590 preserve alignment of accesses.
19592 2017-04-06 Richard Biener <rguenther@suse.de>
19594 PR tree-optimization/80262
19595 * tree-sra.c (build_ref_for_offset): Preserve address-space
19597 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
19598 Drop useless address-space information on MEM_REF offsets.
19600 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
19602 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
19604 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19606 PR rtl-optimization/70703
19607 * ira-color.c (update_conflict_hard_regno_costs): Use
19608 HOST_WIDE_INT instead of long.
19610 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
19613 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
19614 not defined for x86_64 target. Add -mmmx target option when __SSE2__
19616 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
19617 for x86_64 target. Handle -m3dnowa option.
19619 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
19621 PR rtl-optimization/70703
19622 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
19623 (update_conflict_hard_regno_costs): Use long instead of unsigned
19624 arithmetic for cost calculation.
19626 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19627 Bernd Edlinger <bernd.edlinger@hotmail.de>
19630 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
19633 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
19636 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
19637 ptr_mode with Pmode throughout.
19638 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
19639 into probe_stack_range and use DImode.
19641 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
19644 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
19645 call_eh_return is true.
19647 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19649 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
19650 Initialize last_match_fntype_index.
19652 2017-04-05 Jakub Jelinek <jakub@redhat.com>
19655 * tree-nvr.c: Include internal-fn.h.
19656 (pass_return_slot::execute): Ignore internal calls without
19659 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19660 Richard Biener <rguenther@suse.de>
19663 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
19664 captures used multiple times, except for the last use.
19665 * generic-match-head.c: Include gimplify.h.
19667 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19669 PR tree-optimization/79390
19670 * target.h (struct noce_if_info): Declare.
19671 * targhooks.h (default_noce_conversion_profitable_p): Declare.
19672 * target.def (noce_conversion_profitable_p): New target hook.
19673 * ifcvt.h (struct noce_if_info): New type, moved from ...
19674 * ifcvt.c (struct noce_if_info): ... here.
19675 (noce_conversion_profitable_p): Renamed to ...
19676 (default_noce_conversion_profitable_p): ... this. No longer
19678 (noce_try_store_flag_constants, noce_try_addcc,
19679 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
19680 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
19681 instead of noce_conversion_profitable_p.
19682 * config/i386/i386.c: Include ifcvt.h.
19683 (ix86_option_override_internal): Don't override
19684 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
19685 (ix86_noce_conversion_profitable_p): New function.
19686 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
19687 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
19688 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
19689 * doc/tm.texi: Regenerated.
19691 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19693 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
19696 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19699 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
19700 instructions for small multiply cores.
19702 2017-04-04 Jeff Law <law@redhat.com>
19704 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
19706 (mips_expand_vec_perm_const): Initialize elements in orig_perm
19707 that are not set by the loop over the elements.
19709 2017-04-04 Jakub Jelinek <jakub@redhat.com>
19712 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
19713 int mode, convert_modes it to mode as unsigned, otherwise use
19714 lowpart_subreg to mode rather than SImode.
19715 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
19716 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
19717 Use DImode instead of SImode for the shift count operand.
19718 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
19721 2017-04-04 Richard Biener <rguenther@suse.de>
19723 PR middle-end/80281
19724 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
19725 arithmetic done for the negate or the plus. Simplify.
19726 (A - (-B) -> A + B): Likewise.
19727 * fold-const.c (split_tree): Make sure to not negate pointers.
19729 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
19731 PR rtl-optimization/60818
19732 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
19733 a compare of comparisons with the thing compared if this results
19734 in a different machine mode.
19736 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
19738 * alias.c (base_alias_check): Fix typo in comment.
19739 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
19740 * cgraphunit.c (symbol_table::compile): Likewise.
19741 * collect2.c (maybe_run_lto_and_relink): Likewise.
19742 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
19743 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
19744 * config/avr/avr.c (avr_map_op_t): Likewise.
19745 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
19746 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
19747 * config/epiphany/epiphany.md (movcc): Likewise.
19748 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
19749 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
19751 * config/mips/mips.c (mips_save_restore_reg): Likewise.
19752 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
19753 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
19754 * config/sh/sh.c (sh_rtx_costs): Likewise.
19755 * fold-const.c (fold_truth_andor): Likewise.
19756 * genautomata.c (collapse_flag): Likewise.
19757 * gengtype.h (struct type::u::s): Likewise.
19758 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
19759 * input.c (FORMAT_AMOUNT): Likewise.
19760 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
19761 (known_aggs_to_agg_replacement_list): Likewise.
19762 * ipa-inline-analysis.c: Likewise.
19763 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
19764 * ipa-polymorphic-call.c
19765 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
19766 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
19767 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
19769 * modulo-sched.c (apply_reg_moves): Likewise.
19770 * omp-expand.c (build_omp_regions_1): Likewise.
19771 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
19772 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
19773 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
19774 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19775 * value-prof.c: Likewise.
19776 * var-tracking.c (val_reset): Likewise.
19778 2017-04-03 Richard Biener <rguenther@suse.de>
19780 PR tree-optimization/80275
19781 * fold-const.c (split_address_to_core_and_offset): Handle
19784 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
19786 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
19787 descriptors is at least equal to that of functions.
19789 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19791 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
19793 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
19796 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
19797 (mov<IMOD4:mode>): New expander.
19798 (*mov<IMOD4:mode>_internal): New insn and split pattern.
19800 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19802 PR rtl-optimization/79405
19803 * fwprop.c (propagations_left): New variable.
19804 (forward_propagate_into): Decrement it.
19805 (fwprop_init): Initialize it.
19806 (fw_prop): If the variable has reached zero, stop propagating.
19807 (fwprop_addr): Ditto.
19809 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19812 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
19813 a FUNCTION_DECL, pass it as decl instead of origin to
19816 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
19818 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
19821 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
19824 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
19825 TARGET_VSX_SMALL_INTEGER.
19827 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19829 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19830 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
19832 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
19834 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
19835 extraction from odd-numbered MSA register.
19837 2017-03-31 Jakub Jelinek <jakub@redhat.com>
19839 PR middle-end/80173
19840 * expmed.c (store_bit_field_1): Don't attempt to create
19841 a word subreg out of hard registers wider than word if they
19842 have HARD_REGNO_NREGS of 1 for their mode.
19844 PR middle-end/80163
19845 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
19846 conversions to integer types wider than word and pointer.
19849 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
19850 (rtx_equal_for_cselib_p): Pass 0 to it.
19851 * cselib.c (cselib_hasher::equal): Likewise.
19852 (rtx_equal_for_cselib_1): Add depth argument. If depth
19853 is 128, don't look up VALUE locs and punt. Increment
19854 depth in recursive calls when walking VALUE locs.
19856 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
19858 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
19859 (make_gcov_file_name): Use the canonical path name for generating
19861 (read_line): Fix handling of files with ascii null bytes.
19863 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
19865 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
19866 to initialise a vector register instead
19867 of using a const_int.
19869 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19871 PR translation/80189
19872 * gimplify.c (omp_default_clause): Use %qs instead of %s in
19873 diagnostic messages.
19875 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
19878 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
19879 (dfp_diex_<mode>): Update mode of operand 1.
19880 * doc/extend.texi (dxex, dxexq): Document change to return type.
19881 (diex, diexq): Document change to argument type.
19883 2017-03-30 Martin Jambor <mjambor@suse.cz>
19886 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
19887 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
19888 it reflects the signature changes performed at the callee side.
19889 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
19890 to cgraph_build_function_type_skip_args.
19891 (build_function_decl_skip_args): Adjust call to the above function.
19893 2017-03-30 Jakub Jelinek <jakub@redhat.com>
19896 * config/i386/sse.md
19897 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
19898 register as dest whenever it is a MEM not rtx_equal_p to the
19899 corresponding dup operand, and when forcing into reg move the
19900 reg into the memory afterwards.
19901 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
19902 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
19903 for the force_reg mode.
19904 (avx512vl_vextractf128<mode>): Use register as dest either
19905 always when a MEM, or when it is a MEM not rtx_equal_p to the
19906 corresponding dup operand, or even not when it is a CONST_VECTOR
19907 depending on the mode and lo vs. hi.
19908 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
19910 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
19911 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
19912 Likewise. Require that operands[2] is even.
19913 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
19914 Remove extraneous parens. Require that operands[2] is a multiple
19916 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
19917 operands[0] is a MEM if <mask_applied>, the predicates/constraints
19918 disallow memory then.
19920 2017-03-30 Richard Biener <rguenther@suse.de>
19922 PR tree-optimization/77498
19923 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
19924 to non-constants over backedges.
19926 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
19928 PR rtl-optimization/80233
19929 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
19930 as last_combined_insn. Do not test for BARRIER_P separately.
19932 2017-03-29 Andreas Schwab <schwab@suse.de>
19935 * calls.c (prepare_call_address): Convert funexp to Pmode before
19936 copying to temp reg.
19938 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19940 PR tree-optimization/80158
19941 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
19942 Handle possible future case of more than one alternate
19944 (replace_rhs_if_not_dup): Likewise.
19945 (replace_one_candidate): Likewise.
19947 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
19949 PR rtl-optimization/80193
19950 * ira.c (ira): Do not check allocation for LRA.
19952 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
19954 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
19955 (nvptx_output_simt_exit): Declare.
19956 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
19957 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
19958 (init_softstack_frame): Move initialization of crtl->is_leaf to...
19959 (nvptx_declare_function_name): ...here. Emit declaration of local
19960 memory space buffer for omp_simt_enter insn.
19961 (nvptx_output_unisimt_switch): New.
19962 (nvptx_output_softstack_switch): New.
19963 (nvptx_output_simt_enter): New.
19964 (nvptx_output_simt_exit): New.
19965 * config/nvptx/nvptx.h (struct machine_function): New fields
19966 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
19967 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
19968 (UNSPECV_SIMT_EXIT): Ditto.
19969 (omp_simt_enter_insn): New insn.
19970 (omp_simt_enter): New expansion.
19971 (omp_simt_exit): New insn.
19972 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
19974 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
19975 (expand_GOMP_SIMT_ENTER_ALLOC): New.
19976 (expand_GOMP_SIMT_EXIT): New.
19977 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
19978 (GOMP_SIMT_ENTER_ALLOC): Ditto.
19979 (GOMP_SIMT_EXIT): Ditto.
19980 * target-insns.def (omp_simt_enter): New insn.
19981 (omp_simt_exit): Ditto.
19982 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
19984 (lower_rec_simd_input_clauses): Implement SIMT privatization.
19985 (lower_rec_input_clauses): Likewise.
19986 (lower_lastprivate_clauses): Handle SIMT privatization.
19988 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
19989 (ompdevlow_adjust_simt_enter): New.
19990 (find_simtpriv_var_op): New.
19991 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
19992 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
19994 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
19995 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
19996 (copy_decl_for_dup_finish): Ditto.
19998 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
20000 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
20003 * config/i386/i386.c (ix86_option_override_internal): Always
20004 allow -mpreferred-stack-boundary=3 for 64-bit targets.
20006 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20008 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
20010 2017-03-28 Bin Cheng <bin.cheng@arm.com>
20012 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
20013 mark new edge's irreducible flag accordign to it.
20014 (vect_do_peeling): Check loop preheader edge's irreducible flag
20015 and pass it to function slpeel_add_loop_guard.
20017 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
20019 PR tree-optimization/80218
20020 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
20021 Update block frequencies and counts.
20023 2017-03-28 Richard Biener <rguenther@suse.de>
20025 PR tree-optimization/78644
20026 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
20027 of a simplification result we may not use it at all.
20029 2017-03-28 Richard Biener <rguenther@suse.de>
20032 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
20033 without arguments, generate default definition of a SSA name.
20035 2017-03-28 Richard Biener <rguenther@suse.de>
20037 PR middle-end/80222
20038 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
20039 TYPE_REF_CAN_ALIAS_ALL references.
20040 * fold-const.c (fold_indirect_ref_1): Likewise.
20042 2017-03-28 Martin Liska <mliska@suse.cz>
20045 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
20046 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
20048 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20049 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
20051 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
20052 (EXTRA_SPECS): Define.
20053 (SUBTARGET_EXTRA_SPECS): Likewise.
20054 (SUBTARGET_CPP_SPEC): Likewise.
20055 * config/arc/elf.h (EXTRA_SPECS): Renamed to
20056 SUBTARGET_EXTRA_SPECS.
20057 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
20059 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
20061 * config/arc/simdext.md (vst64_insn): Update pattern.
20062 (vld32wh_insn): Likewise.
20063 (vld32wl_insn): Likewise.
20064 (vld64_insn): Likewise.
20065 (vld32_insn): Likewise.
20067 2017-03-28 Marek Polacek <polacek@redhat.com>
20070 * fold-const.c (fold_comparison): Use protected_set_expr_location
20071 instead of SET_EXPR_LOCATION.
20073 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
20075 * tree.c (add_expr): Avoid name lookup warning.
20077 2017-03-27 Jeff Law <law@redhat.com>
20079 PR tree-optimization/80216
20080 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
20081 function name. Limit recursion depth.
20082 (record_temporary_equivalences): Corresponding changes.
20084 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
20086 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
20089 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20092 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
20094 * cfgcleanup.c (reg_note_cfa_p): New array.
20095 (insns_have_identical_cfa_notes): New function.
20096 (old_insns_match_p): Don't cross-jump in between /f
20097 and non-/f instructions. If both i1 and i2 are frame related,
20098 verify all CFA notes, their order and content.
20100 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20103 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
20104 HImode and SImode with zero extend to DImode to one insn.
20105 (bswap<mode>2_extenddi): Likewise.
20106 (bswapsi2_extenddi): Likewise.
20107 (bswaphi2_extendsi): Likewise.
20108 (bswaphi2): Combine bswap HImode and SImode into one insn.
20109 Separate memory insns from swapping register.
20110 (bswapsi2): Likewise.
20111 (bswap<mode>2): Likewise.
20112 (bswaphi2_internal): Delete, no longer used.
20113 (bswapsi2_internal): Likewise.
20114 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
20115 store, and gpr<-gpr swap insns.
20116 (bswap<mode>2_store): Likewise.
20117 (bswaphi2_reg): Register only splitter, combine with the splitter.
20118 (bswaphi2 splitter): Likewise.
20119 (bswapsi2_reg): Likewise.
20120 (bswapsi2 splitter): Likewise.
20121 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
20122 the insns into load, store, and register/register insns.
20123 (bswapdi2_ldbrx): Likewise.
20124 (bswapdi2_load): Likewise.
20125 (bswapdi2_store): Likewise.
20126 (bswapdi2_reg): Likewise.
20128 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
20130 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
20131 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
20133 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
20136 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
20138 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20139 special handling for target option conflicts between dform
20140 options (-mpower9-dform, -mpower9-dform-vector,
20141 -mpower9-dform-scalar) and -mno-direct-move.
20143 2017-03-27 Richard Biener <rguenther@suse.de>
20145 PR tree-optimization/80181
20146 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
20148 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20150 * config/arc/predicates.md (move_double_src_operand): Replace the
20151 call to move_double_src_operand with a call to address_operand.
20153 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20155 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
20156 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
20157 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
20159 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20161 * config/arc/predicates.md (long_immediate_loadstore_operand):
20162 Consider scaled addresses cases.
20164 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
20166 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
20167 restored when in interrupt.
20168 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
20169 doesn't have delay slot.
20171 2017-03-27 Richard Biener <rguenther@suse.de>
20174 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
20175 inlined thunk clones.
20177 2017-03-27 Jakub Jelinek <jakub@redhat.com>
20180 * asan.c (instrument_derefs): Copy over last operand from
20181 original COMPONENT_REF to the new COMPONENT_REF with
20182 DECL_BIT_FIELD_REPRESENTATIVE.
20183 * ubsan.c (instrument_object_size): Likewise.
20185 2017-03-27 Richard Biener <rguenther@suse.de>
20187 PR tree-optimization/80170
20188 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
20189 sure DR/SCEV didnt fold in constants we do not see when looking
20190 at the reference base alignment.
20192 2017-03-27 Richard Biener <rguenther@suse.de>
20194 PR middle-end/80171
20195 * gimple-fold.c (fold_ctor_reference): Properly guard against
20196 NULL return value from canonicalize_constructor_val.
20198 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
20201 * config/i386/i386.c (ix86_expand_builtin)
20202 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
20203 flags reg setting and flags reg using instructions.
20204 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
20205 clobbering instructions to zero extend op2.
20207 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
20209 * doc/install.texi (Configuration) <--with-aix-soname>:
20210 Update link to AIX ld.
20212 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
20214 PR rtl-optimization/80160
20215 PR rtl-optimization/80159
20216 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
20217 reg_alternate_class into account.
20219 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
20222 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
20223 to consider in curr_insn_transform.
20225 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20227 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
20228 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
20229 and emit_mode_inner.
20231 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20233 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
20234 argument to the overloaded builtin variants. Use the new flag to
20235 deprecate certain builtin variants.
20236 * config/s390/s390-builtin-types.def: Add new builtin types.
20237 * config/s390/s390-builtins.h: Support new flags field for
20238 overloaded builtins.
20239 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
20240 (s390_macro_to_expand): Enable vector float data type.
20241 (s390_cpu_cpp_builtins_internal): Indicate support of the new
20242 builtins by incrementing the __VEC__ version number.
20243 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
20245 (s390_resolve_overloaded_builtin): Emit error messages depending
20246 on the builtin flags.
20247 * config/s390/s390.c (s390_expand_builtin): Support additional
20248 flags argument. Change error message to match the messages
20249 emitted in s390-c.c.
20250 * config/s390/s390.md: New UNSPEC_* constants.
20251 (op_type): Add new instruction types.
20252 * config/s390/vecintrin.h: Add new builtins and test data class
20254 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
20255 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
20256 (VEC_INEXACT, VEC_NOINEXACT): New constants.
20257 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
20258 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
20259 ("vec_mergel<mode>"): V_HW -> VEC_HW.
20261 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
20262 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
20263 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
20264 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
20266 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
20267 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
20268 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
20269 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
20271 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
20272 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
20273 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
20274 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
20275 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
20276 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
20277 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
20279 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
20280 ("vec_scatter_element<V_HW_4:mode>_DI")
20281 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
20282 ("vec_fpint<mode>", "vflls")
20283 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
20284 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
20285 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
20286 ("*vec_cmphe<mode>_cc"): ... these.
20288 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
20289 mode constant instead of magic value.
20291 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20293 * config/s390/s390.c (s390_expand_vec_compare): Support other
20294 vector floating point modes than just V2DF.
20295 (s390_expand_vcond): Likewise.
20296 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
20297 (s390_cannot_change_mode_class): Prevent mode changes between TF
20298 and V1TF in vector registers.
20299 * config/s390/s390.md (DF, SF): New mode attributes.
20300 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
20301 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
20302 SFmode support for VRs.
20303 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
20305 (VFT, VF_HW): New mode iterators.
20306 (vw, sdx): New mode attributes.
20307 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
20308 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
20309 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
20310 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
20311 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
20312 also the new vector floating point modes. Renaming to ...
20314 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
20315 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
20316 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
20317 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
20318 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
20319 ("vec_unordered<mode>"): ... these.
20321 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
20322 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
20323 ("*vec_extendv2df"): New insn definitions.
20325 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20327 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
20328 ("mulditi3_2", "*muldi3_sign"): New patterns.
20329 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
20330 rename the pattern definition.
20332 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20334 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
20336 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
20338 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20340 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
20341 instruction if possible.
20342 * config/s390/vector.md (vec_halfnumelts): New mode
20344 ("*vec_vllezlf<mode>"): New pattern.
20346 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20348 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
20349 ("popcountv4si2", "popcountv2di2"): Rename to ...
20350 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
20351 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
20353 ("popcount<mode>2_vxe"): New pattern.
20355 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20357 * common/config/s390/s390-common.c (processor_flags_table): Add
20359 * config.gcc: Add arch12.
20360 * config/s390/driver-native.c (s390_host_detect_local_cpu):
20361 Default to arch12 for unknown CPU model numbers.
20362 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
20363 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
20364 PROCESSOR_max sanity check.
20365 * config/s390/s390-opts.h (enum processor_type): Add
20367 * config/s390/s390.c (processor_table): Add arch12.
20368 (s390_expand_builtin): Add check for B_VXE flag.
20369 (s390_issue_rate): Add PROCESSOR_ARCH12.
20370 (s390_get_sched_attrmask): Likewise.
20371 (s390_get_unit_mask): Likewise.
20372 (s390_sched_score): Enable z13 scheduling for arch12.
20373 (s390_sched_reorder): Likewise.
20374 (s390_sched_variable_issue): Likewise.
20375 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
20377 (s390_tune_attr): Use z13 scheduling also for arch12.
20378 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
20379 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
20380 (TARGET_VXE_P): New macros.
20381 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
20382 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
20383 * config/s390/s390.opt: Add arch12 as processor_type.
20385 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20387 * config/s390/s390.md
20388 ("fixuns_truncdddi2", "fixuns_trunctddi2")
20389 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
20390 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
20392 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
20393 Rename expanders to ...
20395 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
20396 ("fixuns_truncdddi2_emu"): ... these.
20398 ("fixuns_trunc<mode>si2_emu"): New expander.
20400 ("*fixuns_truncdfdi2_z13"): Rename to ...
20401 ("*fixuns_truncdfdi2_vx"): ... this.
20403 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20405 * config/s390/2964.md: Remove the single element vector compare
20406 instructions which are no longer used.
20407 * config/s390/s390.c (s390_select_ccmode): Remove handling of
20409 (s390_canonicalize_comparison): Remove handling of DFmode
20411 (s390_expand_vec_compare_scalar): Remove function.
20412 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
20413 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
20415 ("*cmp<mode>_ccs"): Add wfcdb instruction.
20417 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20419 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
20421 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
20422 will anyway by matched by mov<mode>_64dfp.
20424 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20426 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
20427 vlef/vstef. Add missing operand to vleif.
20429 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20431 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
20432 pair for all vector types with 64 bit elements.
20433 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
20434 * config/s390/vector.md (V_HW_64): ... here.
20435 (V_128_NOSINGLE): New mode iterator.
20436 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
20437 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
20438 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
20439 ("*vec_load_pairv2di"): Change to ...
20440 ("*vec_load_pair<mode>"): ... this one.
20442 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20444 * config/s390/constraints.md: Add comments.
20445 (jKK): Reject element sizes > 8 bytes.
20446 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
20448 * config/s390/s390.md: Add the s_operand checks formerly in
20449 s390_split_ok_p to various splitters where they are still
20451 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
20452 for 128 bit vectors. Plus two splitters.
20454 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20456 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
20459 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20462 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
20463 error if the boundary argument is not constant.
20465 2017-03-24 Jakub Jelinek <jakub@redhat.com>
20467 PR rtl-optimization/80112
20468 * loop-doloop.c (doloop_condition_get): Don't check condition
20469 if cmp isn't SET with IF_THEN_ELSE src.
20471 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20473 PR tree-optimization/80158
20474 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
20475 replacing a candidate statement, also replace it for the
20476 candidate's alternate interpretation.
20477 (replace_rhs_if_not_dup): Likewise.
20478 (replace_one_candidate): Likewise.
20480 2017-03-24 Richard Biener <rguenther@suse.de>
20482 PR tree-optimization/80167
20483 * graphite-isl-ast-to-gimple.c
20484 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
20486 (translate_isl_ast_to_gimple::get_rename): Likewise.
20488 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
20490 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
20491 handling of certain combinations of target options, including the
20492 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
20493 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
20495 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20498 * config/arm/arm.md (*load_multiple): Add reload_completed to
20499 matching condition.
20501 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20502 Richard Biener <rguenth@suse.de>
20504 PR tree-optimization/79908
20505 PR tree-optimization/80136
20506 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20507 been cast away, gimplify_and_add suffices.
20509 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
20511 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
20513 2017-03-23 Richard Biener <rguenther@suse.de>
20515 PR tree-optimization/80032
20516 * gimplify.c (gimple_push_cleanup): Forced unconditional
20517 cleanups still have to go to the conditional_cleanups
20520 2017-03-22 Jakub Jelinek <jakub@redhat.com>
20522 PR tree-optimization/80072
20523 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
20525 (next_operand_entry_id): Change type to unsigned int.
20526 (sort_by_operand_rank): Make sure to return the right return value
20527 even if unsigned fields are bigger than INT_MAX.
20528 (struct oecount): Change cnt and id type to unsigned int.
20529 (oecount_hasher::equal): Formatting fix.
20530 (oecount_cmp): Make sure to return the right return value
20531 even if unsigned fields are bigger than INT_MAX.
20532 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
20535 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
20536 TREE_READONLY on result if writing it more than once.
20539 * doc/invoke.texi (-fsanitize=thread): Document that with
20540 -fnon-call-exceptions atomics are not able to throw
20544 * tsan.c: Include tree-eh.h.
20545 (instrument_builtin_call): Call maybe_clean_eh_stmt or
20546 maybe_clean_or_replace_eh_stmt where needed.
20547 (instrument_memory_accesses): Add cfg_changed argument.
20548 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
20549 if it returned true.
20550 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
20552 PR rtl-optimization/63191
20553 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
20554 wrapper function, moved the whole old content into ...
20555 (ix86_delegitimize_address_1): ... this. New inline function.
20556 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
20557 true as last argument instead of ix86_delegitimize_address.
20559 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20561 * config/aarch64/aarch64.c (generic_branch_cost): Copy
20562 cortexa57_branch_cost.
20564 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
20566 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
20568 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20571 * doc/md.texi (Constraints): Document wA constraint.
20572 * config/rs6000/constraints.md (wA): New.
20573 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
20574 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
20575 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
20576 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
20578 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
20581 * gimplify.c (is_oacc_declared): New function.
20582 (oacc_default_clause): Use it to set default flags for acc declared
20583 variables inside parallel regions.
20584 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
20585 declared variables.
20586 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
20587 declare attribute to any decl as necessary.
20589 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20592 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
20593 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
20594 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
20595 (arm_arch_lpae): This.
20596 * config/arm/arm.c (arm_arch7ve): Rename into ...
20597 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
20598 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
20601 2017-03-22 Martin Liska <mliska@suse.cz>
20604 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
20605 error message instead of an ICE.
20607 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20609 * doc/extend.texi (6.11 Additional Floating Types): Revise.
20611 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
20613 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20615 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
20618 2017-03-21 Martin Sebor <msebor@redhat.com>
20620 * doc/extend.texi: Use "cannot" instead of "can't."
20621 * doc/hostconfig.texi: Same.
20622 * doc/install.texi: Same.
20623 * doc/invoke.texi: Same.
20624 * doc/loop.texi: Same.
20625 * doc/md.texi: Same.
20626 * doc/objc.texi: Same.
20627 * doc/rtl.texi: Same.
20628 * doc/tm.texi: Same.
20629 * doc/tm.texi.in: Same.
20630 * doc/trouble.texi: Same.
20632 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
20635 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
20636 (collect_checksum_attributes): Set it.
20637 (die_checksum_ordered): Use it.
20639 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20641 PR tree-optimization/79908
20642 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
20643 change: For a VA_ARG whose LHS has been cast away, use
20644 force_gimple_operand to construct the side effects.
20646 2017-03-21 David Malcolm <dmalcolm@redhat.com>
20648 PR translation/80001
20649 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
20650 more amenable to translation.
20651 (oacc_loop_auto_partitions): Likewise.
20653 2017-03-21 Marek Polacek <polacek@redhat.com>
20654 Martin Sebor <msebor@redhat.com>
20656 PR tree-optimization/80109
20657 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
20658 on INTEGRAL_TYPE_P.
20660 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20661 Segher Boessenkool <segher@kernel.crashing.org>
20664 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
20665 check reg_used_between_p between insn and one of succ or succ2
20666 depending on if succ is artificial insn not inserted into insn
20669 2017-03-21 Martin Liska <mliska@suse.cz>
20671 PR gcov-profile/80081
20672 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
20673 * doc/gcc.texi: Include gcov-dump stuff.
20674 * doc/gcov-dump.texi: New file.
20676 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
20678 PR rtl-optimization/79150
20679 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
20680 conditional jump, if the jump is the last insn of the loop.
20682 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20683 Richard Biener <rguenth@suse.de>
20685 PR tree-optimization/79908
20686 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
20687 been cast away, use force_gimple_operand to construct the side
20690 2017-03-21 Martin Liska <mliska@suse.cz>
20692 PR libfortran/79956
20693 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
20696 2017-03-21 Brad Spengler <spender@grsecurity.net>
20699 * plugin.c (htab_hash_plugin): New function.
20700 (add_new_plugin): Use it and adjust.
20701 (parse_plugin_arg_opt): Adjust.
20702 (init_one_plugin): Likewise.
20704 2017-03-21 Richard Biener <rguenther@suse.de>
20706 PR tree-optimization/80032
20707 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
20708 if set force the cleanup to happen unconditionally.
20709 (gimplify_target_expr): Push inserted clobbers with force_uncond
20710 to avoid them being removed by control-dependent DCE.
20712 2017-03-21 Richard Biener <rguenther@suse.de>
20714 PR tree-optimization/80122
20715 * tree-inline.c (copy_bb): Do not expans va-arg packs or
20716 va_arg_pack_len when the inlined call stmt requires pack
20718 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
20720 2017-03-21 Jakub Jelinek <jakub@redhat.com>
20723 * tsan.c (instrument_builtin_call): If the memory model argument
20724 is not a constant, assume it is valid.
20727 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
20730 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20732 PR rtl-optimization/79910
20733 * combine.c (can_combine_p): Do not allow combining an I0 or I1
20734 if its dest is used by an insn before I2 (other than the combined
20735 insns themselves, which are properly handled already).
20737 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
20740 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20742 * combine.c (record_used_regs): New static function.
20743 (try_combine): Handle situations where there is an additional
20744 instruction between I2 and I3 which needs to have a LOG_LINK
20748 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20750 * combine.c (try_combine): Delete redundant i1 test. Call
20751 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20753 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20756 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
20757 alternatives 13/14.
20759 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20761 PR tree-optimization/80054
20762 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
20763 the optimization if a PHI or any of its arguments is not dominated
20764 by the candidate's basis. Use gphi* rather than gimple* as
20766 (replace_profitable_candidates): Clean up a gimple* variable that
20767 should be a gphi* variable.
20769 2017-03-20 Martin Sebor <msebor@redhat.com>
20772 * doc/extend.texi (attribute constructor): Document present limitation.
20774 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
20777 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
20778 __POWER9_VECTOR__ #ifdef control, change template definition to
20779 use Power9-specific built-in function.
20780 (vec_any_eq): Likewise.
20781 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
20782 to control outcomes from this test.
20783 (vector_ae_<mode>p): For VEC_F modes, likewise.
20785 2017-03-20 Ian Lance Taylor <iant@google.com>
20787 * config/i386/i386.c (ix86_function_regparm): Save an extra
20788 register for -fsplit-stack with DECL_STATIC_CHAIN.
20790 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20793 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
20794 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
20796 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
20798 * config/riscv/riscv.c (riscv_print_operand): Use "fence
20800 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
20803 2017-03-20 Marek Polacek <polacek@redhat.com>
20806 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
20808 2017-03-20 Richard Biener <rguenther@suse.de>
20810 PR tree-optimization/80113
20811 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
20812 allocate extra SSA name for PHI def.
20813 (add_close_phis_to_outer_loops): Likewise.
20814 (add_close_phis_to_merge_points): Likewise.
20815 (copy_loop_close_phi_args): Likewise.
20816 (copy_cond_phi_nodes): Likewise.
20818 2017-03-20 Martin Liska <mliska@suse.cz>
20820 PR middle-end/79753
20821 * tree-chkp.c (chkp_build_returned_bound): Do not build
20822 returned bounds for a LHS that's not a BOUNDED_P type.
20824 2017-03-20 Martin Liska <mliska@suse.cz>
20828 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
20829 COMPLEX_CST and VECTOR_CST.
20831 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20834 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
20835 target operand. A new splitter adds the clobber statement in case
20836 the target operand is dead anyway.
20838 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
20840 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
20841 to age-old versions of binutils and glibc.
20843 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
20845 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
20847 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20849 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
20851 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
20853 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
20854 requirement for binutils 2.13.
20856 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
20858 * combine.c (try_combine): Delete redundant i1 test. Call
20859 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
20861 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
20863 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
20864 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
20866 <riscv64-*-elf>: Re-arrange section
20867 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
20868 <riscv32-*-linux>: Likewise.
20869 <riscv64-*-elf>: Likewise
20870 <riscv64-*-linux>: Likewise.
20872 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
20875 * aarch64.opt(verbose-cost-dump): Fix typo.
20877 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
20880 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
20881 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
20883 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
20885 * reload.c (find_reloads): When reloading a nonoffsettable address,
20886 use RELOAD_OTHER for it and its address reloads.
20888 PR rtl-optimization/79910
20889 * combine.c (record_used_regs): New static function.
20890 (try_combine): Handle situations where there is an additional
20891 instruction between I2 and I3 which needs to have a LOG_LINK
20894 2017-03-17 Jeff Law <law@redhat.com>
20896 PR tree-optimization/71437
20897 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
20898 conditional in the hash table first.
20899 (vrp_dom_walker::before_dom_children): Extract condition from
20900 ASSERT_EXPR. Record condition, its inverion and any implied
20901 conditions as well.
20903 2017-03-17 Marek Polacek <polacek@redhat.com>
20904 Markus Trippelsdorf <markus@trippelsdorf.de>
20906 PR tree-optimization/80079
20907 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
20910 2017-03-17 Richard Biener <rguenther@suse.de>
20912 PR middle-end/80075
20913 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
20914 Properly verify the LHS before the RHS possibly claims to be
20916 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
20919 2017-03-17 Martin Jambor <mjambor@suse.cz>
20921 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
20922 (List of -O2 options): Likewise.
20923 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
20926 2017-03-17 Tom de Vries <tom@codesourcery.com>
20928 * gcov-dump.c (print_usage): Print bug_report_url.
20930 2017-03-17 Richard Biener <rguenther@suse.de>
20932 PR middle-end/80050
20933 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
20934 (parser::peek): Likewise.
20936 2017-03-17 Richard Biener <rguenther@suse.de>
20938 PR tree-optimization/80048
20939 * sese.c (free_sese_info): Properly release rename_map and
20940 copied_bb_map elements.
20942 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
20944 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
20945 Add linked-list forward and backlinks. Insert on
20946 construction, remove on destruction.
20947 (class pass_store_merging): Add m_stores_head field.
20948 (pass_store_merging::terminate_and_process_all_chains):
20949 Iterate over m_stores_head list.
20950 (pass_store_merging::terminate_all_aliasing_chains):
20952 (pass_store_merging::execute): Check for debug stmts first.
20953 Push new chains onto the m_stores_head stack.
20955 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20958 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
20959 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
20960 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
20962 2017-03-16 Jeff Law <law@redhat.com>
20964 PR tree-optimization/71437
20965 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
20966 member function. Implementation moved into after_dom_children
20967 member function and into the threader's thread_outgoing_edges
20969 (dom_opt_dom_walker::after_dom_children): Simplify by moving
20970 some code into new thread_outgoing_edges.
20971 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
20972 definition. Simplify marker handling (do it here). Assume we always
20973 have the available expression and the const/copies tables.
20974 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
20976 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
20977 * tree-vrp.c (equiv_stack): No longer file scoped.
20978 (vrp_dom_walker): New class.
20979 (vrp_dom_walker::before_dom_children): New member function.
20980 (vrp_dom_walker::after_dom_children): Likewise.
20981 (identify_jump_threads): Setup domwalker. Use it rather than
20982 walking edges in a random order by hand. Simplify setup/finalization.
20983 (finalize_jump_threads): Remove.
20984 (vrp_finalize): Do not call identify_jump_threads here.
20985 (execute_vrp): Do it here instead and call thread_through_all_blocks
20988 PR tree-optimization/71437
20989 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
20991 (simplify_stmt_for_jump_threading): Add basic_block argument. All
20993 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
20994 (dom_opt_dom_walker::thread_across_edge): Remove
20995 handle_dominating_asserts argument. All callers changed.
20996 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
20997 changes. Remove calls to lhs_of_dominating_assert. Other
20998 uses of handle_dominating_asserts turn into unconditional code
20999 (simplify_control_stmt_condition_1): Likewise.
21000 (simplify_control_stmt_condition): Likewise.
21001 (thread_through_normal_block, thread_across_edge): Likewise.
21002 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
21003 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
21004 object if it is not an SSA_NAME.
21005 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
21006 before calling into the VRP specific simplifiers.
21007 (identify_jump_threads): Remove handle_dominating_asserts
21010 2017-03-16 Jakub Jelinek <jakub@redhat.com>
21013 * tree-diagnostic.c (default_tree_printer): No longer static.
21014 * tree-diagnostic.h (default_tree_printer): New prototype.
21016 2017-03-16 Tamar Christina <tamar.christina@arm.com>
21018 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
21019 Change ins into fmov.
21021 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21023 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
21024 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
21025 Use h_con constraint for operand 1.
21026 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
21027 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
21029 2017-03-15 Jeff Law <law@redhat.com>
21031 PR tree-optimization/71437
21032 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
21033 (record_temporary_equivalences): Use it.
21035 PR tree-optimization/71437
21036 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
21037 tree-ssa-scopedtables.
21038 (lookup_avail_expr, build_and_record_new_cond): Likewise.
21039 (record_conditions, record_cond, vuse_eq): Likewise.
21040 (record_edge_info): Adjust to API tweak of record_conditions.
21041 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
21042 (record_temporary_equivalences, optimize_stmt): Likewise.
21043 (eliminate_redundant_computations): Likewise.
21044 (record_equivalences_from_stmt): Likewise.
21045 * tree-ssa-scopedtables.c: Include options.h and params.h.
21046 (vuse_eq): New function, moved from tree-ssa-dom.c
21047 (build_and_record_new_cond): Likewise.
21048 (record_conditions): Likewise. Accept vector of conditions rather
21049 than edge_equivalence structure for first argument.
21050 for the first argument.
21051 (avail_exprs_stack::lookup_avail_expr): New member function, moved
21052 from tree-ssa-dom.c.
21053 (avail_exprs_stack::record_cond): Likewise.
21054 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
21055 from tree-ssa-dom.c.
21056 (avail_exprs_stack): Add new member functions lookup_avail_expr
21058 (record_conditions): Declare.
21060 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
21063 * lra-constraints.c (process_alt_operands): Increase reject for
21064 reloading an input/output operand.
21066 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
21069 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
21070 insns to convert from signed/unsigned char/short to IEEE 128-bit
21072 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
21074 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
21077 * config/i386/i386.c (ix86_vector_duplicate_value): Create
21078 subreg of inner mode for values already in registers.
21080 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
21082 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
21083 iteration reg is used after the loop.
21085 2017-03-14 Martin Sebor <msebor@redhat.com>
21087 PR tree-optimization/79800
21088 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
21089 precision in negative-positive range.
21090 (format_floating): Call non-const overload with adjusted precision.
21092 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21095 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
21098 2017-03-14 Martin Sebor <msebor@redhat.com>
21100 PR middle-end/80020
21101 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
21102 * builtins.def (aligned_alloc): Use it.
21105 * Makefile.in (GTFILES): Add calls.c.
21106 * calls.c: Include "gt-calls.h".
21108 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
21110 PR rtl-optimization/79728
21111 * regs.h (struct target_regs): New field
21112 x_contains_allocatable_regs_of_mode.
21113 (contains_allocatable_regs_of_mode): New macro.
21114 * reginfo.c (init_reg_sets_1): Initialize it, and change
21115 contains_reg_of_mode so it includes global regs as well.
21116 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
21117 rather than contains_regs_of_mode.
21119 2017-03-14 Martin Liska <mliska@suse.cz>
21121 * doc/invoke.texi: Document options that can't be combined with
21122 -fcheck-pointer-bounds.
21124 2017-03-14 Martin Liska <mliska@suse.cz>
21126 PR middle-end/79831
21127 * doc/invoke.texi (-Wchkp): Document the option.
21129 2017-03-14 Martin Liska <mliska@suse.cz>
21131 * Makefile.in: Install gcov-dump.
21133 2017-03-14 Martin Liska <mliska@suse.cz>
21135 * multiple_target.c (expand_target_clones): Bail out for
21136 an invalid attribute.
21138 2017-03-14 Richard Biener <rguenther@suse.de>
21140 * alias.c (struct alias_set_entry): Pack properly.
21141 * cfgloop.h (struct loop): Likewise.
21142 * cse.c (struct set): Likewise.
21143 * ipa-utils.c (struct searchc_env): Likewise.
21144 * loop-invariant.c (struct invariant): Likewise.
21145 * lra-remat.c (struct cand): Likewise.
21146 * recog.c (struct change_t): Likewise.
21147 * rtl.h (struct address_info): Likewise.
21148 * symbol-summary.h (function_summary): Likewise.
21149 * tree-loop-distribution.c (struct partition): Likewise.
21150 * tree-object-size.c (struct object_size_info): Likewise.
21151 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
21152 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
21153 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
21154 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
21155 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
21156 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
21157 (struct _stmt_vec_info): Likewise.
21159 2017-03-14 Martin Liska <mliska@suse.cz>
21162 * multiple_target.c (create_dispatcher_calls): Check that
21163 a target can create a function dispatcher.
21165 2017-03-14 Martin Liska <mliska@suse.cz>
21168 * multiple_target.c (expand_target_clones): Drop local.local
21169 flag for default implementation.
21171 2017-03-14 Richard Biener <rguenther@suse.de>
21173 PR tree-optimization/80030
21174 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
21176 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
21178 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
21179 gcc_fallthrough() instead of __attribute__((fallthrough));
21181 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21183 * doc/gcc.texi: Remove "up" link to (DIR).
21184 * doc/gccint.texi: Ditto.
21186 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
21188 * doc/install.texi (Specific) <avr>: Remove reference to
21191 2017-03-13 Jeff Law <law@redhat.com>
21193 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
21194 attribute rather than comments.
21196 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
21197 match_scratch operand is highest.
21199 2017-03-13 Martin Liska <mliska@suse.cz>
21201 PR middle-end/78339
21202 * ipa-pure-const.c (warn_function_noreturn): If the declarations
21203 is a CHKP clone, use original declaration.
21205 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21207 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
21208 (arc_conditional_register_usage): Use a different allocation order
21209 when optimizing for size.
21210 * common/config/arc/arc-common.c (arc_option_optimization_table):
21211 Section anchors default on when optimizing for size.
21213 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21215 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
21217 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21219 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
21220 * config/arc/arc.md (cpu_facility): Add cd variant.
21221 (*movqi_insn): Add code density variant.
21222 (*movhi_insn): Likewise.
21223 (*movqi_insn): Likewise.
21224 (*addsi3_mixed): Likewise.
21225 (subsi3_insn): Likewise.
21227 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21229 * config/arc/arc.md (movsi_cond_exec): Update constraint.
21231 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
21233 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
21234 expressions with MINUS and UNARY ops.
21236 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21239 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
21241 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
21242 between vec_select and vector argument.
21243 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
21244 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
21245 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
21246 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
21247 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
21248 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
21250 2017-03-13 Richard Biener <rguenther@suse.de>
21253 * params.def (vect-max-peeling-for-alignment): Fix typo.
21255 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21257 * doc/install.texi (Specific) <mips-*-*>: Remove description of
21258 issue that only occurred with binutils below 2.18.
21260 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21262 * doc/install.texi (Specific) <cris-axis-elf>: No longer
21263 refer to binutils 2.11/2.12 minimum.
21265 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
21267 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
21268 ftp.kernel.org and simplify binutils requirement.
21270 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
21272 * doc/invoke.texi (Warning Options): Fix spelling of link-time
21274 (Optimize Options): Ditto. Also remove redundancy.
21276 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21278 PR translation/79848
21279 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
21281 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
21282 to G_ to avoid double translation.
21284 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21286 PR translation/79923
21287 * auto-profile.c (get_combined_location): Convert leading
21288 character of diagnostics to lower case and remove trailing period.
21289 (read_profile): Likewise for various diagnostics.
21290 * config/arm/arm.c (arm_option_override): Remove trailing period
21291 from various diagnostics.
21292 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
21293 (msp430_expand_delay_cycles): Likewise.
21295 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21298 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
21299 full command-line argument, rather than just "str".
21300 (aarch64_validate_march): Likewise.
21301 (aarch64_validate_mtune): Likewise.
21303 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
21305 PR rtl-optimization/78911
21306 * lra-assigns.c (must_not_spill_p): New function.
21307 (spill_for): Use it.
21309 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21311 PR tree-optimization/79981
21312 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
21313 ATOMIC_COMPARE_EXCHANGE ifn result.
21314 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
21315 IFN_ATOMIC_COMPARE_EXCHANGE.
21317 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21320 * opts.c (parse_sanitizer_options): Add missing question mark to
21321 "did you mean" message.
21323 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21325 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
21327 (VMULEUH_UNS): Likewise.
21328 (VMULOUB_UNS): Likewise.
21329 (VMULOUH_UNS): Likewise.
21330 * config/rs6000/rs6000.c (builtin_function_type): Remove
21331 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
21333 2017-03-10 David Malcolm <dmalcolm@redhat.com>
21336 * read-rtl-function.c (function_reader::read_rtx_operand): Update
21337 x with result of extra_parsing_for_operand_code_0.
21338 (function_reader::extra_parsing_for_operand_code_0): Convert
21339 return type from void to rtx, returning x. When reading
21340 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
21341 larger size containing struct block_symbol.
21343 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
21345 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
21346 -mfloat128-hardware without -m64.
21348 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
21351 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
21352 entries to the case statement that marks unsigned arguments to
21353 overloaded functions.
21355 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21357 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
21358 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
21360 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
21363 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
21364 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
21366 2017-03-10 Martin Liska <mliska@suse.cz>
21370 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
21371 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
21374 2017-03-10 Olivier Hainque <hainque@adacore.com>
21376 * tree-switch-conversion (array_value_type): Start by resetting
21377 candidate type to it's main variant.
21379 2017-03-10 Jakub Jelinek <jakub@redhat.com>
21381 PR rtl-optimization/79909
21382 * combine.c (try_combine): Use simplify_replace_rtx on individual
21383 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
21384 of the whole CALL_INSN_FUNCTION_USAGE.
21386 PR tree-optimization/79972
21387 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
21388 get_range_info on SSA_NAMEs. Formatting fixes.
21390 2017-03-10 Richard Biener <rguenther@suse.de>
21391 Jakub Jelinek <jakub@redhat.com>
21393 PR tree-optimization/77975
21394 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
21395 edge to be constant.
21396 (get_val_for): For constant x return it. Formatting fix.
21397 (loop_niter_by_eval): Avoid pointless looping if the next iteration
21398 would use the same bases as the current one.
21400 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21402 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
21403 instead of vec_select for V1TImode.
21404 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
21406 (VSX_LE_128): Add V1TI to this mode iterator.
21407 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
21408 (*vsx_le_perm_store_<mode>): Likewise.
21409 (pre-reload splitter for VSX stores): Likewise.
21410 (post-reload splitter for VSX stores): Likewise.
21411 (*vsx_xxpermdi2_le_<mode>): Likewise.
21412 (*vsx_lxvd2x2_le_<mode>): Likewise.
21413 (*vsx_stxvd2x2_le_<mode>): Likewise.
21415 2017-03-09 Michael Eager <eager@eagercon.com>
21417 Correct failures with --enable-checking=yes,rtl.
21419 * config/microblaze/microblaze.c (microblaze_expand_shift):
21420 Replace GET_CODE test with CONST_INT_P and INTVAL test with
21421 test for const0_rtx.
21422 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
21423 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
21425 2017-03-09 Richard Biener <rguenther@suse.de>
21427 PR tree-optimization/79977
21428 * graphite-scop-detection.c (scop_detection::merge_sese):
21429 Handle the case of extra exits to blocks dominating the entry.
21431 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
21433 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21436 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
21438 PR rtl-optimization/79949
21439 * lra-constraints.c (process_alt_operands): Check memory when
21440 trying to predict a cycle. Print about the overall increase.
21442 2017-03-09 Richard Biener <rguenther@suse.de>
21444 PR middle-end/79971
21445 * gimple-expr.c (useless_type_conversion_p): Preserve
21446 TYPE_SATURATING for fixed-point types.
21448 2017-03-09 Richard Biener <rguenther@suse.de>
21451 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
21452 alignment of BLKmode params.
21454 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21457 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
21458 (VALL_NO_V2Q): Likewise.
21460 * config/aarch64/aarch64-simd.md
21461 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
21463 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
21464 VALL_NO_V2Q mode iterator.
21465 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
21467 2017-03-09 Martin Liska <mliska@suse.cz>
21469 PR tree-optimization/79631
21470 * tree-chkp-opt.c (chkp_is_constant_addr): Call
21471 tree_int_cst_sign_bit just for INTEGER constants.
21473 2017-03-09 Martin Liska <mliska@suse.cz>
21477 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
21480 2017-03-09 Marek Polacek <polacek@redhat.com>
21483 * tree.c (inchash::add_expr): Handle TREE_VEC.
21485 2017-03-09 Martin Liska <mliska@suse.cz>
21488 (chkp_narrow_size_and_offset): New function.
21489 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
21490 (void chkp_parse_bit_field_ref): New function.
21491 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
21492 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
21494 2017-03-09 Martin Liska <mliska@suse.cz>
21497 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
21498 (chkp_find_bounds_1): Remove gcc_unreachable.
21500 2017-03-09 Jakub Jelinek <jakub@redhat.com>
21503 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
21504 BUILT_IN_SYNC*, determine the access type from the size suffix and
21505 always build a MEM_REF with that type. Handle forgotten
21506 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
21509 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
21510 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
21511 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
21512 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
21513 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
21514 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
21515 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
21516 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
21517 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
21518 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
21519 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
21520 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
21521 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
21522 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
21523 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
21524 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
21525 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
21526 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
21527 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
21528 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
21529 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
21530 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
21531 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
21532 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
21533 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
21534 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
21535 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
21536 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
21537 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
21538 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
21539 definitions outside of __OPTIMIZE__ guarded section.
21542 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
21543 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
21544 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
21545 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
21548 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21550 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
21551 ("vfenez<mode>"): Add missing constraints.
21553 2017-03-08 Martin Sebor <msebor@redhat.com>
21556 * config/nds32/nds32.c (nds32_option_override):
21557 Fix misspelled diagnostic.
21559 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21562 * gimplify.c (gimplify_omp_for): Replace index var in outer
21563 taskloop statement with an artificial variable and add
21564 OMP_CLAUSE_PRIVATE clause for it.
21566 2017-03-08 Richard Biener <rguenther@suse.de>
21568 PR tree-optimization/79955
21569 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
21570 for accesses that are completely outside of the variable.
21572 2017-03-08 Andrew Haley <aph@redhat.com>
21574 PR tree-optimization/79943
21575 * tree-ssa-loop-split.c (compute_new_first_bound): When
21576 calculating the new upper bound, (END-BEG) should be added, not
21579 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21581 * config/avr/avr.md (setmemhi): Make sure match_dup
21582 operand number comes before match_scratch.
21584 2017-03-08 Richard Biener <rguenther@suse.de>
21586 PR tree-optimization/79920
21587 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
21588 with ncopies == 1 to ...
21589 (vect_transform_slp_perm_load): ... here. Properly compute
21590 all element loads by iterating VF times over the group. Do
21591 not handle ncopies (computed in a broken way) in
21592 vect_create_mask_and_perm.
21594 2017-03-08 Jakub Jelinek <jakub@redhat.com>
21597 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
21598 is a uniform vector, use uniform_vector_p return value instead of
21599 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
21601 2017-03-07 Marek Polacek <polacek@redhat.com>
21603 PR middle-end/79809
21604 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
21605 (alloca_call_type): Likewise.
21607 2017-03-07 Martin Liska <mliska@suse.cz>
21609 * gcov.c (process_args): Put comment to correct location.
21611 2017-03-07 Martin Liska <mliska@suse.cz>
21613 PR middle-end/68270
21614 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
21615 Use array_at_struct_end_p instead of DECL_CHAIN (field).
21616 (chkp_narrow_bounds_for_field): Likewise.
21617 (chkp_parse_array_and_component_ref): Pass one more argument to
21620 2017-03-07 Richard Biener <rguenther@suse.de>
21622 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
21625 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21627 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
21628 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
21630 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21633 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
21634 to end of description.
21635 (PARAM_MAX_STORES_TO_MERGE): Likewise.
21637 2017-03-07 Jakub Jelinek <jakub@redhat.com>
21639 PR rtl-optimization/79901
21640 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
21642 (*avx512f_<code><mode>3<mask_name>): ... this.
21643 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
21644 iterator instead of VI8_AVX2_AVX512BW.
21646 PR rtl-optimization/79901
21647 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
21648 min/max expander, expand it using expand_vec_cond_expr.
21651 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
21654 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21657 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
21658 to void * for PCH reasons.
21659 * dwarf2out.c (output_loc_operands, output_die): Cast
21660 v.val_vec.array to unsigned char *.
21662 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
21665 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
21668 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
21670 PR rtl-optimization/79571
21671 * lra-constraints.c (process_alt_operands): Calculate static
21672 reject and subtract it from overall when only addresses will be
21675 2017-03-06 Julia Koval <julia.koval@intel.com>
21678 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
21679 incoming stack boundary to 128 for 64-bit targets.
21681 2017-03-06 Richard Biener <rguenther@suse.de>
21683 PR tree-optimization/79894
21684 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
21685 to NULL after folding it.
21687 2017-03-06 Richard Biener <rguenther@suse.de>
21689 PR tree-optimization/79824
21690 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
21691 check disabling peeling for gaps.
21693 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
21695 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21696 attributes): Document gettimeofday.
21698 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21700 * config/s390/s390.c (s390_option_override_internal): Set
21701 PARAM_MIN_VECT_LOOP_BOUND
21703 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
21705 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
21706 * config/s390/s390.md: Likewise.
21708 2017-03-06 Jakub Jelinek <jakub@redhat.com>
21711 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
21712 (<avx2_avx512>_perm<mode>): Rename to ...
21713 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
21715 (<avx512>_perm<mode>_mask): Rename to ...
21716 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
21718 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
21719 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
21720 instead of VI8F_256_512.
21721 (avx512f_perm<mode>): New define_expand.
21722 (avx512f_perm<mode>_mask): Likewise.
21723 (avx512f_perm<mode>_1<mask_name>): New define_insn.
21724 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
21726 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21728 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
21729 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
21731 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
21733 2017-03-06 Martin Liska <mliska@suse.cz>
21736 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
21737 when having a SSA NAME w/o VAR_DECL assigned to it.
21739 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21741 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
21742 msa_dpsub_<su>_d): Fix MODE for vec_select.
21744 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
21746 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
21748 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
21750 2017-03-06 Richard Biener <rguenther@suse.de>
21752 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
21753 * plugin.c (register_plugin_info): Likewise.
21754 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
21756 2017-03-05 Jakub Jelinek <jakub@redhat.com>
21758 * config/i386/sse.md (sse_storehps, sse_storelps,
21759 avx_<castmode><avxsizesuffix>_<castmode>,
21760 avx512f_<castmode><avxsizesuffix>_<castmode>,
21761 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
21762 in condition that at least one operand is not a MEM.
21764 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21766 PR middle-end/79805
21767 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
21768 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
21770 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
21771 gimple_call_nothrow_p flag based on whether original builtin can throw.
21772 If it can, emit following stmts on the fallthrough edge.
21773 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
21774 don't create new bb if inserting just debug stmts on the edge, try to
21775 insert them on the fallthru bb or just reset debug stmts.
21777 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
21780 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
21781 restore recog_data (including the operand rtxes inside it) around
21782 the call to get_insn_template.
21784 2017-03-03 Martin Sebor <msebor@redhat.com>
21786 PR tree-optimization/79699
21787 * context.c (context::~context): Free MPFR caches to avoid
21788 a memory leak on program exit.
21790 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21792 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
21793 Use wide_int::ulow () instead of .elt (0).
21795 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21797 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
21798 (*pushxf): Limit oF constraint to 32bit targets and add oC
21799 constraint for 64bit targets.
21800 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
21801 (*pushdf): Change rmF constraint to rmC.
21803 2017-03-03 Martin Liska <mliska@suse.cz>
21805 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
21806 Remove unused variable.
21808 2017-03-03 Jakub Jelinek <jakub@redhat.com>
21811 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
21812 is a memory operand, increase num_memory.
21813 (ix86_expand_args_builtin): Likewise.
21815 2017-03-03 Jan Hubicka <jh@suse.cz>
21818 * ipa-devirt.c (maybe_record_node): Properly handle
21819 __cxa_pure_virtual visibility.
21821 2017-03-03 Martin Liska <mliska@suse.cz>
21823 PR tree-optimization/79803
21824 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
21826 (pass_loop_prefetch::execute): Disabled optimization if an
21827 assumption about L1 cache size is not met.
21829 2017-03-03 Martin Liska <mliska@suse.cz>
21831 PR rtl-optimization/79574
21832 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
21833 (hash_scan_set): Likewise.
21834 (dump_hash_table): Likewise.
21835 (hoist_code): Likewise.
21837 2017-03-03 Richard Biener <rguenther@suse.de>
21839 * fixed-value.c (fixed_from_string): Restore use of elt (1)
21840 in place of uhigh ().
21841 (fixed_convert_from_real): Likewise.
21843 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
21846 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
21848 2017-03-03 Richard Biener <rguenther@suse.de>
21850 PR middle-end/79818
21851 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
21852 TYPE_OVERFLOW_UNDEFINED check.
21854 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21856 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
21858 (vector_ae_<mode>_p): Likewise.
21859 (vector_nez_<mode>_p): Likewise.
21860 (vector_ne_v2di_p): Likewise.
21861 (vector_ae_v2di_p): Likewise.
21862 (vector_ne_<mode>_p): Likewise.
21863 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
21865 (vsx_tsqrt<mode>2_fe): Likewise.
21867 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
21870 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
21872 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21874 PR rtl-optimization/79780
21875 * cprop.c (one_cprop_pass): When second and further conditional trap
21876 in a single basic block is turned into an unconditional trap, turn it
21877 into a deleted note to avoid RTL verification failures.
21879 2017-03-02 Richard Biener <rguenther@suse.de>
21881 * fold-const.c (const_binop): Use ulow () instead of elt (0).
21883 2017-03-02 Richard Biener <rguenther@suse.de>
21885 PR tree-optimization/79345
21887 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
21888 param and abort the walk, returning -1 if it is hit.
21889 (walk_aliased_vdefs): Take a limit param and pass it on.
21890 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
21891 defaulting to 0 and return a signed int.
21892 * tree-ssa-uninit.c (struct check_defs_data): New struct.
21893 (check_defs): New helper.
21894 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
21895 about uninitialized memory.
21896 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
21897 bogus uninitialized warning.
21898 (fixed_convert_from_real): Likewise.
21900 2017-03-02 Bin Cheng <bin.cheng@arm.com>
21902 PR tree-optimization/66768
21903 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
21904 iv_use if base object can't be determined.
21906 2017-03-02 Jakub Jelinek <jakub@redhat.com>
21908 PR tree-optimization/79345
21909 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
21910 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
21911 (get_pattern_stats): Initialize it.
21912 * genemit.c (gen_expand): Verify match_scratch numbers come after
21913 match_operand/match_dup numbers.
21914 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
21915 match_scratch numbers.
21916 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
21918 * config/s390/s390.md (trunctdsd2): Likewise.
21920 2017-03-02 Richard Biener <rguenther@suse.de>
21922 * wide-int.h (wide_int_storage::operator=): Implement in terms
21925 2017-03-02 Richard Biener <rguenther@suse.de>
21927 PR tree-optimization/79777
21928 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
21929 the to insert expression to sth existing.
21931 2017-03-01 Martin Sebor <msebor@redhat.com>
21933 PR middle-end/79692
21934 * gimple-ssa-sprintf.c
21935 (directive::known_width_and_precision): New function.
21936 (format_integer): Use it.
21937 (get_mpfr_format_length): Consider the full range of precision
21938 when computing %g output with the # flag. Set the likely byte
21939 count to 3 rather than 1 when precision is indeterminate.
21940 (format_floating): Correct the lower bound of precision.
21942 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21944 * doc/invoke.texi: Document default code model for 64-bit Linux.
21946 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21949 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
21950 udiv rather than div since input pattern is unsigned.
21952 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
21954 * config/i386/i386.c (print_reg): Warn for values of
21955 unsupported size in integer register.
21957 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
21960 * config/rs6000/predicates.md (current_file_function_operand): Do
21961 not allow self calls to be local if the function is replaceable.
21963 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
21966 * config/rs6000/altivec.h (vec_ctz and others): Change the
21967 preprocessor macro that controls conditional compilation from
21968 _ARCH_PWR9 to __POWER9_VECTOR__.
21969 (vec_all_ne): Change parameterization of __altivec_scalar_pred
21970 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
21971 control (instead of _ARCH_PWR9 control) so that template
21972 definition uses power9-specific function.
21973 (vec_any_eq): Likewise.
21974 (vec_all_ne): Change macro definition to use a power9-specific
21975 expansion under #ifdef __POWER9_VECTOR__ control (instead of
21976 _ARCH_PWR9 control).
21977 (vec_any_eq) Likewise.
21978 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
21979 expansion for CMPNEF to remove support for xvcmpnesp instruction.
21980 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
21981 support for xvcmpnedp instruction.
21982 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
21983 macro expansion so that Power9 implementation of vec_all_ne does
21984 not use the AltiVec predicate framework.
21985 (VCMPNEH_P): Likewise.
21986 (VCMPNEW_P): Likewise.
21987 (VCMPNED_P): Likewise.
21988 (VCMPNEFP_P): Likewise.
21989 (VCMPNEDP_P): Likewise.
21990 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
21991 implementation of vec_any_eq to not use AltiVec predicate
21993 (VCMPAEH_P): Likewise.
21994 (VCMPAEW_P): Likewise.
21995 (VCMPAED_P): Likewise.
21996 (VCMPAEFP_P): Likewise.
21997 (VCMPAEDP_P): Likewise.
21998 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
21999 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
22000 not use the AltiVec predicate framework.
22001 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
22002 of vec_any_eq to not use AltiVec predicate framework.
22003 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22004 support for predefined __POWER9_VECTOR__ macro to indicate that
22005 Power9 instruction selection is enabled.
22006 (altivec_overloaded_builtins): Remove extraneous
22007 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
22008 function argument types RS6000_BTI_bool_V16QI and
22009 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
22010 entry for overloaded function argument types RS6000_BTI_bool_V4SI
22011 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
22012 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
22013 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
22014 Power9 for implementations of vec_cmpne. Change the signature for
22015 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
22016 (representing vec_all_ne) to remove the previously described first
22017 argument of type RS6000_BTI_INTSI, as this was an artifact of
22018 reliance on the AltiVec predicate framework, which is no longer
22019 used in the implementation of these functions. Add
22020 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
22021 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
22022 since, unlike the AltiVec predicate framework implementation, we
22023 do not share function descriptors between vec_alle and vec_anyeq.
22024 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
22025 set of modes that receive special treatment even when
22026 TARGET_P9_VECTOR is true. The special treatment emits code that
22027 does not depend on Power9 instructions.
22028 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
22029 define_expand to not rely on AltiVec predicate framework.
22030 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22032 (vector_ne_v2di_p): Change this define_expand to not rely on
22033 AltiVec predicate framework.
22034 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
22036 (vector_ne_<mode>_p): Change this define_expand to not rely on
22037 AltiVec predicate framework.
22038 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
22040 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
22041 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
22042 define_insn pattern.
22043 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
22044 define_insn pattern because the xvcmpne<VSs>. instruction is not
22046 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
22047 instruction is not supported.
22049 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22051 * config/nvptx/nvptx.c: Include intl.h.
22053 2017-03-01 Martin Jambor <mjambor@suse.cz>
22056 * ipa-prop.h (ipa_bits): Removed field known.
22057 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
22058 to pointers. Adjusted their comments to warn about their sharing.
22059 (ipcp_transformation_summary): Change bits to a vector of pointers.
22060 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
22061 (ipa_get_ipa_bits_for_value): Declare.
22062 * tree-vrp.h (value_range): Mark as GTY((for_user)).
22063 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
22064 (ipa_bits_hash_table): Likewise.
22065 (ipa_vr_ggc_hash_traits): Likewise.
22066 (ipa_vr_hash_table): Likewise.
22067 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
22068 being pointers and vr_known being removed.
22069 (ipa_set_jf_unknown): Likewise.
22070 (ipa_get_ipa_bits_for_value): New function.
22071 (ipa_set_jfunc_bits): Likewise.
22072 (ipa_get_value_range): New overloaded functions.
22073 (ipa_set_jfunc_vr): Likewise.
22074 (ipa_compute_jump_functions_for_edge): Use the above functions to
22075 construct bits and vr parts of jump functions.
22076 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
22077 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22079 (ipcp_grow_transformations_if_necessary): Also allocate
22080 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
22082 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
22083 them. Fix too long lines.
22084 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
22085 vr_known being removed.
22086 (ipa_read_jump_function): Use new setter functions to construct bits
22087 and vr parts of jump functions or set them to NULL.
22088 (write_ipcp_transformation_info): Adjust for bits being pointers.
22089 (read_ipcp_transformation_info): Likewise.
22090 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
22092 Include gt-ipa-prop.h.
22093 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
22095 (ipcp_store_bits_results): Likewise.
22096 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
22097 Do not write to existing jump functions but use a temporary instead.
22099 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22102 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
22103 attempt to use its first operand as BIT_FIELD_REF base.
22105 2017-03-01 Richard Biener <rguenther@suse.de>
22107 PR middle-end/79721
22108 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
22109 interpolating formula in wrapping arithmetic.
22110 (chrec_apply): Convert chrec_evaluate return value to wanted type.
22112 2017-03-01 Jakub Jelinek <jakub@redhat.com>
22114 PR tree-optimization/79734
22115 * tree-vect-generic.c (expand_vector_condition): Optimize
22116 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
22117 Handle VEC_COND_EXPR where comparison has different inner width from
22118 type's inner width.
22120 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
22122 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
22123 markup, and similar issues. Remove @opindex entries for things
22124 that aren't options. Add missing -mmpy-option entries.
22126 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22128 PR tree-optimization/79737
22129 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
22130 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
22131 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
22132 instead of byte_size. Formatting fix.
22133 (shift_bytes_in_array_right): Formatting fix.
22135 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
22138 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
22139 condition on optimize for the leaf function test.
22141 2017-02-28 Martin Liska <mliska@suse.cz>
22144 * read-rtl-function.c (function_reader::handle_unknown_directive):
22145 Bail out when one uses -flto.
22147 2017-02-28 Martin Liska <mliska@suse.cz>
22149 * common.opt: Replace space with tabular for options of <number>
22151 * config/i386/i386.opt: Show <number> value for
22152 -mlarge-data-threshold.
22153 * opts.c (print_filtered_help): Do not display number in hexadecimal
22156 2017-02-28 Martin Liska <mliska@suse.cz>
22158 * common.opt: Fix --help=option -Q for options which are of
22161 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
22163 * config/i386/i386.c (print_reg): Error out for values
22164 of 8-bit size in invalid integer register.
22166 2017-02-28 Martin Sebor <msebor@redhat.com>
22168 PR tree-optimization/79691
22169 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
22171 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22174 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
22175 gcc_unreachable with output_operand_lossage.
22177 2017-02-28 Richard Biener <rguenther@suse.de>
22179 PR tree-optimization/79740
22180 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
22182 (visit_nary_op): Insert the nary into the hashtable if we
22183 pattern-matched sth.
22184 * tree-ssa-pre.c (eliminate_insert): Robustify.
22186 2017-02-28 Richard Biener <rguenther@suse.de>
22188 PR middle-end/79731
22189 * fold-const.c (decode_field_reference): Reject out-of-bound
22192 2017-02-28 Jakub Jelinek <jakub@redhat.com>
22194 * config/i386/i386.c: Include intl.h.
22195 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
22196 instead of just cond ? "..." : "...".
22197 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
22198 * coverage.c (read_counts_file): Likewise.
22199 * omp-offload.c: Include intl.h.
22200 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
22201 of just cond ? "..." : "...".
22202 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
22203 of just cond ? "..." : "...".
22205 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
22208 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
22210 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
22211 'tune for' CPU name.
22212 * config/arm/arm-cpu-data.h: Regenerated.
22214 2017-02-28 Richard Biener <rguenther@suse.de>
22216 PR tree-optimization/79732
22217 * tree-inline.c (expand_call_inline): Do not shadow var.
22219 2017-02-28 Richard Biener <rguenther@suse.de>
22221 PR tree-optimization/79723
22222 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
22223 address-space properly.
22225 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
22227 * doc/optinfo.texi (Optimization groups): Fix option used for
22229 * doc/invoke.texi (-fopt-info): Document "omp".
22230 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
22231 (OPTGROUP_ALL): Add OPTGROUP_OMP.
22232 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
22233 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
22234 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
22236 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
22238 * dumpfile.c (optgroup_options): Instead of "openmp", associate
22239 OPTGROUP_OMP with "omp".
22241 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
22244 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
22245 for arithmetic shift of unsigned V2DI.
22247 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
22249 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
22250 arc/linux.h headers.
22251 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
22252 (LINK_SPEC): Likewise.
22253 (ARC_TLS_EXTRA_START_SPEC): Likewise.
22254 (EXTRA_SPECS): Likewise.
22255 (STARTFILE_SPEC): Likewise.
22256 (ENDFILE_SPEC): Likewise.
22257 (LIB_SPEC): Likewise.
22258 (TARGET_SDATA_DEFAULT): Likewise.
22259 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
22260 (MULTILIB_DEFAULTS): Likewise.
22261 (DWARF2_UNWIND_INFO): Likewise.
22262 * config/arc/big.h: New file.
22263 * config/arc/elf.h: Likewise.
22264 * config/arc/linux.h: Likewise.
22265 * config/arc/t-uClibc: Remove.
22267 2017-02-27 Bin Cheng <bin.cheng@arm.com>
22269 PR tree-optimization/77536
22270 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
22271 (tree_transform_and_unroll_loop): Use above function to compute the
22272 estimated niter of unrolled loop and use it when scaling profile.
22273 Also use count info rather than frequency if it's non-zero.
22274 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
22275 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
22276 (vect_transform_loop): Call above function.
22278 2017-02-27 Richard Biener <rguenther@suse.de>
22280 PR tree-optimization/45397
22281 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
22282 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
22283 (visit_nary_op): Add pattern matching for CSEing sign-changed
22284 or truncated operations with wider ones.
22286 2017-02-27 Richard Biener <rguenther@suse.de>
22288 PR tree-optimization/79690
22289 * tree-vect-stmts.c (vectorizable_store): Use vector type
22290 built from the DR with address-space.
22292 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
22294 * doc/invoke.texi (Optimize Options): Refine the description
22295 of asan-use-after-return.
22297 2017-02-25 Alan Modra <amodra@gmail.com>
22299 PR rtl-optimization/79584
22300 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
22301 base, not ad->base_term, the reg within base. Remove assertion
22302 that ad->base == ad->base_term. Replace gen_int_mode using
22303 bogus mode with const0_rtx.
22305 2017-02-25 Jakub Jelinek <jakub@redhat.com>
22307 PR middle-end/79396
22308 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
22309 FMA_EXPR like tcc_binary or tcc_unary.
22311 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
22314 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
22316 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
22317 (output_loc_operands): Handle DW_OP_call_ref and
22318 DW_OP_GNU_variable_value.
22319 (struct variable_value_struct): New type.
22320 (struct variable_value_hasher): Likewise.
22321 (variable_value_hash): New variable.
22322 (string_types): Remove.
22323 (copy_loc_descr): New function.
22324 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
22325 (prepend_loc_descr_to_each): New function.
22326 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
22327 instead of add_loc_descr_to_each if the first argument is single
22328 location list and the second has multiple.
22329 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
22330 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
22331 when looking for variable value which doesn't have other location info.
22332 (loc_list_from_tree): Formatting fix.
22333 (gen_array_type_die): Simplify DW_AT_string_length handling.
22334 (adjust_string_types): Remove.
22335 (gen_subprogram_die): Don't call adjust_string_types nor test/set
22336 string_types. Call resolve_variable_values.
22337 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
22338 (resolve_addr_in_expr): Likewise. Add A argument.
22339 (copy_deref_exprloc): Remove deref argument. Adjust for the
22340 original expression being DW_OP_GNU_variable_value with optionally
22341 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
22342 optionally after it.
22343 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
22344 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
22345 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
22346 (variable_value_hasher::hash, variable_value_hasher::equal): New
22348 (resolve_variable_value_in_expr, resolve_variable_value,
22349 resolve_variable_values, note_variable_value_in_expr,
22350 note_variable_value): New functions.
22351 (dwarf2out_early_finish): Call note_variable_value on all toplevel
22354 2017-02-24 Jakub Jelinek <jakub@redhat.com>
22357 * opts.h (handle_generated_option): Add GENERATED_P argument.
22358 * opts-common.c (handle_option): Adjust function comment.
22359 (handle_generated_option): Add GENERATED_P argument, pass it to
22361 (control_warning_option): Pass false to handle_generated_option
22363 * opts.c (maybe_default_option): Pass true to handle_generated_option
22365 * optc-gen.awk: Likewise.
22367 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22369 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
22370 a REG, look at the REG it is a SUBREG of.
22371 (splitter for cmpeqsi_t): Ditto.
22373 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22375 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
22376 the special USEs with the pattern of the insn, not the insn itself.
22378 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
22381 * doc/invoke.texi: Document -mload-store-pairs.
22383 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22384 Sandra Loosemore <sandra@codesourcery.com>
22386 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
22387 argument isn't a CONST_INT.
22388 (nios2_alternate_compare_const): Assert op is a CONST_INT.
22389 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
22390 (nios2_validate_compare): Bypass alternate compare logic if *op2
22391 is not a CONST_INT.
22392 (ldstwm_operation_p): Return false if first_base is not a REG or
22393 if first_offset is not a CONST_INT.
22395 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22397 * config/cris/cris.md: Use correct operand in a define_peephole2.
22399 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22401 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
22403 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
22405 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
22406 this_insn if it is an INSN or JUMP_INSN.
22407 (force_offsettable): Look at base, not at addr.
22408 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
22409 on things that aren't necessarily CONST_INTs.
22411 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
22413 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
22414 -mfpmath=sse is the default also for x86-32 targets with SSE2
22415 instruction set when @option{-ffast-math} is enabled
22417 2017-02-24 Jeff Law <law@redhat.com>
22419 PR rtl-optimizatoin/79286
22420 * ira.c (update_equiv_regs): Drop may_trap_p exception to
22423 2017-02-24 Richard Biener <rguenther@suse.de>
22425 PR tree-optimization/79389
22426 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
22429 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
22431 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
22432 function comment to reflect reality.
22433 (loop_exits_before_overflow): Fix typo in function description.
22435 2017-02-24 Richard Biener <rguenther@suse.de>
22437 PR tree-optimization/79389
22438 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
22439 properly that a threading opportunity exists. Detect conditional
22440 copy/constant propagation opportunities.
22442 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
22444 * config/visium/visium.md (type): Add trap.
22445 (b): New mode attribute.
22446 (*btst): Rename into...
22447 (*btst<mode>): ...this and adjust.
22448 (*cbranchsi4_btst_insn): Rename into...
22449 (*cbranch<mode>4_btst_insn): ...this and adjust.
22450 (trap): New define_insn.
22452 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22454 PR tree-optimization/79389
22455 * ifcvt.c (struct noce_if_info): Add rev_cond field.
22456 (noce_reversed_cond_code): New function.
22457 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
22458 reversed_comparison_code. Formatting fix.
22459 (noce_try_store_flag): Test rev_cond != NULL in addition to
22460 reversed_comparison_code.
22461 (noce_try_store_flag_constants): Likewise.
22462 (noce_try_store_flag_mask): Likewise.
22463 (noce_try_addcc): Use rev_cond if non-NULL instead of
22464 reversed_comparison_code.
22465 (noce_try_cmove_arith): Likewise. Formatting fixes.
22466 (noce_try_minmax, noce_try_abs): Clear rev_cond.
22467 (noce_find_if_block): Initialize rev_cond.
22468 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
22469 instead of false as last argument never attempt to reverse it
22472 2017-02-23 Bin Cheng <bin.cheng@arm.com>
22474 PR tree-optimization/79663
22475 * tree-predcom.c (combine_chains): Process refs in reverse order
22476 only for ZERO length chains, and add explaining comment.
22478 2017-02-23 Jeff Law <law@redhat.com>
22480 PR tree-optimization/79578
22481 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
22482 in call to operand_equal_p.
22484 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
22487 * config/i386/cpuid.h: Fix another undefined behavior.
22489 2017-02-23 Richard Biener <rguenther@suse.de>
22491 PR tree-optimization/79683
22492 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
22493 vector types for data-refs.
22495 2017-02-23 Martin Liska <mliska@suse.cz>
22497 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
22499 2017-02-23 Jakub Jelinek <jakub@redhat.com>
22501 PR middle-end/79665
22502 * internal-fn.c (get_range_pos_neg): Moved to ...
22503 * tree.c (get_range_pos_neg): ... here. No longer static.
22504 * tree.h (get_range_pos_neg): New prototype.
22505 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
22506 are known to be in between 0 and signed maximum inclusive, try to
22507 expand both unsigned and signed divmod and use the cheaper one from
22510 2017-02-22 Jeff Law <law@redhat.com>
22512 PR tree-optimization/79578
22513 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
22514 to compare base operands.
22516 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
22519 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
22520 gpc_reg_operand instead of fpr_reg_operand.
22522 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
22524 * config/mips/mips.c (mips_return_in_memory): Force FP
22525 vector types to be returned in memory for o32 ABI.
22527 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22529 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
22530 instead of DW_TAG_member for static data member declarations and don't
22531 set no_linkage_name for static inline data members.
22532 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
22535 2017-02-22 Martin Liska <mliska@suse.cz>
22537 * doc/invoke.texi: Replace inequality signs with square brackets
22540 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22542 PR tree-optimization/68644
22543 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
22545 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
22548 * lra-constraints.c (simplify_operand_subreg): Handle
22549 WORD_REGISTER_OPERATIONS targets.
22551 2017-02-22 Jakub Jelinek <jakub@redhat.com>
22554 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
22555 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
22556 elimination by swapping fld*.
22558 2017-02-22 Richard Biener <rguenther@suse.de>
22560 PR tree-optimization/79673
22561 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
22562 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
22563 irrelevant address-space qualifiers and avoiding a
22564 ADDR_SPACE_CONVERT_EXPR from fold_convert.
22566 2017-02-22 Richard Biener <rguenther@suse.de>
22568 PR tree-optimization/79666
22569 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
22570 to not symbolically negate if that may introduce undefined
22573 2017-02-22 Martin Liska <mliska@suse.cz>
22576 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
22577 * data-streamer-out.c (streamer_write_gcov_count_stream):
22579 * value-prof.c (stream_out_histogram_value): Make assert more
22580 precise based on type of counter.
22582 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
22585 * config/i386/i386.md (standard_x87sse_constant_load splitter):
22586 Use nonimmediate_operand instead of memory_operand for operand 1.
22587 (float-extend standard_x87sse_constant_load splitter): Ditto.
22589 2017-02-21 Jeff Law <law@redhat.com>
22591 PR tree-optimization/79621
22592 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
22593 blocks with edges to themselves.
22595 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22598 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
22599 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
22600 Use gimple_call_builtin_p.
22603 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
22604 on temporarily removed DEBUG_INSNs.
22606 PR tree-optimization/79649
22607 * tree-loop-distribution.c (classify_partition): Give up on
22608 non-generic address space loads/stores.
22610 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
22612 * doc/loop.texi (Loop manipulation): Remove nonexistent
22613 tree_ssa_loop_version from the documentation.
22614 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
22616 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22619 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
22620 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
22621 * config/rs6000/rs6000.c: Include except.h.
22622 (rs6000_expand_split_stack_prologue): Call
22623 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
22625 2017-02-21 Martin Jambor <mjambor@suse.cz>
22628 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
22629 have been analyzed.
22631 2017-02-21 Martin Jambor <mjambor@suse.cz>
22633 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
22634 for backward compatibility only.
22635 * doc/invoke.texi (Option Summary): Remove all references to
22636 -fipa-cp-alignment.
22638 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
22642 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22644 * lra-constraints.c (curr_insn_transform): Handle
22645 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22647 2017-02-21 Martin Liska <mliska@suse.cz>
22649 * config/i386/i386.opt: Replace -masm-dialect with -masm.
22651 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
22653 PR translation/79638
22654 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
22656 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
22659 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
22660 (arm_function_ok_for_sibcall): Return false for an indirect call by
22661 descriptor if all the argument registers are used.
22662 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
22663 alignment of the function.
22665 2017-02-21 Jakub Jelinek <jakub@redhat.com>
22667 PR tree-optimization/61441
22668 * simplify-rtx.c (simplify_const_unary_operation): For
22669 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
22670 the sNaN unmodified.
22672 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22674 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22675 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22676 instead of SYSTEM_HEADER_DIR.
22678 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
22679 Martin Liška <mliska@suse.cz>
22681 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
22682 Fix typos and grammar, use active voice, and clarify.
22684 2017-02-20 Marek Polacek <polacek@redhat.com>
22686 PR middle-end/79537
22687 * gimplify.c (gimplify_expr): Handle unused *&&L;.
22690 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
22692 2017-02-20 Jakub Jelinek <jakub@redhat.com>
22695 * config/i386/i386.c (ix86_expand_builtin): Handle
22696 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
22697 ix86_builtins_isa[fcode].isa as a requirement of those
22698 flags and any other flag in the bitmask.
22699 (ix86_init_mmx_sse_builtins): Use 0 instead of
22700 ~OPTION_MASK_ISA_64BIT as mask.
22701 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
22702 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
22703 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
22704 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
22706 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22709 * lra-constraints.c (split_reg): Check requested split mode
22710 is supported by the register.
22712 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22714 * lra-constraints.c (simplify_operand_subreg): Remove early
22717 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22720 * lra-constraints.c (curr_insn_transform): Tighten condition
22721 for converting SUBREG reloads from OP_OUT to OP_INOUT.
22723 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
22726 * lra-constraints.c (curr_insn_transform): Handle
22727 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
22729 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
22732 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
22734 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
22736 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
22739 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
22742 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22744 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
22747 2017-02-18 Jakub Jelinek <jakub@redhat.com>
22749 * final.c (last_columnnum, override_columnnum): New variables.
22750 (final_start_function): Set last_columnnum, pass it to begin_prologue
22751 hook and pass 0 to dwarf2out_begin_prologue.
22752 (final_scan_insn): Update override_columnnum. Pass last_columnnum
22753 to source_line debug hook.
22754 (notice_source_line): Compute last_columnnum and for debug_column_info
22755 return true on column changes.
22756 * debug.h (struct gcc_debug_hooks): Add column argument to
22757 source_line and begin_prologue hooks.
22758 (debug_nothing_int_charstar_int_bool): Remove prototype.
22759 (debug_nothing_int_int_charstar,
22760 debug_nothing_int_int_charstar_int_bool): New prototypes.
22761 (dwarf2out_begin_prologue): Add column argument.
22762 * debug.c (do_nothing_debug_hooks): Adjust source_line and
22763 begin_prologue hooks.
22764 (debug_nothing_int_charstar_int_bool): Remove.
22765 (debug_nothing_int_int_charstar,
22766 debug_nothing_int_int_charstar_int_bool): New functions.
22767 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
22768 through to dwarf2out_source_line.
22769 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
22770 (dwarf2out_source_line): Add column argument, emit it if requested.
22771 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
22773 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22774 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
22775 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
22776 through to dwarf2out_begin_prologue.
22777 (vmsdbgout_source_line): Add column argument, pass it through to
22778 dwarf2out_source_line.
22779 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
22780 dbxout_source_line caller.
22781 (dbxout_source_line): Add column argument.
22783 * common.opt (gno-column-info, gcolumn-info): New options.
22784 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
22785 (check_die): Also test for multiple DW_AT_decl_column attributes.
22786 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
22787 DW_AT_decl_column if requested.
22788 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
22790 (gen_variable_die): Likewise.
22791 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
22792 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
22795 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
22796 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
22797 (ix86_handle_option): Handle OPT_m3dnowa.
22798 * doc/invoke.texi (-m3dnowa): Document.
22799 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
22800 -m3dnowa instead of -m3dnow -march=athlon.
22803 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
22804 instead of gcc_assert for K, r and R code checks. Formatting fixes.
22806 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22809 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
22810 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
22811 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
22812 generator for vsx_xxpermdi_<mode>_be.
22813 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
22814 force big-endian semantics.
22815 (vsx_xxpermdi_<mode>_be): New define_expand with same
22816 implementation as previous version of vsx_xxpermdi_<mode>.
22818 2017-02-17 Jakub Jelinek <jakub@redhat.com>
22820 PR tree-optimization/79327
22821 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
22822 variable, its initialization and use.
22824 2017-02-17 Julia Koval <julia.koval@intel.com>
22826 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
22827 (OPTION_MASK_ISA_PKU_UNSET): New.
22828 (ix86_handle_option): Handle -mrdpid.
22829 * config/i386/cpuid.h (bit_RDPID): New.
22830 * config/i386/driver-i386.c (host_detect_local_cpu):
22831 Detect RDPID feature.
22832 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
22833 * config/i386/i386-c.c (ix86_target_macros_internal):
22835 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
22836 (ix86_valid_target_attribute_inner_p): Add "rdpid".
22837 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22838 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
22839 * config/i386/i386.md (define_insn "rdpid"): New.
22840 * config/i386/i386.opt Add -mrdpid.
22841 * config/i386/immintrin.h (_rdpid_u32): New.
22843 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
22845 PR rtl-optimization/79541
22846 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
22847 instead of transforming it into USE.
22849 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
22851 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
22852 If HONOR_SNANS (SFmode) force the input to a register.
22853 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
22854 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
22855 an frsp or similar insn.
22857 2017-02-17 Martin Liska <mliska@suse.cz>
22859 PR rtl-optimization/79577
22860 * params.def (selsched-max-sched-times): Increase minimum to 1.
22862 2017-02-17 Martin Liska <mliska@suse.cz>
22864 PR rtl-optimization/79574
22865 * gcse.c (want_to_gcse_p): Prevent integer overflow.
22867 2017-02-17 Martin Liska <mliska@suse.cz>
22869 PR tree-optimization/79529
22870 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
22871 ssa_defined_default_def_p to handle cases which are implicitly
22873 * tree-ssa.c (ssa_defined_default_def_p): New function.
22874 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
22875 which are implicitly defined.
22876 * tree-ssa.h (ssa_defined_default_def_p): Declare.
22878 2017-02-17 Richard Biener <rguenther@suse.de>
22880 PR middle-end/79576
22881 * params.def (max-ssa-name-query-depth): Limit to 10.
22883 2017-02-17 Richard Biener <rguenther@suse.de>
22885 PR tree-optimization/79552
22886 * tree-ssa-structalias.c (visit_loadstore): Properly verify
22889 2017-02-17 Richard Biener <rguenther@suse.de>
22892 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
22894 2017-02-17 Marek Polacek <polacek@redhat.com>
22896 PR middle-end/79536
22897 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
22898 (fold_negate_expr): New wrapper.
22900 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
22902 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
22903 Correct terminology and de-emphasize pre-standard behavior.
22905 2017-02-16 Alan Modra <amodra@gmail.com>
22907 PR rtl-optimization/79286
22908 * ira.c (def_dominates_uses): New function.
22909 (update_equiv_regs): Don't create an equivalence for insns that
22910 may trap where the register def does not dominate the use.
22912 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
22914 PR rtl-optimization/78127
22915 * lra.c (lra): Call lra_eliminate before finish the loop after
22918 2017-02-16 Richard Biener <rguenther@suse.de>
22920 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
22922 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
22923 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
22924 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
22925 (isl_val_int_from_wi): New function.
22926 (extract_affine_gmp): Rename to ...
22927 (extract_affine_wi): ... this, take a widest_int.
22928 (extract_affine_int): Just wrap extract_affine_wi.
22929 (add_param_constraints): Use isl_val_int_from_wi.
22930 (add_loop_constraints): Likewise, and extract_affine_wi.
22932 2017-02-15 Jeff Law <law@redhat.com>
22934 PR middle-end/79521
22935 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
22936 ira_init_register_move_cost_if_necessary.
22938 2017-02-15 Martin Sebor <msebor@redhat.com>
22940 PR middle-end/32003
22941 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
22942 removed in a prior commit.
22944 2017-02-15 Bin Cheng <bin.cheng@arm.com>
22946 PR tree-optimization/79347
22947 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
22948 counters during peeling.
22950 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
22952 * Makefile.in (site.exp): Remove "set ISLVER".
22954 2017-02-15 Jakub Jelinek <jakub@redhat.com>
22957 * real.c (real_from_integer): Call real_convert even for decimal.
22959 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
22962 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
22964 2017-02-14 Andrew Pinski <apinski@cavium.com>
22966 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
22967 cores and change the partno/implementer to be correct.
22968 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
22969 the 'B" as the implementer.
22970 * config/aarch64/aarch64-tune.md: Regenerate.
22972 2017-02-14 Carl Love <cel@us.ibm.com>
22974 * config/rs6000/rs6000.c: Add case statement entry to make the
22975 xvcvuxdsp built-in argument unsigned.
22976 * config/rs6000/vsx.md: Fix the source and return operand types so they
22977 match the instruction definitions from the ISA document. Fix typo
22978 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
22981 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
22984 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
22985 member early_clobber_alts.
22986 * lra-lives.c (reg_early_clobber_p): New.
22987 (process_bb_lives): Use it.
22988 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
22989 (debug_operand_data): Initialize early_clobber_alts.
22990 (setup_operand_alternative): Set up early_clobber_alts.
22991 (collect_non_operand_hard_regs): Ditto. Pass early clobber
22992 alternatives to new_insn_reg.
22993 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
22995 (lra_update_insn_regno_info): Pass the new arg.
22997 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22999 PR middle-end/79505
23000 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
23001 (new_oacc_loop_raw): Don't clear already cleared fields.
23004 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
23005 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
23006 _mm512_prefetch_i64gather_ps): New inline functions and macros.
23008 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
23011 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
23013 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
23016 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
23017 the extra instruction to the right place to store 128-bit constant
23020 2017-02-14 Martin Sebor <msebor@redhat.com>
23022 PR middle-end/79448
23023 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
23024 warning for strings of unknown length.
23026 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23028 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
23030 2017-02-14 Jeff Law <law@redhat.com>
23033 * ira-costs.c (scan_one_insn): Initialize register move costs
23034 for pseudos seen in USE/CLOBBER insns.
23036 PR tree-optimization/79095
23037 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
23038 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
23039 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
23040 if the operands are known to be not equal, then the resulting range
23042 (intersect_ranges): If the new range is ~[0,0] and the old range is
23043 wide, then prefer ~[0,0].
23044 * tree-vrp.c (overflow_comparison_p_1): New function.
23045 (overflow_comparison_p): New function.
23046 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
23047 if NAME is used in an overflow test.
23048 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
23049 overflow check that can be expressed as an equality test, then adjust
23050 ops to be that equality test.
23052 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23054 * config/s390/s390-builtin-types.def: Remove flags argument.
23055 * config/s390/s390.c (s390_init_builtins): Likewise.
23057 2017-02-14 Martin Liska <mliska@suse.cz>
23059 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
23060 vector. Fix trailing white spaces.
23062 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
23064 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
23067 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23069 PR rtl-optimization/68664
23070 * config/arm/arm.c (arm_sched_can_speculate_insn):
23071 New function. Declare prototype.
23072 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23074 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23076 PR rtl-optimization/68664
23077 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
23079 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
23081 2017-02-14 Amit Pawar <amit.pawar@amd.com>
23083 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
23084 max skip bytes for function, loop and jump.
23086 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23088 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
23089 ABS_EXPR for gimple dump.
23091 2017-02-14 Jakub Jelinek <jakub@redhat.com>
23094 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
23096 PR tree-optimization/79408
23097 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
23098 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
23099 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
23100 also if rhs1 is INTEGER_CST.
23102 2017-02-14 Richard Biener <rguenther@suse.de>
23104 PR middle-end/79432
23105 * tree-into-ssa.c (insert_phi_nodes): When the function can
23106 have abnormal edges rewrite SSA names with broken use-def
23107 dominance out of SSA and register them for PHI insertion.
23109 2017-02-13 Martin Sebor <msebor@redhat.com>
23111 PR middle-end/79496
23112 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
23113 clearing info.nowrite flag when snprintf size argument is a range.
23115 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23117 * cprop.c (cprop_jump): Add missing space in string literal.
23118 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
23119 (get_constraint_for_component_ref): Likewise.
23120 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
23121 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
23122 * lra-constraints.c (process_alt_operands): Likewise.
23123 * ipa-inline.c (inline_small_functions): Likewise.
23124 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
23125 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23126 * trans-mem.c (diagnose_tm_1_op): Likewise.
23127 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
23128 (grid_parallel_clauses_gridifiable): Likewise.
23130 * config/nvptx/mkoffload.c (process): Add space in between
23133 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
23134 "MOD4_SSE_REGS" and "ALL_REGS".
23136 * spellcheck.c (test_data): Add , in between "foo" and "food".
23138 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23141 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
23142 boundary crossing check and subsequent code generation agree.
23144 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23146 * config/aarch64/aarch64.c (has_memory_op): Delete.
23147 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
23150 2017-02-13 Jakub Jelinek <jakub@redhat.com>
23152 PR rtl-optimization/79388
23153 PR rtl-optimization/79450
23154 * combine.c (distribute_notes): When removing TEM_INSN for which
23155 corresponding dest has last value recorded, invalidate that last
23158 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23160 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
23161 of explicit '@'. Add missing assembly comment marker on branch costs
23164 2017-02-13 Nathan Sidwell <nathan@acm.org>
23166 * gengtype-lex.l (<in_struct>): Add '/'.
23168 2017-02-13 Martin Liska <mliska@suse.cz>
23171 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
23173 2017-02-13 Richard Biener <rguenther@suse.de>
23175 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
23177 * configure: Re-generate.
23178 * config.in: Likewise.
23179 * graphite-dependences.c: Simplify as if
23180 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
23181 * graphite-isl-ast-to-gimple.c: Likewise.
23182 * graphite-optimize-isl.c: Likewise.
23183 * graphite-poly.c: Likewise.
23184 * graphite-sese-to-poly.c: Likewise.
23185 * graphite.h: Likewise.
23186 * toplev.c: Include isl/version.h and use isl_version () for
23187 printing the ISL version.
23188 * doc/install.texi: Update ISL requirement.
23190 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23192 * doc/standards.texi (Standards): Update reference to
23195 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
23197 * doc/extend.texi (Named Address Spaces): sourceware.org now
23199 * doc/install.texi (Binaries): Ditto.
23202 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23204 * doc/cpp.texi: Replace "stringify"/"stringification" with C
23205 standard terminology "stringize"/"stringizing" throughout.
23206 * doc/cppinternals.texi: Likewise.
23208 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
23210 * doc/extend.texi: Fix some spelling mistakes and typos.
23211 * doc/invoke.texi: Likewise.
23213 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23216 * params.def (inline-min-speedup) Change from 10 to 8.
23218 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23220 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
23223 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
23226 * ipa-inline-analysis.c (get_minimal_bb): New function.
23227 (record_modified): Use it.
23228 (remap_edge_change_prob): Handle also ancestor functions.
23230 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
23232 * doc/contrib.texi (Contributors): Remove broken link into
23233 the Mauve CVS repository.
23235 2017-02-11 Jakub Jelinek <jakub@redhat.com>
23237 PR middle-end/79454
23238 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
23239 result computation whenever lhs doesn't have vector mode, not
23240 just when it has BLKmode.
23242 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23244 * doc/makefile.texi (profiledbootstrap): Refer to the
23245 installation instructions only in textual form.
23247 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23250 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
23252 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
23254 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
23255 (Specific): Update mingw-w64 reference.
23257 (Specific): Remove broken link to Renesas RX processor.
23259 2017-02-10 Richard Biener <rguenther@suse.de>
23261 * toplev.c (process_options): Do not mention obsolete graphite
23262 options when printing sorry message about missing graphite support.
23263 Mention -floop-nest-optimize.
23265 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
23267 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
23268 (vtst_p16): Likewise.
23269 (vtstq_p8): Likewise.
23270 (vtstq_p16): Likewise.
23272 (vtstq_p64): Likewise.
23273 * config/arm/arm_neon.h (vgetq_lane_p64): New.
23274 (vset_lane_p64): New.
23275 (vsetq_lane_p64): New.
23277 2017-02-10 Jakub Jelinek <jakub@redhat.com>
23279 PR tree-optimization/79411
23280 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
23281 stmt operands are SSA_NAMEs used in abnormal phis.
23282 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
23285 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23288 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
23291 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
23293 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
23295 2017-02-09 Jakub Jelinek <jakub@redhat.com>
23297 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
23301 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
23302 not arbitrary TREE_CONSTANT.
23305 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
23306 "omp declare target link" attribute unless is_global_var.
23307 * omp-offload.c (find_link_var_op): Likewise.
23309 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
23310 Chung-Lin Tang <cltang@codesourcery.com>
23312 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
23314 (gimplify_adjust_omp_clauses): Don't delete TILE.
23315 (gimplify_omp_for): Deal with TILE.
23316 * internal-fn.c (expand_GOACC_TILE): New function.
23317 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
23319 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
23320 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
23322 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
23323 avoid DIV for outermost collapse var.
23324 (expand_oacc_for): Insert tile element loop as needed. Adjust.
23325 Remove out of date comments, fix whitespace.
23326 * omp-general.c (omp_extract_for_data): Deal with tiling.
23327 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
23328 adjust OLF_DIM_BASE value.
23329 (struct omp_for_data): Add tiling field.
23330 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
23331 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
23332 for auto loops. Remove default auto determining, moved to
23333 oacc_loop_fixed_partitions.
23334 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
23335 stmts, add e_mask field.
23336 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
23337 (oacc_thread_numbers): Use oacc_dim_call.
23338 (oacc_xform_tile): New.
23339 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
23340 (finish_oacc_loop): Adjust for ifns vector.
23341 (oacc_loop_discover_walk): Append loop abstraction sites to list,
23342 add case for GOACC_TILE fns.
23343 (oacc_loop_xform_loop): Delete.
23344 (oacc_loop_process): Iterate over call list directly, and add
23345 handling for GOACC_TILE fns.
23346 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
23348 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
23349 vector partitioning to outer loops. Assign 2 partitions to loops
23350 when available. Add TILE handling.
23351 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
23352 (execite_oacc_device_lower): Process GOACC_TILE fns,
23353 ignore unknown specs.
23354 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
23355 * tree.c (omp_clause_num_ops): Adjust TILE ops.
23356 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
23358 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
23360 * configure.ac (ACX_BUGURL): Update.
23361 * configure: Regenerate.
23363 2017-02-09 Richard Biener <rguenther@suse.de>
23365 PR tree-optimization/69823
23366 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23367 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
23369 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23371 * config/arc/arc-c.def: Add __NPS400__ definition.
23372 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
23373 (TARGET_NPS400): Define.
23375 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
23377 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
23379 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
23380 pointer, arch_info.
23381 (arc_cpu_types): Fill the arch_info field with a pointer into the
23382 arc_arch_types table.
23383 (arc_selected_cpu): Declare.
23384 * config/arc/arc.c (arc_selected_cpu): Make global.
23385 (arc_selected_arch): Delete.
23386 (arc_base_cpu): Delete.
23387 (arc_override_options): Remove references to deleted variables,
23388 update access to arch information.
23389 (ARC_OPT): Update access to arch information.
23390 (ARC_OPTX): Likewise.
23391 * config/arc/arc.h (arc_base_cpu): Remove declaration.
23392 (TARGET_ARC600): Update access to arch information.
23393 (TARGET_ARC601): Likewise.
23394 (TARGET_ARC700): Likewise.
23395 (TARGET_EM): Likewise.
23396 (TARGET_HS): Likewise.
23397 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
23400 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
23403 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
23404 condition/operands for integer GE/LE/GEU/LEU operations.
23406 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
23408 PR translation/79397
23409 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
23412 2017-02-08 Martin Jambor <mjambor@suse.cz>
23415 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
23416 whether allocation happened.
23417 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
23418 nothing was allocated.
23420 2017-02-08 Jakub Jelinek <jakub@redhat.com>
23422 PR tree-optimization/79408
23423 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
23424 constant, but SSA_NAME with a known integer range, use the minimum
23425 of that range instead of op1 to determine if modulo can be replaced
23426 with its first operand.
23428 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23430 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
23432 2017-02-08 Richard Biener <rguenther@suse.de>
23434 PR tree-optimization/71824
23435 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23436 Check all loops contained in the merged region.
23438 2017-02-07 Andrew Pinski <apinski@cavium.com>
23440 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
23442 2017-02-07 Andrew Pinski <apinski@cavium.com>
23444 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
23445 (thunderxt88): Likewise.
23446 (thunderxt81): Disable LSE and change v8.1 to v8.
23447 (thunderxt83): Likewise.
23449 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23450 Richard Biener <rguenther@suse.de>
23452 PR middle-end/79399
23453 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
23454 type from int to size_t.
23455 * ira-costs.c (struct_costs_size): Change type from int to size_t.
23457 2017-02-07 Jakub Jelinek <jakub@redhat.com>
23459 PR rtl-optimization/79386
23460 * cprop.c (bypass_conditional_jumps): Initialize
23461 bypass_last_basic_block already before splitting bbs after
23462 unconditional traps...
23463 (bypass_conditional_jumps): ... rather than here.
23466 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
23467 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
23468 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
23469 fix -masm=intel patterns.
23471 2017-02-07 Richard Biener <rguenther@suse.de>
23473 PR tree-optimization/79256
23474 PR middle-end/79278
23475 * builtins.c (get_object_alignment_2): Use min_align_of_type
23476 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
23477 and ADJUST_FIELD_ALIGN.
23479 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
23481 * doc/tm.texi: Regenerate.
23482 * stor-layout.c (layout_decl): Adjust.
23483 (update_alignment_for_field): Likewise.
23484 (place_field): Likewise.
23485 (min_align_of_type): Likewise.
23486 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
23487 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
23488 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
23489 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
23490 * config/frv/frv.c (frv_adjust_field_align): Likewise.
23491 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
23492 * config/i386/i386.c (x86_field_alignment): Likewise.
23493 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
23494 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
23495 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
23496 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
23497 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
23498 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
23502 2017-01-30 Richard Biener <rguenther@suse.de>
23504 PR tree-optimization/79256
23505 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23506 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23509 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
23511 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
23512 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
23513 builtins to SImode and emit a zero-extend, if necessary.
23515 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23517 * docs/invoke.texi (RISC-V Options): Alphabetize.
23519 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23521 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
23524 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
23526 * config/riscv/riscv.c: New file.
23527 * common/config/riscv/riscv-common.c: Likewise.
23528 * config.gcc: Likewise.
23529 * config/riscv/constraints.md: Likewise.
23530 * config/riscv/elf.h: Likewise.
23531 * config/riscv/generic.md: Likewise.
23532 * config/riscv/linux.h: Likewise.
23533 * config/riscv/multilib-generator: Likewise.
23534 * config/riscv/peephole.md: Likewise.
23535 * config/riscv/pic.md: Likewise.
23536 * config/riscv/predicates.md: Likewise.
23537 * config/riscv/riscv-builtins.c: Likewise.
23538 * config/riscv/riscv-c.c: Likewise.
23539 * config/riscv/riscv-ftypes.def: Likewise.
23540 * config/riscv/riscv-modes.def: Likewise.
23541 * config/riscv/riscv-opts.h: Likewise.
23542 * config/riscv/riscv-protos.h: Likewise.
23543 * config/riscv/riscv.h: Likewise.
23544 * config/riscv/riscv.md: Likewise.
23545 * config/riscv/riscv.opt: Likewise.
23546 * config/riscv/sync.md: Likewise.
23547 * config/riscv/t-elf-multilib: Likewise.
23548 * config/riscv/t-linux: Likewise.
23549 * config/riscv/t-linux-multilib: Likewise.
23550 * config/riscv/t-riscv: Likewise.
23551 * configure.ac: Likewise.
23552 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
23553 Waterman as RISC-V maintainers.
23554 * doc/install.texi: Add RISC-V entries.
23555 * doc/invoke.texi: Add RISC-V options section.
23556 * doc/md.texi: Add RISC-V constraints section.
23557 * configure: Regenerated.
23559 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23562 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
23563 false values to be constant vectors with all 0 or all 1 bits set.
23564 (vcondu<mode><mode>): Likewise.
23565 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
23567 (fpmask_comparison_operator): Update comment.
23568 (vecint_comparison_operator): New predicate.
23569 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
23570 vector conditionals when the true and false values are constant
23571 vectors with all 0 bits or all 1 bits set.
23573 2017-02-06 Martin Sebor <msebor@redhat.com>
23575 PR tree-optimization/79376
23576 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
23578 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
23580 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
23581 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
23582 to simplify split condition.
23584 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23586 * omp-expand.c (oxpand_omp_atomic_fetch_op,
23587 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
23590 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
23592 PR rtl-optimization/68664
23593 * target.def (can_speculate_insn): New hook.
23594 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
23595 * doc/tm.texi: Regenerate.
23596 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
23597 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
23598 (rs6000_sched_can_speculate_insn): New function.
23600 2017-02-06 Jakub Jelinek <jakub@redhat.com>
23602 PR tree-optimization/79284
23603 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
23604 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
23605 vectorizable_mask_load_store, vectorizable_operation,
23606 vect_is_simple_cond, get_same_sized_vectype): Use it instead
23607 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
23608 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
23609 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
23610 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
23611 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23612 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
23613 is_gimple_assign (stmt). Replace another such test with
23614 is_gimple_assign (stmt).
23616 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23619 * config/avr/avr.c (rtl-iter.h): Include it.
23620 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
23621 (avr_legitimate_combined_insn): ...and implementation.
23623 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
23625 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
23626 * config/s390/s390.c (s390_const_operand_ok)
23627 (s390_canonicalize_comparison, s390_extract_part)
23628 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
23629 (s390_contiguous_bitmask_p, s390_rtx_costs)
23630 (legitimize_pic_address): Likewise.
23631 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
23632 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
23633 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
23634 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
23635 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
23637 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
23639 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
23640 REGNO($0) == REGNO($1).
23642 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23644 * config/s390/linux.h(SIZE_TYPE): Add comment.
23646 2017-02-06 Julian Brown <julian@codesourcery.com>
23647 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23648 Virendra Pathak <virendra.pathak@broadcom.com>
23650 * config/aarch64/aarch64-cores.def: Change the scheduler
23652 * config/aarch64/aarch64.md: Include thunderx2t99.md.
23653 * config/aarch64/thunderx2t99.md: New file.
23655 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23657 * doc/standards.texi (Go Language): Update link to language
23660 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23662 * tree-eh.c (lower_resx): Sanitize profile.
23663 (cleanup_empty_eh_move_lp): Likewise.
23665 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
23668 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
23670 * cfgloopmanip.h (loop_version): Update prototype.
23671 * modulo-sched.c (sms_schedule): Update call of loop_version.
23672 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
23673 * tree-parloops.c (gen_parallel_loop): Likewise.
23674 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
23675 * tree-ssa-loop-split.c (split_loop): Likewise.
23676 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
23677 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23679 2017-02-05 Martin Liska <mliska@suse.cz>
23682 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
23684 (print_operand_address): Initialize a struct to zero.
23686 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23688 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
23689 garbage collector only in textual form.
23691 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
23693 * doc/extend.texi (x86 specific memory model extensions for
23694 transactional memory): Simplify a phrase.
23696 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
23699 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
23700 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
23701 (atomic_storedi_1): Likewise.
23703 2017-02-04 Jakub Jelinek <jakub@redhat.com>
23705 PR tree-optimization/79338
23706 * tree-parloops.c (gather_scalar_reductions): Don't call
23707 vect_analyze_loop_form for loop->inner before destroying loop's
23710 2017-02-03 Martin Sebor <msebor@redhat.com>
23712 PR tree-optimization/79327
23713 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
23714 when precision has resulted in leading zeros.
23715 (format_integer): Adjust the likely counter to assume an unknown
23716 argument that may be zero is non-zero.
23718 2017-02-03 Jason Merrill <jason@redhat.com>
23721 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
23722 avoid copying non-taken branch.
23724 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23726 PR tree-optimization/79340
23727 * tree-vect-loop.c (vectorizable_reduction): Release
23728 vec_defs elements after safe_splicing them into other vectors.
23731 PR tree-optimization/79327
23732 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
23733 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
23735 (format_integer): Use wide_int_to_tree instead of build_int_cst
23736 + to_?hwi. If argmin is NULL, just set argmin and argmax to
23737 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
23738 of shortest and longest sequence.
23740 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
23742 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
23743 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
23745 2017-02-03 Walter Lee <walt@tilera.com>
23748 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
23749 after initial stackframe link reg save.
23750 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
23752 2017-02-03 Jakub Jelinek <jakub@redhat.com>
23755 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
23756 wu for stxssp alternative.
23758 2017-02-03 Martin Sebor <msebor@redhat.com>
23760 PR tree-optimization/79352
23761 * gimple-fold.c (get_range_strlen): Add argument.
23762 (get_range_strlen): Change return type to bool.
23763 (get_maxval_strlen): Pass in a dummy argument.
23764 * gimple-fold.h (get_range_strlen): Change return type to bool.
23765 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
23766 * tree.h (array_at_struct_end_p): Add argument.
23767 * tree.c (array_at_struct_end_p): Handle it.
23769 2017-02-03 Martin Liska <mliska@suse.cz>
23772 * multiple_target.c (create_dispatcher_calls): Redirect edge
23773 from a caller of a dispatcher.
23774 (expand_target_clones): Make the clones local.
23775 (ipa_target_clone): Do both target clones and resolvers.
23776 (ipa_dispatcher_calls): Remove the pass.
23777 (pass_dispatcher_calls::gate): Likewise.
23778 (make_pass_dispatcher_calls): Likewise.
23779 * passes.def (pass_target_clone): Put as very first IPA early
23782 2017-02-03 Martin Liska <mliska@suse.cz>
23784 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
23785 in case of a function with ifunc attribute.
23787 2017-02-03 Martin Liska <mliska@suse.cz>
23789 * cgraph.c (cgraph_node::dump): Dump function version info.
23790 * symtab.c (symtab_node::dump_base): Add missing new line.
23792 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23794 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
23795 (ifcombine_ifandif): Use it.
23797 2017-02-03 Martin Liska <mliska@suse.cz>
23799 * doc/invoke.texi: Document default value for
23800 use-after-scope-direct-emission-threshold.
23802 2017-02-03 Martin Liska <mliska@suse.cz>
23804 PR tree-optimization/79339
23805 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
23806 (format_floating): Likewise.
23808 2017-02-03 Martin Liska <mliska@suse.cz>
23811 * ipa-prop.c (ipa_node_params_t::insert): Remove current
23813 (ipa_node_params_t::remove): Likewise.
23814 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
23815 initialization from removed ipa_node_params_t::insert.
23816 (ipa_node_params::~ipa_node_params): Move from removed
23817 ipa_node_params_t::release.
23818 * symbol-summary.h (symbol_summary::m_released): New member.
23819 Do not release a summary twice. Do not allow to call finalizer
23820 for types of a summary that live in GGC memory.
23822 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23824 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
23827 2017-02-02 Martin Sebor <msebor@redhat.com>
23829 PR middle-end/79275
23830 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
23831 (format_string): Tighten up the range of output for non-constant
23832 strings and correct the expected range for wide non-constant strings.
23834 2017-02-02 Martin Sebor <msebor@redhat.com>
23836 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
23838 PR middle-end/32003
23839 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
23841 (-fdump-tree-@var): Add to index and document how to come up
23842 with pass-specific option and dump file names.
23843 (-fdump-passes): Clarify where to look for output.
23845 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
23847 PR middle-end/77445
23848 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
23849 statistics of the analyzed path; allow threading for speed when
23850 any of BBs along the path are optimized for speed.
23852 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
23854 PR middle-end/78468
23855 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
23856 settings of the virtual registers.
23859 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
23861 * explow.c (get_dynamic_stack_size): Take known alignment of stack
23862 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
23865 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23867 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
23868 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
23870 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23872 * config/s390/s390.md: Add missing comments with the expanded
23874 * config/s390/vector.md: Likewise.
23875 * config/s390/vx-builtins.md: Likewise.
23877 2017-02-02 Jakub Jelinek <jakub@redhat.com>
23880 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
23881 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
23882 conditions on a single line.
23884 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23886 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
23887 __S390_VX__ to __VX__.
23889 2017-02-01 Andrew Pinski <apinski@cavium.com>
23891 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
23892 stmt_info to record_stmt_cost.
23893 (vect_get_known_peeling_cost): Pass stmt_info if known to
23895 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
23896 cpu_vector_cost field into
23897 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
23898 field into vec_int_stmt_cost and vec_fp_stmt_cost.
23899 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
23900 splitting of scalar_stmt_cost and vec_stmt_cost.
23901 (thunderx_vector_cost): Likewise.
23902 (cortexa57_vector_cost): LIkewise.
23903 (exynosm1_vector_cost): Likewise.
23904 (xgene1_vector_cost): Likewise.
23905 (thunderx2t99_vector_cost): Improve after the splitting of the two
23907 (aarch64_builtin_vectorization_cost): Update for the splitting of
23908 scalar_stmt_cost and vec_stmt_cost.
23910 2017-02-01 Torvald Riegel <triegel@redhat.com>
23911 Richard Henderson <rth@redhat.com>
23913 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
23914 conditional on existance of a fast atomic load.
23915 * optabs-query.c (can_atomic_load_p): New function.
23916 * optabs-query.h (can_atomic_load_p): Declare it.
23917 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
23918 no fast atomic load is available for the particular size of access.
23919 (expand_atomic_compare_and_swap): Likewise.
23920 (expand_atomic_load): Likewise.
23921 (expand_atomic_store): Likewise.
23922 (expand_atomic_fetch_op): Likewise.
23923 * testsuite/lib/target-supports.exp
23924 (check_effective_target_sync_int_128): Remove x86 because it provides
23925 no fast atomic load.
23926 (check_effective_target_sync_int_128_runtime): Likewise.
23928 2017-02-01 Richard Biener <rguenther@suse.de>
23930 * graphite.c: Include tree-vectorizer.h for find_loop_location.
23931 (graphite_transform_loops): Provide opt-info for optimized nests.
23932 * tree-parloop.c (parallelize_loops): Provide opt-info for
23933 parallelized loops.
23935 2017-02-01 Richard Biener <rguenther@suse.de>
23937 PR middle-end/79315
23938 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
23939 was not set before.
23941 2017-02-01 Richard Biener <rguenther@suse.de>
23943 PR tree-optimization/71824
23944 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
23945 Verify the loops are valid in the merged SESE region.
23946 (scop_detection::can_represent_loop_1): Check analyzing the
23947 evolution of the number of iterations in the region succeeds.
23949 2017-01-31 Ian Lance Taylor <iant@golang.org>
23951 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
23952 REG_ARGS_SIZE note to 32-bit push insns and call insn.
23954 2017-01-31 David Malcolm <dmalcolm@redhat.com>
23956 PR preprocessor/79210
23957 * input.c (get_substring_ranges_for_loc): Replace line_width
23958 assertion with error-handling.
23960 2017-01-31 Richard Biener <rguenther@suse.de>
23962 PR tree-optimization/77318
23963 * graphite-sese-to-poly.c (extract_affine): Fix assert.
23964 (create_pw_aff_from_tree): Take loop parameter.
23965 (add_condition_to_pbb): Pass loop of the condition to
23966 create_pw_aff_from_tree.
23968 2017-01-31 Jakub Jelinek <jakub@redhat.com>
23970 * config/s390/s390.c (s390_asan_shadow_offset): New function.
23971 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
23973 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
23977 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
23979 (convert_int_to_float128): Likewise.
23980 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
23981 (convert_int_to_float128): Likewise.
23982 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
23983 (UNSPEC_IEEE128_CONVERT): Likewise.
23984 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
23985 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
23986 Use local variables for IBM extended format.
23987 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
23988 (fix_trunc<mode>si2_fprs): Likewise.
23989 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
23990 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
23991 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
23992 to know that we can now have integers of all sizes in vector
23994 (fix<uns>_<mode>di2_hw): Likewise.
23995 (float<uns>_<mode>si2_hw): Likewise.
23996 (fix_<mode>si2_hw): Likewise.
23997 (fixuns_<mode>si2_hw): Likewise.
23998 (float<uns>_<mode>di2_hw): Likewise.
23999 (float_<mode>di2_hw): Likewise.
24000 (float_<mode>si2_hw): Likewise.
24001 (floatuns_<mode>di2_hw): Likewise.
24002 (floatuns_<mode>si2_hw): Likewise.
24003 (xscvqp<su>wz_<mode>): Delete, no longer used.
24004 (xscvqp<su>dz_<mode>): Likewise.
24005 (xscv<su>dqp_<mode>): Likewise.
24006 (ieee128_mfvsrd_64bit): Likewise.
24007 (ieee128_mfvsrd_32bit): Likewise.
24008 (ieee128_mfvsrwz): Likewise.
24009 (ieee128_mtvsrw): Likewise.
24010 (ieee128_mtvsrd_64bit): Likewise.
24011 (ieee128_mtvsrd_32bit): Likewise.
24013 2017-01-31 Martin Liska <mliska@suse.cz>
24016 * ipa-prop.c (ipa_free_all_node_params): Call release method
24017 instead of ~sumbol_summary to not to trigger double times
24020 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
24022 PR tree-optimization/71691
24023 * bitmap.h (class auto_bitmap): New.
24024 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
24025 is_maybe_undefined instead of ssa_undefined_value_p.
24027 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24029 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
24030 __S390_ARCH_LEVEL__ to __ARCH__.
24032 2017-01-31 Jakub Jelinek <jakub@redhat.com>
24034 PR tree-optimization/79267
24035 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
24036 if should_remove_lhs_p is true.
24038 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
24041 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
24042 (add_alignment_attribute): New.
24043 (base_type_die): Add alignment attribute.
24044 (subrange_type_die): Likewise.
24045 (modified_type_die): Likewise.
24046 (gen_array_type_die): Likewise.
24047 (gen_descr_array_type_die: Likewise.
24048 (gen_enumeration_type_die): Likewise.
24049 (gen_subprogram_die): Likewise.
24050 (gen_variable_die): Likewise.
24051 (gen_field_die): Likewise.
24052 (gen_ptr_to_mbr_type_die): Likewise.
24053 (gen_struct_or_union_type_die): Likewise.
24054 (gen_subroutine_type_die): Likewise.
24055 (gen_typedef_die): Likewise.
24056 (base_type_cmp): Compare alignment attribute.
24058 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24061 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
24062 (setb_unsigned) New pattern for setb with CCUNS.
24063 * config/rs6000/rs6000.c (expand_block_compare): Use a different
24064 subfc./subfe sequence to avoid overflow problems. Generate a
24065 shorter sequence with cmpld/setb for power9.
24066 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
24067 for generating subfc. instruction.
24068 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
24069 now uses this instruction.
24071 2017-01-30 Ian Lance Taylor <iant@google.com>
24074 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
24075 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
24077 2017-01-30 Martin Sebor <msebor@redhat.com>
24079 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
24080 Move constant to the right of a relational operator.
24081 (get_mpfr_format_length, format_character, format_string): Ditto.
24082 (should_warn_p, maybe_warn): Same.
24084 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
24086 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
24089 * asan.c (get_translation_unit_decl): Remove function.
24090 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
24092 2017-01-30 Martin Liska <mliska@suse.cz>
24094 PR gcov-profile/79259
24095 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
24096 -fprofile-generate.
24098 2017-01-30 Martin Liska <mliska@suse.cz>
24101 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
24102 Initialize variables with NULL value.
24104 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
24107 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
24109 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
24111 2017-01-30 Richard Biener <rguenther@suse.de>
24113 PR tree-optimization/79276
24114 * tree-vrp.c (process_assert_insertions): Properly adjust common
24115 when removing a duplicate.
24117 * gcc.dg/torture/pr79276.c: New testcase.
24119 2017-01-30 Richard Biener <rguenther@suse.de>
24121 PR tree-optimization/79256
24122 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
24123 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
24125 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
24127 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
24130 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
24131 ("*r<noxa>sbg_<mode>_sll_bitmask")
24132 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
24133 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
24134 Use contiguous_bitmask_nowrap_operand.
24136 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24139 * config/rs6000/altivec.h (vec_xl): Revise #define.
24140 (vec_xst): Likewise.
24142 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
24144 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
24146 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
24148 PR rtl-optimization/79194
24149 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
24150 traps before call to bypass_conditional_jumps.
24152 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24154 PR tree-optimization/71374
24155 * lra-constraints.c (check_conflict_input_operands): New.
24156 (match_reload): Use it.
24158 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
24161 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
24162 account to calculate conflict_set.
24164 2017-01-27 Bin Cheng <bin.cheng@arm.com>
24166 PR rtl-optimization/78559
24167 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
24168 other_insn in combine.
24170 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24172 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
24173 uint16_type_node for BT_UINT16.
24175 2017-01-27 David Malcolm <dmalcolm@redhat.com>
24177 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
24178 "RTL Tests" to menu.
24179 (GIMPLE Tests): New node.
24180 (RTL Tests): New node.
24182 2017-01-27 Richard Biener <rguenther@suse.de>
24184 PR tree-optimization/79245
24185 * tree-loop-distribution.c (distribute_loop): Apply cost
24186 modeling also to detected patterns.
24188 2017-01-27 Richard Biener <rguenther@suse.de>
24190 PR tree-optimization/71433
24191 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
24192 (compare_assert_loc): New function.
24193 (process_assert_insertions): Sort and optimize assert locations
24194 to remove duplicates and push down identical assertions on
24195 edges to their destination block.
24197 2017-01-27 Richard Biener <rguenther@suse.de>
24199 PR tree-optimization/79244
24200 * tree-vrp.c (remove_range_assertions): Forcefully propagate
24201 out SSA names even if abnormal.
24203 2017-01-27 Jakub Jelinek <jakub@redhat.com>
24205 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
24206 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
24207 instead of MPFR_RNDN.
24209 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
24212 * arm.c (arm_option_override): Don't call build_target_option_node
24213 until after doing all option overrides.
24214 (arm_valid_target_attribute_tree): Likewise.
24216 2017-01-27 Martin Liska <mliska@suse.cz>
24218 * doc/invoke.texi (-fprofile-arcs): Document profiling support
24219 for {cd}tors and C++ {cd}tors.
24221 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
24223 * config/s390/s390.md ("*setmem_long_and")
24224 ("*setmem_long_and_31z"): Use zero_extend instead of and.
24226 2017-01-26 Martin Sebor <msebor@redhat.com>
24228 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
24231 2017-01-26 Martin Sebor <msebor@redhat.com>
24233 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
24234 HAVE_DFmode before using XFmode or DFmode.
24235 (parse_directive): Avoid using the z length modifier to avoid
24236 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
24238 PR middle-end/78703
24239 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
24240 to accept adjustment as an array.
24241 (get_int_range): New function.
24242 (struct directive): Make width and prec arrays.
24243 (directive::set_width, directive::set_precision): Call get_int_range.
24244 (format_integer, format_floating): Handle width and precision ranges.
24245 (format_string, parse_directive): Same.
24247 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24250 * dwarf2out.c (generate_skeleton_bottom_up): For children with
24251 comdat_type_p set, just clone them, but keep the children in the
24255 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
24256 which have direct callers with -fvar-tracking-assignments enabled
24258 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
24259 inside of type units.
24261 2017-01-26 Martin Sebor <msebor@redhat.com>
24263 PR middle-end/78703
24264 * gimple-ssa-sprintf.c (struct result_range): Add likely and
24266 (struct format_result): Replace number_chars, number_chars_min,
24267 and number_chars_max with a single member of struct result_range.
24269 (format_result::operator+=): Adjust.
24270 (struct fmtresult): Remove bounded. Handle likely and unlikely
24272 (fmtresult::adjust_for_width_or_precision): New function.
24273 (fmtresult:type_max_digits): New function.
24274 (bytes_remaining): Handle likely and unlikely counters.
24275 (min_bytes_remaining): Remove.
24276 (format_percent): Simplify.
24277 (format_integer, format_floating): Set likely and unlikely counters.
24278 (get_string_length, format_character, format_string): Same.
24279 (format_plain, should_warn_p): New function.
24280 (maybe_warn): Call should_warn_p. Update diagnostic messages
24281 and handle those for all directives, including plain strings.
24282 (format_directive): Handle likely and unlikely counters.
24283 Remove unnecessary quoting from diagnostics. Add an informational
24285 (add_bytes): Remove.
24286 (pass_sprintf_length::compute_format_length): Simplify.
24287 (try_substitute_return_value): Handle likely and unlikely counters.
24289 2017-01-26 Carl Love <cel@us.ibm.com>
24291 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
24292 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
24294 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
24297 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
24298 endianess for subregs into account.
24299 * lra-constraints.c (lra_constraints): Do risky transformations
24300 always on the first iteration.
24301 * lra-lives.c (check_pseudos_live_through_calls): Add arg
24302 last_call_used_reg_set.
24303 (process_bb_lives): Define and use last_call_used_reg_set.
24304 * lra.c (lra): Always continue after lra_constraints on the first
24307 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
24309 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
24311 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
24313 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24315 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
24316 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
24317 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
24318 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
24319 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
24320 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
24321 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
24322 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
24323 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
24325 2017-01-26 Marek Polacek <polacek@redhat.com>
24328 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
24329 for the third operand.
24331 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24333 PR middle-end/79236
24334 * omp-low.c (struct omp_context): Add simt_stmt field.
24335 (scan_omp_for): Return omp_context *.
24336 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
24337 context to the _simt_ SIMD stmt.
24338 (lower_omp_for): For combined SIMD with sibling _simt_
24339 SIMD, make sure to use the same decls in _looptemp_
24340 clauses as in the sibling.
24342 2017-01-26 David Sherwood <david.sherwood@arm.com>
24344 PR middle-end/79212
24345 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
24348 2017-01-26 Jakub Jelinek <jakub@redhat.com>
24351 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
24352 emit fld b; fld a; if possible.
24354 * brig-builtins.def: Update copyright years.
24355 * config/arm/arm_acle_builtins.def: Update copyright years.
24357 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
24360 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
24361 constraint instead of o for the stxsd instruction.
24363 2017-01-25 Carl Love <cel@us.ibm.com>
24365 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
24366 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
24368 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
24370 * doc/invoke.texi (C++ Dialect Options): Fix typo.
24372 2017-01-25 Richard Biener <rguenther@suse.de>
24374 PR tree-optimization/69264
24375 * target.def (vector_alignment_reachable): Improve documentation.
24376 * doc/tm.texi: Regenerate.
24377 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
24379 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
24380 earlier changes with respect to TYPE_USER_ALIGN.
24381 (vector_alignment_reachable_p): Likewise. Improve dumping.
24383 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24386 * config/arm/arm.md (xordi3): Force constant operand into a register
24389 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24391 * doc/invoke.texi (-fstore-merging): Correct default optimization
24392 levels at which it is enabled.
24393 (-O): Move -fstore-merging from list to...
24396 2017-01-25 Richard Biener <rguenther@suse.de>
24399 * omp-expand.c: Include debug.h.
24400 (expand_omp_taskreg): Make sure to generate early debug before
24401 outlining anything from a function.
24402 (expand_omp_target): Likewise.
24403 (grid_expand_target_grid_body): Likewise.
24405 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
24408 * asan.c (get_translation_unit_decl): New function.
24409 (asan_add_global): Extract modules file name from globals
24410 TRANSLATION_UNIT_DECL name.
24412 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
24415 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
24416 for long calls with APCS frame and VFP.
24418 2017-01-24 David Malcolm <dmalcolm@redhat.com>
24420 * cfg.c (original_copy_tables_initialized_p): New function.
24421 * cfg.h (original_copy_tables_initialized_p): New decl.
24422 * cfgrtl.c (relink_block_chain): Guard the call to
24423 free_original_copy_tables with a call to
24424 original_copy_tables_initialized_p.
24425 * cgraph.h (symtab_node::native_rtl_p): New decl.
24426 * cgraphunit.c (symtab_node::native_rtl_p): New function.
24427 (symtab_node::needed_p): Don't assert for early assembly output
24428 for __RTL functions.
24429 (cgraph_node::finalize_function): Set "force_output" for __RTL
24431 (cgraph_node::analyze): Bail out early for __RTL functions.
24432 (analyze_functions): Update assertion to support __RTL functions.
24433 (cgraph_node::expand): Bail out early for __RTL functions.
24434 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
24436 * function.h (struct function): Update comment for field
24438 * gimple-expr.c: Include "tree-pass.h".
24439 (gimple_has_body_p): Return false for __RTL functions.
24440 * Makefile.in (OBJS): Add run-rtl-passes.o.
24441 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
24443 (gcc::pass_manager::get_clean_slate): New accessor.
24444 * passes.c: Include "insn-addr.h".
24445 (should_skip_pass_p): Add logging. Update logic for running
24446 "expand" to be compatible with both __GIMPLE and __RTL. Guard
24447 property-provider override so it is only done for gimple passes.
24449 (skip_pass): New function.
24450 (execute_one_pass): Call skip_pass when skipping passes.
24451 * read-md.c (md_reader::read_char): Support filtering
24452 the input to a subset of line numbers.
24453 (md_reader::md_reader): Initialize fields
24454 m_first_line and m_last_line.
24455 (md_reader::read_file_fragment): New function.
24456 * read-md.h (md_reader::read_file_fragment): New decl.
24457 (md_reader::m_first_line): New field.
24458 (md_reader::m_last_line): New field.
24459 * read-rtl-function.c (function_reader::create_function): Only
24460 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
24461 curr_properties. Set DECL_INITIAL to a dummy block.
24462 (read_rtl_function_body_from_file_range): New function.
24463 * read-rtl-function.h (read_rtl_function_body_from_file_range):
24465 * run-rtl-passes.c: New file.
24466 * run-rtl-passes.h: New file.
24468 2017-01-24 Jeff Law <law@redhat.com>
24470 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
24473 2017-01-24 Bin Cheng <bin.cheng@arm.com>
24475 PR tree-optimization/79159
24476 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
24477 (record_nonwrapping_iv): Improve boundary using above function if no
24478 value range information.
24480 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
24481 Martin Jambor <mjambor@suse.cz>
24483 * brig-builtins.def: New file.
24484 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
24485 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
24486 (DEF_HSAIL_SAT_BUILTIN): Likewise.
24487 (DEF_HSAIL_INTR_BUILTIN): Likewise.
24488 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
24489 * builtin-types.def (BT_INT8): New.
24490 (BT_INT16): Likewise.
24491 (BT_UINT8): Likewise.
24492 (BT_UINT16): Likewise.
24493 (BT_FN_ULONG): Likewise.
24494 (BT_FN_UINT_INT): Likewise.
24495 (BT_FN_UINT_ULONG): Likewise.
24496 (BT_FN_UINT_LONG): Likewise.
24497 (BT_FN_UINT_PTR): Likewise.
24498 (BT_FN_ULONG_PTR): Likewise.
24499 (BT_FN_INT8_FLOAT): Likewise.
24500 (BT_FN_INT16_FLOAT): Likewise.
24501 (BT_FN_UINT32_FLOAT): Likewise.
24502 (BT_FN_UINT16_FLOAT): Likewise.
24503 (BT_FN_UINT8_FLOAT): Likewise.
24504 (BT_FN_UINT64_FLOAT): Likewise.
24505 (BT_FN_UINT16_UINT32): Likewise.
24506 (BT_FN_UINT32_UINT16): Likewise.
24507 (BT_FN_UINT16_UINT16_UINT16): Likewise.
24508 (BT_FN_INT_PTR_INT): Likewise.
24509 (BT_FN_UINT_PTR_UINT): Likewise.
24510 (BT_FN_LONG_PTR_LONG): Likewise.
24511 (BT_FN_ULONG_PTR_ULONG): Likewise.
24512 (BT_FN_VOID_UINT64_UINT64): Likewise.
24513 (BT_FN_UINT8_UINT8_UINT8): Likewise.
24514 (BT_FN_INT8_INT8_INT8): Likewise.
24515 (BT_FN_INT16_INT16_INT16): Likewise.
24516 (BT_FN_INT_INT_INT): Likewise.
24517 (BT_FN_UINT_FLOAT_UINT): Likewise.
24518 (BT_FN_FLOAT_UINT_UINT): Likewise.
24519 (BT_FN_ULONG_UINT_UINT): Likewise.
24520 (BT_FN_ULONG_UINT_PTR): Likewise.
24521 (BT_FN_ULONG_ULONG_ULONG): Likewise.
24522 (BT_FN_UINT_UINT_UINT): Likewise.
24523 (BT_FN_VOID_UINT_PTR): Likewise.
24524 (BT_FN_UINT_UINT_PTR: Likewise.
24525 (BT_FN_UINT32_UINT64_PTR): Likewise.
24526 (BT_FN_INT_INT_UINT_UINT): Likewise.
24527 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
24528 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
24529 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
24530 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
24531 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
24532 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
24533 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
24534 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
24535 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
24536 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
24537 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
24538 * doc/frontends.texi: List BRIG FE.
24539 * doc/install.texi (Testing): Add BRIG tesring requirements.
24540 * doc/invoke.texi (Overall Options): Mention BRIG.
24541 * doc/standards.texi (Standards): Doucment BRIG HSA version.
24543 2017-01-24 Richard Biener <rguenther@suse.de>
24545 PR translation/79208
24546 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
24548 2017-01-24 Martin Jambor <mjambor@suse.cz>
24551 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
24552 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
24553 and known_contexts.
24555 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
24557 PR middle-end/79123
24558 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
24559 casts from signed to unsigned really don't have a range.
24561 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
24563 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
24564 GMP_RNDx for compatiblity.
24566 2017-01-24 Martin Liska <mliska@suse.cz>
24569 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
24570 that would prevent us to call alloca with -1 as argument.
24572 2017-01-24 Jakub Jelinek <jakub@redhat.com>
24574 * dwarf2out.c (output_compilation_unit_header, output_file_names):
24575 Avoid -Wformat-security warning.
24577 2017-01-23 Andrew Pinski <apinski@cavium.com>
24579 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
24582 2017-01-23 Martin Sebor <msebor@redhat.com>
24584 PR middle-end/78703
24585 * gimple-ssa-sprintf.c (warn_level): New global.
24586 (format_integer): Use it here and throughout the rest of the file.
24587 Use the same switch to compute sign as base.
24588 (maybe_warn): New function.
24589 (format_directive): Factor out warnings into maybe_warn.
24590 Add debugging output. Use warn_level.
24591 (add_bytes): Use warn_level.
24592 (pass_sprintf_length::compute_format_length): Add debugging output.
24593 (try_substitute_return_value): Same.
24594 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
24596 PR middle-end/78703
24597 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
24598 (struct fmtresult, format_integer, format_floating): Adjust.
24599 (fmtresult::fmtresult): Set max correctly in two argument ctor.
24600 (get_string_length, format_string,format_directive): Same.
24601 (pass_sprintf_length::compute_format_length): Same.
24602 (try_substitute_return_value): Simplify slightly.
24604 PR middle-end/78703
24605 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
24606 (fmtresult::operator+=): Outlined.
24607 (struct fmtresult): Add ctors.
24608 (struct conversion_spec): Rename...
24609 (struct directive): ...to this. Add and remove data members.
24610 (directive::set_width, directive::set_precision): New functions.
24611 (format_percent): Use fmtresult ctor.
24612 (get_width_and_precision): Remove.
24613 (format_integer): Make naming changes. Avoid computing width and
24615 (format_floating): Same. Adjust indentation.
24616 (format_character, format_none): New functions.
24617 (format_string): Moved character handling to format_character.
24618 (format_directive): Remove arguments, change return type.
24619 (parse_directive): New function.
24620 (pass_sprintf_length::compute_format_length): Move directive
24621 parsing to parse_directive.
24623 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24625 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
24626 (assign_assembler_name_if_needed): ... this.
24627 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
24628 (assign_assembler_name_if_needed): ... this.
24629 (free_lang_data_in_cgraph): Adjust callers.
24630 * cgraphunit.c (cgraph_node::analyze): Likewise.
24631 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
24634 2017-01-23 Richard Biener <rguenther@suse.de>
24636 PR tree-optimization/79088
24637 PR tree-optimization/79188
24638 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
24639 resetting loop bounds after last path deletion. Reset loop
24640 bounds of the target loop, make code match the comments.
24641 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
24642 Make sure loops need no fixups.
24644 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24646 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
24647 exponent support with double type for first argument.
24648 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
24649 type returned by __builtin_vec_extract_sig,
24650 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
24651 functions from "vector int" to "vector unsigned int" or from
24652 "vector long long int" to "vector unsigned long long int".
24653 Changed type returned by __builtin_vec_extract_exp,
24654 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
24655 functions from "vector int" to "vector unsigned int" or from
24656 "vector long long int" to "vector unsigned long long int".
24657 Changed return type of __builtin_vec_test_data_class,
24658 __builtin_vec_test_data_class_sp, and
24659 __builtin_vec_test_data_class_dp from "vector int" to
24660 "vector bool int" or from "vector long long int" to "vector bool
24661 long long int" and changed second argument type from "unsigned
24662 int" to "int". Added new overloaded function forms "vector float
24663 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
24664 "vector float __builtin_vec_insert_exp_sp (vector float, vector
24665 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
24666 double, vector unsigned long long int)" and "vector double
24667 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
24668 long int)". Changed return type of
24669 __builtin_scalar_test_data_class and
24670 __builtin_scalar_test_data_class_sp and
24671 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
24672 int" and changed second argument from "unsigned int" to "int".
24673 Changed type returned by __builtin_scalar_test_neg,
24674 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
24675 from "int" to "bool int". Added new overloaded function form
24676 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
24677 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
24678 exponent double-precision with floating point first argument.
24679 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
24680 documentation of scalar_test_data_class, scalar_test_neg,
24681 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
24682 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
24683 vec_test_data_class built-in functions to reflect refinements in
24684 their type signatures.
24686 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
24688 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
24690 (aarch64_elf_asm_destructor): Likewise.
24692 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
24694 PR rtl-optimization/78634
24695 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
24696 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
24697 * ifcvt.c (noce_try_cmove): Add missing cost check.
24699 PR rtl-optimization/71724
24700 * combine.c (if_then_else_cond): Look for situations where it is
24701 beneficial to undo the work of one of the recursive calls.
24703 2017-01-23 Bin Cheng <bin.cheng@arm.com>
24705 PR tree-optimization/70754
24706 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
24707 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
24708 combined stmt before it if not NULL.
24709 (combine_chains): Process refs reversely and compute dominance point
24712 2017-01-23 Martin Liska <mliska@suse.cz>
24714 PR tree-optimization/79196
24715 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
24716 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
24718 (strlen_optimize_stmt): Call the renamed function.
24720 2017-01-23 Michael Matz <matz@suse.de>
24722 PR tree-optimization/78384
24723 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
24725 2017-01-23 Richard Biener <rguenther@suse.de>
24727 PR tree-optimization/79186
24728 * tree-vrp.c (register_new_assert_for): Make sure we've seen
24729 both incoming edges before moving an assert.
24731 2017-01-23 Martin Jambor <mjambor@suse.cz>
24733 * ipa-prop.c (load_from_param_1): Removed.
24734 (load_from_unmodified_param): Bits from load_from_param_1 put back
24736 (load_from_param): Removed.
24737 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
24738 with stmt. Reverted back to use of load_from_unmodified_param.
24740 2017-01-23 Martin Jambor <mjambor@suse.cz>
24743 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
24744 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
24745 field a pointer to garbage collected vector, mark lattices and
24746 ipcp_orig_node with GTY((skip)).
24747 (ipa_get_param_count): Adjust to descriptors being a pointer.
24748 (ipa_get_param): Likewise.
24749 (ipa_get_type): Likewise.
24750 (ipa_get_param_move_cost): Likewise.
24751 (ipa_set_param_used): Likewise.
24752 (ipa_get_controlled_uses): Likewise.
24753 (ipa_set_controlled_uses): Likewise.
24754 (ipa_is_param_used): Likewise.
24755 (ipa_node_params_t): Move into garbage collector. New methods insert
24757 (ipa_node_params_sum): Annotate wth GTY(()).
24758 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
24760 (ipa_load_from_parm_agg): Adjust declaration.
24761 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
24762 * ipa-profile.c (ipa_profile): Likewise.
24763 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
24764 (ipa_populate_param_decls): Make descriptors parameter garbage
24766 (ipa_dump_param): Adjust to descriptors being a pointer.
24767 (ipa_alloc_node_params): Likewise.
24768 (ipa_initialize_node_params): Likewise.
24769 (load_from_param_1): Make descriptors parameter garbage collected.
24770 (load_from_unmodified_param): Likewise.
24771 (load_from_param): Likewise.
24772 (ipa_load_from_parm_agg): Likewise.
24773 (ipa_node_params::~ipa_node_params): Removed.
24774 (ipa_free_all_node_params): Remove call to delete operator.
24775 (ipa_node_params_t::insert): New.
24776 (ipa_node_params_t::remove): Likewise.
24777 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
24778 copy known_csts and known_contexts vectors.
24779 (ipa_read_node_info): Adjust to descriptors being a pointer.
24780 (ipcp_modif_dom_walker): Make m_descriptors field garbage
24782 (ipcp_transform_function): Make descriptors variable garbage
24785 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
24787 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
24788 * config/i386/avx512dqintrin.h: Ditto.
24789 * config/i386/avx512fintrin.h: Ditto.
24790 * config/i386/i386.c: Handle new builtins.
24791 * config/i386/i386-builtin.def: Add new builtins.
24792 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
24793 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
24795 2017-01-23 Jakub Jelinek <jakub@redhat.com>
24796 Martin Liska <mliska@suse.cz>
24798 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
24799 * asan.c (asan_expand_poison_ifn): Support stores and use
24800 appropriate ASAN report function.
24801 * internal-fn.c (expand_ASAN_POISON_USE): New function.
24802 * internal-fn.def (ASAN_POISON_USE): Declare.
24803 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
24804 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
24805 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
24806 ASAN_POISON calls w/o LHS.
24807 * tree-ssa.c (execute_update_addresses_taken): Create clobber
24808 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
24809 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
24810 * gimplify.c (asan_poison_variables): Add attribute
24811 use_after_scope_memory to variables that really needs to live
24813 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
24814 having the attribute.
24816 2017-01-23 Martin Liska <mliska@suse.cz>
24818 * asan.c (create_asan_shadow_var): New function.
24819 (asan_expand_poison_ifn): Likewise.
24820 * asan.h (asan_expand_poison_ifn): New declaration.
24821 * internal-fn.c (expand_ASAN_POISON): Likewise.
24822 * internal-fn.def (ASAN_POISON): New builtin.
24823 * sanopt.c (pass_sanopt::execute): Expand
24824 asan_expand_poison_ifn.
24825 * tree-inline.c (copy_decl_for_dup_finish): Make function
24827 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
24828 * tree-ssa.c (is_asan_mark_p): New function.
24829 (execute_update_addresses_taken): Rewrite local variables
24830 (identified just by use-after-scope as addressable) into SSA.
24832 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24834 * doc/install.texi (Specific): opensource.apple.com uses https
24835 now. Remove trailing slash.
24837 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
24839 * README.Portability: Remove note on an Irix compatibility issue.
24841 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
24843 * gcov.c (INCLUDE_ALGORITHM): Define.
24844 (INCLUDE_VECTOR): Define.
24845 No longer include <vector> and <algorithm> directly.
24847 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
24849 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
24851 * doc/invoke.texi (Code Gen Options): Ditto.
24853 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
24856 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
24858 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
24860 rtl-optimization/79125
24861 * cprop.c (local_cprop_pass): Handle cases where we make an
24862 unconditional trap.
24864 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
24868 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
24869 read from, for big endian.
24871 2017-01-20 Jiong Wang <jiong.wang@arm.com>
24873 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
24874 register pauth builtins for LP64 only.
24876 2017-01-20 Marek Polacek <polacek@redhat.com>
24879 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
24882 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24884 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
24886 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
24887 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
24888 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
24890 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24893 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
24894 in big-endian mode when they are not a single duplicated value.
24896 2017-01-20 Richard Biener <rguenther@suse.de>
24898 * BASE-VER: Bump to 7.0.1.
24900 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
24902 * omp-low.c (omplow_simd_context): New struct. Use it...
24903 (lower_rec_simd_input_clauses): ...here and...
24904 (lower_rec_input_clauses): ...here to hold common data. Adjust all
24905 references to idx, lane, max_vf, is_simt.
24907 2017-01-20 Graham Markall <graham.markall@embecosm.com>
24909 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
24912 2017-01-20 Martin Jambor <mjambor@suse.cz>
24914 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
24915 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
24917 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
24918 (GTFILES): Rename hsa.c to hsa-common.c.
24919 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
24920 * hsa-dump.c: Likewise.
24921 * hsa-gen.c: Likewise.
24922 * hsa-regalloc.c: Likewise.
24923 * ipa-hsa.c: Likewise.
24924 * omp-expand.c: Likewise.
24925 * omp-low.c: Likewise.
24926 * toplev.c: Likewise.
24928 2017-01-20 Marek Polacek <polacek@redhat.com>
24931 * doc/invoke.texi: Document -Wduplicated-branches.
24932 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
24933 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
24934 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
24935 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
24936 return 0 only when not OEP_LEXICOGRAPHIC.
24937 (fold_build_cleanup_point_expr): Use the expression
24938 location when building CLEANUP_POINT_EXPR.
24939 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
24940 * tree.c (add_expr): Handle error_mark_node.
24942 2017-01-20 Martin Liska <mliska@suse.cz>
24945 * tree-profile.c (init_ic_make_global_vars): Do not call
24947 (gimple_init_gcov_profiler): Likewise.
24949 2017-01-20 Martin Liska <mliska@suse.cz>
24952 * cgraph.h (maybe_create_reference): Remove argument and
24954 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
24956 * ipa-cp.c (create_specialized_node): Likewise.
24957 * symtab.c (symtab_node::maybe_create_reference): Handle
24958 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
24960 2017-01-20 Martin Liska <mliska@suse.cz>
24962 * read-rtl-function.c (function_reader::create_function): Use
24963 build_decl instread of build_decl_stat.
24965 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
24967 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
24968 * config/i386/avx512dqintrin.h: Ditto.
24969 * config/i386/avx512fintrin.h: Ditto.
24970 * config/i386/i386-builtin-types.def: Add new types.
24971 * config/i386/i386.c: Handle new types.
24972 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
24973 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
24974 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
24975 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
24976 (__builtin_ia32_kshiftridi): New.
24977 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
24979 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
24983 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
24984 define to rs6000_init_stack_protect_guard.
24985 (rs6000_init_stack_protect_guard): New function.
24987 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24988 Yunqiang Su <yunqiang.su@imgtec.com>
24990 * config.gcc (supported_defaults): Add madd4.
24991 (with_madd4): Add validation.
24992 (all_defaults): Add madd4.
24993 * config/mips/mips.opt (mmadd4): New option.
24994 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24996 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24998 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
24999 (ISA_HAS_FUSED_MADD4): Likewise.
25000 * doc/invoke.texi (-mmadd4): Document the new option.
25001 * doc/install.texi (--with-madd4): Document the new option.
25003 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25005 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
25006 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
25007 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
25008 (aarch64_init_pauth_hint_builtins): New.
25009 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
25010 (aarch64_expand_builtin): Expand new builtins.
25012 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25014 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
25015 * combine-stack-adj.c (no_unhandled_cfa): Handle
25016 REG_CFA_TOGGLE_RA_MANGLE.
25017 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
25018 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
25019 info for return address signing.
25020 (aarch64_expand_epilogue): Likewise.
25022 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25024 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
25025 * config/aarch64/aarch64-protos.h
25026 (aarch64_return_address_signing_enabled): New declaration.
25027 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
25029 (aarch64_expand_prologue): Sign return address before it's pushed onto
25031 (aarch64_expand_epilogue): Authenticate return address fetched from
25033 (aarch64_override_options): Sanity check for ILP32 and ISA level.
25034 (aarch64_attributes): New function attributes for "sign-return-address".
25035 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
25036 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
25037 ("*do_return"): Generate combined instructions according to key index.
25038 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
25039 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
25041 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
25042 * config/aarch64/aarch64.opt (msign-return-address=): New.
25043 * doc/extend.texi (AArch64 Function Attributes): Documents
25044 "sign-return-address=".
25045 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
25047 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
25049 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
25050 overall option summary.
25052 2017-01-19 Jiong Wang <jiong.wang@arm.com>
25054 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
25055 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
25056 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
25057 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
25059 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
25061 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
25062 -mpower9-minmax by default for -mcpu=power9.
25063 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
25064 128-bit floating point.
25066 2017-01-20 Alan Modra <amodra@gmail.com>
25068 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
25069 optimizing for size.
25071 2017-01-20 Alan Modra <amodra@gmail.com>
25074 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
25075 for strcmp and strncmp from corresponding builtin decl.
25077 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25079 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
25080 instead of i386/rtems-64.h.
25081 * config/i386/rtems-64.h: Remove.
25083 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
25087 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
25089 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
25091 2017-01-19 Tamar Christina <tamar.christina@arm.com>
25093 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
25094 Change int to HOST_WIDE_INT.
25095 * config/aarch64/aarch64-protos.h
25096 (aarch64_simd_gen_const_vector_dup): Likewise.
25097 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
25099 2017-01-19 David Malcolm <dmalcolm@redhat.com>
25101 * langhooks-def.h (lhd_type_for_size): New decl.
25102 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
25103 * langhooks.c (lhd_type_for_size): New function, taken from
25106 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
25108 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
25109 define_bypass for CR latency.
25110 (power9-cracked-alu): Update bypass latency and remove power9-branch.
25111 (power9-alu2): Add define_bypass for CR latency.
25113 (power9-mul): Update insn latency.
25114 (power9-mul-compare): Update insn latency, bypass latency and remove
25117 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25119 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
25121 * config/aarch64/aarch64.md
25122 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
25123 aarch64_nopcrelative_literal_loads.
25124 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
25126 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
25128 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
25129 TARGET_LOONGSON_3A.
25130 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
25132 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
25135 * config.gcc (supported_defaults): Add lxc1-sxc1.
25136 (with_lxc1_sxc1): Add validation.
25137 (all_defaults): Add lxc1-sxc1.
25138 * config/mips/mips.opt (mlxc1-sxc1): New option.
25139 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
25141 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
25142 __mips_no_lxc1_sxc1.
25143 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
25144 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
25145 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
25147 2017-01-19 Richard Biener <rguenther@suse.de>
25149 PR tree-optimization/72488
25150 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
25151 sure to restore SSA info.
25152 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
25154 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
25156 PR rtl-optimization/79121
25157 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
25158 of the inner type when shifting an extended value.
25160 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25163 * symtab.c (symtab_node::equal_address_to): Fix comparing of
25164 interposable aliases.
25166 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
25169 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
25170 Use the evmergelohi instruction.
25171 (mov_si<mode>_e500_subreg4_2_le): Likewise.
25172 (mov_sitf_e500_subreg8_2_be): Likewise.
25173 (mov_sitf_e500_subreg12_2_le): Likewise.
25174 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
25175 (mov_si<mode>_e500_subreg4_2_be): Likewise.
25176 (mov_sitf_e500_subreg8_2_le): Likewise.
25177 (mov_sitf_e500_subreg12_2_be): Likewise.
25179 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25181 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
25182 attribute from vecsimple to vecperm.
25183 (altivec_vbpermq2): Likewise.
25185 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25188 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
25190 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25191 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
25192 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
25193 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
25194 case where N arg is SIZE_MAX.
25195 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
25196 (cmpstrsi): Add pattern.
25198 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25200 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25201 __builtin_vec_revb builtins.
25202 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
25203 built-in functions to support generation of the ISA 3.0 XXBR<x>
25204 vector byte reverse instructions.
25205 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
25206 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
25207 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
25208 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
25209 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
25210 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
25211 (P9V_BUILTIN_VEC_REVB): Likewise.
25212 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
25213 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
25214 (p9_xxbrq_v16qi): Likewise.
25215 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
25216 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
25217 (p9_xxbrh_v8hi): Likewise.
25218 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
25219 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
25220 vec_revb built-in functions.
25222 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
25224 PR rtl-optimization/78952
25225 * config/i386/i386.md (any_extract): New code iterator.
25226 (*insvqi_2): Use any_extract for source operand.
25227 (*insvqi_3): Use any_shiftrt for source operand.
25229 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25231 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
25233 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
25235 2017-01-18 Matthias Klose <doko@ubuntu.com>
25237 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
25239 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25241 * config/rs6000/altivec.h (vec_bperm): Change #define.
25242 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
25243 (altivec_vbpermq2): New define_insn.
25244 (altivec_vbpermd): Likewise.
25245 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
25246 function interface.
25247 (VBPERMD): Likewise.
25248 (VBPERM): New polymorphic function interface.
25249 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
25250 Add entries for P9V_BUILTIN_VEC_VBPERM.
25251 * doc/extend.texi: Add interfaces for vec_bperm.
25253 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25255 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
25256 first letter of error messages.
25257 (s390_resolve_overloaded_builtin): Likewise.
25258 * config/s390/s390.c (s390_expand_builtin): Likewise.
25259 (s390_invalid_arg_for_unprototyped_fn): Likewise.
25260 (s390_valid_target_attribute_inner_p): Likewise.
25261 * config/s390/s390.md ("tabort"): Likewise.
25263 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
25265 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
25266 (ISA_AVOID_DIV_HILO): New macro.
25267 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
25268 (ISA_HAS_DDIV): Likewise.
25270 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25272 * doc/invoke.texi (fabi-version): Correct number of occurrences.
25274 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25276 * doc/invoke.texi (fabi-version): Spelling fix.
25278 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25281 * doc/invoke.texi (fabi-version): Mention mangling fix for
25284 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
25287 * doc/invoke.texi (fabi-version): Document discriminator mangling.
25289 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
25292 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
25293 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
25295 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
25297 (stack_protect_test): Ditto.
25298 * config/rs6000/rs6000.opt (mstack-protector-guard=,
25299 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
25301 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
25302 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
25303 -mstack-protector-guard-offset=.
25304 (RS/6000 and PowerPC Options): Ditto.
25306 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25308 * config/i386/i386.h (MASK_CLASS_P): New define.
25309 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
25310 there are no registers from different register sets also when
25311 mask registers are used. Update function comment.
25312 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
25313 to (*k/*r) and (*k/*km) alternatives.
25315 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25317 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
25318 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
25319 (EH_RETURN_HANDLER_RTX): New define.
25320 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
25321 Force frame pointer in EH return functions.
25322 (aarch64_expand_epilogue): Add barrier for eh_return.
25323 (aarch64_final_eh_return_addr): Remove.
25324 (aarch64_eh_return_handler_rtx): New function.
25325 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
25327 (aarch64_eh_return_handler_rtx): New prototype.
25329 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25331 * config/rs6000/altivec.h (vec_rlmi): New #define.
25332 (vec_vrlnm): Likewise.
25333 (vec_rlnm): Likewise.
25334 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
25335 (UNSPEC_VRLNM): Likewise.
25336 (VIlong): New mode iterator.
25337 (altivec_vrl<VI_char>mi): New define_insn.
25338 (altivec_vrl<VI_char>nm): Likewise.
25339 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
25341 (VRLDNM): Likewise.
25342 (RLNM): New polymorphic function entry.
25343 (VRLWMI): New monomorphic function entry.
25344 (VRLDMI): Likewise.
25345 (RLMI): New polymorphic function entry.
25346 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
25347 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
25348 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
25351 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25354 * dwarf2out.c (field_byte_offset): Restore the
25355 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
25356 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
25357 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
25360 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25363 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
25365 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25368 * dwarf2out.c (add_data_member_location_attribute): For constant
25369 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
25370 instead of DW_AT_data_member_location, DW_AT_bit_offset and
25371 DW_AT_byte_size attributes.
25373 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
25375 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
25376 after forcing to constant memory when the code model is medium.
25378 2017-01-17 Julia Koval <julia.koval@intel.com>
25381 * config/i386/avx512fintrin.h
25382 (_mm512_i32gather_ps): Change __addr type to void const*.
25383 (_mm512_mask_i32gather_ps): Ditto.
25384 (_mm512_i32gather_pd): Ditto.
25385 (_mm512_mask_i32gather_pd): Ditto.
25386 (_mm512_i64gather_ps): Ditto.
25387 (_mm512_mask_i64gather_ps): Ditto.
25388 (_mm512_i64gather_pd): Ditto.
25389 (_mm512_mask_i64gather_pd): Ditto.
25390 (_mm512_i32gather_epi32): Ditto.
25391 (_mm512_mask_i32gather_epi32): Ditto.
25392 (_mm512_i32gather_epi64): Ditto.
25393 (_mm512_mask_i32gather_epi64): Ditto.
25394 (_mm512_i64gather_epi32): Ditto.
25395 (_mm512_mask_i64gather_epi32): Ditto.
25396 (_mm512_i64gather_epi64): Ditto.
25397 (_mm512_mask_i64gather_epi64): Ditto.
25398 (_mm512_i32scatter_ps): Change __addr type to void*.
25399 (_mm512_mask_i32scatter_ps): Ditto.
25400 (_mm512_i32scatter_pd): Ditto.
25401 (_mm512_mask_i32scatter_pd): Ditto.
25402 (_mm512_i64scatter_ps): Ditto.
25403 (_mm512_mask_i64scatter_ps): Ditto.
25404 (_mm512_i64scatter_pd): Ditto.
25405 (_mm512_mask_i64scatter_pd): Ditto.
25406 (_mm512_i32scatter_epi32): Ditto.
25407 (_mm512_mask_i32scatter_epi32): Ditto.
25408 (_mm512_i32scatter_epi64): Ditto.
25409 (_mm512_mask_i32scatter_epi64): Ditto.
25410 (_mm512_i64scatter_epi32): Ditto.
25411 (_mm512_mask_i64scatter_epi32): Ditto.
25412 (_mm512_i64scatter_epi64): Ditto.
25413 (_mm512_mask_i64scatter_epi64): Ditto.
25414 * config/i386/avx512pfintrin.h
25415 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
25416 (_mm512_mask_prefetch_i32gather_ps): Ditto.
25417 (_mm512_mask_prefetch_i64gather_pd): Ditto.
25418 (_mm512_mask_prefetch_i64gather_ps): Ditto.
25419 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
25420 (_mm512_prefetch_i32scatter_ps): Ditto.
25421 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
25422 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
25423 (_mm512_prefetch_i64scatter_pd): Ditto.
25424 (_mm512_prefetch_i64scatter_ps): Ditto.
25425 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
25426 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
25427 * config/i386/avx512vlintrin.h
25428 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
25429 (_mm_mmask_i32gather_ps): Ditto.
25430 (_mm256_mmask_i32gather_pd): Ditto.
25431 (_mm_mmask_i32gather_pd): Ditto.
25432 (_mm256_mmask_i64gather_ps): Ditto.
25433 (_mm_mmask_i64gather_ps): Ditto.
25434 (_mm256_mmask_i64gather_pd): Ditto.
25435 (_mm_mmask_i64gather_pd): Ditto.
25436 (_mm256_mmask_i32gather_epi32): Ditto.
25437 (_mm_mmask_i32gather_epi32): Ditto.
25438 (_mm256_mmask_i32gather_epi64): Ditto.
25439 (_mm_mmask_i32gather_epi64): Ditto.
25440 (_mm256_mmask_i64gather_epi32): Ditto.
25441 (_mm_mmask_i64gather_epi32): Ditto.
25442 (_mm256_mmask_i64gather_epi64): Ditto.
25443 (_mm_mmask_i64gather_epi64): Ditto.
25444 (_mm256_i32scatter_ps): Change __addr type to void*.
25445 (_mm256_mask_i32scatter_ps): Ditto.
25446 (_mm_i32scatter_ps): Ditto.
25447 (_mm_mask_i32scatter_ps): Ditto.
25448 (_mm256_i32scatter_pd): Ditto.
25449 (_mm256_mask_i32scatter_pd): Ditto.
25450 (_mm_i32scatter_pd): Ditto.
25451 (_mm_mask_i32scatter_pd): Ditto.
25452 (_mm256_i64scatter_ps): Ditto.
25453 (_mm256_mask_i64scatter_ps): Ditto.
25454 (_mm_i64scatter_ps): Ditto.
25455 (_mm_mask_i64scatter_ps): Ditto.
25456 (_mm256_i64scatter_pd): Ditto.
25457 (_mm256_mask_i64scatter_pd): Ditto.
25458 (_mm_i64scatter_pd): Ditto.
25459 (_mm_mask_i64scatter_pd): Ditto.
25460 (_mm256_i32scatter_epi32): Ditto.
25461 (_mm256_mask_i32scatter_epi32): Ditto.
25462 (_mm_i32scatter_epi32): Ditto.
25463 (_mm_mask_i32scatter_epi32): Ditto.
25464 (_mm256_i32scatter_epi64): Ditto.
25465 (_mm256_mask_i32scatter_epi64): Ditto.
25466 (_mm_i32scatter_epi64): Ditto.
25467 (_mm_mask_i32scatter_epi64): Ditto.
25468 (_mm256_i64scatter_epi32): Ditto.
25469 (_mm256_mask_i64scatter_epi32): Ditto.
25470 (_mm_i64scatter_epi32): Ditto.
25471 (_mm_mask_i64scatter_epi32): Ditto.
25472 (_mm256_i64scatter_epi64): Ditto.
25473 (_mm256_mask_i64scatter_epi64): Ditto.
25474 (_mm_i64scatter_epi64): Ditto.
25475 (_mm_mask_i64scatter_epi64): Ditto.
25476 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
25477 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
25478 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
25479 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
25480 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
25481 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
25482 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
25483 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
25484 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
25485 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
25486 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
25487 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
25488 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
25489 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
25490 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
25491 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
25492 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
25493 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
25494 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
25495 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
25496 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
25497 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
25498 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
25499 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
25500 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
25501 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
25502 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
25503 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
25504 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
25505 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
25506 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
25507 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
25508 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
25509 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
25510 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
25511 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
25512 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
25513 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
25514 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
25515 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
25516 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
25517 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
25518 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
25519 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
25520 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
25521 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
25522 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
25523 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
25524 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
25525 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
25526 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
25527 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
25528 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
25529 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
25530 definitions accordingly.
25532 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
25533 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25536 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
25539 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
25542 * ira-conflicts.c (ira_build_conflicts): Update total conflict
25543 hard regs for inner regno.
25545 2017-01-17 Martin Liska <mliska@suse.cz>
25548 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
25549 assumption and add comment.
25551 2017-01-17 Nathan Sidwell <nathan@acm.org>
25553 * ipa-visibility.c (localize_node): New function, broken out of ...
25554 (function_and_variable_visibility): ... here. Call it.
25556 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
25558 PR middle-end/77445
25559 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
25560 correctly set frequency of oudgoing edge.
25561 (duplicate_thread_path): Fix profile updating.
25563 2017-01-17 Jakub Jelinek <jakub@redhat.com>
25566 * configure.ac: Add GCC_BASE_VER.
25567 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
25568 version from BASE-VER file.
25569 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
25570 (gcc.o): Depend on $(BASEVER).
25571 * common.opt (dumpfullversion): New option.
25572 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
25573 * doc/invoke.texi: Document -dumpfullversion.
25574 * doc/install.texi: Document --with-gcc-major-version-only.
25575 * configure: Regenerated.
25577 2017-01-17 Richard Biener <rguenther@suse.de>
25579 PR tree-optimization/71433
25580 * tree-vrp.c (register_new_assert_for): Merge same asserts
25581 on all incoming edges.
25582 (process_assert_insertions_for): Handle insertions at the
25585 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
25587 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
25588 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
25590 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
25593 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
25596 2017-01-17 Alan Modra <amodra@gmail.com>
25599 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
25600 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
25601 symbolic stack limit when pic.
25603 2017-01-16 Martin Sebor <msebor@redhat.com>
25605 PR tree-optimization/78608
25606 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
25608 2017-01-16 Jeff Law <law@redhat.com>
25611 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
25612 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
25613 for several include directories that may be relative to sysroot.
25614 * config/i386/x-mingw32 (gplus_includedir): Define.
25615 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
25616 (native_system_includedir): Likewise.
25617 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
25618 override if TARGET_SYSTEM_ROOT is defined.
25619 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
25621 PR tree-optimization/79090
25622 PR tree-optimization/33562
25623 PR tree-optimization/61912
25624 PR tree-optimization/77485
25625 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
25626 and computed trims into the dump file.
25628 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
25630 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
25632 2017-01-16 Jakub Jelinek <jakub@redhat.com>
25635 * gimplify.c (gimplify_init_constructor): If want_value and
25636 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
25640 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
25641 sequence. Formatting fixes.
25642 (doloop_optimize): Formatting fixes.
25645 * gcc.c (debug_level_greater_than_spec_func): New function.
25646 (static_spec_functions): Add debug-level-gt spec function.
25647 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
25649 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25650 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
25651 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
25652 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
25653 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
25654 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
25656 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
25658 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
25659 QImode fixups to general and mask registers only.
25661 2017-01-16 Carl Love <cel@us.ibm.com>
25663 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
25664 for built-in functions
25665 vector signed char vec_nabs (vector signed char)
25666 vector signed short vec_nabs (vector signed short)
25667 vector signed int vec_nabs (vector signed int)
25668 vector signed long long vec_nabs (vector signed long long)
25669 vector float vec_nabs (vector float)
25670 vector double vec_nabs (vector double)
25671 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
25673 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
25674 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
25675 * doc/extend.texi: Update the documentation file for the new built-in
25678 2017-01-16 Martin Sebor <msebor@redhat.com>
25680 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
25683 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25685 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
25686 UNSPEC_VSX__XXSPLTD to require special splat handling.
25688 2017-01-16 David Malcolm <dmalcolm@redhat.com>
25691 * system.h: Poison strndup.
25693 2017-01-16 Alan Modra <amodra@gmail.com>
25696 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
25699 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
25701 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
25703 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
25705 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
25706 call recog here. Assert that INSN_CODE (insn) is non-negative.
25708 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25711 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
25713 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
25714 in the currently scheduled RTL fragment.
25716 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25718 PR rtl-optimization/78751
25719 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
25722 2017-01-14 Jeff Law <law@redhat.com>
25724 PR tree-optimization/79090
25725 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
25726 variable length stores.
25727 (compute_trims): Delete dead assignment to *trim_tail.
25728 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
25731 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
25733 PR rtl-optimization/78626
25734 PR rtl-optimization/78727
25735 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
25736 of a block, and split such blocks after everything else is finished.
25738 2017-01-14 Alan Modra <amodra@gmail.com>
25741 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
25742 target legitimate_combined_insn.
25743 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
25744 (rs6000_legitimate_combined_insn): New function.
25745 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
25747 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
25748 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
25749 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
25751 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
25753 * doc/frontends.texi (G++ and GCC): Remove references to Java.
25755 2017-01-13 Jeff Law <law@redhat.com>
25757 PR tree-optimization/33562
25758 PR tree-optimization/61912
25759 PR tree-optimization/77485
25760 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
25762 (delete_dead_assignment): Likewise.
25763 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
25764 statement to delete_dead_call and delete_dead_assignment.
25766 2017-01-13 David Malcolm <dmalcolm@redhat.com>
25769 * substring-locations.c (format_warning_va): Strengthen case 1 so
25770 that both endpoints of the substring must be within the format
25771 range for just the substring to be printed.
25773 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
25775 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
25776 * config/i386/i386.c (ix86_target_string): Add missing options
25777 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
25778 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
25779 flags_other and ix86_target_other to flags2_other. Display unknown
25781 (ix86_valid_target_attribute_inner_p): Add missing options and
25782 reorder options by implied ISAs, as in ix86_target_string.
25784 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25786 * hash-table.h (hash_table::too_empty_p): New function.
25787 (hash_table::expand): Use it.
25788 (hash_table::traverse): Likewise.
25789 (hash_table::empty_slot): Use sizeof (value_type) instead of
25790 sizeof (PTR) to convert bytes to elements. Shrink the table
25791 if the current size is excessive for the current number of
25794 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
25796 * ira-costs.c (record_reg_classes): Break from the inner loop
25797 early once alt_fail is known to be true. Update outer loop
25798 handling accordingly.
25800 2017-01-13 Jeff Law <law@redhat.com>
25802 * tree-ssa-dse.c (decrement_count): New function.
25803 (increment_start_addr, maybe_trim_memstar_call): Likewise.
25804 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
25805 when we know the partially dead statement is a mem* function.
25807 PR tree-optimization/61912
25808 PR tree-optimization/77485
25809 * tree-ssa-dse.c: Include expr.h.
25810 (maybe_trim_constructor_store): New function.
25811 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
25813 PR tree-optimization/33562
25814 PR tree-optimization/61912
25815 PR tree-optimization/77485
25816 * doc/invoke.texi: Document new dse-max-object-size param.
25817 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
25818 * tree-ssa-dse.c: Include params.h.
25819 (dse_store_status): New enum.
25820 (initialize_ao_ref_for_dse): New, partially extracted from
25822 (valid_ao_ref_for_dse, normalize_ref): New.
25823 (setup_live_bytes_from_ref, compute_trims): Likewise.
25824 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
25825 (maybe_trim_partially_dead_store): Likewise.
25826 (maybe_trim_complex_store): Likewise.
25827 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
25828 Track what bytes live from the original store. Return tri-state
25829 for dead, partially dead or live.
25830 (dse_dom_walker): Add constructor, destructor and new private members.
25831 (delete_dead_call, delete_dead_assignment): New extracted from
25833 (dse_optimize_stmt): Make a member of dse_dom_walker.
25834 Use initialize_ao_ref_for_dse.
25836 PR tree-optimization/33562
25837 PR tree-optimization/61912
25838 PR tree-optimization/77485
25839 * sbitmap.h (bitmap_count_bits): Prototype.
25840 (bitmap_clear_range, bitmap_set_range): Likewise.
25841 * sbitmap.c (bitmap_clear_range): New function.
25842 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
25844 2017-01-13 Martin Liska <mliska@suse.cz>
25847 * function.c (set_cfun): Add new argument force.
25848 * function.h (set_cfun): Likewise.
25849 * ipa-inline-transform.c (inline_call): Use the function when
25850 strict alising from is dropped for function we inline to.
25852 2017-01-13 Richard Biener <rguenther@suse.de>
25854 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
25855 for dumping GIMPLE INTEGER_CSTs.
25857 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25859 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
25860 to 201112L since C++17.
25862 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
25865 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
25866 if -fsanitize=kernel-address is present.
25868 2017-01-13 Richard Biener <rguenther@suse.de>
25870 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
25871 as _Literal ( type ) number in case usual suffixes do not
25872 preserve all information.
25874 2017-01-13 Richard Biener <rguenther@suse.de>
25876 PR tree-optimization/77283
25877 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
25878 and ssa-iterators.h.
25879 (is_feasible_trace): Implement a cost model based on joiner
25882 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
25885 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
25886 char or short to __float128/_Float128 directly.
25888 2017-01-12 Martin Sebor <msebor@redhat.com>
25890 to -Wformat-overflow.
25891 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
25892 (min_bytes_remaining): Same.
25893 (get_string_length): Same.
25894 (format_string): Same.
25895 (format_directive): Same.
25897 (pass_sprintf_length::handle_gimple_call): Same.
25899 2017-01-12 Jakub Jelinek <jakub@redhat.com>
25901 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
25902 info.nowrite calls with no lhs that can't throw. Return bool
25903 whether gsi_remove has been called or not.
25904 (pass_sprintf_length::handle_gimple_call): Return bool whether
25905 try_substitute_return_value called gsi_remove. Formatting fix.
25906 (pass_sprintf_length::execute): Don't use gsi_remove if
25907 handle_gimple_call returned true.
25910 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
25911 be removed due to side-effects, don't remove following barrier nor
25912 turn the successor edge into fallthru edge.
25914 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25917 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
25918 element-reversing loads and stores as not swappable.
25920 2017-01-12 Nathan Sidwell <nathan@acm.org>
25921 Nicolai Stange <nicstange@gmail.com>
25923 * combine.c (try_combine): Don't ignore result of overlap checking
25924 loop. Combine overlap & asm check into single loop.
25926 2017-01-12 Richard Biener <rguenther@suse.de>
25928 * tree-pretty-print.c (dump_generic_node): Provide -gimple
25929 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
25931 2017-01-12 Richard Biener <rguenther@suse.de>
25933 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
25934 and TS_TARGET_OPTION directly derive from TS_BASE.
25935 * tree-core.h (tree_optimization_option): Derive from tree_base.
25936 (tree_target_option): Likewise.
25938 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
25940 * config/i386/i386.c (memory_address_length): Increase len
25941 only when rip_relative_addr_p returns false.
25943 2017-01-11 Julia Koval <julia.koval@intel.com>
25945 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
25946 (OPTION_MASK_ISA_SGX_SET): New.
25947 (ix86_handle_option): Handle OPT_msgx.
25948 * config.gcc: Added sgxintrin.h.
25949 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
25950 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
25951 * config/i386/i386.c (ix86_target_string): Add -msgx.
25953 (ix86_option_override_internal): Handle new options.
25954 (ix86_valid_target_attribute_inner_p): Add sgx.
25955 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
25956 * config/i386/i386.opt: Add msgx.
25957 * config/i386/sgxintrin.h: New file.
25958 * config/i386/x86intrin.h: Add sgxintrin.h.
25960 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25963 * fold-const.c (maybe_nonzero_address): Return 1 for function
25965 (tree_single_nonzero_warnv_p): Don't handle function local objects
25969 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
25972 2017-01-11 David Malcolm <dmalcolm@redhat.com>
25975 * opts.c: Include "spellcheck.h"
25976 (struct string_fragment): New struct.
25977 (struct edit_distance_traits<const string_fragment &>): New
25979 (get_closest_sanitizer_option): New function.
25980 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
25982 2017-01-11 Jakub Jelinek <jakub@redhat.com>
25984 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
25986 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
25987 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
25988 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
25989 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
25990 for initial die_offset if dwarf_split_debug_info.
25991 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
25992 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
25994 (output_skeleton_debug_sections): Formatting fix. Use
25995 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
25996 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
25998 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
26000 * config/arm/cortex-a53.md: Add bypasses for
26001 cortex_a53_r2f_cvt.
26002 (cortex_a53_r2f): Only use for transfers.
26003 (cortex_a53_f2r): Likewise.
26004 (cortex_a53_r2f_cvt): Add reservation for conversions.
26005 (cortex_a53_f2r_cvt): Likewise.
26007 2017-01-11 Tamar Christina <tamar.christina@arm.com>
26009 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
26010 to all inlined functions, change static to extern.
26012 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
26015 * config/arm/arm.c (legitimize_pic_address): Handle reference to
26017 (arm_assemble_integer): Likewise.
26019 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
26021 * config.gcc: Use new awk script to check CPU, FPU and architecture
26022 parameters for --with-... options.
26023 * config/arm/parsecpu.awk: New file
26024 * config/arm/arm-cpus.in: New file.
26025 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
26027 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
26029 * config/arm/t-arm: Update dependency rules.
26030 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
26031 of processing .def files.
26032 * config/arm/genopt.sh: Deleted.
26033 * config/arm/gentune.sh: Deleted.
26034 * config/arm/arm-cores.def: Deleted.
26035 * config/arm/arm-arches.def: Deleted.
26036 * config/arm/arm-fpus.def: Deleted.
26037 * config/arm/arm-tune.md: Regenerated.
26038 * config/arm/arm-tables.opt: Regenerated.
26039 * config/arm/arm-cpu.h: New generated file.
26040 * config/arm/arm-cpu-data.h: New generated file.
26041 * config/arm/arm-cpu-cdata.h: New generated file.
26043 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
26046 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
26048 (input_varpool_node): Unpack dynamically_initialized bit.
26050 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
26052 PR rtl-optimization/79032
26053 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
26054 the alignment of the adjusted memory reference against that of MODE,
26055 instead of the alignment of the original memory reference.
26057 2017-01-11 Martin Jambor <mjambor@suse.cz>
26059 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
26061 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
26062 decorated functions.
26064 2017-01-11 Richard Biener <rguenther@suse.de>
26066 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
26067 set range/nonnull info for PHI results. Do not set it on
26068 stmts marked for removal.
26070 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
26072 * expr.c (store_field): In the bitfield case, fetch the return value
26073 from the registers before applying a single big-endian adjustment.
26074 Always do a final load for a BLKmode value not larger than a word.
26076 2017-01-10 David Malcolm <dmalcolm@redhat.com>
26079 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26080 that we correctly handle column numbers greater than
26081 LINE_MAP_MAX_COLUMN_NUMBER.
26083 2017-01-10 Martin Sebor <msebor@redhat.com>
26085 PR middle-end/78245
26086 * gimple-ssa-sprintf.c (get_destination_size): Call
26087 {init,fini}object_sizes.
26088 * tree-object-size.c (addr_object_size): Adjust.
26089 (pass_through_call): Adjust.
26090 (pass_object_sizes::execute): Adjust.
26091 * tree-object-size.h (fini_object_sizes): Declare.
26093 2017-01-10 Martin Sebor <msebor@redhat.com>
26095 PR tree-optimization/78775
26096 * builtins.c (get_size_range): Move...
26097 * calls.c: ...to here.
26098 (alloc_max_size): Accept zero argument.
26099 (operand_signed_p): Remove.
26100 (maybe_warn_alloc_args_overflow): Call get_size_range.
26101 * calls.h (get_size_range): Declare.
26103 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
26105 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
26106 from TI's devices.csv file as of September 2016.
26107 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
26109 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
26111 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
26112 * doc/invoke.texi: Likewise.
26113 * doc/md.texi: Likewise.
26114 * doc/objc.texi: Likewise.
26116 2017-01-10 Joshua Conner <joshconner@google.com>
26118 * config/arm/fuchsia-elf.h: New file.
26119 * config/fuchsia.h: New file.
26120 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
26121 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
26123 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
26125 2016-01-10 Richard Biener <rguenther@suse.de>
26127 PR tree-optimization/79034
26128 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26129 Propagate out degenerate PHIs in the joiner.
26131 2017-01-10 Martin Liska <mliska@suse.cz>
26133 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
26134 (sort_congruence_classes_by_decl_uid): Likewise.
26135 (sort_congruence_class_groups_by_decl_uid): Likewise.
26136 (sem_item_optimizer::merge_classes): Sort class, groups in these
26137 classes and members in the groups by DECL_UID of declarations.
26138 This would make merge operations stable.
26140 2017-01-10 Martin Liska <mliska@suse.cz>
26142 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
26143 usage of m_classes_vec.
26144 (sem_item_optimizer::~sem_item_optimizer): Likewise.
26145 (sem_item_optimizer::get_group_by_hash): Likewise.
26146 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
26147 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
26148 (sem_item_optimizer::verify_classes): Likewise.
26149 (sem_item_optimizer::process_cong_reduction): Likewise.
26150 (sem_item_optimizer::dump_cong_classes): Likewise.
26151 (sem_item_optimizer::merge_classes): Likewise.
26152 * ipa-icf.h (congruence_class_hash): Rename from
26153 congruence_class_group_hash. Remove declaration of m_classes_vec.
26155 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
26157 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
26158 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
26159 * config.gcc: Add avx512vpopcntdqintrin.h.
26160 * config/i386/avx512vpopcntdqintrin.h: New.
26161 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
26162 * config/i386/i386-builtin-types.def: Add new types.
26163 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
26164 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
26165 __builtin_ia32_vpopcountq_v8di_mask): New.
26166 * config/i386/i386-c.c (ix86_target_macros_internal): Define
26167 __AVX512VPOPCNTDQ__.
26168 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
26169 (PTA_AVX512VPOPCNTDQ): Define.
26170 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
26171 TARGET_AVX512VPOPCNTDQ_P): Define.
26172 * config/i386/i386.opt: Add mavx512vpopcntdq.
26173 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
26174 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
26176 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26178 PR middle-end/77484
26179 * predict.def (PRED_CALL): Set to 67.
26181 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
26183 * expr.c (store_field): In the bitfield case, if the value comes from
26184 a function call and is of an aggregate type returned in registers, do
26185 not modify the field mode; extract the value in all cases if the mode
26186 is BLKmode and the size is not larger than a word.
26188 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
26191 * config/i386/cpuid.h: Fix undefined behavior.
26193 2017-01-04 Jeff Law <law@redhat.com>
26195 PR tree-optimization/79007
26196 PR tree-optimization/67955
26197 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
26198 conservative for pt.null when flag_non_call_exceptions is on.
26200 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26202 PR translation/79019
26203 PR translation/79020
26204 * params.def (PARAM_INLINE_MIN_SPEEDUP,
26205 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
26206 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
26208 * config/avr/avr.opt (maccumulate-args): Likewise.
26209 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
26210 * common.opt (freport-bug): Likewise.
26211 * cif-code.def (CIF_FINAL_ERROR): Likewise.
26212 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
26213 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
26214 translatable string.
26215 * config/i386/i386.c (function_value_32): Likewise.
26216 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
26217 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
26219 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
26220 * common/config/msp430/msp430-common.c (msp430_handle_option):
26222 * symtab.c (symtab_node::verify_base): Likewise.
26223 * opts.c (set_debug_level): Likewise.
26224 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
26225 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
26226 missing whitespace to translatable strings.
26227 * config/avr/avr.md (bswapsi2): Fix typo in comment.
26228 * config/sh/superh.h: Likewise.
26229 * config/i386/xopintrin.h: Likewise.
26230 * config/i386/znver1.md: Likewise.
26231 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
26232 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
26233 * double-int.h (struct double_int): Likewise.
26234 * double-int.c (div_and_round_double): Likewise.
26235 * wide-int.cc: Likewise.
26236 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
26237 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
26238 * cfgcleanup.c (crossjumps_occured): Renamed to ...
26239 (crossjumps_occurred): ... this.
26240 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
26243 PR tree-optimization/78899
26244 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
26245 returning bool return struct loop *, NULL for failure and the new
26247 (versionable_outer_loop_p): Don't version outer loop if it has
26248 dont_vectorized bit set.
26249 (tree_if_conversion): When versioning outer loop, ensure
26250 tree_if_conversion is performed also on the inner loop of the
26251 non-vectorizable outer loop copy.
26252 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
26253 LOOP_VECTORIZED in inner loop of the scalar outer loop and
26254 prevent vectorization of it.
26255 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
26256 the outer loop vectorization of the non-scalar version is attempted
26257 before vectorization of the inner loop in scalar version. If
26258 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
26259 vectorization of its inner loop.
26260 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
26261 has 2 inner loops, rename also on edges from bb whose single pred
26262 is outer_loop->header. Fix typo in function comment.
26264 2017-01-09 Martin Sebor <msebor@redhat.com>
26267 * asan.c (asan_emit_stack_protection): Increase local buffer size
26268 to avoid snprintf truncation warning.
26270 2017-01-09 Andrew Pinski <apinski@cavium.com>
26272 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
26273 to reference thunderx2t99 for the tuning structure
26274 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
26276 (thunderx2t99_extra_costs): This.
26277 * config/aarch64/aarch64-tune.md: Regenerate.
26278 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
26279 (vulcan_addrcost_table): This.
26280 (vulcan_regmove_cost): Rename to ...
26281 (thunderx2t99_regmove_cost): This.
26282 (vulcan_vector_cost): Rename to ...
26283 (thunderx2t99_vector_cost): this.
26284 (vulcan_branch_cost): Rename to ...
26285 (thunderx2t99_branch_cost): This.
26286 (vulcan_tunings): Rename to ...
26287 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
26288 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
26290 2017-01-09 Martin Jambor <mjambor@suse.cz>
26294 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
26295 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
26296 (propagate_vr_accross_jump_function): Use the above function for all
26297 value range computations for pass-through jump functions and type
26298 converasion from explicit value range values.
26299 (ipcp_propagate_stage): Do not attempt to deduce types of formal
26300 parameters from TYPE_ARG_TYPES.
26301 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
26302 (ipa_write_node_info): Stream type of the actual argument.
26303 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
26305 2017-01-09 Martin Liska <mliska@suse.cz>
26308 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
26309 (lookup_compiler): Do not show error message with have_E.
26311 2017-01-09 Jakub Jelinek <jakub@redhat.com>
26313 PR tree-optimization/78938
26314 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
26315 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
26316 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
26317 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
26320 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26322 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
26325 2017-01-09 Richard Biener <rguenther@suse.de>
26327 PR tree-optimization/78997
26328 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
26329 name condition properly.
26331 2017-01-09 Richard Biener <rguenther@suse.de>
26334 * dwarf2out.c (is_cxx): New overload with context.
26335 (is_naming_typedef_decl): Use it.
26337 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26339 * invoke.texi (Option Summary): Correct spacing in option lists
26340 and add line breaks to fix over-long lines.
26342 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26344 PR middle-end/17660
26346 * extend.texi (Common Variable Attributes): Add xref to GCC
26347 Internals manual to explain mode attribute keywords.
26349 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
26352 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
26353 and Preprocessor Options.
26354 (Options for Linking): Document -pthread here....
26355 (RS/6000 and PowerPC Options): ...not here.
26356 (Solaris 2 Options): ...or here.
26357 * doc/cppopts.texi: Document -pthread.
26359 2017-01-08 Martin Sebor <msebor@redhat.com>
26361 PR middle-end/77708
26362 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
26363 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
26364 New member functions.
26365 (format_directive): Used them.
26367 (pass_sprintf_length::handle_gimple_call): Same.
26368 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
26369 to avoid truncation for any argument.
26370 (extract_affine_mul): Same.
26371 * tree.c (get_file_function_name): Same.
26373 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26375 PR middle-end/77484
26376 * predict.def (PRED_INDIR_CALL): Set to 86.
26378 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26380 PR preprocessor/54124
26381 * doc/cppopts.texi: Reformat -d subtable to list the full name
26382 of the options. Add cross-reference to the docs for the general
26383 compiler -d options.
26384 * doc/invoke.texi (Developer Options): Add cross-reference to the
26385 preprocessor-specific -d option documentation.
26387 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26389 PR preprocessor/13498
26390 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
26391 redudant material, and reflect new command-line options.
26392 (System Headers): Likewise.
26394 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
26396 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
26397 -isystem, and -idirafter. Copy-edit.
26398 * doc/cppopts.texi: Copy-edit. Remove contradiction about
26399 default for -ftrack-macro-expansion. Delete obsolete and
26400 badly-formatted implementation details about -fdebug-cpp output.
26401 * doc/cppwarnopts.texi: Copy-edit.
26403 2017-01-07 David Malcolm <dmalcolm@redhat.com>
26406 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
26407 that the transition from a max line width >= 1<<10 to narrower
26408 lines works correctly.
26410 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
26412 * doc/options.texi (PerFunction): New.
26413 * opt-functions.awk (switch_flags): Map both Optimization and
26414 PerFunction to CL_OPTIMIZATION.
26415 * opth-gen.awk: Test for PerFunction flag along with
26417 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
26418 it only when the latter is present. Skip those that don't in
26419 the hash function generator.
26420 * common.opt (fvar-tracking): Mark as PerFunction instead of
26422 (fvar-tracking-assignments): Likewise.
26423 (fvar-tracking-assignments-toggle): Likewise.
26424 (fvar-tracking-uninit): Likewise.
26426 2017-01-07 Jakub Jelinek <jakub@redhat.com>
26428 PR translation/79018
26429 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
26432 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
26435 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
26436 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
26437 TARGET_LONG_JUMP_TABLE_OFFSETS.
26438 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
26439 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
26440 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
26441 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
26442 * config/m68k/m68k.md (tablejump expander): Likewise.
26443 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
26444 TARGET_LONG_JUMP_TABLE_OFFSETS.
26445 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
26446 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
26448 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
26449 David Holsgrove <david.holsgrove@xilinx.com>
26451 * common/config/microblaze/microblaze-common.c
26452 (TARGET_EXCEPT_UNWIND_INFO): Remove.
26453 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
26455 * config/microblaze/microblaze.c (microblaze_must_save_register)
26456 (microblaze_expand_epilogue, microblaze_return_addr): Handle
26458 (microblaze_eh_return): New function.
26459 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
26460 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
26461 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
26462 * config/microblaze/microblaze.md (eh_return): New pattern.
26464 2017-01-06 Jakub Jelinek <jakub@redhat.com>
26466 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
26467 GCC_DIAGNOSTIC_STRINGIFY): Define.
26469 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
26471 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26473 * config/arm/arm.md (<mcrr>): New.
26475 * config/arm/arm.c (arm_arch5te): New.
26476 (arm_option_override): Set arm_arch5te.
26477 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
26479 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
26480 (arm_mcrr_qualifiers): ... this. New.
26481 (MRRC_QUALIFIERS): Define to...
26482 (arm_mrrc_qualifiers): ... this. New.
26483 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
26485 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
26486 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
26487 (MRRCI, mrrc, MRRC): New.
26488 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
26489 VUNSPEC_MRRC2): New.
26491 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26493 * config/arm/arm.md (<mcr>): New.
26495 * config/arm/arm.c (arm_coproc_builtin_available): Add
26496 support for mcr, mrc, mcr2 and mrc2.
26497 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
26498 (arm_mcr_qualifiers): ... this. New.
26499 (MRC_QUALIFIERS): Define to ...
26500 (arm_mrc_qualifiers): ... this. New.
26501 (MCR_QUALIFIERS): Define to ...
26502 (arm_mcr_qualifiers): ... this. New.
26503 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
26505 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
26506 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
26507 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
26508 VUNSPEC_MRC2): New.
26510 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26512 * config/arm/arm.md (*ldc): New.
26516 * config/arm/arm.c (arm_coproc_builtin_available): Add
26517 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
26518 (arm_coproc_ldc_stc_legitimate_address): New.
26519 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
26520 'qualifier_const_pointer'.
26521 (LDC_QUALIFIERS): Define to...
26522 (arm_ldc_qualifiers): ... this. New.
26523 (STC_QUALIFIERS): Define to...
26524 (arm_stc_qualifiers): ... this. New.
26525 * config/arm/arm-protos.h
26526 (arm_coproc_ldc_stc_legitimate_address): New.
26527 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
26528 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
26529 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
26530 stc2, stcl, stc2l): New.
26531 * config/arm/constraints.md (Uz): New.
26532 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
26533 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
26534 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
26535 VUNSPEC_STC2L): New.
26537 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26539 * config/arm/arm.md (<cdp>): New.
26540 * config/arm/arm.c (neon_const_bounds): Rename this ...
26541 (arm_const_bounds): ... this.
26542 (arm_coproc_builtin_available): New.
26543 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
26544 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
26545 (CDP_QUALIFIERS): Define to...
26546 (arm_cdp_qualifiers): ... this. New.
26548 (arm_expand_builtin_args): Add case for 6 arguments.
26549 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
26550 (arm_const_bounds): ... this.
26551 (arm_coproc_builtin_available): New.
26552 * config/arm/arm_acle.h (__arm_cdp): New.
26554 * config/arm/arm_acle_builtins.def (cdp): New.
26556 * config/arm/iterators.md (CDPI,CDP,cdp): New.
26557 * config/arm/neon.md: Rename all 'neon_const_bounds' to
26558 'arm_const_bounds'.
26559 * config/arm/types.md (coproc): New.
26560 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
26561 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
26562 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
26563 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
26565 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26567 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
26568 (UBINOP_QUALIFIERS): New.
26570 (acle_builtin_data): New. Change comment.
26571 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
26572 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
26573 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
26574 arm_acle_builtins.def.
26575 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
26576 (arm_init_acle_builtins): New.
26577 (CRC32_BUILTIN): Remove.
26578 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
26579 crc32cb, crc32ch and crc32cw.
26580 (arm_init_crc32_builtins): Remove.
26581 (arm_init_builtins): Use arm_init_acle_builtins rather
26582 than arm_init_crc32_builtins.
26583 (arm_expand_acle_builtin): New.
26584 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
26585 * config/arm/arm_acle_builtins.def: New.
26587 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
26589 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
26590 (arm_builtin_datum): ... this.
26591 (arm_init_neon_builtin): Rename to ...
26592 (arm_init_builtin): ... this. Add a new parameters PREFIX
26593 and USE_SIG_IN_NAME.
26594 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
26595 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
26596 'arm_builtin_datum'.
26597 (arm_init_vfp_builtins): Likewise.
26598 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
26599 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
26600 (arm_expand_neon_args): Rename to ...
26601 (arm_expand_builtin_args): ... this. Rename builtin_arg
26602 enum values and differentiate between ARG_BUILTIN_MEMORY
26603 and ARG_BUILTIN_NEON_MEMORY.
26604 (arm_expand_neon_builtin_1): Rename to ...
26605 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
26606 values, arm_expand_builtin_args and add bool parameter NEON.
26607 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
26608 (arm_expand_vfp_builtin): Likewise.
26609 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
26611 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26613 PR middle-end/77484
26614 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
26615 * predict.c (tree_estimate_probability_bb): Reverse direction of
26616 polymorphic call predictor.
26618 2017-01-06 David Malcolm <dmalcolm@redhat.com>
26620 * passes.c (execute_one_pass): Split out pass-skipping logic into...
26621 (determine_pass_name_match): ...this new function and...
26622 (should_skip_pass_p): ...this new function.
26624 2017-01-06 Nathan Sidwell <nathan@acm.org>
26626 * ipa-visibility.c (function_and_variable_visibility): Reformat
26627 comments and long lines. Remove extrneous if.
26628 * symtab.c (symtab_node::make_decl_local): Fix code format.
26629 (symtab_node::set_section_for_node): Fix comment typo.
26631 2017-01-06 Martin Liska <mliska@suse.cz>
26634 * lra-constraints.c: Rename invariant to lra_invariant.
26635 * predict.c (set_even_probabilities): Initialize e to NULL.
26637 2017-01-05 Martin Sebor <msebor@redhat.com>
26639 PR tree-optimization/78910
26640 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
26641 (format_integer): Correct off-by-one error in the handling
26642 of precision with negative numbers in signed conversions..
26644 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
26646 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
26648 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26650 PR tree-optimization/71016
26651 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
26652 factor_out_conditional_conversion. Formatting fix.
26653 (factor_out_conditional_conversion): Add cond_stmt argument.
26654 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
26655 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
26658 2017-01-05 David Malcolm <dmalcolm@redhat.com>
26660 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
26661 read-rtl-function.o, and selftest-rtl.o.
26662 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
26663 (selftest::aarch64_test_loading_full_dump): New function.
26664 (selftest::aarch64_run_selftests): New function.
26665 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
26666 selftest::aarch64_run_selftests.
26667 * config/i386/i386.c
26668 (selftest::ix86_test_loading_dump_fragment_1): New function.
26669 (selftest::ix86_test_loading_call_insn): New function.
26670 (selftest::ix86_test_loading_full_dump): New function.
26671 (selftest::ix86_test_loading_unspec): New function.
26672 (selftest::ix86_run_selftests): Call the new functions.
26673 * emit-rtl.c (maybe_set_max_label_num): New function.
26674 * emit-rtl.h (maybe_set_max_label_num): New decl.
26675 * function.c (instantiate_decls): Guard call to
26676 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
26677 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
26679 * gensupport.c (gen_reader::gen_reader): Pass "false"
26680 for new "compact" param of rtx_reader.
26681 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
26682 rather than an empty string for NULL strings.
26683 * read-md.c: Potentially include config.h rather than bconfig.h.
26684 Wrap include of errors.h with #ifdef GENERATOR_FILE.
26685 (have_error): New global, copied from errors.c.
26686 (md_reader::read_name): Rename to...
26687 (md_reader::read_name_1): ...this, adding "out_loc" param,
26688 and converting "missing name or number" to returning false, rather
26690 (md_reader::read_name): Reimplement in terms of read_name_1.
26691 (md_reader::read_name_or_nil): New function.
26692 (md_reader::read_string): Handle "(nil)" by returning NULL.
26693 (md_reader::md_reader): Add new param "compact".
26694 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
26695 (md_reader::read_file): New method.
26696 * read-md.h (md_reader::md_reader): Add new param "compact".
26697 (md_reader::read_file): New method.
26698 (md_reader::is_compact): New accessor.
26699 (md_reader::read_name): Convert return type from void to file_location.
26700 (md_reader::read_name_or_nil): New decl.
26701 (md_reader::read_name_1): New decl.
26702 (md_reader::m_compact): New field.
26703 (noop_reader::noop_reader): Pass "false" for new "compact" param
26705 (rtx_reader::rtx_reader): Add new "compact" param.
26706 (rtx_reader::read_rtx_operand): Make virtual and convert return
26707 type from void to rtx.
26708 (rtx_reader::read_until): New decl.
26709 (rtx_reader::handle_any_trailing_information): New virtual function.
26710 (rtx_reader::postprocess): New virtual function.
26711 (rtx_reader::finalize_string): New virtual function.
26712 (rtx_reader::m_in_call_function_usage): New field.
26713 (rtx_reader::m_reuse_rtx_by_id): New field.
26714 * read-rtl-function.c: New file.
26715 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
26716 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
26717 (selftest::verify_three_block_rtl_cfg): New decl.
26718 * read-rtl-function.h: New file.
26719 * read-rtl.c: Potentially include config.h rather than bconfig.h.
26720 For host, include function.h, memmodel.h, and emit-rtl.h.
26721 (one_time_initialization): New function.
26722 (struct compact_insn_name): New struct.
26723 (compact_insn_names): New array.
26724 (find_code): Handle insn codes in compact dumps.
26725 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
26726 (bind_subst_iter_and_attr): Likewise.
26727 (add_condition_to_string): Likewise.
26728 (add_condition_to_rtx): Likewise.
26729 (apply_attribute_uses): Likewise.
26730 (add_current_iterators): Likewise.
26731 (apply_iterators): Likewise.
26732 (initialize_iterators): Guard usage of apply_subst_iterator with
26733 #ifdef GENERATOR_FILE.
26734 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
26735 (md_reader::read_mapping): Likewise.
26736 (add_define_attr_for_define_subst): Likewise.
26737 (add_define_subst_attr): Likewise.
26738 (read_subst_mapping): Likewise.
26739 (check_code_iterator): Likewise.
26740 (rtx_reader::read_rtx): Likewise. Move one-time initialization
26742 (one_time_initialization): New function.
26743 (rtx_reader::read_until): New method.
26744 (read_flags): New function.
26745 (parse_reg_note_name): New function.
26746 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
26747 Handle reuse_rtx ids.
26748 Wrap iterator lookup within #ifdef GENERATOR_FILE.
26749 Add parsing support for RTL dumps, mirroring the special-cases in
26750 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
26751 values, and calling handle_any_trailing_information.
26752 (rtx_reader::read_rtx_operand): Convert return type from void
26753 to rtx, returning return_rtx. Handle case 'e'. Call
26754 finalize_string on XSTR and XTMPL fields.
26755 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
26756 "(nil)" values were omitted. Call the postprocess vfunc on the
26758 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
26759 class ctor. Initialize m_in_call_function_usage. Call
26760 one_time_initialization.
26761 * rtl-tests.c (selftest::test_uncond_jump): Call
26762 set_new_first_and_last_insn.
26763 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
26764 * selftest-rtl.c: New file.
26765 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
26766 (selftest::get_insn_by_uid): New decl.
26767 * selftest-run-tests.c (selftest::run_tests): Call
26768 read_rtl_function_c_tests.
26769 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
26770 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
26773 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
26775 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
26776 operands in a special way. Assert that pos+len <= mode precision.
26778 2017-01-05 Jakub Jelinek <jakub@redhat.com>
26780 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
26781 3 argument Alias with unlimited for the negative form.
26782 (fno-vect-cost-model): Removed.
26784 2017-01-05 Martin Liska <mliska@suse.cz>
26786 * hsa-gen.c (gen_hsa_divmod): New function.
26787 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
26789 2017-01-05 Martin Liska <mliska@suse.cz>
26792 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
26795 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26797 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
26798 small constant length operands.
26800 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26802 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
26803 between loop iterations.
26805 2017-01-05 Martin Liska <mliska@suse.cz>
26808 * gimplify.c (gimplify_decl_expr): Compare to
26809 asan_poisoned_variables instread of checking flags.
26810 (gimplify_target_expr): Likewise.
26811 (gimplify_expr): Likewise.
26812 (gimplify_function_tree): Conditionally initialize
26813 asan_poisoned_variables.
26815 2017-01-04 Jeff Law <law@redhat.com>
26817 PR tree-optimizatin/78812
26818 * rtl.h (contains_mem_rtx_p): Prototype.
26819 * ifcvt.c (containts_mem_rtx_p): Move from here to...
26820 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
26821 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
26822 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
26823 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
26825 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26827 * input.c (assert_char_at_range): Default-initialize actual_range.
26829 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26831 * df-scan.c (df_ref_create_structure): Make regno unsigned,
26832 to match the caller.
26834 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26836 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
26837 insns after final jump in test to emit dummy move.
26839 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26841 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
26842 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
26844 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
26846 * multiple_target.c (create_dispatcher_calls): Init e_next.
26847 * tree-ssa-loop-split.c (split_loop): Init border.
26848 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
26851 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
26856 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
26857 (altivec_register_operand): Do not return true if the operand
26858 contains a SUBREG mixing SImode and SFmode.
26859 (vsx_register_operand): Likewise.
26860 (vsx_reg_sfsubreg_ok): New predicate.
26861 (vfloat_operand): Do not return true if the operand contains a
26862 SUBREG mixing SImode and SFmode.
26863 (vint_operand): Likewise.
26864 (vlogical_operand): Likewise.
26865 (gpc_reg_operand): Likewise.
26866 (int_reg_operand): Likewise.
26867 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
26868 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
26869 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
26871 (rs6000_emit_move_si_sf_subreg): New helper function.
26872 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
26873 fixup SUBREGs involving SImode and SFmode.
26874 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
26875 numbers for the new peephole2 optimization.
26876 (peephole2 for SFmode unions): New peephole2 to optimize cases in
26877 the GLIBC math library that do AND/IOR/XOR operations on single
26878 precision floating point.
26879 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
26880 target macros to say whether we need to avoid SUBREGs mixing
26882 (TARGET_ALLOW_SF_SUBREG): Likewise.
26883 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
26884 (UNSPEC_SI_FROM_SF): Likewise.
26885 (iorxor): Change spacing.
26886 (and_ior_xor): New iterator for AND, IOR, and XOR.
26887 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
26888 (movdi_from_sf_zero_ext): Likewise.
26889 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
26890 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
26891 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
26892 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
26893 (fms<mode>4): Likewise.
26894 (fnma<mode>4): Likewise.
26895 (fnms<mode>4): Likewise.
26896 (nfma<mode>4): Likewise.
26897 (nfms<mode>4): Likewise.
26899 2017-01-04 Marek Polacek <polacek@redhat.com>
26902 * doc/invoke.texi: Document -Wpointer-compare.
26904 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26906 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
26909 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
26910 descriptions for -gdwarf-5 and emit them as uleb128 instead of
26913 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
26916 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
26917 documentation of the powerpc_popcntb_ok attribute.
26918 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26919 code to issue warning messages if a requested CPU configuration is
26920 not supported by the binary (assembler and loader) toolchain.
26921 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
26922 made to define a built-in function that has been disabled.
26923 (paired_init_builtins): Add assertion to prevent ICE if attempt is
26924 made to define a built-in function that has been disabled.
26925 (altivec_init_builtins): Add comment explaining why definition
26926 of the DST built-in functions is not preceded by an assertion
26927 check. Add assertions to prevent ICE if attempts are made to
26928 define an altivec predicate or an abs* built-in function that has
26930 (htm_init_builtins): Add comment explaining why definition of the
26931 htm built-in functions is not preceded by an assertion check.
26933 2017-01-04 Jeff Law <law@redhat.com>
26935 PR tree-optimizatin/67955
26936 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
26937 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
26938 the points-to solution does not include pt_null. Use DECL_PT_UID
26941 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
26943 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
26944 Use gen_int_mode instead of gen_lopwart for const_int operands.
26946 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26948 PR tree-optimization/71563
26949 * match.pd: Simplify X << Y into X if Y is known to be 0 or
26950 out of range value - has low bits known to be zero.
26952 2017-01-04 Alan Modra <amodra@gmail.com>
26954 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
26955 * configure: Regenerate.
26956 * config.in: Regenerate.
26958 2017-01-04 Jakub Jelinek <jakub@redhat.com>
26961 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
26962 a substring of the message, but strcmp with the whole message. Ifdef
26963 ENABLE_NLS, translate the message first using dgettext.
26965 2017-01-03 Jeff Law <law@redhat.com>
26967 PR tree-optimizatin/78856
26968 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
26969 (mark_threaded_blocks): Remove code to truncate thread paths that
26970 cross multiple loop headers. Instead invalidate the cached loop
26971 iteration information and handle case of a thread path walking
26972 into an irreducible region.
26974 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
26977 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
26978 assertions. Add support for doing the signbit if the IEEE 128-bit
26979 floating point value is in a GPR.
26980 * config/rs6000/rs6000.md (Fsignbit): Delete.
26981 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
26982 Update the length attribute if the value is in a GPR.
26983 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
26984 the sign or zero extension instruction, since the value is always 0/1.
26985 (signbit<mode>2_dm2): Delete using <Fsignbit>.
26988 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
26989 extracting SImode to a GPR register so that we can generate a
26990 store, limit the vector to be in a traditional Altivec register
26991 for the vextuwrx instruction.
26993 2017-01-03 Ian Lance Taylor <iant@google.com>
26995 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
26997 2017-01-03 Martin Sebor <msebor@redhat.com>
26999 PR tree-optimization/78696
27000 * gimple-ssa-sprintf.c (format_floating): Correct handling of
27001 precision. Use MPFR for %f for greater fidelity. Correct handling
27003 (pass_sprintf_length::compute_format_length): Set width and precision
27004 specified by asrerisk to void_node for vararg functions.
27005 (try_substitute_return_value): Adjust dump output.
27007 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
27009 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
27011 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
27013 * doc/invoke.texi (SPARC options): Document -mlra as the default.
27014 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
27015 -mlra/-mno-lra was passed to the compiler.
27017 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
27019 PR rtl-optimization/65618
27020 * emit-rtl.c (try_split): Move initialization of "before" and
27021 "after" to just before the call to emit_insn_after_setloc.
27023 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
27025 * doc/md.texi (Standard Names): Remove reference to Java frontend.
27027 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
27029 * dwarf2out.c (gen_enumeration_type_die): When
27030 -gno-strict-dwarf, add a DW_AT_encoding attribute.
27032 2017-01-03 Jakub Jelinek <jakub@redhat.com>
27034 PR tree-optimization/78965
27035 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
27036 Change first argument from const call_info & to call_info &. For %n
27037 set info.nowrite to false.
27039 PR middle-end/78901
27040 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
27041 possibly throwing calls.
27043 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
27044 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
27045 and fns handling, rather than in a separate case SSA_NAME.
27047 2017-01-02 Jeff Law <law@redhat.com>
27049 * config/darwin-driver.c (darwin_driver_init): Const-correctness
27050 fixes for first_period and second_period variables.
27052 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
27055 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
27056 (*insvqi_1): New insn pattern.
27057 (*insvqi_1_mem_rex64): Ditto.
27058 (*insvqi_2): Ditto.
27059 (*insvqi_3): Rename from *insvqi.
27061 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
27063 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
27065 * doc/cfg.texi (Edges): Remove reference to Java.
27066 (Maintaining the CFG): Ditto.
27068 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27070 PR middle-end/77674
27071 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
27072 transparent aliases.
27074 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
27076 PR middle-end/77484
27077 * predict.def (PRED_CALL): Update hitrate.
27078 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
27079 * predict.c (tree_estimate_probability_bb): Split CALL predictor
27080 into direct/indirect/polymorphic variants.
27082 2017-01-01 Jakub Jelinek <jakub@redhat.com>
27084 Update copyright years.
27086 * gcc.c (process_command): Update copyright notice dates.
27087 * gcov-dump.c (print_version): Ditto.
27088 * gcov.c (print_version): Ditto.
27089 * gcov-tool.c (print_version): Ditto.
27090 * gengtype.c (create_file): Ditto.
27091 * doc/cpp.texi: Bump @copying's copyright year.
27092 * doc/cppinternals.texi: Ditto.
27093 * doc/gcc.texi: Ditto.
27094 * doc/gccint.texi: Ditto.
27095 * doc/gcov.texi: Ditto.
27096 * doc/install.texi: Ditto.
27097 * doc/invoke.texi: Ditto.
27099 Copyright (C) 2017 Free Software Foundation, Inc.
27101 Copying and distribution of this file, with or without modification,
27102 are permitted in any medium without royalty provided the copyright
27103 notice and this notice are preserved.