1 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * coretypes.h (pad_direction): New enum.
6 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
7 (FUNCTION_ARG_PADDING): Likewise.
8 * target.def (function_arg_padding): New hook.
9 * targhooks.h (default_function_arg_padding): Declare.
10 * targhooks.c (default_function_arg_padding): New function.
11 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
12 (TARGET_FUNCTION_ARG_PADDING): ...this.
13 * doc/tm.texi: Regenerate.
14 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16 (compute_argument_addresses): Likewise.
17 (load_register_parameters): Likewise.
18 (emit_library_call_value_1): Likewise.
19 (store_one_arg): Use targetm.calls.function_arg_padding instead
20 of FUNCTION_ARG_PADDING.
21 (must_pass_in_stack_var_size_or_pad): Likewise.
22 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
23 (emit_group_store): Likewise.
24 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
25 instead of FUNCTION_ARG_PADDING.
26 (emit_push_insn): Likewise, and propagate enum change throughout
28 * function.h (direction): Delete.
29 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
31 * function.c (assign_parm_find_stack_rtl): Likewise.
32 (assign_parm_setup_block_p): Likewise.
33 (assign_parm_setup_block): Likewise.
34 (gimplify_parameters): Likewise.
35 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
36 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
38 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
39 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
40 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
41 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
42 (aarch64_function_arg_padding): ...this new function.
43 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
44 (TARGET_FUNCTION_ARG_PADDING): Redefine.
45 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
46 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
47 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
48 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
49 (arm_pad_arg_upward): Replace with...
50 (arm_function_arg_padding): ...this new function.
51 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
53 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
54 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
55 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
56 (ia64_hpux_function_arg_padding): Replace with...
57 (ia64_function_arg_padding): ...this new function. Use pad_direction
58 instead of direction. Check for TARGET_HPUX.
59 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
60 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
61 (iq2000_function_arg_padding): New function.
62 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
63 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
64 (mips_function_arg_padding): ...this new function.
65 (mips_pad_reg_upward): Update accordingly.
66 (TARGET_FUNCTION_ARG_PADDING): Redefine.
67 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
68 targetm.calls.function_arg_padding.
69 (FUNCTION_ARG_PADDING): Delete.
70 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
71 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
72 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
73 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
74 (nios2_block_reg_padding): Return pad_direction instead of direction.
75 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
77 (nios2_function_arg_padding): Likewise. Make static.
78 (TARGET_FUNCTION_ARG_PADDING): Redefine.
79 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
80 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
81 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
82 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
83 (pa_function_arg_padding): Make static. Return pad_direction instead
85 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
86 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
87 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
88 instead of direction. Use targetm.calls.function_arg_padding.
89 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
90 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
91 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
92 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
93 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
95 (function_arg_padding): Rename to...
96 (rs6000_function_arg_padding): ...this. Make static. Return
97 pad_direction instead of direction.
98 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
99 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
100 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
101 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
102 instead of direction. Use targetm.calls.function_arg_padding.
103 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
104 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
105 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
106 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
107 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
108 (function_arg_padding): Rename to...
109 (rs6000_function_arg_padding): ...this. Make static. Return
110 pad_direction instead of direction.
111 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
112 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
113 * config/s390/s390.c (s390_function_arg_padding): New function.
114 (TARGET_FUNCTION_ARG_PADDING): Redefine.
115 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
116 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
117 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
118 (function_arg_padding): Rename to...
119 (sparc_function_arg_padding): ...this. Make static. Return
120 pad_direction instead of direction.
121 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
122 * config/spu/spu.c (spu_function_arg_padding): New function.
123 (TARGET_FUNCTION_ARG_PADDING): Redefine.
124 * system.h (FUNCTION_ARG_PADDING): Poison.
126 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
127 Alan Hayward <alan.hayward@arm.com>
128 David Sherwood <david.sherwood@arm.com>
130 * target.def (modes_tieable_p): New hook.
131 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
132 (TARGET_MODES_TIEABLE_P): ...this.
133 * doc/tm.texi.in: Regenerate.
134 * hooks.h (hook_bool_mode_mode_true): Declare.
135 * hooks.c (hook_bool_mode_mode_true): New function.
136 * combine.c (subst): Use targetm.modes_tieable_p instead of
138 * dse.c (find_shift_sequence): Likewise.
139 * expmed.c (extract_low_bits): Likewise.
140 * lower-subreg.c: Include target.h.
141 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
143 * rtlanal.c (rtx_cost): Likewise.
144 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
145 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
146 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
147 (TARGET_MODES_TIEABLE_P): Redefine.
148 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
149 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
150 (TARGET_MODES_TIEABLE_P): Redefine.
151 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
152 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
153 (arc_modes_tieable_p): New function.
154 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
155 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
156 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
157 (arm_modes_tieable_p): Make static.
158 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
159 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
160 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
161 (TARGET_MODES_TIEABLE_P): Redefine.
162 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
163 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
164 (TARGET_MODES_TIEABLE_P): Redefine.
165 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
166 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
167 (cr16_modes_tieable_p): New function.
168 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
169 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
170 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
171 (TRULY_NOOP_TRUNCATION): Update comment.
172 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
173 (TRULY_NOOP_TRUNCATION): Update comment.
174 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
175 (frv_modes_tieable_p): New function.
176 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
177 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
178 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
179 (TARGET_MODES_TIEABLE_P): Redefine.
180 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
181 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
182 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
183 (TARGET_MODES_TIEABLE_P): Redefine.
184 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
185 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
186 (ia64_modes_tieable_p): New function.
187 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
188 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
189 (iq2000_modes_tieable_p): New function.
190 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
191 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
192 (lm32_modes_tieable_p): New function.
193 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
194 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
195 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
196 (TARGET_MODES_TIEABLE_P): Redefine.
197 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
198 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
199 (m32r_modes_tieable_p): New function.
200 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
201 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
202 (m68k_modes_tieable_p): New function.
203 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
204 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
205 (mcore_modes_tieable_p): New function.
206 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
207 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
209 (TARGET_MODES_TIEABLE_P): Redefine.
210 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
211 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
212 * config/mips/mips.c (mips_modes_tieable_p): Make static.
213 (TARGET_MODES_TIEABLE_P): Redefine.
214 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
215 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
216 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
217 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
218 (mn10300_modes_tieable_p): ...this and make static.
219 (TARGET_MODES_TIEABLE_P): Redefine.
220 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
221 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
222 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
223 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
224 (msp430_modes_tieable_p): Make static.
225 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
226 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
227 (TARGET_MODES_TIEABLE_P): Redefine.
228 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
229 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
230 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
231 (TARGET_MODES_TIEABLE_P): Redefine.
232 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
233 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
234 * config/pa/pa.c (pa_modes_tieable_p): Make static.
235 (TARGET_MODES_TIEABLE_P): Redefine.
236 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
237 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
238 (pdp11_modes_tieable_p): New function.
239 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
240 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
241 (rs6000_modes_tieable_p): New function.
242 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
243 * config/powerpcspe/powerpcspe.md: Update comment.
244 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
245 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
246 (TARGET_MODES_TIEABLE_P): Redefine.
247 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
248 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
249 (rl78_modes_tieable_p): New function.
250 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
251 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
252 (rs6000_modes_tieable_p): New function.
253 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
254 * config/rs6000/rs6000.md: Update comment.
255 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
256 * config/rx/rx.c (rx_modes_tieable_p): New function.
257 (TARGET_MODES_TIEABLE_P): Redefine.
258 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
259 * config/s390/s390.c (s390_modes_tieable_p): New function.
260 (TARGET_MODES_TIEABLE_P): Redefine.
261 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
262 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
263 (sh_modes_tieable_p): New function.
264 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
265 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
266 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
267 (sparc_modes_tieable_p): Make static.
268 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
269 * config/spu/spu.c (spu_modes_tieable_p): New function.
270 (TARGET_MODES_TIEABLE_P): Redefine.
271 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
272 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
273 (TARGET_MODES_TIEABLE_P): Redefine.
274 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
275 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
276 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
277 * config/v850/v850.c (v850_modes_tieable_p): New function.
278 (TARGET_MODES_TIEABLE_P): Redefine.
279 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
280 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
281 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
282 (visium_modes_tieable_p): New function.
283 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
284 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
285 (xtensa_modes_tieable_p): New function.
286 * system.h (MODES_TIEABLE_P): Poison.
288 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
289 Alan Hayward <alan.hayward@arm.com>
290 David Sherwood <david.sherwood@arm.com>
292 * target.def (hard_regno_mode_ok): New hook.
293 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
294 (TARGET_HARD_REGNO_MODE_OK): ...this.
295 * doc/tm.texi.in: Regenerate.
296 * hooks.h (hook_bool_uint_mode_true): Declare.
297 * hooks.c (hook_bool_uint_mode_true): New function.
298 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
300 * genpreds.c (write_insn_preds_c): Add an include of target.h.
301 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
302 instead of HARD_REGNO_MODE_OK.
303 * caller-save.c: Include target.h.
304 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
306 * combine.c (can_combine_p): Likewise.
307 (combinable_i3pat): Likewise.
308 (can_change_dest_mode): Likewise.
309 * expr.c (init_expr_target): Likewise.
310 (convert_move): Likewise.
311 (convert_modes): Likewise.
312 * ira.c (setup_prohibited_class_mode_regs): Likewise.
313 (setup_prohibited_mode_move_regs): Likewise.
314 * ira.h (target_ira): Likewise.
315 * lra-assigns.c (find_hard_regno_for_1): Likewise.
316 * lra-constraints.c (process_alt_operands): Likewise.
317 (split_reg): Likewise.
318 * recog.c (peep2_find_free_register): Likewise.
319 * ree.c (combine_reaching_defs): Likewise.
320 * regcprop.c (maybe_mode_change): Likewise.
321 * reginfo.c (init_reg_sets_1): Likewise.
322 (choose_hard_reg_mode): Likewise.
323 (simplifiable_subregs): Likewise.
324 * regrename.c (check_new_reg_p): Likewise.
325 * reload.c (find_valid_class): Likewise.
326 (find_valid_class_1): Likewise.
327 (reload_inner_reg_of_subreg): Likewise.
328 (push_reload): Likewise.
329 (combine_reloads): Likewise.
330 (find_dummy_reload): Likewise.
331 (find_reloads): Likewise.
332 * reload1.c (find_reg): Likewise.
333 (set_reload_reg): Likewise.
334 (allocate_reload_reg): Likewise.
335 (choose_reload_regs): Likewise.
336 (reload_adjust_reg_for_temp): Likewise.
337 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
338 (simplify_subreg_regno): Likewise.
339 * sel-sched.c (init_regs_for_mode): Likewise.
340 * varasm.c (make_decl_rtl): Likewise.
341 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
342 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
344 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
345 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
346 (TARGET_HARD_REGNO_MODE_OK): Redefine.
347 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
348 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
349 (TARGET_HARD_REGNO_MODE_OK): Redefine.
350 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
351 (arc_mode_class): Delete.
352 (HARD_REGNO_MODE_OK): Delete.
353 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
354 (arc_hard_regno_mode_ok): Rename old array to...
355 (arc_hard_regno_mode_ok_modes): ...this.
356 (arc_conditional_register_usage): Update accordingly.
357 (arc_mode_class): Make static.
358 (arc_hard_regno_mode_ok): New function.
359 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
360 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
361 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
362 (arm_hard_regno_mode_ok): Make static.
363 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
365 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
366 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
367 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
369 (TARGET_HARD_REGNO_MODE_OK): Redefine.
370 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
371 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
372 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
373 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
374 (TARGET_HARD_REGNO_MODE_OK): Redefine.
375 * config/bfin/predicates.md (valid_reg_operand): Use
376 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
377 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
378 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
379 (TARGET_HARD_REGNO_MODE_OK): Redefine.
380 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
381 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
382 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
383 (cr16_hard_regno_mode_ok): Make static and return a bool.
384 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
385 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
386 (cris_hard_regno_mode_ok): New function.
387 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
388 (epiphany_mode_class): Delete.
389 (HARD_REGNO_MODE_OK): Delete.
390 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
391 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
392 (hard_regno_mode_ok): Rename to...
393 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
394 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
395 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
397 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
398 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
399 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
400 (frv_hard_regno_mode_ok): Make static and return a bool.
401 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
403 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
404 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
405 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
406 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
408 (TARGET_HARD_REGNO_MODE_OK): Redefine.
409 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
410 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
411 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
413 (TARGET_HARD_REGNO_MODE_OK): Redefine.
414 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
415 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
416 (ia64_hard_regno_mode_ok): New function.
417 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
418 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
419 (iq2000_hard_regno_mode_ok): New function.
420 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
421 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
422 (lm32_hard_regno_mode_ok): New function.
423 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
424 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
425 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
426 instead of HARD_REGNO_MODE_OK.
427 (m32c_hard_regno_ok): Rename to...
428 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
429 (m32c_cannot_change_mode_class): Update accordingly.
430 (TARGET_HARD_REGNO_MODE_OK): Redefine.
431 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
432 (m32r_mode_class): Delete.
433 (HARD_REGNO_MODE_OK): Delete.
434 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
435 (m32r_hard_regno_mode_ok): Rename to...
436 (m32r_hard_regno_modes): ...this.
437 (m32r_mode_class): Make static.
438 (m32r_hard_regno_mode_ok): New function.
439 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
440 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
441 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
442 (m68k_hard_regno_mode_ok): Make static.
443 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
444 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
445 (mcore_hard_regno_mode_ok): New function.
446 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
447 (HARD_REGNO_MODE_OK): Delete.
448 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
450 (microblaze_hard_regno_mode_ok_p): ...this and make static.
451 (microblaze_hard_regno_mode_ok): New function.
452 (TARGET_HARD_REGNO_MODE_OK): Redefine.
453 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
454 (mips_hard_regno_mode_ok): Delete.
455 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
456 (mips_hard_regno_mode_ok_p): ...this and make static.
457 (mips_hard_regno_mode_ok_p): Rename to...
458 (mips_hard_regno_mode_ok_uncached): ...this.
459 (mips_hard_regno_mode_ok): New function.
460 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
461 of HARD_REGNO_MODE_OK.
462 (mips_option_override): Update after above name changes.
463 (TARGET_HARD_REGNO_MODE_OK): Redefine.
464 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
465 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
466 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
467 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
468 (TARGET_HARD_REGNO_MODE_OK): Redefine.
469 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
470 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
471 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
472 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
473 (msp430_hard_regno_mode_ok): Make static and return a bool.
474 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
475 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
476 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
478 (TARGET_HARD_REGNO_MODE_OK): Redefine.
479 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
480 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
481 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
482 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
483 (PA_HARD_REGNO_MODE_OK): ...this
484 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
485 (PA_HARD_REGNO_MODE_OK): ...this.
486 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
487 (pa_hard_regno_mode_ok): New function.
488 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
489 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
490 (pdp11_hard_regno_mode_ok): New function.
491 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
492 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
494 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
496 (TARGET_HARD_REGNO_MODE_OK): Redefine.
497 (rs6000_hard_regno_mode_ok): Rename to...
498 (rs6000_hard_regno_mode_ok_uncached): ...this.
499 (rs6000_init_hard_regno_mode_ok): Update accordingly.
500 (rs6000_hard_regno_mode_ok): New function.
501 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
502 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
503 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
504 (riscv_hard_regno_mode_ok): ...this and make static.
505 (TARGET_HARD_REGNO_MODE_OK): Redefine.
506 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
507 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
508 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
509 (rl78_hard_regno_mode_ok): Make static and return bool.
510 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
511 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
513 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
514 (TARGET_HARD_REGNO_MODE_OK): Redefine.
515 (rs6000_hard_regno_mode_ok): Rename to...
516 (rs6000_hard_regno_mode_ok_uncached): ...this.
517 (rs6000_init_hard_regno_mode_ok): Update accordingly.
518 (rs6000_hard_regno_mode_ok): New function.
519 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
520 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
521 (TARGET_HARD_REGNO_MODE_OK): Redefine.
522 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
523 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
524 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
525 (TARGET_HARD_REGNO_MODE_OK): Redefine.
526 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
527 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
528 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
529 (sh_hard_regno_mode_ok): Make static.
530 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
531 instead of HARD_REGNO_MODE_OK.
532 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
533 (sparc_mode_class): Delete.
534 (HARD_REGNO_MODE_OK): Delete.
535 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
536 (hard_regno_mode_classes): Make static.
537 (sparc_mode_class): Likewise.
538 (sparc_hard_regno_mode_ok): New function.
539 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
540 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
541 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
543 (TARGET_HARD_REGNO_MODE_OK): Redefine.
544 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
545 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
546 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
547 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
548 (TARGET_HARD_REGNO_MODE_OK): Redefine.
549 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
550 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
551 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
552 (visium_hard_regno_mode_ok): New function.
553 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
554 instead of HARD_REGNO_MODE_OK.
555 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
556 (HARD_REGNO_MODE_OK): Delete.
557 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
558 (xtensa_hard_regno_mode_ok_p): ...this and make static.
559 (xtensa_option_override): Update accordingly.
560 (TARGET_HARD_REGNO_MODE_OK): Redefine.
561 (xtensa_hard_regno_mode_ok): New function.
562 * system.h (HARD_REGNO_MODE_OK): Poison.
564 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
565 Alan Hayward <alan.hayward@arm.com>
566 David Sherwood <david.sherwood@arm.com>
568 * target.def (hard_regno_call_part_clobbered): New hook.
569 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
570 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
571 * doc/tm.texi: Regenerate.
572 * hooks.h (hook_bool_uint_mode_false): Declare.
573 * hooks.c (hook_bool_uint_mode_false): New function.
574 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
575 * cselib.c (cselib_process_insn): Use
576 targetm.hard_regno_call_part_clobbered instead of
577 HARD_REGNO_CALL_PART_CLOBBERED.
578 * ira-conflicts.c (ira_build_conflicts): Likewise.
579 * ira-costs.c (ira_tune_allocno_costs): Likewise.
580 * lra-constraints.c (need_for_call_save_p): Likewise.
581 * lra-lives.c: Include target.h.
582 (check_pseudos_live_through_calls): Use
583 targetm.hard_regno_call_part_clobbered instead of
584 HARD_REGNO_CALL_PART_CLOBBERED.
585 * regcprop.c: Include target.h.
586 (copyprop_hardreg_forward_1): Use
587 targetm.hard_regno_call_part_clobbered instead of
588 HARD_REGNO_CALL_PART_CLOBBERED.
589 * reginfo.c (choose_hard_reg_mode): Likewise.
590 * regrename.c (check_new_reg_p): Likewise.
591 * reload.c (find_equiv_reg): Likewise.
592 * reload1.c (emit_reload_insns): Likewise.
593 * sched-deps.c (deps_analyze_insn): Likewise.
594 * sel-sched.c (init_regs_for_mode): Likewise.
595 (mark_unavailable_hard_regs): Likewise.
596 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
597 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
598 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
600 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
601 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
602 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
604 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
606 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
607 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
608 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
610 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
611 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
612 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
614 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
615 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
617 * config/powerpcspe/powerpcspe.c
618 (rs6000_hard_regno_call_part_clobbered): New function.
619 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
620 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
621 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
623 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
624 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
625 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
627 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
628 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
629 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
631 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
632 Alan Hayward <alan.hayward@arm.com>
633 David Sherwood <david.sherwood@arm.com>
635 * rtl.h (subreg_memory_offset): Declare.
636 * emit-rtl.c (subreg_memory_offset): New function.
637 * expmed.c (store_bit_field_1): Use it.
638 * expr.c (undefined_operand_subword_p): Likewise.
639 * simplify-rtx.c (simplify_subreg): Likewise.
641 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
643 PR rtl-optimization/57448
646 * optabs.c (expand_atomic_load): Place compiler memory barriers if
647 using atomic_load pattern.
648 (expand_atomic_store): Likewise.
650 2017-09-04 Jakub Jelinek <jakub@redhat.com>
653 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
654 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
655 handling. Use replace_call_with_value with NULL instead of
656 gsi_replace, unlink_stmt_vdef and release_defs.
658 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
661 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
663 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
666 * rtl.h (emit_library_call_value_1): Declare.
667 (emit_library_call): Replace declaration with a series of overloads.
668 Remove the parameter count argument.
669 (emit_library_call_value): Likewise.
670 * calls.c (emit_library_call_value_1): Make global. Replace varargs
671 with an "rtx_mode_t *".
672 (emit_library_call_value): Delete.
673 (emit_library_call): Likewise.
674 * asan.c (asan_emit_stack_protection): Update calls accordingly.
675 (asan_emit_allocas_unpoison): Likewise.
676 * builtins.c (expand_builtin_powi): Likewise.
677 (expand_asan_emit_allocas_unpoison): Likewise.
678 * cfgexpand.c (expand_main_function): Likewise.
679 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
680 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
681 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
682 * config/arm/arm.c (arm_trampoline_init): Likewise.
683 (arm_call_tls_get_addr): Likewise.
684 (arm_expand_divmod_libfunc): Likewise.
685 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
686 (smulsi3_highpart): Likewise.
687 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
688 (c6x_expand_compare): Likewise.
689 (c6x_expand_movmem): Likewise.
690 * config/frv/frv.c (frv_trampoline_init): Likewise.
691 * config/i386/i386.c (ix86_trampoline_init): Likewise.
692 (ix86_expand_divmod_libfunc): Likewise.
693 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
694 (ia64_expand_compare): Likewise.
695 (ia64_profile_hook): Likewise.
696 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
697 (nonlocal_goto): Likewise.
698 (restore_stack_nonlocal): Likewise.
699 * config/m32r/m32r.c (block_move_call): Likewise.
700 (m32r_trampoline_init): Likewise.
701 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
702 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
703 (m68k_call_m68k_read_tp): Likewise.
704 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
705 (microblaze_expand_divide): Likewise.
706 * config/mips/mips.h (mips_args): Likewise.
707 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
708 (MIPS_ICACHE_SYNC): Likewise.
709 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
710 (nios2_trampoline_init): Likewise.
711 * config/pa/pa.c (hppa_tls_call): Likewise.
712 (pa_trampoline_init): Likewise.
713 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
714 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
715 (expand_strn_compare): Likewise.
716 (rs6000_generate_compare): Likewise.
717 (rs6000_expand_float128_convert): Likewise.
718 (output_profile_hook): Likewise.
719 (rs6000_trampoline_init): Likewise.
720 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
721 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
722 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
723 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
724 (rs6000_generate_compare): Likewise.
725 (rs6000_expand_float128_convert): Likewise.
726 (output_profile_hook): Likewise.
727 (rs6000_trampoline_init): Likewise.
728 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
729 * config/sh/sh.c (sh_trampoline_init): Likewise.
730 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
731 (sparc_emit_float_lib_cmp): Likewise.
732 (sparc32_initialize_trampoline): Likewise.
733 (sparc64_initialize_trampoline): Likewise.
734 (sparc_profile_hook): Likewise.
735 * config/spu/spu.c (ea_load_store): Likewise.
736 * config/spu/spu.md (floatunssidf2): Likewise.
737 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
738 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
739 * config/visium/visium.c (expand_block_move_4): Likewise.
740 (expand_block_move_2): Likewise.
741 (expand_block_move_1): Likewise.
742 (expand_block_set_4): Likewise.
743 (expand_block_set_2): Likewise.
744 (expand_block_set_1): Likewise.
745 (visium_trampoline_init): Likewise.
746 (visium_profile_hook): Likewise.
747 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
748 (xtensa_setup_frame_addresses): Likewise.
749 (xtensa_trampoline_init): Likewise.
750 * except.c (sjlj_emit_function_enter): Likewise.
751 (sjlj_emit_function_exit): Likewise.
752 * explow.c (allocate_dynamic_stack_space): Likewise.
753 (probe_stack_range): Likewise.
754 * expr.c (convert_mode_scalar): Likewise.
755 * optabs.c (expand_binop): Likewise.
756 (expand_twoval_binop_libfunc): Likewise.
757 (expand_unop): Likewise.
758 (prepare_cmp_insn): Likewise.
759 (prepare_float_lib_cmp): Likewise.
760 (expand_float): Likewise.
761 (expand_fix): Likewise.
762 (expand_fixed_convert): Likewise.
763 (maybe_emit_sync_lock_test_and_set): Likewise.
764 (expand_atomic_compare_and_swap): Likewise.
765 (expand_mem_thread_fence): Likewise.
766 (expand_atomic_fetch_op): Likewise.
768 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
770 * doc/generic.texi (OpenACC): Adjust URL.
771 * doc/invoke.texi (C Dialect Options): Ditto.
773 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
775 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
776 predicate for operand 1. Add (m,<S>) constraint.
777 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
778 Prevent memory operand 1 with register operand 2.
780 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
782 PR rtl-optimization/82024
783 * combine.c (try_combine): If the combination result is a PARALLEL,
784 and we only need to retain the SET in there that would be placed
785 at I2, check that we can place that at I3 instead, before doing so.
787 2017-09-01 Jakub Jelinek <jakub@redhat.com>
790 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
792 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
793 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
796 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
797 Jeff Law <law@redhat.com>
799 * varasm.c (bss_initializer_p): Do not put constants into .bss
800 (categorize_decl_for_section): Handle bss_initializer_p returning
801 false when DECL_INITIAL is NULL.
803 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
806 * config/s390/s390.c (s390_can_inline_p): New function.
808 2017-09-01 Jeff Law <law@redhat.com>
810 PR tree-optimization/82052
811 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
812 Always initialize the returned slot after a hash table miss
815 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
817 * config/s390/s390.md (mem_signal_fence): Remove.
818 * doc/md.texi (mem_signal_fence): Remove.
819 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
821 * target-insns.def (mem_signal_fence): Remove.
823 2017-09-01 Jakub Jelinek <jakub@redhat.com>
826 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
829 * asan.c (create_odr_indicator): Strip name encoding from assembler
830 name before appending it after __odr_asan_.
832 2017-09-01 Martin Liska <mliska@suse.cz>
834 PR tree-optimization/82059
835 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
836 frequency only when an edge is redirected.
838 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
840 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
841 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
842 (arc_conditional_register_usage): Remove ARC600 lp_count
844 (arc_file_start): Emit Tag_ARC_CPU_variation.
845 (arc_can_use_doloop_p): New conditions to use ZOLs.
846 (hwloop_fail): New function.
847 (hwloop_optimize): Likewise.
848 (hwloop_pattern_reg): Likewise.
849 (arc_doloop_hooks): New struct, to be used with reorg_loops.
850 (arc_reorg_loops): New function, calls reorg_loops.
851 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
852 (arc600_corereg_hazard): Remove ZOL checking, case handled by
854 (arc_loop_hazard): Remove function, functionality moved into
856 (arc_hazard): Remove arc_loop_hazard call.
857 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
858 into hwloop_optimize.
859 (arc_label_align): Remove ZOL handling.
860 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
861 * config/arc/arc.md (doloop_begin): Remove pattern.
862 (doloop_begin_i): Likewise.
863 (doloop_end_i): Likewise.
864 (doloop_fallback): Likewise.
865 (doloop_fallback_m): Likewise.
866 (doloop_end): Reimplement expand.
867 (arc_lp): New pattern for LP instruction.
868 (loop_end): New pattern.
869 (loop_fail): Likewise.
870 (decrement_and_branch_until_zero): Likewise.
871 * config/arc/arc.opt (mlpc-width): New option.
872 * doc/invoke.texi (mlpc-width): Document option.
874 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
876 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
877 (arc_ccfsm_advance): Fix checking for delay slots.
878 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
880 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
882 * config/arc/arc.md (movqi_insn): Add stores to save constant long
884 (movhi_insn): Update store instruction constraint which are saving
885 6-bit short immediates.
886 (movsi_insn): Consider also short scaled load operations.
887 (zero_extendhisi2_i): Use Usd constraint instead of T.
888 (extendhisi2_i): Add q constraint.
889 (arc_clzsi2): Add type and length attributes.
890 (arc_ctzsi2): Likewise.
891 * config/arc/constraints.md (Usc): Update constraint, the
892 assembler can parse two relocations for a single instruction.
894 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
896 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
897 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
899 2017-08-31 Olivier Hainque <hainque@adacore.com>
901 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
902 match as powerpc-wrs-vxworks*.
904 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
906 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
907 register constraint for by-element operand.
908 (aarch64_mls_elt_merge<mode>): Likewise.
910 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
912 * config/arc/arc.c (arc_can_follow_jump): Check for short
915 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
917 * config.gcc: Use g.opt for arc.
918 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
919 functionality moved to ...
920 (legitimate_scaled_address_p): New function, ...here.
921 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
922 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
923 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
925 (arc_override_options): Handle G option.
926 (arc_output_pic_addr_const): Correct function definition.
927 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
928 (arc_decl_anon_ns_mem_p): Delete.
929 (arc_in_small_data_p): Overhaul this function to take into
930 consideration the value given via G option.
931 (arc_rewrite_small_data_1): Renamed and corrected old
932 arc_rewrite_small_data function.
933 (arc_rewrite_small_data): New function.
934 (small_data_pattern): Don't use pic_offset_table_rtx.
935 * config/arc/arc.h (CC1_SPEC): Recognize G option.
936 * config/arc/simdext.md (movmisalignv2hi): Use
937 prepare_move_operands function.
939 (movmisalign*): Likewise.
940 * doc/invoke.texi (ARC options): Document -G option.
942 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
944 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
946 * config/arc/arc.c (arc_print_operand): Output scalled address for
947 sdata whenever is possible.
948 (arc_in_small_data_p): Allow sdata for 64bit datum when double
949 load/stores are available.
950 (compact_sda_memory_operand): Check for the alignment required by
951 code density instructions.
952 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
954 * config/arc/constraints.md (Usd): Update constraint.
955 (Us0): New constraint.
956 (Usc): Update constraint.
958 2017-08-31 Richard Biener <rguenther@suse.de>
961 * dwarf2out.c (dwarf2out_early_global_decl): Process each
964 2017-08-31 Tamar Christina <tamar.christina@arm.com>
966 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
967 Resize type_signature.
969 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
970 Alan Hayward <alan.hayward@arm.com>
971 David Sherwood <david.sherwood@arm.com>
973 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
974 subregs whose inner modes can be stored in GPRs.
975 (aarch64_classify_index): Likewise.
977 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
978 Alan Hayward <alan.hayward@arm.com>
979 David Sherwood <david.sherwood@arm.com>
981 * config/aarch64/iterators.md (V_cmp_result): Rename to...
982 (V_INT_EQUIV): ...this.
983 (v_cmp_result): Rename to...
984 (v_int_equiv): ...this.
985 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
986 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
987 (copysign<mode>3): Likewise.
988 (aarch64_simd_bsl<mode>_internal): Likewise.
989 (aarch64_simd_bsl<mode>): Likewise.
990 (vec_cmp<mode><mode>): Likewise.
991 (vcond<mode><mode>): Likewise.
992 (vcond<v_cmp_mixed><mode>): Likewise.
993 (vcondu<mode><v_cmp_mixed>): Likewise.
994 (aarch64_cm<optab><mode>): Likewise.
995 (aarch64_cmtst<mode>): Likewise.
996 (aarch64_fac<optab><mode>): Likewise.
997 (vec_perm_const<mode>): Likewise.
998 (vcond_mask_<mode><v_cmp_result>): Rename to...
999 (vcond_mask_<mode><v_int_equiv>): ...this.
1000 (vec_cmp<mode><v_cmp_result>): Rename to...
1001 (vec_cmp<mode><v_int_equiv>): ...this.
1003 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1004 Alan Hayward <alan.hayward@arm.com>
1005 David Sherwood <david.sherwood@arm.com>
1007 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1009 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1010 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1011 (UNSPEC_LD4_DREG): New unspecs.
1012 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1013 (aarch64_ld2<mode>_dreg_be): Replace with...
1014 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1016 (aarch64_ld3<mode>_dreg_le)
1017 (aarch64_ld3<mode>_dreg_be): Replace with...
1018 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1020 (aarch64_ld4<mode>_dreg_le)
1021 (aarch64_ld4<mode>_dreg_be): Replace with...
1022 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1025 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1027 PR tree-optimization/81987
1028 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1029 insert an initializer in a location not dominated by the stride
1032 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1034 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1035 on the entire header of the finally block in the fallthru case.
1037 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1039 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1041 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1043 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1044 rs6000_emit_move_from_cr and call renamed function.
1045 (rs6000_emit_prologue): Call renamed functions.
1046 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1047 movesi_from_cr, remove volatile CRs.
1049 2017-08-30 Jon Beniston <jon@beniston.com>
1050 Richard Biener <rguenther@suse.de>
1052 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1053 of VECTOR_MODE_P check.
1054 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1055 element vector types.
1057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1059 * df.h (df_read_modify_subreg_p): Remove in favor of...
1060 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1061 const_rtx instead of an rtx.
1062 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1063 * df-problems.c (df_word_lr_mark_ref): Likewise.
1064 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1065 (mark_pseudo_reg_dead): Likewise.
1066 (mark_ref_dead): Likewise.
1067 * reginfo.c (init_subregs_of_mode): Likewise.
1068 * sched-deps.c (sched_analyze_1): Likewise.
1069 * df-scan.c (df_def_record_1): Likewise.
1070 (df_uses_record): Likewise.
1071 (df_read_modify_subreg_p): Remove in favor of...
1072 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1073 const_rtx instead of an rtx.
1075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1076 Alan Hayward <alan.hayward@arm.com>
1077 David Sherwood <david.sherwood@arm.com>
1079 * rtl.h (partial_subreg_p): New function.
1080 * caller-save.c (save_call_clobbered_regs): Use it.
1081 * calls.c (expand_call): Likewise.
1082 * combine.c (combinable_i3pat): Likewise.
1083 (simplify_set): Likewise.
1084 (make_extraction): Likewise.
1085 (make_compound_operation_int): Likewise.
1086 (gen_lowpart_or_truncate): Likewise.
1087 (force_to_mode): Likewise.
1088 (make_field_assignment): Likewise.
1089 (reg_truncated_to_mode): Likewise.
1090 (record_truncated_value): Likewise.
1091 (move_deaths): Likewise.
1092 * cse.c (record_jump_cond): Likewise.
1093 (cse_insn): Likewise.
1094 * cselib.c (cselib_lookup_1): Likewise.
1095 * expmed.c (extract_bit_field_using_extv): Likewise.
1096 * function.c (assign_parm_setup_reg): Likewise.
1097 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1098 * ira-build.c (create_insn_allocnos): Likewise.
1099 * lra-coalesce.c (merge_pseudos): Likewise.
1100 * lra-constraints.c (match_reload): Likewise.
1101 (simplify_operand_subreg): Likewise.
1102 (curr_insn_transform): Likewise.
1103 * lra-lives.c (process_bb_lives): Likewise.
1104 * lra.c (new_insn_reg): Likewise.
1105 (lra_substitute_pseudo): Likewise.
1106 * regcprop.c (mode_change_ok): Likewise.
1107 (maybe_mode_change): Likewise.
1108 (copyprop_hardreg_forward_1): Likewise.
1109 * reload.c (push_reload): Likewise.
1110 (find_reloads): Likewise.
1111 (find_reloads_subreg_address): Likewise.
1112 * reload1.c (alter_reg): Likewise.
1113 (eliminate_regs_1): Likewise.
1114 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1116 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
1118 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
1119 back to if statements, including unpack.
1121 2017-08-30 Martin Liska <mliska@suse.cz>
1124 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
1126 (func_checker::compare_asm_inputs_outputs): ... this function.
1127 (func_checker::compare_gimple_asm): Use the function to compare
1128 also ASM constrains.
1129 * ipa-icf-gimple.h: Rename the function.
1131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1132 Alan Hayward <alan.hayward@arm.com>
1133 David Sherwood <david.sherwood@arm.com>
1135 * coretypes.h (complex_mode): New type.
1136 * gdbhooks.py (build_pretty_printer): Handle it.
1137 * machmode.h (complex_mode): New class.
1138 (complex_mode::includes_p): New function.
1139 (is_complex_int_mode): Likewise.
1140 (is_complex_float_mode): Likewise.
1141 * genmodes.c (get_mode_class): Handle complex mode classes.
1142 * function.c (expand_function_end): Use is_complex_int_mode.
1144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1145 Alan Hayward <alan.hayward@arm.com>
1146 David Sherwood <david.sherwood@arm.com>
1148 * coretypes.h (scalar_mode_pod): New typedef.
1149 * gdbhooks.py (build_pretty_printer): Handle it.
1150 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
1151 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
1152 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
1153 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
1156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1157 Alan Hayward <alan.hayward@arm.com>
1158 David Sherwood <david.sherwood@arm.com>
1160 * machmode.h (mode_for_vector): Take a scalar_mode instead
1162 * stor-layout.c (mode_for_vector): Likewise.
1163 * explow.c (promote_mode): Use as_a <scalar_mode>.
1164 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
1166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1167 Alan Hayward <alan.hayward@arm.com>
1168 David Sherwood <david.sherwood@arm.com>
1170 * target.def (preferred_simd_mode): Take a scalar_mode
1171 instead of a machine_mode.
1172 * targhooks.h (default_preferred_simd_mode): Likewise.
1173 * targhooks.c (default_preferred_simd_mode): Likewise.
1174 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
1175 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
1176 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
1177 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
1178 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
1179 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
1180 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
1181 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
1183 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
1184 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
1185 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
1186 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
1187 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
1188 * doc/tm.texi: Regenerate.
1189 * optabs-query.c (can_vec_mask_load_store_p): Return false for
1192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1193 Alan Hayward <alan.hayward@arm.com>
1194 David Sherwood <david.sherwood@arm.com>
1196 * target.def (scalar_mode_supported_p): Take a scalar_mode
1197 instead of a machine_mode.
1198 * targhooks.h (default_scalar_mode_supported_p): Likewise.
1199 * targhooks.c (default_scalar_mode_supported_p): Likewise.
1200 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
1201 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1202 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
1203 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
1204 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
1205 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
1206 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
1207 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
1208 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
1209 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
1210 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
1211 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
1213 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
1214 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
1215 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1216 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1217 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
1219 * doc/tm.texi: Regenerate.
1221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1222 Alan Hayward <alan.hayward@arm.com>
1223 David Sherwood <david.sherwood@arm.com>
1225 * coretypes.h (opt_scalar_mode): New typedef.
1226 * gdbhooks.py (build_pretty_printers): Handle it.
1227 * machmode.h (mode_iterator::get_2xwider): Add overload for
1229 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
1231 * expr.c (convert_mode_scalar): Likewise.
1232 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1233 * optabs.c (expand_float): Likewise.
1234 (expand_fix): Likewise.
1235 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1238 Alan Hayward <alan.hayward@arm.com>
1239 David Sherwood <david.sherwood@arm.com>
1241 * optabs.c (expand_float): Explicitly check for scalars before
1242 using a branching expansion.
1243 (expand_fix): Likewise.
1245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1246 Alan Hayward <alan.hayward@arm.com>
1247 David Sherwood <david.sherwood@arm.com>
1249 * expr.c (convert_mode): Split scalar handling out into...
1250 (convert_mode_scalar): ...this new function. Treat the modes
1253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1254 Alan Hayward <alan.hayward@arm.com>
1255 David Sherwood <david.sherwood@arm.com>
1257 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
1259 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
1261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1262 Alan Hayward <alan.hayward@arm.com>
1263 David Sherwood <david.sherwood@arm.com>
1265 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
1266 rather than a machine_mode.
1267 (fixed_from_string): Likewise.
1268 (fixed_convert): Likewise.
1269 (fixed_convert_from_int): Likewise.
1270 (fixed_convert_from_real): Likewise.
1271 (real_convert_from_fixed): Likewise.
1272 * fixed-value.c (fixed_from_double_int): Likewise.
1273 (fixed_from_string): Likewise.
1274 (fixed_convert): Likewise.
1275 (fixed_convert_from_int): Likewise.
1276 (fixed_convert_from_real): Likewise.
1277 (real_convert_from_fixed): Likewise.
1278 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
1280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1281 Alan Hayward <alan.hayward@arm.com>
1282 David Sherwood <david.sherwood@arm.com>
1284 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
1285 of separate mode class checks. Do not allow vector modes here.
1286 (immed_wide_int_const): Use as_a <scalar_mode>.
1287 * explow.c (trunc_int_for_mode): Likewise.
1288 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
1289 (wi::shwi): Likewise.
1290 (wi::min_value): Likewise.
1291 (wi::max_value): Likewise.
1292 * dwarf2out.c (loc_descriptor): Likewise.
1293 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
1296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1297 Alan Hayward <alan.hayward@arm.com>
1298 David Sherwood <david.sherwood@arm.com>
1300 * tree.h (SCALAR_TYPE_MODE): New macro.
1301 * expr.c (expand_expr_addr_expr_1): Use it.
1302 (expand_expr_real_2): Likewise.
1303 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
1304 (fold_convert_const_fixed_from_int): Likewise.
1305 (fold_convert_const_fixed_from_real): Likewise.
1306 (native_encode_fixed): Likewise
1307 (native_encode_complex): Likewise
1308 (native_encode_vector): Likewise.
1309 (native_interpret_fixed): Likewise.
1310 (native_interpret_real): Likewise.
1311 (native_interpret_complex): Likewise.
1312 (native_interpret_vector): Likewise.
1313 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
1314 (simd_clone_adjust_argument_types): Likewise.
1315 (simd_clone_init_simd_arrays): Likewise.
1316 (simd_clone_adjust): Likewise.
1317 * stor-layout.c (layout_type): Likewise.
1318 * tree.c (build_minus_one_cst): Likewise.
1319 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1320 * tree-inline.c (estimate_move_cost): Likewise.
1321 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
1322 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
1323 (vectorizable_reduction): Likewise.
1324 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
1325 (vect_recog_mixed_size_cond_pattern): Likewise.
1326 (check_bool_pattern): Likewise.
1327 (adjust_bool_pattern): Likewise.
1328 (search_type_for_mask_1): Likewise.
1329 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1330 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1331 (vectorizable_load): Likewise.
1332 (vectorizable_store): Likewise.
1333 * ubsan.c (ubsan_encode_value): Likewise.
1334 * varasm.c (output_constant): Likewise.
1336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1337 Alan Hayward <alan.hayward@arm.com>
1338 David Sherwood <david.sherwood@arm.com>
1340 * coretypes.h (scalar_mode): New class.
1341 * machmode.h (scalar_mode): Likewise.
1342 (scalar_mode::includes_p): New function.
1343 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
1344 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
1345 * genmodes.c (get_mode_class): Handle remaining scalar modes.
1346 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
1347 * expmed.c (store_bit_field_1): Likewise.
1348 (extract_bit_field_1): Likewise.
1349 * expr.c (write_complex_part): Likewise.
1350 (read_complex_part): Likewise.
1351 (emit_move_complex_push): Likewise.
1352 (expand_expr_real_2): Likewise.
1353 * function.c (assign_parm_setup_reg): Likewise.
1354 (assign_parms_unsplit_complex): Likewise.
1355 * optabs.c (expand_binop): Likewise.
1356 * rtlanal.c (subreg_get_info): Likewise.
1357 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1358 * varasm.c (output_constant_pool_2): Likewise.
1360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1361 Alan Hayward <alan.hayward@arm.com>
1362 David Sherwood <david.sherwood@arm.com>
1364 * expmed.c (extract_high_half): Use scalar_int_mode and remove
1366 (expmed_mult_highpart_optab): Likewise.
1367 (expmed_mult_highpart): Likewise.
1369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1370 Alan Hayward <alan.hayward@arm.com>
1371 David Sherwood <david.sherwood@arm.com>
1373 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
1374 instead of a machine_mode.
1375 (builtin_memset_read_str): Likewise.
1376 * builtins.c (c_readstr): Likewise.
1377 (builtin_memcpy_read_str): Likewise.
1378 (builtin_strncpy_read_str): Likewise.
1379 (builtin_memset_read_str): Likewise.
1380 (builtin_memset_gen_str): Likewise.
1381 (expand_builtin_signbit): Use scalar_int_mode for local variables.
1382 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
1383 instead of a machine_mode.
1384 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
1386 (make_extraction): Likewise.
1387 (try_widen_shift_mode): Take and return scalar_int_modes instead
1389 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
1390 a scalar_int_mode instead of a machine_mode.
1391 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
1392 (avr_addr_space_pointer_mode): Likewise.
1393 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
1394 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
1395 (msp430_unwind_word_mode): Likewise.
1396 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
1397 (spu_addr_space_pointer_mode): Likewise.
1398 (spu_addr_space_address_mode): Likewise.
1399 (spu_libgcc_cmp_return_mode): Likewise.
1400 (spu_libgcc_shift_count_mode): Likewise.
1401 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
1402 (rl78_addr_space_pointer_mode): Likewise.
1403 (fl78_unwind_word_mode): Likewise.
1404 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
1406 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
1407 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
1408 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
1409 (mips_valid_pointer_mode): Likewise.
1410 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
1411 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
1412 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
1414 (ft32_addr_space_address_mode): Likewise.
1415 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
1416 scalar_int_mode instead of a machine_mode.
1417 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
1419 (m32c_addr_space_address_mode): Likewise.
1420 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
1421 (rs6000_eh_return_filter_mode): Likewise.
1422 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
1423 (rs6000_eh_return_filter_mode): Likewise.
1424 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
1425 (s390_libgcc_shift_count_mode): Likewise.
1426 (s390_unwind_word_mode): Likewise.
1427 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
1429 * target.def (mode_rep_extended): Likewise.
1430 (valid_pointer_mode): Likewise.
1431 (addr_space.valid_pointer_mode): Likewise.
1432 (eh_return_filter_mode): Return a scalar_int_mode rather than
1434 (libgcc_cmp_return_mode): Likewise.
1435 (libgcc_shift_count_mode): Likewise.
1436 (unwind_word_mode): Likewise.
1437 (addr_space.pointer_mode): Likewise.
1438 (addr_space.address_mode): Likewise.
1439 * doc/tm.texi: Regenerate.
1440 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
1442 (do_jump): Use scalar_int_mode for local variables.
1443 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
1444 rather than a machine_mode.
1445 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
1446 (scompare_loc_descriptor_wide): Likewise.
1447 (scompare_loc_descriptor_narrow): Likewise.
1448 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
1450 * except.c (sjlj_emit_dispatch_table): Likewise.
1451 (expand_builtin_eh_copy_values): Likewise.
1452 * explow.c (convert_memory_address_addr_space_1): Likewise.
1453 Take a scalar_int_mode rather than a machine_mode.
1454 (convert_memory_address_addr_space): Take a scalar_int_mode rather
1455 than a machine_mode.
1456 (memory_address_addr_space): Use scalar_int_mode for local variables.
1457 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
1458 rather than a machine_mode.
1459 * expmed.c (mask_rtx): Likewise.
1460 (init_expmed_one_conv): Likewise.
1461 (expand_mult_highpart_adjust): Likewise.
1462 (extract_high_half): Likewise.
1463 (expmed_mult_highpart_optab): Likewise.
1464 (expmed_mult_highpart): Likewise.
1465 (expand_smod_pow2): Likewise.
1466 (expand_sdiv_pow2): Likewise.
1467 (emit_store_flag_int): Likewise.
1468 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
1470 (extract_low_bits): Likewise.
1471 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
1473 * expr.c (pieces_addr::adjust): Likewise.
1474 (can_store_by_pieces): Likewise.
1475 (store_by_pieces): Likewise.
1476 (clear_by_pieces_1): Likewise.
1477 (expand_expr_addr_expr_1): Likewise.
1478 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
1479 (expand_expr_real_1): Likewise.
1480 (try_casesi): Likewise.
1481 * final.c (shorten_branches): Likewise.
1482 * fold-const.c (fold_convert_const_int_from_fixed): Change the
1483 type of "mode" to machine_mode.
1484 * internal-fn.c (expand_arith_overflow_result_store): Take a
1485 scalar_int_mode rather than a machine_mode.
1486 (expand_mul_overflow): Use scalar_int_mode for local variables.
1487 * loop-doloop.c (doloop_modify): Likewise.
1488 (doloop_optimize): Likewise.
1489 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
1490 than a machine_mode.
1491 (expand_doubleword_shift_condmove): Likewise.
1492 (expand_doubleword_shift): Likewise.
1493 (expand_doubleword_clz): Likewise.
1494 (expand_doubleword_popcount): Likewise.
1495 (expand_doubleword_parity): Likewise.
1496 (expand_absneg_bit): Use scalar_int_mode for local variables.
1497 (prepare_float_lib_cmp): Likewise.
1498 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
1499 rather than a machine_mode.
1500 (convert_memory_address_addr_space): Likewise.
1501 (get_mode_bounds): Likewise.
1502 (get_address_mode): Return a scalar_int_mode rather than a
1504 * rtlanal.c (get_address_mode): Likewise.
1505 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
1506 than a machine_mode.
1507 * targhooks.c (default_mode_rep_extended): Likewise.
1508 (default_valid_pointer_mode): Likewise.
1509 (default_addr_space_valid_pointer_mode): Likewise.
1510 (default_eh_return_filter_mode): Return a scalar_int_mode rather
1511 than a machine_mode.
1512 (default_libgcc_cmp_return_mode): Likewise.
1513 (default_libgcc_shift_count_mode): Likewise.
1514 (default_unwind_word_mode): Likewise.
1515 (default_addr_space_pointer_mode): Likewise.
1516 (default_addr_space_address_mode): Likewise.
1517 * targhooks.h (default_eh_return_filter_mode): Likewise.
1518 (default_libgcc_cmp_return_mode): Likewise.
1519 (default_libgcc_shift_count_mode): Likewise.
1520 (default_unwind_word_mode): Likewise.
1521 (default_addr_space_pointer_mode): Likewise.
1522 (default_addr_space_address_mode): Likewise.
1523 (default_mode_rep_extended): Take a scalar_int_mode rather than
1525 (default_valid_pointer_mode): Likewise.
1526 (default_addr_space_valid_pointer_mode): Likewise.
1527 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
1529 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
1530 rather than a machine_mode.
1531 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
1532 for local variables.
1533 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1534 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
1535 than a machine_mode.
1537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1538 Alan Hayward <alan.hayward@arm.com>
1539 David Sherwood <david.sherwood@arm.com>
1541 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
1542 the mode argument to scalar_int_mode.
1543 (do_jump_by_parts_zero_rtx): Likewise.
1544 (do_jump_by_parts_equality_rtx): Likewise.
1545 (do_jump_by_parts_greater): Take a mode argument.
1546 (do_jump_by_parts_equality): Likewise.
1547 (do_jump_1): Update calls accordingly.
1549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1550 Alan Hayward <alan.hayward@arm.com>
1551 David Sherwood <david.sherwood@arm.com>
1553 * is-a.h (safe_dyn_cast): New function.
1554 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
1555 (jump_table_for_label): Likewise.
1556 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
1557 instead of an rtx_insn *.
1558 (shorten_branches): Use dyn_cast instead of LABEL_P and
1559 JUMP_TABLE_DATA_P. Use jump_table_for_label and
1560 rtx_jump_table_data::get_data_mode.
1561 (final_scan_insn): Likewise.
1563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1564 Alan Hayward <alan.hayward@arm.com>
1565 David Sherwood <david.sherwood@arm.com>
1567 * combine.c (try_combine): Use is_a <scalar_int_mode> when
1568 trying to combine a full-register integer set with a subreg
1571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1572 Alan Hayward <alan.hayward@arm.com>
1573 David Sherwood <david.sherwood@arm.com>
1575 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
1576 that is always either address_mode or pointer_mode.
1578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1579 Alan Hayward <alan.hayward@arm.com>
1580 David Sherwood <david.sherwood@arm.com>
1582 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
1583 when the two are known to be equal.
1585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1586 Alan Hayward <alan.hayward@arm.com>
1587 David Sherwood <david.sherwood@arm.com>
1589 * simplify-rtx.c (simplify_const_unary_operation): Use
1590 is_a <scalar_int_mode> instead of checking for a nonzero
1591 precision. Forcibly convert op_mode to a scalar_int_mode
1592 in that case. More clearly differentiate the operand and
1593 result modes and use the former when deciding what the value
1594 of a count-bits operation should be. Use is_int_mode instead
1595 of checking for a MODE_INT. Remove redundant check for whether
1596 this mode has a zero precision.
1598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1599 Alan Hayward <alan.hayward@arm.com>
1600 David Sherwood <david.sherwood@arm.com>
1602 * optabs.c (widen_leading): Change the type of the mode argument
1603 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
1604 (widen_bswap): Likewise.
1605 (expand_parity): Likewise.
1606 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
1607 (expand_ffs): Likewise.
1608 (epand_unop): Check for scalar integer modes before calling the
1611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1612 Alan Hayward <alan.hayward@arm.com>
1613 David Sherwood <david.sherwood@arm.com>
1615 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
1617 (expand_expr_real_1): Update call accordingly.
1619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1620 Alan Hayward <alan.hayward@arm.com>
1621 David Sherwood <david.sherwood@arm.com>
1623 * expmed.c (store_bit_field_using_insv): Add op0_mode and
1624 value_mode arguments. Use scalar_int_mode internally.
1625 (store_bit_field_1): Rename the new integer mode from imode
1626 to op0_mode and use it instead of GET_MODE (op0). Update calls
1627 to store_split_bit_field, store_bit_field_using_insv and
1628 store_fixed_bit_field.
1629 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
1630 Use scalar_int_mode internally. Use a bit count rather than a mode
1631 when calculating the largest bit size for get_best_mode.
1632 Update calls to store_split_bit_field and store_fixed_bit_field_1.
1633 (store_fixed_bit_field_1): Add mode and value_mode arguments.
1634 Remove assertion that OP0 has a scalar integer mode.
1635 (store_split_bit_field): Add op0_mode and value_mode arguments.
1636 Update calls to extract_fixed_bit_field.
1637 (extract_bit_field_using_extv): Add an op0_mode argument.
1638 Use scalar_int_mode internally.
1639 (extract_bit_field_1): Rename the new integer mode from imode to
1640 op0_mode and use it instead of GET_MODE (op0). Update calls to
1641 extract_split_bit_field, extract_bit_field_using_extv and
1642 extract_fixed_bit_field.
1643 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
1644 to extract_split_bit_field and extract_fixed_bit_field_1.
1645 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
1646 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
1648 (extract_split_bit_field): Add an op0_mode argument. Update call
1649 to extract_fixed_bit_field.
1651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1652 Alan Hayward <alan.hayward@arm.com>
1653 David Sherwood <david.sherwood@arm.com>
1655 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
1656 * explow.c (hard_function_value): Likewise.
1657 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
1658 convert_to_mode call outside the loop.
1659 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
1660 for the mode iterator. Require the mode specified by max_pieces
1662 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
1664 (copy_blkmode_to_reg): Likewise.
1665 (set_storage_via_setmem): Likewise.
1666 * optabs.c (prepare_cmp_insn): Likewise.
1667 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1668 * stor-layout.c (finish_bitfield_representative): Likewise.
1670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1671 Alan Hayward <alan.hayward@arm.com>
1672 David Sherwood <david.sherwood@arm.com>
1674 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
1675 * expr.c (convert_move): Use them.
1676 (convert_modes): Likewise.
1677 (store_expr_with_bounds): Likewise.
1679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1680 Alan Hayward <alan.hayward@arm.com>
1681 David Sherwood <david.sherwood@arm.com>
1683 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
1684 parameter for the mode of "x". Remove the "known_x", "known_mode"
1685 and "known_ret" arguments. Change the type of the mode argument
1687 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
1688 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
1689 (reg_num_sign_bit_copies_for_combine): Likewise.
1690 * rtlanal.c (nonzero_bits1): Likewise.
1691 (num_sign_bit_copies1): Likewise.
1692 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
1693 (reg_num_sign_bit_copies_general): Likewise.
1694 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
1695 (reg_nonzero_bits_general): Likewise.
1697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1698 Alan Hayward <alan.hayward@arm.com>
1699 David Sherwood <david.sherwood@arm.com>
1701 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
1702 than in subroutines. Return 1 for non-integer modes.
1703 (cached_num_sign_bit_copies): Change the type of the mode parameter
1705 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
1706 classes. Handle CONST_INT_P first and then check whether X also
1707 has a scalar integer mode. Check the same thing for inner registers
1708 of a SUBREG and for values that are being extended or truncated.
1710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1711 Alan Hayward <alan.hayward@arm.com>
1712 David Sherwood <david.sherwood@arm.com>
1714 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
1715 in subroutines. Return the mode mask for non-integer modes.
1716 (cached_nonzero_bits): Change the type of the mode parameter
1718 (nonzero_bits1): Likewise. Remove early exit for other mode
1719 classes. Handle CONST_INT_P first and then check whether X
1720 also has a scalar integer mode.
1722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1723 Alan Hayward <alan.hayward@arm.com>
1724 David Sherwood <david.sherwood@arm.com>
1726 * expr.c (widest_int_mode_for_size): Make the comment match the code.
1727 Return a scalar_int_mode and assert that the size is greater than
1729 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
1730 (op_by_pieces_d::op_by_pieces_d): Likewise.
1731 (op_by_pieces_d::run): Likewise.
1732 (can_store_by_pieces): Likewise.
1734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1735 Alan Hayward <alan.hayward@arm.com>
1736 David Sherwood <david.sherwood@arm.com>
1738 * combine.c (extract_left_shift): Add a mode argument and update
1740 (make_compound_operation_int): Change the type of the mode parameter
1741 to scalar_int_mode and update the call to extract_left_shift.
1743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1744 Alan Hayward <alan.hayward@arm.com>
1745 David Sherwood <david.sherwood@arm.com>
1747 * combine.c (simplify_and_const_int): Change the type of the mode
1748 parameter to scalar_int_mode.
1749 (simplify_and_const_int_1): Likewise. Update recursive call.
1751 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1752 Alan Hayward <alan.hayward@arm.com>
1753 David Sherwood <david.sherwood@arm.com>
1755 * combine.c (simplify_compare_const): Check that the mode is a
1756 scalar_int_mode (rather than VOIDmode) before testing its
1758 (simplify_comparison): Move COMPARISON_P handling out of the
1759 loop and restrict the latter part of the loop to scalar_int_modes.
1760 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
1761 and when considering SUBREG_REGs. Use is_int_mode instead of
1762 checking GET_MODE_CLASS against MODE_INT.
1764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1765 Alan Hayward <alan.hayward@arm.com>
1766 David Sherwood <david.sherwood@arm.com>
1768 * combine.c (try_widen_shift_mode): Move check for equal modes to...
1769 (simplify_shift_const_1): ...here. Use scalar_int_mode for
1770 shift_unit_mode and for modes involved in scalar shifts.
1772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1773 Alan Hayward <alan.hayward@arm.com>
1774 David Sherwood <david.sherwood@arm.com>
1776 * combine.c (force_int_to_mode): New function, split out from...
1777 (force_to_mode): ...here. Keep xmode up-to-date and use it
1778 instead of GET_MODE (x).
1780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1781 Alan Hayward <alan.hayward@arm.com>
1782 David Sherwood <david.sherwood@arm.com>
1784 * optabs-query.h (extraction_insn::struct_mode): Change type to
1785 opt_scalar_int_mode and update comment.
1786 (extraction_insn::field_mode): Change type to scalar_int_mode.
1787 (extraction_insn::pos_mode): Likewise.
1788 * combine.c (make_extraction): Update accordingly.
1789 * optabs-query.c (get_traditional_extraction_insn): Likewise.
1790 (get_optab_extraction_insn): Likewise.
1791 * recog.c (simplify_while_replacing): Likewise.
1792 * expmed.c (narrow_bit_field_mem): Change the type of the mode
1793 parameter to opt_scalar_int_mode.
1795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1796 Alan Hayward <alan.hayward@arm.com>
1797 David Sherwood <david.sherwood@arm.com>
1799 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
1800 to a scalar_int_mode instead of a machine_mode.
1801 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
1802 (get_best_mode): Return a boolean and use a pointer argument to store
1803 the selected mode. Replace the limit mode parameter with a bit limit.
1804 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
1805 for the values returned by bit_field_mode_iterator::next_mode.
1806 (store_bit_field): Update call to get_best_mode.
1807 (store_fixed_bit_field): Likewise.
1808 (extract_fixed_bit_field): Likewise.
1809 * expr.c (optimize_bitfield_assignment_op): Likewise.
1810 * fold-const.c (optimize_bit_field_compare): Likewise.
1811 (fold_truth_andor_1): Likewise.
1812 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
1813 Update for new type of m_mode.
1814 (get_best_mode): As above.
1816 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1817 Alan Hayward <alan.hayward@arm.com>
1818 David Sherwood <david.sherwood@arm.com>
1820 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
1821 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
1822 (store_bit_field): Check is_a <scalar_int_mode> before calling
1823 strict_volatile_bitfield_p.
1824 (extract_bit_field): Likewise.
1826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1827 Alan Hayward <alan.hayward@arm.com>
1828 David Sherwood <david.sherwood@arm.com>
1830 * target.def (cstore_mode): Return a scalar_int_mode.
1831 * doc/tm.texi: Regenerate.
1832 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
1833 * targhooks.h (default_cstore_mode): Likewise.
1834 * targhooks.c (default_cstore_mode): Likewise, using a forced
1836 * expmed.c (emit_cstore): Expect the target of the cstore to be
1839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1840 Alan Hayward <alan.hayward@arm.com>
1841 David Sherwood <david.sherwood@arm.com>
1843 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
1845 (niter_desc): Likewise mode.
1846 (iv_analyze): Add a mode parameter.
1848 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
1849 and change its type to scalar_int_mode.
1850 * loop-iv.c: Update commentary at head of file.
1851 (iv_constant): Pass the mode paraeter before the rtx it describes
1852 and change its type to scalar_int_mode. Remove VOIDmode handling.
1853 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
1854 (iv_extend): Likewise.
1855 (shorten_into_mode): Likewise.
1856 (iv_add): Use scalar_int_mode.
1857 (iv_mult): Likewise.
1858 (iv_shift): Likewise.
1859 (canonicalize_iv_subregs): Likewise.
1860 (get_biv_step_1): Pass the outer_mode parameter before the rtx
1861 it describes and change its mode to scalar_int_mode. Also change
1862 the type of the returned inner_mode to scalar_int_mode.
1863 (get_biv_step): Likewise, turning outer_mode from a pointer
1864 into a direct parameter. Update call to get_biv_step_1.
1865 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
1866 iv_constant and get_biv_step.
1867 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
1868 and change its type to scalar_int_mode. Don't initialise iv->mode
1869 to VOIDmode and remove later checks for its still being VOIDmode.
1870 Update calls to iv_analyze_op and iv_analyze_expr. Check
1871 is_a <scalar_int_mode> when changing the mode under consideration.
1872 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
1873 Update call to iv_analyze_expr.
1874 (iv_analyze_op): Add a mode parameter. Reject subregs whose
1875 inner register is not also a scalar_int_mode. Update call to
1877 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
1878 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
1879 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
1880 separate mode class checks. Update calls to iv_analyze. Remove
1881 fix-up of VOIDmodes after iv_analyze_biv.
1882 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
1883 don't have a scalar_int_mode. Update call to biv_p.
1885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1886 Alan Hayward <alan.hayward@arm.com>
1887 David Sherwood <david.sherwood@arm.com>
1889 * cfgexpand.c (convert_debug_memory_address): Use
1890 as_a <scalar_int_mode>.
1891 * combine.c (expand_compound_operation): Likewise.
1892 (make_extraction): Likewise.
1893 (change_zero_ext): Likewise.
1894 (simplify_comparison): Likewise.
1895 * cse.c (cse_insn): Likewise.
1896 * dwarf2out.c (minmax_loc_descriptor): Likewise.
1897 (mem_loc_descriptor): Likewise.
1898 (loc_descriptor): Likewise.
1899 * expmed.c (init_expmed_one_mode): Likewise.
1900 (synth_mult): Likewise.
1901 (emit_store_flag_1): Likewise.
1902 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
1903 of a comparison with size.
1904 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
1905 (reduce_to_bit_field_precision): Likewise.
1906 * function.c (expand_function_end): Likewise.
1907 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
1908 * loop-doloop.c (doloop_modify): Likewise.
1909 * optabs.c (expand_binop): Likewise.
1910 (expand_unop): Likewise.
1911 (expand_copysign_absneg): Likewise.
1912 (prepare_cmp_insn): Likewise.
1913 (maybe_legitimize_operand): Likewise.
1914 * recog.c (const_scalar_int_operand): Likewise.
1915 * rtlanal.c (get_address_mode): Likewise.
1916 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1917 (simplify_cond_clz_ctz): Likewise.
1918 * tree-nested.c (get_nl_goto_field): Likewise.
1919 * tree.c (build_vector_type_for_mode): Likewise.
1920 * var-tracking.c (use_narrower_mode): Likewise.
1922 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1923 Alan Hayward <alan.hayward@arm.com>
1924 David Sherwood <david.sherwood@arm.com>
1926 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
1927 * builtins.c (expand_builtin_signbit): Use it.
1928 * cfgexpand.c (expand_debug_expr): Likewise.
1929 * dojump.c (do_jump): Likewise.
1930 (do_compare_and_jump): Likewise.
1931 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
1932 * expmed.c (make_tree): Likewise.
1933 * expr.c (expand_expr_real_2): Likewise.
1934 (expand_expr_real_1): Likewise.
1935 (try_casesi): Likewise.
1936 * fold-const-call.c (fold_const_call_ss): Likewise.
1937 * fold-const.c (unextend): Likewise.
1938 (extract_muldiv_1): Likewise.
1939 (fold_single_bit_test): Likewise.
1940 (native_encode_int): Likewise.
1941 (native_encode_string): Likewise.
1942 (native_interpret_int): Likewise.
1943 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
1944 * internal-fn.c (expand_addsub_overflow): Likewise.
1945 (expand_neg_overflow): Likewise.
1946 (expand_mul_overflow): Likewise.
1947 (expand_arith_overflow): Likewise.
1948 * match.pd: Likewise.
1949 * stor-layout.c (layout_type): Likewise.
1950 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1951 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
1952 * tree-ssanames.c (get_range_info): Likewise.
1953 * tree-switch-conversion.c (array_value_type) Likewise.
1954 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
1955 (vect_recog_divmod_pattern): Likewise.
1956 (vect_recog_mixed_size_cond_pattern): Likewise.
1957 * tree-vrp.c (extract_range_basic): Likewise.
1958 (simplify_float_conversion_using_ranges): Likewise.
1959 * tree.c (int_fits_type_p): Likewise.
1960 * ubsan.c (instrument_bool_enum_load): Likewise.
1961 * varasm.c (mergeable_string_section): Likewise.
1962 (narrowing_initializer_constant_valid_p): Likewise.
1963 (output_constant): Likewise.
1965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1966 Alan Hayward <alan.hayward@arm.com>
1967 David Sherwood <david.sherwood@arm.com>
1969 * machmode.h (NARROWEST_INT_MODE): New macro.
1970 * expr.c (alignment_for_piecewise_move): Use it instead of
1971 GET_CLASS_NARROWEST_MODE (MODE_INT).
1972 (push_block): Likewise.
1973 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1975 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1978 Alan Hayward <alan.hayward@arm.com>
1979 David Sherwood <david.sherwood@arm.com>
1981 * postreload.c (move2add_valid_value_p): Change the type of the
1982 mode parameter to scalar_int_mode.
1983 (move2add_use_add2_insn): Add a mode parameter and use it instead
1985 (move2add_use_add3_insn): Likewise.
1986 (reload_cse_move2add): Update accordingly.
1988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1989 Alan Hayward <alan.hayward@arm.com>
1990 David Sherwood <david.sherwood@arm.com>
1992 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
1994 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
1995 * optabs.c (expand_unop): Likewise.
1997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1998 Alan Hayward <alan.hayward@arm.com>
1999 David Sherwood <david.sherwood@arm.com>
2001 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2002 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2003 (popcount_loc_descriptor): Likewise.
2004 (bswap_loc_descriptor): Likewise.
2005 (rotate_loc_descriptor): Likewise.
2006 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2007 calling the functions above.
2009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2010 Alan Hayward <alan.hayward@arm.com>
2011 David Sherwood <david.sherwood@arm.com>
2013 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2015 (try_combine): Likewise.
2016 (simplify_if_then_else): Likewise.
2017 * cse.c (cse_insn): Likewise.
2018 * dwarf2out.c (mem_loc_descriptor): Likewise.
2019 * emit-rtl.c (gen_lowpart_common): Likewise.
2020 * simplify-rtx.c (simplify_truncation): Likewise.
2021 (simplify_binary_operation_1): Likewise.
2022 (simplify_const_relational_operation): Likewise.
2023 (simplify_ternary_operation): Likewise.
2024 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
2026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2027 Alan Hayward <alan.hayward@arm.com>
2028 David Sherwood <david.sherwood@arm.com>
2030 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2031 * reload.c (push_reload): Likewise.
2032 (find_reloads): Likewise.
2034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2035 Alan Hayward <alan.hayward@arm.com>
2036 David Sherwood <david.sherwood@arm.com>
2038 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2039 (make_compound_operation_int): Likewise.
2040 (change_zero_ext): Likewise.
2041 * expr.c (convert_move): Likewise.
2042 (convert_modes): Likewise.
2043 * fwprop.c (forward_propagate_subreg): Likewise.
2044 * loop-iv.c (get_biv_step_1): Likewise.
2045 * optabs.c (widen_operand): Likewise.
2046 * postreload.c (move2add_valid_value_p): Likewise.
2047 * recog.c (simplify_while_replacing): Likewise.
2048 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2049 (simplify_binary_operation_1): Likewise. Remove redundant
2050 mode equality check.
2052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2053 Alan Hayward <alan.hayward@arm.com>
2054 David Sherwood <david.sherwood@arm.com>
2056 * combine.c (combine_simplify_rtx): Add checks for
2057 is_a <scalar_int_mode>.
2058 (simplify_if_then_else): Likewise.
2059 (make_field_assignment): Likewise.
2060 (simplify_comparison): Likewise.
2061 * ifcvt.c (noce_try_bitop): Likewise.
2062 * loop-invariant.c (canonicalize_address_mult): Likewise.
2063 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2066 Alan Hayward <alan.hayward@arm.com>
2067 David Sherwood <david.sherwood@arm.com>
2069 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2070 is_a <scalar_int_mode> instead of != BLKmode.
2072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2073 Alan Hayward <alan.hayward@arm.com>
2074 David Sherwood <david.sherwood@arm.com>
2076 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2077 instead of != VOIDmode.
2078 * combine.c (if_then_else_cond): Likewise.
2079 (change_zero_ext): Likewise.
2080 * dwarf2out.c (mem_loc_descriptor): Likewise.
2081 (loc_descriptor): Likewise.
2082 * rtlanal.c (canonicalize_condition): Likewise.
2083 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
2085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2086 Alan Hayward <alan.hayward@arm.com>
2087 David Sherwood <david.sherwood@arm.com>
2089 * simplify-rtx.c (simplify_binary_operation_1): Use
2090 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2093 Alan Hayward <alan.hayward@arm.com>
2094 David Sherwood <david.sherwood@arm.com>
2096 * wide-int.h (int_traits<unsigned char>) New class.
2097 (int_traits<unsigned short>) Likewise.
2098 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2099 Use GET_MODE_UNIT_PRECISION and remove redundant test for
2101 * combine.c (set_nonzero_bits_and_sign_copies): Use
2102 is_a <scalar_int_mode>.
2103 (find_split_point): Likewise.
2104 (combine_simplify_rtx): Likewise.
2105 (simplify_logical): Likewise.
2106 (expand_compound_operation): Likewise.
2107 (expand_field_assignment): Likewise.
2108 (make_compound_operation): Likewise.
2109 (extended_count): Likewise.
2110 (change_zero_ext): Likewise.
2111 (simplify_comparison): Likewise.
2112 * dwarf2out.c (scompare_loc_descriptor): Likewise.
2113 (ucompare_loc_descriptor): Likewise.
2114 (minmax_loc_descriptor): Likewise.
2115 (mem_loc_descriptor): Likewise.
2116 (loc_descriptor): Likewise.
2117 * expmed.c (init_expmed_one_mode): Likewise.
2118 * lra-constraints.c (lra_constraint_offset): Likewise.
2119 * optabs.c (prepare_libcall_arg): Likewise.
2120 * postreload.c (move2add_note_store): Likewise.
2121 * reload.c (operands_match_p): Likewise.
2122 * rtl.h (load_extend_op): Likewise.
2123 * rtlhooks.c (gen_lowpart_general): Likewise.
2124 * simplify-rtx.c (simplify_truncation): Likewise.
2125 (simplify_unary_operation_1): Likewise.
2126 (simplify_binary_operation_1): Likewise.
2127 (simplify_const_binary_operation): Likewise.
2128 (simplify_const_relational_operation): Likewise.
2129 (simplify_subreg): Likewise.
2130 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2131 * var-tracking.c (adjust_mems): Likewise.
2132 (prepare_call_arguments): Likewise.
2134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2135 Alan Hayward <alan.hayward@arm.com>
2136 David Sherwood <david.sherwood@arm.com>
2138 * machmode.h (is_int_mode): New fuction.
2139 * combine.c (find_split_point): Use it.
2140 (combine_simplify_rtx): Likewise.
2141 (simplify_if_then_else): Likewise.
2142 (simplify_set): Likewise.
2143 (simplify_shift_const_1): Likewise.
2144 (simplify_comparison): Likewise.
2145 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
2146 * cse.c (notreg_cost): Likewise.
2147 (cse_insn): Likewise.
2148 * cselib.c (cselib_lookup_1): Likewise.
2149 * dojump.c (do_jump_1): Likewise.
2150 (do_compare_rtx_and_jump): Likewise.
2151 * dse.c (get_call_args): Likewise.
2152 * dwarf2out.c (rtl_for_decl_init): Likewise.
2153 (native_encode_initializer): Likewise.
2154 * expmed.c (emit_store_flag_1): Likewise.
2155 (emit_store_flag): Likewise.
2156 * expr.c (convert_modes): Likewise.
2157 (store_field): Likewise.
2158 (expand_expr_real_1): Likewise.
2159 * fold-const.c (fold_read_from_constant_string): Likewise.
2160 * gimple-ssa-sprintf.c (get_format_string): Likewise.
2161 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
2162 * optabs.c (expand_binop): Likewise.
2163 (expand_unop): Likewise.
2164 (expand_abs_nojump): Likewise.
2165 (expand_one_cmpl_abs_nojump): Likewise.
2166 * simplify-rtx.c (mode_signbit_p): Likewise.
2167 (val_signbit_p): Likewise.
2168 (val_signbit_known_set_p): Likewise.
2169 (val_signbit_known_clear_p): Likewise.
2170 (simplify_relational_operation_1): Likewise.
2171 * tree.c (vector_type_mode): Likewise.
2173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2174 Alan Hayward <alan.hayward@arm.com>
2175 David Sherwood <david.sherwood@arm.com>
2177 * machmode.h (smallest_mode_for_size): Fix formatting.
2178 (smallest_int_mode_for_size): New function.
2179 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
2180 instead of smallest_mode_for_size.
2181 * combine.c (make_extraction): Likewise.
2182 * config/arc/arc.c (arc_expand_movmem): Likewise.
2183 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
2184 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2185 * config/s390/s390.c (s390_expand_insv): Likewise.
2186 * config/sparc/sparc.c (assign_int_registers): Likewise.
2187 * config/spu/spu.c (spu_function_value): Likewise.
2188 (spu_function_arg): Likewise.
2189 * coverage.c (get_gcov_type): Likewise.
2190 (get_gcov_unsigned_t): Likewise.
2191 * dse.c (find_shift_sequence): Likewise.
2192 * expmed.c (store_bit_field_1): Likewise.
2193 * expr.c (convert_move): Likewise.
2194 (store_field): Likewise.
2195 * internal-fn.c (expand_arith_overflow): Likewise.
2196 * optabs-query.c (get_best_extraction_insn): Likewise.
2197 * optabs.c (expand_twoval_binop_libfunc): Likewise.
2198 * stor-layout.c (layout_type): Likewise.
2199 (initialize_sizetypes): Likewise.
2200 * targhooks.c (default_get_mask_mode): Likewise.
2201 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
2203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2204 Alan Hayward <alan.hayward@arm.com>
2205 David Sherwood <david.sherwood@arm.com>
2207 * machmode.h (opt_mode::else_blk): New function.
2208 (int_mode_for_mode): Declare.
2209 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
2210 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
2212 * cfgexpand.c (expand_debug_expr): Likewise.
2213 * combine.c (gen_lowpart_or_truncate): Likewise.
2214 (gen_lowpart_for_combine): Likewise.
2215 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
2216 * config/avr/avr.c (avr_to_int_mode): Likewise.
2217 (avr_out_plus_1): Likewise.
2218 (avr_out_plus): Likewise.
2219 (avr_out_round): Likewise.
2220 * config/i386/i386.c (ix86_split_to_parts): Likewise.
2221 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2222 (s390_expand_vcond): Likewise.
2223 * config/spu/spu.c (spu_split_immediate): Likewise.
2224 (spu_expand_mov): Likewise.
2225 * dse.c (get_stored_val): Likewise.
2226 * expmed.c (store_bit_field_1): Likewise.
2227 (convert_extracted_bit_field): Use int_mode_for_mode instead of
2229 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
2230 (extract_low_bits): Likewise.
2231 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
2232 handling rather than repeating the check.
2233 (emit_group_store): Likewise.
2234 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
2235 * optabs.c (expand_absneg_bit): Likewise.
2236 (expand_copysign_absneg): Likewise.
2237 (expand_copysign_bit): Likewise.
2238 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2239 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
2240 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
2241 * var-tracking.c (prepare_call_arguments): Likewise.
2242 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
2243 int_mode_for_mode instead of mode_for_size.
2244 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2247 Alan Hayward <alan.hayward@arm.com>
2248 David Sherwood <david.sherwood@arm.com>
2250 * machmode.h (int_mode_for_size): New function.
2251 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
2252 instead of mode_for_size.
2253 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
2255 * combine.c (expand_field_assignment): Use int_mode_for_size
2256 instead of mode_for_size.
2257 (make_extraction): Likewise.
2258 (simplify_shift_const_1): Likewise.
2259 (simplify_comparison): Likewise.
2260 * dojump.c (do_jump): Likewise.
2261 * dwarf2out.c (mem_loc_descriptor): Likewise.
2262 * emit-rtl.c (init_derived_machine_modes): Likewise.
2263 * expmed.c (flip_storage_order): Likewise.
2264 (convert_extracted_bit_field): Likewise.
2265 * expr.c (copy_blkmode_from_reg): Likewise.
2266 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
2267 * internal-fn.c (expand_mul_overflow): Likewise.
2268 * lower-subreg.c (simple_move): Likewise.
2269 * optabs-libfuncs.c (init_optabs): Likewise.
2270 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2271 * tree.c (vector_type_mode): Likewise.
2272 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
2273 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
2274 * tree-vect-generic.c (expand_vector_parallel): Likewise.
2275 * tree-vect-stmts.c (vectorizable_load): Likewise.
2276 (vectorizable_store): Likewise.
2278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2279 Alan Hayward <alan.hayward@arm.com>
2280 David Sherwood <david.sherwood@arm.com>
2282 * coretypes.h (pod_mode): New type.
2283 (scalar_int_mode_pod): New typedef.
2284 * machmode.h (pod_mode): New class.
2285 (int_n_data_t::m): Change type to scalar_int_mode_pod.
2286 * genmodes.c (emit_mode_int_n): Update accordingly.
2287 * lower-subreg.h (target_lower_subreg): Change type to
2288 scalar_int_mode_pod.
2289 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
2290 scalar_int_mode_pod.
2292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2293 Alan Hayward <alan.hayward@arm.com>
2294 David Sherwood <david.sherwood@arm.com>
2296 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
2297 machine_mode to scalar_int_mode.
2298 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
2299 (rs6000_option_override_internal): Remove cast to int.
2300 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
2301 machine_mode to scalar_int_mode.
2302 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
2303 (rs6000_option_override_internal): Remove cast to int.
2304 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
2305 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
2307 * config/s390/s390.c (s390_expand_builtin): Likewise.
2308 * coretypes.h (scalar_int_mode): New type.
2309 (opt_scalar_int_mode): New typedef.
2310 * machmode.h (scalar_int_mode): New class.
2311 (scalar_int_mode::includes_p): New function.
2312 (byte_mode): Change type to scalar_int_mode.
2313 (word_mode): Likewise.
2314 (ptr_mode): Likewise.
2315 * emit-rtl.c (byte_mode): Likewise.
2316 (word_mode): Likewise.
2317 (ptr_mode): Likewise.
2318 (init_derived_machine_modes): Update accordingly.
2319 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
2320 and MODE_PARTIAL_INT.
2321 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
2322 opt_scalar_int_mode.
2324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2325 Alan Hayward <alan.hayward@arm.com>
2326 David Sherwood <david.sherwood@arm.com>
2328 * target.def (libgcc_floating_mode_supported_p): Take a
2330 * doc/tm.texi: Regenerate.
2331 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
2333 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2334 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
2337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2338 Alan Hayward <alan.hayward@arm.com>
2339 David Sherwood <david.sherwood@arm.com>
2341 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
2342 * doc/tm.texi: Regenerate.
2343 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
2344 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
2345 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
2346 * targhooks.h (default_floatn_mode): Likewise.
2347 * targhooks.c (default_floatn_mode): Likewise.
2348 * tree.c (build_common_tree_nodes): Update accordingly.
2350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2351 Alan Hayward <alan.hayward@arm.com>
2352 David Sherwood <david.sherwood@arm.com>
2354 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
2355 (mode_iterator::iterate_p): Likewise.
2356 (mode_iterator::get_wider): Likewise.
2357 * expr.c (init_expr_target): Use opt_scalar_float_mode.
2359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2360 Alan Hayward <alan.hayward@arm.com>
2361 David Sherwood <david.sherwood@arm.com>
2363 * coretypes.h (opt_scalar_float_mode): New typedef.
2364 * machmode.h (float_mode_for_size): New function.
2365 * emit-rtl.c (double_mode): Delete.
2366 (init_emit_once): Use float_mode_for_size.
2367 * stor-layout.c (layout_type): Likewise.
2368 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
2370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2371 Alan Hayward <alan.hayward@arm.com>
2372 David Sherwood <david.sherwood@arm.com>
2374 * output.h (assemble_real): Take a scalar_float_mode.
2375 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
2376 * config/arm/arm.md (consttable_4): Likewise.
2377 (consttable_8): Likewise.
2378 (consttable_16): Likewise.
2379 * config/mips/mips.md (consttable_float): Likewise.
2380 * config/s390/s390.c (s390_output_pool_entry): Likewise.
2381 * varasm.c (assemble_real): Take a scalar_float_mode.
2382 (output_constant_pool_2): Update accordingly.
2383 (output_constant): Likewise.
2385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2386 Alan Hayward <alan.hayward@arm.com>
2387 David Sherwood <david.sherwood@arm.com>
2389 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
2390 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
2391 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
2392 (native_encode_real): Likewise.
2393 (native_interpret_real): Likewise.
2394 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
2395 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2398 Alan Hayward <alan.hayward@arm.com>
2399 David Sherwood <david.sherwood@arm.com>
2401 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
2402 <scalar_float_mode>. Simplify.
2403 (gen_extend_conv_libfunc): Likewise.
2405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2406 Alan Hayward <alan.hayward@arm.com>
2407 David Sherwood <david.sherwood@arm.com>
2409 * coretypes.h (scalar_float_mode): New type.
2410 * machmode.h (mode_traits::from_int): Use machine_mode if
2411 USE_ENUM_MODES is defined.
2412 (is_a): New function.
2414 (dyn_cast): Likewise.
2415 (scalar_float_mode): New class.
2416 (scalar_float_mode::includes_p): New function.
2417 (is_float_mode): Likewise.
2418 * gdbhooks.py (MachineModePrinter): New class.
2419 (build_pretty_printer): Use it for scalar_float_mode.
2420 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
2421 (format_helper::format_helper): Turn into a template.
2422 * genmodes.c (get_mode_class): New function.
2423 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
2424 or machine_mode if none.
2425 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
2426 as_a <scalar_float_mode>.
2427 * dwarf2out.c (mem_loc_descriptor): Likewise.
2428 (insert_float): Likewise.
2429 (add_const_value_attribute): Likewise.
2430 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2431 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
2432 (expand_unop): Update accordingly.
2433 (expand_abs_nojump): Likewise.
2434 (expand_copysign_absneg): Take a scalar_float_mode.
2435 (expand_copysign_bit): Likewise.
2436 (expand_copysign): Update accordingly.
2438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2439 Alan Hayward <alan.hayward@arm.com>
2440 David Sherwood <david.sherwood@arm.com>
2442 * coretypes.h (opt_mode): New class.
2443 * machmode.h (opt_mode): Likewise.
2444 (opt_mode::else_void): New function.
2445 (opt_mode::require): Likewise.
2446 (opt_mode::exists): Likewise.
2447 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
2448 (GET_MODE_2XWIDER_MODE): Likewise.
2449 (mode_iterator::get_wider): Update accordingly.
2450 (mode_iterator::get_2xwider): Likewise.
2451 (mode_iterator::get_known_wider): Likewise, turning into a template.
2452 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
2453 forcing a wider mode to exist.
2454 * config/cr16/cr16.h (LONG_REG_P): Likewise.
2455 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2456 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
2457 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
2458 * lower-subreg.c (init_lower_subreg): Likewise.
2459 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
2460 on the final iteration.
2461 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
2462 a wider mode exists before asking for a move pattern.
2463 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
2464 forcing a wider mode to exist.
2465 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
2466 returning false if no such mode exists.
2467 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
2468 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
2469 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
2470 Avoid checking for a MODE_INT if we already know the mode is not a
2472 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
2473 forcing a wider mode to exist.
2474 (expmed_mult_highpart_optab): Likewise.
2475 (expmed_mult_highpart): Likewise.
2476 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
2478 * lto-streamer-in.c (lto_input_mode_table): Likewise.
2479 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
2480 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2481 * internal-fn.c (expand_mul_overflow): Update use of
2482 GET_MODE_2XWIDER_MODE.
2483 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2484 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
2485 GET_MODE_WIDER_MODE.
2486 (convert_plusminus_to_widen): Likewise.
2487 * tree-switch-conversion.c (array_value_type): Likewise.
2488 * var-tracking.c (emit_note_insn_var_location): Likewise.
2489 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2490 Return false inside rather than outside the loop if no wider mode
2492 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
2493 and GET_MODE_2XWIDER_MODE
2494 (can_compare_p): Use else_void.
2495 * gdbhooks.py (OptMachineModePrinter): New class.
2496 (build_pretty_printer): Use it for opt_mode.
2498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2499 Alan Hayward <alan.hayward@arm.com>
2500 David Sherwood <david.sherwood@arm.com>
2502 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
2503 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
2505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2506 Alan Hayward <alan.hayward@arm.com>
2507 David Sherwood <david.sherwood@arm.com>
2509 * machmode.h (mode_traits): New structure.
2510 (get_narrowest_mode): New function.
2511 (mode_iterator::start): Likewise.
2512 (mode_iterator::iterate_p): Likewise.
2513 (mode_iterator::get_wider): Likewise.
2514 (mode_iterator::get_known_wider): Likewise.
2515 (mode_iterator::get_2xwider): Likewise.
2516 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
2517 (FOR_EACH_MODE): Likewise.
2518 (FOR_EACH_MODE_FROM): Likewise.
2519 (FOR_EACH_MODE_UNTIL): Likewise.
2520 (FOR_EACH_WIDER_MODE): Likewise.
2521 (FOR_EACH_2XWIDER_MODE): Likewise.
2522 * builtins.c (expand_builtin_strlen): Use new mode iterators.
2523 * combine.c (simplify_comparison): Likewise
2524 * config/i386/i386.c (type_natural_mode): Likewise.
2525 * cse.c (cse_insn): Likewise.
2526 * dse.c (find_shift_sequence): Likewise.
2527 * emit-rtl.c (init_derived_machine_modes): Likewise.
2528 (init_emit_once): Likewise.
2529 * explow.c (hard_function_value): Likewise.
2530 * expmed.c (extract_fixed_bit_field_1): Likewise.
2531 (extract_bit_field_1): Likewise.
2532 (expand_divmod): Likewise.
2533 (emit_store_flag_1): Likewise.
2534 * expr.c (init_expr_target): Likewise.
2535 (convert_move): Likewise.
2536 (alignment_for_piecewise_move): Likewise.
2537 (widest_int_mode_for_size): Likewise.
2538 (emit_block_move_via_movmem): Likewise.
2539 (copy_blkmode_to_reg): Likewise.
2540 (set_storage_via_setmem): Likewise.
2541 (compress_float_constant): Likewise.
2542 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2543 * optabs-query.c (get_best_extraction_insn): Likewise.
2544 * optabs.c (expand_binop): Likewise.
2545 (expand_twoval_unop): Likewise.
2546 (expand_twoval_binop): Likewise.
2547 (widen_leading): Likewise.
2548 (widen_bswap): Likewise.
2549 (expand_parity): Likewise.
2550 (expand_unop): Likewise.
2551 (prepare_cmp_insn): Likewise.
2552 (prepare_float_lib_cmp): Likewise.
2553 (expand_float): Likewise.
2554 (expand_fix): Likewise.
2555 (expand_sfix_optab): Likewise.
2556 * postreload.c (move2add_use_add2_insn): Likewise.
2557 * reg-stack.c (reg_to_stack): Likewise.
2558 * reginfo.c (choose_hard_reg_mode): Likewise.
2559 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2560 * stor-layout.c (mode_for_size): Likewise.
2561 (smallest_mode_for_size): Likewise.
2562 (mode_for_vector): Likewise.
2563 (finish_bitfield_representative): Likewise.
2564 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
2565 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
2566 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2567 * var-tracking.c (prepare_call_arguments): Likewise.
2569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2570 Alan Hayward <alan.hayward@arm.com>
2571 David Sherwood <david.sherwood@arm.com>
2573 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
2574 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
2575 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
2576 * machmode.h (mode_size): Move earlier in file.
2577 (mode_precision): Likewise.
2578 (mode_inner): Likewise.
2579 (mode_nunits): Likewise.
2580 (mode_unit_size): Likewise.
2581 (unit_unit_precision): Likewise.
2582 (mode_wider): Likewise.
2583 (mode_2xwider): Likewise.
2584 (machine_mode): New class.
2585 (mode_to_bytes): New function.
2586 (mode_to_bits): Likewise.
2587 (mode_to_precision): Likewise.
2588 (mode_to_inner): Likewise.
2589 (mode_to_unit_size): Likewise.
2590 (mode_to_unit_precision): Likewise.
2591 (mode_to_nunits): Likewise.
2592 (GET_MODE_SIZE): Use mode_to_bytes.
2593 (GET_MODE_BITSIZE): Use mode_to_bits.
2594 (GET_MODE_PRECISION): Use mode_to_precision.
2595 (GET_MODE_INNER): Use mode_to_inner.
2596 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
2597 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
2598 (GET_MODE_NUNITS): Use mode_to_nunits.
2599 * system.h (ALWAYS_INLINE): New macro.
2600 * config/powerpcspe/powerpcspe-c.c
2601 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
2602 int for arg1_mode and arg2_mode.
2604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2605 Alan Hayward <alan.hayward@arm.com>
2606 David Sherwood <david.sherwood@arm.com>
2608 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2609 Prefix mode names with E_ in case statements.
2610 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2611 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
2612 (aarch64_split_simd_move): Likewise.
2613 (aarch64_gen_storewb_pair): Likewise.
2614 (aarch64_gen_loadwb_pair): Likewise.
2615 (aarch64_gen_store_pair): Likewise.
2616 (aarch64_gen_load_pair): Likewise.
2617 (aarch64_get_condition_code_1): Likewise.
2618 (aarch64_constant_pool_reload_icode): Likewise.
2619 (get_rsqrte_type): Likewise.
2620 (get_rsqrts_type): Likewise.
2621 (get_recpe_type): Likewise.
2622 (get_recps_type): Likewise.
2623 (aarch64_gimplify_va_arg_expr): Likewise.
2624 (aarch64_simd_container_mode): Likewise.
2625 (aarch64_emit_load_exclusive): Likewise.
2626 (aarch64_emit_store_exclusive): Likewise.
2627 (aarch64_expand_compare_and_swap): Likewise.
2628 (aarch64_gen_atomic_cas): Likewise.
2629 (aarch64_emit_bic): Likewise.
2630 (aarch64_emit_atomic_swap): Likewise.
2631 (aarch64_emit_atomic_load_op): Likewise.
2632 (aarch64_evpc_trn): Likewise.
2633 (aarch64_evpc_uzp): Likewise.
2634 (aarch64_evpc_zip): Likewise.
2635 (aarch64_evpc_ext): Likewise.
2636 (aarch64_evpc_rev): Likewise.
2637 (aarch64_evpc_dup): Likewise.
2638 (aarch64_gen_ccmp_first): Likewise.
2639 (aarch64_gen_ccmp_next): Likewise.
2640 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2641 (alpha_emit_xfloating_libcall): Likewise.
2642 (emit_insxl): Likewise.
2643 (alpha_arg_type): Likewise.
2644 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
2645 (arc_preferred_simd_mode): Likewise.
2646 (arc_secondary_reload): Likewise.
2647 (get_arc_condition_code): Likewise.
2648 (arc_print_operand): Likewise.
2649 (arc_legitimate_constant_p): Likewise.
2650 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2651 * config/arc/arc.md (casesi_load): Likewise.
2652 (casesi_compact_jump): Likewise.
2653 * config/arc/predicates.md (proper_comparison_operator): Likewise.
2654 (cc_use_register): Likewise.
2655 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2656 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2657 (arm_init_iwmmxt_builtins): Likewise.
2658 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
2659 (neon_expand_vector_init): Likewise.
2660 (arm_attr_length_move_neon): Likewise.
2661 (maybe_get_arm_condition_code): Likewise.
2662 (arm_emit_vector_const): Likewise.
2663 (arm_preferred_simd_mode): Likewise.
2664 (arm_output_iwmmxt_tinsr): Likewise.
2665 (thumb1_output_casesi): Likewise.
2666 (thumb2_output_casesi): Likewise.
2667 (arm_emit_load_exclusive): Likewise.
2668 (arm_emit_store_exclusive): Likewise.
2669 (arm_expand_compare_and_swap): Likewise.
2670 (arm_evpc_neon_vuzp): Likewise.
2671 (arm_evpc_neon_vzip): Likewise.
2672 (arm_evpc_neon_vrev): Likewise.
2673 (arm_evpc_neon_vtrn): Likewise.
2674 (arm_evpc_neon_vext): Likewise.
2675 (arm_validize_comparison): Likewise.
2676 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
2677 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
2678 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
2679 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
2680 (c6x_preferred_simd_mode): Likewise.
2681 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
2682 (epiphany_rtx_costs): Likewise.
2683 * config/epiphany/predicates.md (proper_comparison_operator):
2685 * config/frv/frv.c (condexec_memory_operand): Likewise.
2686 (frv_emit_move): Likewise.
2687 (output_move_single): Likewise.
2688 (output_condmove_single): Likewise.
2689 (frv_hard_regno_mode_ok): Likewise.
2690 (frv_matching_accg_mode): Likewise.
2691 * config/h8300/h8300.c (split_adds_subs): Likewise.
2692 (h8300_rtx_costs): Likewise.
2693 (h8300_print_operand): Likewise.
2694 (compute_mov_length): Likewise.
2695 (output_logical_op): Likewise.
2696 (compute_logical_op_length): Likewise.
2697 (compute_logical_op_cc): Likewise.
2698 (h8300_shift_needs_scratch_p): Likewise.
2699 (output_a_shift): Likewise.
2700 (compute_a_shift_length): Likewise.
2701 (compute_a_shift_cc): Likewise.
2702 (expand_a_rotate): Likewise.
2703 (output_a_rotate): Likewise.
2704 * config/i386/i386.c (classify_argument): Likewise.
2705 (function_arg_advance_32): Likewise.
2706 (function_arg_32): Likewise.
2707 (function_arg_64): Likewise.
2708 (function_value_64): Likewise.
2709 (ix86_gimplify_va_arg): Likewise.
2710 (ix86_legitimate_constant_p): Likewise.
2711 (put_condition_code): Likewise.
2712 (split_double_mode): Likewise.
2713 (ix86_avx256_split_vector_move_misalign): Likewise.
2714 (ix86_expand_vector_logical_operator): Likewise.
2715 (ix86_split_idivmod): Likewise.
2716 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
2717 (ix86_build_const_vector): Likewise.
2718 (ix86_build_signbit_mask): Likewise.
2719 (ix86_match_ccmode): Likewise.
2720 (ix86_cc_modes_compatible): Likewise.
2721 (ix86_expand_branch): Likewise.
2722 (ix86_expand_sse_cmp): Likewise.
2723 (ix86_expand_sse_movcc): Likewise.
2724 (ix86_expand_int_sse_cmp): Likewise.
2725 (ix86_expand_vec_perm_vpermi2): Likewise.
2726 (ix86_expand_vec_perm): Likewise.
2727 (ix86_expand_sse_unpack): Likewise.
2728 (ix86_expand_int_addcc): Likewise.
2729 (ix86_split_to_parts): Likewise.
2730 (ix86_vectorize_builtin_gather): Likewise.
2731 (ix86_vectorize_builtin_scatter): Likewise.
2732 (avx_vpermilp_parallel): Likewise.
2733 (inline_memory_move_cost): Likewise.
2734 (ix86_tieable_integer_mode_p): Likewise.
2735 (x86_maybe_negate_const_int): Likewise.
2736 (ix86_expand_vector_init_duplicate): Likewise.
2737 (ix86_expand_vector_init_one_nonzero): Likewise.
2738 (ix86_expand_vector_init_one_var): Likewise.
2739 (ix86_expand_vector_init_concat): Likewise.
2740 (ix86_expand_vector_init_interleave): Likewise.
2741 (ix86_expand_vector_init_general): Likewise.
2742 (ix86_expand_vector_set): Likewise.
2743 (ix86_expand_vector_extract): Likewise.
2744 (emit_reduc_half): Likewise.
2745 (ix86_emit_i387_round): Likewise.
2746 (ix86_mangle_type): Likewise.
2747 (ix86_expand_round_sse4): Likewise.
2748 (expand_vec_perm_blend): Likewise.
2749 (canonicalize_vector_int_perm): Likewise.
2750 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
2751 (expand_vec_perm_1): Likewise.
2752 (expand_vec_perm_interleave3): Likewise.
2753 (expand_vec_perm_even_odd_pack): Likewise.
2754 (expand_vec_perm_even_odd_1): Likewise.
2755 (expand_vec_perm_broadcast_1): Likewise.
2756 (ix86_vectorize_vec_perm_const_ok): Likewise.
2757 (ix86_expand_vecop_qihi): Likewise.
2758 (ix86_expand_mul_widen_hilo): Likewise.
2759 (ix86_expand_sse2_abs): Likewise.
2760 (ix86_expand_pextr): Likewise.
2761 (ix86_expand_pinsr): Likewise.
2762 (ix86_preferred_simd_mode): Likewise.
2763 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
2764 * config/i386/sse.md (*andnot<mode>3): Likewise.
2765 (<mask_codefor><code><mode>3<mask_name>): Likewise.
2766 (*<code><mode>3): Likewise.
2767 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
2768 (ia64_expand_atomic_op): Likewise.
2769 (ia64_arg_type): Likewise.
2770 (ia64_mode_to_int): Likewise.
2771 (ia64_scalar_mode_supported_p): Likewise.
2772 (ia64_vector_mode_supported_p): Likewise.
2773 (expand_vec_perm_broadcast): Likewise.
2774 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
2775 (iq2000_function_arg_advance): Likewise.
2776 (iq2000_function_arg): Likewise.
2777 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
2778 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
2779 (m68k_libcall_value): Likewise.
2780 (m68k_function_value): Likewise.
2781 (sched_attr_op_type): Likewise.
2782 * config/mcore/mcore.c (mcore_output_move): Likewise.
2783 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
2785 (microblaze_function_arg): Likewise.
2786 * config/mips/mips.c (mips16_build_call_stub): Likewise.
2787 (mips_print_operand): Likewise.
2788 (mips_mode_ok_for_mov_fmt_p): Likewise.
2789 (mips_vector_mode_supported_p): Likewise.
2790 (mips_preferred_simd_mode): Likewise.
2791 (mips_expand_vpc_loongson_even_odd): Likewise.
2792 (mips_expand_vec_unpack): Likewise.
2793 (mips_expand_vi_broadcast): Likewise.
2794 (mips_expand_vector_init): Likewise.
2795 (mips_expand_vec_reduc): Likewise.
2796 (mips_expand_msa_cmp): Likewise.
2797 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
2798 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
2799 (cc_flags_for_mode): Likewise.
2800 * config/msp430/msp430.c (msp430_print_operand): Likewise.
2801 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
2802 (nds32_output_casesi_pc_relative): Likewise.
2803 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2804 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
2805 (nvptx_gen_unpack): Likewise.
2806 (nvptx_gen_pack): Likewise.
2807 (nvptx_gen_shuffle): Likewise.
2808 (nvptx_gen_wcast): Likewise.
2809 (nvptx_preferred_simd_mode): Likewise.
2810 * config/pa/pa.c (pa_secondary_reload): Likewise.
2811 * config/pa/predicates.md (base14_operand): Likewise.
2812 * config/powerpcspe/powerpcspe-c.c
2813 (altivec_resolve_overloaded_builtin): Likewise.
2814 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
2816 (rs6000_preferred_simd_mode): Likewise.
2817 (output_vec_const_move): Likewise.
2818 (rs6000_expand_vector_extract): Likewise.
2819 (rs6000_split_vec_extract_var): Likewise.
2820 (reg_offset_addressing_ok_p): Likewise.
2821 (rs6000_legitimate_offset_address_p): Likewise.
2822 (rs6000_legitimize_address): Likewise.
2823 (rs6000_emit_set_const): Likewise.
2824 (rs6000_const_vec): Likewise.
2825 (rs6000_emit_move): Likewise.
2826 (spe_build_register_parallel): Likewise.
2827 (rs6000_darwin64_record_arg_recurse): Likewise.
2828 (swap_selector_for_mode): Likewise.
2829 (spe_init_builtins): Likewise.
2830 (paired_init_builtins): Likewise.
2831 (altivec_init_builtins): Likewise.
2832 (do_load_for_compare): Likewise.
2833 (rs6000_generate_compare): Likewise.
2834 (rs6000_expand_float128_convert): Likewise.
2835 (emit_load_locked): Likewise.
2836 (emit_store_conditional): Likewise.
2837 (rs6000_output_function_epilogue): Likewise.
2838 (rs6000_handle_altivec_attribute): Likewise.
2839 (rs6000_function_value): Likewise.
2840 (emit_fusion_gpr_load): Likewise.
2841 (emit_fusion_p9_load): Likewise.
2842 (emit_fusion_p9_store): Likewise.
2843 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
2844 (fusion_gpr_mem_load): Likewise.
2845 (fusion_addis_mem_combo_load): Likewise.
2846 (fusion_addis_mem_combo_store): Likewise.
2847 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
2848 (fusion_gpr_mem_load): Likewise.
2849 (fusion_addis_mem_combo_load): Likewise.
2850 (fusion_addis_mem_combo_store): Likewise.
2851 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2853 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
2854 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
2855 (rs6000_preferred_simd_mode): Likewise.
2856 (output_vec_const_move): Likewise.
2857 (rs6000_expand_vector_extract): Likewise.
2858 (rs6000_split_vec_extract_var): Likewise.
2859 (reg_offset_addressing_ok_p): Likewise.
2860 (rs6000_legitimate_offset_address_p): Likewise.
2861 (rs6000_legitimize_address): Likewise.
2862 (rs6000_emit_set_const): Likewise.
2863 (rs6000_const_vec): Likewise.
2864 (rs6000_emit_move): Likewise.
2865 (rs6000_darwin64_record_arg_recurse): Likewise.
2866 (swap_selector_for_mode): Likewise.
2867 (paired_init_builtins): Likewise.
2868 (altivec_init_builtins): Likewise.
2869 (rs6000_expand_float128_convert): Likewise.
2870 (emit_load_locked): Likewise.
2871 (emit_store_conditional): Likewise.
2872 (rs6000_output_function_epilogue): Likewise.
2873 (rs6000_handle_altivec_attribute): Likewise.
2874 (rs6000_function_value): Likewise.
2875 (emit_fusion_gpr_load): Likewise.
2876 (emit_fusion_p9_load): Likewise.
2877 (emit_fusion_p9_store): Likewise.
2878 * config/rx/rx.c (rx_gen_move_template): Likewise.
2879 (flags_from_mode): Likewise.
2880 * config/s390/predicates.md (s390_alc_comparison): Likewise.
2881 (s390_slb_comparison): Likewise.
2882 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
2883 (s390_vector_mode_supported_p): Likewise.
2884 (s390_cc_modes_compatible): Likewise.
2885 (s390_match_ccmode_set): Likewise.
2886 (s390_canonicalize_comparison): Likewise.
2887 (s390_emit_compare_and_swap): Likewise.
2888 (s390_branch_condition_mask): Likewise.
2889 (s390_rtx_costs): Likewise.
2890 (s390_secondary_reload): Likewise.
2891 (__SECONDARY_RELOAD_CASE): Likewise.
2892 (s390_expand_cs): Likewise.
2893 (s390_preferred_simd_mode): Likewise.
2894 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
2895 * config/sh/sh.c (sh_print_operand): Likewise.
2896 (dump_table): Likewise.
2897 (sh_secondary_reload): Likewise.
2898 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2899 * config/sh/sh.md (casesi_worker_1): Likewise.
2900 (casesi_worker_2): Likewise.
2901 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
2902 (fcc_comparison_operator): Likewise.
2903 * config/sparc/sparc.c (sparc_expand_move): Likewise.
2904 (emit_soft_tfmode_cvt): Likewise.
2905 (sparc_preferred_simd_mode): Likewise.
2906 (output_cbranch): Likewise.
2907 (sparc_print_operand): Likewise.
2908 (sparc_expand_vec_perm_bmask): Likewise.
2909 (vector_init_bshuffle): Likewise.
2910 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2911 (spu_vector_mode_supported_p): Likewise.
2912 (spu_expand_insv): Likewise.
2913 (spu_emit_branch_or_set): Likewise.
2914 (spu_handle_vector_attribute): Likewise.
2915 (spu_builtin_splats): Likewise.
2916 (spu_builtin_extract): Likewise.
2917 (spu_builtin_promote): Likewise.
2918 (spu_expand_sign_extend): Likewise.
2919 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2920 (tilegx_simd_int): Likewise.
2921 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
2922 (tilepro_simd_int): Likewise.
2923 * config/v850/v850.c (const_double_split): Likewise.
2924 (v850_print_operand): Likewise.
2925 (ep_memory_offset): Likewise.
2926 * config/vax/vax.c (vax_rtx_costs): Likewise.
2927 (vax_output_int_move): Likewise.
2928 (vax_output_int_add): Likewise.
2929 (vax_output_int_subtract): Likewise.
2930 * config/visium/predicates.md (visium_branch_operator): Likewise.
2931 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
2932 (visium_print_operand_address): Likewise.
2933 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2934 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
2935 (xtensa_expand_conditional_branch): Likewise.
2936 (xtensa_copy_incoming_a7): Likewise.
2937 (xtensa_output_literal): Likewise.
2938 * dfp.c (decimal_real_maxval): Likewise.
2939 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2942 Alan Hayward <alan.hayward@arm.com>
2943 David Sherwood <david.sherwood@arm.com>
2945 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
2946 (mode_nunits_inline): Likewise.
2947 (mode_inner_inline): Likewise.
2948 (mode_unit_size_inline): Likewise.
2949 (mode_unit_precision_inline): Likewise.
2950 (emit_insn_modes_h): Likewise. Also emit a #define of the
2952 (emit_mode_wider): Add an E_ prefix to mode names.
2953 (emit_mode_complex): Likewise.
2954 (emit_mode_inner): Likewise.
2955 (emit_mode_adjustments): Likewise.
2956 (emit_mode_int_n): Likewise.
2957 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
2958 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
2959 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
2960 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
2961 (CRC32_BUILTIN, ENTRY): Likewise.
2962 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
2963 (aarch64_pop_regs): Likewise.
2964 (aarch64_process_components): Likewise.
2965 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
2966 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
2967 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
2968 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
2969 * config/arm/arm.c (arm_init_libfuncs): Likewise.
2970 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
2972 * config/i386/i386-builtin.def (pcmpestr): Likewise.
2973 (pcmpistr): Likewise.
2974 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
2975 * config/mmix/mmix.c (mmix_output_condition): Likewise.
2976 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2978 * config/rl78/rl78.c (mduc_regs): Likewise.
2979 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2980 (htm_expand_builtin): Likewise.
2981 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
2982 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
2983 * config/xtensa/xtensa.c (print_operand): Likewise.
2984 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
2985 (NUM_MODE_VECTOR_INT): Likewise.
2986 * genoutput.c (null_operand): Likewise.
2987 (output_operand_data): Likewise.
2988 * genrecog.c (print_parameter_value): Likewise.
2989 * lra.c (debug_operand_data): Likewise.
2991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2993 * dwarf2out.c (scompare_loc_descriptor_wide)
2994 (scompare_loc_descriptor_narrow): New functions, split out from...
2995 (scompare_loc_descriptor): ...here.
2996 * expmed.c (emit_store_flag_int): New function, split out from...
2997 (emit_store_flag): ...here.
2999 2017-08-30 Richard Biener <rguenther@suse.de>
3001 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3002 (dwarf2out_early_finish): Move setting of AT_pubnames from
3003 early debug output to early finish.
3005 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3007 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3008 and -mdata-region to the assembler.
3010 2017-08-30 Richard Biener <rguenther@suse.de>
3012 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3014 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3015 (dwarf2out_early_global_decl): Only generate a DIE for the
3016 abstract origin if it doesn't already exist or is a declaration DIE.
3017 (resolve_addr): Do not add the linkage name twice when
3018 generating a stub DIE for the DW_TAG_GNU_call_site target.
3020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3022 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3023 Use machine_mode rather than int for arg1_mode.
3025 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3028 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3029 that the second argument of the built-in functions to unpack
3030 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3031 switch statement instead a lot of if statements.
3032 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3033 Allow 64-bit values to be in Altivec registers as well as
3034 traditional floating point registers.
3035 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3037 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3039 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3040 MAX_REGS_PER_ADDRESS == 1.
3042 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3044 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3045 * config/i386/i386.c (ix86_option_override_internal): Simplify
3046 setting of opts->x_flag_entry.
3048 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3049 Jakub Jelinek <jakub@redhat.com>
3050 Richard Biener <rguenther@suse.de>
3052 PR tree-optimization/81503
3053 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3054 folded constant fits in the target type; reorder tests for clarity.
3056 2017-08-29 Martin Liska <mliska@suse.cz>
3058 * passes.def: Include pass_lower_switch.
3059 * stmt.c (dump_case_nodes): Remove and move to
3060 tree-switch-conversion.
3061 (case_values_threshold): Likewise.
3062 (expand_switch_as_decision_tree_p): Likewise.
3063 (emit_case_decision_tree): Likewise.
3064 (expand_case): Likewise.
3065 (balance_case_nodes): Likewise.
3066 (node_has_low_bound): Likewise.
3067 (node_has_high_bound): Likewise.
3068 (node_is_bounded): Likewise.
3069 (emit_case_nodes): Likewise.
3070 (struct simple_case_node): New struct.
3071 (add_case_node): Remove.
3072 (emit_case_dispatch_table): Use vector instead of case_list.
3073 (reset_out_edges_aux): Remove.
3074 (compute_cases_per_edge): Likewise.
3075 (expand_case): Build list of simple_case_node.
3076 (expand_sjlj_dispatch_table): Use it.
3077 * tree-switch-conversion.c (struct case_node): Moved from
3078 stmt.c and adjusted.
3079 (emit_case_nodes): Likewise.
3080 (node_has_low_bound): Likewise.
3081 (node_has_high_bound): Likewise.
3082 (node_is_bounded): Likewise.
3083 (case_values_threshold): Likewise.
3084 (reset_out_edges_aux): Likewise.
3085 (compute_cases_per_edge): Likewise.
3086 (add_case_node): Likewise.
3087 (dump_case_nodes): Likewise.
3088 (balance_case_nodes): Likewise.
3089 (expand_switch_as_decision_tree_p): Likewise.
3090 (emit_jump): Likewise.
3091 (emit_case_decision_tree): Likewise.
3092 (try_switch_expansion): Likewise.
3093 (do_jump_if_equal): Likewise.
3094 (emit_cmp_and_jump_insns): Likewise.
3095 (fix_phi_operands_for_edge): New function.
3096 (record_phi_operand_mapping): Likewise.
3097 (class pass_lower_switch): New pass.
3098 (pass_lower_switch::execute): New function.
3099 (make_pass_lower_switch): Likewise.
3100 (conditional_probability):
3101 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
3102 * tree-pass.h: Add make_pass_lower_switch.
3104 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3107 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
3110 2017-08-29 Richard Biener <rguenther@suse.de>
3112 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
3113 we do not add a DW_AT_inline attribute twice.
3114 (gen_subprogram_die): Remove code setting DW_AT_inline on
3115 DECL_ABSTRACT_P nodes.
3117 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
3119 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
3120 calls to internal functions.
3121 (gimplify_modify_expr): Likewise.
3122 * tree-call-cdce.c (use_internal_fn): Likewise.
3123 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3124 (convert_to_divmod): Set the nothrow flag.
3125 * tree-if-conv.c (predicate_mem_writes): Likewise.
3126 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3127 (vectorizable_call): Likewise.
3128 (vectorizable_store): Likewise.
3129 (vectorizable_load): Likewise.
3130 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
3131 (vect_recog_mask_conversion_pattern): Likewise.
3133 2017-08-29 Martin Liska <mliska@suse.cz>
3136 * gcc.c (driver_handle_option): Add new argument.
3137 * opts-common.c (handle_option): Pass
3138 target_option_override_hook.
3139 * opts-global.c (lang_handle_option): Add new option.
3140 (set_default_handlers): Add new argument.
3141 (decode_options): Likewise.
3142 * opts.c (target_handle_option): Likewise.
3143 (common_handle_option): Call target_option_override_hook.
3144 * opts.h (struct cl_option_handler_func): Add hook for
3145 target option override.
3146 (struct cl_option_handlers): Likewise.
3147 (set_default_handlers): Add new argument.
3148 (decode_options): Likewise.
3149 (common_handle_option): Likewise.
3150 (target_handle_option): Likewise.
3151 * toplev.c (toplev::main): Pass targetm.target_option.override
3154 2017-08-29 Richard Biener <rguenther@suse.de>
3155 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3157 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
3158 life to the active subtree.
3160 2017-08-28 Jeff Law <law@redhat.com>
3162 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
3163 derive_equivalences.
3164 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
3166 (edge_info::derive_equivalences): New private member function
3168 * tree-ssa-dom.c (class edge_info): Changed from a struct
3169 to a class. Add ctor/dtor, methods and data members.
3170 (edge_info::edge_info): Renamed from allocate_edge_info.
3171 Initialize additional members.
3172 (edge_info::~edge_info): New.
3173 (free_dom_edge_info): Delete the edge info.
3174 (record_edge_info): Use new class & associated member functions.
3175 Tighten forms for testing for edge equivalences.
3176 (record_temporary_equivalences): Iterate over the simple
3177 equivalences rather than assuming there's only one per edge.
3178 (cprop_into_successor_phis): Iterate over the simple
3179 equivalences rather than assuming there's only one per edge.
3180 (optimize_stmt): Use operand_equal_p rather than pointer
3181 equality for mini-DSE code.
3183 2017-08-28 Nathan Sidwell <nathan@acm.org>
3185 * gcc.c (execute): Fold SIGPIPE handling into switch
3186 statement. Adjust internal error message.
3188 2017-08-28 Richard Biener <rguenther@suse.de>
3191 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
3192 Do nothing for removed DIEs.
3194 2017-08-28 Richard Biener <rguenther@suse.de>
3196 PR tree-optimization/81977
3197 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
3200 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
3203 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
3204 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
3205 using targetm.gen_mem_thread_fence.
3207 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
3210 * config/i386/i386.md (*<btsc><mode>): Change operand 2
3211 predicate to register_operand. Reorder operands.
3212 (*btr<mode>): Ditto.
3213 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
3214 (*btr<mode>_mask): Ditto.
3216 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
3218 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
3219 * config/rs6000/xmmintrin.h: New file.
3220 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
3222 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3225 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
3226 parameter and_insn and return it.
3227 (recombine_lvx_pattern): Insert a copy to ensure availability of
3228 the base register of the copied masking operation at the point of
3229 the instruction replacement.
3230 (recombine_stvx_pattern): Likewise.
3232 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3234 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
3235 undocumented switches.
3236 (-mpower9-dform-vector): Likewise.
3237 (-mpower9-dform): Likewise.
3238 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
3239 comments to delete references to -mpower9-dform* switches.
3240 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
3241 Delete reference to -mpower9-dform* switches, test for
3242 -mpower9-vector instead.
3243 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
3244 (OTHER_P9_VECTOR_MASKS): Likewise.
3245 (POWERPC_MASKS): Likewise.
3246 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
3247 tests against -mpower9-dform* to -mpower9-vector. Delete code
3248 that checked for -mpower9-dform* consistancy with other options.
3249 Add test for -mpower9-misc to enable other power9 switches.
3250 (rs6000_init_hard_regno_mode_ok): Likewise.
3251 (rs6000_option_override_internal): Likewise.
3252 (rs6000_emit_prologue): Likewise.
3253 (rs6000_emit_epilogue): Likewise.
3254 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
3255 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
3256 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
3258 (emit_fusion_p9_store): Likewise.
3259 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
3260 resetting these macros if the assembler does not support ISA 3.0
3262 (TARGET_P9_DFORM_VECTOR): Likewise.
3263 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
3264 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
3266 2017-08-25 Alan Modra <amodra@gmail.com>
3268 PR rtl-optimization/81747
3269 * cse.c (cse_extended_basic_block): Don't attempt to record
3270 equivalences for degenerate conditional branches that branch
3271 to their fall-through.
3273 2017-08-24 Martin Sebor <msebor@redhat.com>
3276 * gimple-fold.c (size_must_be_zero_p): New function.
3277 (gimple_fold_builtin_memory_op): Call it.
3279 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
3281 * config/rs6000/mm_malloc.h: New file.
3283 2017-08-24 Bin Cheng <bin.cheng@arm.com>
3285 PR tree-optimization/81913
3286 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
3287 analysis when either IVs in condition can wrap.
3289 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
3291 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
3292 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
3294 2017-08-24 Richard Biener <rguenther@suse.de>
3297 * targhooks.c (default_target_can_inline_p): Properly
3298 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
3299 is present and always compare.
3300 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
3301 infer -mfpmath=sse from TARGET_SSE_P.
3302 (ix86_can_inline_p): Properly use target_option_default_node when
3303 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
3305 2017-08-24 Richard Biener <rguenther@suse.de>
3308 * dwarf2out.c (output_die): Handle flag_generate_offload like
3310 (output_comp_unit): Likewise.
3311 (gen_array_type_die): Likewise.
3312 (dwarf2out_early_finish): Likewise.
3313 (note_variable_value_in_expr): Likewise.
3314 (dwarf2out_finish): Likewise. Adjust assert.
3315 * cgraphunit.c (symbol_table::compile): Move setting of
3316 flag_generate_offload earlier ...
3317 (symbol_table::finalize_compilation_unit): ... here, before
3318 early debug finalization.
3320 2017-08-24 Richard Biener <rguenther@suse.de>
3322 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
3323 and ipa-fnsummary.h.
3324 (ix86_can_inline_p): When ix86_fpmath flags do not match
3325 check whether the callee uses FP math at all.
3327 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
3330 * tree-ssanames.c (get_nonzero_bits): Use element_precision
3331 instead of TYPE_PRECISION.
3333 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
3334 Alan Hayward <alan.hayward@arm.com>
3335 David Sherwood <david.sherwood@arm.com>
3337 * combine.c (make_extraction): Use subreg_offset_from_lsb.
3339 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
3341 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
3343 (ix86_frame::stack_realign_allocate): New field.
3344 (struct machine_frame_state): Modify comments.
3345 (machine_frame_state::sp_realigned_fp_end): New field.
3346 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
3348 (sp_valid_at): Add assertion to assure no attempt to access invalid
3349 offset of a realigned stack.
3350 (fp_valid_at): Likewise.
3351 (choose_baseaddr): Modify comments.
3352 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
3353 ix86_expand_prologue.
3354 (ix86_expand_prologue): Modify stack realignment and allocation.
3355 (ix86_expand_epilogue): Modify comments.
3356 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
3357 avx2_runtime, avx512f, and avx512f_runtime.
3359 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
3361 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
3362 (mstackrealign): Do not init to -1.
3363 * config/i386/i386.c (ix86_option_override_internal):
3364 Check opts_set, not opts when setting default value of
3365 opts->x_ix86_force_align_arg_pointer.
3367 2017-08-23 Richard Biener <rguenther@suse.de>
3369 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
3370 lang_hooks.decl_printable_name.
3371 * print-rtl-function.c (print_rtx_function): Likewise.
3372 * tree-pretty-print.c (dump_function_header): Likewise.
3374 2017-08-23 Richard Biener <rguenther@suse.de>
3377 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
3378 -g0 at compile-time.
3380 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3383 * doc/sourcebuild.texi (Other hardware attributes):
3386 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3389 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
3390 Fix single-use check.
3392 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
3394 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
3396 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3398 * doc/install.texi: Modify to add more details on running selected
3401 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3403 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
3404 is combined with -mabi=ms.
3405 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
3408 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3410 PR tree-optimization/81488
3411 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
3412 and cached_basis fields.
3413 (MAX_SPREAD): New constant.
3414 (alloc_cand_and_find_basis): Initialize new fields.
3415 (clear_visited): New function.
3416 (create_phi_basis_1): Rename from create_phi_basis, set visited
3417 and cached_basis fields.
3418 (create_phi_basis): New wrapper function.
3419 (phi_add_costs_1): Rename from phi_add_costs, add spread
3420 parameter, set visited field, short-circuit when limits reached.
3421 (phi_add_costs): New wrapper function.
3422 (record_phi_increments_1): Rename from record_phi_increments, set
3424 (record_phi_increments): New wrapper function.
3425 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
3426 (phi_incr_cost): New wrapper function.
3427 (all_phi_incrs_profitable_1): Rename from
3428 all_phi_incrs_profitable, set visited field.
3429 (all_phi_incrs_profitable): New wrapper function.
3431 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3432 Alan Hayward <alan.hayward@arm.com>
3433 David Sherwood <david.sherwood@arm.com>
3435 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
3436 that takes the outer and inner modes.
3437 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
3438 comparison as the canonical test for a paradoxical subreg.
3439 * combine.c (simplify_set): Use paradoxical_subreg_p.
3440 (make_extraction): Likewise.
3441 (force_to_mode): Likewise.
3442 (rtx_equal_for_field_assignment_p): Likewise.
3443 (gen_lowpart_for_combine): Likewise.
3444 (simplify_comparison): Likewise.
3445 * cse.c (equiv_constant): Likewise.
3446 * expmed.c (store_bit_field_1): Likewise.
3447 * final.c (alter_subreg): Likewise.
3448 * fwprop.c (propagate_rtx): Likewise.
3449 (forward_propagate_subreg): Likewise.
3450 * ira-conflicts.c (ira_build_conflicts): Likewise.
3451 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
3452 * lra-constraints.c (curr_insn_transform): Likewise.
3453 (split_reg): Likewise.
3454 * lra-eliminations.c (move_plus_up): Likewise.
3455 (lra_eliminate_regs_1): Likewise.
3456 * recog.c (general_operand): Likewise.
3457 * ree.c (combine_reaching_defs): Likewise.
3458 * reload.c (push_reload): Likewise.
3459 (find_reloads): Likewise.
3460 * reload1.c (elimination_effects): Likewise.
3461 (compute_reload_subreg_offset): Likewise.
3462 (choose_reload_regs): Likewise.
3463 * rtlanal.c (subreg_lsb_1): Likewise.
3464 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3465 (simplify_subreg): Likewise.
3466 * var-tracking.c (track_loc_p): Likewise.
3467 * emit-rtl.c (byte_lowpart_offset): Likewise.
3468 (paradoxical_subreg_p): Delete out-of-line definition.
3470 2017-08-22 Jeff Law <law@redhat.com>
3472 PR tree-optimization/81741
3473 PR tree-optimization/71947
3474 * tree-ssa-dom.c: Include tree-inline.h.
3475 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
3476 equivalences if one is more expensive to compute than the other.
3477 * tree-ssa-scopedtables.h (class const_or_copies): Make
3478 record_const_or_copy_raw method private.
3479 (class avail_exprs_stack): New method simplify_binary_operation.
3480 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
3481 avail_exprs_stack::simplify_binary_operation as needed.
3482 (avail_exprs_stack::simplify_binary_operation): New function.
3484 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
3486 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
3487 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
3488 (DOT_SYMBOLS): Likewise.
3489 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
3490 (RELOCATABLE_NEEDS_FIXUP): Likewise.
3491 (RS6000_ABI_NAME): Likewise.
3492 (TARGET_CMODEL): Likewise.
3493 (TOC_SECTION_ASM_OP): Likewise.
3494 (SET_CMODEL): New macro.
3495 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
3497 2017-08-22 Richard Biener <rguenther@suse.de>
3499 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
3500 to allow for free-lang-data replacements similar to verify_type_variant.
3502 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3503 Alan Hayward <alan.hayward@arm.com>
3504 David Sherwood <david.sherwood@arm.com>
3506 * config/aarch64/aarch64.md (casesi): Use DImode rather than
3507 VOIDmode for the LABEL_REF.
3509 2017-08-22 Richard Biener <rguenther@suse.de>
3511 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
3513 2017-08-22 Richard Biener <rguenther@suse.de>
3515 * common.opt (feliminate-dwarf2-dups): Ignore.
3516 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
3517 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
3518 same_die_p_wrap, compute_section_prefix,
3519 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
3520 (comdat_symbol_id, comdat_symbol_number): Likewise.
3521 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
3523 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
3524 (output_die): Mark unreachable path unreachable.
3525 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
3526 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
3527 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
3528 (dwarf2out_early_finish): Likewise.
3530 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
3532 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
3534 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
3537 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
3538 not VAR_P. Filter attribute warnings with OPT_Wattributes.
3539 (avr_attribute_table) <io, io_low, address>: Initialize
3540 .decl_required with true.
3542 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
3544 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
3545 undocumented debugging options.
3546 (-mvsx-scalar-double): Likewise.
3547 (-mallow-df-permute): Likewise.
3548 (-mvectorize-builtins): Likewise.
3549 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3550 (rs6000_builtin_vectorized_function): Likewise.
3551 (rs6000_builtin_md_vectorized_function): Likewise.
3552 (rs6000_opt_vars): Likewise.
3554 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
3557 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
3558 (*btrq_imm): Rename from *btrq.
3559 (*btcq_imm): Rename from *btcq.
3560 (btsc): New code attribute.
3561 (*<btsc><mode>): New insn pattern.
3562 (*btr<mode>): Ditto.
3563 (*<btsc><mode>_mask): New insn_and_split pattern.
3564 (*btr<mode>_mask): Ditto.
3566 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3567 Alan Hayward <alan.hayward@arm.com>
3568 David Sherwood <david.sherwood@arm.com>
3570 * function.c (pad_below): Simplify padding calculation.
3572 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3573 Alan Hayward <alan.hayward@arm.com>
3574 David Sherwood <david.sherwood@arm.com>
3576 * target.def (function_prologue): Remove frame size argument.
3577 (function_epilogue): Likewise.
3578 * doc/tm.texi: Regenerate.
3579 * final.c (final_start_function): Update call to function_prologue.
3580 (final_end_function): Update call to function_epilogue.
3581 (default_function_pro_epilogue): Remove frame size argument.
3582 * output.h (default_function_pro_epilogue): Likewise.
3583 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
3584 (arm_output_function_prologue): Likewise.
3585 * config/frv/frv.c (frv_function_prologue): Likewise.
3586 (frv_function_epilogue): Likewise.
3587 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
3588 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
3589 (ia64_output_function_epilogue): Likewise.
3590 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
3591 (m32r_output_function_epilogue): Likewise.
3592 * config/microblaze/microblaze.c (microblaze_function_prologue)
3593 (microblaze_function_epilogue): Likewise.
3594 * config/mips/mips.c (mips_output_function_prologue): Likewise.
3595 (mips_output_function_epilogue): Likewise.
3596 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
3597 (mmix_target_asm_function_epilogue): Likewise.
3598 * config/msp430/msp430.c (msp430_start_function): Likewise.
3599 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
3600 (nds32_asm_function_epilogue): Likewise.
3601 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
3602 * config/pa/pa.c (pa_output_function_prologue): Likewise.
3603 (pa_output_function_epilogue): Likewise.
3604 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
3605 (rs6000_output_function_epilogue): Likewise.
3606 * config/rl78/rl78.c (rl78_start_function): Likewise.
3607 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
3608 (rs6000_output_function_epilogue): Likewise.
3609 * config/rx/rx.c (rx_output_function_prologue): Likewise.
3610 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
3611 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
3612 (sparc_asm_function_epilogue): Likewise.
3614 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3616 * tree.h (type_has_mode_precision_p): New function.
3617 * convert.c (convert_to_integer_1): Use it.
3618 * expr.c (expand_expr_real_2): Likewise.
3619 (expand_expr_real_1): Likewise.
3620 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
3621 * match.pd: Likewise.
3622 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
3623 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
3624 * tree-tailcall.c (process_assignment): Likewise.
3625 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3626 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
3627 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
3628 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3629 (vectorizable_assignment): Likewise.
3630 (vectorizable_shift): Likewise.
3631 (vectorizable_operation): Likewise.
3632 * tree-vrp.c (register_edge_assert_for_2): Likewise.
3634 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
3636 * match.pd: Add pow (C, x) simplification.
3638 2017-08-21 Richard Biener <rguenther@suse.de>
3640 PR tree-optimization/81900
3641 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
3642 for blocks with abnormal predecessors.
3643 (compute_antic): Do not set visited flag prematurely.
3645 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
3648 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
3650 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3652 * stor-layout.h (vector_type_mode): Move to...
3653 * tree.h (vector_type_mode): ...here.
3654 * stor-layout.c (vector_type_mode): Move to...
3655 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
3657 2017-08-21 Richard Biener <rguenther@suse.de>
3659 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
3660 register_external_die hooks.
3661 (debug_false_tree_charstarstar_uhwistar): Declare.
3662 (debug_nothing_tree_charstar_uhwi): Likewise.
3663 * debug.c (do_nothing_debug_hooks): Adjust.
3664 (debug_false_tree_charstarstar_uhwistar): New do nothing.
3665 (debug_nothing_tree_charstar_uhwi): Likewise.
3666 * dbxout.c (dbx_debug_hooks): Adjust.
3667 (xcoff_debug_hooks): Likewise.
3668 * sdbout.c (sdb_debug_hooks): Likewise.
3669 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3670 * dwarf2out.c (macinfo_label_base): New global.
3671 (dwarf2out_register_external_die): New function for the
3672 register_external_die hook.
3673 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
3674 (dwarf2_debug_hooks): Use them.
3675 (dwarf2_lineno_debug_hooks): Adjust.
3676 (struct die_struct): Add with_offset flag.
3677 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
3678 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
3679 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
3680 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
3681 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
3682 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
3683 defining section names for the early LTO debug variants.
3684 (reset_indirect_string): New helper.
3685 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
3686 (print_dw_val): Add support for offsetted symbol references.
3687 (get_ultimate_context): Split out from is_cxx.
3688 (is_cxx): Use get_ultimate_context.
3689 (is_fortran): Add decl overload.
3690 (compute_comp_unit_symbol): Split out worker from
3691 compute_section_prefix.
3692 (compute_section_prefix): Call compute_comp_unit_symbol and
3693 set comdat_type_p here.
3694 (output_die): Skip DIE symbol output for the LTO added one.
3695 Handle DIE symbol references with offset.
3696 (output_comp_unit): Guard section name mangling properly.
3697 For LTO debug sections emit a symbol at the section beginning
3698 which we use to refer to its DIEs.
3699 (add_abstract_origin_attribute): For DIEs registered via
3700 dwarf2out_register_external_die directly refer to the early
3701 DIE rather than indirectly through the shadow one we created.
3702 Remove obsolete call to dwarf2out_abstract_function for
3703 non-function/block origins.
3704 (gen_array_type_die): When generating early LTO debug do
3705 not emit DW_AT_string_length.
3706 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
3707 late when in LTO. As suggested place a gcc_unreachable for
3708 the DECL_ABSTRACT_P case.
3709 (gen_subprogram_die): Avoid another specification DIE
3710 for early built declarations/definitions for the late LTO case.
3711 (gen_variable_die): Add type references for late duplicated VLA dies
3713 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
3714 we have the abstract instance already.
3715 (process_scope_var): Adjust decl DIE contexts in LTO which
3716 first puts them in limbo.
3717 (gen_decl_die): Do not generate type DIEs late apart from
3718 types for VLAs or for decls we do not yet have a DIE. Do not
3719 call dwarf2out_abstract_function late.
3720 (dwarf2out_early_global_decl): Make sure to create DIEs
3721 for abstract instances of a decl first.
3722 (dwarf2out_late_global_decl): Adjust comment.
3723 (output_macinfo_op): With multiple macro sections use
3724 macinfo_label_base to distinguish labels.
3725 (output_macinfo): Likewise. Update macinfo_label_base.
3726 Pass in the line info label.
3727 (note_variable_value_in_expr): When generating LTO resolve
3728 all variable values here by generating DIEs as needed.
3729 (init_sections_and_labels): Add early LTO debug flag parameter
3730 and generate different sections and names if set. Add generation
3731 counter for the labels so we can have multiple of them.
3732 (reset_dies): Helper to allow DIEs to be output multiple times.
3733 (dwarf2out_finish): When outputting DIEs to the fat part of an
3734 LTO object first reset DIEs.
3735 (dwarf2out_early_finish): Output early DIEs when generating LTO.
3736 (modified_type_die): Check for decl_ultimate_origin being self
3738 (gen_type_die_with_usage): Likewise.
3739 (gen_typedef_die): Allow decl_ultimate_origin being self.
3740 (set_decl_abstract_flags): Remove.
3741 (set_block_abstract_flags): Likewise.
3742 (dwarf2out_abstract_function): Treat the early generated DIEs
3743 as the abstract copy and only add DW_AT_inline and
3744 DW_AT_artificial here and call set_decl_origin_self.
3745 If the DIE has an abstract origin don't do anything.
3746 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
3747 if we have none yet (Go fails to build one, PR78628).
3748 (variably_modified_type_p): Prevent endless recursion for Ada
3749 cyclic pointer types.
3750 * lto-streamer-in.c: Include debug.h.
3751 (dref_queue): New global.
3752 (lto_read_tree_1): Stream in DIE references.
3753 (lto_input_tree): Register DIE references.
3754 (input_function): Stream DECL_DEBUG_ARGS.
3755 * lto-streamer-out.c: Include debug.h.
3756 (lto_write_tree_1): Output DIE references.
3757 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
3758 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
3759 (output_function): Stream DECL_DEBUG_ARGS.
3760 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
3761 Stream DECL_ABSTRACT_ORIGIN.
3762 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
3763 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
3764 DECL_CONTEXT for file-scope decls.
3765 * lto-streamer.h (struct dref_entry): Declare.
3766 (dref_queue): Likewise.
3767 * cfgexpand.c (pass_expand::execute): Do not call the
3768 outlining_inline_function hook here.
3769 * lto-wrapper.c (debug_obj): New global.
3770 (tool_cleanup): Unlink it if required.
3771 (debug_objcopy): New function.
3772 (run_gcc): Handle early debug sections in the IL files by
3773 extracting them to separate files, partially linkin them and
3774 feeding the result back as result to the linker.
3775 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
3776 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
3777 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
3778 sections into a separate segment.
3779 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
3781 (darwin_asm_dwarf_section): Likewise.
3782 (darwin_asm_output_dwarf_offset): Likewise.
3783 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
3785 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3786 Alan Hayward <alan.hayward@arm.com>
3787 David Sherwood <david.sherwood@arm.com>
3789 * read-md.h (md_reader::record_potential_iterator_use): Replace
3790 pointer argument with an rtx and an index.
3791 * read-rtl.c (iterator_group::apply_iterator): Likewise.
3792 (apply_mode_iterator): Likewise.
3793 (apply_code_iterator): Likewise.
3794 (apply_int_iterator): Likewise.
3795 (apply_subst_iterator): Likewise.
3796 (record_iterator_use): Likewise.
3797 (record_attribute_use): Likewise.
3798 (md_reader::record_potential_iterator_use): Likewise. Update calls
3799 to record_iterator_use and apply_iterator.
3800 (iterator_use): Replace ptr with x and index.
3801 (attribute_use): Likewise.
3802 (apply_attribute_uses): Update calls to apply_iterator.
3803 (apply_iterators): Likewise. Update initialization of iterator_use.
3804 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
3805 and record_potential_iterator_use.
3806 (rtx_reader::read_rtx_operand): Likewise.
3808 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3809 Alan Hayward <alan.hayward@arm.com>
3810 David Sherwood <david.sherwood@arm.com>
3812 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
3815 2017-08-21 Richard Biener <rguenther@suse.de>
3818 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
3819 at struct end conservatively when comparing common bases.
3821 2017-08-21 Richard Biener <rguenther@suse.de>
3823 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
3824 (mem_ref_in_stmt): Remove.
3825 (determine_max_movement): Use ref index to get at the reference.
3826 (invariantness_dom_walker::before_dom_children): Deal with
3827 lim data already initialized.
3828 (gather_mem_refs_stmt): Initialize lim data and record ref index.
3830 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
3832 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
3833 (TARGET_ISA_ROUND): Ditto.
3834 (TARGET_ROUND): Ditto.
3835 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
3836 * config/i386/i386.md: Ditto.
3837 * config/i386/sse.md: Ditto.
3838 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
3839 with OPTION_MASK_ISA_SSE4_1.
3841 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
3844 * doc/extend.texi (x86 Built-in Functions): Correct the name of
3845 __builtin_ia32_lzcnt_u16.
3847 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3850 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
3851 (rs6000_set_current_function): Rewrite function to use it.
3853 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
3856 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
3857 and TYPE_WARN_IF_NOT_ALIGN.
3858 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
3859 (handle_warn_if_not_align): New.
3860 (place_union_field): Call handle_warn_if_not_align.
3861 (place_field): Call handle_warn_if_not_align.
3862 Copy TYPE_WARN_IF_NOT_ALIGN.
3863 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
3864 (layout_type): Likewise.
3865 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
3867 (tree_decl_common): Add warn_if_not_align.
3868 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
3869 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
3870 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
3871 (DECL_WARN_IF_NOT_ALIGN): Likewise.
3872 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
3873 * doc/extend.texi: Document warn_if_not_aligned attribute.
3874 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
3876 2017-08-17 Martin Liska <mliska@suse.cz>
3879 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
3880 (get_data_dependence): Use it as pointer type.
3881 (distribute_loop): Likewise.
3883 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3885 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
3886 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
3887 (p8_vmrgow_<mode>_direct): New define_insn.
3888 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
3889 handle endianness for vmrgew and vmrgow permute patterns.
3891 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
3893 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
3894 * config/rs6000/rs6000-cpus.def: Remove comment.
3895 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
3896 (POWERPC_MASKS): Likewise.
3897 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
3898 use of TARGET_VSX_TIMODE.
3899 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
3900 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
3901 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
3902 (rs6000_option_override_internal): Remove dead code.
3903 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
3904 (rs6000_legitimize_reload_address): Likewise.
3905 (rs6000_legitimate_address_p): Likewise.
3906 (rs6000_opt_masks): Delete "vsx-timode".
3907 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
3908 from function comment.
3909 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
3910 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
3911 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
3913 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
3914 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
3915 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
3918 (define_peephole2): Likewise.
3920 2017-08-17 Martin Sebor <msebor@redhat.com>
3923 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
3924 past the end of an array.
3925 (test_pp_format): Add test cases.
3927 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
3929 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
3930 missing ECF_NOTHROW flags.
3932 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
3935 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
3936 operands residing in integer registers.
3937 (*vsx_le_perm_load_<mode>): Likewise.
3938 (*vsx_le_perm_store_<mode>): Likewise.
3939 (define_peephole2): Add peepholes to optimize the above.
3941 2017-08-17 Marek Polacek <polacek@redhat.com>
3944 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
3945 to mimic what shorten_compare did. Change the return type to bool.
3946 (fold_cond_expr_with_comparison): Update call to
3947 operand_equal_for_comparison_p.
3948 (fold_ternary_loc): Likewise.
3950 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
3952 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
3954 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
3956 2017-08-17 Richard Biener <rguenther@suse.de>
3958 * tree-ssa-structalias.c (solve_graph): When propagating
3959 to successors update the graphs succ edges and avoid duplicate work.
3961 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
3964 * config/i386/i386.c (ix86_option_override_internal): Save target
3965 specific options after ix86_stack_protector_guard_reg was changed.
3967 2017-08-17 Richard Biener <rguenther@suse.de>
3969 PR tree-optimization/81827
3970 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
3971 (new_var_info): Initialize it conservatively.
3972 (get_call_vi): Mark register vars.
3973 (new_scalar_tmp_constraint_exp): Likewise.
3974 (handle_rhs_call): Likewise.
3975 (handle_const_call): Likewise.
3976 (create_function_info_for): Likewise.
3977 (solve_constraints): Sort varinfos to separate register from
3978 non-register vars to pack points-to solution bitmaps during
3981 2017-08-17 Marek Polacek <polacek@redhat.com>
3983 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
3985 2017-08-17 Richard Biener <rguenther@suse.de>
3987 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
3988 to true when overflow is undefined and we saturated the result.
3990 2017-08-17 Alan Modra <amodra@gmail.com>
3993 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
3994 Don't use store multiple if only one reg needs saving.
3995 (interesting_frame_related_regno): New function.
3996 (rs6000_frame_related): Don't emit frame info for regs that
3998 (rs6000_emit_epilogue): Likewise.
4000 2017-08-16 Nathan Sidwell <nathan@acm.org>
4002 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4003 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4004 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4005 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4006 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4007 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4008 (verify_type): Adjust for TYPE_BINFO move.
4009 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4010 process TYPE_BINFO directly.
4011 (hash_tree): Likewise.
4012 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4014 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4017 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4019 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4021 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4024 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4025 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4026 Add ix86_binary_operator_ok to insn constraint.
4027 (*iordi_1_bts): Ditto.
4028 (*xordi_1_btc): Ditto.
4029 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4030 Update corresponding peephole2 pattern.
4034 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4036 PR tree-optimization/81832
4037 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4038 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4040 2017-08-16 Marek Polacek <polacek@redhat.com>
4043 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4044 regarding size_zero_node.
4046 2017-08-16 Martin Liska <mliska@suse.cz>
4049 * config.gcc: Respect previously set extra_objs in case
4052 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4054 PR tree-optimization/81835
4055 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4056 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4057 not depend on the phi.
4059 2017-08-16 Alan Modra <amodra@gmail.com>
4061 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4064 2017-08-16 Alan Modra <amodra@gmail.com>
4066 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4067 (save_reg_p): ..into this. Update all callers.
4068 (first_reg_to_save): Simplify.
4070 2017-08-16 Alan Modra <amodra@gmail.com>
4072 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4075 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4079 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4080 constant count if that count is less than 32.
4082 2017-08-15 Nathan Sidwell <nathan@acm.org>
4084 * gcc.c (execute): Emit friendlier message if inferior is killed
4085 by an external cause.
4087 2017-08-15 Richard Biener <rguenther@suse.de>
4089 PR tree-optimization/81790
4090 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4091 CONSTRUCTORs from simplifying and VN.
4093 2017-08-14 Martin Sebor <msebor@redhat.com>
4095 * builtin-attrs.def: Add comments.
4097 2017-08-14 Martin Sebor <msebor@redhat.com>
4100 * doc/extend.texi (attribute nonstring): Document new attribute.
4102 2017-08-14 Martin Sebor <msebor@redhat.com>
4105 * tree-diagnostic.c (default_tree_printer): Handle %G.
4106 * gimple-pretty-print.h (percent_G_format): Declare new function.
4107 * gimple-pretty-print.c (percent_G_format): Define.
4108 * tree-pretty-print.c (percent_K_format): Add argument.
4110 2017-08-14 Martin Sebor <msebor@redhat.com>
4112 PR translation/79998
4113 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
4114 Remove a stray space.
4116 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
4119 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
4120 (*iordi_1_bts): Ditto.
4121 (*xordi_1_btc): Ditto.
4123 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4126 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
4127 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4129 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
4130 quoted strings, and make more translator-friendly.
4131 (darwin_rs6000_override_options): Likewise.
4132 (rs6000_option_override_internal): Likewise.
4133 (rs6000_return_in_memory): Fix overlong line.
4134 (init_cmulative_args): Use quoted strings, and make more
4135 translator-friendly.
4136 (rs6000_pass_by_reference): Fix overlong line.
4137 (def_builtin): Use quoted strings.
4138 (altivec_expand_predicate_builtin): Use quoted strings, and make
4139 more translator-friendly.
4140 (htm_expand_builtin): Use quoted strings.
4141 (cpu_expand_builtin): Use quoted strings, and make more
4142 translator-friendly.
4143 (altivec_expand_builtin): Likewise.
4144 (paired_expand_predicate_builtin): Likewise.
4145 (rs6000_invalid_builtin): Likewise.
4146 (builtin_function_type): Use quoted strings.
4147 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
4148 more translator-friendly.
4149 (rs6000_trampoline_init): Likewise.
4150 (rs6000_handle_altivec_attribute): Likewise.
4151 (rs6000_inner_target_options): Use quoted strings.
4152 (rs6000_disable_incompatible_switches): Likewise.
4153 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
4154 strings, and make more translator-friendly.
4155 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
4157 2017-08-14 Bin Cheng <bin.cheng@arm.com>
4159 PR tree-optimization/81799
4160 * tree-loop-distribution.c (version_loop_by_alias_check): Force
4161 cond_expr to simple gimple operand.
4163 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
4166 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
4168 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
4172 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
4173 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
4174 TARGET_GASISR_PROLOGUES.
4175 * config/avr/avr.c (avr_option_override): Same.
4176 (avr_pass_pre_proep::execute): Same.
4178 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
4181 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
4182 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
4184 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
4186 * config/i386/i386.md (*load_tp_<mode>): Redefine as
4187 define_insn_and_split. Split to a memory load from 0 in
4188 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
4189 using PTR mode iterator.
4190 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
4191 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
4192 (*add_tp_<mode>): Redefine as define_insn_and_split.
4193 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
4194 address space. Merge with *add_tp_x32 using PTR mode iterator.
4195 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
4196 Split to an add with a memory load from 0 in
4197 DEFAULT_TLS_SEG_REG address space.
4199 2017-08-12 Andrew Pinski <apinski@cavium.com>
4201 * config/aarch64/aarch64-option-extensions.def (rdma):
4202 Fix feature string to what Linux prints out in /proc/cpuinfo.
4204 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
4207 * dwarf2out.c (modified_type_die): For C typedef types that have
4208 an ultimate origin, process the ultimate origin instead of the
4210 (gen_typedef_die): Assert that input DECLs have no ultimate
4212 (gen_type_die_with_usage): For typedef variants that have an
4213 ultimate origin, just call gen_decl_die on the original DECL.
4214 (process_scope_var): Avoid creating DIEs for local typedefs and
4215 concrete static variables.
4217 2017-08-12 Alan Modra <amodra@gmail.com>
4221 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
4222 match gnu-user.h startfile.
4223 (ENDFILE_LINUX_SPEC): Similarly.
4225 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
4228 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
4230 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
4232 2017-08-11 Tamar Christina <tamar.christina@arm.com>
4233 * config/aarch64/aarch64.md (mov<mode>): Change.
4234 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4235 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
4236 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
4238 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
4240 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
4241 for storage order barriers.
4243 2017-08-11 Martin Liska <mliska@suse.cz>
4246 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
4247 variables of void type.
4249 2017-08-11 Martin Liska <mliska@suse.cz>
4251 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
4252 TARGET_SUPPORTS_ALIASES.
4253 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
4254 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
4255 (optimize_weakref): Likewise.
4256 * symtab.c (symtab_node::noninterposable_alias): Likewise.
4257 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
4258 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
4260 2017-08-11 Martin Liska <mliska@suse.cz>
4263 * config/i386/i386.c (make_resolver_func): Do complete
4264 refactoring of the function.
4266 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4269 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
4270 * config/i386/i386.c (ix86_stack_protect_guard): Use
4271 ix86_stack_protect_guard_symbol_str to generate varible declaration.
4272 * doc/invoke.texi (x86 Options): Document
4273 -mstack-protector-guard-symbol= option.
4275 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4277 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
4278 * config/i386/i386.c (ix86_split_stack_guard): New function.
4279 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
4280 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
4281 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
4282 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
4283 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
4284 (split_stack_space_check): Call ix86_split_stack_guard.
4286 2017-08-10 Martin Sebor <msebor@redhat.com>
4288 * print-tree.c (print_node): Print location using the established
4290 Replace spaces with colons.
4291 (debug_raw, debug): Ditto.
4293 2017-08-10 Martin Sebor <msebor@redhat.com>
4296 * pretty-print.c (pp_format): Correct the handling of %s precision.
4298 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
4301 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4303 (ix86_finalize_stack_frame_flags): This. Also clear
4304 frame_pointer_needed if -fno-omit-frame-pointer is used without
4306 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4307 with ix86_finalize_stack_frame_flags.
4308 (ix86_expand_epilogue): Likewise.
4309 (ix86_expand_split_stack_prologue): Likewise.
4310 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
4312 2017-08-10 Martin Liska <mliska@suse.cz>
4315 * c-attribs.c (handle_target_attribute):
4316 Report warning for an empty string argument of target attribute.
4318 2017-08-09 Jakub Jelinek <jakub@redhat.com>
4321 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
4323 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
4324 or DECL_NONLOCAL labels.
4325 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
4326 or DECL_NONLOCAL labels here.
4328 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
4330 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
4331 to indicate when early gimple folding has been disabled.
4332 (rs6000_gimple_fold_builtin): Add debug content.
4333 (rs6000_invalid_builtin): Fix whitespace.
4334 (rs6000_expand_builtin): Fix whitespace.
4335 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
4337 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
4340 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
4341 SAVE_MULTIPLE if not all the registers that saves, should be saved.
4343 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
4345 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
4346 (qdf24xx): Likewise.
4347 * config/aarch64/aarch64.md: Include falkor.md.
4348 * config/aarch64/falkor.md: New.
4350 2017-08-09 Marek Polacek <polacek@redhat.com>
4353 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
4354 * diagnostic.c (emit_diagnostic): Add a comment.
4355 (emit_diagnostic_valist): New function.
4357 2017-08-09 Marek Polacek <polacek@redhat.com>
4360 * input.c (make_location): New overload.
4361 * input.h (make_location): Declare.
4363 2017-08-08 Alan Modra <amodra@gmail.com>
4364 H.J. Lu <hongjiu.lu@intel.com>
4367 * gcc.c (NO_PIE_SPEC): Delete.
4368 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
4370 (LINK_PIE_SPEC): ..to here.
4371 (LINK_COMMAND_SPEC): Support -no-pie.
4372 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
4373 chain of crtbegin*.o selection, update for PIE_SPEC changes and
4375 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
4376 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
4377 (ENDFILE_CRTEND_SPEC): Similarly.
4379 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
4382 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
4383 (mstack-protector-guard-offset=): Ditto.
4384 * config/i386/i386.c (ix86_option_override): Handle
4385 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
4387 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
4388 ix86_stack_protect_guard_offset variables.
4389 (TARGET_STACK_PROTECT_GUARD): Always define.
4390 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
4391 and -mstack-protector-guard-offset= options.
4393 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4395 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
4396 boundary case for the last candidate.
4398 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4400 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
4401 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
4403 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4406 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
4407 * config/aarch64/aarch64-builtins.c
4408 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
4409 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
4410 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
4412 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4413 Andrew Pinski <pinskia@gmail.com>
4416 * internal-fn.def (XORSIGN): New.
4417 * optabs.def (xorsign_optab): New.
4418 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
4419 (convert_expand_mult_copysign): New.
4420 (pass_optimize_widening_mul::execute): Call
4421 convert_expand_mult_copysign.
4423 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4425 PR tree-optimization/81354
4426 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
4427 Insert on edges rather than explicitly creating landing pads.
4428 (analyze_candidates_and_replace): Commit edge inserts.
4430 2017-08-08 Richard Biener <rguenther@suse.de>
4433 * tree-ssa-loop-niter.c: Include tree-dfa.h.
4434 (expand_simple_operations): Also look through ADDR_EXPRs with
4435 MEM_REF bases treating them as POINTER_PLUS_EXPR.
4437 2017-08-08 Richard Biener <rguenther@suse.de>
4439 PR tree-optimization/81723
4440 * tree-vect-slp.c (struct bst_traits): New hash traits.
4441 (bst_fail): New global.
4442 (vect_build_slp_tree_2): New worker, split out from ...
4443 (vect_build_slp_tree): ... this now wrapping it with using
4444 bst_fail set to cache SLP tree build fails. Properly handle
4446 (vect_analyze_slp_instance): Allocate and free bst_fail.
4448 2017-08-08 Martin Liska <mliska@suse.cz>
4451 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
4452 LABEL_DECLs that can be from a different function.
4454 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4456 PR tree-optimization/81744
4457 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
4458 loop's number of iterations.
4460 2017-08-08 Martin Liska <mliska@suse.cz>
4462 * asan.c: Include header files.
4463 * attribs.c (build_decl_attribute_variant): New function moved
4465 (build_type_attribute_qual_variant): Likewise.
4466 (cmp_attrib_identifiers): Likewise.
4467 (simple_cst_list_equal): Likewise.
4468 (omp_declare_simd_clauses_equal): Likewise.
4469 (attribute_value_equal): Likewise.
4470 (comp_type_attributes): Likewise.
4471 (build_type_attribute_variant): Likewise.
4472 (lookup_ident_attribute): Likewise.
4473 (remove_attribute): Likewise.
4474 (merge_attributes): Likewise.
4475 (merge_type_attributes): Likewise.
4476 (merge_decl_attributes): Likewise.
4477 (merge_dllimport_decl_attributes): Likewise.
4478 (handle_dll_attribute): Likewise.
4479 (attribute_list_equal): Likewise.
4480 (attribute_list_contained): Likewise.
4481 * attribs.h (lookup_attribute): New function moved from tree.[ch].
4482 (lookup_attribute_by_prefix): Likewise.
4483 * bb-reorder.c: Include header files.
4484 * builtins.c: Likewise.
4485 * calls.c: Likewise.
4486 * cfgexpand.c: Likewise.
4487 * cgraph.c: Likewise.
4488 * cgraphunit.c: Likewise.
4489 * convert.c: Likewise.
4490 * dwarf2out.c: Likewise.
4491 * final.c: Likewise.
4492 * fold-const.c: Likewise.
4493 * function.c: Likewise.
4494 * gimple-expr.c: Likewise.
4495 * gimple-fold.c: Likewise.
4496 * gimple-pretty-print.c: Likewise.
4497 * gimple.c: Likewise.
4498 * gimplify.c: Likewise.
4499 * hsa-common.c: Likewise.
4500 * hsa-gen.c: Likewise.
4501 * internal-fn.c: Likewise.
4502 * ipa-chkp.c: Likewise.
4503 * ipa-cp.c: Likewise.
4504 * ipa-devirt.c: Likewise.
4505 * ipa-fnsummary.c: Likewise.
4506 * ipa-inline.c: Likewise.
4507 * ipa-visibility.c: Likewise.
4509 * lto-cgraph.c: Likewise.
4510 * omp-expand.c: Likewise.
4511 * omp-general.c: Likewise.
4512 * omp-low.c: Likewise.
4513 * omp-offload.c: Likewise.
4514 * omp-simd-clone.c: Likewise.
4515 * opts-global.c: Likewise.
4516 * passes.c: Likewise.
4517 * predict.c: Likewise.
4518 * sancov.c: Likewise.
4519 * sanopt.c: Likewise.
4520 * symtab.c: Likewise.
4521 * toplev.c: Likewise.
4522 * trans-mem.c: Likewise.
4523 * tree-chkp.c: Likewise.
4524 * tree-eh.c: Likewise.
4525 * tree-into-ssa.c: Likewise.
4526 * tree-object-size.c: Likewise.
4527 * tree-parloops.c: Likewise.
4528 * tree-profile.c: Likewise.
4529 * tree-ssa-ccp.c: Likewise.
4530 * tree-ssa-live.c: Likewise.
4531 * tree-ssa-loop.c: Likewise.
4532 * tree-ssa-sccvn.c: Likewise.
4533 * tree-ssa-structalias.c: Likewise.
4534 * tree-ssa.c: Likewise.
4535 * tree-streamer-in.c: Likewise.
4536 * tree-vectorizer.c: Likewise.
4537 * tree-vrp.c: Likewise.
4539 * ubsan.c: Likewise.
4540 * varasm.c: Likewise.
4541 * varpool.c: Likewise.
4542 * tree.c: Remove functions moved to attribs.[ch].
4544 * config/aarch64/aarch64.c: Add attrs.h header file.
4545 * config/alpha/alpha.c: Likewise.
4546 * config/arc/arc.c: Likewise.
4547 * config/arm/arm.c: Likewise.
4548 * config/avr/avr.c: Likewise.
4549 * config/bfin/bfin.c: Likewise.
4550 * config/c6x/c6x.c: Likewise.
4551 * config/cr16/cr16.c: Likewise.
4552 * config/cris/cris.c: Likewise.
4553 * config/darwin.c: Likewise.
4554 * config/epiphany/epiphany.c: Likewise.
4555 * config/fr30/fr30.c: Likewise.
4556 * config/frv/frv.c: Likewise.
4557 * config/ft32/ft32.c: Likewise.
4558 * config/h8300/h8300.c: Likewise.
4559 * config/i386/winnt.c: Likewise.
4560 * config/ia64/ia64.c: Likewise.
4561 * config/iq2000/iq2000.c: Likewise.
4562 * config/lm32/lm32.c: Likewise.
4563 * config/m32c/m32c.c: Likewise.
4564 * config/m32r/m32r.c: Likewise.
4565 * config/m68k/m68k.c: Likewise.
4566 * config/mcore/mcore.c: Likewise.
4567 * config/microblaze/microblaze.c: Likewise.
4568 * config/mips/mips.c: Likewise.
4569 * config/mmix/mmix.c: Likewise.
4570 * config/mn10300/mn10300.c: Likewise.
4571 * config/moxie/moxie.c: Likewise.
4572 * config/msp430/msp430.c: Likewise.
4573 * config/nds32/nds32-isr.c: Likewise.
4574 * config/nds32/nds32.c: Likewise.
4575 * config/nios2/nios2.c: Likewise.
4576 * config/nvptx/nvptx.c: Likewise.
4577 * config/pa/pa.c: Likewise.
4578 * config/pdp11/pdp11.c: Likewise.
4579 * config/powerpcspe/powerpcspe.c: Likewise.
4580 * config/riscv/riscv.c: Likewise.
4581 * config/rl78/rl78.c: Likewise.
4582 * config/rx/rx.c: Likewise.
4583 * config/s390/s390.c: Likewise.
4584 * config/sh/sh.c: Likewise.
4585 * config/sol2.c: Likewise.
4586 * config/sparc/sparc.c: Likewise.
4587 * config/spu/spu.c: Likewise.
4588 * config/stormy16/stormy16.c: Likewise.
4589 * config/tilegx/tilegx.c: Likewise.
4590 * config/tilepro/tilepro.c: Likewise.
4591 * config/v850/v850.c: Likewise.
4592 * config/vax/vax.c: Likewise.
4593 * config/visium/visium.c: Likewise.
4594 * config/xtensa/xtensa.c: Likewise.
4596 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4599 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
4600 constraints since the -mupper-regs-* switches have been
4602 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
4603 into a vector from a double word element that was extracted from
4604 another vector, and eliminate extra XXPERMDI instructions.
4605 (vsx_concat_<mode>_2): Likewise.
4606 (vsx_concat_<mode>_3): Likewise.
4607 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
4608 concat to allow optimizing inserts from previous extracts.
4610 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
4612 * config/i386/i386.c (ix86_stack_protect_guard): Generate
4613 memory reference to a SSP offset in TLS address space.
4614 (ix86_print_operand) <case '@'>: Remove.
4615 (ix86_print_operand_punct_valid_p): Remove '@' code.
4616 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
4618 (stack_tls_protect_set_<mode>): Remove.
4619 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
4620 (stack_tls_protect_test_<mode>): Remove.
4621 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
4623 2017-08-07 Olivier Hainque <hainque@adacore.com>
4626 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
4628 2017-08-07 Douglas Rupp <rupp@adacore.com>
4630 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
4631 variable was referenced as multidir in command.
4633 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4636 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
4639 2017-08-07 Martin Liska <mliska@suse.cz>
4641 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
4642 * config/rl78/rl78.c: Add include of attribs.h.
4643 * config/sh/sh.c: Likewise.
4644 * config/v850/v850.c: Likewise.
4646 2017-08-07 Tom de Vries <tom@codesourcery.com>
4649 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
4651 2017-08-07 Martin Liska <mliska@suse.cz>
4653 * config/mips/mips.c: Include attribs.h.
4655 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
4658 * doc/invoke.texi: Document change in behvaior for -Ofast for
4661 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
4663 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
4665 (aarch64_pop_regs): Likewise.
4666 (aarch64_gen_load_pair): Likewise.
4667 (aarch64_save_callee_saves): Likewise.
4668 (aarch64_restore_callee_saves): Likewise.
4670 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4672 * config/i386/i386.c: Revert the last change.
4674 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4677 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4679 (ix86_finalize_stack_frame_flags): This. Also clear
4680 frame_pointer_needed if -fno-omit-frame-pointer is used without
4682 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4683 with ix86_finalize_stack_frame_flags.
4684 (ix86_expand_epilogue): Likewise.
4685 (ix86_expand_split_stack_prologue): Likewise.
4687 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4690 * config/i386/i386.c (get_builtin_code_for_version): Set priority
4691 to P_AES for Westmere.
4693 2017-08-07 Jonathan Yong <10walls@gmail.com>
4695 * config/i386/mingw.opt (fset-stack-executable): Removed.
4696 * config/i386/cygming.opt (fset-stack-executable): Moved
4698 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
4700 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
4702 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
4704 2017-08-07 Marek Polacek <polacek@redhat.com>
4707 * fold-const.c (fold_indirect_ref_1): Check type_domain.
4709 2017-08-07 Martin Liska <mliska@suse.cz>
4711 * attribs.h (canonicalize_attr_name): New function.
4712 (cmp_attribs): Move from c-format.c and adjusted.
4713 (is_attribute_p): Moved from tree.h.
4714 * tree-inline.c: Add new includes.
4715 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
4716 (private_is_attribute_p): Remove.
4717 (private_lookup_attribute): Likewise.
4718 (private_lookup_attribute_by_prefix): Simplify.
4719 (remove_attribute): Use is_attribute_p.
4720 * tree.h: Remove removed declarations.
4722 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4725 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
4726 instead of computing it in the function. Formatting fix.
4727 (expand_case): Don't rely on default_edge being the first edge,
4728 clear it if removing it, pass default_edge to
4729 emit_case_dispatch_table.
4730 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
4733 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
4735 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
4736 insn in the function, emit NOP after the insn.
4738 2017-08-06 Tom de Vries <tom@codesourcery.com>
4740 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
4743 2017-08-06 Tom de Vries <tom@codesourcery.com>
4745 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
4748 2017-08-04 Yury Gribov <tetra2005@gmail.com>
4750 PR tree-optimization/57371
4751 * match.pd: New pattern.
4753 2017-08-04 Marek Polacek <polacek@redhat.com>
4756 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
4757 perform the computation in offset_int.
4759 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4761 PR tree-optimization/81136
4762 * tree-vectorizer.h: Include tree-hash-traits.h.
4763 (vec_base_alignments): New typedef.
4764 (vec_info): Add a base_alignments field.
4765 (vect_record_base_alignments): Declare.
4766 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
4768 (DR_IS_CONDITIONAL_IN_STMT): New macro.
4769 (create_data_ref): Add an is_conditional_in_stmt argument.
4770 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
4771 the is_conditional_in_stmt field.
4772 (data_ref_loc): Add an is_conditional_in_stmt field.
4773 (get_references_in_stmt): Set the is_conditional_in_stmt field.
4774 (find_data_references_in_stmt): Update call to create_data_ref.
4775 (graphite_find_data_references_in_stmt): Likewise.
4776 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
4777 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
4778 (vect_record_base_alignment): New function.
4779 (vect_record_base_alignments): Likewise.
4780 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
4781 for nested statements even if we fail to compute a misalignment.
4782 Use pooled base alignments for unconditional references.
4783 (vect_find_same_alignment_drs): Compare base addresses instead
4785 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
4786 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
4788 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4790 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
4791 Add an explicit name for the enum. Use auto_vec for slp_instances
4793 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
4795 (_bb_vec_info): Add a constructor and destructor.
4796 (vinfo_for_stmt): Return NULL for uids of -1 as well.
4797 (destroy_loop_vec_info): Delete.
4798 (vect_destroy_datarefs): Likewise.
4799 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
4800 (vec_info::vec_info): New function.
4801 (vec_info::~vec_info): Likewise.
4802 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
4803 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
4804 destroy_loop_vec_info.
4805 * tree-vect-loop.c (new_loop_vec_info): Replace with...
4806 (_loop_vec_info::_loop_vec_info): ...this.
4807 (destroy_loop_vec_info): Replace with...
4808 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
4809 the stmt_vec_infos. Leave handling of vec_info information to its
4810 destructor. Remove explicit vector releases.
4811 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
4812 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
4813 * tree-vect-slp.c (new_bb_vec_info): Replace with...
4814 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
4815 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
4816 (destroy_bb_vec_info): Replace with...
4817 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
4818 information to its destructor.
4819 (vect_slp_analyze_bb_1): Use new and delete instead of
4820 new_bb_vec_info and destroy_bb_vec_info.
4821 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
4824 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4826 * tree-data-ref.h (subscript): Add access_fn field.
4827 (data_dependence_relation): Add could_be_independent_p.
4828 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
4829 (same_access_functions): Move to tree-data-ref.c.
4830 * tree-data-ref.c (ref_contains_union_access_p): New function.
4831 (access_fn_component_p): Likewise.
4832 (access_fn_components_comparable_p): Likewise.
4833 (dr_analyze_indices): Add a reference to access_fn_component_p.
4834 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
4836 (constant_access_functions): Likewise.
4837 (add_other_self_distances): Likewise.
4838 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
4839 (initialize_data_dependence_relation): Use XCNEW and remove
4840 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
4841 of access functions that have the same type. Allow the
4842 subsequence to end with different bases in some circumstances.
4843 Record the chosen access functions in SUB_ACCESS_FN.
4844 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
4845 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
4846 (subscript_dependence_tester_1): Likewise dra and drb.
4847 (build_classic_dist_vector): Update calls accordingly.
4848 (subscript_dependence_tester): Likewise.
4849 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
4850 DDR_COULD_BE_INDEPENDENT_P.
4851 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
4852 comp_alias_ddrs instead of may_alias_ddrs.
4853 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
4855 (vect_analyze_data_ref_dependence): Use it if
4856 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
4857 distance vectors if that fails.
4858 (dependence_distance_ge_vf): New function.
4859 (vect_prune_runtime_alias_test_list): Use it. Don't clear
4860 LOOP_VINFO_MAY_ALIAS_DDRS.
4862 2017-08-04 Richard Biener <rguenther@suse.de>
4865 * fold-const.c (fold_binary_loc): Properly restrict
4866 minus_var0 && minus_var1 case when associating undefined overflow
4869 2017-08-04 Tom de Vries <tom@codesourcery.com>
4871 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
4873 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4875 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4876 Don't start diagnostic messages with a capital letter.
4877 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4879 (rs6000_invalid_builtin): Likewise.
4880 (rs6000_trampoline_init): Likewise.
4882 2017-08-03 Jakub Jelinek <jakub@redhat.com>
4885 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
4886 after setting changeable df flags.
4888 2017-08-03 Richard Biener <rguenther@suse.de>
4890 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
4891 up if the use is in USE - X.
4893 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
4895 * toplev.c (dumpfile.h): New include.
4896 (internal_error_reentered): New static function. Use it...
4897 (internal_error_function): ...here to handle reentered internal_error.
4899 2017-08-03 Richard Biener <rguenther@suse.de>
4902 * fold-const.c (split_tree): Add minus_var and minus_con
4903 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
4904 here but always use minus_*.
4905 (associate_trees): Assert we never associate with MINUS_EXPR
4906 and NULL first operand. Do not recurse for PLUS_EXPR operands
4907 when associating as MINUS_EXPR either.
4908 (fold_binary_loc): Track minus_var and minus_con.
4910 2017-08-03 Tom de Vries <tom@codesourcery.com>
4913 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
4914 ACCEL_COMPILER, apply finish_options on
4915 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
4917 2017-08-03 Tom de Vries <tom@codesourcery.com>
4920 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
4921 function_entry_patch_area_size > 0.
4923 2017-08-03 Jakub Jelinek <jakub@redhat.com>
4926 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
4927 instead of 10??LU, perform unit multiplication in wide_int,
4928 don't change alloc_object_size_limit if the limit is larger
4931 PR tree-optimization/81655
4932 PR tree-optimization/81588
4933 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
4934 the case when ranges[i].low and high are 1 for unsigned type with
4938 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
4939 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
4941 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4943 * tree-vrp.h: Add include guard.
4945 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
4948 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
4949 (ud2): New insn pattern.
4950 * config/i386/i386.c (ix86_expand_epilogue):
4951 For naked functions, generate ud2 instead of trap insn.
4953 2017-08-02 Marek Polacek <polacek@redhat.com>
4956 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
4958 2017-08-02 Tom de Vries <tom@codesourcery.com>
4959 Cesar Philippidis <cesar@codesourcery.com>
4961 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
4962 Add missing edge probabilities.
4964 2017-08-02 Tamar Christina <tamar.christina@arm.com>
4966 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
4969 2017-08-02 Jakub Jelinek <jakub@redhat.com>
4972 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
4973 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
4974 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
4975 prologue_seq sequences - if any.
4977 2017-08-02 Richard Biener <rguenther@suse.de>
4979 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
4980 via vectors if supported, integer extracts via punning if supported
4981 or otherwise vector extracts.
4983 2017-08-02 Richard Biener <rguenther@suse.de>
4985 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
4987 (bitmap_insert_into_set): ... this.
4989 2017-08-02 Richard Biener <rguenther@suse.de>
4991 PR tree-optimization/81633
4993 2015-08-17 Alan Hayward <alan.hayward@arm.com>
4995 PR tree-optimization/71752
4996 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
4998 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5000 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5001 (machine_function::call_ms2sysv_pad_out): Remove field.
5002 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5003 (ix86_compute_frame_layout): Likewise.
5005 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5008 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5009 attribute with interrupt attribute.
5011 2017-08-01 Andrew Pinski <apinski@cavium.com>
5013 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5014 BIT_INSERT_EXPR's operand 1
5015 to see if the types precision matches.
5017 2017-08-01 Martin Liska <mliska@suse.cz>
5020 * builtins.c (expand_builtin_memcpy_args): Remove.
5021 (expand_builtin_memcpy): Call newly added function
5022 expand_builtin_memory_copy_args.
5023 (expand_builtin_memcpy_with_bounds): Likewise.
5024 (expand_builtin_mempcpy): Remove last argument.
5025 (expand_builtin_mempcpy_with_bounds): Likewise.
5026 (expand_builtin_memory_copy_args): New function created from
5027 expand_builtin_mempcpy_args with small modifications.
5028 (expand_builtin_mempcpy_args): Remove.
5029 (expand_builtin_stpcpy): Remove unused argument.
5030 (expand_builtin): Likewise.
5031 (expand_builtin_with_bounds): Likewise.
5033 2017-08-01 Martin Liska <mliska@suse.cz>
5036 Make mempcpy more optimal (PR middle-end/70140).
5038 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5041 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5042 __builtin_vec_cmpne verify both arguments are compatible vectors
5043 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5044 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5045 move computation of aligned to after checking the argument types.
5049 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5050 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5053 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5054 Doug Rupp <rupp@adacore.com>
5055 Olivier Hainque <hainque@adacore.com>
5057 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5058 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5060 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5061 for TARGET_OS_CPP_BUILTIN.
5062 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5063 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5065 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5066 passing required abi options to the assembler for EABI configurations.
5067 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5068 of .text.hot and .text.unlikely sections for kernel modules when
5069 using ARM style exceptions.
5070 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5071 options. Add EXTRA_CC1_SPEC.
5072 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5074 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5076 (ARM_TARGET2_DWARF_FORMAT): Define.
5077 * config/arm/t-vxworks: Adjust multilib control to removal of the
5078 Diab command line options.
5080 2017-08-01 Martin Liska <mliska@suse.cz>
5082 PR gcov-profile/81561
5083 * gcov.c (unblock): Make unblocking safe as we need to preserve
5084 index correspondence of blocks and block_lists.
5086 2017-08-01 Richard Biener <rguenther@suse.de>
5088 PR tree-optimization/81181
5089 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5090 (compute_antic): ... end of iteration here.
5092 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5094 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5095 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5096 (ftree-slp-vectorize): Likewise.
5097 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5098 can no longer be set independent of flag_tree_loop_vectorize.
5099 * omp-general.c (emp_max_vf): Likewise.
5100 * opts.c (enable_fdo_optimizations): Remove references to
5101 flag_tree_vectorize, these are now implicit.
5102 (common_handle_option): Remove handling for OPT_ftree_vectorize,
5103 and leave it for the options machinery.
5105 2017-08-01 Martin Liska <mliska@suse.cz>
5108 * builtins.c (expand_builtin_memcpy_args): Remove.
5109 (expand_builtin_memcpy): Call newly added function
5110 expand_builtin_memory_copy_args.
5111 (expand_builtin_memcpy_with_bounds): Likewise.
5112 (expand_builtin_mempcpy): Remove last argument.
5113 (expand_builtin_mempcpy_with_bounds): Likewise.
5114 (expand_builtin_memory_copy_args): New function created from
5115 expand_builtin_mempcpy_args with small modifications.
5116 (expand_builtin_mempcpy_args): Remove.
5117 (expand_builtin_stpcpy): Remove unused argument.
5118 (expand_builtin): Likewise.
5119 (expand_builtin_with_bounds): Likewise.
5121 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5124 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
5125 print "ds:" only for immediates in generic address space.
5127 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5130 * config/i386/i386.c (ix86_funciton_naked): New prototype.
5131 (ix86_function_ok_for_sibcall): Return false for naked functions.
5133 2017-08-01 Richard Biener <rguenther@suse.de>
5135 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
5136 (compute_antic): Seed worklist with exit block predecessors.
5137 * cfganal.c (dfs_find_deadend): For a cycle return the source
5138 of the edge closing it.
5140 2017-08-01 Tamar Christina <tamar.christina@arm.com>
5142 * config/aarch64/aarch64.c
5143 (aarch64_can_const_movi_rtx_p): Move 0 check.
5145 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5147 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
5148 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
5150 * match.pd: Ditto in address comparison pattern.
5152 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5154 PR tree-optimization/81627
5155 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
5156 closed ssa form for store-store chain.
5158 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5160 PR tree-optimization/81620
5161 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
5162 for store-store chain.
5164 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5166 PR tree-optimization/81588
5167 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5168 ranges[i].in_p, invert comparison code ccode. For >/>=,
5169 swap rhs1 and rhs2 and comparison code unconditionally,
5170 for </<= don't do that. Don't swap rhs1/rhs2 again if
5171 ranges[i].in_p, instead invert comparison code ccode if
5172 opcode or oe->rank is BIT_IOR_EXPR.
5175 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5176 a direct optab to conversion optab.
5177 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
5178 with GET_MODE_INNER as last argument instead of optab_handler.
5179 * expmed.c (extract_bit_field_1): Likewise. Use vector from
5180 vector extraction if possible and optab is available.
5181 * expr.c (store_constructor): Use convert_optab_handler instead
5182 of optab_handler. Use vector initialization from smaller
5183 vectors if possible and optab is available.
5184 * tree-vect-stmts.c (vectorizable_load): Likewise.
5185 * doc/md.texi (vec_extract, vec_init): Document that the optabs
5187 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
5188 of vec_init from half-sized vectors with the same element mode.
5189 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
5190 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
5191 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
5192 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
5193 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
5194 after mode in gen_vec_extract* calls.
5195 (vec_extract<mode>): Renamed to ...
5196 (vec_extract<mode><ssescalarmodelower>): ... this.
5197 (vec_extract<mode><ssehalfvecmodelower>): New expander.
5198 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
5199 element mode after mode in gen_vec_init* calls.
5200 (VEC_INIT_HALF_MODE): New mode iterator.
5201 (vec_init<mode>): Renamed to ...
5202 (vec_init<mode><ssescalarmodelower>): ... this.
5203 (vec_init<mode><ssehalfvecmodelower>): New expander.
5204 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
5205 (vec_extractv2sfsf): ... this.
5206 (vec_initv2sf): Renamed to ...
5207 (vec_initv2sfsf): ... this.
5208 (vec_extractv2si): Renamed to ...
5209 (vec_extractv2sisi): ... this.
5210 (vec_initv2si): Renamed to ...
5211 (vec_initv2sisi): ... this.
5212 (vec_extractv4hi): Renamed to ...
5213 (vec_extractv4hihi): ... this.
5214 (vec_initv4hi): Renamed to ...
5215 (vec_initv4hihi): ... this.
5216 (vec_extractv8qi): Renamed to ...
5217 (vec_extractv8qiqi): ... this.
5218 (vec_initv8qi): Renamed to ...
5219 (vec_initv8qiqi): ... this.
5220 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
5221 (vec_init<mode>): Renamed to ...
5222 (vec_init<mode><VEC_base_l>): ... this.
5223 (vec_extract<mode>): Renamed to ...
5224 (vec_extract<mode><VEC_base_l>): ... this.
5225 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
5226 (vec_initv2sfsf): ... this.
5227 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
5228 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5229 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
5230 element mode after mode in gen_vec_init* calls.
5231 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
5232 (vec_init<mode><Vel>): ... this.
5233 (vec_extract<mode>): Renamed to ...
5234 (vec_extract<mode><Vel>): ... this.
5235 * config/aarch64/iterators.md (Vel): New mode attribute.
5236 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
5237 Add element mode after mode in gen_vec_extract* calls.
5238 * config/s390/vector.md (non_vec_l): New mode attribute.
5239 (vec_extract<mode>): Renamed to ...
5240 (vec_extract<mode><non_vec_l>): ... this.
5241 (vec_init<mode>): Renamed to ...
5242 (vec_init<mode><non_vec_l>): ... this.
5243 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
5244 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
5246 * config/arm/iterators.md (V_elem_l): New mode attribute.
5247 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
5248 (vec_extract<mode><V_elem_l>): ... this.
5249 (vec_extractv2di): Renamed to ...
5250 (vec_extractv2didi): ... this.
5251 (vec_init<mode>): Renamed to ...
5252 (vec_init<mode><V_elem_l>): ... this.
5253 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
5254 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
5255 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
5256 Add element mode after gen_vec_extract* calls.
5257 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
5258 (vec_init<mode><unitmode>): ... this.
5259 (vec_extract<mode>): Renamed to ...
5260 (vec_extract<mode><unitmode>): ... this.
5261 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
5262 (vec_init<mode><unitmode>): ... this.
5263 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
5264 (vec_initv2sfsf): ... this.
5265 (vec_extractv2sf): Renamed to ...
5266 (vec_extractv2sfsf): ... this.
5267 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
5268 Add element mode after gen_vec_extract* calls.
5269 * config/mips/mips.md (unitmode): New mode iterator.
5270 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
5271 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
5272 * config/spu/spu.md (inner_l): New mode attribute.
5273 (vec_init<mode>): Renamed to ...
5274 (vec_init<mode><inner_l>): ... this.
5275 (vec_extract<mode>): Renamed to ...
5276 (vec_extract<mode><inner_l>): ... this.
5277 * config/sparc/sparc.md (veltmode): New mode iterator.
5278 (vec_init<VMALL:mode>): Renamed to ...
5279 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
5280 * config/ia64/vect.md (vec_initv2si): Renamed to ...
5281 (vec_initv2sisi): ... this.
5282 (vec_initv2sf): Renamed to ...
5283 (vec_initv2sfsf): ... this.
5284 (vec_extractv2sf): Renamed to ...
5285 (vec_extractv2sfsf): ... this.
5286 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
5287 (vec_init<mode>): Renamed to ...
5288 (vec_init<mode><VEC_base_l>): ... this.
5289 (vec_extract<mode>): Renamed to ...
5290 (vec_extract<mode><VEC_base_l>): ... this.
5291 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
5292 (vec_initv2sfsf): ... this.
5293 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
5294 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5295 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
5296 gen_vec_init* calls.
5298 2017-08-01 Richard Biener <rguenther@suse.de>
5300 PR tree-optimization/81297
5301 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
5302 TREE_OVERFLOW from INTEGER_CSTs.
5304 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
5306 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
5308 2017-07-31 Carl Love <cel@us.ibm.com>
5310 * config/rs6000/rs6000-c: Add support for built-in functions
5311 vector signed char vec_xl_be (signed long long, signed char *);
5312 vector unsigned char vec_xl_be (signed long long, unsigned char *);
5313 vector signed int vec_xl_be (signed long long, signed int *);
5314 vector unsigned int vec_xl_be (signed long long, unsigned int *);
5315 vector signed long long vec_xl_be (signed long long, signed long long *);
5316 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
5317 vector signed short vec_xl_be (signed long long, signed short *);
5318 vector unsigned short vec_xl_be (signed long long, unsigned short *);
5319 vector double vec_xl_be (signed long long, double *);
5320 vector float vec_xl_be (signed long long, float *);
5321 * config/rs6000/altivec.h (vec_xl_be): Add #define.
5322 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
5323 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
5325 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
5326 (altivec_expand_builtin): Add switch statement to call
5327 altivec_expand_xl_be for each builtin.
5328 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
5329 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
5330 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
5331 __builtin_vsx_le_be_v16qi.
5332 * doc/extend.texi: Update the built-in documentation file for the
5333 new built-in functions.
5335 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5338 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
5340 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5342 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5344 * config.gcc: Add z14.
5345 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5346 CPU model numbers for z13s and z14.
5347 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
5349 * config/s390/s390-opts.h (enum processor_type): Rename
5350 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5351 * config/s390/s390.c (processor_table): Add field for CPU name to
5352 be passed to Binutils.
5353 (s390_asm_output_machine_for_arch): Use the new field in
5354 processor_table for Binutils.
5355 (s390_expand_builtin): Replace arch12 with z14.
5356 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5357 (s390_get_sched_attrmask): Likewise.
5358 (s390_get_unit_mask): Likewise.
5359 * config/s390/s390.opt: Add z14 to processor_type enum.
5361 2017-07-31 Martin Jambor <mjambor@suse.cz>
5364 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
5365 regardless of optimization level.
5367 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
5368 Martin Liska <mliska@suse.cz>
5370 * predict.def: Remove old comment and adjust probability.
5371 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
5374 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5377 * config/i386/i386.c (ix86_function_naked): New function.
5378 (ix86_can_use_return_insn_p): Return false for naked functions.
5379 (ix86_expand_prologue): Skip prologue for naked functions.
5380 (ix86_expand_epilogue): Skip epilogue for naked functions
5381 and emit trap instruction.
5382 (ix86_warn_func_return): New function.
5383 (ix86_attribute_table): Add "naked" attribute specification.
5384 (TARGET_WARN_FUNC_RETURN): Define.
5385 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
5387 2017-07-31 Martin Liska <mliska@suse.cz>
5389 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
5390 (dump_gimple_bb_header): Always dump BB info.
5391 (pp_cfg_jump): Do not append info about BB when dumping a jump.
5393 2017-07-31 Martin Liska <mliska@suse.cz>
5396 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
5397 also with current_function_decl non-null equality.
5399 2017-07-31 Jakub Jelinek <jakub@redhat.com>
5402 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
5403 change type to the element type, instead add eltype variable and
5404 use it where we are interested in the element type.
5406 PR tree-optimization/81603
5407 * ipa-polymorphic-call.c
5408 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
5409 offset arithmetic in offset_int, bail out if the resulting bit offset
5410 doesn't fit into shwi.
5412 2017-07-31 Martin Liska <mliska@suse.cz>
5414 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
5415 (gimplify_save_expr): Fix comment.
5417 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
5420 * config/i386/i386.c (ix86_function_arg): Update arguments for
5422 (ix86_compute_frame_layout): Set the initial stack offset to
5423 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
5424 INCOMING_FRAME_SP_OFFSET.
5425 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
5426 stack before exception handler returns.
5427 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
5428 the 'ERROR_CODE' for exception handler.
5430 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
5432 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
5433 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
5434 (ASM_OUTPUT_REG_POP): Ditto.
5435 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
5436 instead of asm_fprintf to output pure string.
5438 2017-07-29 Jakub Jelinek <jakub@redhat.com>
5440 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
5441 to imported_module_or_decl hook.
5442 (debug_nothing_tree_tree_tree_bool): Remove.
5443 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
5444 * debug.c (do_nothing_debug_hooks): Use
5445 debug_nothing_tree_tree_tree_bool_bool instead of
5446 debug_nothing_tree_tree_tree_bool.
5447 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5448 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
5449 * sdbout.c (sdb_debug_hooks): Likewise.
5450 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
5451 (gen_namespace_die): Add DW_AT_export_symbols attribute if
5453 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
5454 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
5455 attribute, don't add anything.
5457 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5459 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
5460 (fold_build2_stat_loc): Likewise.
5461 (fold_build3_stat_loc): Likewise.
5462 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
5463 (fold_build1_loc): Remove macro.
5464 (fold_build2_loc): Likewise.
5465 (fold_build3_loc): Likewise.
5467 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5469 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
5470 (gimple_build_debug_bind_source_stat): Likewise.
5471 * gimple.h (gimple_build_debug_bind): Remove macro.
5472 (gimple_build_debug_bind_source): Likewise.
5474 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5476 * bitmap.c (bitmap_alloc): Adjust.
5477 (bitmap_gc_alloc): Likewise.
5478 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
5480 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5482 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
5483 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
5484 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
5485 (bitmap_gc_alloc_stat): Likewise.
5486 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
5488 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5490 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
5491 * rtl.h (shallow_copy_rtx): Remove macro.
5493 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5495 * emit-rtl.c (gen_raw_REG): Adjust.
5496 * gengenrtl.c (gendef): Likewise.
5497 * rtl.c (rtx_alloc_stat): Remove _stat from name.
5498 * rtl.h (rtx_alloc): Remove macro.
5500 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5502 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
5503 (build_tree_list_stat): Likewise.
5504 * tree.h (build_tree_list): Remove macro.
5505 (build_tree_list_vec): Likewise.
5507 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5509 * tree.c (make_vector_stat): Remove _stat from name.
5510 (build_vector_stat): Likewise.
5511 * tree.h (make_vector_stat): Remove macro.
5512 (build_vector_stat): Likewise.
5514 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5516 * tree.h (build_var_debug_value): Remove prototype.
5518 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5520 * tree.c (tree_cons_stat): Remove _stat from name.
5521 * tree.h (tree_cons): Remove macro.
5523 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5525 * tree.c (build_vl_exp_stat): Remove _stat from name.
5526 * tree.h (build_vl_exp): Remove macro.
5528 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5530 * tree.c (build_decl_stat): Remove _stat from name.
5531 * tree.h (build_decl): Remove macro.
5533 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5535 * gimple.c (gimple_build_with_ops_stat): Adjust.
5536 (gimple_alloc_stat): Remove _stat from name.
5537 * gimple.h (gimple_alloc): Remove macro.
5539 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5541 * tree.c (make_tree_vec_stat): Remove _stat from name.
5542 (grow_tree_vec_stat): Likewise.
5543 * tree.h (make_tree_vec_stat): Adjust prototype.
5544 (grow_tree_vec_stat): Likewise.
5545 (make_tree_vec): Remove macro.
5546 (grow_tree_vec): Likewise.
5548 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5550 * fold-const.c (fold_build1_stat_loc): Adjust.
5551 (fold_build2_stat_loc): Likewise.
5552 (fold_build3_stat_loc): Likewise.
5553 * tree.c (build0_stat): Remove _stat from name.
5554 (build1_stat): Likewise.
5555 (build2_stat): Likewise.
5556 (build3_stat): Likewise.
5557 (build4_stat): Likewise.
5558 (build5_stat): Likewise.
5559 * tree.h (build1_loc): Remove macro, and rename _stat function
5561 (build2_loc): Likewise.
5562 (build3_loc): Likewise.
5563 (build4_loc): Likewise.
5564 (build5_loc): Likewise.
5566 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5568 * tree.c (make_int_cst_stat): Remove _stat from name.
5569 * tree.h (make_int_cst_stat): Adjust prototype.
5570 (make_int_cst): Remove macro.
5572 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5574 * tree.c (make_tre_binfo_stat): Remove _stat from name.
5575 * tree.h (make_tree_binfo_stat): Adjust prototype.
5576 (make_tree_binfo): Remove.
5578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5580 * tree.c (copy_node_stat): Rename to copy_node.
5581 (build_distinct_type_copy): Adjust.
5582 * tree.h (copy_node_stat): Adjust prototype.
5583 (copy_node): Remove macro.
5585 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5587 * tree.c (make_node_stat): rename to make_node.
5588 (build_tree_list_stat): Adjust.
5589 (build0_stat): Likewise.
5590 (build2_stat): Likewise.
5591 (build3_stat): Likewise.
5592 (build4_stat): Likewise.
5593 (build5_stat): Likewise.
5594 (build_decl_stat): Likewise.
5595 * tree.h (make_node_stat): Adjust prototype.
5596 (make_node): remove macro.
5598 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
5600 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
5601 (PPC_FEATURE2_SCV): Likewise.
5602 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
5604 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5606 * config/aarch64/aarch64.c
5607 (aarch64_internal_mov_immediate): Add new special pattern.
5608 * config/aarch64/aarch64.md (*movdi_aarch64):
5609 Add reg/32bit const mov case.
5611 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5612 Richard Sandiford <richard.sandiford@linaro.org>
5614 * config/aarch64/aarch64.md (mov<mode>): Generalize.
5615 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5616 Add integer and movi cases.
5617 (movi-split-hf-df-sf split, fp16): New.
5618 (enabled): Added TARGET_FP_F16INST.
5619 * config/aarch64/iterators.md (GPF_HF): New.
5620 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
5622 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5624 * config/aarch64/aarch64.c
5625 (aarch64_simd_container_mode): Add prototype.
5626 (aarch64_expand_mov_immediate): Add HI support.
5627 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
5628 (aarch64_can_const_movi_rtx_p): New.
5629 (aarch64_preferred_reload_class):
5630 Remove restrictions of using FP registers for certain SIMD operations.
5631 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
5632 (aarch64_valid_floating_const): Add integer move validation.
5633 (aarch64_simd_imm_scalar_p): Remove.
5634 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
5635 (aarch64_legitimate_constant_p): Expand list of supported cases.
5636 * config/aarch64/aarch64-protos.h
5637 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
5638 (aarch64_reinterpret_float_as_int): New.
5639 (aarch64_simd_imm_scalar_p): Remove.
5640 * config/aarch64/constraints.md (Uvi): New.
5641 (Dd): Split into Ds and new Dd.
5642 * config/aarch64/aarch64.md (*movsi_aarch64):
5644 (*movdi_aarch64): Add SIMD mov case.
5646 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5648 * tree-predcom.c: (struct chain): Handle store-store chain in which
5649 stores for elimination only store loop invariant values.
5650 (execute_pred_commoning_chain): Ditto.
5651 (prepare_initializers_chain_store_elim): Ditto.
5652 (prepare_finalizers): Ditto.
5653 (is_inv_store_elimination_chain): New function.
5654 (initialize_root_vars_store_elim_1): New function.
5656 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5658 * tree-predcom.c: Revise general description of the pass.
5659 (enum chain_type): New enum type for store elimination.
5660 (struct chain): New field supporting store elimination.
5661 (struct component): Ditto.
5662 (dump_chain): Dump store-stores chain.
5663 (release_chain): Release resources.
5664 (split_data_refs_to_components): Compute and create component
5665 contains only stores for elimination.
5666 (get_chain_last_ref_at): New function.
5667 (make_invariant_chain): Initialization.
5668 (make_rooted_chain): Specify chain type in parameter and record it.
5669 (add_looparound_copies): Skip for store-stores chain.
5670 (determine_roots_comp): Compute type of chain and pass it to
5672 (initialize_root_vars_store_elim_2): New function.
5673 (finalize_eliminated_stores): New function.
5674 (remove_stmt): Handle store for elimination.
5675 (execute_pred_commoning_chain): Execute predictive commoning on
5677 (determine_unroll_factor): Skip unroll for store-stores chain.
5678 (prepare_initializers_chain_store_elim): New function.
5679 (prepare_initializers_chain): Hanlde store-store chain.
5680 (prepare_finalizers_chain, prepare_finalizers): New function.
5681 (tree_predictive_commoning_loop): Return integer value indicating
5682 if loop is unrolled or lcssa form is corrupted.
5683 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
5685 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5687 * tree-predcom.c (initialize_root): Delete.
5688 (execute_pred_commoning_chain): Initialize root vars and replace
5689 reference of non-combined chain directly, rather than call above
5692 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5694 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
5695 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
5697 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5699 * tree-predcom.c (struct chain): New field init_seq.
5700 (release_chain): Release init_seq.
5701 (prepare_initializers_chain): Record intialization stmts in above
5703 (insert_init_seqs): New function.
5704 (tree_predictive_commoning_loop): Call insert_init_seqs.
5706 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5708 * tree-predcom.c (determine_roots_comp): Skip trivial components.
5710 2017-07-28 Richard Biener <rguenther@suse.de>
5712 * match.pd: Remove superfluous :c.
5713 * genmatch.c (simplify::id): Add member.
5714 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
5716 (current_id): New global.
5717 (dt_node::parent): Move from ...
5718 (dt_operand::parent): ... here. Add for_id member.
5719 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
5720 (decision_tree::find_node): Relax order requirement when
5721 merging DT_TRUE nodes to ones inbetween the current simplify
5722 and the one we try to merge with. Add diagnostic whenever
5723 we need to enforce pattern order by not merging.
5724 (decision_tree::insert): Set current_id.
5725 (decision_tree::print_node): Dump parent node and for_id.
5726 (parser::last_id): Add member.
5727 (parser::push_simplify): Assign unique id.
5728 (parser::parser): Initialize last_id.
5730 2017-07-28 Martin Liska <mliska@suse.cz>
5733 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
5734 gimple_build_debug_bind.
5736 2017-07-28 Richard Biener <rguenther@suse.de>
5738 PR tree-optimization/81502
5739 * match.pd: Add pattern combining BIT_INSERT_EXPR with
5741 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
5743 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
5744 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
5745 for BIT_FIELD_REF args.
5746 * fold-const.c (make_bit_field_ref): Likewise.
5747 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5749 2017-07-28 Jakub Jelinek <jakub@redhat.com>
5752 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
5753 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
5754 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
5755 Or it into SANITIZER_UNDEFINED.
5756 * ubsan.c: Include gimple-fold.h and varasm.h.
5757 (ubsan_expand_ptr_ifn): New function.
5758 (instrument_pointer_overflow): New function.
5759 (maybe_instrument_pointer_overflow): New function.
5760 (instrument_object_size): Formatting fix.
5761 (pass_ubsan::execute): Call instrument_pointer_overflow
5762 and maybe_instrument_pointer_overflow.
5763 * internal-fn.c (expand_UBSAN_PTR): New function.
5764 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
5765 * sanitizer.def (__ubsan_handle_pointer_overflow,
5766 __ubsan_handle_pointer_overflow_abort): New builtins.
5767 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
5768 * internal-fn.def (UBSAN_PTR): New internal function.
5769 * opts.c (sanitizer_opts): Add pointer-overflow.
5770 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
5771 * fold-const.c (build_range_check): Compute pointer range check in
5772 integral type if pointer arithmetics would be needed. Formatting
5775 2017-07-28 Martin Liska <mliska@suse.cz>
5778 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
5779 parameters that are of a variable-length.
5781 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
5783 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
5785 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
5786 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
5787 (CRT_CALL_STATIC_FUNCTION): Likewise.
5788 (ASM_DEFAULT_SPEC): New define.
5789 (ASM_SPEC32): Likewise.
5790 (ASM_SPEC64): Likewise.
5791 (ASM_SPEC_COMMON): Likewise.
5792 (ASM_SPEC): Likewise.
5793 (INVALID_64BIT): Likewise.
5794 (LINK_OS_DEFAULT_SPEC): Likewise.
5795 (LINK_OS_SPEC32): Likewise.
5796 (LINK_OS_SPEC64): Likewise.
5797 (POWERPC_LINUX): Likewise.
5798 (PTRDIFF_TYPE): Likewise.
5799 (RESTORE_FP_PREFIX): Likewise.
5800 (RESTORE_FP_SUFFIX): Likewise.
5801 (SAVE_FP_PREFIX): Likewise.
5802 (SAVE_FP_SUFFIX): Likewise.
5803 (SIZE_TYPE): Likewise.
5804 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
5805 (TARGET_64BIT): Likewise.
5806 (TARGET_64BIT): Likewise.
5807 (TARGET_AIX): Likewise.
5808 (WCHAR_TYPE_SIZE): Likewise.
5809 (WCHAR_TYPE): Undefine.
5810 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
5811 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
5812 (CPP_OS_RTEMS_SPEC): Delete.
5813 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
5814 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
5816 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
5818 2017-07-28 Jakub Jelinek <jakub@redhat.com>
5820 PR tree-optimization/81578
5821 * tree-parloops.c (build_new_reduction): Bail out if
5822 reduction_code isn't one of the standard OpenMP reductions.
5823 Move the details printing after that decision.
5825 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5827 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
5828 related to reload_in_progress.
5829 (splat_input_operand): Likewise.
5830 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
5832 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
5834 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
5835 (TARGET_INSTANTIATE_DECLS): Likewise.
5836 (legitimate_indexed_address_p): Delete reload_in_progress code.
5837 (rs6000_debug_legitimate_address_p): Likewise.
5838 (rs6000_eliminate_indexed_memrefs): Likewise.
5839 (rs6000_emit_le_vsx_store): Likewise.
5840 (rs6000_emit_move_si_sf_subreg): Likewise.
5841 (rs6000_emit_move): Likewise.
5842 (register_to_reg_type): Likewise.
5843 (rs6000_pre_atomic_barrier): Likewise.
5844 (rs6000_machopic_legitimize_pic_address): Likewise.
5845 (rs6000_allocate_stack_temp): Likewise.
5846 (rs6000_address_for_fpconvert): Likewise.
5847 (rs6000_address_for_altivec): Likewise.
5848 (rs6000_secondary_memory_needed_rtx): Delete function.
5849 (rs6000_check_sdmode): Likewise.
5850 (rs6000_alloc_sdmode_stack_slot): Likewise.
5851 (rs6000_instantiate_decls): Likewise.
5852 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
5853 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
5854 Delete reload_in_progress.
5855 (*vec_reload_and_plus_<mptrsize>): Likewise.
5856 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
5857 (vsx_div_v2di): Likewise.
5858 (vsx_udiv_v2di): Likewise.
5860 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5862 * config/rs6000/rs6000.opt (mlra): Replace with stub.
5863 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
5864 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
5865 (rs6000_debug_reg_global): Delete print of LRA status.
5866 (rs6000_option_override_internal): Delete dead LRA related code.
5867 (rs6000_lra_p): Delete function.
5868 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
5870 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
5872 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
5873 * config/riscv/rtems.h: New file.
5875 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5876 Sudakshina Das <sudi.das@arm.com>
5878 * config/aarch64/aarch64.md
5879 (define_split for and<mode>3nr_compare): Move
5880 non aarch64_logical_operand to a register.
5881 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
5882 register immediate operand to a register.
5883 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
5885 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5888 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
5890 2017-07-27 Richard Biener <rguenther@suse.de>
5892 PR tree-optimization/81573
5893 PR tree-optimization/81494
5894 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
5895 multi defuse cycle case.
5897 2017-07-27 Richard Biener <rguenther@suse.de>
5899 PR tree-optimization/81571
5900 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
5903 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
5905 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
5906 earlier and only if MASK_FPU is set. Adjust formatting.
5908 2017-07-27 Martin Liska <mliska@suse.cz>
5910 * opt-functions.awk: Add validation of value of Init.
5911 * optc-gen.awk: Pass new argument.
5913 2017-07-27 Martin Liska <mliska@suse.cz>
5915 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
5916 Fix wrong condition.
5918 2017-07-27 Martin Liska <mliska@suse.cz>
5920 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
5921 BBs and edges seen by autoFDO.
5923 2017-07-27 Richard Biener <rguenther@suse.de>
5925 PR tree-optimization/81502
5926 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
5927 with incompatible but same sized type.
5928 (execute_update_addresses_taken): Likewise.
5930 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
5932 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
5933 flag_tree_loop_vectorize rather than flag_tree_vectorize.
5935 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5938 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
5939 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
5940 Change s_operand to memory_operand.
5942 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
5944 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
5945 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
5946 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
5947 Emit instructions rather than returning an expression. Handle TFmode
5948 and KFmode by casting to TImode.
5949 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
5950 (rs6000_emit_le_vsx_store): Likewise.
5951 * config/rs6000/vsx.md (VSX_TI): New iterator.
5952 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
5953 (*vsx_le_undo_permute_<mode>): Likewise.
5954 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
5955 emit the split sequence.
5956 (*vsx_le_perm_store_<mode>): Likewise.
5958 2017-07-27 Jakub Jelinek <jakub@redhat.com>
5960 PR tree-optimization/81555
5961 PR tree-optimization/81556
5962 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
5963 if true, force CHANGED for the recursive invocation.
5964 (reassociate_bb): Remember original length of ops array, pass
5965 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
5967 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
5968 attributes for noipa attribute. For naked attribute use
5969 lookup_attribute first before lookup_attribute_spec.
5970 * final.c (rest_of_handle_final): Disable IPA RA for functions with
5972 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
5973 for functions with noipa attribute.
5974 (cgraph_externally_visible_p): Return true for functions with noipa
5976 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
5977 for functions with noipa attribute.
5978 * doc/extend.texi: Document noipa function attribute.
5979 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
5980 also for functions with noipa attribute.
5981 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
5983 2017-07-26 Andrew Pinski <apinski@cavium.com>
5985 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
5986 vec_unalign_load_cost and vec_unalign_store_cost.
5988 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5990 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
5991 -mvsx-small-integer option.
5992 (ISA_3_0_MASKS_IEEE): Likewise.
5993 (OTHER_VSX_VECTOR_MASKS): Likewise.
5994 (POWERPC_MASKS): Likewise.
5995 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
5996 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
5997 code, only testing for DImode being allowed in non-VSX floating
5999 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6000 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6002 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6003 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6004 TARGET_P8_VECTOR test.
6005 (rs6000_secondary_reload_simple_move): Likewise.
6006 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6007 since TARGET_P9_VECTOR was already tested.
6008 (rs6000_opt_masks): Remove -mvsx-small-integer.
6009 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6010 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6012 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6013 test for TARGET_VEXTRACTUB was used, and that uses
6015 (p9 extract splitter): Likewise.
6016 (vsx_extract_<mode>_di_p9): Likewise.
6017 (vsx_extract_<mode>_store_p9): Likewise.
6018 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6019 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6020 the elimination of TARGET_VSX_SMALL_INTEGER.
6021 (vsx_extract_<mode>_p8): Likewise.
6022 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6023 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6024 (vsx_set_<mode>_p9): Likewise.
6025 (vsx_set_v4sf_p9): Likewise.
6026 (vsx_set_v4sf_p9_zero): Likewise.
6027 (vsx_insert_extract_v4sf_p9): Likewise.
6028 (vsx_insert_extract_v4sf_p9_2): Likewise.
6029 * config/rs6000/rs6000.md (sign extend splitter): Change
6030 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6031 (floatsi<mode>2_lfiwax_mem): Likewise.
6032 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6033 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6034 since a test for TARGET_P9_VECTOR was used.
6035 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6036 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6037 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6038 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6039 TARGET_P8_VECTOR test.
6040 (fix_trunc<mode>si2_stfiwx): Likewise.
6041 (fix_trunc<mode>si2_internal): Likewise.
6042 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6043 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6045 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6046 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6047 TARGET_P8_VECTOR test.
6048 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6049 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6050 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6052 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6053 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6054 since a test for TARGET_P9_VECTOR was used.
6055 (splitter for loading small constants): Likewise.
6057 2017-07-26 Andrew Pinski <apinski@cavium.com>
6059 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6062 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6065 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6066 (fp_valid_at): Likewise.
6068 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6070 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6071 (qdf24xx_addrcost_table): Likewise.
6072 (cortexa57_tunings): Update to use generic_branch_cost.
6073 (cortexa72_tunings): Likewise.
6074 (cortexa73_tunings): Likewise.
6075 (qdf24xx_tunings): Likewise.
6077 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6079 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6080 (thunderx2t99_branch_cost): Likewise.
6081 (cortexa35_tunings): Update to use generic_branch_cost.
6082 (cortexa53_tunings): Likewise.
6083 (cortexa57_tunings): Likewise.
6084 (cortexa72_tunings): Likewise.
6085 (cortexa73_tunings): Likewise.
6086 (thunderx2t99_tunings): Likewise.
6088 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6090 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6091 (sparc_option_override): Honour MASK_FSMULD.
6092 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6093 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6094 * config/sparc/sparc.opt (mfsmuld): New option.
6095 * doc/invoke.texi (mfsmuld): Document option.
6097 2017-07-26 Marek Polacek <polacek@redhat.com>
6100 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
6102 2017-07-26 Richard Biener <rguenther@suse.de>
6104 * gimple-match-head.c (do_valueize): Return OP if valueize
6106 (get_def): New helper to get at the def stmt of a SSA name
6108 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
6109 do_valueize to get at the def stmt.
6110 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
6112 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
6115 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
6117 2017-07-26 Martin Liska <mliska@suse.cz>
6120 * function.c (expand_function_start): Make expansion of
6121 nonlocal_goto_save_area after parm_birth_insn.
6123 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6125 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
6126 from all CPU target flags enable members.
6128 2017-07-26 Richard Biener <rguenther@suse.de>
6130 * genmatch.c (dt_simplify::gen): Make iterator vars const.
6131 (decision_tree::gen): Make 'type' const.
6132 (write_predicate): Likewise.
6134 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6136 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
6137 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
6138 (rs6000_option_override_internal): Likewise.
6139 (rs6000_expand_vector_set): Likewise.
6140 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
6141 (TARGET_UPPER_REGS_SF): Likewise.
6142 (TARGET_UPPER_REGS_DI): Likewise.
6143 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
6144 (TARGET_DIRECT_MOVE_64BIT): Likewise.
6145 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
6146 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6147 (Splitters for DI constants in Altivec registers): Likewise.
6148 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
6149 (vsx_set_v4sf_p9): Likewise.
6150 (vsx_set_v4sf_p9_zero): Likewise.
6151 (vsx_insert_extract_v4sf_p9): Likewise.
6152 (vsx_insert_extract_v4sf_p9_2): Likewise.
6154 2017-07-25 Carl Love <cel@us.ibm.com>
6156 * doc/extend.texi: Update the built-in documentation file for the
6157 existing built-in functions
6158 vector signed char vec_cnttz (vector signed char);
6159 vector unsigned char vec_cnttz (vector unsigned char);
6160 vector signed short vec_cnttz (vector signed short);
6161 vector unsigned short vec_cnttz (vector unsigned short);
6162 vector signed int vec_cnttz (vector signed int);
6163 vector unsigned int vec_cnttz (vector unsigned int);
6164 vector signed long long vec_cnttz (vector signed long long);
6165 vector unsigned long long vec_cnttz (vector unsigned long long);
6167 2017-07-25 Andrew Pinski <apinski@cavium.com>
6169 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
6170 accesses where the use is for the first operand of a BIT_INSERT.
6172 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6175 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6176 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6178 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6180 * config/i386/gstabs.h: Delete.
6181 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
6183 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
6185 * config/i386/i386.c (ix86_decompose_address): Do not check for
6186 register RTX when looking at index_reg or base_reg.
6187 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
6189 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
6191 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
6192 to update EH info here.
6194 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6196 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
6198 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6200 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
6202 2017-07-25 Torsten Duwe <duwe@suse.de>
6204 * common.opt: Introduce -fpatchable-function-entry
6205 command line option, and its variables function_entry_patch_area_size
6206 and function_entry_patch_area_start.
6207 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
6208 including a two-value parser.
6209 * target.def (print_patchable_function_entry): New target hook.
6210 * targhooks.h (default_print_patchable_function_entry): New function.
6211 * targhooks.c (default_print_patchable_function_entry): Likewise.
6212 * toplev.c (process_options): Switch off IPA-RA if
6213 patchable function entries are being generated.
6214 * varasm.c (assemble_start_function): Look at the
6215 patchable-function-entry command line switch and current
6216 function attributes and maybe generate NOP instructions by
6217 calling the print_patchable_function_entry hook.
6218 * doc/extend.texi: Document patchable_function_entry attribute.
6219 * doc/invoke.texi: Document -fpatchable_function_entry
6220 command line option.
6221 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
6223 * doc/tm.texi: Re-generate.
6225 2017-07-25 Jakub Jelinek <jakub@redhat.com>
6228 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
6229 TARGET_AVX512DQ rather than TARGET_AVX512BW.
6231 2017-07-25 Tamar Christina <tamar.christina@arm.com>
6233 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
6235 2017-07-25 Richard Biener <rguenther@suse.de>
6237 PR tree-optimization/81455
6238 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
6239 not walk in cycles when looking for guards.
6241 2017-07-25 Richard Biener <rguenther@suse.de>
6243 PR tree-optimization/81529
6244 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
6245 when optimizing backedge uses.
6247 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
6249 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
6251 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
6252 to dl_section_ref. On AIX, append an expression to subtract
6253 the size of the section length to dl_section_ref.
6255 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
6257 * configure.ac: If any of the config.* scripts fail, exit 1.
6258 * configure: Regenerate.
6260 2017-07-25 Richard Biener <rguenther@suse.de>
6263 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
6264 of immediate uses, be more verbose on errors.
6266 2017-07-25 Richard Biener <rguenther@suse.de>
6268 PR tree-optimization/81510
6269 * tree-vect-loop.c (vect_is_simple_reduction): When the
6270 reduction stmt is not inside the loop bail out.
6272 2017-07-25 Richard Biener <rguenther@suse.de>
6274 PR tree-optimization/81303
6275 * tree-vect-loop-manip.c (vect_loop_versioning): Build
6276 profitability check against LOOP_VINFO_NITERSM1.
6278 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6280 * domwalk.c (cmp_bb_postorder): Simplify.
6281 (sort_bbs_postorder): New function. Use it...
6282 (dom_walker::walk): ...here to optimize common cases.
6284 2017-07-25 Martin Liska <mliska@suse.cz>
6287 * ipa-visibility.c (function_and_variable_visibility): Make the
6288 redirection just on target that supports aliasing.
6289 Fix GNU coding style.
6291 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
6294 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
6296 * config/arm/rtems.h: Likewise.
6297 * config/bfin/rtems.h: Likewise.
6298 * config/i386/rtemself.h: Likewise.
6299 * config/lm32/rtems.h: Likewise.
6300 * config/m32c/rtems.h: Likewise.
6301 * config/m68k/rtemself.h: Likewise.
6302 * config/microblaze/rtems.h: Likewise.
6303 * config/mips/rtems.h: Likewise.
6304 * config/moxie/rtems.h: Likewise.
6305 * config/nios2/rtems.h: Likewise.
6306 * config/powerpcspe/rtems.h: Likewise.
6307 * config/rs6000/rtems.h: Likewise.
6308 * config/rtems.h: Likewise.
6309 * config/sh/rtems.h: Likewise.
6310 * config/sh/rtemself.h: Likewise.
6311 * config/sparc/rtemself.h: Likewise.
6313 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6316 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
6317 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
6318 * tree-ssa-structalias.c (alias_get_name): Same.
6320 2017-07-25 Bin Cheng <bin.cheng@arm.com>
6323 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
6324 instructions if no du chain is found.
6326 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6328 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
6330 2017-07-25 Richard Biener <rguenther@suse.de>
6333 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
6336 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6338 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
6340 (ISA_2_7_MASKS_SERVER): Likewise.
6341 (ISA_3_0_MASKS_IEEE): Likewise.
6342 (OTHER_P8_VECTOR_MASKS): Likewise.
6343 (OTHER_VSX_VECTOR_MASKS): Likewise.
6344 (POWERPC_MASKS): Likewise.
6345 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
6346 duplicate list of options.
6347 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
6348 explicit -mupper-regs options.
6349 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
6350 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
6351 alias for -mupper-regs-df.
6352 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6353 (rs6000_init_hard_regno_mode_ok): Likewise.
6354 (rs6000_option_override_internal): Likewise.
6355 (rs6000_opt_masks): Likewise.
6356 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
6357 options in terms of whether -mvsx or -mpower8-vector was used.
6358 (TARGET_UPPER_REGS_DI): Likewise.
6359 (TARGET_UPPER_REGS_SF): Likewise.
6360 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
6361 -mupper-regs-* options.
6363 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6365 * passes.c (emergency_dump_function): Print some empty lines and a
6366 header before the RTL dump.
6368 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6370 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
6372 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
6375 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6376 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
6378 2017-07-24 Carl Love <cel@us.ibm.com>
6380 * config/rs6000/rs6000-c.c: Add support for built-in functions
6381 vector float vec_extract_fp32_from_shorth (vector unsigned short);
6382 vector float vec_extract_fp32_from_shortl (vector unsigned short);
6383 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
6384 vec_extract_fp_from_shortl): Add defines for the two builtins.
6385 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
6386 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
6388 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
6389 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
6390 * doc/extend.texi: Update the built-in documentation file for the
6391 new built-in function.
6393 2017-07-24 Jakub Jelinek <jakub@redhat.com>
6396 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
6398 * doc/generic.texi: Likewise.
6399 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6400 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6402 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
6404 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
6405 (aarch64_mls_elt_merge<mode>): Likewise.
6407 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
6409 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6410 having __cxa_atexit.
6412 2017-07-23 Michael Collison <michael.collison@arm.com>
6414 * config/arm/arm.c (arm_option_override): Deprecate
6415 use of -mstructure-size-boundary.
6416 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
6417 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
6419 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6422 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
6423 Reduce cost estimate for direct moves.
6425 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
6428 * config/i386/i386.c (ix86_option_override_internal): Disable
6429 BMI, BMI2 and TBM instructions for -m16.
6431 2017-07-21 Carl Love <cel@us.ibm.com>
6433 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6434 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6435 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6436 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6437 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6438 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6439 VMULOSW): New enum "unspec" values.
6440 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6441 altivec_vmulosw): New patterns.
6442 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6443 VMULOSW): Add definitions.
6445 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
6447 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
6448 (qdf24xx): Likewise.
6449 * config/aarch64/aarch64-options-extensions.def (rdma); New.
6450 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
6451 (AARCH64_FL_V8_1): Renumber.
6452 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
6453 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
6454 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
6455 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
6456 rdma to feature modifiers list.
6458 2017-07-21 Yury Gribov <tetra2005@gmail.com>
6461 * ipa-visibility (function_and_variable_visibility): Convert
6462 recursive PLT call to direct call if appropriate.
6464 2017-07-21 Andrew Pinski <apinski@cavium.com>
6466 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
6467 operand 1 to see if the types precision matches.
6468 * fold-const.c (operand_equal_p): Likewise.
6470 2017-07-21 Richard Biener <rguenther@suse.de>
6472 PR tree-optimization/81303
6473 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
6474 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
6475 (vect_peeling_hash_get_lowest_cost): Adjust.
6476 (vect_enhance_data_refs_alignment): Likewise. Use
6477 vect_get_peeling_costs_all_drs to compute the penalty for no
6478 peeling to match up costs.
6480 2017-07-21 Richard Biener <rguenther@suse.de>
6482 PR tree-optimization/81500
6483 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
6484 we didn't identify a reduction path.
6486 2017-07-21 Tom de Vries <tom@codesourcery.com>
6487 Cesar Philippidis <cesar@codesourcery.com>
6489 PR gcov-profile/81442
6490 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
6493 2017-07-21 Tom de Vries <tom@codesourcery.com>
6496 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
6498 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
6499 nvptx_override_options_after_change.
6501 2017-07-21 Ulrich Drepper <drepper@redhat.com>
6503 * dwarf2out.c (output_file_names): Avoid double testing for
6506 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
6508 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
6510 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6512 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
6514 (try_crossjump_to_edge): Do not punt on partitioned functions.
6516 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6518 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
6519 Put all BBs reachable only via paths crossing cold region to cold
6521 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
6523 2016-07-21 Richard Biener <rguenther@suse.de>
6525 PR tree-optimization/81303
6526 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
6527 into account prologue and epilogue iterations when raising
6528 min_profitable_iters to sth at least covering one vector iteration.
6530 2017-07-21 Tamar Christina <tamar.christina@arm.com>
6532 * config/arm/arm.c (arm_test_cpu_arch_dat):
6535 2017-07-20 Nathan Sidwell <nathan@acm.org>
6537 Remove TYPE_METHODS.
6538 * tree.h (TYPE_METHODS): Delete.
6539 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
6540 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
6541 (dbxout_type_methods): Scan TYPE_FIELDS.
6542 (dbxout_type): Don't check TYPE_METHODS here.
6543 * function.c (use_register_for_decl): Always ignore register for
6544 class types when not optimizing.
6545 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
6546 * tree.c (free_lang_data_in_type): Stitch out member functions and
6547 templates from TYPE_FIELDS.
6548 (build_distinct_type_copy, verify_type_variant,
6549 verify_type): Member fns are on TYPE_FIELDS.
6550 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
6551 * tree-pretty-print.c (dump_generic_node): Likewise.
6553 2017-07-20 Jakub Jelinek <jakub@redhat.com>
6556 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
6557 V2TImode and V4TImode.
6558 (ix86_expand_vector_extract): Likewise.
6559 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
6560 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
6561 (ssescalarmode): Handle V4TImode and V2TImode.
6562 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
6563 (*vec_extractv2ti, *vec_extractv4ti): New insns.
6564 (VEXTRACTI128_MODE): New mode iterator.
6565 (splitter for *vec_extractv?ti first element): New.
6566 (VEC_INIT_MODE): New mode iterator.
6567 (vec_init<mode>): Consolidate 3 expanders into one using
6568 VEC_INIT_MODE mode iterator.
6570 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6572 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
6573 non_spilled_static_chain_regno_p.
6575 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6577 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
6579 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
6581 * bb-reorder.c (connect_traces): Allow copying of blocks within
6584 2017-07-20 Richard Biener <rguenther@suse.de>
6586 * gimple.h (gimple_phi_result): Add gphi * overload.
6587 (gimple_phi_result_ptr): Likewise.
6588 (gimple_phi_arg): Likewise. Adjust index assert to only
6589 allow actual argument accesses rather than all slots available
6591 (gimple_phi_arg_def): Add gphi * overload.
6592 * tree-phinodes.c (make_phi_node): Initialize only actual
6594 (resize_phi_node): Clear memory not covered by old node,
6595 do not initialize excess argument slots.
6596 (reserve_phi_args_for_new_edge): Initialize new argument slot
6599 2017-07-20 Bin Cheng <bin.cheng@arm.com>
6601 PR tree-optimization/81388
6603 2016-07-21 Bin Cheng <bin.cheng@arm.com>
6605 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
6606 by removing computation of may_be_zero.
6608 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6609 Tom de Vries <tom@codesourcery.com>
6612 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
6613 when gimple level profile disagrees with what RTL expander did.
6615 2017-07-20 Richard Biener <rguenther@suse.de>
6617 PR tree-optimization/61171
6618 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
6619 (vect_analyze_stmt): Add slp instance parameter.
6620 (vectorizable_reduction): Likewise.
6621 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
6622 (vect_is_simple_reduction): Deal with chains not detected
6623 as SLP reduction chain, specifically not properly associated
6624 chains containing a mix of plus/minus.
6625 (get_reduction_op): Remove.
6626 (get_initial_defs_for_reduction): Simplify, pass in whether
6627 this is a reduction chain, pass in the SLP node for the PHIs.
6628 (vect_create_epilog_for_reduction): Get the SLP instance as
6630 (vectorizable_reduction): Get the SLP instance as arg.
6631 During analysis remember the SLP node with the PHIs in the
6632 instance. Simplify getting at the vectorized reduction PHIs.
6633 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
6634 through SLP instance.
6635 (vect_slp_analyze_operations): Likewise.
6636 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
6637 (vect_transform_stmt): Likewise.
6639 2017-07-20 Tom de Vries <tom@codesourcery.com>
6641 PR tree-optimization/81489
6642 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
6643 read of phi arg location to before loop that modifies phi.
6645 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
6647 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
6650 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6653 * except.c (execute): Fix ordering issue.
6655 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6657 PR rtl-optimization/81423
6658 * combine.c (make_compound_operation_int): Don't try to optimize
6659 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
6661 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6663 PR rtl-optimization/81423
6664 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
6665 with a constant that is -1 in the truncated to mode.
6667 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6669 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
6670 (determine_unlikely_bbs): ... here.
6671 * predict.h (propagate_unlikely_bbs_forward): Declare.
6672 * cfgexpand.c (pass_expand::execute): Use it.
6673 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
6675 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
6676 propagate_unlikely_bbs_forward.
6678 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6681 * except.c (maybe_add_nop_after_section_switch): New function.
6684 2017-07-19 Tom de Vries <tom@codesourcery.com>
6686 * gimple.h (gimple_phi_set_arg): Make assert more strict.
6688 2017-07-19 Tom de Vries <tom@codesourcery.com>
6690 * gimple.h (gimple_phi_arg): Make assert more strict.
6692 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
6694 * config.gcc (powerpc*-*-*): Add mmintrin.h.
6695 * config/rs6000/mmintrin.h: New file.
6696 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
6698 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6700 PR tree-optimization/81346
6701 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
6703 2017-07-19 Tom de Vries <tom@codesourcery.com>
6705 * config/nvptx/nvptx.md (VECIM): Add V2DI.
6707 2017-07-19 Tom de Vries <tom@codesourcery.com>
6709 * config/nvptx/nvptx-modes.def: Add V2DImode.
6710 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
6711 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
6712 (nvptx_output_mov_insn): Handle lack of mov.b128.
6713 (nvptx_print_operand): Handle 'H' and 'L' codes.
6714 (nvptx_vector_mode_supported): Allow V2DImode.
6715 (nvptx_preferred_simd_mode): New function.
6716 (nvptx_data_alignment): New function.
6717 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
6718 nvptx_preferred_simd_mode.
6719 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
6721 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
6723 2017-07-19 Tom de Vries <tom@codesourcery.com>
6725 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
6726 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
6727 (nvptx_vector_mode_supported): New function. Allow V2SImode.
6728 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
6729 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
6730 (mov<VECIM>_insn): New define_insn.
6731 (define_expand "mov<VECIM>): New define_expand.
6733 2017-07-19 Tom de Vries <tom@codesourcery.com>
6735 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
6737 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6739 PR tree-optimization/81346
6740 * fold-const.h (fold_div_compare, range_check_type): Declare.
6741 * fold-const.c (range_check_type): New function.
6742 (build_range_check): Use range_check_type.
6743 (fold_div_compare): No longer static, rewritten into
6744 a match.pd helper function.
6745 (fold_comparison): Don't call fold_div_compare here.
6746 * match.pd (X / C1 op C2): New optimization using fold_div_compare
6749 2017-07-19 Nathan Sidwell <nathan@acm.org>
6751 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
6752 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
6753 * tree.c (find_decls_types_r, verify_type): Use
6754 TYPE_{MIN,MAX}_VALUE_RAW.
6755 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
6756 (hash_tree): Likewise.
6757 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6759 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6762 2017-07-18 Tom de Vries <tom@codesourcery.com>
6765 * omp-expand.c (expand_omp_for_static_chunk): Handle
6766 equal-argument loop exit phi.
6768 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
6771 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
6772 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
6773 operand 2 predicate.
6774 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
6775 operand 2 predicate.
6776 (ror,rol -> rorx splitters): Use const_int_operand as
6777 operand 2 predicate.
6779 2017-06-18 Richard Biener <rguenther@suse.de>
6781 PR tree-optimization/81410
6782 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
6783 the gap in the ! slp_perm SLP case after each group.
6785 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6788 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
6791 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6794 * predict.c (set_even_probabilities): Cleanup; do not affect
6795 probabilities that are already known.
6796 (combine_predictions_for_bb): Call even when count is set.
6798 2017-07-18 Nathan Sidwell <nathan@acm.org>
6800 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
6803 2017-07-18 Bin Cheng <bin.cheng@arm.com>
6806 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
6807 optimization for loop niter analysis.
6809 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
6812 * config/avr/avr.c (avr_optimize_casesi): Don't use
6813 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
6815 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6817 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
6818 body_cost_vec from _vect_peel_extended_info.
6819 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
6820 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
6823 2017-07-18 Bin Cheng <bin.cheng@arm.com>
6825 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
6827 2017-07-18 Richard Biener <rguenther@suse.de>
6829 PR tree-optimization/80620
6830 PR tree-optimization/81403
6831 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
6832 info when re-using a VN table entry.
6834 2017-07-18 Richard Biener <rguenther@suse.de>
6836 PR tree-optimization/81418
6837 * tree-vect-loop.c (vectorizable_reduction): Properly compute
6838 vectype_in. Verify that with lane-reducing reduction operations
6839 we have a single def-use cycle.
6841 2017-07-17 Carl Love <cel@us.ibm.com>
6843 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
6845 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6846 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6847 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6848 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6849 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6850 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6851 VMULOSW): New enum "unspec" values.
6852 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
6853 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
6854 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6855 altivec_vmulosw): New patterns.
6856 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6857 VMULOSW): Add definitions.
6858 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
6860 * config/alpha/alpha.c: Include predict.h.
6862 2017-07-17 Yury Gribov <tetra2005@gmail.com>
6864 * tree-vrp.c (compare_assert_loc): Fix comparison function
6865 to return predictable results.
6867 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6869 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
6872 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
6873 * doc/invoke.texi (mexpand-adddi): Update text.
6875 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6877 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
6878 that also clobbers the CC register. The old expand code is moved
6880 (*arc_clzsi2): ... here.
6881 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
6882 the CC register. The old expand code is moved to ...
6883 (arc_ctzsi2): ... here.
6885 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6887 * config/arc/arc.opt (mindexed-loads): Use initial value
6888 TARGET_INDEXED_LOADS_DEFAULT.
6889 (mauto-modify-reg): Use initial value
6890 TARGET_AUTO_MODIFY_REG_DEFAULT.
6891 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
6892 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
6893 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
6894 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
6896 2017-07-17 Martin Liska <mliska@suse.cz>
6899 * opts.c (finish_options): Do not allow -fgnu-tm
6900 w/ -fsanitize={kernel-,}address. Say sorry.
6902 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6905 * tree-loop-distribution.c (classify_partition): Only assert on
6906 numer of iterations.
6907 (merge_dep_scc_partitions): Delete prameter. Update function call.
6908 (distribute_loop): Remove code handling loop with unknown niters.
6909 (pass_loop_distribution::execute): Skip loop with unknown niters.
6911 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6914 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
6915 function sort_partitions_by_post_order.
6917 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6919 PR tree-optimization/81374
6920 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
6921 the max index of basic blocks, rather than number of basic blocks.
6923 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6925 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
6927 (arc_legitimate_pic_operand_p): Likewise.
6928 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
6930 (arc_needs_pcl_p): Likewise.
6931 (arc_legitimate_pc_offset_p): Likewise.
6932 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
6933 function is also used in constrains.md.
6934 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
6935 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
6936 PLUS. Only return true/false in known cases, otherwise assert.
6937 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
6938 is already called in arc_legitimate_constant_p.
6939 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
6941 (LEGITIMATE_PIC_OPERAND_P): Use
6942 arc_raw_symbolic_reference_mentioned_p function.
6943 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
6948 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6949 Andrew Burgess <andrew.burgess@embecosm.com>
6951 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
6952 (arc_return_address_register): New function.
6953 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
6954 (arc_handle_fndecl_attribute): Add naked attribute.
6955 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6956 (TARGET_WARN_FUNC_RETURN): Likewise.
6957 (arc_allocate_stack_slots_for_args): New function.
6958 (arc_warn_func_return): Likewise.
6959 (machine_function): Change type fn_type.
6960 (arc_compute_function_type): Consider new naked function type,
6961 change function return type.
6962 (arc_must_save_register): Adapt to handle new
6963 arc_compute_function_type's return type.
6964 (arc_expand_prologue): Likewise.
6965 (arc_expand_epilogue): Likewise.
6966 (arc_return_address_regs): Delete.
6967 (arc_return_address_register): New function.
6968 (arc_epilogue_uses): Use above function.
6969 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
6970 (arc_function_type): Change encoding, add naked type.
6971 (ARC_INTERRUPT_P): Change to handle the new encoding.
6972 (ARC_FAST_INTERRUPT_P): Likewise.
6973 (ARC_NORMAL_P): Define.
6974 (ARC_NAKED_P): Likewise.
6975 (arc_compute_function_type): Delete prototype.
6976 * config/arc/arc.md (in_ret_delay_slot): Use
6977 arc_return_address_register function.
6978 (simple_return): Likewise.
6979 (p_return_i): Likewise.
6981 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6983 PR tree-optimization/81428
6984 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
6985 can't be built for those types.
6987 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
6989 Remove stuff dead since r239246.
6991 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
6992 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
6993 (avr_inform_devices): Remove dead stuff.
6995 2017-07-17 Tamar Christina <tamar.christina@arm.com>
6997 * config/arm/arm_neon.h: Fix softp typo.
6999 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7001 PR tree-optimization/81365
7002 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7003 aggregate moves onto bb predecessor edges, make sure there are no
7004 loads that could alias the lhs in between the start of bb and the
7007 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7010 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7011 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7012 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7014 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7016 PR tree-optimization/81396
7017 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7018 (init_symbolic_number): Initialize it to 1.
7019 (perform_symbolic_merge): Add n_ops from both operands into the new
7021 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7022 without base_addr as useless if they need more than one operation.
7023 (bswap_replace): Handle !bswap case for NULL base_addr.
7025 2017-07-17 Tom de Vries <tom@codesourcery.com>
7028 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7031 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7033 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7034 conditional builtin define __FIX_LEON3FT_B2BST.
7036 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7038 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7039 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7042 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7044 PR rtl-optimization/81424
7045 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7046 to remove potential trapping from operands if -fnon-call-exceptions.
7048 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7050 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7051 profile_proability for scalling.
7052 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7054 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7056 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7058 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7060 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7061 fixpoint arithmetics.
7063 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7065 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7066 fixpoint arithmetics.
7068 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7070 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7071 fixpoint arithmetics.
7073 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7075 * profile-count.h (profile_probability::from_reg_br_prob_note,
7076 profile_probability::to_reg_br_prob_note): New functions.
7077 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7078 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7079 * predict.c (probability_reliable_p): Update.
7080 (edge_probability_reliable_p): Update.
7081 (br_prob_note_reliable_p): Update.
7082 (invert_br_probabilities): Update.
7083 (add_reg_br_prob_note): New function.
7084 (combine_predictions_for_insn): Update.
7085 * asan.c (asan_clear_shadow): Update.
7086 * cfgbuild.c (compute_outgoing_frequencies): Update.
7087 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7088 (update_br_prob_note): Update.
7089 (rtl_verify_edges): Update.
7090 (purge_dead_edges): Update.
7091 (fixup_reorder_chain): Update.
7092 * emit-rtl.c (try_split): Update.
7093 * ifcvt.c (cond_exec_process_insns): Update.
7094 (cond_exec_process_if_block): Update.
7095 (dead_or_predicable): Update.
7096 * internal-fn.c (expand_addsub_overflow): Update.
7097 (expand_neg_overflow): Update.
7098 (expand_mul_overflow): Update.
7099 * loop-doloop.c (doloop_modify): Update.
7100 * loop-unroll.c (compare_and_jump_seq): Update.
7101 * optabs.c (emit_cmp_and_jump_insn_1): Update.
7102 * predict.h: Update.
7103 * reorg.c (mostly_true_jump): Update.
7105 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
7106 * config/alpha/alpha.c (emit_unlikely_jump): Update.
7107 * config/arc/arc.c: (emit_unlikely_jump): Update.
7108 * config/arm/arm.c: (emit_unlikely_jump): Update.
7109 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
7110 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
7111 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
7112 (ix86_print_operand): Update.
7113 (ix86_split_fp_branch): Update.
7114 (predict_jump): Update.
7115 * config/ia64/ia64.c (ia64_print_operand): Update.
7116 * config/mmix/mmix.c (mmix_print_operand): Update.
7117 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
7118 (rs6000_expand_split_stack_prologue): Update.
7119 * config/rs6000/rs6000.c: Update.
7120 * config/s390/s390.c (s390_expand_vec_strlen): Update.
7121 (s390_expand_vec_movstr): Update.
7122 (s390_expand_cs_tdsi): Update.
7123 (s390_expand_split_stack_prologue): Update.
7124 * config/sh/sh.c (sh_print_operand): Update.
7125 (expand_cbranchsi4): Update.
7126 (expand_cbranchdi4): Update.
7127 * config/sparc/sparc.c (output_v9branch): Update.
7128 * config/spu/spu.c (get_branch_target): Update.
7129 (ea_load_store_inline): Update.
7130 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
7131 * config/tilepro/tilepro.c: Update.
7133 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7135 * gimplify.c (mostly_copy_tree_r): Revert latest change.
7136 (gimplify_save_expr): Likewise.
7138 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7140 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
7142 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7144 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
7146 * timevar.def (TV_IPA_FNSUMMARY): Define.
7148 2017-07-16 Daniel Cederman <cederman@gaisler.com>
7150 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
7151 to back store errata sensitive sequence from being generated.
7152 (sqrtdf2_fix): Likewise.
7154 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7156 * tree-ssa-threadupdate.c (compute_path_counts,
7157 update_joiner_offpath_counts): Use profile_probability.
7159 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
7162 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7164 * config/arm/arm-c.c (arm_cpu_builtins): Define
7165 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7167 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
7169 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7170 array entries to represent __ieee128 versions of the
7171 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
7172 scalar_extract_sig, and scalar_insert_exp built-in functions.
7173 (altivec_resolve_overloaded_builtin): Add special case handling
7174 for the __builtin_scalar_insert_exp function, as represented by
7175 the P9V_BUILTIN_VEC_VSIEDP constant.
7176 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
7177 exponent support for __ieee128 argument.
7178 (VSESQP): Add scalar extract signature support for __ieee128
7180 (VSTDCNQP): Add scalar test negative support for __ieee128
7182 (VSIEQP): Add scalar insert exponent support for __int128 argument
7183 with __ieee128 result.
7184 (VSIEQPF): Add scalar insert exponent support for __ieee128
7185 argument with __ieee128 result.
7186 (VSTDCQP): Add scalar test data class support for __ieee128
7188 (VSTDCNQP): Add overload support for scalar test negative with
7190 (VSTDCQP): Add overload support for scalar test data class
7192 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
7194 (UNSPEC_VSX_SIEXPQP): New constant.
7195 (xsxexpqp): New insn for VSX scalar extract exponent quad
7197 (xsxsigqp): New insn for VSX scalar extract significand quad
7199 (xsiexpqpf): New insn for VSX scalar insert exponent quad
7200 precision with floating point argument.
7201 (xststdcqp): New expand for VSX scalar test data class quad
7203 (xststdcnegqp): New expand for VSX scalar test negative quad
7205 (xststdcqp): New insn to match expansions for VSX scalar test data
7206 class quad precision and VSX scalar test negative quad precision.
7207 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7208 special case operand checking to enforce that second operand of
7209 VSX scalar test data class with quad precision argument is a 7-bit
7211 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7212 prototypes and descriptions of __ieee128 versions of
7213 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
7214 scalar_test_data_class, and scalar_test_neg built-in functions.
7216 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7218 PR tree-optimization/81162
7219 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
7220 replace a negate with an add.
7222 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
7224 * doc/invoke.texi (arm/-mcpu): Document +crypto.
7226 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7228 * config/arm/arm-c.c (arm_cpu_builtins): Define
7229 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7231 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7233 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
7234 (armv8-r): Set ARM Cortex-R52 as default CPU.
7235 * config/arm/arm-tables.opt: Regenerate.
7236 * config/arm/arm-tune.md: Regenerate.
7237 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7239 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
7240 extension for -mcpu=cortex-r52.
7242 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7244 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
7245 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
7246 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
7247 (fp-armv8): Define it as FP_ARMv8 only.
7248 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
7249 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
7251 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
7252 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
7253 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
7254 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
7255 than TARGET_FPU_ARMV8.
7256 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
7257 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
7258 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
7260 * config/arm/neon.md (neon_vrint): Likewise.
7261 (neon_vcvt): Likewise.
7262 (neon_<fmaxmin_op><mode>): Likewise.
7263 (<fmaxmin><mode>3): Likewise.
7264 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
7265 * config/arm/predicates.md (arm_cond_move_operator): Check against
7266 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
7268 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
7270 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
7273 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7275 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
7276 loop in comment with memset.
7278 2017-07-14 Martin Liska <mliska@suse.cz>
7280 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
7281 * dwarf2out.c (is_java): Remove the function.
7282 (output_pubname): Remove usage of the function.
7283 (lower_bound_default): Remove usage of DW_LANG_Java.
7284 (gen_compile_unit_die): Likewise.
7285 * gcc.c: Remove compiler defaults for .java and .zip files.
7286 * gimple-expr.c (remove_suffix): Change as there's no longer
7287 extension than 4-letter one.
7288 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
7289 (gimplify_save_expr): Likewise.
7290 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
7291 as it's possible even for other languages than Java.
7292 * langhooks.h (struct lang_hooks): Remove Java from a comment.
7293 * lto-opts.c (lto_write_options): Remove reference to Java.
7294 * opts.c (strip_off_ending): Update file extension handling.
7295 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
7296 * tree-eh.c (lower_resx): Likewise.
7297 * tree.c (free_lang_data_in_type): Remove dead code.
7298 (find_decls_types_r): Likewise.
7299 (build_common_builtin_nodes): Remove Java from a comment.
7300 (verify_type): Remove dead code.
7301 * varasm.c (assemble_external): Remove Java from a comment.
7303 2017-07-14 Martin Liska <mliska@suse.cz>
7305 * opts.c (finish_options): Add quotes.
7306 (common_handle_option): Likewise.
7308 2017-07-14 Martin Liska <mliska@suse.cz>
7310 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
7311 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
7313 * dwarf2out.c (lower_bound_default): Do not handle
7315 (gen_compile_unit_die): Likewise.
7316 * gcc.c: Remove default extension binding for GNU Pascal.
7317 * stmt.c: Remove Pascal language from a comment.
7318 * xcoffout.c: Likewise.
7320 2017-07-13 David Malcolm <dmalcolm@redhat.com>
7323 * diagnostic-show-locus.c (fixit_cmp): New function.
7324 (layout::layout): Sort m_fixit_hints.
7325 (column_range::column_range): Assert that the values are valid.
7326 (struct char_span): New struct.
7327 (correction::overwrite): New method.
7328 (struct source_line): New struct.
7329 (line_corrections::add_hint): Add assertions. Reimplement memcpy
7330 calls in terms of classes source_line and char_span, and
7331 correction::overwrite.
7332 (selftest::test_overlapped_fixit_printing_2): New function.
7333 (selftest::diagnostic_show_locus_c_tests): Call it.
7335 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7337 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
7338 early if there is no lhs.
7340 2017-07-13 Martin Liska <mliska@suse.cz>
7342 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
7343 (gen_reference_type_die): Likewise.
7344 * stor-layout.c: Remove Pascal-related comment.
7346 2017-07-13 Martin Liska <mliska@suse.cz>
7348 * opts.c (finish_options): Add quotes to error messages.
7349 (parse_sanitizer_options): Likewise.
7351 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
7353 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
7355 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
7357 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
7359 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7361 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
7363 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
7365 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7368 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
7369 provides the hardware capability bits, define the macro
7370 __BUILTIN_CPU_SUPPORTS__.
7371 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
7372 if GLIBC does not provide the hardware capability bits. Add a
7373 gcc_unreachable call if the built-in cpu function is neither
7374 __builtin_cpu_is nor __builtin_cpu_supports.
7375 (rs6000_get_function_versions_dispatcher): Change the warning
7376 that an old GLIBC is used which does not export the capability
7377 bits to be an error.
7378 * doc/extend.texi (target_clones attribute): Document the
7379 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
7380 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
7381 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
7382 the macros defined by GCC if the newer GLIBC is available.
7384 2017-07-12 Jeff Law <law@redhat.com>
7386 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
7387 remaining includes slightly.
7388 * config/riscv/riscv-builtins.c: Include profile-count.h.
7390 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7393 * config/avr/avr.c (avr_set_current_function): In diagnostic
7394 messages: Quote keywords and (parts of) identifiers.
7395 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
7398 2017-07-12 Carl Love <cel@us.ibm.com>
7400 * config/rs6000/rs6000-c.c: Add support for built-in functions
7401 vector bool char vec_revb (vector bool char);
7402 vector bool short vec_revb (vector short char);
7403 vector bool int vec_revb (vector bool int);
7404 vector bool long long vec_revb (vector bool long long);
7405 * doc/extend.texi: Update the built-in documentation file for the
7406 new built-in functions.
7408 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7410 * config/s390/s390.md: Remove movcc splitter.
7412 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7414 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
7415 load/store on condition.
7417 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7420 * config/avr/avr.c (avr_encode_section_info)
7421 [progmem && !TREE_READONLY]: Error if progmem object needs
7424 2017-07-11 Michael Collison <michael.collison@arm.com>
7426 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
7429 2017-07-11 Carl Love <cel@us.ibm.com>
7431 * config/rs6000/rs6000-c.c: Add support for builtins
7432 vector unsigned int vec_parity_lsbb (vector signed int);
7433 vector unsigned int vec_parity_lsbb (vector unsigned int);
7434 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
7435 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
7436 vector unsigned long long vec_parity_lsbb (vector signed long long);
7437 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
7438 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
7439 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
7440 * doc/extend.texi: Update the built-in documentation file for the
7441 new built-in functions.
7443 2017-07-11 David Malcolm <dmalcolm@redhat.com>
7445 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
7446 (layout::m_primary_loc): New field.
7447 (layout::layout): Initialize new field. Move location filtering
7448 logic from here to...
7449 (layout::maybe_add_location_range): ...this new method. Add
7450 support for filtering to just the lines already specified by other
7452 (layout::will_show_line_p): New method.
7453 (gcc_rich_location::add_location_if_nearby): New method.
7454 (selftest::test_add_location_if_nearby): New test function.
7455 (selftest::diagnostic_show_locus_c_tests): Call it.
7456 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
7459 2017-07-11 Tom de Vries <tom@codesourcery.com>
7461 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
7462 (bb_first_real_insn): New function.
7463 (nvptx_single): Add extra initialization of broadcasted condition
7466 2017-07-11 Nathan Sidwell <nathan@acm.org>
7468 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
7470 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
7472 * doc/extend.texi (AVR Function Attributes): Remove weblink to
7473 Binutils doc as TEXI will mess them up.
7474 * doc/invoke.texi (AVR Options): Same here.
7476 2017-07-11 Daniel Cederman <cederman@gaisler.com>
7478 * config/sparc/sparc.opt (mfix-ut700): New option.
7479 (mfix-gr712rc): Likewise.
7480 (sparc_fix_b2bst): New variable.
7481 * doc/invoke.texi (SPARC options): Document them.
7482 (ARM options): Fix warnings.
7483 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
7484 instructions to prevent sequences that can trigger the store-store
7485 errata for certain LEON3FT processors.
7486 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
7487 (sparc_option_override): Set sparc_fix_b2bst appropriately.
7488 * config/sparc/sparc.md (fix_b2bst): New attribute.
7489 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
7491 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
7494 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
7496 (*rsqrtsf2_sse): Ditto.
7498 (div<mode>3): Macroize insn from divdf3 and divsf3
7499 using MODEF mode iterator.
7501 2017-07-10 Martin Sebor <msebor@redhat.com>
7503 PR tree-optimization/80397
7504 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
7505 instead of testing for equality to INTEGER_TYPE.
7507 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
7509 * config.gcc: Remove uclibc from arc target spec.
7511 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
7513 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
7515 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7518 * lto-wrapper.c (remove_option): New function.
7519 (merge_and_complain): Merge PIC/PIE options more realistically.
7521 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7523 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
7527 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
7528 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
7529 * config.in: Regenerate.
7530 * configure: Regenerate.
7531 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
7532 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
7533 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
7534 (TARGET_GASISR_PROLOGUES): ...target mask.
7535 * common/config/avr/avr-common.c
7536 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
7537 Set -mgas-isr-prologues.
7538 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
7539 INSERT_PASS_BEFORE for it.
7540 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
7541 * config/avr/avr.c (avr_option_override)
7542 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
7543 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
7544 (avr_attribute_table) <no_gccisr>: Add new function attribute.
7545 (avr_set_current_function) <is_no_gccisr>: Init machine field.
7546 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
7548 (make_avr_pass_pre_proep): New function.
7549 (emit_push_sfr) <treg>: Add argument to function and use it
7551 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
7552 and set machine->gasisr.yes.
7553 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
7554 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
7555 __gcc_isr.n_pushed to .L__stack_usage.
7556 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
7557 (avr_asm_final_postscan_insn): ...this new static function.
7558 * config/avr/avr.h (machine_function)
7559 <is_no_gccisr, use_L__stack_usage>: New fields.
7560 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
7561 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
7562 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
7563 (gasisr, *gasisr): New expander and insn.
7564 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
7565 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
7566 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
7568 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
7570 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
7573 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7575 Move jump-tables out of .text again.
7578 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
7579 (ASM_OUTPUT_ADDR_VEC): New function.
7580 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
7581 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
7582 INSN_ADDRESSes as asm comment.
7583 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
7584 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
7585 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
7586 * config/avr/avr.md (*tablejump): Adjust comment.
7587 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
7588 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
7590 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
7591 (avr_output_addr_vec): New proto.
7592 (avr_log_t) <insn_addresses>: New field.
7594 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
7597 * config/i386/i386.c (ix86_function_arg_advance): Set
7598 outgoing_args_on_stack to true if there are outgoing arguments
7600 (ix86_function_arg): Likewise.
7601 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
7602 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
7603 * config/i386/i386.h (machine_function): Add
7604 outgoing_args_on_stack.
7606 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
7608 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7610 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
7612 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
7614 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
7615 (REAL_H): Remove $(MACHMODE_H).
7616 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
7618 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
7619 $(MACHMODE_H) and double-int.h.
7620 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
7622 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
7625 2017-07-07 Andrew Pinski <apinski@cavium.com>
7627 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
7628 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
7630 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7632 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
7633 Add warning if GCC was not configured to link against a GLIBC that
7634 exports the hardware capability bits.
7635 (make_resolver_func): Make resolver function private and not a
7636 COMDAT function. Create the name with clone_function_name instead
7637 of make_unique_name.
7640 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
7641 correct operand in doing the split.
7643 2017-07-07 Carl Love <cel@us.ibm.com>
7645 * config/rs6000/rs6000-c: Add support for built-in function
7646 vector unsigned short vec_pack_to_short_fp32 (vector float,
7648 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
7649 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
7650 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
7651 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
7652 (convert_4f32_8i16): Add define_expand.
7653 * doc/extend.texi: Update the built-in documentation file for the
7654 new built-in function.
7656 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7658 * config/sparc/m8.md: New file.
7659 * config/sparc/sparc.md: Include m8.md.
7661 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7663 * config/sparc/sparc.opt: New option -mvis4b.
7664 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
7665 (sparc_option_override): Handle VIS4B.
7666 (enum sparc_builtins): Define
7667 SPARC_BUILTIN_DICTUNPACK{8,16,32},
7668 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
7669 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
7670 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
7671 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
7672 (check_constant_argument): New function.
7673 (sparc_vis_init_builtins): Define builtins
7674 __builtin_vis_dictunpack{8,16,32},
7675 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
7676 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
7677 __builtin_vis_fpcmpde{8,16,32}shl and
7678 __builtin_vis_fpcmpur{8,16,32}shl.
7679 (sparc_expand_builtin): Check that the constant operands to
7680 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
7681 constant and in range.
7682 * config/sparc/sparc-c.c (sparc_target_macros): Handle
7684 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
7685 (SPARC_IMM5_P): Likewise.
7686 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
7687 (enabled): Handle vis4b.
7688 (UNSPEC_DICTUNPACK): New unspec.
7689 (UNSPEC_FPCMPSHL): Likewise.
7690 (UNSPEC_FPUCMPSHL): Likewise.
7691 (UNSPEC_FPCMPDESHL): Likewise.
7692 (UNSPEC_FPCMPURSHL): Likewise.
7693 (cpu_feature): New CPU feature `vis4b'.
7694 (dictunpack{8,16,32}): New insns.
7695 (FPCSMODE): New mode iterator.
7696 (fpcscond): New code iterator.
7697 (fpcsucond): Likewise.
7698 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
7699 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
7700 (fpcmpde{8,16,32}{si,di}shl): Likewise.
7701 (fpcmpur{8,16,32}{si,di}shl): Likewise.
7702 * config/sparc/constraints.md: Define constraints `q' for unsigned
7703 2-bit integer constants and `t' for unsigned 5-bit integer
7705 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
7707 (imm5_operand_dictunpack16): Likewise.
7708 (imm5_operand_dictunpack32): Likewise.
7709 (imm2_operand): Likewise.
7710 * doc/invoke.texi (SPARC Options): Document -mvis4b.
7711 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
7712 ditunpack* and fpcmp*shl builtins.
7714 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7716 * config.gcc: Handle m8 in --with-{cpu,tune} options.
7717 * config.in: Add HAVE_AS_SPARC6 define.
7718 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
7720 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
7722 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
7723 (CPP_CPU_SPEC): Handle m8.
7724 (ASM_CPU_SPEC): Likewise.
7725 * config/sparc/sparc-opts.h (enum processor_type): Add
7727 * config/sparc/sparc.c (m8_costs): New struct.
7728 (sparc_option_override): Handle TARGET_CPU_m8.
7729 (sparc32_initialize_trampoline): Likewise.
7730 (sparc64_initialize_trampoline): Likewise.
7731 (sparc_issue_rate): Likewise.
7732 (sparc_register_move_cost): Likewise.
7733 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
7734 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
7735 (ASM_CPU64_DEFAULT_SPEC): Likewise.
7736 (CPP_CPU_SPEC): Handle M8.
7737 (ASM_CPU_SPEC): Likewise.
7738 (AS_M8_FLAG): Define.
7739 * config/sparc/sparc.md: Add m8 to the cpu attribute.
7740 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
7741 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
7743 * configure: Regenerate.
7744 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
7747 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7749 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
7751 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
7752 ("*movdi_insn_sp32"): Do not set v3pipe.
7753 ("*movsi_insn"): Likewise.
7754 ("*movdi_insn_sp64"): Likewise.
7755 ("*movsf_insn"): Likewise.
7756 ("*movdf_insn_sp32"): Likewise.
7757 ("*movdf_insn_sp64"): Likewise.
7758 ("*zero_extendsidi2_insn_sp64"): Likewise.
7759 ("*sign_extendsidi2_insn"): Likewise.
7760 ("*mov<VM32:mode>_insn"): Likewise.
7761 ("*mov<VM64:mode>_insn_sp64"): Likewise.
7762 ("*mov<VM64:mode>_insn_sp32"): Likewise.
7763 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
7764 ("<vlop:code><VL:mode>3"): Likewise.
7765 ("*not_<vlop:code><VL:mode>3"): Likewise.
7766 ("*nand<VL:mode>_vis"): Likewise.
7767 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
7768 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
7769 ("one_cmpl<VL:mode>2"): Likewise.
7770 ("faligndata<VM64:mode>_vis"): Likewise.
7771 ("alignaddrsi_vis"): Likewise.
7772 ("alignaddrdi_vis"): Likweise.
7773 ("alignaddrlsi_vis"): Likewise.
7774 ("alignaddrldi_vis"): Likewise.
7775 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
7776 ("bmaskdi_vis"): Likewise.
7777 ("bmasksi_vis"): Likewise.
7778 ("bshuffle<VM64:mode>_vis"): Likewise.
7779 ("cmask8<P:mode>_vis"): Likewise.
7780 ("cmask16<P:mode>_vis"): Likewise.
7781 ("cmask32<P:mode>_vis"): Likewise.
7782 ("pdistn<P:mode>_vis"): Likewise.
7783 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
7785 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7787 * config/sparc/sparc.md ("subtype"): New insn attribute.
7788 ("*wrgsr_sp64"): Set insn subtype.
7789 ("*rdgsr_sp64"): Likewise.
7790 ("alignaddrsi_vis"): Likewise.
7791 ("alignaddrdi_vis"): Likewise.
7792 ("alignaddrlsi_vis"): Likewise.
7793 ("alignaddrldi_vis"): Likewise.
7794 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
7795 ("fexpand_vis"): Likewise.
7796 ("fpmerge_vis"): Likewise.
7797 ("faligndata<VM64:mode>_vis"): Likewise.
7798 ("bshuffle<VM64:mode>_vis"): Likewise.
7799 ("cmask8<P:mode>_vis"): Likewise.
7800 ("cmask16<P:mode>_vis"): Likewise.
7801 ("cmask32<P:mode>_vis"): Likewise.
7802 ("fchksm16_vis"): Likewise.
7803 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
7804 ("fmean16_vis"): Likewise.
7805 ("fp<plusminus_insn>64_vis"): Likewise.
7806 ("<plusminus_insn>v8qi3"): Likewise.
7807 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
7808 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
7809 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
7810 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
7811 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
7812 ("*movqi_insn"): Likewise.
7813 ("*movhi_insn"): Likewise.
7814 ("*movsi_insn"): Likewise.
7815 ("movsi_pic_gotdata_op"): Likewise.
7816 ("*movdi_insn_sp32"): Likewise.
7817 ("*movdi_insn_sp64"): Likewise.
7818 ("movdi_pic_gotdata_op"): Likewise.
7819 ("*movsf_insn"): Likewise.
7820 ("*movdf_insn_sp32"): Likewise.
7821 ("*movdf_insn_sp64"): Likewise.
7822 ("*zero_extendhisi2_insn"): Likewise.
7823 ("*zero_extendqihi2_insn"): Likewise.
7824 ("*zero_extendqisi2_insn"): Likewise.
7825 ("*zero_extendqidi2_insn"): Likewise.
7826 ("*zero_extendhidi2_insn"): Likewise.
7827 ("*zero_extendsidi2_insn_sp64"): Likewise.
7828 ("ldfsr"): Likewise.
7829 ("prefetch_64"): Likewise.
7830 ("prefetch_32"): Likewise.
7831 ("tie_ld32"): Likewise.
7832 ("tie_ld64"): Likewise.
7833 ("*tldo_ldub_sp32"): Likewise.
7834 ("*tldo_ldub1_sp32"): Likewise.
7835 ("*tldo_ldub2_sp32"): Likewise.
7836 ("*tldo_ldub_sp64"): Likewise.
7837 ("*tldo_ldub1_sp64"): Likewise.
7838 ("*tldo_ldub2_sp64"): Likewise.
7839 ("*tldo_ldub3_sp64"): Likewise.
7840 ("*tldo_lduh_sp32"): Likewise.
7841 ("*tldo_lduh1_sp32"): Likewise.
7842 ("*tldo_lduh_sp64"): Likewise.
7843 ("*tldo_lduh1_sp64"): Likewise.
7844 ("*tldo_lduh2_sp64"): Likewise.
7845 ("*tldo_lduw_sp32"): Likewise.
7846 ("*tldo_lduw_sp64"): Likewise.
7847 ("*tldo_lduw1_sp64"): Likewise.
7848 ("*tldo_ldx_sp64"): Likewise.
7849 ("*mov<VM32:mode>_insn"): Likewise.
7850 ("*mov<VM64:mode>_insn_sp64"): Likewise.
7851 ("*mov<VM64:mode>_insn_sp32"): Likewise.
7853 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7855 * config/sparc/sparc.md ("type"): New insn type viscmp.
7856 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
7858 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
7859 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
7860 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
7861 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
7863 ("n7_vis_logical_11cycle"): Likewise.
7864 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
7865 * config/sparc/niagara2.md ("niag3_vis": Likewise.
7866 * config/sparc/niagara.md ("niag_vis"): Likewise.
7867 * config/sparc/ultra3.md ("us3_fga"): Likewise.
7868 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
7870 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7872 * config/sparc/sparc.md: New instruction type `bmask'.
7873 (bmaskdi_vis): Use the `bmask' type.
7874 (bmasksi_vis): Likewise.
7875 * config/sparc/ultra3.md (us3_array): Likewise.
7876 * config/sparc/niagara7.md (n7_array): Likewise.
7877 * config/sparc/niagara4.md (n4_array): Likewise.
7878 * config/sparc/niagara2.md (niag2_vis): Likewise.
7879 (niag3_vis): Likewise.
7880 * config/sparc/niagara.md (niag_vis): Likewise.
7882 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7884 * ipa-comdats.c: Remove optimize check from gate.
7885 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
7886 for functions not optimized.
7887 (ipa_fn_summary_read): Skip optimize check.
7888 (ipa_fn_summary_write): Likewise.
7889 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
7891 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
7893 (can_inline_edge_p): Check flag_pcc_struct_return for match.
7894 (check_callers): Give up on caller which is not optimized.
7895 (inline_small_functions): Likewise.
7896 (ipa_inline): Do not give up when not optimizing.
7897 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
7898 away unoptimizes cdtors.
7899 (whole_program_function_and_variable_visibility): Do
7900 ipa_discover_readonly_nonaddressable_vars in LTO mode.
7901 * ipa.c (process_references): Do not check optimize.
7902 (symbol_table::remove_unreachable_nodes): Update optimize check.
7903 (set_writeonly_bit): Update optimize check.
7904 (pass_ipa_cdtor_merge::gate): Do not check optimize.
7905 (pass_ipa_single_use::gate): Remove.
7907 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7909 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
7910 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
7911 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
7912 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
7913 permute_load, permute_store, adjust_extract, adjust_splat,
7914 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
7915 replace_swap_with_copy, dump_swap_insn_table,
7916 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
7917 recombine_lvx_pattern, recombine_stvx_pattern,
7918 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
7919 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
7920 to file rs6000-p8swap.c.
7921 * config/rs6000/rs6000-p8swap.c: New file.
7922 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
7923 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
7924 and rs6000*-*-* targets.
7926 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7928 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
7930 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7932 * lto-wrapper.c (merge_and_complain): Do not merge
7933 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
7934 fsigned_zeros, ftrapping_math, fwrapv.
7935 (append_compiler_options): Do not track these options.
7936 (append_linker_options): Likewie
7938 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7940 * cgraphunit.c (cgraph_node::finalize_function): When
7941 !flag_toplevel_reorde set no_reorder flag.
7942 (varpool_node::finalize_decl): Likewise.
7943 (symbol_table::compile): Drop no toplevel reorder path.
7945 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7947 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
7948 edges; zero probability is not better than uninitialized.
7950 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
7952 * asan.h (asan_sanitize_allocas_p): Declare.
7953 * asan.c (asan_sanitize_allocas_p): New function.
7954 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
7955 (handle_builtin_alloca): Likewise.
7956 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
7957 if !asan_sanitize_allocas_p.
7958 * params.def (asan-instrument-allocas): Add new option.
7959 * params.h (ASAN_PROTECT_ALLOCAS): Define.
7960 * opts.c (common_handle_option): Disable allocas sanitization for
7963 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
7965 * asan.c: Include gimple-fold.h.
7966 (get_last_alloca_addr): New function.
7967 (handle_builtin_stackrestore): Likewise.
7968 (handle_builtin_alloca): Likewise.
7969 (asan_emit_allocas_unpoison): Likewise.
7970 (get_mem_refs_of_builtin_call): Add new parameter, remove const
7971 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
7972 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
7973 (instrument_builtin_call): Pass gimple iterator to
7974 get_mem_refs_of_builtin_call.
7975 (last_alloca_addr): New global.
7976 * asan.h (asan_emit_allocas_unpoison): Declare.
7977 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
7978 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
7979 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
7980 if function calls alloca.
7981 * gimple-fold.c (replace_call_with_value): Remove static keyword.
7982 * gimple-fold.h (replace_call_with_value): Declare.
7983 * internal-fn.c: Include asan.h.
7984 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
7985 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
7987 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7989 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
7990 (C_SELFTEST_FLAGS): New.
7991 (CPP_SELFTEST_FLAGS): New.
7992 (SELFTEST_DEPS): New, from deps of s-selftest.
7993 (C_SELFTEST_DEPS): New, from deps of s-selftest.
7994 (CPP_SELFTEST_DEPS): New.
7995 (selftest): Add dependency on s-selftest-c++.
7996 (s-selftest): Rename to...
7997 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
7998 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
7999 than SELFTEST_FLAGS.
8000 (selftest-gdb): Rename to...
8001 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8003 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8004 (selftest-valgrind): Rename to...
8005 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8007 (selftest-valgrind): Reintroduce as an alias for
8008 selftest-c-valgrind.
8009 (s-selftest-c++): New.
8010 (selftest-c++-gdb): New.
8011 (selftest-c++-valgrind): New.
8013 2017-07-06 Olivier Hainque <hainque@adacore.com>
8015 * gcc.c (process_command): When deciding if undefined variables
8016 should be ignored when processing specs, accept "gcc -v" as well.
8018 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8020 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8021 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8023 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8025 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8026 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8027 * config/arm/arm-tables.opt: Regenerate.
8028 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8030 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8032 2017-07-06 Carl Love <cel@us.ibm.com>
8034 * ChangeLog: Clean up from mid air collision
8036 2017-07-06 Carl Love <cel@us.ibm.com>
8038 * config/rs6000/rs6000-c.c: Add support for built-in functions
8039 vector signed int vec_subc (vector signed int, vector signed int);
8040 vector signed __int128 vec_subc (vector signed __int128,
8041 vector signed __int128);
8042 vector unsigned __int128 vec_subc (vector unsigned __int128,
8043 vector unsigned __int128);
8044 vector signed int vec_sube (vector signed int, vector signed int,
8046 vector unsigned int vec_sube (vector unsigned int,
8047 vector unsigned int,
8048 vector unsigned int);
8049 vector signed __int128 vec_sube (vector signed __int128,
8050 vector signed __int128,
8051 vector signed__int128);
8052 vector unsigned __int128 vec_sube (vector unsigned __int128,
8053 vector unsigned __int128,
8054 vector unsigned __int128);
8055 vector signed int vec_subec (vector signed int, vector signed int,
8057 vector unsigned int vec_subec (vector unsigned int,
8058 vector unsigned int,
8059 vector unsigned int);
8060 vector signed __int128 vec_subec (vector signed __int128,
8061 vector signed __int128,
8062 vector signed__int128);
8063 vector unsigned __int128 vec_subec (vector unsigned __int128,
8064 vector unsigned __int128,
8065 vector unsigned __int128);
8066 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8067 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8068 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8069 BU_ALTIVEC_OVERLOAD_X definitions.
8070 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8071 * doc/extend.texi: Update the built-in documentation file for the new
8074 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8077 * diagnostic-show-locus.c (layout::layout): Use start and finish
8078 spelling location for the start and finish of each range.
8079 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8080 Add unused aspect param.
8081 * input.c (expand_location_1): Add "aspect" param, and use it
8082 to access the correct part of the location.
8083 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8085 (expand_location_to_spelling_point): Likewise.
8086 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8087 param, and pass it to expand_location_1.
8089 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8091 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8092 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8093 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8094 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8095 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8096 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8097 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8098 _mm_maskz_getmant_ss): New intrinsics.
8099 (__builtin_ia32_getexpss128_mask): Changed to ...
8100 __builtin_ia32_getexpss128_round ... this.
8101 (__builtin_ia32_getexpsd128_mask): Changed to ...
8102 __builtin_ia32_getexpsd128_round ... this.
8103 * config/i386/i386-builtin-types.def
8104 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
8105 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
8106 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
8107 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
8108 __builtin_ia32_getmantss_mask_round): New builtins.
8109 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
8110 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
8111 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
8112 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
8113 * config/i386/sse.md
8114 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
8115 avx512f_sgetexp<mode><mask_scalar_name>
8116 <round_saeonly_scalar_name> ... this.
8117 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
8118 %0, %1, %2<round_saeonly_op3>}): Changed to ...
8119 vgetexp<ssescalarmodesuffix>
8120 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8121 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
8122 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
8123 avx512f_vgetmant<mode><mask_scalar_name>
8124 <round_saeonly_scalar_name> ... this.
8125 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
8126 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
8127 vgetmant<ssescalarmodesuffix>
8128 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
8129 %0<mask_scalar_operand4>, %1, %2
8130 <round_saeonly_scalar_mask_op4>, %3} ... this.
8131 * config/i386/subst.md (mask_scalar_operand4,
8132 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
8133 round_saeonly_scalar_nimm_predicate): New subst attributes.
8135 2017-07-06 Julia Koval <julia.koval@intel.com>
8137 * config/i386/i386.c (ix86_erase_embedded_rounding):
8138 Remove code for old rounding pattern.
8140 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
8142 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
8144 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
8146 * doc/sourcebuild.texi (Test Directives, Variants of
8147 dg-require-support): Add documentation for dg-require-stack-check.
8149 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
8151 * config/i386/subst.md (mask_scalar, round_scalar,
8152 round_saeonly_scalar): New meta-templates.
8153 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
8154 round_scalar_mask_operand3, round_scalar_mask_op3,
8155 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
8156 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
8157 round_saeonly_scalar_constraint,
8158 round_saeonly_scalar_prefix): New subst attribute.
8159 * config/i386/sse.md
8160 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
8161 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
8162 <round_scalar_name> ... this.
8163 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
8164 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
8165 <round_scalar_name> ... this.
8166 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
8167 <sse>_vm<code><mode>3<mask_scalar_name>
8168 <round_saeonly_scalar_name> ... this.
8169 (v<plusminus_mnemonic><ssescalarmodesuffix>
8170 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8171 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8172 v<plusminus_mnemonic><ssescalarmodesuffix>
8173 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8174 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8175 (v<multdiv_mnemonic><ssescalarmodesuffix>
8176 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8177 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8178 v<multdiv_mnemonic><ssescalarmodesuffix>
8179 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8180 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8181 (v<maxmin_float><ssescalarmodesuffix>
8182 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
8183 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
8184 v<maxmin_float><ssescalarmodesuffix>
8185 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8186 %0<mask_scalar_operand3>, %1, %<iptr>2
8187 <round_saeonly_scalar_mask_op3>} ... this.
8189 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
8191 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
8192 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
8194 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
8195 Alan Hayward <alan.hayward@arm.com>
8196 David Sherwood <david.sherwood@arm.com>
8198 * combine.c (simplify_if_then_else): Remove "enum" before
8200 * compare-elim.c (can_eliminate_compare): Likewise.
8201 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8203 (aarch64_lookup_simd_builtin_type): Likewise.
8204 (aarch64_simd_builtin_type): Likewise.
8205 (aarch64_init_simd_builtin_types): Likewise.
8206 (aarch64_simd_expand_args): Likewise.
8207 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
8209 (aarch64_reverse_mask): Likewise.
8210 (aarch64_simd_emit_reg_reg_move): Likewise.
8211 (aarch64_gen_adjusted_ldpstp): Likewise.
8212 (aarch64_ccmp_mode_to_code): Likewise.
8213 (aarch64_operands_ok_for_ldpstp): Likewise.
8214 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8215 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8217 (aarch64_min_divisions_for_recip_mul): Likewise.
8218 (aarch64_reassociation_width): Likewise.
8219 (aarch64_get_condition_code_1): Likewise.
8220 (aarch64_simd_emit_reg_reg_move): Likewise.
8221 (aarch64_simd_attr_length_rglist): Likewise.
8222 (aarch64_reverse_mask): Likewise.
8223 (aarch64_operands_ok_for_ldpstp): Likewise.
8224 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8225 (aarch64_gen_adjusted_ldpstp): Likewise.
8226 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
8228 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
8229 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8230 (arm_lookup_simd_builtin_type): Likewise.
8231 (arm_simd_builtin_type): Likewise.
8232 (arm_init_simd_builtin_types): Likewise.
8233 (arm_expand_builtin_args): Likewise.
8234 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
8235 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
8236 (ft32_setup_incoming_varargs): Likewise.
8237 (ft32_function_arg): Likewise.
8238 (ft32_function_arg_advance): Likewise.
8239 (ft32_pass_by_reference): Likewise.
8240 (ft32_arg_partial_bytes): Likewise.
8241 (ft32_valid_pointer_mode): Likewise.
8242 (ft32_addr_space_pointer_mode): Likewise.
8243 (ft32_addr_space_legitimate_address_p): Likewise.
8244 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
8246 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
8247 (ix86_emit_outlined_ms2sysv_restore): Likewise.
8248 (iamcu_alignment): Likewise.
8249 (canonicalize_vector_int_perm): Likewise.
8250 (ix86_noce_conversion_profitable_p): Likewise.
8251 (ix86_mpx_bound_mode): Likewise.
8252 (ix86_operands_ok_for_move_multiple): Likewise.
8253 * config/microblaze/microblaze-protos.h
8254 (microblaze_expand_conditional_branch_reg): Likewise.
8255 * config/microblaze/microblaze.c
8256 (microblaze_expand_conditional_branch_reg): Likewise.
8257 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8259 (rs6000_reassociation_width): Likewise.
8260 (rs6000_invalid_binary_op): Likewise.
8261 (fusion_p9_p): Likewise.
8262 (emit_fusion_p9_load): Likewise.
8263 (emit_fusion_p9_store): Likewise.
8264 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
8266 (riscv_hard_regno_mode_ok_p): Likewise.
8267 (riscv_address_insns): Likewise.
8268 (riscv_split_symbol): Likewise.
8269 (riscv_legitimize_move): Likewise.
8270 (riscv_function_value): Likewise.
8271 (riscv_hard_regno_nregs): Likewise.
8272 (riscv_expand_builtin): Likewise.
8273 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
8274 (riscv_build_integer): Likewise.
8275 (riscv_split_integer): Likewise.
8276 (riscv_legitimate_constant_p): Likewise.
8277 (riscv_cannot_force_const_mem): Likewise.
8278 (riscv_regno_mode_ok_for_base_p): Likewise.
8279 (riscv_valid_base_register_p): Likewise.
8280 (riscv_valid_offset_p): Likewise.
8281 (riscv_valid_lo_sum_p): Likewise.
8282 (riscv_classify_address): Likewise.
8283 (riscv_legitimate_address_p): Likewise.
8284 (riscv_address_insns): Likewise.
8285 (riscv_load_store_insns): Likewise.
8286 (riscv_force_binary): Likewise.
8287 (riscv_split_symbol): Likewise.
8288 (riscv_force_address): Likewise.
8289 (riscv_legitimize_address): Likewise.
8290 (riscv_move_integer): Likewise.
8291 (riscv_legitimize_const_move): Likewise.
8292 (riscv_legitimize_move): Likewise.
8293 (riscv_address_cost): Likewise.
8294 (riscv_subword): Likewise.
8295 (riscv_output_move): Likewise.
8296 (riscv_canonicalize_int_order_test): Likewise.
8297 (riscv_emit_int_order_test): Likewise.
8298 (riscv_function_arg_boundary): Likewise.
8299 (riscv_pass_mode_in_fpr_p): Likewise.
8300 (riscv_pass_fpr_single): Likewise.
8301 (riscv_pass_fpr_pair): Likewise.
8302 (riscv_get_arg_info): Likewise.
8303 (riscv_function_arg): Likewise.
8304 (riscv_function_arg_advance): Likewise.
8305 (riscv_arg_partial_bytes): Likewise.
8306 (riscv_function_value): Likewise.
8307 (riscv_pass_by_reference): Likewise.
8308 (riscv_setup_incoming_varargs): Likewise.
8309 (riscv_print_operand): Likewise.
8310 (riscv_elf_select_rtx_section): Likewise.
8311 (riscv_save_restore_reg): Likewise.
8312 (riscv_for_each_saved_reg): Likewise.
8313 (riscv_register_move_cost): Likewise.
8314 (riscv_hard_regno_mode_ok_p): Likewise.
8315 (riscv_hard_regno_nregs): Likewise.
8316 (riscv_class_max_nregs): Likewise.
8317 (riscv_memory_move_cost): Likewise.
8318 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
8319 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
8320 (rl78_addr_space_address_mode): Likewise.
8321 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8323 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8324 (rs6000_reassociation_width): Likewise.
8325 (rs6000_invalid_binary_op): Likewise.
8326 (fusion_p9_p): Likewise.
8327 (emit_fusion_p9_load): Likewise.
8328 (emit_fusion_p9_store): Likewise.
8329 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
8330 (ok_for_simple_move_operands): Likewise.
8331 (ok_for_simple_move_strict_operands): Likewise.
8332 (ok_for_simple_arith_logic_operands): Likewise.
8333 (visium_legitimize_reload_address): Likewise.
8334 (visium_select_cc_mode): Likewise.
8335 (output_cbranch): Likewise.
8336 (visium_split_double_move): Likewise.
8337 (visium_expand_copysign): Likewise.
8338 (visium_expand_int_cstore): Likewise.
8339 (visium_expand_fp_cstore): Likewise.
8340 * config/visium/visium.c (visium_pass_by_reference): Likewise.
8341 (visium_function_arg): Likewise.
8342 (visium_function_arg_advance): Likewise.
8343 (visium_libcall_value): Likewise.
8344 (visium_setup_incoming_varargs): Likewise.
8345 (visium_legitimate_constant_p): Likewise.
8346 (visium_legitimate_address_p): Likewise.
8347 (visium_legitimize_address): Likewise.
8348 (visium_secondary_reload): Likewise.
8349 (visium_register_move_cost): Likewise.
8350 (visium_memory_move_cost): Likewise.
8351 (prepare_move_operands): Likewise.
8352 (ok_for_simple_move_operands): Likewise.
8353 (ok_for_simple_move_strict_operands): Likewise.
8354 (ok_for_simple_arith_logic_operands): Likewise.
8355 (visium_function_value_1): Likewise.
8356 (rtx_ok_for_offset_p): Likewise.
8357 (visium_legitimize_reload_address): Likewise.
8358 (visium_split_double_move): Likewise.
8359 (visium_expand_copysign): Likewise.
8360 (visium_expand_int_cstore): Likewise.
8361 (visium_expand_fp_cstore): Likewise.
8362 (visium_split_cstore): Likewise.
8363 (visium_select_cc_mode): Likewise.
8364 (visium_split_cbranch): Likewise.
8365 (output_cbranch): Likewise.
8366 (visium_print_operand_address): Likewise.
8367 * expmed.c (flip_storage_order): Likewise.
8368 * expmed.h (emit_cstore): Likewise.
8369 (flip_storage_order): Likewise.
8370 * genrecog.c (validate_pattern): Likewise.
8371 * hsa-gen.c (gen_hsa_addr): Likewise.
8372 * internal-fn.c (expand_arith_overflow): Likewise.
8373 * ira-color.c (allocno_copy_cost_saving): Likewise.
8374 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8375 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
8376 (process_invariant_for_inheritance): Likewise.
8377 * lra-eliminations.c (move_plus_up): Likewise.
8378 * omp-low.c (lower_oacc_reductions): Likewise.
8379 * simplify-rtx.c (simplify_subreg): Likewise.
8380 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
8381 (TARGET_CHKP_BOUND_MODE): Likewise..
8382 * targhooks.c (default_chkp_bound_mode): Likewise.
8383 (default_setup_incoming_vararg_bounds): Likewise.
8384 * targhooks.h (default_chkp_bound_mode): Likewise.
8385 (default_setup_incoming_vararg_bounds): Likewise.
8386 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
8387 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
8388 (have_whole_vector_shift): Likewise.
8389 * tree-vect-stmts.c (vectorizable_load): Likewise.
8390 * doc/tm.texi: Regenerate.
8392 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8394 Graceful degrade if Binutils PR21472 is not available.
8397 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
8398 .rodata in flash test fails.
8399 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
8400 * confgure: Regenerate.
8401 * config.in: Regenerate.
8402 * config/avr/avr.c (avr_asm_named_section)
8403 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
8404 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
8405 (avr_asm_init_sections): Same.
8407 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8409 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
8410 (fma<VH:mode>4_intrinsic): Likewise.
8411 (*fmsub<VCVTF:mode>4): Likewise.
8412 (*fmsub<VH:mode>4_intrinsic): Likewise.
8414 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8417 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
8418 Don't depend on "optimize > 0".
8419 (out_movhi_r_mr, out_movqi_mr_r): Same.
8420 (out_movhi_mr_r, out_movqi_r_mr): Same.
8421 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
8422 io_address_operand on "optimize > 0".
8424 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8426 * tree-loop-distribution.c: Add general explanantion on the pass.
8427 (generate_loops_for_partition): Mark distributed loop.
8428 (pg_add_dependence_edges): New parameter. Handle alias data
8429 dependence specially and record it in the parameter if asked.
8430 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
8431 (init_partition_graph_vertices, add_partition_graph_edge): New.
8432 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
8433 (free_partition_graph_vdata, build_partition_graph): New.
8434 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
8435 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
8436 (data_ref_segment_size, latch_dominated_by_data_ref): New.
8437 (compute_alias_check_pairs, version_loop_by_alias_check): New.
8438 (version_for_distribution_p, finalize_partitions): New.
8439 (distribute_loop): Handle alias data dependence specially. Factor
8440 out loop fusion code as functions and call these functions.
8442 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8444 * tree-loop-distribution.c (classify_partition): New parameter and
8445 better handle reduction statement.
8446 (rdg_build_partitions): Revise comment.
8447 (distribute_loop): Compute statements in all partitions and pass it
8448 to classify_partition.
8450 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8452 * tree-loop-distribution.c (enum partition_type): New.
8453 (struct partition): New field type.
8454 (partition_merge_into): Add parameter. Update partition type.
8455 (data_dep_in_cycle_p, update_type_for_merge): New functions.
8456 (build_rdg_partition_for_vertex): Compute partition type.
8457 (rdg_build_partitions): Dump partition type.
8458 (distribute_loop): Update calls to partition_merge_into.
8460 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8462 * tree-loop-distribution.c (struct ddr_hasher): New.
8463 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
8465 (classify_partition): Call get_data_dependence.
8466 (pg_add_dependence_edges): Ditto.
8467 (distribute_loop): Release data dependence hash table.
8469 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8471 * tree-loop-distribution.c (ref_base_address): Delete.
8472 (similar_memory_accesses): Rename ...
8473 (share_memory_accesses): ... to this. Check if partitions access
8474 the same memory reference.
8475 (distribute_loop): Call share_memory_accesses.
8477 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8479 * tree-loop-distribution.c (struct partition): New field recording
8481 (partition_alloc, partition_free): Init and release data refs.
8482 (partition_merge_into): Merge data refs.
8483 (build_rdg_partition_for_vertex): Collect data refs for partition.
8484 (pg_add_dependence_edges): Change parameters from vector to bitmap.
8486 (distribute_loop): Remve data refs from vertice data of partition
8489 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8491 * tree-loop-distribution.c (params.h): Include header file.
8492 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
8493 (datarefs_vec): New global var.
8494 (create_rdg_vertices): Use datarefs_vec directly.
8495 (free_rdg): Don't free data references.
8496 (build_rdg): Update use. Don't free data references.
8497 (distribute_loop): Compute global variable for data references.
8498 Bail out if there are too many data references.
8500 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8502 * tree-loop-distribution.c (loop_nest): New global var.
8503 (build_rdg): Use loop directly, rather than loop nest.
8504 (pg_add_dependence_edges): Remove loop nest parameter. Use global
8506 (distribute_loop): Compute global variable loop nest. Update use.
8508 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8510 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
8511 (partition_merge_into): New parameter. Dump reason for fusion.
8512 (distribute_loop): Update use of partition_merge_into.
8514 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8516 * tree-loop-distribution.c (bb_top_order_index): New.
8517 (bb_top_order_index_size, bb_top_order_cmp): New.
8518 (stmts_from_loop): Use topological order.
8519 (pass_loop_distribution::execute): Compute and release topological
8520 order for basic blocks.
8522 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8524 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
8527 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8529 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
8530 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
8531 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
8532 * internal-fn.def (LOOP_DIST_ALIAS): New.
8533 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
8534 (fold_loop_internal_call): ... this.
8535 (vect_loop_dist_alias_call): New function.
8536 (set_uid_loop_bbs): Call fold_loop_internal_call.
8537 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
8540 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
8543 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
8544 Require dead FLAGS_REG at the beginning of a peephole.
8546 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
8549 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
8550 arguments in the call to __builtin_ia32_sbb_u32.
8551 (_subborrow_u64): Swap _X and _Y arguments in the call to
8552 __builtin_ia32_sbb_u64.
8554 2017-07-04 Jakub Jelinek <jakub@redhat.com>
8557 * tree-vrp.c (compare_assert_loc): Turn into a function template
8558 with stable template parameter. Only test if a->e is NULL,
8559 !a->e == !b->e has been verified already. Use e == NULL or
8560 e != NULL instead of e or ! e tests. If stable is true, don't use
8561 iterative_hash_expr, on the other side allow a or b or both NULL
8562 and sort the NULLs last.
8563 (process_assert_insertions): Sort using compare_assert_loc<false>
8564 instead of compare_assert_loc, later sort using
8565 compare_assert_loc<true> before calling process_assert_insertions_for
8566 in a loop. Use break instead of continue once seen NULL pointer.
8568 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
8570 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8571 Cortex-R7 and Cortex-R8 processors.
8573 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8575 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
8576 uninitialized while src is not.
8578 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
8580 * common/config/arm/arm-common.c: Adjust include path for
8582 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
8583 (arm-cpu.h): Create in build directory. Adjust dependency rules.
8584 (arm-cpu-data.h): Likewise.
8585 (arm-cpu-cdata.h): Likewise.
8586 * config/arm/arm-cpu.h: Delete.
8587 * config/arm/arm-cpu-cdata.h: Delete.
8588 * config/arm/arm-cpu-data.h: Delete.
8590 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
8592 * config/arm/arm-cpus.in (cortex-a55): New.
8593 (cortex-a75): Likewise.
8594 (cortex-a75.cortex-a55): Likewise.
8595 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
8597 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
8598 * config/arm/arm-cpu-cdata.h: Regenerate.
8599 * config/arm/arm-cpu-data.h: Regenerate.
8600 * config/arm/arm-cpu.h: Regenerate.
8601 * config/arm/arm-tables.opt: Regenerate.
8602 * config/arm/arm-tune.md: Regenerate.
8604 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8606 * haifa-sched.c (sched_create_recovery_edges): Update profile.
8608 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8610 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
8613 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
8615 PR tree-optimization/81292
8616 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
8617 full_string_p, also call adjust_related_strinfos if the adjustment
8618 is simple, otherwise invalidate related strinfos.
8620 2017-07-04 Martin Liska <mliska@suse.cz>
8623 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
8624 newly created variable as DECL_IGNORED_P.
8626 2017-07-04 Martin Liska <mliska@suse.cz>
8629 * ipa-inline.c (inline_small_functions):
8630 Use xstrdup_for_dump.
8632 2017-07-04 Tom de Vries <tom@codesourcery.com>
8634 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
8636 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8639 * config/darwin.c (darwin_function_switched_text_sections):
8642 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
8644 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
8646 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8648 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
8650 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8652 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
8653 min_profitable_iters, and th as inclusive lower bounds.
8654 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
8655 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
8656 for min_profitable_iters and min_profitable_estimate.
8657 (vect_transform_loop): Treat th as an inclusive lower bound.
8658 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
8660 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8663 * config/darwin.c (darwin_function_switched_text_sections):
8664 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
8665 in two pieces, and suppress the use of buf.
8667 2017-07-03 Nathan Sidwell <nathan@acm.org>
8669 * hash-table.h (hash_table_mod1): Fix indentation.
8671 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8674 * predict.c (force_edge_cold): Be more careful about propagation
8676 * profile-count.h (profile_probability::guessed,
8677 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
8679 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
8681 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
8683 * doc/invoke.texi (rcpc architecture extension): Document it.
8685 2017-07-03 Richard Biener <rguenther@suse.de>
8687 PR tree-optimization/60510
8688 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
8689 the scalar reduction PHI and use it.
8690 (vectorizable_reduction): Properly guard the single_defuse_cycle
8691 path for non-SLP reduction chains where we cannot use it.
8692 Rework reduc_def/index and vector type deduction. Rework
8693 vector operand gathering during reduction op code-gen.
8694 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
8695 chains dissolve the chain and leave it to non-SLP reduction
8698 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8700 * tree-data-ref.h (dr_alignment): Declare.
8701 * tree-data-ref.c (dr_alignment): New function.
8702 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
8703 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
8705 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
8707 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8709 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
8710 and base_misalignment fields.
8711 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
8712 * tree-data-ref.c: Include builtins.h.
8713 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
8714 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
8715 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
8716 * tree-vect-data-refs.c: Include tree-cfg.h.
8717 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
8718 fields instead of calculating an alignment here.
8719 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
8720 innermost_loop_behavior fields.
8722 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8724 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
8726 (DR_STEP_ALIGNMENT): New macro.
8727 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
8728 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
8729 (create_data_ref): Print it.
8730 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
8731 to tell whether the step preserves vector (mis)alignment.
8732 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8733 Move the check for an integer step and generalise to all INTEGER_CST.
8734 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
8735 Print the outer step alignment.
8737 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8739 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
8740 with offset_alignment.
8741 (DR_ALIGNED_TO): Delete.
8742 (DR_OFFSET_ALIGNMENT): New macro.
8743 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
8744 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
8745 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
8746 (create_data_ref): Likewise.
8747 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8748 (vect_analyze_data_refs): Likewise.
8749 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
8750 creating dummy innermost behavior.
8752 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8754 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
8755 with a "innermost_loop_behavior *" and refeence tree.
8756 * tree-data-ref.c (dr_analyze_innermost): Likewise.
8757 (create_data_ref): Update call accordingly.
8758 * tree-predcom.c (find_looparound_phi): Likewise.
8760 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8762 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
8763 fields with dr_wrt_vec_loop.
8764 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
8765 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
8766 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
8767 (vect_dr_behavior): New function.
8768 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
8769 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
8770 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
8771 track whether the step preserves the misalignment.
8772 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
8773 Use vect_dr_behavior.
8774 (vect_setup_realignment): Update call accordingly.
8775 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
8776 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
8777 call to vect_create_addr_base_for_vector_ref.
8778 (vect_create_cond_for_align_checks): Likewise.
8779 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
8780 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
8781 (vect_recog_mask_conversion_pattern): Likewise.
8782 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
8783 (new_stmt_vec_info): Remove redundant zeroing.
8785 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8787 * common/config/arm/arm-common.c (arm_be8_option): New function.
8788 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
8789 (ISA_ARMv6): Add isa_bit_be8.
8790 * config/arm/arm.h (arm_be8_option): Add prototype.
8791 (BE8_SPEC_FUNCTION): New define.
8792 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
8793 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
8794 (mlittle-endian): Similarly.
8795 (mbe8, mbe32): New options.
8796 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
8797 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
8799 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8801 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
8803 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8805 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
8806 (cleanup_tree_cfg_bb): Use it.
8807 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
8809 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
8811 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8814 * loop-doloop.c (add_test): Update profile.
8816 2017-07-03 Martin Liska <mliska@suse.cz>
8819 * sanopt.c (rewrite_usage_of_param): New function.
8820 (sanitize_rewrite_addressable_params): Likewise.
8821 (pass_sanopt::execute): Call rewrite_usage_of_param.
8823 2017-07-03 Richard Biener <rguenther@suse.de>
8825 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
8826 back to using VIEW_CONVERT_EXPR.
8828 2017-07-03 Martin Liska <mliska@suse.cz>
8831 * doc/extend.texi: Document when a resolver function is
8832 generated for target_clones.
8834 2017-07-03 Martin Liska <mliska@suse.cz>
8836 * asan.c (asan_emit_stack_protection): Unpoison just red zones
8837 and shadow memory of auto variables which are subject of
8838 use-after-scope sanitization.
8839 (asan_expand_mark_ifn): Add do set only when is_poison.
8841 2016-07-03 Richard Biener <rguenther@suse.de>
8843 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
8845 (vect_force_simple_reduction): Record reduction def -> phi mapping.
8846 (vectorizable_reduction): Perform reduction PHI creation when
8847 visiting a reduction PHI and adjust and simplify code generation
8848 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
8849 (vect_transform_loop): Visit reduction PHIs.
8850 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
8851 defs into the SLP tree.
8852 (vect_build_slp_tree): Reduction defs terminate the recursion.
8853 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
8855 (vect_get_vec_defs_for_stmt_copy): Export.
8856 (vect_get_vec_defs): Likewise.
8857 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
8859 (vect_get_vec_defs_for_stmt_copy): Declare.
8860 (vect_get_vec_defs): Likewise.
8862 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8864 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
8865 parameter with a "loop" parameter and use it instead of the
8866 loop containing DR_STMT. Don't check simple_iv when doing
8867 BB analysis. Describe the two analysis modes in the comment.
8869 2017-07-03 Tom de Vries <tom@codesourcery.com>
8871 PR tree-optimization/69468
8872 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
8873 (find_same_succ_bb): Handle ignore_edge_flags.
8875 2017-07-03 Tom de Vries <tom@codesourcery.com>
8877 PR tree-optimization/81192
8878 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
8880 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
8882 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
8884 2017-07-03 Tom de Vries <tom@codesourcery.com>
8886 PR tree-optimization/81192
8887 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
8888 BB_SAME_SUCC (bb) == NULL.
8890 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8892 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
8895 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8897 * dumpfile.c: Include profile-count.h
8898 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
8900 (insert_cond_bb): Update profile.
8901 * tree-cfg.h (insert_cond_bb): Update prototype.
8902 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
8903 * tree-dump.c: Do not include tree-cfg.
8905 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8907 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
8909 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8911 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
8914 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8916 * tree-complex.c (expand_complex_div_wide): update profile.
8918 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8919 Alan Hayward <alan.hayward@arm.com>
8920 David Sherwood <david.sherwood@arm.com>
8922 * Makefile.in (MACHMODE_H): Remove insn-modes.h
8923 (CORETYPES_H): New define.
8924 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
8925 (insn-modes-inline.h, s-modes-inline-h): New rules.
8926 (generated_files): Add insn-modes-inline.h.
8927 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
8928 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
8929 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
8930 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
8931 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
8932 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
8933 (build/gencodes.o, build/genconditions.o): Likewise.
8934 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
8935 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
8936 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
8937 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
8938 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
8939 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
8940 * coretypes.h: Include everything up to real.h for generators.
8941 Include insn-modes.h first. Include wide-int-print.h after
8942 wide-int.h. Include insn-modes-inline.h and then machmode.h.
8943 * machmode.h: Don't include insn-modes.h here.
8944 * function-tests.c: Remove includes of signop.h, machmode.h,
8945 double-int.h and wide-int.h.
8947 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
8949 * optc-save-gen.awk: Likewise.
8950 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
8951 * godump.c: Remove include of wide-int-print.h.
8952 * pretty-print.h: Likewise.
8953 * wide-int-print.cc: Likewise.
8954 * wide-int.cc: Likewise.
8955 * hash-map-tests.c: Remove include of signop.h.
8956 * hash-set-tests.c: Likewise.
8957 * rtl-tests.c: Likewise.
8958 * mkconfig.sh: Remove include of machmode.h.
8959 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
8961 (emit_insn_modes_inline_h): ...this new function. Emit the code
8962 into an insn-modes-inline.h header file, adding appropriate
8963 include guards and end comments.
8964 (emit_insn_modes_c_header): Remove include of machmode.h.
8965 (emit_min_insn_modes_c_header): Include coretypes.h rather than
8967 (main): Handle -i flag and call emit_insn_modes_inline_h when
8970 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8972 * tree-ssa-strlen.c (strinfo): Rename the length field to
8973 nonzero_chars. Add a full_string_p field.
8974 (compare_nonzero_chars, zero_length_string_p): New functions.
8975 (get_addr_stridx): Add an offset_out parameter.
8976 Use compare_nonzero_chars.
8977 (get_stridx): Update accordingly. Use compare_nonzero_chars.
8978 (new_strinfo): Update after above changes to strinfo.
8979 (set_endptr_and_length): Set full_string_p.
8980 (get_string_length): Update after above changes to strinfo.
8981 (unshare_strinfo): Update call to new_strinfo.
8982 (maybe_invalidate): Likewise.
8983 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
8984 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
8985 as a uhwi instead of an shwi. Update after above changes to
8986 strinfo and new_strinfo.
8987 (zero_length_string): Assert that chainsi contains full strings.
8988 Use zero_length_string_p. Update call to new_strinfo.
8989 (adjust_related_strinfos): Update after above changes to strinfo.
8990 Copy full_string_p from origsi.
8991 (adjust_last_stmt): Use zero_length_string_p.
8992 (handle_builtin_strlen): Update after above changes to strinfo and
8993 new_strinfo. Install the lhs as the string length if the previous
8994 entry didn't describe a full string.
8995 (handle_builtin_strchr): Update after above changes to strinfo
8997 (handle_builtin_strcpy): Likewise.
8998 (handle_builtin_strcat): Likewise.
8999 (handle_builtin_malloc): Likewise.
9000 (handle_pointer_plus): Likewise.
9001 (handle_builtin_memcpy): Likewise. Track nonzero characters
9002 that aren't necessarily followed by a nul terminator.
9003 (handle_char_store): Likewise.
9005 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9007 PR tree-optimization/80769
9008 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9009 for malloc and calloc. Document the new invariant that all related
9010 strinfos have delayed lengths or none do.
9011 (verify_related_strinfos): Move earlier in file.
9012 (set_endptr_and_length): New function, split out from...
9013 (get_string_length): ...here. Also set the lengths of related
9015 (zero_length_string): Assert that chainsi has known (rather than
9017 (adjust_related_strinfos): Likewise.
9019 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9021 PR tree-optimization/81136
9022 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9023 assert that two references with the same misalignment have the same
9024 compile-time misalignment if those compile-time misalignments
9027 2017-07-01 Andi Kleen <ak@linux.intel.com>
9029 * print-tree.c (print_node): Print all attributes.
9031 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9033 * cfg.c (scale_bbs_frequencies): New function.
9034 * cfg.h (scale_bbs_frequencies): Declare it.
9035 * cfgloopanal.c (single_likely_exit): Cleanup.
9036 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9038 (scale_loop_profile): Likewise.
9039 (loop_version): Likewise.
9040 (create_empty_loop_on_edge): Update.
9041 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9042 scale_loop_frequencies, scale_loop_profile, loopify,
9043 loop_version): Update prototypes.
9044 * modulo-sched.c (sms_schedule): Update.
9045 * predict.c (unlikely_executed_edge_p): Also check probability.
9046 (probably_never_executed_edge_p): Fix typo.
9047 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9048 * tree-parloops.c (gen_parallel_loop): Update.
9049 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9050 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9051 * tree-ssa-loop-split.c (split_loop): Update.
9052 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9053 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9054 (vect_loop_versioning): Update.
9055 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9057 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9059 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9061 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9063 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9064 to keep profile consistent.
9066 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9068 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9069 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9070 * profile-count.h (max_safe_multiplier): Make unsigned.
9071 (profile_count::guessed_zero): New.
9073 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9075 * bb-reorder.c (fix_up_crossing_landing_pad,
9076 fix_crossing_conditional_branches): Use make_single_succ_edge
9077 to keep profile consistent.
9079 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9081 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9084 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9087 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9088 the right scopes, make sure cond_jump isn't preserved between multiple
9089 iterations. Search for fallthru edge whenever there are 3+ edges and
9090 use find_fallthru_edge for it.
9092 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9094 Patch by Alexander Monakov <amonakov@ispras.ru>
9095 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9096 probabilities consistently.
9098 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9100 * pa.c (pa_expand_compare_and_swap_loop): Update call of
9101 emit_cmp_and_jump_insns.
9103 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9106 * tree-inline.c (expand_call_inline): Combine profile statuses.
9108 2017-06-30 Andrew Pinski <apinski@cavium.com>
9110 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
9111 fold_stmt returned true.
9113 2017-06-30 Nathan Sidwell <nathan@acm.org>
9115 * ggc.h (empty_string): Delete.
9116 * cfgexpand.c (expand_asm_stmt): Use plain "".
9117 * optabs.c (expand_asm_memory_barrier): Likewise.
9118 * stringpool.c (empty_string): Delete.
9119 (digit_vector, digit_string): Delete.
9120 (ggc_alloc_string): Use plain "", don't optimize single digit
9121 strings. Use ggc_alloc_atomic.
9123 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
9125 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
9126 comparison set and one other set, use the cost of the non-comparison
9129 2017-06-30 Nathan Sidwell <nathan@acm.org>
9131 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
9134 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
9136 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
9137 loops. Remove now unneeded calls to gimple_switch_set_label() that
9138 just set removed labels to NULL_TREE.
9140 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
9142 * tree-ssanames.c (set_range_info_raw): Abstract from ...
9143 (set_range_info): ...here. Only call set_range_info_raw if domain
9145 (set_nonzero_bits): Call set_range_info_raw.
9146 * tree-ssanames.h (set_range_info_raw): New.
9148 2017-06-30 Jakub Jelinek <jakub@redhat.com>
9151 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
9152 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
9153 of nonimmediate_operand and <store_mask_constraint> instead of m
9154 for the input operand. For V8FI iterator, always split if input
9155 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
9156 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
9157 <store_mask_predicate> instead of register_operand and
9158 <store_mask_constraint> instead of v for the input operand. Make
9159 sure both operands aren't MEMs for if not <mask_applied>.
9161 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
9163 * lto-wrapper.c (copy_file) Close both file descriptors before
9166 2017-06-30 Martin Liska <mliska@suse.cz>
9169 * multiple_target.c (create_dispatcher_calls): Make ifunc
9170 also for function that don't have calls or are not referenced.
9172 2017-06-30 Richard Biener <rguenther@suse.de>
9174 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
9175 analyze the first scalar stmt. Move vector type computation
9176 for the BB case here from ...
9177 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
9178 live operation processing in the SLP case properly.
9180 2017-06-30 Richard Biener <rguenther@suse.de>
9182 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
9184 2017-06-30 Martin Liska <mliska@suse.cz>
9187 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
9188 before BUILT_IN_UNWIND_RESUME when ASAN is used.
9190 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
9192 * doc/invoke.texi (AArch64): Add missing options and remove redundant
9195 2017-06-30 Richard Biener <rguenther@suse.de>
9197 PR tree-optimization/81249
9198 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
9199 condition reduction result to original scalar type.
9201 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9203 * profile-count.h (enum profile_quality): Fix typos and whitespace
9206 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9208 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
9209 type for branch probabilities.
9211 2017-06-29 Julian Brown <julian@codesourcery.com>
9212 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9214 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
9215 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
9216 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
9217 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
9219 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9221 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
9222 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
9223 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
9224 CC usage from generic code to here.
9225 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
9226 CC usage into the target macros.
9228 2017-06-29 Maya Rashish <coypu@sdf.org>
9230 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
9233 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9235 * arm/arm-builtins.c: Include profile-count.h
9236 * except.c (sjlj_emit_function_enter): Use
9237 profile_probability::unlikely.
9239 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9241 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
9242 and tocrel_offset be pointer args rather than implicitly using
9244 (legitimate_constant_pool_address_p, rs6000_emit_move,
9245 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
9246 tocrel_offset and use in toc_relative_expr_p call.
9247 (print_operand, print_operand_address): Use static tocrel_base_oac
9248 and tocrel_offset_oac.
9249 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
9252 2017-06-29 Maya Rashish <coypu@sdf.org>
9254 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
9256 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
9258 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
9259 objects, take into account only the alignment of 'op0' and 'mode1' if
9262 2017-06-29 Steve Ellcey <sellcey@cavium.com>
9264 * ccmp.c (ccmp_tree_comparison_p): New function.
9265 (ccmp_candidate_p): Update to use above function.
9266 (get_compare_parts): New function.
9267 (expand_ccmp_next): Update to use new functions.
9268 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
9270 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
9271 take mode as argument.
9272 * ccmp.h (expand_ccmp_expr): Add mode as argument.
9273 * expr.c (expand_expr_real_1): Pass mode as argument.
9275 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
9277 * combine.c (combine_instructions): Print insns to dump_file, together
9280 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9282 * asan.c (asan_emit_stack_protection): Update.
9283 (create_cond_insert_point): Update.
9284 * auto-profile.c (afdo_propagate_circuit): Update.
9285 * basic-block.h (struct edge_def): Turn probability to
9286 profile_probability.
9287 (EDGE_FREQUENCY): Update.
9288 * bb-reorder.c (find_traces_1_round): Update.
9289 (better_edge_p): Update.
9290 (sanitize_hot_paths): Update.
9291 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
9292 (make_single_succ_edge): Update.
9293 (check_bb_profile): Update.
9294 (dump_edge_info): Update.
9295 (update_bb_profile_for_threading): Update.
9296 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
9297 probabilitycount to 0.
9298 * cfgbuild.c (compute_outgoing_frequencies): Update.
9299 * cfgcleanup.c (try_forward_edges): Update.
9300 (outgoing_edges_match): Update.
9301 (try_crossjump_to_edge): Update.
9302 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
9303 (expand_gimple_tailcall): Update.
9304 (construct_init_block): Use make_single_succ_edge.
9305 (construct_exit_block): Use make_single_succ_edge.
9306 * cfghooks.c (verify_flow_info): Update.
9307 (redirect_edge_succ_nodup): Update.
9308 (split_edge): Update.
9309 (account_profile_record): Update.
9310 * cfgloopanal.c (single_likely_exit): Update.
9311 * cfgloopmanip.c (scale_loop_profile): Update.
9312 (set_zero_probability): Remove.
9313 (duplicate_loop_to_header_edge): Update.
9314 * cfgloopmanip.h (loop_version): Update prototype.
9315 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9316 (force_nonfallthru_and_redirect): Update.
9317 (update_br_prob_note): Update.
9318 (rtl_verify_edges): Update.
9319 (purge_dead_edges): Update.
9320 (rtl_lv_add_condition_to_bb): Update.
9321 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
9322 * cgraphunit.c (init_lowered_empty_function): Update.
9323 (cgraph_node::expand_thunk): Update.
9324 * cilk-common.c: Include profile-count.h
9325 * dojump.c (inv): Remove.
9326 (jumpifnot): Update.
9327 (jumpifnot_1): Update.
9328 (do_jump_1): Update.
9330 (do_jump_by_parts_greater_rtx): Update.
9331 (do_compare_rtx_and_jump): Update.
9332 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
9333 do_jump_1. do_compare_rtx_and_jump): Update prototype.
9334 * dwarf2cfi.c: Include profile-count.h
9335 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
9336 (sjlj_emit_dispatch_table): Likewise.
9337 * explow.c: Include profile-count.h
9338 * expmed.c (emit_store_flag_force): Update.
9339 (do_cmp_and_jump): Update.
9340 * expr.c (compare_by_pieces_d::generate): Update.
9341 (compare_by_pieces_d::finish_mode): Update.
9342 (emit_block_move_via_loop): Update.
9343 (store_expr_with_bounds): Update.
9344 (store_constructor): Update.
9345 (expand_expr_real_2): Update.
9346 (expand_expr_real_1): Update.
9347 * expr.h (try_casesi, try_tablejump): Update prototypes.
9348 * gimple-pretty-print.c (dump_probability): Update.
9349 (dump_profile): New.
9350 (dump_gimple_label): Update.
9351 (dump_gimple_bb_header): Update.
9352 * graph.c (draw_cfg_node_succ_edges): Update.
9353 * hsa-gen.c (convert_switch_statements): Update.
9354 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
9355 (find_if_case_1): Update.
9356 (find_if_case_2): Update.
9357 * internal-fn.c (expand_arith_overflow_result_store): Update.
9358 (expand_addsub_overflow): Update.
9359 (expand_neg_overflow): Update.
9360 (expand_mul_overflow): Update.
9361 (expand_vector_ubsan_overflow): Update.
9362 * ipa-cp.c (good_cloning_opportunity_p): Update.
9363 * ipa-split.c (split_function): Use make_single_succ_edge.
9364 * ipa-utils.c (ipa_merge_profiles): Update.
9365 * loop-doloop.c (add_test): Update.
9366 (doloop_modify): Update.
9367 * loop-unroll.c (compare_and_jump_seq): Update.
9368 (unroll_loop_runtime_iterations): Update.
9369 * lra-constraints.c (lra_inheritance): Update.
9370 * lto-streamer-in.c (input_cfg): Update.
9371 * lto-streamer-out.c (output_cfg): Update.
9372 * mcf.c (adjust_cfg_counts): Update.
9373 * modulo-sched.c (sms_schedule): Update.
9374 * omp-expand.c (expand_omp_for_init_counts): Update.
9375 (extract_omp_for_update_vars): Update.
9376 (expand_omp_ordered_sink): Update.
9377 (expand_omp_for_ordered_loops): Update.
9378 (expand_omp_for_generic): Update.
9379 (expand_omp_for_static_nochunk): Update.
9380 (expand_omp_for_static_chunk): Update.
9381 (expand_cilk_for): Update.
9382 (expand_omp_simd): Update.
9383 (expand_omp_taskloop_for_outer): Update.
9384 (expand_omp_taskloop_for_inner): Update.
9385 * omp-simd-clone.c (simd_clone_adjust): Update.
9386 * optabs.c (expand_doubleword_shift): Update.
9387 (expand_abs): Update.
9388 (emit_cmp_and_jump_insn_1): Update.
9389 (expand_compare_and_swap_loop): Update.
9390 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
9391 * predict.c (predictable_edge_p): Update.
9392 (edge_probability_reliable_p): Update.
9393 (set_even_probabilities): Update.
9394 (combine_predictions_for_insn): Update.
9395 (combine_predictions_for_bb): Update.
9396 (propagate_freq): Update.
9397 (estimate_bb_frequencies): Update.
9398 (force_edge_cold): Update.
9399 * profile-count.c (profile_count::dump): Add missing space into dump.
9400 (profile_count::debug): Add newline.
9401 (profile_count::differs_from_p): Explicitly convert to unsigned.
9402 (profile_count::stream_in): Update.
9403 (profile_probability::dump): New member function.
9404 (profile_probability::debug): New member function.
9405 (profile_probability::differs_from_p): New member function.
9406 (profile_probability::differs_lot_from_p): New member function.
9407 (profile_probability::stream_in): New member function.
9408 (profile_probability::stream_out): New member function.
9409 * profile-count.h (profile_count_quality): Rename to ...
9410 (profile_quality): ... this one.
9411 (profile_probability): New.
9412 (profile_count): Update.
9413 * profile.c (compute_branch_probabilities): Update.
9414 * recog.c (peep2_attempt): Update.
9415 * sched-ebb.c (schedule_ebbs): Update.
9416 * sched-rgn.c (find_single_block_region): Update.
9417 (compute_dom_prob_ps): Update.
9418 (schedule_region): Update.
9419 * sel-sched-ir.c (compute_succs_info): Update.
9420 * stmt.c (struct case_node): Update.
9421 (do_jump_if_equal): Update.
9422 (get_outgoing_edge_probs): Update.
9423 (conditional_probability): Update.
9424 (emit_case_dispatch_table): Update.
9425 (expand_case): Update.
9426 (expand_sjlj_dispatch_table): Update.
9427 (emit_case_nodes): Update.
9428 * targhooks.c: Update.
9429 * tracer.c (better_p): Update.
9430 (find_best_successor): Update.
9431 * trans-mem.c (expand_transaction): Update.
9432 * tree-call-cdce.c: Update.
9433 * tree-cfg.c (gimple_split_edge): Upate.
9434 (move_sese_region_to_fn): Upate.
9435 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
9436 * tree-eh.c (lower_resx): Upate.
9437 (cleanup_empty_eh_move_lp): Upate.
9438 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9439 * tree-inline.c (copy_edges_for_bb): Update.
9440 (copy_cfg_body): Update.
9441 * tree-parloops.c (gen_parallel_loop): Update.
9442 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
9443 (gimple_gen_time_profiler): Update.
9444 * tree-ssa-dce.c (remove_dead_stmt): Update.
9445 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9446 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9447 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9448 (unloop_loops): Update.
9449 (try_peel_loop): Update.
9450 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9451 * tree-ssa-loop-split.c (connect_loops): Update.
9452 (split_loop): Update.
9453 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9454 (hoist_guard): Update.
9455 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9456 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9457 (value_replacement): Update.
9458 * tree-ssa-reassoc.c (branch_fixup): Update.
9459 * tree-ssa-tail-merge.c (replace_block_by): Update.
9460 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9461 (create_edge_and_update_destination_phis): Update.
9462 (compute_path_counts): Update.
9463 (recompute_probabilities): Update.
9464 (update_joiner_offpath_counts): Update.
9465 (freqs_to_counts_path): Update.
9466 (duplicate_thread_path): Update.
9467 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9468 (struct switch_conv_info): Update.
9469 (gen_inbound_check): Update.
9470 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9471 (vect_do_peeling): Update.
9472 (vect_loop_versioning): Update.
9473 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9474 (optimize_mask_stores): Update.
9475 * ubsan.c (ubsan_expand_null_ifn): Update.
9476 * value-prof.c (gimple_divmod_fixed_value): Update.
9477 (gimple_divmod_fixed_value_transform): Update.
9478 (gimple_mod_pow2): Update.
9479 (gimple_mod_pow2_value_transform): Update.
9480 (gimple_mod_subtract): Update.
9481 (gimple_mod_subtract_transform): Update.
9482 (gimple_ic): Update.
9483 (gimple_stringop_fixed_value): Update.
9484 (gimple_stringops_transform): Update.
9485 * value-prof.h: Update.
9487 2017-06-29 Carl Love <cel@us.ibm.com>
9489 * config/rs6000/rs6000-c.c: Add support for built-in functions
9490 vector signed int vec_signed (vector float);
9491 vector signed long long vec_signed (vector double);
9492 vector signed int vec_signed2 (vector double, vector double);
9493 vector signed int vec_signede (vector double);
9494 vector signed int vec_signedo (vector double);
9495 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
9496 instruction generator.
9497 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9498 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
9499 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
9501 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
9502 vunsignede_v2df): Add define_expands.
9503 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
9504 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
9505 VEC_UNSIGNEDO): Add definitions.
9506 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9507 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
9508 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
9509 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
9510 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
9511 * config/rs6000/altivec.h (vec_signed, vec_signed2,
9512 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
9513 vec_unsignede, vec_unsignedo): Add builtin defines.
9514 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
9516 * doc/extend.texi: Update the built-in documentation file for the
9517 new built-in functions.
9519 2017-06-29 Richard Biener <rguenther@suse.de>
9521 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
9522 reduction chains to LOOP_VINFO_REDUCTIONS.
9523 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
9524 SLP reductions after processing reduction chains.
9526 2017-06-29 Nathan Sidwell <nathan@acm.org>
9528 * builtins.c (fold_builtin_FUNCTION): Use
9529 lang_hooks.decl_printable_name.
9531 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
9534 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
9535 with only one label.
9536 * stmt.c (expand_case): Assert NCASES is greater than one.
9538 2017-06-29 Richard Biener <rguenther@suse.de>
9540 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
9542 (group_case_labels): Likewise.
9543 (find_taken_edge): Push sanity checking on val to workers...
9544 (find_taken_edge_cond_expr): ... here
9545 (find_taken_edge_switch_expr): ... and here, handle cases
9546 with just a default label.
9547 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
9548 (group_case_labels): Likewise.
9549 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
9550 group_case_labels does anything cleanup the CFG again.
9552 2017-06-29 Bin Cheng <bin.cheng@arm.com>
9554 PR tree-optimization/81196
9555 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
9556 exit condition comparing two IVs.
9558 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
9560 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
9561 profile to the dummy entry at the end of the list of architectures.
9562 * config/arm/arm-cpu-cdata.h: Regenerated.
9564 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9565 Michael Collison <michael.collison@arm.com>
9568 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
9570 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
9571 (*aarch64_reg_<mode>3_minus_mask): New pattern.
9572 (*aarch64_<optab>_reg_di3_mask2): New pattern.
9573 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
9574 of shift when the shift amount is masked with constant equal to
9575 the size of the mode.
9576 * config/aarch64/predicates.md (subreg_lowpart_operator): New
9579 2017-06-29 Martin Liska <mliska@suse.cz>
9581 * config/i386/i386.opt: Change range from [1,5] to [0,5].
9583 2017-06-29 Yury Gribov <tetra2005@gmail.com>
9586 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
9589 (edge_growth_cache_entry::edge_growth_cache_entry): New
9591 (reset_edge_growth_cache): Update to use constructor.
9593 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9595 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
9596 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9597 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9599 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
9601 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
9602 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
9604 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
9606 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
9607 (*-linux-uclibc*): Add t-uclibc tmake_file.
9608 * config/t-musl: New.
9609 * config/t-uclibc: New.
9611 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
9613 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
9615 (gen_comm_data): Emit architectural setting of arch_prof.
9616 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
9618 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
9619 (armv8-m.base, armv8-m.main): Likewise.
9620 * arm-protos.h (arm_build_target): Add profile field.
9621 (arch_option): Likewise.
9622 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
9624 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
9625 arm_active_target.profile.
9627 2017-06-28 Richard Biener <rguenther@suse.de>
9630 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
9631 TYPE_OVERFLOW_WRAPS.
9632 * match.pd (negate_expr_p): Likewise.
9633 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
9634 fold_build2, not fold_binary.
9636 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9638 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
9639 Convert memory address to Pmode.
9640 (aarch64_print_operand): Assert MEM operands are always Pmode.
9642 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9645 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
9646 Remove redundant if.
9647 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
9648 * config/arm/aarch-common-protos.h
9649 (aarch_forward_to_shift_is_not_shifted_re): Remove.
9650 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
9652 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
9655 * multiple_target.c (create_dispatcher_calls): Set the default
9656 clone to be static, not public.
9658 2017-06-28 Richard Biener <rguenther@suse.de>
9660 * tree-vect-loop.c (vectorizable_reduction): Move special
9661 cond reduction IV var creation ...
9662 (vect_create_epilog_for_reduction): ... here. Remove induction_index
9663 parameter. Use STMT_VINFO_VECTYPE.
9664 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
9667 2017-06-28 Martin Liska <mliska@suse.cz>
9670 * ipa-visibility.c (non_local_p): Handle visibility.
9672 2017-06-28 Martin Liska <mliska@suse.cz>
9675 * common.opt: Add IntegerRange to various options.
9676 * opt-functions.awk (integer_range_info): New function.
9677 * optc-gen.awk: Add integer_range_info to cl_options struct.
9678 * opts-common.c (decode_cmdline_option): Handle
9679 CL_ERR_INT_RANGE_ARG.
9680 (cmdline_handle_error): Likewise.
9681 * opts.c (print_filtered_help): Show valid interval in
9682 when --help is provided.
9683 * opts.h (struct cl_option): Add range_min and range_max fields.
9684 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
9686 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
9688 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
9689 (x * C EQ/NE y * C): New transformation.
9691 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
9693 * genmultilib (combination_space): Accept '+' in option names.
9695 2017-06-28 Martin Liska <mliska@suse.cz>
9698 * asan.c (instrument_derefs): Bail out inner references
9699 that are hard register variables.
9701 2017-06-28 Jakub Jelinek <jakub@redhat.com>
9704 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
9705 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
9707 2017-06-28 Richard Biener <rguenther@suse.de>
9709 * tree-vectorizer.h (vect_get_vec_defs): Remove.
9710 (vect_get_slp_defs): Adjust.
9711 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
9713 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
9715 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9716 get_initial_defs_for_reduction instead of vect_get_vec_defs.
9717 (vectorizable_reduction): Adjust.
9718 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
9720 (vect_get_slp_defs): Likewise.
9721 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
9722 (vectorizable_bswap): Adjust.
9723 (vectorizable_call): Likewise.
9724 (vectorizable_conversion): Likewise.
9725 (vectorizable_assignment): Likewise.
9726 (vectorizable_shift): Likewise.
9727 (vectorizable_operation): Likewise.
9728 (vectorizable_store): Likewise.
9729 (vectorizable_condition): Likewise.
9730 (vectorizable_comparison): Likewise.
9732 2017-06-28 Michael Collison <michael.collison@arm.com>
9735 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
9737 (arm_gen_movmemqi): Removed unused variable 'i'.
9738 Convert 'for' loop into 'while' loop.
9739 (arm_expand_prologue): Remove last unnecessary set of insn.
9740 (thumb_pop): Remove unused variable 'pushed_words'.
9741 (thumb_exit): Remove last unnecessary set of regs_to_pop.
9743 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9745 * config/s390/predicates.md: Use s390_rel_address_ok_p.
9746 * config/s390/s390-protos.h: Add prototype of
9747 s390_rel_address_ok_p.
9748 * config/s390/s390.c (s390_got_symbol): New function.
9749 (s390_rel_address_ok_p): New function.
9750 (legitimize_pic_address): Use s390_rel_address_ok_p.
9751 (s390_load_got): Use s390_got_symbol.
9752 (s390_option_override): Issue error if
9753 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
9754 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
9756 * config/s390/s390.opt: New option mpic-data-is-text-relative.
9758 2017-06-27 Andrew Pinski <apinski@cavium.com>
9760 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
9761 (X * copysign (1.0, X)): New pattern.
9762 (X * copysign (1.0, -X)): New pattern.
9763 (copysign (-1.0, CST)): New pattern.
9765 2017-06-27 Joseph Myers <joseph@codesourcery.com>
9767 * genmultilib (combination_space): Remove variable.
9768 Validate reuse rules against regular expression for any sequence
9769 of multilib options in any order.
9771 2017-06-27 Michael Collison <michael.collison@arm.com>
9773 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
9774 call aarch64_split_simd_combine.
9775 * (aarch64_combine_internal<mode>): Delete pattern.
9776 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
9777 Allow register and subreg operands.
9779 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9781 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
9782 specific need, just fallback on defaults.
9783 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
9785 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9786 Olivier Hainque <hainque@adacore.com>
9788 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
9790 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
9791 targets. Pick a default if no particular attempt applied.
9792 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
9795 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9797 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
9798 (x86_64-wrs-vxworks7): Likewise.
9800 2017-06-27 Marek Polacek <polacek@redhat.com>
9803 * ubsan.c (instrument_null): Check get_base_address's result for null.
9805 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
9807 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
9809 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
9811 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
9812 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
9813 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
9814 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
9816 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
9817 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
9818 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
9819 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
9820 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
9821 BUILT_IN_FEUPDATEENV): New builtins.
9822 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
9823 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
9824 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
9825 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
9827 (builtin_structptr_types): Adjust size.
9828 * tree.c (builtin_structptr_types): Add four entries.
9830 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9831 Olivier Hainque <hainque@adacore.com>
9833 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
9834 (TLS_SYM): New local macro, forcing reference to __tls__ on
9835 link command lines for VxWorks 7 RTPs, triggering initialization
9837 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
9838 OS features TLS support, true for RTPs on VxWorks 7.
9839 * config/vxworks.c (vxworks_override_options): Setup emutls
9842 2017-06-27 Jakub Jelinek <jakub@redhat.com>
9844 * predict.c (test_prediction_value_range): Use -1U instead of -1
9845 to avoid narrowing conversion warning.
9846 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
9847 to avoid narrowing conversion warning.
9848 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
9850 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
9852 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9854 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
9855 64bit configurations.
9856 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
9857 (SIZE_TYPE): Likewise.
9858 * config/vxworks.c (vxworks_emutls_var_fields): Use
9859 long_unsigned_type_node instead of unsigned_type_node as the offset
9860 field type, which is "pointer" mode in emutls.c.
9862 2017-06-27 Jakub Jelinek <jakub@redhat.com>
9865 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
9868 * gimple-fold.c (replace_call_with_call_and_fold): Handle
9869 gimple_vuse copying separately from gimple_vdef copying.
9871 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9873 * value-prof.c (free_hist): Remove call to memset and the enclosing if
9876 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
9877 Olivier Hainque <hainque@adacore.com>
9879 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
9880 for all vxworks7 targets.
9881 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
9882 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
9883 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
9884 variations for VX6/VX7 and 32/64bits later on in ...
9885 (VXWORKS_LIB_SPEC): Leverage new macros.
9886 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
9887 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
9889 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
9891 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
9892 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
9894 2017-06-26 Carl Love <cel@us.ibm.com>
9896 * config/rs6000/rs6000-c.c: Add support for built-in functions
9897 vector bool char vec_reve (vector bool char);
9898 vector signed char vec_reve (vector signed char);
9899 vector unsigned char vec_reve (vector unsigned char);
9900 vector bool int vec_reve (vector bool int);
9901 vector signed int vec_reve (vector signed int);
9902 vector unsigned int vec_reve (vector unsigned int);
9903 vector bool long long vec_reve (vector bool long long);
9904 vector signed long long vec_reve (vector signed long long);
9905 vector unsigned long long vec_reve (vector unsigned long long);
9906 vector bool short vec_reve (vector bool short);
9907 vector signed short vec_reve (vector signed short);
9908 vector double vec_reve (vector double);
9909 vector float vec_reve (vector float);
9910 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
9911 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
9912 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
9913 (altivec_vreve): New pattern.
9914 * config/rs6000/altivec.h (vec_reve): New define.
9915 * doc/extend.texi (vec_rev): Update the built-in documentation file
9916 for the new built-in functions.
9918 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9920 PR tree-optimization/71815
9921 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
9923 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
9925 (slsr_process_phi): Likewise.
9926 (replace_uncond_cands_and_profitable_phis): Don't replace a
9927 multiply candidate with a stride of 1 (copy or cast).
9928 (phi_incr_cost): Call uses_consumed_by_stmt rather than
9930 (lowest_cost_path): Likewise.
9931 (total_savings): Likewise.
9933 2017-06-26 Richard Biener <rguenther@suse.de>
9936 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
9937 Use def_builtin_pure for all gather builtins.
9939 2017-06-26 Richard Biener <rguenther@suse.de>
9941 PR tree-optimization/81203
9942 * tree-tailcall.c (find_tail_calls): Do not move stmts into
9945 2017-06-26 Marek Polacek <polacek@redhat.com>
9948 * doc/invoke.texi: Document -Wmultistatement-macros.
9950 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
9952 * doc/sourcebuild.texi (ARM-specific attributes): Document new
9953 arm_neon_ok_no_float_abi effective target.
9955 2017-06-26 Richard Biener <rguenther@suse.de>
9957 PR tree-optimization/80928
9958 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
9959 (copy_bbs): Set BB_DUPLICATED flag early.
9960 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
9962 (execute_on_shrinking_pred): Likewise.
9963 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
9964 BB_DUPLICATED blocks.
9965 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
9966 iterate over all PHIs considering removal of *gsi.
9968 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
9970 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
9973 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9975 * config/rs6000/rs6000-string.c: (expand_block_clear,
9976 do_load_for_compare, select_block_compare_mode,
9977 compute_current_alignment, expand_block_compare,
9978 expand_strncmp_align_check, expand_strn_compare,
9979 expand_block_move, rs6000_output_load_multiple)
9980 Move functions related to string/block move/compare
9982 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
9983 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
9984 for this function which is now used in two files.
9985 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
9986 * config.gcc: Add rs6000-string.o to extra_objs for
9987 targets powerpc*-*-* and rs6000*-*-*.
9989 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
9992 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
9993 32-bit, since indexed is not valid for DImode.
9994 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
9995 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
9996 (define_peephole2 for Altivec d-form load): Add 32-bit support.
9997 (define_peephole2 for Altivec d-form store): Likewise.
10000 * multiple_target.c (create_dispatcher_calls): Only create the
10001 dispatcher call if the function is the default clone of a
10002 versioned function.
10004 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10006 PR middle-end/80902
10007 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10008 a call, force the call to not be a tail call.
10010 2017-06-23 Jeff Law <law@redhat.com>
10012 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10015 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10017 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10018 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10019 handling for early expansion of vector shifts (sl,sr,sra,rl).
10020 (builtin_function_type): Add vector shift right instructions
10021 to the unsigned argument list.
10023 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10025 rtl-optimizatoin/79286
10026 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10027 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10028 trap. PIC register plus a const unspec without offset can never trap.
10030 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10032 * tree.h (builtin_structptr_type): New type.
10033 (builtin_structptr_types): Declare new array.
10034 * tree.c (builtin_structptr_types): New array.
10035 (free_lang_data, build_common_tree_nodes): Use it.
10037 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10040 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10043 2017-06-22 Matt Turner <mattst88@gmail.com>
10045 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10046 Lake models to skylake case. Assume skylake for unknown
10047 models with clflushopt.
10049 2017-06-22 Jeff Law <law@redhat.com>
10051 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10052 frame sizes that do not satisfy aarch64_uimm12_shift.
10054 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10056 * profile-count.h (apply_probability,
10057 apply_scale, probability_in): Fix checks for zero.
10059 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10061 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10062 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10064 2016-06-22 Richard Biener <rguenther@suse.de>
10066 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10067 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10068 REDUC_MAX_EXPR support.
10069 (vectorizable_reduction): Likewise.
10070 (vect_create_epilog_for_reduction): Likewise.
10072 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10074 * match.pd (A / (1 << B) -> A >> B): New.
10075 * generic-match-head.c: Include optabs-tree.h.
10076 * gimple-match-head.c: Likewise.
10077 * optabs-tree.h (target_supports_op_p): New.
10078 * optabs-tree.c (target_supports_op_p): New.
10080 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10082 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10083 $gcc_cv_ld --help output.
10084 (gcc_cv_ld_demangle): Likewise.
10085 (gcc_cv_ld_eh_frame_hdr): Likewise.
10086 (gcc_cv_ld_pie): Likewise.
10087 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10088 (gcc_cv_ld_buildid): Likewise.
10089 (gcc_cv_ld_sysroot): Likewise.
10090 (ld_bndplt_support): Likewise.
10091 (ld_pushpopstate_support): Likewise.
10092 * configure: Regenerate.
10093 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10095 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10098 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10099 operands indexes to avoid gap between operands and match_dups.
10101 2017-06-21 Andrew Pinski <apinski@cavium.com>
10103 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
10104 Increment Arith_shift and Arith_shift_reg by 1.
10105 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
10107 * config/aarch64/aarch64.c (thunderx_tunings): Enable
10108 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
10109 (aarch64_strip_extend): Add new argument and test for it.
10110 (aarch64_cheap_mult_shift_p): New function.
10111 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
10112 add a cost if it is true.
10113 Update calls to aarch64_strip_extend.
10114 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
10116 2017-06-21 Andrew Pinski <apinski@cavium.com>
10118 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
10120 (thunderxt88): Likewise.
10121 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
10122 (thunderx_prefetch_tune): New variable.
10123 (thunderx2t99_prefetch_tune): Update for the correct values.
10124 (thunderxt88_tunings): New variable.
10125 (thunderx_tunings): Use thunderx_prefetch_tune instead of
10126 generic_prefetch_tune.
10127 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
10129 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10131 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
10132 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
10133 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
10134 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
10135 (aarch64_atomic_cas<mode>, GPI): Likewise.
10137 2017-06-21 Martin Liska <mliska@suse.cz>
10139 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
10140 statements on cold and hot labels.
10141 * predict.c (tree_estimate_probability_bb): Remove the
10142 prediction from this place.
10144 2017-06-21 Martin Liska <mliska@suse.cz>
10146 PR tree-optimization/79489
10147 * gimplify.c (maybe_add_early_return_predict_stmt): New
10149 (gimplify_return_expr): Call the function.
10150 * predict.c (tree_estimate_probability_bb): Remove handling
10152 * predict.def: Update comment about early return predictor.
10153 * gimple-predict.h (is_gimple_predict): New function.
10154 * predict.def: Change default value of early return to 66.
10155 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
10157 * passes.def: Put pass_strip_predict_hints to the beginning of
10160 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
10162 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
10163 FUNCTION_DECL declarations.
10164 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
10166 (dwaf2out_decl): Likewise.
10167 * godump.c (go_early_global_decl): Skip call to the real debug hook
10168 for FUNCTION_DECL declarations.
10169 * passes.c (rest_of_decl_compilation): Skip call to the
10170 early_global_decl debug hook for FUNCTION_DECL declarations, unless
10171 -fdump-go-spec is passed.
10173 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10175 * config/i386/i386.c (struct builtin_isa): New field pure_p.
10176 Reorder for compactness.
10177 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
10178 (def_builtin_pure, def_builtin_pure2): New functions.
10179 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
10181 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10183 * match.pd (nop_convert): New predicate.
10184 ((A +- CST1) +- CST2): Allow some NOP conversions.
10186 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10189 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
10190 with ctors/dtors if GOVD_SHARED is set.
10192 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10194 * config/aarch64/aarch64.md (movti_aarch64):
10195 Emit mov rather than orr.
10196 (movtf_aarch64): Likewise.
10197 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
10198 Emit mov rather than orr.
10200 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10202 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
10203 Swap alternatives, make integer dup more expensive.
10205 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10207 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
10208 Return true for non-tls symbols.
10210 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
10212 * config/aarch64/aarch64-cores.def (cortex-a55): New.
10213 (cortex-a75): Likewise.
10214 (cortex-a75.cortex-a55): Likewise.
10215 * config/aarch64/aarch64-tune.md: Regenerate.
10216 * doc/invoke.texi (-mtune): Document new values for -mtune.
10218 2017-06-21 Tom de Vries <tom@codesourcery.com>
10220 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
10221 stack_size feature.
10222 (Effective-Target Keywords, Other attributes): Suggest using
10223 dg-add-options stack_size feature to get stack limit in stack_size
10224 effective target documentation.
10226 2017-06-21 Julian Brown <julian@codesourcery.com>
10227 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10229 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
10230 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
10231 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
10233 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
10234 attribute type list for neon_multiply.
10235 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
10236 attribute type list for neon_multiply.
10237 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
10238 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
10239 attribute type list for neon_multiply.
10240 * config/arm/types.md (crypto_pmull): Add.
10241 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
10242 attribute type list.
10244 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
10246 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
10249 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10251 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
10252 to make sure not to dereference a NULL cost_classes_ptr pointer.
10254 2017-06-20 Carl Love <cel@us.ibm.com>
10256 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10257 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10258 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10259 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10260 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10261 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10262 VMULOSW): New enum "unspec" values.
10263 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10264 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10265 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10266 altivec_vmulosw): New patterns.
10267 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10268 VMULOSW): Add definitions.
10270 2017-06-20 Julia Koval <julia.koval@intel.com>
10272 * config/i386/i386.c: Fix rounding expand for new pattern.
10273 * config/i386/subst.md: Fix pattern (parallel -> unspec).
10275 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10277 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
10278 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
10280 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10282 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
10285 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10287 * config/aarch64/aarch64-cores.def: Rearrange to sort by
10288 architecture, then by implementer ID.
10289 * config/aarch64/aarch64-tune.md: Regenerate.
10291 2017-06-20 Richard Biener <rguenther@suse.de>
10293 PR middle-end/81097
10294 * fold-const.c (split_tree): Fold to type before negating.
10296 2017-06-20 David Malcolm <dmalcolm@redhat.com>
10298 * diagnostic-show-locus.c
10299 (selftest::test_fixit_deletion_affecting_newline): New function.
10300 (selftest::diagnostic_show_locus_c_tests): Call it.
10302 2017-06-20 Andreas Schwab <schwab@suse.de>
10305 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
10308 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10310 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
10311 __ARM_FEATURE_COPROC according to support.
10313 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10315 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10316 Rewritten to avoid overflow for > 32-bit pointers.
10319 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
10320 by removing enum keyword.
10321 (ubsan_type_descriptor): Likewise. Formatting fix.
10324 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
10325 splitter): Require TARGET_SSE2 in the condition.
10327 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
10330 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
10331 for doing vector set of SFmode on ISA 3.0.
10332 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
10333 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
10335 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
10336 SFmode value into a V4SF variable that was extracted from another
10337 V4SF variable without converting the element to double precision
10338 and back to single precision vector format.
10339 (vsx_insert_extract_v4sf_p9_2): Likewise.
10341 2017-06-19 Jakub Jelinek <jakub@redhat.com>
10343 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
10344 in UWHI to avoid undefined overflow.
10347 * ubsan.h (enum ubsan_encode_value_phase): New.
10348 (ubsan_encode_value): Change second argument to
10349 enum ubsan_encode_value_phase with default value of
10350 UBSAN_ENCODE_VALUE_GENERIC.
10351 * ubsan.c (ubsan_encode_value): Change second argument to
10352 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
10353 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
10354 create_tmp_var_raw instead of create_tmp_var and use a
10356 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
10357 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
10358 ubsan_encode_value callers.
10361 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
10362 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
10363 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
10365 2017-06-19 Richard Biener <rguenther@suse.de>
10367 PR middle-end/81118
10368 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
10369 estimates if we changed anything.
10371 2017-06-19 Richard Biener <rguenther@suse.de>
10373 PR tree-optimization/80887
10374 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
10375 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
10376 simplified lookups, then reset mprts_hook.
10377 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
10379 (try_to_simplify): Likewise.
10381 2017-06-19 Martin Liska <mliska@suse.cz>
10384 * gimplify.c (gimplify_switch_expr):
10385 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
10387 2017-06-19 Martin Liska <mliska@suse.cz>
10389 * doc/install.texi: Document that PGO runs in 4 stages.
10391 2017-06-19 Martin Liska <mliska@suse.cz>
10394 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
10395 to dispatcher function name.
10396 * multiple_target.c (replace_function_decl): New function.
10397 (create_dispatcher_calls): Redirect both edges and references.
10399 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
10401 * profile-count.c (profile_count::dump): Dump quality.
10402 (profile_count::differs_from_p): Update for unsigned val.
10403 * profile-count.h (profile_count_quality): New enum.
10404 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
10406 2017-06-19 Richard Biener <rguenther@suse.de>
10408 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
10409 struct function as arg.
10410 (estimate_numbers_of_iterations): Export overload with loop arg.
10411 (free_numbers_of_iterations_estimates_loop): Use an overload of
10412 free_numbers_of_iterations_estimates instead.
10413 * tree-cfg.c (remove_bb): Adjust.
10414 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
10415 * tree-parloops.c (gen_parallel_loop): Likewise.
10416 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
10418 (tree_unroll_loops_completely): Likewise.
10419 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
10420 Use an overload instead and export.
10421 (estimated_loop_iterations): Adjust.
10422 (max_loop_iterations): Likewise.
10423 (likely_max_loop_iterations): Likewise.
10424 (estimate_numbers_of_iterations): Take struct function as arg
10426 (loop_exits_before_overflow): Adjust.
10427 (free_numbers_of_iterations_estimates_loop): Use an overload.
10428 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
10429 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
10431 2017-06-19 Richard Biener <rguenther@suse.de>
10434 * ipa-prop.c (find_constructor_constant_at_offset): Handle
10435 RANGE_EXPR conservatively.
10437 2017-06-16 Carl Love <cel@us.ibm.com>
10439 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10440 definitions for vec_float, vec_float2, vec_floato,
10441 vec_floate built-ins.
10442 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
10443 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
10445 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
10446 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
10447 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
10448 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
10449 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
10450 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
10451 vec_floato): Add builtin defines.
10452 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
10453 Update the built-in documentation file for the new built-in
10456 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10458 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
10459 (mthumb): Mark as the negative of -marm.
10461 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10463 * doc/invoke.texi (ARM Options, -mcpu): Document supported
10465 (ARM Options, -mtune): Document that this accepts the same
10466 extension options as -mcpu.
10467 (ARM Options, -mfpu): Document addition of -mfpu=auto.
10469 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10471 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
10472 permitted extensions.
10474 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10476 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
10477 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
10478 (armv8-m.main): Add option +nodsp.
10479 * config/arm/arm-cpu-cdata.h: Regenerated.
10481 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10483 * config/arm/t-fuchsia: New file.
10484 * config.gcc (arm*-*-fuchsia*): Use it.
10486 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10488 * config/arm/t-symbian: Rewrite for new option infrastructure.
10490 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10492 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
10493 (MULTILIB_REQUIRED): Likewise.
10495 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10497 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
10498 (MULTILIB_RESUE): Likewise.
10499 (MULTILIB_MATCHES): Likewise.
10500 (MULTLIB_REQUIRED): Likewise.
10502 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10504 * config/arm/t-rtems: Rewrite for new option framework.
10506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10508 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
10509 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
10510 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
10511 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
10512 * config/arm/t-multilib: ... here.
10513 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
10514 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
10515 armv7-a and armv8*-a when A-profile libraries have not been built.
10516 * config/arm/t-rmprofile: Rewrite.
10518 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10520 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
10521 with a backslash. Remove the backslash after substituting unescaped
10523 * doc/fragments.texi (MULTILIB_REUSE): Document it.
10525 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10527 * config.gcc: (arm*-*-*): When building a-profile libraries, force
10528 the driver to pass through the default setting of -mfloat-abi.
10529 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
10531 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
10532 (all_feat_combs): New rule.
10533 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
10535 * config/arm/t-aprofile: Rewrite.
10537 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10539 * config/arm/arm.h (FPUTYPE_AUTO): Define.
10540 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
10541 fpu is not specified by the user/command-line.
10542 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
10543 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
10544 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
10545 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
10546 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
10547 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
10549 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10551 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
10552 * config/arm/t-arm-elf: Rewritten.
10554 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10556 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
10557 have some floating-point instructions.
10558 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
10559 (TARGET_MAYBE_HARD_FLOAT): New macro.
10560 * config/arm/arm-builtins.c (arm_init_builtins): Use
10561 TARGET_MAYBE_HARD_FLOAT.
10562 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
10564 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10566 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
10567 (configargs.h): Include it.
10568 (arm_print_hint_for_fpu_option): New function.
10569 (arm_parse_fpu_option): New function.
10570 (candidate_extension): New class.
10571 (arm_canon_for_multilib): New function.
10572 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
10573 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
10574 (ARCH_CANONICAL_SPECS): New macro.
10575 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
10577 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10579 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
10580 are set after handling multilib fragments. Set target_cpu_default2
10583 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10585 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
10587 (arm*-*-*): Set target_cpu_default2 to a quoted string.
10588 * config/arm/parsecpu.awk (check_cpu): Validate any extension
10590 (check_arch): Likewise.
10591 * config/arm/arm.c (arm_configure_build_target): Handle
10592 TARGET_CPU_DEFAULT being a string constant. Scan any feature
10593 options in the default.
10595 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10597 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
10598 when an option is an alias of another.
10599 * config/arm/parsecpu.awk (optalias): New parser token.
10600 (gen_comm_data): Mark non-alias options as such. Emit entries
10601 for extension aliases.
10602 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
10603 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
10604 (armv6kz, armv6zk, armv6t2): Likewise.
10605 (armv7): Make vfpv3-d16 an alias.
10606 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
10608 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
10609 Sort in canonical order.
10610 (armv8-a): Sort in canonical order.
10611 (armv8.1-a, armv8.2-a): Likewise.
10612 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
10614 (cortex-a9): Sort in canonical order.
10615 * config/arm/arm.c (selftests.h): Include it.
10616 (arm_test_cpu_arch_data): New function.
10617 (arm_run_self_tests): New function.
10618 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
10619 (targetm): Move declaration to the end of the file.
10620 * arm-cpu-cdata.h: Regenerated.
10622 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10624 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
10625 call to target_mode_check describing the type of option passed.
10626 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
10627 (arm_target_thumb_only): Use arm_parse_arch_option_name or
10628 arm_parse_cpu_option_name to match parameters against list of
10630 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
10631 arm_arch_core_flags data structure.
10632 * config/arm/arm-cpu_cdata.h: Regenerated.
10634 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10636 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
10638 (arm_print_hint_for_cpu_option): Likewise.
10639 (arm_print_hint_for_arch_option): Likewise.
10640 (arm_parse_cpu_option_name): Likewise.
10641 (arm_parse_arch_option_name): Likewise.
10642 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
10643 of entries in the all_fpus list.
10644 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
10645 (arm_parse_cpu_option_name): Declare.
10646 (arm_parse_arch_option_name): Declare.
10647 (arm_parse_option_features): Declare.
10648 (arm_intialize_isa): Declare.
10649 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
10651 (gen_comm_data): ... here. Make definitions non-static.
10652 * config/arm/arm-cpu-data.h: Regenerated.
10653 * config/arm/arm-cpu-cdata.h: Regenerated.
10655 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10657 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
10658 (cpu_arch_extension): New structure.
10659 (cpu_arch_option, arch_option, cpu_option): New structures.
10660 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
10661 architecture types.
10662 (gen_data): Generate new format data tables.
10663 * config/arm/arm.c (cpu_tune): New structure.
10664 (cpu_option, processors): Delete.
10665 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
10666 (arm_print_hint_for_cpu_option): ... this and ...
10667 (arm_print_hint_for_arch_option): ... this.
10668 (arm_parse_arch_cpu_name): Delete. Replace with ...
10669 (arm_parse_cpu_option_name): ... this and ...
10670 (arm_parse_arch_option_name): ... this.
10671 (arm_unrecognized_feature): Change type of target parameter to
10673 (arm_parse_arch_cpu_features): Delete. Replace with ...
10674 (arm_parse_option_features): ... this.
10675 (arm_configure_build_target): Rework to use new configuration data
10677 (arm_print_tune_info): Rework for new configuration data tables.
10678 * config/arm/arm-cpu-data.h: Regenerated.
10679 * config/arm/arm-cpu.h: Regenerated.
10681 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10683 * Makefile.in (OBJS): Move sbitmap.o from here ...
10684 (OBJS-libcommon): ... to here.
10686 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10688 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
10689 (ISA_ALL_CRYPTO): New macro.
10690 (ISA_ALL_SIMD): New macro
10691 (ISA_ALL_FP): New macro.
10692 * config/arm/arm.c (fpu_bitlist): Update initializer.
10693 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
10695 (arm9e): Add fpu. Add option for nofp
10696 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
10697 (arm926ej-s, arm1026ej-s): Likewise.
10698 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
10699 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
10700 neon-fp16, neon-vfpv4, nofp and nosimd.
10701 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
10702 (cortex-a8): Add fpu. Add option for nofp.
10703 (cortex-a9): Add fpu. Add options for nosimd and nofp.
10704 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
10705 (cortex-r4f): Add fpu.
10706 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
10707 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
10709 (cortex-r8): Likewise.
10710 (cortex-m4): Add fpu. Add option for nofp.
10711 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
10712 (cortex-a17.cortex-a7): Likewise.
10713 (cortex-a32): Add fpu. Add options for crypto and nofp.
10714 (cortex-a35, cortex-a53): Likewise.
10715 (cortex-a57): Add fpu. Add option for crypto.
10716 (cortex-a72, cortex-a73): Likewise.
10717 (exynos-m1): Likewise.
10718 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
10719 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
10720 (cortex-m33): Add fpu. Add option for nofp.
10721 * config/arm/arm-cpu-cdata.h: Regenerated
10722 * config/arm/arm-cpu-data.h: Regenerated.
10724 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10726 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
10727 (armv5te, armv5tej): Likewise.
10728 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
10729 (armv7): Add options fp and vfpv3-d16.
10730 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
10731 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
10733 (armv7ve): Likewise.
10734 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
10735 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
10736 (armv8-a): Add nocrypto option.
10737 (armv8.1-a, armv8.2-a): Likewise.
10738 (armv8-m.main): add options fp, fp.dp and nofp.
10740 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10742 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
10744 (armv8-a+crc): Delete.
10745 (armv8.1-a): Add options simd, crypto and nofp.
10746 (armv8.2-a): Add options fp16, simd, crypto and nofp.
10747 (armv8.2-a+fp16): Delete.
10748 (armv8-m.main): Add option dsp.
10749 (armv8-m.main+dsp): Delete.
10750 (cortex-a8): Add fpu. Add nofp option.
10751 (cortex-a9): Add fpu. Add nofp and nosimd options.
10752 * config/arm/parsecpu.awk (gen_data): Generate option tables and
10753 link to main cpu and architecture data structures.
10754 (gen_comm_data): Only put isa attributes from the main architecture
10756 (option): New statement for architecture and CPU entries.
10757 * arm.c (struct cpu_option): New structure.
10758 (struct processors): Add entry for options.
10759 (arm_unrecognized_feature): New function.
10760 (arm_parse_arch_cpu_name): Ignore any characters after the first
10762 (arm_parse_arch_cpu_feature): New function.
10763 (arm_configure_build_target): Separate out any CPU and architecture
10764 features and parse separately. Don't error out if -mfpu=auto is
10765 used with only an architecture string.
10766 (arm_print_asm_arch_directives): New function.
10767 (arm_file_start): Call it.
10768 * config/arm/arm-cpu-cdata.h: Regenerated.
10769 * config/arm/arm-cpu-data.h: Likewise.
10770 * config/arm/arm-tables.opt: Likewise.
10772 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10774 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
10775 assembler when it is not -mfpu=auto.
10777 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10779 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
10780 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
10781 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
10782 (ASM_CPU_SPEC): Rewrite.
10783 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
10784 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
10785 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
10786 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
10787 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
10788 copied string is NUL-terminated. Also strip any characters prefixed
10790 (arm_rewrite_selected_arch): New function.
10791 (arm_rewrite_march): New function.
10793 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10795 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
10796 (x_arm_cpu_string, x_arm_tune_string): Likewise.
10797 (march, mcpu, mtune): Convert to string-based options.
10798 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
10799 (arm_parse_arch_cpu_name): New function.
10800 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
10801 identify selected architecture or CPU.
10802 (arm_option_save): New function.
10803 (TARGET_OPTION_SAVE): Redefine.
10804 (arm_option_restore): Restore string options.
10805 (arm_option_print): Print string options.
10807 2017-06-16 Martin Sebor <msebor@redhat.com>
10809 PR tree-optimization/80933
10810 PR tree-optimization/80934
10811 * builtins.c (fold_builtin_3): Do not handle bcmp here.
10812 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
10813 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
10814 (gimple_fold_builtin): Call them.
10816 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10818 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
10819 as unlikely; update profile.
10821 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10823 * predict.c (force_edge_cold): Handle declaring edges impossible
10826 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10828 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
10830 (try_unroll_loop_completely): Fix reporting.
10832 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10834 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
10836 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
10839 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
10840 if given a non-constant argument for an intrinsic which requires a
10843 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10845 * profile.c (compare_freqs): New function.
10846 (branch_prob): Sort edge list.
10847 (find_spanning_tree): Assume that the list is priority sorted.
10849 2017-06-16 Richard Biener <rguenther@suse.de>
10851 PR tree-optimization/81090
10852 * passes.def (pass_record_bounds): Remove.
10853 * tree-pass.h (make_pass_record_bounds): Likewise.
10854 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
10855 make_pass_record_bounds): Likewise.
10856 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
10857 not free niter estimates at the beginning but at the end.
10858 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
10860 2017-06-16 Richard Biener <rguenther@suse.de>
10862 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
10863 initializer to workaround ICE in host GCC 4.8.
10865 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10867 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
10869 (clone_inlined_nodes): Update.
10871 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10873 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
10874 prefetch settings, and enable prefetching by default at -O3.
10876 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10878 * config/aarch64/aarch64.c (aarch64_override_options_internal):
10879 Set flag_prefetch_loop_arrays according to tuning data.
10881 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10883 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
10884 New tune structure.
10885 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
10886 [Unrelated to main purpose of the patch] Place the pointer field last
10887 to enable type checking errors when tune structure are wrongly merged.
10888 * config/aarch64/aarch64.c (generic_prefetch_tune,)
10889 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
10890 (thunderx2t99_prefetch_tune): New tune constants.
10891 (tune_params *_tunings): Update all tunings (no functional change).
10892 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
10893 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
10894 from tunings structures.
10896 2017-06-16 Jakub Jelinek <jakub@redhat.com>
10899 * ubsan.c (instrument_null): Add T argument, use it instead
10900 of computing it based on IS_LHS.
10901 (instrument_object_size): Likewise.
10902 (pass_ubsan::execute): Adjust instrument_null and
10903 instrument_object_size callers to pass gimple_get_lhs or
10904 gimple_assign_rhs1 result to it. Use instrument_null instead of
10905 calling get_base_address and instrument_mem_ref. Handle
10906 aggregate call arguments for object-size sanitization.
10908 2017-06-16 Yury Gribov <tetra2005@gmail.com>
10910 PR tree-optimization/81089
10911 * tree-vrp.c (is_masked_range_test): Validate operands of
10914 2017-06-15 Martin Sebor <msebor@redhat.com>
10917 * dumpfile.c (dump_register): Avoid calling memset to initialize
10918 a class with a default ctor.
10919 * gcc.c (struct compiler): Remove const qualification.
10920 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
10921 * hash-table.h: Ditto.
10922 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
10924 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
10925 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
10927 * params.h (struct param_info): Make struct members non-const.
10928 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
10929 with default initialization.
10930 * vec.h (vec_copy_construct, vec_default_construct): New helper
10932 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
10933 with vec_copy_construct.
10934 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
10935 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
10936 * doc/invoke.texi (-Wclass-memaccess): Document.
10938 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10940 * emit-rtl.h (is_leaf): Update comment about local
10941 register allocator.
10943 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10946 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
10947 for a variable to have a section before checking if the section has a
10949 Set section to.persistent if persistent attribute is set.
10950 Warn if .persistent attribute is used on an automatic variable.
10952 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
10954 PR rtl-optimization/80474
10955 * reorg.c (update_block): Do not ignore instructions in a delay slot.
10957 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
10959 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
10962 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
10964 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
10965 (casesi): Emit bounds checking as RTL.
10966 (casesi_internal_mips16_<mode>): Remove bounds checking.
10968 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
10970 * config/xtensa/xtensa.c (xtensa_option_override): Append
10971 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
10972 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
10973 xtensa_doloop_hooks): Define unconditionally.
10974 (xtensa_reorg_loops): Only call reorg_loops in the presence of
10976 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
10977 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
10978 for it in xtensa_option_override.
10979 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
10980 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
10982 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
10984 * doc/cppopts.texi: Document '-' special value to -MF.
10986 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
10988 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
10989 (cortex_a53_fconst): Likewise.
10990 (cortex_a53_fpmul): Likewise.
10991 (cortex_a53_f_load_64): Likewise.
10992 (cortex_a53_f_load_many): Likewise.
10993 (cortex_a53_advsimd_alu): Likewise.
10994 (cortex_a53_advsimd_alu_q): Likewise.
10995 (cortex_a53_advsimd_mul): Likewise.
10996 (cortex_a53_advsimd_mul_q): Likewise.
10997 (fpmac bypass): Add new bypass for fpmac-fpmac case.
10998 Add missing fmul, r2f_cvt and fconst cases.
11000 2017-06-14 Richard Biener <rguenther@suse.de>
11002 PR middle-end/81088
11003 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11005 (fold_binary_loc): When associating do not treat pre-existing
11006 TREE_OVERFLOW on literal constants as a reason to allow
11007 TREE_OVERFLOW on associated literal constants.
11009 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11011 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11012 (MASK_FEATURES): New macro.
11013 * config/sparc/sparc.c (sparc_option_override): Remove the special
11014 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11016 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11018 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11019 a division of 0 if non-call exceptions are enabled.
11021 2017-06-14 Andrew Pinski <apinski@cavium.com>
11022 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11025 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11026 Improve vector initialization code gen for only variable case.
11028 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11030 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11032 2017-06-14 Richard Biener <rguenther@suse.de>
11034 PR tree-optimization/81083
11035 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11038 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11040 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11041 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11042 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11043 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11044 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11045 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11047 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11049 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11050 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11052 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11054 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11056 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11058 * config/rs6000/t-rtems: Don't handle SPE.
11060 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11062 * config/rs6000/t-linux: Don't handle SPE.
11064 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11066 * config/rs6000/eabispe.h: Delete file.
11068 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11070 * config/rs6000/t-spe: Delete file.
11072 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11074 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11075 (rs6000_legitimate_offset_address_p): Return false for anything in
11076 V2SImode or V2SFmode.
11078 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11080 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11081 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11082 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11084 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11085 (rs6000_legitimate_offset_address_p): Ditto.
11086 (rs6000_emit_move): Ditto.
11087 (rs6000_init_builtins): Remove V4HI_type_node.
11089 2017-06-13 Martin Liska <mliska@suse.cz>
11092 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11093 (gate_asan): Likewise.
11094 * asan.h (asan_no_sanitize_address_p): Remove the function.
11095 (sanitize_flags_p): New function.
11096 * builtins.def: Fix coding style.
11097 * common.opt: Use renamed enum value.
11098 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11099 * doc/extend.texi: Document no_sanitize attribute.
11100 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
11101 to SANITIZE_UNDEFINED_NONDEFAULT.
11102 * gcc.c (sanitize_spec_function): Use the renamed enum value.
11103 * gimple-fold.c (optimize_atomic_compare_exchange_p):
11104 Use sanitize_flags_p.
11105 * gimplify.c (gimplify_function_tree): Likewise.
11106 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
11107 * opts.c (parse_no_sanitize_attribute): New function.
11108 (common_handle_option): Use renamed enum value.
11109 * opts.h (parse_no_sanitize_attribute): Declare.
11110 * tree.c (sanitize_flags_p): New function.
11111 * tree.h: Declared here.
11112 * tsan.c: Use sanitize_flags_p.
11113 * ubsan.c (ubsan_expand_null_ifn): Likewise.
11114 (instrument_mem_ref): Likewise.
11115 (instrument_bool_enum_load): Likewise.
11116 (do_ubsan_in_current_function): Remove the function.
11117 (pass_ubsan::execute): Use sanitize_flags_p.
11118 * ubsan.h: Remove do_ubsan_in_current_function
11119 * tree-cfg.c (print_no_sanitize_attr_value): New function.
11120 (dump_function_to_file): Use it here.
11122 2017-06-13 Martin Jambor <mjambor@suse.cz>
11124 PR tree-optimization/80803
11125 PR tree-optimization/81063
11126 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
11127 (propagate_subaccesses_across_link): Enqueue subtree whenever
11128 necessary instead of relying on the caller.
11130 2017-06-13 Martin Jambor <mjambor@suse.cz>
11132 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
11133 that have a first_link.
11134 (sort_and_splice_var_accesses): Do not check first_link before
11136 (subtree_mark_written_and_enqueue): Likewise.
11137 (propagate_all_subaccesses): Likewise and do not stop at first
11138 parent with a first_link.
11140 2017-06-13 Martin Jambor <mjambor@suse.cz>
11142 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
11145 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11147 * match.pd: New pattern.
11149 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11151 * tree-vrp.c (is_masked_range_test): New function.
11152 (register_edge_assert_for): Determine ranges for
11155 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11157 PR tree-optimization/67328
11158 * fold-const.c (maskable_range_p): New function.
11159 (build_range_check): Generate bittests if possible.
11161 2017-06-13 Martin Liska <mliska@suse.cz>
11163 * gimple-pretty-print.c (dump_probability): Add new argument.
11164 (dump_edge_probability): Dump both probability and count.
11165 (dump_gimple_label): Likewise.
11166 (dump_gimple_bb_header): Likewise.
11168 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
11171 * config/avr/avr-devices.c: Fix indentation.
11172 * config/avr/gen-avr-mmcu-specs.c: Dito.
11174 2017-06-13 Richard Biener <rguenther@suse.de>
11176 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
11177 instead get vector type from stmt_info.
11178 (vectorizable_reduction): Adjust. Remove dead code.
11180 2017-06-13 Richard Biener <rguenther@suse.de>
11182 PR middle-end/81065
11183 * fold-const.c (extract_muldiv_1): Remove bogus distribution
11184 case of C * (x * C2 + C3).
11185 (fold_addr_of_array_ref_difference): Properly fold index difference.
11187 2017-06-12 David S. Miller <davem@davemloft.net>
11190 * config/sparc/sparc.md (return expander): Emit frame blockage if
11191 function uses alloca.
11193 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11195 * combine.c (make_field_assignment): Check len rather than the mode
11196 precision when calling force_to_mode.
11198 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
11200 Support multilibs and devices that see flash in RAM address range.
11203 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
11204 (avr_mcu_t) <flash_pm_offset>: New field.
11205 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
11206 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
11207 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
11208 (AVR_TINY_PM_OFFSET): Remove macro.
11209 * config/avr/avr.opt (-mshort-calls): New option.
11210 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11211 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
11212 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
11213 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
11214 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
11215 instead of avr_arch->have_jmp_call.
11216 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
11217 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
11218 avr_arch->flash_pm_offset to define.
11219 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
11220 new field flash_pm_offset. Add entry for avrxmega3.
11221 (avr_texinfo): Add entry for avrxmega3.
11222 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
11223 attiny212, attiny214,
11224 attiny412, attiny414, attiny416, attiny417,
11225 attiny814, attiny816, attiny817,
11226 attiny1614, attiny1616, attiny1617,
11227 attiny3214, attiny3216, attiny3217.
11228 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
11229 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
11230 (avr_print_operand_address) [AVR_TINY]: Same.
11231 (avr_asm_init_sections) <readonly_data_section>: Only patch
11232 callback if avr_arch->flash_pm_offset = 0.
11233 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
11234 for rodata if avr_arch->flash_pm_offset != 0.
11235 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
11236 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
11237 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
11238 (m_options): Append opt_rcall.
11239 (m_dirnames): Append dir_rcall.
11240 * config/avr/t-multilib: Regenerate.
11242 * configure.ac [target=avr]: Check whether avrxmega3 default
11243 linker description file works as needed.
11244 * configure: Regenerate.
11245 * doc/avr-mmcu.texi: Regenerate.
11246 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
11247 <__AVR_ARCH__>: Document avrxmega3 and 103.
11248 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
11249 <__AVR_SHORT_CALLS__>: Document it.
11250 <__AVR_PM_BASE_ADDRESS__>: Document it.
11251 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
11252 (AVR Variable Attributes) <progmem>: Document this is
11253 not needed for avrxmega3.
11254 (AVR Named Address Spaces) <__flash>: Dito.
11256 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
11258 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
11260 2017-06-12 Doug Rupp <rupp@adacore.com>
11262 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
11263 Append vxworks-stdint.h to the tm_file list.
11264 * config/vxworks-stdint.h: New file.
11266 2017-06-12 Martin Liska <mliska@suse.cz>
11268 PR tree-optimization/81041
11269 * tree-profile.c (gimple_gen_ic_func_profiler):
11270 Create an extra BB in profile-generate
11271 (gimple_gen_time_profiler): Likewise.
11273 2017-06-12 Jakub Jelinek <jakub@redhat.com>
11275 PR tree-optimization/81003
11276 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
11277 (update_range_test): Use it instead of force_gimple_operand_gsi.
11279 2017-06-12 Richard Biener <rguenther@suse.de>
11281 PR tree-optimization/81053
11282 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
11283 with backedge value not defined in loop. Simplify def stmt
11286 2017-06-11 Tom de Vries <tom@codesourcery.com>
11289 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
11291 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
11292 nvptx_cannot_force_const_mem.
11294 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11296 * opts.c (finish_options): Move test for flag_split_stack after
11297 it has been initialized.
11299 2017-06-11 Jason Merrill <jason@redhat.com>
11301 * tree.h (id_equal): New.
11302 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
11303 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
11304 instead of strcmp of IDENTIFIER_POINTER.
11306 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11308 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
11309 (mark_all_inlined_calls_cdtor): Fix formating.
11310 (inline_transform): Rescale profile before inlining.
11312 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11314 * cgraph.h (cgraph_edge::clone): Update prototype.
11315 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
11316 (cgraph_node::create_clone): Update.
11317 (cgraph_node::create_version_clone): Update.
11318 * tree-inline.c (copy_bb): Update.
11319 (expand_call_inline): Update.
11321 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11323 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
11324 factored out from ...
11325 (rs6000_emit_prologue): ... here.
11327 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11329 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
11330 factored out from ...
11331 (rs6000_emit_prologue): ... here.
11333 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11335 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
11337 (handle_missing_profiles): Fix computation of tp_first_run.
11338 (counts_to_freqs): Do not touch freqs when count is 0.
11340 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11342 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
11345 2017-06-10 Tom de Vries <tom@codesourcery.com>
11347 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
11348 attributes): Document signal effective target.
11350 2017-06-10 Tom de Vries <tom@codesourcery.com>
11352 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
11353 Document effective target stack_size.
11355 2017-06-09 David Malcolm <dmalcolm@redhat.com>
11357 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
11358 to the edit_context if they can be auto-applied.
11360 2017-06-9 Ian Lance Taylor <iant@golang.org>
11362 * opts.c (finish_options): If -fsplit-stack, disable implicit
11363 -forder-blocks-and-partition.
11364 * doc/invoke.texi (Optimize Options): Document that when using
11365 -fsplit-stack -forder-blocks-and-partition is not implicitly
11368 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11370 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
11371 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
11372 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
11373 * builtins.def (abort, trap, unreachable): Declare cold.
11374 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
11375 * tree-core.h (ECF_COLD): New.
11376 * tree.c (set_call_expr_flags): Handle ECF_COLD.
11377 (build_common_builtin_nodes): Mark unreachable and abort as cold.
11379 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11381 * predict.c (unlikely_executed_stmt_p): Cleanup.
11383 2017-06-09 Richard Biener <rguenther@suse.de>
11385 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
11386 model if the ref is always written to.
11388 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11390 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
11392 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11394 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
11397 2017-06-09 Tom de Vries <tom@codesourcery.com>
11400 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
11401 "target cannot support label values" when encountering LABEL_REF.
11403 2017-06-09 Martin Liska <mliska@suse.cz>
11405 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
11406 (gimple_gen_ic_func_profiler): Emit direct comparison
11407 of __gcov_indirect_call_callee with NULL.
11408 (gimple_gen_time_profiler): Change probability from
11409 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
11411 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11413 * profile.c (edge_gcov_counts): Turn to pointer.
11414 (compute_branch_probabilities, compute_branch_probabilities): Update.
11415 (branch_prob): Do not clear edge_gcov_count.
11416 * profile.h (edge_gcov_counts): Turn to pointer.
11417 (edge_gcov_count): Update.
11419 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11421 * gimple.h (gimple_check_failed): Mark cold.
11423 2017-06-09 Richard Biener <rguenther@suse.de>
11425 PR tree-optimization/66623
11426 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
11427 refactor check_reduction into two parts, properly computing
11428 whether we have to check reduction validity for outer loop
11431 2017-06-09 Richard Biener <rguenther@suse.de>
11433 PR tree-optimization/79483
11434 * graphite-scop-detection.c (order): New global.
11435 (get_order): Compute bb to order mapping that satisfies code
11436 generation constraints.
11437 (cmp_pbbs): New helper.
11438 (build_scops): Start domwalk at entry block, sort generated
11441 2017-06-09 Richard Biener <rguenther@suse.de>
11443 PR middle-end/81007
11444 * ipa-polymorphic-call.c
11445 (ipa_polymorphic_call_context::restrict_to_inner_class):
11446 Skip FIELD_DECLs with error_mark_node type.
11447 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
11450 2017-06-09 Martin Liska <mliska@suse.cz>
11452 * predict.c (struct branch_predictor): New struct.
11453 (test_prediction_value_range): New test.
11454 (predict_c_tests): New function.
11455 * selftest-run-tests.c (selftest::run_tests): Run the function.
11456 * selftest.h: Declare new tests.
11458 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
11461 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
11462 gen_add3_insn did not fail.
11463 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
11464 r0, construct that number in a temporary reg and add that reg to r0.
11465 If asked to put the result in r0 as well, fail.
11467 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
11469 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11470 for early expansion of vec_eqv.
11472 2017-06-08 Jakub Jelinek <jakub@redhat.com>
11474 PR middle-end/81005
11475 * ubsan.c (instrument_null): Avoid pointless code temporary.
11476 (pass_ubsan::execute): Instrument aggregate arguments of calls.
11478 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
11482 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
11485 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
11488 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11490 * predict.c (unlikely_executed_edge_p): Move ahead.
11491 (probably_never_executed_edge_p): Use it.
11493 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11495 PR middle-end/79988
11496 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
11497 gimple_call_builtin_p call.
11499 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11501 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
11502 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
11503 rtl_check_failed_type2, rtl_check_failed_code1,
11504 rtl_check_failed_code2, rtl_check_failed_code_mode,
11505 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
11506 rtvec_check_failed_bounds, rtl_check_failed_flag,
11507 _fatal_insn_not_found, _fatal_insn): Likewise.
11508 * tree.h (tree_contains_struct_check_failed,
11509 tree_check_failed, tree_not_check_failed,
11510 tree_class_check_failed, tree_range_check_failed,
11511 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
11512 tree_vec_elt_check_failed, phi_node_elt_check_failed,
11513 tree_operand_check_failed, omp_clause_check_failed,
11514 omp_clause_operand_check_failed, omp_clause_range_check_failed):
11517 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11519 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
11520 flag_branch_probabilities.
11521 * ipa-inline.c (edge_badness): Likewise.
11522 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
11523 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
11524 * predict.c (maybe_hot_frequency_p): Likewise.
11525 (probably_never_executed): Likewise.
11526 * sched-ebb.c (schedule_ebbs): Likewise.
11527 * sched-rgn.c (find_single_block_region): Likewise.
11528 * tracer.c (tail_duplicate): Likewise.
11530 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11532 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
11533 longer requires x_flag_profile_use.
11535 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11537 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
11538 instead of flag_reorder_blocks_and_partition.
11539 * dbxout.c (dbxout_function_end): Likewise.
11540 * dwarf2out.c (gen_subprogram_die): Likewise.
11541 * haifa-sched.c (sched_create_recovery_edges): Likewise.
11542 * hw-doloop.c (reorg_loops): Likewise.
11543 * varasm.c (assemble_start_function,
11544 assemble_end_function): Likewise.
11545 (decide_function_section): Do not check for
11546 flag_reorder_blocks_and_partition.
11548 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11550 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11552 (chkp_get_hard_register_fake_addr_expr): Ditto.
11553 (chkp_build_addr_expr): Add check for hard reg case.
11554 (chkp_parse_array_and_component_ref): Ditto.
11555 (chkp_find_bounds_1): Ditto.
11556 (chkp_process_stmt): Don't generate bounds store for
11559 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11561 * predict.c (maybe_hot_bb_p): Do not check profile status.
11562 (maybe_hot_edge_p): Likewise.
11563 (probably_never_executed): Check for zero counts even if profile
11565 (unlikely_executed_edge_p): New function.
11566 (unlikely_executed_stmt_p): New function.
11567 (unlikely_executed_bb_p): New function.
11568 (set_even_probabilities): Use unlikely predicates.
11569 (combine_predictions_for_bb): Likewise.
11570 (predict_paths_for_bb): Likewise.
11571 (predict_paths_leading_to_edge): Likewise.
11572 (determine_unlikely_bbs): New function.
11573 (estimate_bb_frequencies): Use it.
11574 (compute_function_frequency): Use zero counts even if profile is
11576 * profile-count.h: Fix typo.
11578 2017-08-08 Julia Koval <julia.koval@intel.com>
11580 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
11581 _mm512_mask_cvtsepi16_storeu_epi8,
11582 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
11583 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
11584 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
11585 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
11586 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
11587 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
11588 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
11589 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
11590 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
11591 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
11592 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
11593 __builtin_ia32_pmovuswb256mem_mask,
11594 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
11595 __builtin_ia32_pmovwb512mem_mask): New builtins.
11597 2017-08-08 Julia Koval <julia.koval@intel.com>
11599 PR target/73350,80862
11600 * config/i386/subst.md (round): Fix round pattern.
11601 * config/i386/i386.c (ix86_erase_embedded_rounding):
11602 Fix erasing rounding for the fixed pattern.
11604 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11606 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
11608 2017-06-08 Martin Liska <mliska@suse.cz>
11610 PR gcov-profile/80911
11611 * gcov.c (block_info::block_info): New constructor.
11613 2017-06-07 Carl Love <cel@us.ibm.com>
11615 * config/rs6000/rs6000-c: The return type of the following
11616 built-in functions was implemented as int not long long. Fix sign
11617 of return value for the unsigned version of vec_mulo and vec_mule.
11618 vector unsigned long long vec_bperm (vector unsigned long long,
11619 vector unsigned char)
11620 vector signed long long vec_mule (vector signed int,
11622 vector unsigned long long vec_mule (vector unsigned int,
11623 vector unsigned int)
11624 vector signed long long vec_mulo (vector signed int,
11626 vector unsigned long long vec_mulo (vector unsigned int,
11627 vector unsigned int)
11628 * doc/extend.texi: Fix the documentation for the built-in
11631 2017-06-07 Carl Love <cel@us.ibm.com>
11634 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
11637 2017-06-07 Carl Love <cel@us.ibm.com>
11639 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
11640 support, Generate doublehv for signed int/float for BE case only.
11642 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
11644 * doc/invoke.texi (mcx16): Rewrite.
11646 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11648 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
11649 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
11650 *mov<mode>_softfloat, and an anonymous splitter): Use
11651 nonimmediate_operand instead of rs6000_nonimmediate_operand.
11653 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11655 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
11657 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11658 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
11659 (rs6000_debug_reg_global): Adjust.
11660 (rs6000_init_hard_regno_mode_ok): Adjust.
11661 (rs6000_dbx_register_number): Adjust.
11662 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
11663 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11664 Remove SPE_ACC and SPEFSCR.
11665 (REG_ALLOC_ORDER): Ditto.
11666 (FRAME_POINTER_REGNUM): Change to 111.
11667 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
11668 (REG_CLASS_NAMES): Ditto.
11669 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
11670 (REGISTER_NAMES): Ditto.
11671 (ADDITIONAL_REG_NAMES): Ditto.
11672 (rs6000_reg_names): Ditto.
11673 * config/rs6000/rs6000.md: Renumber some register number
11676 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11678 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
11680 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11681 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
11683 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
11684 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11685 Delete the SPE high registers.
11686 (REG_ALLOC_ORDER): Ditto.
11687 (enum reg_class): Remove SPE_HIGH_REGS.
11688 (REG_CLASS_NAMES): Ditto.
11689 (REG_CLASS_CONTENTS): Delete the SPE high registers.
11690 (REGISTER_NAMES): Ditto.
11691 (rs6000_reg_names): Ditto.
11692 * doc/tm.texi.in: Remove SPE as example.
11693 * doc/tm.texi: Regenerate.
11695 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11697 * config/rs6000/8540.md (ppc8540_brinc): Delete.
11698 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
11699 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
11700 * config/rs6000/rs6000.md (type): Remove "brinc".
11702 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11704 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
11705 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
11706 * config/rs6000/linuxspe.h: Delete file.
11707 * config/rs6000/rs6000.md: Don't include spe.md.
11708 * config/rs6000/spe.h: Delete file.
11709 * config/rs6000/spe.md: Delete file.
11710 * config/rs6000/t-rs6000: Remove spe.md.
11712 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11714 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
11715 (reg_or_none500mem_operand): Delete.
11716 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
11717 instead of reg_or_none500mem_operand.
11719 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11721 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11722 handling of SPE flags.
11723 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
11725 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11727 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
11729 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
11730 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
11731 paired_divv2sf3): Similar.
11732 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
11733 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
11734 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
11736 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
11737 Rename the paired_* instruction patterns.
11738 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
11740 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
11741 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
11742 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
11743 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
11744 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
11745 PAIRED_VECTOR_MODE.
11746 (struct machine_function): Delete field spe_insn_chain_scanned_p.
11747 (spe_func_has_64bit_regs_p): Delete.
11748 (spe_expand_predicate_builtin): Delete.
11749 (spe_expand_evsel_builtin): Delete.
11750 (TARGET_DWARF_REGISTER_SPAN): Do not define.
11751 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
11752 (invalid_e500_subreg): Delete.
11753 (rs6000_legitimize_address): Always force_reg op2 as well, for
11754 paired single memory accesses.
11755 (rs6000_member_type_forces_blk): Delete.
11756 (rs6000_spe_function_arg): Delete.
11757 (rs6000_expand_unop_builtin): Delete SPE handling.
11758 (rs6000_expand_binop_builtin): Ditto.
11759 (spe_expand_stv_builtin): Delete.
11760 (bdesc_2arg_spe): Delete.
11761 (spe_expand_builtin): Delete.
11762 (spe_expand_predicate_builtin): Delete.
11763 (spe_expand_evsel_builtin): Delete.
11764 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
11765 (spe_init_builtins): Delete.
11766 (spe_func_has_64bit_regs_p): Delete.
11767 (savres_routine_name): Delete "info" parameter. Adjust callers.
11768 (rs6000_emit_stack_reset): Ditto.
11769 (rs6000_dwarf_register_span): Delete.
11770 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
11771 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
11772 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
11774 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
11776 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
11777 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
11778 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
11779 mulv2sf3, divv2sf3): Delete expanders.
11781 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11783 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
11785 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11787 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
11788 * config/rs6000/rs6000.c: Ditto.
11790 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11792 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
11793 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
11794 comparison_operator.
11796 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11798 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
11799 * config/rs6000/rs6000.opt: Ditto.
11800 * config/rs6000/t-rtems: Ditto.
11802 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11804 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
11805 TARGET_E500_SINGLE by 0, simplify.
11806 * config/rs6000/rs6000.c: Ditto.
11807 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
11808 (spe_build_register_parallel): Delete.
11809 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
11810 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
11811 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
11812 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
11813 (E500_CONVERT): Delete.
11814 * config/rs6000/spe.md: Remove many patterns and all define_constants.
11816 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11818 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
11819 * config/rs6000/dfp.md: Ditto.
11820 (negdd2, *negdd2_fpr): Merge.
11821 (absdd2, *absdd2_fpr): Merge.
11822 (negtd2, *negtd2_fpr): Merge.
11823 (abstd2, *abstd2_fpr): Merge.
11824 * config/rs6000/e500.h: Delete file.
11825 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
11826 TARGET_FPRS by 1 and simplify.
11827 * config/rs6000/rs6000-c.c: Ditto.
11828 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
11829 TARGET_DF_SPE by 0.
11830 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
11832 * config/rs6000/rs6000.md: Ditto.
11833 (floatdidf2, *floatdidf2_fpr): Merge.
11834 (move_from_CR_gt_bit): Delete.
11835 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
11836 (E500_CR_IOR_COMPARE): Delete.
11837 (All patterns that require !TARGET_FPRS): Delete.
11838 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
11840 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11842 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
11844 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11846 * graphds.c (add_edge): Intitialize edge's attached data.
11847 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
11848 pointer parameter. Call pointed function on each edge during
11849 graph traversing. Skip traversing the edge when the function
11851 (graphds_dfs, graphds_scc): Ditto.
11852 (for_each_edge): New parameter. Pass the new parameter to callback
11854 * graphds.h (skip_edge_callback): New function pointer type.
11855 (graphds_dfs, graphds_scc): New function pointer parameter.
11856 (graphds_edge_callback, for_each_edge): New parameter.
11858 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11860 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
11861 out code checking if runtime alias check is possible to below ...
11862 Call the new function.
11863 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
11864 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
11866 2017-06-07 Marek Polacek <polacek@redhat.com>
11869 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
11870 TYPE_OVERFLOW_WRAPS check.
11872 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11874 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
11875 if versioning is required.
11876 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
11877 peeling with the check for versioning.
11879 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11881 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
11882 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
11883 Set true to new parameter if new ssa variable is defined.
11884 (vect_gen_vector_loop_niters): Refactor. Set range information
11885 for the new vector loop bound variable.
11886 (vect_do_peeling): Ditto.
11888 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11890 * tree-affine.c (ssa.h): Include header file.
11891 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
11892 has wrapping overflow behavior.
11894 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11896 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
11898 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11900 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
11901 (tree_to_aff_combination): ... here.
11903 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11905 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
11906 reg_pressure model function.
11907 (ivopts_global_cost_for_size): Delete.
11908 (determine_set_costs, iv_ca_recount_cost): Call new model function
11909 ivopts_estimate_reg_pressure.
11911 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11913 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
11914 expensive than udiv. Remove floating point cases from mod.
11916 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11918 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
11919 Increase idiv cost.
11921 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11923 * config/aarch64/aarch64.md
11924 (copysignsf3): Fix mask generation.
11926 2017-06-07 Jakub Jelinek <jakub@redhat.com>
11928 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
11930 (class dump_manager): Add register_dumps method.
11931 * dumpfile.c: Include langhooks.h.
11932 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
11933 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
11934 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
11935 (dump_manager::dump_register): Start with 512 entries instead of 32.
11936 (dump_manager::register_dumps): New method.
11937 * toplev.c (general_init): Instead of invoking register_dumps
11938 langhook, invoke register_dumps method on the dump manager.
11939 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
11942 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
11944 * doc/md.texi: Clarify the restrictions on a define_insn condition.
11945 Say that # requires an associated define_split to exist, and that
11946 the define_split must be suitable for use after register allocation.
11948 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11950 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
11951 (compute_outgoing_frequencies): Also initialize zero counts.
11952 (find_many_sub_basic_blocks): Do not produce uninitialized profile
11953 around loops; preserve more of profile when nothing changes.
11955 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
11957 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
11959 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
11960 * config/arm/arm-cpu-cdata.h: Regenerate.
11961 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
11962 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
11963 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
11964 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
11965 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
11967 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
11969 * config/arm/t-rmprofile: Likewise.
11970 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
11972 2017-06-06 David S. Miller <davem@davemloft.net>
11975 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
11976 blockage if function uses alloca.
11978 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11980 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
11981 New "uid" fields to hold pretty-print IDs of group and ref.
11982 Memory references are now identified as <group_id>:<ref_id>
11983 instead of using [random] addresses.
11984 (dump_mem_details): Simplify, no functional change.
11985 (dump_mem_ref): Simplify and make output more concise.
11986 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
11987 (find_or_create_group): Initialize group uid.
11988 (record_ref): Initialize ref uid. Improve debug output.
11989 (prune_group_by_reuse, should_issue_prefetch_p,)
11990 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
11991 (mark_nontemporal_store, determine_loop_nest_reuse):
11992 Improve debug output.
11994 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11996 * dbgcnt.def (prefetch): New debug counter.
11997 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
11998 (schedule_prefetches): Stop issueing prefetches if debug counter
12001 2017-06-06 Tom de Vries <tom@codesourcery.com>
12003 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12004 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12006 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12008 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12009 Use aarch64_reg_or_zero predicate for operand 4.
12010 (aarch64_compare_and_swap<mode> define_insn_and_split):
12011 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12012 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12014 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12016 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12017 (arm_compute_save_core_reg_mask): This.
12018 (thumb1_compute_save_reg_mask): Rename into ...
12019 (thumb1_compute_save_core_reg_mask): This.
12020 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12021 (arm_compute_frame_layout): Likewise.
12023 2017-06-06 Richard Biener <rguenther@suse.de>
12025 PR tree-optimization/80974
12026 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12027 keep or clear leaders SSA info.
12029 2017-06-06 Tom de Vries <tom@codesourcery.com>
12031 * config/nvptx/nvptx.c (split_mode_p): New function.
12032 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12034 2017-06-06 Tom de Vries <tom@codesourcery.com>
12036 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12038 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12041 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12044 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12046 * shrink-wrap.c (handle_simple_exit): Update profile.
12047 (try_shrink_wrapping): Upate profile.
12049 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12051 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12052 (tree_guess_outgoing_edge_probabilities): New.
12053 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12054 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12056 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12058 * ipa-split.c (split_function): Initialize return bb profile.
12060 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12062 * profile.c (compute_branch_probabilities): Also initialize
12063 EXIT_BLOCK profile.
12065 2017-06-06 Richard Biener <rguenther@suse.de>
12067 PR tree-optimization/80928
12068 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12069 (vect_analyze_loop_operations): Properly guard analysis for
12071 (vect_transform_loop): Likewise.
12072 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12073 (vect_model_induction_cost): Do not cost for pure SLP.
12074 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12075 of induction in inner loop vectorization.
12076 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12077 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12078 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12080 (vect_analyze_slp_cost_1): Cost induction.
12081 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12082 (vect_get_slp_vect_defs): Likewise.
12083 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12084 (vect_transform_stmt): Handle SLP reductions.
12085 * tree-vectorizer.h (vectorizable_induction): Adjust.
12087 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12089 * config/rs6000/rs6000.c (make_resolver_func): Update
12090 init_lowered_empty_function call.
12092 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12094 * doc/invoke.texi: Document the -fprofile-abs-path option.
12095 * common.opt (fprofile-abs-path): New option.
12096 * gcov-io.h (gcov_write_filename): Declare.
12097 * gcov-io.c (gcov_write_filename): New function.
12098 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12099 * profile.c (output_location): Likewise.
12101 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12103 * shring-wrap.c: Revert accidental commit.
12105 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
12107 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
12109 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12111 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
12113 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
12114 profile in callgraph edge.
12115 * profile-count.h (apply_probability): If THIS is 0, then result is 0
12116 (apply_scale): Likewise.
12117 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
12118 Also scale profile when inlining function with zero profile.
12119 (initialize_cfun): Update exit block profile even when it is zero.
12120 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
12121 when profile is read.
12123 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12125 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
12126 (CLONE_*): New constants to define the processors we can generate
12127 code for with the target_clone attribute.
12128 (rs6000_clone_map): New array to identify which clone processors
12129 the current program is running on.
12130 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
12131 target_clone attribute.
12132 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
12133 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
12134 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
12135 (cpu_expand_builtin): Add support for target_clone attribute.
12136 (rs6000_valid_attribute_p): Allow "default" attribute.
12137 (get_decl_name): New debug function to simplify printing the
12138 current function name in debugging statements.
12139 (rs6000_clone_priority): New functions to support the target_clone
12140 attribute, and be able to generate code to switch between ISA 2.05
12141 through ISA 3.0 (power6 through power9).
12142 (rs6000_compare_version_priority): Likewise.
12143 (rs6000_get_function_versions_dispatcher): Likewise.
12144 (make_resolver_func): Likewise.
12145 (add_condition_to_bb): Likewise.
12146 (dispatch_function_versions): Likewise.
12147 (rs6000_generate_version_dispatcher_body): Likewise.
12148 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
12149 (fusion_gpr_load_p): Fix a spacing issue.
12150 * doc/extend.texi (Common Function Attributes): Document that the
12151 PowerPC supports the target_clone attribute.
12153 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
12155 * config/arm/arm.h: explain F symbol found in description of ARM
12156 register allocation in its legend.
12158 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12160 * config/mips/frame-header-opt.c: Include profile-count.h.
12161 * config/riscv/riscv.c: Include profile-count.h
12163 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12165 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
12167 (sm_set_flag_if_changed): Add bbs field.
12168 (execute_sm_if_changed_flag_set): Pass BBS.
12169 (execute_sm): Update.
12171 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12173 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
12176 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12178 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
12179 (peephole2): New peephole2 to emit the above.
12180 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
12182 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12184 * config/aarch64/aarch64.c (define_peephole2 above
12185 *sub_<shift>_<mode>): New peephole.
12187 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12189 * config/i386/i386.c (make_resolver_func): Update.
12190 * Makefile.in: Add profile-count.h and profile-count.o
12191 * auto-profile.c (afdo_indirect_call): Update to new API.
12192 (afdo_set_bb_count): Update.
12193 (afdo_propagate_edge): Update.
12194 (afdo_propagate_circuit): Update.
12195 (afdo_calculate_branch_prob): Update.
12196 (afdo_annotate_cfg): Update.
12197 * basic-block.h: Include profile-count.h
12198 (struct edge_def): Turn count to profile_count.
12199 (struct basic_block_def): Likewie.
12200 (REG_BR_PROB_BASE): Move to profile-count.h
12201 (RDIV): Move to profile-count.h
12202 * bb-reorder.c (max_entry_count): Turn to profile_count.
12203 (find_traces): Update.
12204 (rotate_loop):Update.
12205 (connect_traces):Update.
12206 (sanitize_hot_paths):Update.
12207 * bt-load.c (migrate_btr_defs): Update.
12208 * cfg.c (RDIV): Remove.
12209 (init_flow): Use alloc_block.
12210 (alloc_block): Uninitialize count.
12211 (unchecked_make_edge): Uninitialize count.
12212 (check_bb_profile): Update.
12213 (dump_edge_info): Update.
12214 (dump_bb_info): Update.
12215 (update_bb_profile_for_threading): Update.
12216 (scale_bbs_frequencies_int): Update.
12217 (scale_bbs_frequencies_gcov_type): Update.
12218 (scale_bbs_frequencies_profile_count): New.
12219 * cfg.h (update_bb_profile_for_threading): Update.
12220 (scale_bbs_frequencies_profile_count): Declare.
12221 * cfgbuild.c (compute_outgoing_frequencies): Update.
12222 (find_many_sub_basic_blocks): Update.
12223 * cfgcleanup.c (try_forward_edges): Update.
12224 (try_crossjump_to_edge): Update.
12225 * cfgexpand.c (expand_gimple_tailcall): Update.
12226 (construct_exit_block): Update.
12227 * cfghooks.c (verify_flow_info): Update.
12228 (dump_bb_for_graph): Update.
12229 (split_edge): Update.
12230 (make_forwarder_block): Update.
12231 (duplicate_block): Update.
12232 (account_profile_record): Update.
12233 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
12234 (get_estimated_loop_iterations): Update.
12235 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
12236 (single_likely_exit): Update.
12237 * cfgloopmanip.c (scale_loop_profile): Update.
12239 (set_zero_probability): Update.
12240 (lv_adjust_loop_entry_edge): Update.
12241 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12242 (purge_dead_edges): Update.
12243 (rtl_account_profile_record): Update.
12244 * cgraph.c (cgraph_node::create): Uninitialize count.
12245 (symbol_table::create_edge): Uninitialize count.
12246 (cgraph_update_edges_for_call_stmt_node): Update.
12247 (cgraph_edge::dump_edge_flags): Update.
12248 (cgraph_node::dump): Update.
12249 (cgraph_edge::maybe_hot_p): Update.
12250 * cgraph.h: Include profile-count.h
12251 (create_clone), create_edge, create_indirect_edge): Update.
12252 (cgraph_node): Turn count to profile_count.
12253 (cgraph_edge0: Likewise.
12254 (make_speculative, clone): Update.
12255 (create_edge): Update.
12256 (init_lowered_empty_function): Update.
12257 * cgraphclones.c (cgraph_edge::clone): Update.
12258 (duplicate_thunk_for_node): Update.
12259 (cgraph_node::create_clone): Update.
12260 * cgraphunit.c (cgraph_node::analyze): Update.
12261 (cgraph_node::expand_thunk): Update.
12262 * final.c (dump_basic_block_info): Update.
12263 * gimple-streamer-in.c (input_bb): Update.
12264 * gimple-streamer-out.c (output_bb): Update.
12265 * graphite.c (print_global_statistics): Update.
12266 (print_graphite_scop_statistics): Update.
12267 * hsa-brig.c: Include basic-block.h.
12268 * hsa-dump.c: Include basic-block.h.
12269 * hsa-gen.c (T sum_slice): Update.
12270 (convert_switch_statements):Update.
12271 * hsa-regalloc.c: Include basic-block.h.
12272 * ipa-chkp.c (chkp_produce_thunks): Update.
12273 * ipa-cp.c (struct caller_statistics): Update.
12274 (init_caller_stats): Update.
12275 (gather_caller_stats): Update.
12276 (ipcp_cloning_candidate_p): Update.
12277 (good_cloning_opportunity_p): Update.
12278 (get_info_about_necessary_edges): Update.
12279 (dump_profile_updates): Update.
12280 (update_profiling_info): Update.
12281 (update_specialized_profile): Update.
12282 (perhaps_add_new_callers): Update.
12283 (decide_about_value): Update.
12284 (ipa_cp_c_finalize): Update.
12285 * ipa-devirt.c (struct odr_type_warn_count): Update.
12286 (struct decl_warn_count): Update.
12287 (struct final_warning_record): Update.
12288 (possible_polymorphic_call_targets): Update.
12289 (ipa_devirt): Update.
12290 * ipa-fnsummary.c (redirect_to_unreachable): Update.
12291 * ipa-icf.c (sem_function::merge): Update.
12292 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
12293 * ipa-inline.c (compute_uninlined_call_time): Update.
12294 (compute_inlined_call_time): Update.
12295 (want_inline_small_function_p): Update.
12296 (want_inline_self_recursive_call_p): Update.
12297 (edge_badness): Update.
12298 (lookup_recursive_calls): Update.
12299 (recursive_inlining): Update.
12300 (inline_small_functions): Update.
12301 (dump_overall_stats): Update.
12302 (dump_inline_stats): Update.
12303 * ipa-profile.c (ipa_profile_generate_summary): Update.
12304 (ipa_propagate_frequency): Update.
12305 (ipa_profile): Update.
12306 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12307 * ipa-utils.c (ipa_merge_profiles): Update.
12308 * loop-doloop.c (doloop_modify): Update.
12309 * loop-unroll.c (report_unroll): Update.
12310 (unroll_loop_runtime_iterations): Update.
12311 * lto-cgraph.c (lto_output_edge): Update.
12312 (lto_output_node): Update.
12313 (input_node): Update.
12314 (input_edge): Update.
12315 (merge_profile_summaries): Update.
12316 * lto-streamer-in.c (input_cfg): Update.
12317 * lto-streamer-out.c (output_cfg): Update.
12318 * mcf.c (create_fixup_graph): Update.
12319 (adjust_cfg_counts): Update.
12320 (sum_edge_counts): Update.
12321 * modulo-sched.c (sms_schedule): Update.
12322 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
12323 * predict.c (maybe_hot_count_p): Update.
12324 (probably_never_executed): Update.
12325 (dump_prediction): Update.
12326 (combine_predictions_for_bb): Update.
12327 (propagate_freq): Update.
12328 (handle_missing_profiles): Update.
12329 (counts_to_freqs): Update.
12330 (rebuild_frequencies): Update.
12331 (force_edge_cold): Update.
12332 * predict.h: Include profile-count.h
12333 (maybe_hot_count_p, counts_to_freqs): UPdate.
12334 * print-rtl-function.c: Do not include cfg.h
12335 * print-rtl.c: Include basic-block.h
12336 * profile-count.c: New file.
12337 * profile-count.h: New file.
12338 * profile.c (is_edge_inconsistent): Update.
12339 (correct_negative_edge_counts): Update.
12340 (is_inconsistent): Update.
12341 (set_bb_counts): Update.
12342 (read_profile_edge_counts): Update.
12343 (compute_frequency_overlap): Update.
12344 (compute_branch_probabilities): Update; Initialize and deinitialize
12346 (branch_prob): Update.
12347 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
12348 (edge_gcov_count): New.
12349 (bb_gcov_count): New.
12350 * shrink-wrap.c (try_shrink_wrapping): Update.
12351 * tracer.c (better_p): Update.
12352 * trans-mem.c (expand_transaction): Update.
12353 (ipa_tm_insert_irr_call): Update.
12354 (ipa_tm_insert_gettmclone_call): Update.
12355 * tree-call-cdce.c: Update.
12356 * tree-cfg.c (gimple_duplicate_sese_region): Update.
12357 (gimple_duplicate_sese_tail): Update.
12358 (gimple_account_profile_record): Update.
12359 (execute_fixup_cfg): Update.
12360 * tree-inline.c (copy_bb): Update.
12361 (copy_edges_for_bb): Update.
12362 (initialize_cfun): Update.
12363 (freqs_to_counts): Update.
12364 (copy_cfg_body): Update.
12365 (expand_call_inline): Update.
12366 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12367 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
12368 (try_unroll_loop_completely): Update.
12369 (try_peel_loop): Update.
12370 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12371 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
12372 * tree-ssa-loop-split.c (connect_loops): Update.
12373 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
12374 * tree-ssa-reassoc.c (branch_fixup): Update.
12375 * tree-ssa-tail-merge.c (replace_block_by): Update.
12376 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
12377 (compute_path_counts): Update.
12378 (update_profile): Update.
12379 (recompute_probabilities): Update.
12380 (update_joiner_offpath_counts): Update.
12381 (estimated_freqs_path): Update.
12382 (freqs_to_counts_path): Update.
12383 (clear_counts_path): Update.
12384 (ssa_fix_duplicate_block_edges): Update.
12385 (duplicate_thread_path): Update.
12386 * tree-switch-conversion.c (case_bit_test_cmp): Update.
12387 (struct switch_conv_info): Update.
12388 * tree-tailcall.c (decrease_profile): Update.
12389 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12390 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12391 * value-prof.c (check_counter): Update.
12392 (gimple_divmod_fixed_value): Update.
12393 (gimple_mod_pow2): Update.
12394 (gimple_mod_subtract): Update.
12395 (gimple_ic_transform): Update.
12396 (gimple_stringop_fixed_value): Update.
12397 * value-prof.h (gimple_ic): Update.
12399 2017-06-02 Carl Love <cel@us.ibm.com>
12401 * config/rs6000/rs6000-c: Add support for built-in functions
12402 vector double vec_doublee (vector signed int);
12403 vector double vec_doublee (vector unsigned int);
12404 vector double vec_doublee (vector float);
12405 vector double vec_doubleh (vector signed int);
12406 vector double vec_doubleh (vector unsigned int);
12407 vector double vec_doubleh (vector float);
12408 vector double vec_doublel (vector signed int);
12409 vector double vec_doublel (vector unsigned int);
12410 vector double vec_doublel (vector float);
12411 vector double vec_doubleo (vector signed int);
12412 vector double vec_doubleo (vector unsigned int);
12413 vector double vec_doubleo (vector float);.
12414 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
12415 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
12417 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
12418 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
12419 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
12421 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
12422 vec_doublel, vec_doubleh.
12423 * doc/extend.texi: Update the built-in documentation file for the
12424 new built-in functions.
12426 2017-06-02 David Malcolm <dmalcolm@redhat.com>
12429 * ipa-inline-analysis.c (free_growth_caches): Set
12430 edge_removal_hook_holder to NULL after removing it.
12432 2017-06-02 Sudakshina Das <sudi.das@arm.com>
12434 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
12435 comparision with zero.
12437 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12438 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12439 for early expansion of vec_min and vec_max builtins.
12440 (builtin_function_type): Add min/max unsigned variants to those
12441 identified as having unsigned arguments.
12443 2017-06-02 Olivier Hainque <hainque@adacore.com>
12445 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
12447 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12449 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
12450 Use VALL_F16 iterator rather than VALL.
12452 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12454 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
12455 Emit CBNZ inside loop when doing a strong exchange and comparing
12456 against zero. Generate the CC flags after the loop.
12458 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
12460 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
12461 (dl_section_ref): New.
12462 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
12463 On AIX, append an expression to subtract the size of the
12464 section length to dl_section_ref.
12466 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12468 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12469 for early expansion of vector absolute builtins.
12471 2017-06-02 Richard Biener <rguenther@suse.de>
12473 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
12474 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
12476 2017-06-02 Richard Biener <rguenther@suse.de>
12478 PR tree-optimization/80948
12479 * tree-tailcall.c (find_tail_calls): Track stmts to move in
12480 stmt order as well.
12482 2017-06-02 Richard Biener <rguenther@suse.de>
12484 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
12486 * tree-vect-stmts.c (process_use): Do not mark backedge defs
12487 for inductions as relevant.
12489 2017-06-02 Richard Biener <rguenther@suse.de>
12491 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
12492 (vectorizable_induction): ... this. Remove dead code.
12494 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
12496 * builtins. (expand_builtin_alloca): Remove second parameter and
12497 infer its value from the first parameter instead.
12498 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
12500 2017-06-02 Jakub Jelinek <jakub@redhat.com>
12502 PR rtl-optimization/80903
12503 * loop-doloop.c (add_test): Unshare sequence.
12505 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
12507 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
12509 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
12511 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
12513 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
12514 xlogue_layout::get_instance, logue_layout::xlogue_layout,
12515 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
12516 (xlogue_layout::get_stub_rtx): Make static.
12517 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
12518 (xlogue_layout::compute_stub_managed_regs): Rename to...
12519 (xlogue_layout::count_stub_managed_regs): ...this.
12520 (xlogue_layout::is_stub_managed_reg): New function.
12521 (xlogue_layout::m_stub_names): Rename to...
12522 (xlogue_layout::s_stub_names): ...this, make static.
12523 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
12524 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
12525 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
12526 xlogue_layout::s_stub_names): Instantiate statics.
12527 (stub_managed_regs): Remove.
12528 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
12529 (disable_call_ms2sysv_xlogues): Rename to...
12530 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
12531 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
12533 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
12534 change after reload_completed.
12535 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
12537 (ix86_expand_prologue): Likewise.
12538 (ix86_expand_epilogue): Likewise.
12539 (ix86_expand_split_stack_prologue): Likewise.
12540 (ix86_compute_frame_layout): Remove frame parameter ...
12541 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
12542 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
12544 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
12545 (ix86_frame): Move from here ...
12546 * config/i386/i386.h (ix86_frame): ... to here.
12547 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
12548 complete ix86_frame data structure instead. Remove some_ld_name.
12550 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12552 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
12553 symbols that hold a DECL_VALUE_EXPR.
12555 2017-06-01 Martin Jambor <mjambor@suse.cz>
12557 PR tree-optimization/80898
12558 * tree-sra.c (process_subtree_disqualification): Removed.
12559 (disqualify_candidate): Do not acll
12560 process_subtree_disqualification.
12561 (subtree_mark_written_and_enqueue): New function.
12562 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
12563 RHS has been disqualified and re-queue LHS if necessary. Apart
12564 from that, ignore disqualified RHS.
12566 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12568 * config/s390/s390.c (s390_emit_epilogue): Disable early return
12569 address fetch for z10 or later.
12571 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12573 * config/arc/arc.md (tst_movb): Add guard when splitting.
12575 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12577 * config/arc/arc.c (arc_can_eliminate): Test against
12578 arc_frame_pointer_needed.
12580 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12582 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
12583 to prevent store reordering.
12584 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
12585 (type): Add block type.
12586 (stack_tie): Define special instruction to be used in
12589 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12591 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
12592 constraint. It is not valid for the pattern.
12593 (noncommutative_binary_comparison): Likewise.
12595 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12597 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
12600 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12602 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
12603 be used by the reg-alloc.
12605 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12607 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
12608 reg-alloc when having mul64 or mul32x16 instructions.
12609 (mulsidi3): Likewise.
12610 (umulsidi3): Likewise.
12611 (mulsi32x16): New pattern.
12612 (mulsi64): Likewise.
12613 (mulsidi64): Likewise.
12614 (umulsidi64): Likewise.
12615 (MUL32x16_REG): Define.
12616 (mul64_600): Use MUL32x16_REG.
12617 (mac64_600): Likewise.
12618 (umul64_600): Likewise.
12619 (umac64_600): Likewise.
12621 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12623 * config/arc/arc.md (mulsi3_700): Make it commutative.
12625 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
12627 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
12629 (*sign_extendsidi2_insn): Likewise for movstosw.
12631 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12633 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
12634 the type of the input discriminant value. Convert the
12635 discriminant value of signedness vary.
12637 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
12639 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
12640 Add to -Wall section.
12642 2017-06-01 Richard Biener <rguenther@suse.de>
12644 PR middle-end/66313
12645 * fold-const.c (fold_plusminus_mult_expr): If the factored
12646 factor may be zero use a wrapping type for the inner operation.
12647 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
12648 and handle moved defs.
12649 (process_assignment): Properly guard the unary op case. Return a
12650 tri-state indicating that moving the stmt before the call may allow
12651 to continue. Pass through to_move.
12652 (find_tail_calls): Handle moving unrelated defs before
12655 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
12658 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
12659 splitter result in the canonical way.
12661 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
12663 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
12664 also for 32bit target. Update insn attributes.
12665 (zero-extendsidi2 splitter): Allow all registers for operand 1.
12667 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
12669 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
12670 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
12671 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
12672 (_mm_maskz_min_ss): New intrinsics.
12674 2017-05-31 Martin Liska <mliska@suse.cz>
12676 * tree-vect-loop.c (vect_create_epilog_for_reduction):
12677 Change comment style to one we normally use.
12678 (vectorizable_reduction): Likewise.
12679 (vectorizable_induction): Likewise.
12680 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12681 (vectorizable_call): Likewise.
12682 (vectorizable_simd_clone_call): Likewise.
12683 (vectorizable_conversion): Likewise.
12684 (vectorizable_assignment): Likewise.
12685 (vectorizable_shift): Likewise.
12686 (vectorizable_operation): Likewise.
12687 (vectorizable_store): Likewise.
12688 (vectorizable_load): Likewise.
12689 * tree-vectorizer.h: Likewise.
12691 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
12693 * passes.c (emergency_dump_function): New.
12694 * tree-pass.h (emergency_dump_function): Declare.
12695 * plugin.c (plugins_internal_error_function): Remove.
12696 * plugin.h (plugins_internal_error_function): Remove declaration.
12697 * toplev.c (internal_error_function): New static function. Use it...
12698 (general_init): ...here.
12700 2017-05-31 Graham Markall <graham.markall@embecosm.com>
12702 * config/arc/arc.c (arc_print_operand): Handle constant operands.
12703 (arc_rtx_costs): Add costs for new patterns.
12704 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
12705 * config/arc/predicates.md: Add _1_2_3_operand predicate.
12707 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12709 * tree-ssa-strlen.c (get_next_strinfo): New function.
12710 (get_stridx_plus_constant): Use it.
12711 (zero_length_string): Likewise.
12712 (adjust_related_strinfos): Likewise.
12713 (adjust_last_stmt): Likewise.
12715 2017-05-31 Richard Biener <rguenther@suse.de>
12718 * config/i386/i386.c (ix86_expand_builtin): Remove assert
12719 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
12721 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12723 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
12724 loop_vinfo argument and use of dependence distance vectors.
12725 Check instead whether the two references differ only in their
12726 initial value and assume that they have the same alignment if the
12727 difference is a multiple of the vector alignment.
12728 (vect_analyze_data_refs_alignment): Update call accordingly.
12730 2017-05-31 Martin Liska <mliska@suse.cz>
12733 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
12735 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12737 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
12738 (create_intersect_range_checks): Move from ...
12739 * tree-data-ref.c (create_intersect_range_checks_index)
12740 (create_intersect_range_checks): ... to here.
12741 (create_runtime_alias_checks): New function factored from ...
12742 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
12743 here. Call above function.
12744 * tree-data-ref.h (create_runtime_alias_checks): New function.
12746 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12748 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
12749 segment length for dr_b and compute it in wide_int.
12751 2017-05-31 Richard Biener <rguenther@suse.de>
12753 PR tree-optimization/80906
12754 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
12755 and pass through iv_map.
12756 (copy_bb_and_scalar_dependences): Adjust.
12757 (translate_pending_phi_nodes): Likewise.
12758 (copy_loop_close_phi_args): Handle code-generating IVs instead
12761 2017-05-30 David Malcolm <dmalcolm@redhat.com>
12763 * diagnostic-color.c (color_dict): Add "type-diff".
12764 (parse_gcc_colors): Update comment.
12765 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
12766 -fdiagnostics-show-template-tree and -fno-elide-type.
12767 (GCC_COLORS): Add type-diff to example.
12769 (-fdiagnostics-show-template-tree): New.
12770 (-fno-elide-type): New.
12771 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
12772 the pp_format_decoder callback. Call any m_format_postprocessor's
12774 (pretty_printer::pretty_printer): Initialize
12775 m_format_postprocessor.
12776 (pretty_printer::~pretty_printer): Delete any
12777 m_format_postprocessor.
12778 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
12779 (class format_postprocessor): New class.
12780 (struct pretty_printer::format_decoder): Document the new parameters.
12781 (struct pretty_printer::m_format_postprocessor): New field.
12782 * tree-diagnostic.c (default_tree_printer): Update for new
12783 bool and const char ** params.
12784 * tree-diagnostic.h (default_tree_printer): Likewise.
12786 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
12788 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
12789 (lwa_operand): Delete rs6000_gen_cell_microcode test.
12790 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12791 rs6000_gen_cell_microcode code.
12792 (rs6000_final_prescan_insn): Delete.
12793 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
12794 "warn-cell-microcode" entries.
12795 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
12796 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
12797 throughout. Change cc_reg_not_micro_cr0_operand to
12798 cc_reg_not_cr0_operand throughout.
12799 (*extendhi<mode>2_noload): Delete.
12800 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
12801 (mwarn-cell-microcode): Delete.
12802 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
12803 -mgen-cell-microcode and -mwarn-cell-microcode.
12805 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
12808 * config/i386/constraints.md (Yd): New constraint.
12810 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
12811 and (?Yd, r) alternatives. Update insn attributes.
12812 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
12813 and (?*Yd, r) alternatives. Update insn attributes.
12814 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
12816 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
12818 * gimplify.c (gimplify_modify_expr): Don't create a
12819 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
12822 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
12824 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
12826 2017-05-30 Richard Biener <rguenther@suse.de>
12828 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
12829 and reduc_def fields.
12830 (STMT_VINFO_REDUC_TYPE): New define.
12831 (STMT_VINFO_REDUC_DEF): Likewise.
12832 (vect_force_simple_reduction): Adjust prototype.
12833 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
12834 (vect_is_simple_reduction): Remove check_reduction argument.
12835 (vect_force_simple_reduction): Adjust and set
12836 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
12837 (vectorizable_reduction): Do not re-do reduction analysis
12838 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
12839 * tree-parloops.c (gather_scalar_reductions): Adjust.
12841 2017-05-30 Richard Biener <rguenther@suse.de>
12843 PR middle-end/80901
12844 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
12847 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
12849 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
12850 Introduce unknown_misalignment parameter and remove vf.
12851 (vect_peeling_hash_get_lowest_cost):
12852 Pass unknown_misalignment parameter.
12853 (vect_enhance_data_refs_alignment):
12854 Fix unsupportable data ref treatment.
12856 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12858 * tree-vect-data-refs.c (vect_get_data_access_cost):
12859 Workaround for SLP handling.
12860 (vect_enhance_data_refs_alignment):
12861 Compute costs for doing no peeling at all, compare to the best
12862 peeling costs so far and avoid peeling if cheaper.
12864 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12866 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
12867 Return peeling info and set costs to zero for unlimited cost
12869 (vect_enhance_data_refs_alignment): Also inspect all datarefs
12870 with unknown misalignment. Compute and costs for unknown
12871 misalignment, compare them to the costs for known misalignment
12872 and choose the cheapest for peeling.
12874 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12876 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
12877 (vect_get_peeling_costs_all_drs): Create function.
12878 (vect_peeling_hash_get_lowest_cost):
12879 Use vect_get_peeling_costs_all_drs.
12880 (vect_peeling_supportable): Create function.
12881 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
12883 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12885 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
12886 DR_HAS_NEGATIVE_STEP.
12887 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
12888 (vect_enhance_data_refs_alignment): Use.
12889 (vect_duplicate_ssa_name_ptr_info): Use.
12890 * tree-vectorizer.h (dr_misalignment): Use.
12891 (known_alignment_for_access_p): Use.
12893 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12896 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
12898 (has_section_name): New function.
12900 2017-05-30 Martin Liska <mliska@suse.cz>
12903 * auto-profile.c (get_function_decl_from_block): Fix
12906 2017-05-30 Richard Biener <rguenther@suse.de>
12908 PR middle-end/80876
12909 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
12911 2017-05-30 Martin Liska <mliska@suse.cz>
12913 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
12914 * dumpfile.h (struct dump_file_info): Remove ctors.
12916 2017-05-30 Martin Liska <mliska@suse.cz>
12918 * predict.def: Fix GNU coding style.
12920 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
12922 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
12923 Mark 'to' argument with ATTRIBUTE_UNUSED.
12925 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
12927 * config/xtensa/xtensa.c (xtensa_emit_call): Use
12928 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
12929 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
12932 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
12934 * doc/install.texi (Options specification): Restore entry of
12935 --enable-sjlj-exceptions.
12937 2017-05-27 Michael Eager <eager@eagercon.com>
12940 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
12942 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
12944 * config/microblaze/microblaze.h
12945 (FIXED_REGISTERS): Update in macro.
12946 (CALL_USED_REGISTERS): Update in macro.
12948 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
12950 * doc/install.texi: Add links to macOS binary distributions.
12952 2017-05-27 Jakub Jelinek <jakub@redhat.com>
12956 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12958 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
12960 2017-05-26 Martin Liska <mliska@suse.cz>
12962 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
12964 2017-05-26 Martin Liska <mliska@suse.cz>
12966 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
12967 always leading ';; '.
12968 (dump_bb_info): Likewise.
12969 (brief_dump_cfg): Likewise.
12970 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
12971 * dumpfile.c: Remove usage of TDF_VERBOSE.
12972 * dumpfile.h (enum dump_kind): Likewise.
12973 (dump_gimple_bb_header): Do not use TDF_COMMENT.
12974 * print-tree.c (debug_verbose): Remove.
12975 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
12976 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
12977 * tree-diagnostic.c (default_tree_printer): Replace
12978 TDF_DIAGNOSTIC with TDF_SLIM.
12980 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12982 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
12983 in parameter loop, rather than loop_vinfo.
12984 (create_intersect_range_checks): Ditto.
12985 (vect_create_cond_for_alias_checks): Update call to above functions.
12987 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12989 PR tree-optimization/80815
12990 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
12991 for merging runtime alias checks. Handle negative DR_STEPs.
12993 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12995 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
12997 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
12998 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
12999 out code pruning runtime alias checks.
13000 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13001 factored out from above.
13002 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13004 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13006 (prune_runtime_alias_test_list): New decalaration.
13008 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13010 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13011 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13012 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13013 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13014 (operator==, comp_dr_with_seg_len_pair): Ditto.
13015 (vect_prune_runtime_alias_test_list): Ditto.
13017 2017-05-26 Martin Liska <mliska@suse.cz>
13020 * params.def: Bound partial-inlining-entry-probability param.
13022 2017-05-26 Marek Polacek <polacek@redhat.com>
13025 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13028 2017-05-26 Richard Biener <rguenther@suse.de>
13030 PR tree-optimization/80842
13031 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13034 2017-05-26 Richard Biener <rguenther@suse.de>
13036 PR tree-optimization/80844
13037 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13039 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13041 * doc/md.texi (Machine Constraints): Update x86 family
13042 machine constraints section to match 'config/i386/constraints.md'.
13044 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13046 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13048 2017-05-25 Nathan Sidwell <nathan@acm.org>
13050 * doc/invoke.texi (--enable-languages): Update documentation.
13052 2017-05-25 Martin Liska <mliska@suse.cz>
13054 * dumpfile.c: Add TDF_FOLDING.
13055 * dumpfile.h (enum dump_kind): Likewise.
13056 * genmatch.c (dt_simplify::gen_1): Use it.
13058 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13060 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13062 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13064 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13065 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13067 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13069 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13070 * match.pd (X == C): Rewrite it here.
13071 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13072 with_certain_nonzero_bits2): New predicates.
13073 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13075 2017-05-24 Nathan Sidwell <nathan@acm.org>
13077 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13080 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13083 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13085 * config/powerpcspe: New port. Files are copied from the rs6000
13086 port, with "rs6000" in filenames replaced by "powerpcspe".
13088 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13090 PR rtl-optimization/80754
13091 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13093 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13095 * config/sparc/sparc.md (length): Return the correct value for -mflat
13096 sibcalls to match output_sibcall.
13098 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13102 * config/rs6000/rs6000.c (struct machine_function): Add new field
13104 (rs6000_get_separate_components): Init that field, use it.
13105 (rs6000_components_for_bb): Use the field.
13107 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
13109 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
13111 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
13113 PR middle-end/80823
13114 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
13116 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13119 * config/s390/s390.c (s390_check_qrst_address): Check incoming
13120 address against address_operand predicate.
13121 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
13123 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
13125 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
13126 parameters passed indirectly.
13128 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
13130 * config/i386/i386.md (*movdi_internal): Remove SSE4
13131 alternative 18 (?r, *v). Update insn attributes.
13132 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
13133 Update insn attributes.
13134 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
13135 Update insn attributes.
13136 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
13137 alternative 1 (r, v). Remove isa attribute.
13138 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
13139 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
13140 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13142 2017-05-23 Tom de Vries <tom@codesourcery.com>
13144 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
13147 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13149 * cgraphunit.c (symbol_table::process_new_functions): Update.
13150 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
13151 (inline_generate_summary): Rename to ...
13152 (ipa_fn_summary_generate): ... this one.
13153 (inline_read_summary): Rename to ...
13154 (ipa_fn_summary_read): ... this one.
13155 (inline_write_summary): Rename to ...
13156 (ipa_fn_summary_write): ... this one.
13157 (inline_free_summary): Rename to ...
13158 (ipa_free_fn_summary): ... this one.
13159 (pass_data_local_fn_summary, pass_local_fn_summary,
13160 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
13161 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
13162 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
13163 make_pass_ipa_fn_summary): New.
13164 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
13165 inline_write_summary, inline_free_summary): Remove.
13166 (ipa_free_fn_summary) : New.
13167 * ipa-inline.c (ipa_inline): Update.
13168 (pass_ipa_inline): Do not generate summaries.
13169 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
13171 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
13172 and add pass_ipa_fn_summary.
13173 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
13175 (make_pass_inline_parameters): Remove.
13177 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
13179 * omp-low.c (struct omp_context): Remove "default_kind" member.
13182 * omp-offload.c (execute_oacc_device_lower): Remove the
13183 parallelism dimensions function attributes for unparallelized
13184 OpenACC kernels constructs.
13186 2017-05-23 Martin Liska <mliska@suse.cz>
13188 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
13190 (cgraph_edge::make_speculative): Likewise.
13191 (cgraph_edge::resolve_speculation): Likewise.
13192 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13193 (cgraph_node::dump): Likewise.
13194 * cgraph.h: Likewise.
13195 * cgraphunit.c (analyze_functions): Likewise.
13196 (symbol_table::compile): Likewise.
13197 * ipa-cp.c (print_all_lattices): Likewise.
13198 (determine_versionability): Likewise.
13199 (initialize_node_lattices): Likewise.
13200 (ipcp_verify_propagated_values): Likewise.
13201 (estimate_local_effects): Likewise.
13202 (update_profiling_info): Likewise.
13203 (create_specialized_node): Likewise.
13204 (perhaps_add_new_callers): Likewise.
13205 (decide_about_value): Likewise.
13206 (decide_whether_version_node): Likewise.
13207 (identify_dead_nodes): Likewise.
13208 (ipcp_store_bits_results): Likewise.
13209 * ipa-devirt.c (dump_targets): Likewise.
13210 (ipa_devirt): Likewise.
13211 * ipa-icf.c (sem_item::dump): Likewise.
13212 (sem_function::equals): Likewise.
13213 (sem_variable::equals): Likewise.
13214 (sem_item_optimizer::read_section): Likewise.
13215 (sem_item_optimizer::execute): Likewise.
13216 (congruence_class::dump): Likewise.
13217 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
13218 (dump_inline_summary): Likewise.
13219 (estimate_node_size_and_time): Likewise.
13220 (inline_analyze_function): Likewise.
13221 * ipa-inline-transform.c (inline_call): Likewise.
13222 * ipa-inline.c (report_inline_failed_reason): Likewise.
13223 (want_early_inline_function_p): Likewise.
13224 (edge_badness): Likewise.
13225 (update_edge_key): Likewise.
13226 (inline_small_functions): Likewise.
13227 * ipa-profile.c (ipa_profile): Likewise.
13228 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
13229 (ipa_make_edge_direct_to_target): Likewise.
13230 (remove_described_reference): Likewise.
13231 (ipa_impossible_devirt_target): Likewise.
13232 (propagate_controlled_uses): Likewise.
13233 (ipa_print_node_params): Likewise.
13234 (ipcp_transform_function): Likewise.
13235 * ipa-pure-const.c (pure_const_read_summary): Likewise.
13236 (propagate_pure_const): Likewise.
13237 * ipa-reference.c (generate_summary): Likewise.
13238 (read_write_all_from_decl): Likewise.
13239 (propagate): Likewise.
13240 (ipa_reference_read_optimization_summary): Likewise.
13241 * ipa-utils.c (ipa_merge_profiles): Likewise.
13242 * ipa.c (walk_polymorphic_call_targets): Likewise.
13243 (symbol_table::remove_unreachable_nodes): Likewise.
13244 (ipa_single_use): Likewise.
13245 * passes.c (execute_todo): Likewise.
13246 * predict.c (drop_profile): Likewise.
13247 * symtab.c (symtab_node::get_dump_name): New function.
13248 (symtab_node::dump_name): Likewise.
13249 (symtab_node::dump_asm_name): Likewise.
13250 (symtab_node::dump_references): Likewise.
13251 (symtab_node::dump_referring): Likewise.
13252 (symtab_node::dump_base): Likewise.
13253 (symtab_node::debug_symtab): Likewise.
13254 * tree-sra.c (convert_callers_for_node): Likewise.
13255 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
13256 * value-prof.c (init_node_map): Likewise.
13258 2017-05-23 Martin Liska <mliska@suse.cz>
13260 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
13261 and symtab_node::debug_symtab to symbol_table::debug.
13262 * cgraphunit.c (analyze_functions): Use the renamed function.
13263 (symbol_table::compile): Likewise.
13264 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
13265 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
13266 * passes.c (execute_todo): Likewise.
13267 * symtab.c (symbol_table::dump): New function.
13268 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
13270 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13272 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
13273 that nonconst implies exec.
13275 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13277 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13278 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13279 (inline_edge_summary_vec): Turn into ...
13280 (ipa_call_summaries): ... this one.
13281 (redirect_to_unreachable, edge_set_predicate,
13282 evaluate_properties_for_edge, inline_summary_alloc,
13283 reset_ipa_call_summary, reset_inline_summary,
13284 inline_summary_t::duplicate): Update.
13285 (inline_edge_duplication_hook): Turn to ...
13286 (ipa_call_summary_t::duplicate): ... this one.
13287 (inline_edge_removal_hook): Turn to ...
13288 (ipa_call_summary_t::remove): ... this one.
13289 (dump_inline_edge_summary): Turn to ...
13290 (dump_ipa_call_summary): ... this one.
13291 (estimate_function_body_sizes): Update.
13292 (inline_update_callee_summaries): Update.
13293 (remap_edge_change_prob): Update.
13294 (remap_edge_summaries): Update.
13295 (inline_merge_summary): Update.
13296 (do_estimate_edge_time): Update.
13297 (inline_generate_summary): Update.
13298 (inline_read_section): Update.
13299 (inline_read_summary): Update.
13300 (inline_free_summary): Update.
13301 * ipa-inline.c (can_inline_edge_p): Update.
13302 (compute_inlined_call_time): Update.
13303 (want_inline_small_function_p): Update.
13304 (edge_badness): Update.
13305 (early_inliner): Update.
13306 * ipa-inline.h (inline_edge_summary): Turn to ...
13307 (ipa_call_summary): ... this one.
13308 (ipa_call_summary_t): New class.
13309 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13310 (ipa_call_summaries): New.
13311 (inline_edge_summary): Remove.
13312 (estimate_edge_growth): Update.
13313 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13314 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13315 * ipa-split.c (execute_split_functions): Update.
13316 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13318 2017-05-23 Tom de Vries <tom@codesourcery.com>
13320 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13321 attributes): Document rdrand effective target.
13323 2017-05-23 Tom de Vries <tom@codesourcery.com>
13325 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13326 attributes): Sort alphabetically.
13328 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
13330 * config/avr/genmultilib.awk: Use gsub instead of gensub.
13332 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
13335 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
13336 V2DF/V2DI splat into two separate patterns, one that handles
13337 registers, and the other that only handles memory. Drop support
13338 for splatting from a GPR on ISA 2.07 and then splitting the
13339 splat into direct move and splat.
13340 (vsx_splat_<mode>_reg): Likewise.
13341 (vsx_splat_<mode>_mem): Likewise.
13343 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
13345 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
13347 2017-05-22 Jakub Jelinek <jakub@redhat.com>
13349 PR middle-end/80809
13350 * omp-low.c (finish_taskreg_remap): New function.
13351 (finish_taskreg_scan): If unit size of ctx->record_type
13352 is non-constant, unshare the size expression and replace
13353 decls in it with possible outer var refs.
13355 PR middle-end/80809
13356 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
13357 GOVD_SHARED rather than GOVD_PRIVATE with it.
13358 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
13359 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
13361 PR middle-end/80853
13362 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
13363 as last argument to build_outer_var_ref for pointer bases of array
13364 section reductions.
13366 2017-05-19 Martin Sebor <msebor@redhat.com>
13368 * print-tree.c (print_node): Print DECL_READ_P flag.
13370 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13372 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
13373 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
13374 * cgraph.c: Likewise.
13375 * cgraphunit.c: Likewise.
13376 * gengtype.c: Likewise.
13377 * ipa-cp.c: Likewise.
13378 * ipa-devirt.c: Likewise.
13379 * ipa-icf.c: Likewise.
13380 * ipa-predicate.c: Likewise.
13381 * ipa-profile.c: Likewise.
13382 * ipa-prop.c: Likewise.
13383 * ipa-split.c: Likewise.
13385 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
13386 edge_predicate_pool, dump_inline_hints,
13387 inline_summary::account_size_time, redirect_to_unreachable,
13388 edge_set_predicate, set_hint_predicate,
13389 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
13390 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
13391 inline_summary_t::remove, remap_hint_predicate_after_duplication,
13392 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
13393 ipa_call_summary_t::remove, initialize_growth_caches,
13394 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
13395 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
13396 mark_modified, unmodified_parm_1, unmodified_parm,
13397 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
13398 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
13399 compute_bb_predicates, will_be_nonconstant_expr_predicate,
13400 will_be_nonconstant_predicate, record_modified_bb_info,
13401 get_minimal_bb, record_modified, param_change_prob,
13402 phi_result_unknown_predicate, predicate_for_phi_result,
13403 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
13404 estimate_function_body_sizes, compute_inline_parameters,
13405 compute_inline_parameters_for_curren, pass_data_inline_parameters,
13406 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
13407 inline_update_callee_summaries, remap_edge_change_prob,
13408 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
13409 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
13410 inline_analyze_function, inline_summary_t::insert,
13411 inline_generate_summary, read_ipa_call_summary, inline_read_section,
13412 inline_read_summary, write_ipa_call_summary, inline_write_summary,
13413 inline_free_summary): Move to ipa-fnsummary.h
13414 (predicate_t): Remove.
13415 * ipa-fnsummary.c: New file.
13416 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
13417 (enum inline_hints_vals, inline_hints, agg_position_info,
13418 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
13419 inline_summaries, ipa_call_summary, ipa_call_summary_t,
13420 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
13421 dump_inline_summary, dump_inline_hints, inline_generate_summary,
13422 inline_read_summary, inline_write_summary, inline_free_summary,
13423 inline_analyze_function, initialize_inline_failed,
13424 inline_merge_summary, inline_update_overall_summary,
13425 compute_inline_parameters): Move to ipa-fnsummary.h
13426 * ipa-fnsummary.h: New file.
13427 * ipa-inline-transform.h: Include ipa-inline.h.
13428 * ipa-inline.c: LIkewise.
13430 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13432 * ipa-inline.c (edge_badness): Use inlined_time instead of
13433 inline_summaries->get.
13435 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13437 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
13439 2017-05-22 Nathan Sidwell <nathan@acm.org>
13441 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
13442 (fdump-lang): Document 'raw' option.
13443 * dumpfile.h (TDI_tu): Delete.
13444 * dumpfile.c (dump_files): Remove translation-unit.
13445 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
13447 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
13449 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
13450 command option from $(AWK) call.
13451 * config/avr/genmultilib.awk: Simplify and rewrite so that it
13452 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
13453 [FORMAT]: Remove handling of variable.
13454 * config/avr/t-multilib: Regenerate.
13456 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13458 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
13460 (dump_inline_summary): Do not print self_time.
13461 (estimate_function_body_sizes): Do not set self_time.
13462 (compute_inline_parameters): Likewise.
13463 (inline_read_section, inline_write_summary): Do not stream self_time.
13464 * ipa-inline.h (inline_summary): Drop self_time.
13466 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13468 * ipa-inline-analysis.c (account_size_time): Rename to ...
13469 (inline_summary::account_size_time): ... this one.
13470 (reset_ipa_call_summary): Turn to ...
13471 (ipa_call_summary::reset): ... this one.
13472 (reset_inline_summary): Turn to ...
13473 (inline_summary::reset): ... this one.
13474 (inline_summary_t::remove): Update.
13475 (inline_summary_t::duplicate): Update.
13476 (ipa_call_summary_t::remove): Update.
13477 (dump_inline_summary): Update.
13478 (estimate_function_body_sizes): Update.
13479 (compute_inline_parameters): Update.
13480 (estimate_node_size_and_time): Update.
13481 (inline_merge_summary): Update.
13482 (inline_update_overall_summary): Update.
13483 (inline_read_section): Update.
13484 (inline_write_summary): Update.
13485 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
13486 add account_size_time and reset member functions.
13487 (ipa_call_summary): Add reset function.
13488 * ipa-predicate.h (predicate::operator &): Constify.
13490 2017-05-22 Richard Biener <rguenther@suse.de>
13492 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
13494 2017-05-19 Jason Merrill <jason@redhat.com>
13496 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
13498 2017-05-19 Marek Polacek <polacek@redhat.com>
13501 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
13502 TYPE_OVERFLOW_WRAPS checks.
13504 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
13506 * tree-core.h (enum omp_clause_default_kind): Add
13507 "OMP_CLAUSE_DEFAULT_PRESENT".
13508 * tree-pretty-print.c (dump_omp_clause): Handle it.
13509 * gimplify.c (enum gimplify_omp_var_data): Add
13510 "GOVD_MAP_FORCE_PRESENT".
13511 (gimplify_adjust_omp_clauses_1): Map it to
13512 "GOMP_MAP_FORCE_PRESENT".
13513 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
13515 * gimplify.c (oacc_default_clause): Clarify.
13517 2017-05-19 Nathan Sidwell <nathan@acm.org>
13519 LANG_HOOK_REGISTER_DUMPS
13520 * toplev.c (general_init): Call register dump lang hook.
13521 * doc/invoke.texi: Document -fdump-lang option family.
13522 * dumpfile.c (dump_files): Remove class dump here.
13523 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
13524 * dumpfile.h (tree_dump_index): Remove TDI_class.
13525 * langhooks-def.h (lhd_register_dumps): Declare.
13526 (LANG_HOOKS_REGISTER_DUMPS): Define.
13527 (LANG_HOOKS_INITIALIZER): Add it.
13528 * langhooks.c (lhd_register_dumps): Define.
13529 * langhooks.h (struct lang_hooks): Add register_dumps.
13531 2017-05-19 Nathan Sidwell <nathan@acm.org>
13533 * context.h (context::set_passes): New.
13534 * context.c (context::context): Do not create pass manager.
13535 * toplev.c (general_init): Create pass manager here.
13537 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
13539 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
13540 use this splitter if two add or or instructions would also work for
13541 the constant we want to generate.
13543 2017-05-19 Richard Biener <rguenther@suse.de>
13546 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
13547 predicate evaluation.
13549 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
13551 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
13553 * ipa-inline.c (want_inline_small_function_p): Do not cast to
13556 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
13558 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13559 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13560 (inline_edge_summary_vec): Turn into ...
13561 (ipa_call_summaries): ... this one.
13562 (redirect_to_unreachable, edge_set_predicate,
13563 evaluate_properties_for_edge, inline_summary_alloc,
13564 reset_ipa_call_summary, reset_inline_summary,
13565 inline_summary_t::duplicate): Update.
13566 (inline_edge_duplication_hook): Turn to ...
13567 (ipa_call_summary_t::duplicate): ... this one.
13568 (inline_edge_removal_hook): Turn to ...
13569 (ipa_call_summary_t::remove): ... this one.
13570 (dump_inline_edge_summary): Turn to ...
13571 (dump_ipa_call_summary): ... this one.
13572 (estimate_function_body_sizes): Update.
13573 (inline_update_callee_summaries): Update.
13574 (remap_edge_change_prob): Update.
13575 (remap_edge_summaries): Update.
13576 (inline_merge_summary): Update.
13577 (do_estimate_edge_time): Update.
13578 (inline_generate_summary): Update.
13579 (inline_read_section): Update.
13580 (inline_read_summary): Update.
13581 (inline_free_summary): Update.
13582 * ipa-inline.c (can_inline_edge_p): Update.
13583 (compute_inlined_call_time): Update.
13584 (want_inline_small_function_p): Update.
13585 (edge_badness): Update.
13586 (early_inliner): Update.
13587 * ipa-inline.h (inline_edge_summary): Turn to ...
13588 (ipa_call_summary): ... this one.
13589 (ipa_call_summary_t): New class.
13590 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13591 (ipa_call_summaries): New.
13592 (inline_edge_summary): Remove.
13593 (estimate_edge_growth): Update.
13594 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13595 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13596 * ipa-split.c (execute_split_functions): Update.
13597 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13599 2017-05-19 Richard Biener <rguenther@suse.de>
13601 PR middle-end/80764
13602 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
13604 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
13606 * config/rs6000/rs6000.c (struct machine_function): Add field
13607 fpr_is_wrapped_separately.
13608 (rs6000_get_separate_components): Use 64 components. Handle the
13609 new FPR components.
13610 (rs6000_components_for_bb): Handle the FPR components.
13611 (rs6000_emit_prologue_components): Handle the FPR components.
13612 (rs6000_emit_epilogue_components): Handle the FPR components.
13613 (rs6000_set_handled_components): Handle the FPR components.
13614 (rs6000_emit_prologue): Don't output prologue code for those FPRs
13615 that are already separately shrink-wrapped.
13616 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
13617 that are already separately shrink-wrapped.
13619 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
13622 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
13625 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
13626 (define_peephole2 for Altivec d-form load): Add peepholes to catch
13627 cases where the register allocator uses a move and an offsettable
13628 memory operation to/from a FPR register on ISA 2.06/2.07.
13629 (define_peephole2 for Altivec d-form store): Likewise.
13631 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
13634 * config/i386/mmx.md (*mov<mode>_internal): Enable
13635 alternatives 11, 12, 13 and 14 also for 32bit targets.
13636 Remove alternatives 15, 16, 17 and 18.
13637 * config/i386/sse.md (vec_concatv2di): Change
13638 alternative (!x, *y) to (x, ?!*Yn).
13640 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
13642 * dumpfile.h (enum dump_kind): Remove stray comma.
13644 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13646 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
13647 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
13648 predicate::num_conditions
13649 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
13650 (CHANGED): turn into predicate::changed.
13651 (agg_position_info): Move to ipa-predicate.h
13652 (add_condition, predicate::add_clause, predicate::operator &=,
13653 predicate::or_with, predicate::evaluate, predicate::probability,
13654 dump_condition, dump_clause, predicate::dump,
13655 predicate::remap_after_duplication, predicate::remap_after_inlining,
13656 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
13657 (evaluate_conditions_for_known_args): Update.
13658 (set_cond_stmt_execution_predicate): Update.
13659 * ipa-inline.h: Include ipa-predicate.h
13660 (condition, inline_param_summary, conditions, agg_position_info,
13661 predicate): Move to ipa-predicate.h
13662 * ipa-predicate.c: New file.
13663 * ipa-predicate.h: New file.
13665 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
13667 * final.c (leaf_function_p): Check we are not in a sequence.
13669 2017-05-18 Martin Liska <mliska@suse.cz>
13671 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
13672 * dumpfile.c (dump_register): Use new enum dump_kind.
13673 (get_dump_file_name): Likewise.
13674 (dump_enable_all): Likewise.
13675 (dump_switch_p_1): Likewise.
13676 (enable_rtl_dump_file): Remove usage of TDF_RTL.
13677 * dumpfile.h (enum dump_kind): New enum type.
13678 (struct dump_file_info): Create constructor and
13679 format fields and comments.
13680 * passes.c (pass_manager::register_one_dump_file):
13682 * statistics.c (statistics_early_init): Likewise.
13683 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
13684 TDF_TREE with TDF_SLIM.
13685 (gather_memory_references_ref): Likewise.
13687 2017-05-18 Martin Liska <mliska@suse.cz>
13689 * vec.h (struct vnull): Use it.
13691 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13693 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
13694 (true_predicate, false_predicate, true_predicate_p,
13695 false_predicate_p): Remove.
13696 (single_cond_predicate, not_inlined_predicate): Turn to member function
13698 (add_condition): Update.
13699 (add_clause): Turn to...
13700 (predicate::add_clause): ... this one; update; allow passing NULL
13702 (and_predicates): Turn to ...
13703 (predicate::operator &=): ... this one.
13704 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
13705 (or_predicates): Turn to ...
13706 (predicate::or_with): ... this one.
13707 (evaluate_predicate): Turn to ...
13708 (predicate::evaluate): ... this one.
13709 (predicate_probability): Turn to ...
13710 (predicate::probability): ... this one.
13711 (dump_condition): Update.
13712 (dump_predicate): Turn to ...
13713 (predicate::dump): ... this one.
13714 (account_size_time): Update.
13715 (edge_set_predicate): Update.
13716 (set_hint_predicate): UPdate.
13717 (evaluate_conditions_for_known_args): Update.
13718 (evaluate_properties_for_edge): Update.
13719 (remap_predicate_after_duplication): Turn to...
13720 (predicate::remap_after_duplication): ... this one.
13721 (remap_hint_predicate_after_duplication): Update.
13722 (inline_summary_t::duplicate): UPdate.
13723 (dump_inline_edge_summary): Update.
13724 (dump_inline_summary): Update.
13725 (set_cond_stmt_execution_predicate): Update.
13726 (set_switch_stmt_execution_predicate): Update.
13727 (compute_bb_predicates): Update.
13728 (will_be_nonconstant_expr_predicate): Update.
13729 (will_be_nonconstant_predicate): Update.
13730 (phi_result_unknown_predicate): Update.
13731 (predicate_for_phi_result): Update.
13732 (array_index_predicate): Update.
13733 (estimate_function_body_sizes): Update.
13734 (estimate_node_size_and_time): Update.
13735 (estimate_ipcp_clone_size_and_time): Update.
13736 (remap_predicate): Rename to ...
13737 (predicate::remap_after_inlining): ... this one.
13738 (remap_hint_predicate): Update.
13739 (inline_merge_summary): Update.
13740 (inline_update_overall_summary): Update.
13741 (estimate_size_after_inlining): Update.
13742 (read_predicate): Rename to ...
13743 (predicate::stream_in): ... this one.
13744 (read_inline_edge_summary): Update.
13745 (write_predicate): Rename to ...
13746 (predicate::stream_out): ... this one.
13747 (write_inline_edge_summary): Update.
13748 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
13749 (clause_t): Turn to uint32_t
13750 (predicate): Turn to class; implement constructor and operators
13752 (size_time_entry): Update.
13753 (inline_summary): Update.
13754 (inline_edge_summary): Update.
13756 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
13758 * fold-const.c (fold_binary_loc): Move transformation...
13759 * match.pd (C - X CMP X): ... here.
13761 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
13763 * config/sparc/sparc.c (sparc_option_override): Set function
13764 alignment for -mcpu=niagara7 to 64 to match the I$ line.
13765 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
13767 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
13769 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
13771 2017-05-18 Marek Polacek <polacek@redhat.com>
13774 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
13775 (pass_ubsan::execute): Call gimple_assign_single_p instead of
13776 gimple_assign_load_p.
13778 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
13780 PR middle-end/80692
13781 * real.c (do_compare): Give decimal_do_compare preference over
13782 comparing just the signs.
13784 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
13786 * doc/md.texi (Canonicalization of Instructions): Describe the
13787 canonical form of instructions that inherently set a condition
13790 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
13792 PR middle-end/80775
13793 * tree-cfg.c: Move deletion of unreachable case statements to after
13794 the merging of consecutive case labels.
13796 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
13798 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
13799 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
13800 restoring of callee-saved registers.
13802 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
13804 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
13805 * config/visium/visium.c (single_set_and_flags): Likewise.
13806 * config/visium/visium.md (Substitutions): Likewise.
13808 2017-05-17 Martin Liska <mliska@suse.cz>
13810 * cfg.c: Introduce dump_flags_t type and
13811 use it instead of int type.
13813 * cfghooks.c: Likewise.
13814 * cfghooks.h (struct cfg_hooks): Likewise.
13815 * cfgrtl.c: Likewise.
13816 * cfgrtl.h: Likewise.
13817 * cgraph.c (cgraph_node::get_body): Likewise.
13818 * coretypes.h: Likewise.
13819 * domwalk.c: Likewise.
13820 * domwalk.h: Likewise.
13821 * dumpfile.c (struct dump_option_value_info): Likewise.
13822 (dump_enable_all): Likewise.
13823 (dump_switch_p_1): Likewise.
13824 (opt_info_switch_p): Likewise.
13825 * dumpfile.h (enum tree_dump_index): Likewise.
13826 (struct dump_file_info): Likewise.
13827 * genemit.c: Likewise.
13828 * generic-match-head.c: Likewise.
13829 * gengtype.c (open_base_files): Likewise.
13830 * gimple-pretty-print.c: Likewise.
13831 * gimple-pretty-print.h: Likewise.
13832 * graph.c (print_graph_cfg): Likewise.
13833 * graphite-scop-detection.c (dot_all_sese): Likewise.
13834 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
13835 * loop-unroll.c (report_unroll): Likewise.
13836 * passes.c (pass_manager::register_one_dump_file): Likewise.
13837 * print-tree.c: Likewise.
13838 * statistics.c: Likewise.
13839 * tree-cfg.c: Likewise.
13840 * tree-cfg.h: Likewise.
13841 * tree-dfa.c: Likewise.
13842 * tree-dfa.h: Likewise.
13843 * tree-dump.c (dump_function): Likewise.
13844 * tree-dump.h (struct dump_info): Likewise.
13845 * tree-pretty-print.c: Likewise.
13846 * tree-pretty-print.h: Likewise.
13847 * tree-ssa-live.c: Likewise.
13848 * tree-ssa-live.h: Likewise.
13849 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
13850 * tree-vect-loop.c: Likewise.
13851 * tree-vect-slp.c: Likewise.
13853 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
13854 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13856 PR tree-optimization/80457
13857 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
13858 of all arguments to a statement as scalar_to_vec operations.
13859 (vectorizable_call): Adjust call to vect_model_simple_cost for
13861 (vectorizable_conversion): Likewise.
13862 (vectorizable_assignment): Likewise.
13863 (vectorizable_shift): Likewise.
13864 (vectorizable_operation): Likewise.
13865 (vectorizable_comparison): Likewise.
13866 (vect_is_simple_cond): Record the def types for operands.
13867 (vectorizable_condition): Likewise, call vect_model_simple_cost.
13868 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
13869 for statement argument count.
13871 2017-05-16 Carl Love <cel@us.ibm.com>
13873 * config/rs6000/rs6000-c: Add support for built-in functions
13874 vector unsigned long long vec_bperm (vector unsigned long long,
13875 vector unsigned char)
13876 vector signed long long vec_mule (vector signed int,
13878 vector unsigned long long vec_mule (vector unsigned int,
13879 vector unsigned int)
13880 vector signed long long vec_mulo (vector signed int,
13882 vector unsigned long long vec_mulo (vector unsigned int,
13883 vector unsigned int)
13884 vector signed char vec_sldw (vector signed char,
13885 vector signed char,
13887 vector unsigned char vec_sldw (vector unsigned char,
13888 vector unsigned char,
13890 vector signed short vec_sldw (vector signed short,
13891 vector signed short,
13893 vector unsigned short vec_sldw (vector unsigned short,
13894 vector unsigned short,
13896 vector signed int vec_sldw (vector signed int,
13899 vector unsigned int vec_sldw (vector unsigned int,
13900 vector unsigned int,
13902 vector signed long long vec_sldw (vector signed long long,
13903 vector signed long long,
13905 vector unsigned long long vec_sldw (vector unsigned long long,
13906 vector unsigned long long,
13908 * config/rs6000/rs6000-c: Add support for built-in functions
13909 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
13910 * config/rs6000/altivec.h: Add defintion for vec_sldw.
13911 * doc/extend.texi: Update the built-in documentation for the
13912 new built-in functions.
13914 2017-05-16 Marek Polacek <polacek@redhat.com>
13918 * tree.c (save_expr): Don't fold the expression.
13920 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
13922 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
13923 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
13924 and (?*y,m). Update insn attributes.
13926 2017-05-16 Martin Liska <mliska@suse.cz>
13928 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
13929 flags argument of print_gimple_stmt, print_gimple_expr,
13930 print_generic_stmt and print_generic_expr.
13931 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
13932 * coretypes.h: Likewise.
13933 * except.c (dump_eh_tree): Likewise.
13934 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13935 * gimple-pretty-print.h: Likewise.
13936 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
13937 (backprop::push_to_worklist): Likewise.
13938 (backprop::pop_from_worklist): Likewise.
13939 (backprop::process_use): Likewise.
13940 (backprop::intersect_uses): Likewise.
13941 (note_replacement): Likewise.
13942 * gimple-ssa-store-merging.c
13943 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
13944 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
13945 (pass_store_merging::execute): Likewise.
13946 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
13947 (ssa_base_cand_dump_callback): Likewise.
13948 (dump_incr_vec): Likewise.
13949 (replace_refs): Likewise.
13950 (replace_mult_candidate): Likewise.
13951 (create_add_on_incoming_edge): Likewise.
13952 (create_phi_basis): Likewise.
13953 (insert_initializers): Likewise.
13954 (all_phi_incrs_profitable): Likewise.
13955 (introduce_cast_before_cand): Likewise.
13956 (replace_one_candidate): Likewise.
13957 * gimplify.c (gimplify_expr): Likewise.
13958 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
13959 (set_rename): Likewise.
13960 (rename_uses): Likewise.
13961 (copy_loop_phi_nodes): Likewise.
13962 (add_close_phis_to_merge_points): Likewise.
13963 (copy_loop_close_phi_args): Likewise.
13964 (copy_cond_phi_args): Likewise.
13965 (graphite_copy_stmts_from_block): Likewise.
13966 (translate_pending_phi_nodes): Likewise.
13967 * graphite-poly.c (print_pdr): Likewise.
13968 (dump_gbb_cases): Likewise.
13969 (dump_gbb_conditions): Likewise.
13970 (print_scop_params): Likewise.
13971 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
13972 (build_cross_bb_scalars_use): Likewise.
13973 (gather_bbs::before_dom_children): Likewise.
13974 * hsa-dump.c (dump_hsa_immed): Likewise.
13975 * ipa-cp.c (print_ipcp_constant_value): Likewise.
13976 (get_replacement_map): Likewise.
13977 * ipa-inline-analysis.c (dump_condition): Likewise.
13978 (estimate_function_body_sizes): Likewise.
13979 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
13980 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
13981 * ipa-prop.c (ipa_dump_param): Likewise.
13982 (ipa_print_node_jump_functions_for_edge): Likewise.
13983 (ipa_modify_call_arguments): Likewise.
13984 (ipa_modify_expr): Likewise.
13985 (ipa_dump_param_adjustments): Likewise.
13986 (ipa_dump_agg_replacement_values): Likewise.
13987 (ipcp_modif_dom_walker::before_dom_children): Likewise.
13988 * ipa-pure-const.c (check_stmt): Likewise.
13989 (pass_nothrow::execute): Likewise.
13990 * ipa-split.c (execute_split_functions): Likewise.
13991 * omp-offload.c (dump_oacc_loop_part): Likewise.
13992 (dump_oacc_loop): Likewise.
13993 * trans-mem.c (tm_log_emit): Likewise.
13994 (tm_memopt_accumulate_memops): Likewise.
13995 (dump_tm_memopt_set): Likewise.
13996 (dump_tm_memopt_transform): Likewise.
13997 * tree-cfg.c (gimple_verify_flow_info): Likewise.
13998 (print_loop): Likewise.
13999 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14000 (chkp_gather_checks_info): Likewise.
14001 (chkp_get_check_result): Likewise.
14002 (chkp_remove_check_if_pass): Likewise.
14003 (chkp_use_outer_bounds_if_possible): Likewise.
14004 (chkp_reduce_bounds_lifetime): Likewise.
14005 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14006 (chkp_mark_completed_bounds): Likewise.
14007 (chkp_register_incomplete_bounds): Likewise.
14008 (chkp_mark_invalid_bounds): Likewise.
14009 (chkp_maybe_copy_and_register_bounds): Likewise.
14010 (chkp_build_returned_bound): Likewise.
14011 (chkp_get_bound_for_parm): Likewise.
14012 (chkp_build_bndldx): Likewise.
14013 (chkp_get_bounds_by_definition): Likewise.
14014 (chkp_generate_extern_var_bounds): Likewise.
14015 (chkp_get_bounds_for_decl_addr): Likewise.
14016 * tree-chrec.c (chrec_apply): Likewise.
14017 * tree-data-ref.c (dump_data_reference): Likewise.
14018 (dump_subscript): Likewise.
14019 (dump_data_dependence_relation): Likewise.
14020 (analyze_overlapping_iterations): Likewise.
14021 * tree-inline.c (expand_call_inline): Likewise.
14022 (tree_function_versioning): Likewise.
14023 * tree-into-ssa.c (dump_defs_stack): Likewise.
14024 (dump_currdefs): Likewise.
14025 (dump_names_replaced_by): Likewise.
14026 (dump_update_ssa): Likewise.
14027 (update_ssa): Likewise.
14028 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14029 * tree-parloops.c (build_new_reduction): Likewise.
14030 (try_create_reduction_list): Likewise.
14031 (ref_conflicts_with_region): Likewise.
14032 (oacc_entry_exit_ok_1): Likewise.
14033 (oacc_entry_exit_single_gang): Likewise.
14034 * tree-pretty-print.h: Likewise.
14035 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14036 (get_scalar_evolution): Likewise.
14037 (add_to_evolution): Likewise.
14038 (get_loop_exit_condition): Likewise.
14039 (analyze_evolution_in_loop): Likewise.
14040 (analyze_initial_condition): Likewise.
14041 (analyze_scalar_evolution): Likewise.
14042 (instantiate_scev): Likewise.
14043 (number_of_latch_executions): Likewise.
14044 (gather_chrec_stats): Likewise.
14045 (final_value_replacement_loop): Likewise.
14046 (scev_const_prop): Likewise.
14047 * tree-sra.c (dump_access): Likewise.
14048 (disqualify_candidate): Likewise.
14049 (create_access): Likewise.
14050 (reject): Likewise.
14051 (maybe_add_sra_candidate): Likewise.
14052 (create_access_replacement): Likewise.
14053 (analyze_access_subtree): Likewise.
14054 (analyze_all_variable_accesses): Likewise.
14055 (sra_modify_assign): Likewise.
14056 (initialize_constant_pool_replacements): Likewise.
14057 (find_param_candidates): Likewise.
14058 (decide_one_param_reduction): Likewise.
14059 (replace_removed_params_ssa_names): Likewise.
14060 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14061 * tree-ssa-copy.c (dump_copy_of): Likewise.
14062 (copy_prop_visit_cond_stmt): Likewise.
14063 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14064 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14065 (record_equivalences_from_stmt): Likewise.
14066 * tree-ssa-dse.c (compute_trims): Likewise.
14067 (delete_dead_call): Likewise.
14068 (delete_dead_assignment): Likewise.
14069 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14070 (forward_propagate_into_cond): Likewise.
14071 (pass_forwprop::execute): Likewise.
14072 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14073 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14075 (move_computations_worker): Likewise.
14076 (execute_sm): Likewise.
14077 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14078 (remove_exits_and_undefined_stmts): Likewise.
14079 (remove_redundant_iv_tests): Likewise.
14080 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14081 (adjust_iv_update_pos): Likewise.
14082 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14083 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14084 (value_replacement): Likewise.
14085 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14086 * tree-ssa-pre.c (print_pre_expr): Likewise.
14087 (get_representative_for): Likewise.
14088 (create_expression_by_pieces): Likewise.
14089 (insert_into_preds_of_block): Likewise.
14090 (eliminate_insert): Likewise.
14091 (eliminate_dom_walker::before_dom_children): Likewise.
14092 (eliminate): Likewise.
14093 (remove_dead_inserted_code): Likewise.
14094 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14095 * tree-ssa-reassoc.c (get_rank): Likewise.
14096 (eliminate_duplicate_pair): Likewise.
14097 (eliminate_plus_minus_pair): Likewise.
14098 (eliminate_not_pairs): Likewise.
14099 (undistribute_ops_list): Likewise.
14100 (eliminate_redundant_comparison): Likewise.
14101 (update_range_test): Likewise.
14102 (optimize_range_tests_var_bound): Likewise.
14103 (optimize_vec_cond_expr): Likewise.
14104 (rewrite_expr_tree): Likewise.
14105 (rewrite_expr_tree_parallel): Likewise.
14106 (linearize_expr): Likewise.
14107 (break_up_subtract): Likewise.
14108 (linearize_expr_tree): Likewise.
14109 (attempt_builtin_powi): Likewise.
14110 (attempt_builtin_copysign): Likewise.
14111 (transform_stmt_to_copy): Likewise.
14112 (transform_stmt_to_multiply): Likewise.
14113 (dump_ops_vector): Likewise.
14114 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
14115 (print_scc): Likewise.
14116 (set_ssa_val_to): Likewise.
14117 (visit_reference_op_store): Likewise.
14118 (visit_use): Likewise.
14119 (sccvn_dom_walker::before_dom_children): Likewise.
14120 (run_scc_vn): Likewise.
14121 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14123 (expr_hash_elt::print): Likewise.
14124 (const_and_copies::pop_to_marker): Likewise.
14125 (const_and_copies::record_const_or_copy_raw): Likewise.
14126 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
14127 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
14128 (dump_predicates): Likewise.
14129 (find_uninit_use): Likewise.
14130 (warn_uninitialized_phi): Likewise.
14131 (pass_late_warn_uninitialized::execute): Likewise.
14132 * tree-ssa.c (verify_vssa): Likewise.
14133 (verify_ssa): Likewise.
14134 (maybe_optimize_var): Likewise.
14135 * tree-vrp.c (dump_value_range): Likewise.
14136 (dump_all_value_ranges): Likewise.
14137 (dump_asserts_for): Likewise.
14138 (register_edge_assert_for_2): Likewise.
14139 (vrp_visit_cond_stmt): Likewise.
14140 (vrp_visit_switch_stmt): Likewise.
14141 (vrp_visit_stmt): Likewise.
14142 (vrp_visit_phi_node): Likewise.
14143 (simplify_cond_using_ranges_1): Likewise.
14144 (fold_predicate_in): Likewise.
14145 (evrp_dom_walker::before_dom_children): Likewise.
14146 (evrp_dom_walker::push_value_range): Likewise.
14147 (evrp_dom_walker::pop_value_range): Likewise.
14148 (execute_early_vrp): Likewise.
14150 2017-05-16 Richard Biener <rguenther@suse.de>
14152 * dwarf2out.c (loc_list_from_tree_1): Do not create
14153 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
14155 2017-05-16 Richard Biener <rguenther@suse.de>
14157 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
14159 (note_variable_value_in_expr): If we resolved the decl ref
14160 do not push to the stack.
14162 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
14164 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
14165 operations in fast-math mode.
14166 (vaddq_f16): Likewise.
14167 (vmul_f16): Likewise.
14168 (vmulq_f16): Likewise.
14169 (vsub_f16): Likewise.
14170 (vsubq_f16): Likewise.
14171 * config/arm/neon.md (add<mode>3): New.
14173 (fma:<VH:mode>3): New. Also remove outdated comment.
14176 2017-05-16 Martin Liska <mliska@suse.cz>
14180 * ipa-devirt.c (warn_types_mismatch): Fix typo.
14181 (odr_types_equivalent_p): Likewise.
14183 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
14185 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
14187 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
14190 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
14191 non-interunit SSE move alternatives with '?'.
14192 (zero-extendsidi peephole2): New peephole to skip intermediate
14193 general register in SSE zero-extend sequence.
14195 2017-05-15 Jeff Law <law@redhat.com>
14197 * reorg.c (relax_delay_slots): Create a new variable to hold
14198 the temporary target rather than clobbering TARGET_LABEL.
14200 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
14201 missing argument to extract_bit_field call.
14202 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
14204 2017-05-15 Martin Liska <mliska@suse.cz>
14207 * gcc.c (process_command): Do not allow empty argument of -o option.
14209 2017-05-15 Renlin Li <renlin.li@arm.com>
14211 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
14212 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
14213 * config/aarch64/constraints.md (Usf): Add long call check.
14214 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
14215 (call_value): Likewise.
14216 (sibcall): Likewise.
14217 (sibcall_value): Likewise.
14219 (call_value_insn): New.
14220 (sibcall_insn): Update rtx pattern.
14221 (sibcall_value_insn): Likewise.
14222 (call_internal): Remove.
14223 (call_value_internal): Likewise.
14224 (sibcall_internal): Likewise.
14225 (sibcall_value_internal): Likewise.
14226 (call_reg): Likewise.
14227 (call_symbol): Likewise.
14228 (call_value_reg): Likewise.
14229 (call_value_symbol): Likewise.
14231 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
14234 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
14236 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
14238 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
14239 compatible with CCGOCmode and with CCZmode.
14241 2017-05-14 Martin Sebor <msebor@redhat.com>
14243 PR middle-end/77671
14244 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
14245 (gimple_fold_builtin_snprintf): Same.
14246 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
14247 (gimple_fold_builtin_snprintf): Same.
14248 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
14249 of character types.
14250 (is_call_safe): New function.
14251 (try_substitute_return_value): Call it.
14252 (try_simplify_call): New function.
14253 (pass_sprintf_length::handle_gimple_call): Call it.
14255 2017-05-14 Martin Sebor <msebor@redhat.com>
14257 PR middle-end/80669
14258 * builtins.c (expand_builtin_stpncpy): Simplify.
14260 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
14262 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
14263 * config/i386/i386.h
14264 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
14265 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
14266 (struct machine_function): Add new members call_ms2sysv,
14267 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
14268 (struct machine_frame_state): New fields sp_realigned and
14269 sp_realigned_offset.
14270 * config/i386/i386.c
14271 (enum xlogue_stub): New enum.
14272 (enum xlogue_stub_sets): New enum.
14273 (class xlogue_layout): New class.
14274 (struct ix86_frame): New fields stack_realign_allocate_offset,
14275 stack_realign_offset and outlined_save_offset. Modify comments to
14276 detail stack layout when using out-of-line stubs.
14277 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
14278 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
14279 -mcall-ms2sysv-xlogues.
14280 (stub_managed_regs): New static variable.
14281 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
14282 registers managed by out-of-line stub.
14283 (disable_call_ms2sysv_xlogues): New function.
14284 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
14285 m->call_ms2sysv when appropriate and compute frame layout for
14287 (sp_valid_at, fp_valid_at): New inline functions.
14288 (choose_basereg): New function.
14289 (choose_baseaddr): Add align parameter, use choose_basereg and modify
14291 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
14292 Use align parameter of choose_baseaddr to generated aligned SSE movs
14294 (pro_epilogue_adjust_stack): Modify to track
14295 machine_frame_state::sp_realigned.
14296 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
14297 (ix86_nsaved_sseregs): Likewise.
14298 (ix86_emit_save_regs): Likewise.
14299 (ix86_emit_save_regs_using_mov): Likewise.
14300 (ix86_emit_save_sse_regs_using_mov): Likewise.
14301 (get_scratch_register_on_entry): Likewise.
14302 (gen_frame_set): New function.
14303 (gen_frame_load): Likewise.
14304 (gen_frame_store): Likewise.
14305 (emit_outlined_ms2sysv_save): Likewise.
14306 (emit_outlined_ms2sysv_restore): Likewise.
14307 (ix86_expand_prologue): Modify stack re-alignment code and call
14308 emit_outlined_ms2sysv_save when appropriate.
14309 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
14310 parameter rtx_insn *insn, which allows the function to be used to only
14311 generate the notes.
14312 (ix86_expand_epilogue): Modify validity checks of frame and stack
14313 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
14314 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
14315 * config/i386/predicates.md
14316 (save_multiple): New predicate.
14317 (restore_multiple): Likewise.
14318 * config/i386/sse.md
14319 (save_multiple<mode>): New pattern.
14320 (save_multiple_realign<mode>): Likewise.
14321 (restore_multiple<mode>): Likewise.
14322 (restore_multiple_and_return<mode>): Likewise.
14323 (restore_multiple_leave_return<mode>): Likewise.
14324 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
14326 2017-05-14 Julia Koval <julia.koval@intel.com>
14328 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
14329 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
14330 (__builtin_ia32_xsetbv): New builtins.
14331 * config/i386/i386.c (ix86_expand_special_args_builtin):
14333 (ix86_expand_builtin): Special expand for new intrinsics.
14334 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
14335 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
14336 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
14338 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14340 * cfganal.c (inverted_post_order_compute): Change argument type
14342 * cfganal.h (inverted_post_order_compute): Adjust prototype.
14343 * df-core.c (rest_of_handle_df_initialize): Adjust.
14344 (rest_of_handle_df_finish): Likewise.
14345 (df_analyze_1): Likewise.
14346 (df_analyze): Likewise.
14347 (loop_inverted_post_order_compute): Change argument to be a vec *.
14348 (df_analyze_loop): Adjust.
14349 (df_get_n_blocks): Likewise.
14350 (df_get_postorder): Likewise.
14351 * df.h (struct df_d): Change field to be a vec.
14352 * lcm.c (compute_laterin): Adjust.
14353 (compute_available): Likewise.
14354 * lra-lives.c (lra_create_live_ranges_1): Likewise.
14355 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
14356 * tree-ssa-pre.c (compute_antic): Likewise.
14358 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14360 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
14361 (depth_first_search::depth_first_search): Change structure init
14362 function to this constructor.
14363 (depth_first_search::add_bb): Rename function to this member.
14364 (depth_first_search::execute): Likewise.
14365 (flow_dfs_compute_reverse_finish): Adjust.
14367 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14369 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
14370 (longest_simple_path): Likewise.
14371 * shrink-wrap.c (spread_components): Likewise.
14372 (disqualify_problematic_components): Likewise.
14373 (emit_common_heads_for_components): Likewise.
14374 (emit_common_tails_for_components): Likewise.
14375 (insert_prologue_epilogue_for_components): Likewise.
14377 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14379 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
14382 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14384 * df-core.c (df_set_blocks): Start using auto_bitmap.
14385 (df_compact_blocks): Likewise.
14386 * df-problems.c (df_rd_confluence_n): Likewise.
14387 * df-scan.c (df_insn_rescan_all): Likewise.
14388 (df_process_deferred_rescans): Likewise.
14389 (df_update_entry_block_defs): Likewise.
14390 (df_update_exit_block_uses): Likewise.
14391 (df_entry_block_bitmap_verify): Likewise.
14392 (df_exit_block_bitmap_verify): Likewise.
14393 (df_scan_verify): Likewise.
14394 * lra-constraints.c (lra_constraints): Likewise.
14395 (undo_optional_reloads): Likewise.
14396 (lra_undo_inheritance): Likewise.
14397 * lra-remat.c (calculate_gen_cands): Likewise.
14398 (do_remat): Likewise.
14399 * lra-spills.c (assign_spill_hard_regs): Likewise.
14400 (spill_pseudos): Likewise.
14401 * tree-ssa-pre.c (bitmap_set_and): Likewise.
14402 (bitmap_set_subtract_values): Likewise.
14404 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14406 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
14407 management with auto_bitmap.
14408 (fix_inter_tick): Likewise.
14409 (fix_recovery_deps): Likewise.
14410 * ira.c (add_store_equivs): Likewise.
14411 (find_moveable_pseudos): Likewise.
14412 (split_live_ranges_for_shrink_wrap): Likewise.
14413 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
14414 (rtx_reuse_manager::seen_def_p): Likewise.
14415 (rtx_reuse_manager::set_seen_def): Likewise.
14416 * print-rtl.h (class rtx_reuse_manager): Likewise.
14418 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14420 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
14422 (migrate_btr_def): Likewise.
14423 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
14424 * df-core.c (loop_post_order_compute): Likewise.
14425 (loop_inverted_post_order_compute): Likewise.
14426 * hsa-common.h: Likewise.
14427 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
14428 * init-regs.c (initialize_uninitialized_regs): Likewise.
14429 * ipa-inline.c (resolve_noninline_speculation): Likewise.
14430 (inline_small_functions): Likewise.
14431 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
14432 * ira.c (combine_and_move_insns): Likewise.
14433 (build_insn_chain): Likewise.
14434 * loop-invariant.c (find_invariants): Likewise.
14435 * lower-subreg.c (propagate_pseudo_copies): Likewise.
14436 * predict.c (tree_predict_by_opcode): Likewise.
14437 (predict_paths_leading_to): Likewise.
14438 (predict_paths_leading_to_edge): Likewise.
14439 (estimate_loops_at_level): Likewise.
14440 (estimate_loops): Likewise.
14441 * shrink-wrap.c (try_shrink_wrapping): Likewise.
14442 (spread_components): Likewise.
14443 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
14444 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
14445 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
14446 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
14447 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
14448 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
14449 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
14450 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
14451 (mark_threaded_blocks): Likewise.
14452 (thread_through_all_blocks): Likewise.
14453 * tree-ssa.c (verify_ssa): Likewise.
14454 (execute_update_addresses_taken): Likewise.
14455 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
14457 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14459 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
14461 (post_order_compute): Likewise.
14462 (inverted_post_order_compute): Likewise.
14463 (pre_and_rev_post_order_compute_fn): Likewise.
14465 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14467 * genrecog.c (int_set::int_set): Explicitly construct our
14468 auto_vec base class.
14469 * vec.h (auto_vec::auto_vec): New constructor.
14471 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14473 * bitmap.h (class auto_bitmap): New constructor taking
14474 bitmap_obstack * argument.
14476 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14478 * bitmap.h (class auto_bitmap): Change type of m_bits to
14479 bitmap_head, and adjust ctor / dtor and member operators.
14481 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
14483 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
14484 when returned register mode doesn't match original mode.
14486 2017-05-12 Jeff Law <law@redhat.com>
14487 Jakub Jelinek <jakub@redhat.com>
14489 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
14490 we look for cc setter after the compare-elim changes.
14491 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
14492 within the vector to match what compare-elim now expects.
14493 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
14494 (xorsi3_flags, one_cmplsi2_flags): Likewise.
14496 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
14497 after the compare-elim changes.
14498 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
14499 the vector to match what compare-elim now expects.
14500 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
14501 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
14502 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
14503 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
14504 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
14506 * config/visium/visium.c (single_set_and_flags): Fix where
14507 we look for cc setter after the compare-elim changes.
14508 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
14509 with the vector to match what compare-elim now expects.
14510 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
14511 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
14512 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
14513 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
14514 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
14515 (neg<mode>2_insn_set_overflow): Likewise.
14517 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
14519 PR middle-end/79794
14520 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
14521 maybe_expand_insn call, set ops[0].target. If still set after call,
14522 set alt_rtl. Add extra arg to recursive calls.
14523 (extract_bit_field): Add alt_rtl argument. Pass to
14525 * expmed.h (extract_bit_field): Fix prototype.
14526 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
14527 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
14528 to extract_bit_field_calls.
14529 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
14530 Pass alt_rtl to extract_bit_field calls.
14531 * calls.c (store_unaligned_arguments_into_psuedos)
14532 load_register_parameters): Pass extra NULL to extract_bit_field calls.
14533 * optabs.c (maybe_legitimize_operand): Clear op->target when call
14535 * optabs.h (struct expand_operand): Add target bitfield.
14537 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
14539 * compare-elim.c (try_eliminate_compare): Canonicalize
14540 operation with embedded compare to
14541 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
14542 (set (reg) (operation)].
14544 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
14546 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
14549 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
14550 cost of adding a carry flag for ADC instruction.
14551 [case MINUS]: Ignore the cost of subtracting a carry flag
14552 for SBB instruction.
14554 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
14556 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
14558 * config/rs6000/bmiintrin.h: New file.
14559 * config/rs6000/bmi2intrin.h: New file.
14560 * config/rs6000/x86intrin.h: New file.
14562 2017-05-12 Jeff Law <law@redhat.com>
14564 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
14567 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
14569 PR middle-end/80707
14570 * tree-cfg.c: Remove cfg edges of unreachable case statements.
14572 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14574 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14575 early expansion of vector divide builtins.
14576 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
14577 builtins identified as having unsigned arguments.
14579 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14581 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
14582 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
14583 expansion of vector logical operations (and, andc, or, xor,
14586 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14588 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
14589 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
14591 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14593 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14594 early GIMPLE expansion of vector multiplies.
14596 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14598 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
14599 TARGET_HAVE_MOVT conditional.
14600 (movt splitter): Likewise.
14602 2017-05-12 Richard Biener <rguenther@suse.de>
14604 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
14605 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14606 Fold all stmts not inplace.
14608 2017-05-12 Richard Biener <rguenther@suse.de>
14610 PR tree-optimization/80713
14611 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
14612 inserted_exprs bit for not removed stmts.
14614 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
14616 PR middle-end/69921
14617 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
14618 parallelized" attribute for parallelized OpenACC kernels.
14619 * omp-offload.c (execute_oacc_device_lower): Use it.
14621 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
14622 Set "oacc kernels" attribute.
14623 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
14624 parameter. Adjust all users.
14625 (oacc_fn_attrib_kernels_p): Remove function.
14626 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
14627 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
14628 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
14629 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
14630 assert "oacc kernels" attribute is set.
14632 2017-05-11 Carl Love <cel@us.ibm.com>
14634 * config/rs6000/rs6000-c: Add support for built-in functions
14635 vector unsigned char vec_popcnt (vector signed char)
14636 vector unsigned char vec_popcnt (vector unsigned char)
14637 vector unsigned short vec_popcnt (vector signed short)
14638 vector unsigned short vec_popcnt (vector unsigned short)
14639 vector unsigned int vec_popcnt (vector signed int)
14640 vector unsigned int vec_popcnt (vector unsigned int)
14641 vector unsigned long long vec_popcnt (vector signed long long)
14642 vector unsigned long long vec_popcnt (vector unsigned long long)
14643 vector signed long long vec_slo (vector signed long long,
14644 vector signed char)
14645 vector signed long long vec_slo (vector signed long long,
14646 vector unsigned char)
14647 vector unsigned long long vec_slo (vector unsigned long long,
14648 vector signed char)
14649 vector unsigned long long vec_slo (vector unsigned long long,
14650 vector unsigned char)
14651 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
14652 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
14653 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
14654 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
14655 * doc/extend.texi: Update the built-in documentation file for the
14656 new built-in functions.
14658 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
14660 * attribs.h (sorted_attr_string): Move machine independent
14661 functions for target clone support from the i386 port to common
14662 code. Rename ix86_function_versions to common_function_versions.
14663 Rename make_name to make_unique_name.
14664 (common_function_versions): Likewise.
14665 (make_unique_name): Likewise.
14666 (make_dispatcher_decl): Likewise.
14667 (is_function_default_version): Likewise.
14668 * attribs.c (attr_strcmp): Likewise.
14669 (sorted_attr_string): Likewise.
14670 (common_function_versions): Likewise.
14671 (make_unique_name): Likewise.
14672 (make_dispatcher_decl): Likewise.
14673 (is_function_default_version): Likewise.
14674 * config/i386/i386.c (attr_strcmp): Likewise.
14675 (sorted_attr_string): Likewise.
14676 (ix86_function_versions): Likewise.
14677 (make_name): Likewise.
14678 (make_dispatcher_decl): Likewise.
14679 (is_function_default_version): Likewise.
14680 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
14682 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14685 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14686 Account for direct move costs for vec_construct of integer
14689 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
14692 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
14693 (UNSPEC_STX_ATOMIC): Ditto.
14694 (loaddi_via_sse): New insn.
14695 (storedi_via_sse): Ditto.
14696 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
14697 Update corresponding peephole2 patterns.
14698 (atomic_storedi_fpu): Ditto.
14700 2017-05-11 Julia Koval <julia.koval@intel.com>
14702 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
14703 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
14705 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
14706 (__builtin_ia32_rsqrt14ss_mask): New builtins.
14707 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
14709 2017-05-11 Nathan Sidwell <nathan@acm.org>
14711 * graphite-poly.c: Include dumpfile.h.
14713 * dumpfle.h (dump_function): Declare here ...
14714 * tree-dump.h (dump_function): ... not here.
14715 * dumpfile.c: #include tree-cfg.h.
14716 (dump_function): Move here from ...
14717 * tree-dump.c (dump_function): ... here.
14718 * gimplify.c: #include splay-tree.h, not tree-dump.h.
14719 * graphite-poly.c: Don't include tree-dump.h.
14720 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
14721 * print-tree.c: Likewise.
14722 * stor-layout.c: Likewise.
14723 * tree-nested.c: Likewise.
14725 * dumpfile.c (dump_start): Use TDF_FLAGS.
14726 (dump_enable_all): Fix TDF_KIND check thinko.
14728 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
14730 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14731 array entries to represent two legal parameterizations of the
14732 overloaded __builtin_cmpb function, as represented by the
14733 P6_OV_BUILTIN_CMPB constant.
14734 (altivec_resolve_overloaded_builtin): Add special case handling
14735 for the __builtin_cmpb function, as represented by the
14736 P6_OV_BUILTIN_CMPB constant.
14737 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
14738 (BU_P6_64BIT_2): New macro.
14739 (BU_P6_OVERLOAD_2): New macro
14740 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
14741 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
14742 (CMPB): Add overload support to represent both 32-bit and 64-bit
14743 compare-bytes function.
14744 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14745 support for TARGET_CMPB.
14746 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
14747 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14748 documentation of the __builtin_cmpb overloaded built-in function.
14750 2017-05-11 Richard Biener <rguenther@suse.de>
14752 PR tree-optimization/80705
14753 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
14754 bases are not vectorizable.
14756 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14758 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
14759 when counting register pressure.
14761 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14763 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
14764 (iv_ca_more_deps): Renamed to ...
14765 (iv_ca_compare_deps): ... this.
14766 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
14768 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14770 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
14772 (determine_group_iv_costs): ... here.
14773 (find_inv_vars_cb): Record inv var if it's not recorded before.
14775 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14777 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
14778 (get_shiftadd_cost): Ditto.
14780 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14782 * tree-ssa-address.c: Include header file.
14783 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
14785 (add_to_parts): Refactor.
14786 (addr_to_parts): New parameter. Update use of move_hint_to_base.
14787 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
14790 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14792 PR tree-optimization/53090
14793 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
14795 (extract_cond_operands): Detect condition with IV on both sides
14796 and return COMP_IV_EXPR_2.
14797 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
14798 (rewrite_use_compare): Simplify by removing call to function
14799 extract_cond_operands.
14801 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14803 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
14804 (extract_cond_operands): Detect condition comparing against non-
14805 invariant bound and return appropriate enum value.
14806 (find_interesting_uses_cond): Update use of extract_cond_operands.
14807 Handle its return value accordingly.
14808 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
14810 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14812 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
14813 nonlinear iv_use computation in loop invariant sensitive way.
14815 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14817 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
14818 (find_iv_candidates): Call relate_compare_use_with_all_cands.
14820 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14822 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
14823 (dump_cand): Support iv_cand.inv_exprs.
14824 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
14826 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
14829 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14831 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
14833 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
14834 as local function. Include necessary header files.
14835 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
14837 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14839 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
14841 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14843 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
14844 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
14845 RSHIFT_EXPR and BIT_NOT_EXPR.
14847 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14849 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
14850 (adjust_setup_cost): New parameter supporting round up adjustment.
14851 (struct address_cost_data): Delete.
14852 (force_expr_to_var_cost): Don't bound cost with spill_cost.
14853 (split_address_cost, ptr_difference_cost): Delete.
14854 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
14855 (struct ainc_cost_data): New struct.
14856 (get_address_cost_ainc): New function.
14857 (get_address_cost, get_computation_cost): Reimplement.
14858 (determine_group_iv_cost_address): Record inv_expr for all uses of
14860 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
14861 (iv_ca_has_deps): Reimplemented to ...
14862 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
14864 (iv_ca_extend): Call iv_ca_more_deps.
14866 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14868 * tree-ssa-address.c (struct mem_address): Move to header file.
14869 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
14870 * tree-ssa-address.h (struct mem_address): Move from C file.
14871 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
14873 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14875 * tree-affine.h (aff_combination_type): New interface.
14876 (aff_combination_zero_p): Remove static.
14877 (aff_combination_const_p): New interface.
14878 (aff_combination_singleton_var_p): New interfaces.
14880 2017-05-11 Richard Biener <rguenther@suse.de>
14882 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14883 Skip unreachable blocks and destinations.
14884 (eliminate): Move stmt removal and fixup ...
14885 (fini_eliminate): ... here. Skip inserted exprs.
14886 (pass_pre::execute): Move fini_pre after fini_eliminate.
14887 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
14888 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
14889 PRE to get rid of dead code that has invalid SSA form and
14890 split critical edges again.
14892 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14894 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14896 2017-05-11 Richard Biener <rguenther@suse.de>
14898 * passes.c (execute_function_todo): Verify loops if they are
14899 said to be up-to-date.
14900 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
14901 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
14903 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
14906 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
14907 handle calling assemble_external ourself.
14910 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
14911 modes with zero size. Enhance comment.
14913 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14915 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
14916 built-ins for vec_xl and vec_xst with short and char pointer
14919 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
14921 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
14922 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
14923 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
14924 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
14925 (_mm_maskz_min_round_ss): New intrinsics.
14926 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
14927 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
14928 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
14929 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
14930 (__builtin_ia32_minss_mask_round): New builtins.
14931 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14932 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14933 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
14935 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
14936 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
14938 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
14941 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
14943 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
14944 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
14945 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
14946 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
14947 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
14948 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
14949 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
14950 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14951 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
14952 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
14953 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
14954 (__builtin_ia32_mulss_mask_round): New builtins.
14955 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14956 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14957 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
14959 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
14960 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
14962 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
14965 2017-05-10 Julia Koval <julia.koval@intel.com>
14967 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
14968 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
14969 (_mm256_setr_m128i): New intrinsics.
14971 2017-05-10 Julia Koval <julia.koval@intel.com>
14973 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
14974 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
14975 (_mm_maskz_rcp14_ss): New intrinsics.
14976 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
14977 (__builtin_ia32_rcp14ss_mask): New builtins.
14978 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
14980 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
14982 PR tree-optimization/51513
14983 * tree-cfg.c (gimple_seq_unreachable_p): New function.
14984 (assert_unreachable_fallthru_edge_p): Use it.
14985 (group_case_labels_stmt): Likewise.
14986 * tree-cfg.h: Prototype it.
14987 * stmt.c: Include cfghooks.h and tree-cfg.h.
14988 (emit_case_dispatch_table) <gap_label>: New local variable.
14989 Use it to fill dispatch table gaps.
14990 Test for default_label before updating probabilities.
14991 (expand_case) <default_label>: Remove unneeded initialization.
14992 Test for unreachable default case statement and remove its edge.
14993 Set default_label accordingly.
14994 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
14996 2017-05-10 Carl Love <cel@us.ibm.com>
14998 * config/rs6000/rs6000-c: Add support for built-in functions
14999 vector signed char vec_neg (vector signed char)
15000 vector signed short int vec_neg (vector short int)
15001 vector signed int vec_neg (vector signed int)
15002 vector signed long long vec_neg (vector signed long long)
15003 vector float vec_neg (vector float)
15004 vector double vec_neg (vector double)
15005 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15007 * config/rs6000/altivec.h: Add define for vec_neg
15008 * doc/extend.texi: Update the built-in documentation for the
15009 new built-in functions.
15011 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15013 PR tree-optimization/77644
15014 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15016 2017-05-10 Nathan Sidwell <nathan@acm.org>
15018 * dumpfile.h (TDI_lang_all): New.
15019 (TDF_KIND): New. Renumber others
15020 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15022 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15024 (get_dump_file_name): Adjust suffix generation.
15025 (dump_enable_all): Use TDF_KIND.
15026 * doc/invoke.texi (-fdump-lang-all): Document.
15028 * dumpfile.h: Tabify.
15030 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15033 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15034 Move member access before delete.
15036 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15038 * tree-inline.c (expand_call_inline): Split block at stmt
15041 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15044 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15045 are now unused after splitting mov{sf,sd}_hardfloat.
15046 (f32_lr2): Likewise.
15047 (f32_lm): Likewise.
15048 (f32_lm2): Likewise.
15049 (f32_li): Likewise.
15050 (f32_li2): Likewise.
15051 (f32_lv): Likewise.
15052 (f32_sr): Likewise.
15053 (f32_sr2): Likewise.
15054 (f32_sm): Likewise.
15055 (f32_sm2): Likewise.
15056 (f32_si): Likewise.
15057 (f32_si2): Likewise.
15058 (f32_sv): Likewise.
15059 (f32_dm): Likewise.
15060 (f32_vsx): Likewise.
15061 (f32_av): Likewise.
15062 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15063 For movsf, order stores so the VSX stores occur before the GPR
15064 store which encourages the register allocator to use a traditional
15065 FPR instead of a GPR. For movsd, order the stores so that the GPR
15066 store comes before the VSX stores to allow the power6 to work.
15067 This is due to the power6 not having a 32-bit integer store
15068 instruction from a FPR.
15069 (movsf_hardfloat): Likewise.
15070 (movsd_hardfloat): Likewise.
15072 2017-05-09 Martin Sebor <msebor@redhat.com>
15074 PR translation/80280
15075 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15078 PR translation/80280
15079 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15080 data member added in r247778.
15081 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15083 2017-05-09 Nathan Sidwell <nathan@acm.org>
15085 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15087 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15090 2017-05-09 Marek Polacek <polacek@redhat.com>
15092 * doc/invoke.texi: Fix typo.
15094 2017-05-09 Richard Biener <rguenther@suse.de>
15096 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15097 (vrp_val_is_min): Likewise.
15098 (set_value_range_to_value): Likewise.
15099 (set_value_range_to_nonnegative): Likewise.
15100 (gimple_assign_nonzero_p): Likewise.
15101 (gimple_stmt_nonzero_p): Likewise.
15102 (vrp_int_const_binop): Likewise. Remove unreachable case.
15103 (adjust_range_with_scev): Adjust comments.
15104 (compare_range_with_value): Likewise.
15105 (extract_range_from_phi_node): Likewise.
15106 (test_for_singularity): Likewise.
15108 2017-05-09 Richard Biener <rguenther@suse.de>
15110 * tree-vrp.c (get_single_symbol): Add assert that we don't
15111 get overflowed constants as invariant part.
15112 (compare_values_warnv): Add comment before the TREE_NO_WARNING
15113 checks. Use wi::cmp instead of recursing for integer constants.
15114 (compare_values): Just ignore whether we assumed undefined
15115 overflow instead of failing the compare.
15116 (extract_range_for_var_from_comparison_expr): Add comment before the
15117 TREE_NO_WARNING sets.
15118 (test_for_singularity): Likewise.
15119 (extract_range_from_comparison): Do not disable optimization
15120 when we assumed undefined overflow.
15121 (extract_range_basic): Remove init of unused var.
15123 2017-05-09 Richard Biener <rguenther@suse.de>
15125 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
15126 (extract_range_from_multiplicative_op_1): Adjust.
15127 (extract_range_from_binary_expr_1): Use int_const_binop.
15129 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15132 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
15133 rs6000_store_data_bypass_p in seven define_bypass directives and
15134 in several comments.
15135 * config/rs6000/rs6000-protos.h: Add prototype for
15136 rs6000_store_data_bypass_p function.
15137 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
15138 function implements slightly different (rs6000-specific) semantics
15139 than store_data_bypass_p, returning false rather than aborting
15140 with assertion error when arguments do not satisfy the
15141 requirements of store data bypass.
15142 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
15143 rs6000_store_data_bypass_p.
15145 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
15147 * config/xtensa/xtensa-protos.h
15148 (xtensa_initial_elimination_offset): New declaration.
15149 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15150 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
15151 macro definition, add case for FRAME_POINTER_REGNUM when
15152 FRAME_GROWS_DOWNWARD.
15153 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
15154 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
15155 xtensa_initial_elimination_offset.
15157 2017-05-08 Nathan Sidwell <nathan@acm.org>
15159 * doc/invoke.texi: Alphabetize -fdump options.
15161 2017-05-08 Martin Sebor <msebor@redhat.com>
15163 PR translation/80280
15164 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
15166 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15168 * target.def (compute_frame_layout): New optional target hook.
15169 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
15170 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
15171 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
15173 * reload1.c (verify_initial_elim_offsets): Likewise.
15174 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
15175 (use_simple_return_p): Call arm_compute_frame_layout if needed.
15176 (arm_get_frame_offsets): Split up into this ...
15177 (arm_compute_frame_layout): ... and this function.
15179 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
15181 * config/aarch64/constraints.md (Usa): New constraint.
15182 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
15184 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
15186 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
15187 with_multilib_list after it has been checked.
15189 2017-05-08 Richard Biener <rguenther@suse.de>
15191 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
15192 (bitmap_set_subtract_values): Likewise.
15194 2017-05-08 Richard Biener <rguenther@suse.de>
15196 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
15197 (gimple_assign_nonzero): ... this and remove strict_overflow_p
15199 (gimple_stmt_nonzero_warnv_p): Rename to ...
15200 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
15202 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
15203 (extract_range_basic): Adjust, do not disable propagation on
15204 strict overflow sensitive simplification.
15205 (vrp_visit_cond_stmt): Likewise.
15207 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
15209 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
15210 body size unconditionally.
15212 2017-05-07 Jeff Law <law@redhat.com>
15215 2017-05-06 Jeff Law <law@redhat.com>
15216 PR tree-optimization/78496
15217 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15220 PR tree-optimization/78496
15221 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15222 (simplify_stmt_using_ranges): Call it.
15223 (vrp_dom_walker::before_dom_children): Extract equivalences
15224 from an ASSERT_EXPR with an equality comparison against a
15227 2017-05-06 Jeff Law <law@redhat.com>
15229 PR tree-optimization/78496
15230 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15233 PR tree-optimization/78496
15234 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15235 (simplify_stmt_using_ranges): Call it.
15236 (vrp_dom_walker::before_dom_children): Extract equivalences
15237 from an ASSERT_EXPR with an equality comparison against a
15240 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15242 * lra-constraints.c (lra_copy_reg_equiv): New function.
15243 (split_reg): Use it to copy equivalence information from the
15244 original register to the spill register.
15246 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15248 PR rtl-optimization/75964
15249 * simplify-rtx.c (simplify_const_relational_operation): Remove
15250 invalid handling of comparisons of integer ABS.
15252 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
15254 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
15255 initialize to zero.
15256 (init_regs): Remove declaration.
15257 (function_arg_advance_32): Initialize error_p as boolean variable.
15259 2017-05-05 Nathan Sidwell <nathan@acm.org>
15261 * store-motion.c (remove_reachable_equiv_notes): Reformat long
15262 lines. Use for (;;).
15264 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15266 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
15267 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
15268 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
15269 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
15270 VF=2 that require versioning.
15272 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15274 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
15277 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15279 * diagnostic.h (diagnostic_override_option_index): Convert from
15280 macro to inline function.
15282 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15284 * diagnostic.c (last_module_changed_p): New function.
15285 (set_last_module): New function.
15286 (diagnostic_report_current_module): Convert macro usage to
15287 the above functions.
15288 * diagnostic.h (diagnostic_context::last_module): Strengthen
15289 from const line_map * to const line_map_ordinary *.
15290 (diagnostic_last_module_changed): Delete macro.
15291 (diagnostic_set_last_module): Delete macro.
15293 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15295 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
15296 with diagnostic_report_diagnostic.
15297 (diagnostic_n_impl_richloc): Likewise.
15298 * diagnostic.h (report_diagnostic): Delete macro.
15299 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
15300 with diagnostic_report_diagnostic.
15301 * substring-locations.c (format_warning_va): Likewise.
15303 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15305 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
15306 save/restor of format_spec. Move option-printing code to...
15307 (print_option_information): ...this new function, and
15308 reimplement by simply printing to the pretty_printer,
15309 rather than appending to the format string.
15311 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15313 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
15314 handling logic into...
15315 (update_effective_level_from_pragmas): ...this new function.
15317 2017-05-04 Andrew Waterman <andrew@sifive.com>
15319 * config/riscv/riscv.opt (mstrict-align): New option.
15320 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
15321 (SLOW_UNALIGNED_ACCESS): Define.
15322 (riscv_slow_unaligned_access): Declare.
15323 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
15325 (riscv_slow_unaligned_access): New variable.
15326 (rocket_tune_info): Set slow_unaligned_access to true.
15327 (optimize_size_tune_info): Set slow_unaligned_access to false.
15328 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
15329 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
15330 (riscv_option_override): Set riscv_slow_unaligned_access.
15331 * doc/invoke.texi: Add -mstrict-align to RISC-V.
15333 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
15335 * config/riscv/riscv.md: Unify indentation.
15337 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15342 * config/rs6000/rs6000.md (u code attribute): Add FIX and
15344 (extendsi<mode>2): Add support for doing sign extension via
15345 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
15346 don't have ISA 3.0 instructions.
15347 (extendsi<mode>2 splitter): Likewise.
15348 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
15349 generate the normal insns since SImode can now go in vector
15350 registers. Disallow the special UNSPECs needed for previous
15351 machines to hide SImode being used. Add new insns
15352 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
15353 (fix_trunc<mode>si2_stfiwx): Likewise.
15354 (fix_trunc<mode>si2_internal): Likewise.
15355 (fixuns_trunc<mode>si2): Likewise.
15356 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15357 (fctiw<u>z_<mode>_smallint): Likewise.
15358 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
15359 of floating point to 32-bit integer from doing a direct move to
15360 the GPR registers to do a store.
15361 (fctiwz_<mode>): Break long line.
15363 2017-05-05 Bin Cheng <bin.cheng@arm.com>
15365 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
15366 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
15367 (addr_list, addr_offset_valid_p): New.
15368 (split_address_groups): Check offset validity with above function.
15369 (gt-tree-ssa-loop-ivopts.h): Include header file.
15371 2017-05-05 Nathan Sidwell <nathan@acm.org>
15373 * config.gcc (arm*-*-*): Add missing 'fi'.
15375 2017-05-05 Steve Ellcey <sellcey@cavium.com>
15377 * doc/invoke.texi (-fopt-info): Explicitly say order of options
15378 included in -fopt-info does not matter.
15379 * doc/optinfo.texi (-fopt-info): Fix description of default
15380 behavour. Explicitly say order of options included in -fopt-info
15383 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15385 * config.gcc: Allow combinations of aprofile and rmprofile values for
15386 --with-multilib-list.
15387 * config/arm/t-multilib: New file.
15388 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
15389 variables. Remove setting of ISA and floating-point ABI in
15390 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
15391 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
15392 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
15393 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
15395 * config/arm/t-rmprofile: Likewise except for the matches changes.
15396 * doc/install.texi (--with-multilib-list): Document the combination of
15397 aprofile and rmprofile values and warn about pitfalls in doing that.
15399 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15401 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
15402 (movdi_aarch64): Likewise.
15404 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15406 PR tree-optimization/80632
15407 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
15409 (build_arrays): Initialize it for virtual phis.
15410 (fix_phi_nodes): Use it for virtual phis.
15412 PR tree-optimization/80558
15413 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
15414 [x, y] op z into [x op, y op z] for op & or | if conditions
15417 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
15418 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15421 * config/arm/arm.md (use_literal_pool): Remove.
15422 (64-bit immediate split): No longer takes cost into consideration
15423 if arm_disable_literal_pool is enabled.
15424 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
15425 used when arm_disable_literal_pool is enabled.
15426 (arm_max_const_double_inline_cost): Remove use of
15427 arm_disable_literal_pool.
15428 (push_minipool_fix): Add assert.
15429 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
15430 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
15431 (no_literal_pool_sf_immediate): New.
15433 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15435 PR tree-optimization/80613
15436 * tree-ssa-dce.c (propagate_necessity): Remove cases for
15437 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
15439 2017-05-05 Richard Biener <rguenther@suse.de>
15441 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
15443 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
15445 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
15446 of this flag from insn conditions due to removal from r247495.
15448 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15450 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
15452 (arm_early_store_addr_dep_ptr): Likewise.
15453 * config/arm/aarch-common-protos.h
15454 (arm_early_load_addr_dep_ptr): Add prototype.
15455 (arm_early_store_addr_dep_ptr): Likewise.
15456 * config/arm/cortex-a53.md: Add new bypasses.
15458 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15460 * tree.c (next_type_uid): Change type to unsigned.
15461 (type_hash_canon): Decrement back next_type_uid if
15462 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
15463 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
15466 2017-05-04 Martin Sebor <msebor@redhat.com>
15468 * builtins.c: Fix a trivial typo in a comment.
15470 PR middle-end/79234
15471 * builtins.c (check_sizes): Adjust to handle reading past the end.
15472 Avoid printing excessive upper bound of ranges. Use %E to print
15473 tree nodes instead of converting them to %wu.
15474 (expand_builtin_memchr): New function.
15475 (compute_dest_size): Rename...
15476 (compute_objsize): ...to this.
15477 (expand_builtin_memcpy): Adjust.
15478 (expand_builtin_mempcpy): Adjust.
15479 (expand_builtin_strcat): Adjust.
15480 (expand_builtin_strcpy): Adjust.
15481 (check_strncat_sizes): Adjust.
15482 (expand_builtin_strncat): Adjust.
15483 (expand_builtin_strncpy): Adjust and simplify.
15484 (expand_builtin_memset): Adjust.
15485 (expand_builtin_bzero): Adjust.
15486 (expand_builtin_memcmp): Adjust.
15487 (expand_builtin): Handle memcmp.
15488 (maybe_emit_chk_warning): Check strncat just once.
15490 2017-05-04 Martin Sebor <msebor@redhat.com>
15492 PR preprocessor/79214
15493 PR middle-end/79222
15494 PR middle-end/79223
15495 * builtins.c (check_sizes): Add inlining context and issue
15496 warnings even when -Wno-system-headers is set.
15497 (check_strncat_sizes): Same.
15498 (expand_builtin_strncat): Same.
15499 (expand_builtin_memmove): New function.
15500 (expand_builtin_stpncpy): Same.
15501 (expand_builtin): Handle memmove and stpncpy.
15503 2017-05-04 Bin Cheng <bin.cheng@arm.com>
15505 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
15506 which is not used any more.
15508 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15510 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
15512 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15514 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
15515 (cortexa53_tunings): Likewise.
15516 (cortexa57_tunings): Likewise.
15517 (cortexa72_tunings): Likewise.
15518 (cortexa73_tunings): Likewise.
15520 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15522 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
15523 Set loop alignment to 8.
15525 2017-05-04 Martin Sebor <msebor@redhat.com>
15527 PR translation/80280
15528 * builtins.c (expand_builtin_object_size): Add missing quoting to
15529 %D and like directives.
15530 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
15531 (hsa_type_for_tree_type): Same.
15532 (verify_function_arguments): Same.
15533 * symtab.c (symbol_table::change_decl_assembler_name): Same.
15534 * varasm.c (get_section): Same.
15537 2017-05-04 Martin Sebor <msebor@redhat.com>
15539 PR translation/80280
15540 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
15542 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15544 * config/aarch64/aarch64.c (generic_addrcost_table):
15545 Change HI/TI mode setting.
15547 2017-05-04 Martin Jambor <mjambor@suse.cz>
15549 PR tree-optimization/80622
15550 * tree-sra.c (comes_initialized_p): New function.
15551 (build_accesses_from_assign): Only set write lazily when
15552 comes_initialized_p is false.
15553 (analyze_access_subtree): Use comes_initialized_p.
15554 (propagate_subaccesses_across_link): Assert !comes_initialized_p
15555 instead of testing for PARM_DECL.
15557 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15559 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
15560 constraint on operand 0 to allow more general addressing modes.
15561 Adjust output template.
15562 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
15564 * config/aarch64/aarch64-protos.h
15565 (aarch64_address_valid_for_prefetch_p): Declare prototype.
15566 * config/aarch64/constraints.md (Dp): New address constraint.
15567 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
15570 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
15572 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
15573 update use of estimate_ipcp_clone_size_and_time.
15574 (estimate_local_effects): Update use of
15575 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
15576 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
15577 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
15578 Return nonspecialized time.
15580 2017-05-04 Richard Biener <rguenther@suse.de>
15582 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
15583 for the last VUSE which def dominates the PHI. Directly call
15585 (get_continuation_for_phi_1): Remove.
15587 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
15589 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
15590 to explain the use of truncating division. Cap the number of
15591 iterations to the maximum given by nb_iterations_upper_bound,
15594 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15596 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
15597 * configure: Regenerate.
15598 * config.in: Regenerate.
15599 * config/i386/driver-mingw32.c: new file.
15600 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
15601 * config.host: Link driver-mingw32.o on MinGW host.
15602 * doc/install.texi: Document new --enable-mingw-wildcard configure
15605 2017-05-04 Marek Polacek <polacek@redhat.com>
15607 PR tree-optimization/80612
15608 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
15610 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15611 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
15613 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
15614 (movt splitter): Likewise.
15615 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
15616 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
15617 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
15618 block for Thumb-1 with MOVT.
15619 (thumb2_legitimate_address_p): Move code block ...
15620 (can_avoid_literal_pool_for_label_p): ... into this new function.
15621 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
15623 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
15624 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
15625 "M-profile targets with the MOVT instruction".
15627 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15629 * config/arm/arm-builtins.c (arm_init_builtins): Rename
15630 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
15631 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
15633 2017-05-04 Martin Liska <mliska@suse.cz>
15635 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
15636 variable cond_code.
15638 2017-05-04 Richard Biener <rguenther@suse.de>
15640 * tree.c (array_at_struct_end_p): Handle arrays at struct
15641 end with flexarrays more conservatively. Refactor and treat
15642 arrays of arrays or aggregates more strict. Fix
15643 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
15644 * tree.c (array_at_struct_end_p): Adjust prototype.
15645 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
15646 * gimple-fold.c (get_range_strlen): Likewise.
15647 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
15649 2017-05-04 Richard Biener <rguenther@suse.de>
15651 PR tree-optimization/31130
15652 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
15654 (supports_overflow_infinity): Likewise.
15655 (is_negative_overflow_infinity): Likewise.
15656 (is_positive_overflow_infinity): Likewise.
15657 (is_overflow_infinity): Likewise.
15658 (stmt_overflow_infinity): Likewise.
15659 (overflow_infinity_range_p): Likewise.
15660 (usable_range_p): Remove as always returning true.
15661 (make_overflow_infinity): Remove.
15662 (negative_overflow_infinity): Likewise.
15663 (positive_overflow_infinity): Likewise.
15664 (avoid_overflow_infinity): Likewise.
15665 (set_value_range): Adjust accordingly.
15666 (set_value_range_to_nonnegative): Likewise, remove now unused
15667 overflow_infinity arg.
15668 (vrp_operand_equal_p): Adjust.
15669 (update_value_range): Likewise.
15670 (range_int_cst_singleton_p): Likewise.
15671 (operand_less_p): Likewise.
15672 (compare_values_warnv): Likewise.
15673 (extract_range_for_var_from_comparison_expr): Likewise.
15674 (vrp_int_const_binop): Likewise.
15675 (zero_nonzero_bits_from_vr): Likewise.
15676 (extract_range_from_multiplicative_op_1): Likewise.
15677 (extract_range_from_binary_expr_1): Likewise.
15678 (extract_range_from_unary_expr): Likewise.
15679 (extract_range_from_comparison): Likewise.
15680 (extract_range_basic): Likewise.
15681 (adjust_range_with_scev): Likewise.
15682 (compare_ranges): Likewise.
15683 (compare_range_with_value): Likewise.
15684 (dump_value_range): Likewise.
15685 (test_for_singularity): Likewise, remove strict_overflow_p parameter
15687 (simplify_cond_using_ranges): Adjust.
15689 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
15691 * brig-builtins.def: Added a builtin for class_f64.
15692 * builtin-types.def: Added a builtin type needed by class_f64.
15694 2017-05-03 Jason Merrill <jason@redhat.com>
15696 * timevar.def: Add TV_CONSTEXPR.
15698 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15700 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
15702 2017-05-03 Martin Jambor <mjambor@suse.cz>
15704 * ipa-prop.c (ipa_update_after_lto_read): Removed.
15705 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
15706 * ipa-cp.c (ipcp_propagate_stage): Do not call
15707 ipa_update_after_lto_read.
15708 * ipa-inline.c (ipa_inline): Likewise.
15710 2017-05-03 Martin Jambor <mjambor@suse.cz>
15712 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
15713 tag. Added a default constructor and a destructor.
15714 (ipa_edge_args_sum_t): New class;
15715 (ipa_edge_args_sum): Declare.
15716 (ipa_edge_args_vector): Remove declaration.
15717 (IPA_EDGE_REF): Use ipa_edge_args_sum.
15718 (ipa_free_edge_args_substructures): Remove declaration.
15719 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
15720 (ipa_edge_args_info_available_for_edge_p): Likewise.
15721 * ipa-prop.c (ipa_edge_args_vector): Removed.
15722 (edge_removal_hook_holder): Likewise.
15723 (edge_duplication_hook_holder): Likewise.
15724 (ipa_edge_args_sum): New variable.
15725 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
15726 ipa_edge_args_vector.
15727 (ipa_free_edge_args_substructures): Likewise.
15728 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
15729 ipa_edge_args_vector.
15730 (ipa_edge_removal_hook): Turned into method
15731 ipa_edge_args_sum_t::remove.
15732 (ipa_edge_duplication_hook): Turned into method
15733 ipa_edge_args_sum_t::duplicate.
15734 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
15735 registering edge hooks.
15736 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
15737 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
15738 ipa_edge_args_sum instead of ipa_edge_args_vector.
15739 * ipa-profile.c (ipa_profile): Likewise.
15741 2017-05-03 Martin Jambor <mjambor@suse.cz>
15743 * symbol-summary.h (function_summary): New method exists.
15744 (function_summary::symtab_removal): Deallocate through release.
15745 (call_summary): New class.
15746 (gt_ggc_mx): New overload.
15747 (gt_pch_nx): Likewise.
15748 (gt_pch_nx): Likewise.
15750 2017-05-03 Jeff Law <law@redhat.com>
15752 PR tree-optimization/78496
15753 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
15754 from simplify_cond_using_ranges. Split off code to walk
15755 backwards through casts into ...
15756 (simplify_cond_using_ranges_2): New function.
15757 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
15758 (execute_vrp): After identifying jump threads, call
15759 simplify_cond_using_ranges_2.
15761 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
15764 * ipa-inline.h (inline_summary): Add ctor.
15765 (create_ggc): Do not use ggc_cleared_alloc.
15767 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
15768 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15770 * gcc.c (handle_braces): Support escaping in switch matching
15772 * doc/invoke.texi (Spec Files): Document it.
15773 Remove superfluous @code markup in items.
15775 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15777 * diagnostic-show-locus.c (struct column_range): New struct.
15778 (get_affected_columns): New function.
15779 (get_printed_columns): New function.
15780 (struct correction): New struct.
15781 (correction::ensure_capacity): New function.
15782 (correction::ensure_terminated): New function.
15783 (struct line_corrections): New struct.
15784 (line_corrections::~line_corrections): New dtor.
15785 (line_corrections::add_hint): New function.
15786 (layout::print_trailing_fixits): Reimplement in terms of the new
15788 (selftest::test_overlapped_fixit_printing): New function.
15789 (selftest::diagnostic_show_locus_c_tests): Call it.
15791 2017-05-03 Nathan Sidwell <nathan@acm.org>
15793 Canonicalize canonical type hashing
15794 * tree.h (type_hash_canon_hash): Declare.
15795 * tree.c (type_hash_list, attribute_hash_list): Move into
15796 type_hash_canon_hash.
15797 (build_type_attribute_qual_variant): Break out hash code calc into
15798 type_hash_canon_hash.
15799 (type_hash_canon_hash): New. Generic type hash computation.
15800 (build_range_type_1, build_array_type_1, build_function_type,
15801 build_method_type_directly, build_offset_type, build_complex_type,
15802 make_vector_type): Call it.
15804 2017-05-03 Richard Biener <rguenther@suse.de>
15806 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
15807 When all DRs have unknown misaligned do not always peel
15808 when there is a store but apply the same costing model as if
15809 there were only loads.
15811 2017-05-03 Richard Biener <rguenther@suse.de>
15814 PR tree-optimization/80492
15815 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15816 compare_base_decls returning dont-know properly.
15818 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
15820 * config/arm/iterators.md (CCSI): New mode iterator.
15821 (arch): New mode attribute.
15822 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
15823 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
15824 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
15825 code iterator for success result mode.
15826 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
15827 the corresponding new insn generators.
15829 2017-05-03 Bin Cheng <bin.cheng@arm.com>
15832 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15833 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15835 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
15837 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
15838 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
15839 (DDR_A): Wrap DDR argument in brackets.
15840 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
15841 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
15842 (DDR_REVERSED_P): Likewise.
15844 2017-05-03 Jakub Jelinek <jakub@redhat.com>
15846 PR tree-optimization/79472
15847 * tree-switch-conversion.c (struct switch_conv_info): Add
15848 contiguous_range and default_case_nonstandard fields.
15849 (collect_switch_conv_info): Compute contiguous_range and
15850 default_case_nonstandard fields, don't clear final_bb if
15851 contiguous_range and only the default case doesn't have the required
15853 (check_all_empty_except_final): Set default_case_nonstandard instead
15854 of failing if contiguous_range and the default case doesn't have empty
15856 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
15857 and only the default case doesn't have the required constants. Skip
15859 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
15860 if default_case_nonstandard.
15861 (build_constructors): Build constant 1 just once. Assert that default
15862 values aren't inserted in between cases if contiguous_range. Skip
15864 (build_arrays): Skip virtual phis.
15865 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
15866 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
15867 Handle virtual phis.
15868 (gen_inbound_check): Handle default_case_nonstandard case.
15869 (process_switch): Adjust check_final_bb caller. Call
15870 gather_default_values with the first non-default case instead of
15871 default case if default_case_nonstandard.
15873 2017-05-02 Nathan Sidwell <nathan@acm.org>
15875 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
15876 check. Fix formatting.
15878 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
15880 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
15881 errors when comparing specialized and unspecialized times.
15883 2017-05-02 David Malcolm <dmalcolm@redhat.com>
15885 * diagnostic-show-locus.c
15886 (layout::should_print_annotation_line_p): Make private.
15887 (layout::print_annotation_line): Make private.
15888 (layout::annotation_line_showed_range_p): Make private.
15889 (layout::show_ruler): Make private.
15890 (layout::print_source_line): Make private. Pass in line and
15891 line_width, rather than calling location_get_source_line. Drop
15893 (layout::print_leading_fixits): New method.
15894 (layout::print_any_fixits): Rename to...
15895 (layout::print_trailing_fixits): ...this, and make private.
15896 Don't print newline fixits.
15897 (diagnostic_show_locus): Move logic for printing one row into...
15898 (layout::print_line): ...this new function. Move the
15899 location_get_source_line call and error-handling from
15900 print_source_line to here. Call print_leading_fixits, and rename
15901 print_any_fixits to print_trailing_fixits.
15902 (selftest::test_fixit_insert_containing_newline): Update now that
15903 newlines are partially supported.
15904 (selftest::test_fixit_insert_containing_newline_2): New test.
15905 (selftest::test_fixit_replace_containing_newline): Update comments.
15906 (selftest::diagnostic_show_locus_c_tests): Call the new test.
15907 * edit-context.c (class added_line): New class.
15908 (class edited_line): Describe newline handling in comment.
15909 (edited_line::actually_edited_p): New method.
15910 (edited_line::print_content): Delete redundant decl.
15911 (edited_line::m_predecessors): New field.
15912 (edited_file::print_content): Call edited_line::print_content.
15913 (edited_file::print_diff): Update to support newlines.
15914 (edited_file::print_diff_hunk): Likewise.
15915 (edited_file::print_run_of_changed_lines): New function.
15916 (edited_file::print_diff_line): Convert to...
15917 (print_diff_line): ...this.
15918 (edited_file::get_effective_line_count): New function.
15919 (edited_line::edited_line): Initialize new field m_predecessors.
15920 (edited_line::~edited_line): Clean up m_predecessors.
15921 (edited_line::apply_fixit): Handle newlines.
15922 (edited_line::get_effective_line_count): New function.
15923 (edited_line::print_content): New function.
15924 (edited_line::print_diff_lines): New function.
15925 (selftest::test_applying_fixits_insert_containing_newline): New
15927 (selftest::test_applying_fixits_replace_containing_newline): New
15929 (selftest::insert_line): New function.
15930 (selftest::test_applying_fixits_multiple_lines): Add example of
15932 (selftest::edit_context_c_tests): Call the new tests.
15934 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15936 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
15937 parameter cand. Update dump information.
15938 (get_computation_cost): Update uses.
15940 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15942 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
15943 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
15944 (get_computation_at, rewrite_use_address): Update use of
15945 get_computation_aff.
15947 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15949 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
15950 (get_computation): Delete.
15951 (get_computation_cost): Implement like get_computation_cost_at.
15952 Use get_computation_at.
15953 (get_computation_cost_at): Delete.
15954 (rewrite_use_nonlinear_expr): Use get_computation_at.
15955 (rewrite_use_compare, remove_unused_ivs): Ditto.
15957 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15959 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
15961 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15963 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
15964 (ivopts_global_cost_for_size): Rename parameter and update uses.
15965 (iv_ca_recount_cost): Update uses.
15966 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
15967 candidates seperately in n_invs and n_cands.
15968 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
15970 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15972 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
15973 (find_inv_vars_cb): New.
15974 (find_depends): Renamed to ...
15975 (find_inv_vars): ... this.
15976 (add_candidate_1, force_var_cost): Call find_inv_vars.
15977 (split_address_cost, determine_group_iv_cost_cond): Ditto.
15979 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15981 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
15982 inv_vars. Add inv_exprs.
15983 (struct iv_cand): Rename depends_on to inv_vars.
15984 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
15985 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
15986 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
15987 (dump_cand): Dump inv_vars.
15988 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
15989 (record_invariant, find_depends, add_candidate_1): Ditto.
15990 (set_group_iv_cost, force_var_cost): Ditto.
15991 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
15992 (get_computation_cost_at, get_computation_cost): Ditto.
15993 (determine_group_iv_cost_generic): Ditto.
15994 (determine_group_iv_cost_address): Ditto.
15995 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
15996 (determine_group_iv_costs): Ditto.
15997 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
15998 (iv_ca_set_remove_invariants): Renamed to ...
15999 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16000 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16001 (iv_ca_set_add_invariants): Renamed to ...
16002 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16003 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16004 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16005 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16006 (create_new_ivs): Remove useless dump.
16008 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16010 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16012 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16013 (iv_ca_set_no_cp, create_new_iv): Ditto.
16015 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16017 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16019 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16021 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16022 function tree_check2.
16024 2017-05-02 Martin Liska <mliska@suse.cz>
16026 * doc/gcov.texi: Add missing preposition.
16027 * gcov.c (function_info::function_info): Properly fill up
16028 all member variables.
16030 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16032 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16034 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16036 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16038 2017-05-02 Martin Liska <mliska@suse.cz>
16041 * lto-streamer-in.c (lto_read_tree_1): Remove
16042 LTO_STREAMER_DEBUG.
16043 * lto-streamer.c (struct tree_hash_entry): Likewise.
16044 (struct tree_entry_hasher): Likewise.
16045 (tree_entry_hasher::hash): Likewise.
16046 (tree_entry_hasher::equal): Likewise.
16047 (lto_streamer_init): Likewise.
16048 (lto_orig_address_map): Likewise.
16049 (lto_orig_address_get): Likewise.
16050 (lto_orig_address_remove): Likewise.
16051 * lto-streamer.h: Likewise.
16052 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16053 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16055 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16057 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16058 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16059 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16060 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16061 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16062 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16063 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16064 (mm_maskz_sub_ss): New intrinsics.
16065 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16066 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16067 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16068 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16069 (__builtin_ia32_subss_mask_round): New builtins.
16070 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16071 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16072 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16074 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16075 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16077 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16080 2017-05-02 Martin Jambor <mjambor@suse.cz>
16082 PR tree-optimization/78687
16083 * tree-sra.c (access): New field parent.
16084 (process_subtree_disqualification): New function.
16085 (disqualify_candidate): Call it.
16086 (build_accesses_from_assign): Reset write flag if creating an
16088 (build_access_subtree): Fill in parent field and also prpagate
16089 down grp_write flag.
16090 (create_artificial_child_access): New parameter set_grp_write, set
16091 grp_write to its value.
16092 (propagate_subaccesses_across_link): Also propagate grp_write flag
16094 (propagate_all_subaccesses): Push the closest parent back to work
16095 queue if add_access_to_work_queue returned true.
16097 2017-05-02 Richard Biener <rguenther@suse.de>
16099 * common.opt (fstrict-overflow): Alias negative to fwrapv.
16100 * doc/invoke.texi (fstrict-overflow): Remove all traces of
16101 -fstrict-overflow documentation.
16102 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
16103 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
16104 flag_strict_overflow.
16105 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
16106 * lto-opts.c (lto_write_options): Do not stream it.
16107 * lto-wrapper.c (merge_and_complain): Do not handle it.
16108 * opts.c (default_options_table): Do not set -fstrict-overflow.
16109 (finish_options): Likewise do not clear it when sanitizing.
16110 * simplify-rtx.c (simplify_const_relational_operation): Do not
16111 test flag_strict_overflow.
16113 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16115 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
16116 using enabled attribute.
16117 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
16118 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
16119 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
16120 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
16121 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
16122 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
16123 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
16124 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
16125 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
16126 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
16128 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16130 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
16132 2017-05-02 Richard Biener <rguenther@suse.de>
16134 PR tree-optimization/80591
16136 2017-04-10 Richard Biener <rguenther@suse.de>
16138 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16141 2017-05-02 Richard Biener <rguenther@suse.de>
16143 PR tree-optimization/80549
16144 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
16145 (cleanup_tree_cfg_noloop): Create forwarders to known loop
16146 headers if they do not have a preheader.
16148 2017-05-02 Martin Liska <mliska@suse.cz>
16151 * common.opt: Fix typo.
16152 * doc/invoke.texi: Likewise.
16154 2017-05-01 Jan Beulich <jbeulich@suse.com>
16156 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
16157 swapping, add (x,x,m,x,n) alternative.
16159 2017-05-01 Nathan Sidwell <nathan@acm.org>
16161 * calls.c (combine_pending_stack_adjustment_and_call): Remove
16162 unnecessary unadjusted_alignment check.
16164 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16167 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
16169 * gimplify.c (gimplify_cilk_detach): New function.
16170 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
16171 * tree-core.h: Document EXPR_CILK_SPAWN.
16172 * tree.h (EXPR_CILK_SPAWN): Define.
16174 2017-05-01 David Malcolm <dmalcolm@redhat.com>
16176 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
16177 to use new fixit_hint representation, using the "replace" logic.
16178 (get_line_span_for_fixit_hint): Likewise.
16179 (layout::print_any_fixits): Likewise.
16180 (selftest::test_one_liner_many_fixits): Rename to...
16181 (selftest::test_one_liner_many_fixits_1): ...this, and update
16182 comment and expected output to reflect that the multiple fix-it
16183 hints are now consolidated into one insertion.
16184 (selftest::test_one_liner_many_fixits_2): New test.
16185 (selftest::test_diagnostic_show_locus_one_liner): Update for
16187 (selftest::test_fixit_consolidation): Update for fix-it API
16189 * diagnostic.c (print_parseable_fixits): Likewise.
16190 * edit-context.c (edited_line::m_line_events): Convert from
16191 auto_vec <line_event *> to auto_vec <line_event>.
16192 (class line_event): Convert from abstract base class to a concrete
16193 class, taking over the role of replace_event.
16194 (class insert_event): Delete.
16195 (class replace_event): Rename to class line_event. Convert to
16197 (edit_context::add_fixits): Reimplement.
16198 (edit_context::apply_insert): Delete.
16199 (edit_context::apply_replace): Rename to...
16200 (edit_context::apply_fixit): ...this. Convert to half-open range.
16201 (edited_file::apply_insert): Delete.
16202 (edited_file::apply_replace): Rename to...
16203 (edited_file::apply_fixit): ...this.
16204 (edited_line::~edited_line): Drop deletion of events.
16205 (edited_line::apply_insert): Delete.
16206 (edited_line::apply_replace): Rename to...
16207 (edited_line::apply_fixit): ...this. Convert to half-open range.
16208 Update for change to type of m_line_events.
16209 * edit-context.h (edit_context::apply_insert): Delete.
16210 (edit_context::apply_replace): Rename to...
16211 (edit_context::apply_fixit): ...this.
16213 2017-05-01 Martin Sebor <msebor@redhat.com>
16215 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
16218 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
16221 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
16222 __get_cpuid_max returns 0.
16223 (__get_cpuid_count): Ditto.
16225 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
16227 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
16228 replacement expression is another instance of one of its arguments.
16230 2017-05-01 Jakub Jelinek <jakub@redhat.com>
16233 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
16234 check for stack push/pop autoinc.
16235 * config/i386/i386.c (ix86_agi_dependent): Return false
16236 if the only reason why modified_in_p returned true is that
16237 addr is SP based and set_insn is a push or pop.
16239 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16241 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
16244 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16247 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
16248 (account_size_time): Use two predicates - exec_pred and
16250 (evaluate_conditions_for_known_args): Compute both clause and
16252 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
16253 (inline_summary_t::duplicate): Update.
16254 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
16256 (compute_inline_parameters): Likewise.
16257 (estimate_edge_size_and_time): Update caluclation of time.
16258 (estimate_node_size_and_time): Compute both time and nonspecialized
16260 (estimate_ipcp_clone_size_and_time): Update.
16261 (inline_merge_summary): Update.
16262 (do_estimate_edge_time): Update.
16263 (do_estimate_edge_size): Update.
16264 (do_estimate_edge_hints): Update.
16265 (inline_read_section, inline_write_summary): Stream both new predicates.
16266 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
16268 (compute_inlined_call_time): Cleanup.
16269 (big_speedup_p): Update.
16270 (edge_badness): Update.
16271 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
16272 (size_time_entry): Replace predicate by exec_predicate and
16273 nonconst_predicate.
16274 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
16275 (estimate_edge_time): Return also nonspec_time.
16276 (reset_edge_growth_cache): Update.
16278 2017-04-29 Jakub Jelinek <jakub@redhat.com>
16280 PR rtl-optimization/80491
16281 * ifcvt.c (noce_process_if_block): When looking for x setter
16282 with missing else_bb, don't check only the insn right before
16283 cond_earliest, but look for the last insn that x is modified in
16284 within the same bb.
16286 PR rtl-optimization/80491
16287 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
16289 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
16291 PR tree-optimization/80487
16292 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
16294 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16296 PR tree-optimization/79697
16297 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
16298 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
16299 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
16301 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
16302 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
16304 2017-04-28 Martin Sebor <msebor@redhat.com>
16306 PR tree-optimization/80523
16307 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
16308 (init_target_to_host_charmap, target_to_host, target_strtol10): New
16310 (maybe_warn, format_directive, parse_directive): Use new functions.
16311 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
16313 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
16315 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
16317 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
16319 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
16320 target_header_dir): Set correctly.
16321 * configure: Regenerated.
16322 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16323 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16324 instead of SYSTEM_HEADER_DIR.
16326 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
16328 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
16329 (estimate_local_effects): Likewise.
16330 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
16331 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
16332 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
16333 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
16334 do_estimate_edge_time, estimate_edge_time): Likewise.
16335 * ipa-inline-analysis.c (estimate_node_size_and_time,
16336 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
16337 (estimate_time_after_inlining): Remove.
16339 2017-04-28 Martin Liska <mliska@suse.cz>
16341 * doc/gcov.texi: Enhance documentation of gcov.
16343 2017-04-28 Martin Liska <mliska@suse.cz>
16345 * doc/gcov.texi: Sort options in alphabetic order.
16346 * doc/gcov-dump.texi: Likewise.
16347 * doc/gcov-tool.texi: Likewise.
16348 * gcov.c (print_usage): Likewise.
16349 * gcov-dump.c (print_usage): Likewise.
16350 * gcov-tool.c (print_merge_usage_message): Likewise.
16351 (print_rewrite_usage_message): Likewise.
16352 (print_overlap_usage_message): Likewise.
16354 2017-04-28 Martin Liska <mliska@suse.cz>
16356 PR gcov-profile/53915
16357 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
16359 2017-04-28 Martin Liska <mliska@suse.cz>
16361 PR gcov-profile/79891
16362 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
16363 is marked by compiler as living on a line.
16364 (get_cycles_count): Remove usage of the union.
16365 (output_intermediate_file): Likewise.
16366 (find_source): Fix GNU coding style.
16367 (accumulate_line_counts): Remove old non-all block mode.
16368 (output_lines): Remove usage of the union.
16369 * profile.c (output_location): Include all BBs, even if
16370 belonging to a same line (and file) as a previous BB.
16372 2017-04-28 Martin Liska <mliska@suse.cz>
16374 * gcov.c (process_args): Handle new argument 'w'.
16375 (read_graph_file): Assign ID to BBs.
16376 (output_branch_count): Display BB # if verbose flag is set.
16377 (output_lines): Likewise for arcs.
16378 (print_usage): Add '--verbose' option help.
16379 * doc/gcov.texi: Document --verbose (-w) option.
16381 2017-04-28 Martin Liska <mliska@suse.cz>
16383 * gcov.c (struct block_location_info): New struct.
16384 (process_file): Fill up the new structure.
16385 (read_graph_file): Replace usage of encoding by the newly added
16387 (add_line_counts): Likewise.
16388 (accumulate_line_counts): Remove usage of the union.
16389 (function_info::function_info): New function.
16390 (function_info::~function_info): Likewise.
16391 (process_file): Call delete instead of release_function.
16392 (release_function): Release the function.
16393 (release_structures): Call delete instead of release_function.
16394 (solve_flow_graph): Replace usage of num_blocks.
16395 (find_exception_blocks): Likewise.
16396 (output_lines): Fix GNU coding style.
16398 2017-04-28 Martin Liska <mliska@suse.cz>
16401 * coverage.c (coverage_remove_note_file): New function.
16402 * coverage.h: Declare the function.
16403 * toplev.c (finalize): Clean if an error has been seen.
16405 2017-04-28 Martin Liska <mliska@suse.cz>
16407 PR gcov-profile/80031
16408 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
16409 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
16410 * gcov.c (read_graph_file): Read just number of blocks.
16411 * profile.c (branch_prob): Do not stream 0 flags per a basic
16414 2017-04-28 Martin Liska <mliska@suse.cz>
16416 * gcov-dump.c (tag_*): Add new argument to declarations.
16417 (dump_gcov_file): Likewise.
16418 (tag_blocks): Add and use new argument depth.
16419 (tag_arcs): Likewise.
16420 (tag_lines): Likewise.
16421 (tag_counters): Likewise.
16422 (tag_summary): Likewise.
16423 (dump_working_sets): Use depth to do a proper indentation.
16425 2017-04-28 Jakub Jelinek <jakub@redhat.com>
16428 * cgraph.h (symtab_node::debug_symtab): No longer inline.
16429 * symtab.c (symtab_node::debug_symtab): Move definition here.
16431 2017-04-28 Richard Biener <rguenther@suse.de>
16433 * lto-streamer.h (LTO_major_version): Bump to 7.
16435 2017-04-28 Richard Biener <rguenther@suse.de>
16437 * tree-vrp.c (assert_info): New struct.
16438 (add_assert_info): New helper.
16439 (register_edge_assert_for_2): Refactor to add asserts to a vector
16441 (register_edge_assert_for_1): Likewise.
16442 (register_edge_assert_for): Likewise.
16443 (finish_register_edge_assert_for): New helper actually registering
16444 asserts where live on edge.
16445 (find_conditional_asserts): Adjust.
16446 (find_switch_asserts): Likewise.
16447 (evrp_dom_walker::try_find_new_range): Generalize.
16448 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
16450 2017-04-27 Marek Polacek <polacek@redhat.com>
16453 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
16454 arg10 and arg11 to itype.
16456 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
16458 * doc/extend.texi (Object Size Checking): Improve grammar.
16460 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
16463 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
16464 that the logic for permitting reciprocal estimates matches that
16467 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16470 * tree.c (type_cache_hasher::equal): Only compare
16471 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
16472 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
16473 non-aggregate element types.
16474 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
16475 about the flag on ARRAY_TYPEs in the comment, formatting fix.
16477 2017-04-27 Richard Biener <rguenther@suse.de>
16479 PR middle-end/80533
16480 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
16481 stripping ARRAY_REFs from MEM_EXPR make sure we're not
16482 keeping a reference to a trailing array.
16484 2017-04-27 Richard Biener <rguenther@suse.de>
16486 PR middle-end/80539
16487 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
16488 being in loop-closed SSA form conservatively.
16489 (chrec_fold_multiply_poly_poly): Likewise.
16491 2017-04-27 Tamar Christina <tamar.christina@arm.com>
16493 PR middle-end/79665
16494 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
16495 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
16497 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16500 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
16501 (aarch64_function_arg_alignment): Return unsigned int again, but still
16502 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
16503 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16504 Don't emit -Wpsabi note.
16505 (aarch64_function_arg_boundary): Likewise.
16506 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16509 2017-04-26 Nathan Sidwell <nathan@acm.org>
16511 * tree.h (crc32_unsigned_n): Declare.
16512 (crc32_unsigned, crc32_unsigned): Make inline.
16513 * tree.c (crc32_unsigned_bits): Replace with ...
16514 (crc32_unsigned_n): ... this.
16515 (crc32_unsigned, crc32_byte): Remove.
16516 (crc32_string): Remove unnecessary braces.
16518 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16520 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
16521 * ipa-inline-analysis.c (MAX_TIME): Remove.
16522 (account_size_time): Use sreal for time.
16523 (dump_inline_summary): Update.
16524 (estimate_function_body_sizes): Update.
16525 (estimate_edge_size_and_time): Update.
16526 (estimate_calls_size_and_time): Update.
16527 (estimate_node_size_and_time): Update.
16528 (inline_merge_summary): Update.
16529 (inline_update_overall_summary): Update.
16530 (estimate_time_after_inlining): Update.
16531 (inline_read_section): Update.
16532 (inline_write_summary): Update.
16533 * ipa-inline.c (compute_uninlined_call_time): Update.
16534 (compute_inlined_call_time): Update.
16535 (recursive_inlining): Update.
16536 (inline_small_functions): Update.
16537 (dump_overall_stats): Update.
16538 * ipa-inline.h: Include sreal.h.
16539 (size_time_entry): Turn time to sreal.
16540 (inline_summary): Turn self_time nad time to sreal.
16542 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16544 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
16546 (sreal::stream_out, sreal::stream_in): New.
16547 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
16549 2017-04-25 Jakub Jelinek <jakub@redhat.com>
16551 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
16554 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
16557 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
16559 (dimode_scalar_chain::compute_convert_gain): Ditto.
16560 (dimode_scalar_chain::make_vector_copies): Ditto.
16561 (dimode_scalar_chain::convert_reg): Ditto.
16562 (dimode_scalar_chain::convert_insn): Ditto.
16563 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
16564 (VI248_AVX512BW_1): New mode iterator.
16565 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
16566 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
16569 2017-04-25 Martin Sebor <msebor@redhat.com>
16571 PR tree-optimization/80497
16572 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
16573 constants are representable in HOST_WIDE_INT.
16574 (parse_directive): Ditto.
16576 2017-04-25 Martin Sebor <msebor@redhat.com>
16579 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
16580 (new_zero_array): Adjust signature.
16581 (dom_info::dom_init): Used unsigned rather that size_t.
16582 (dom_info::dom_info): Same.
16584 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16585 Jakub Jelinek <jakub@redhat.com>
16588 * config/arm/arm.c: Include gimple.h.
16589 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16590 returns negative, increment ncrn only if it returned positive.
16591 (arm_needs_doubleword_align): Return int instead of bool,
16592 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
16593 members, but if there is any such non-FIELD_DECL
16594 > PARM_BOUNDARY aligned decl, return -1 instead of false.
16595 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16596 returns negative, increment nregs only if it returned positive.
16597 (arm_setup_incoming_varargs): Likewise.
16598 (arm_function_arg_boundary): Emit -Wpsabi note if
16599 arm_needs_doubleword_align returns negative, return
16600 DOUBLEWORD_ALIGNMENT only if it returned positive.
16602 2017-04-25 Marek Polacek <polacek@redhat.com>
16605 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16606 first argument to type.
16608 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
16611 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
16612 type checks to test for compatibility instead of equality.
16614 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16615 Jakub Jelinek <jakub@redhat.com>
16618 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
16620 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
16621 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
16622 the alignment computation, but return their maximum in warn_alignment.
16623 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16624 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
16626 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
16627 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16630 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16632 * config/arc/simdext.md (dmpyh): Fix typo.
16634 2017-04-25 Richard Biener <rguenther@suse.de>
16636 PR tree-optimization/80492
16637 * alias.c (compare_base_decls): Handle registers with asm
16638 specification conservatively.
16639 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16640 compare_base_decls returning dont-know properly.
16642 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16644 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
16645 (legitimate_offset_address_p): New function.
16646 (arc_legitimate_address_p): Use above function.
16648 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16650 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
16652 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16654 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
16655 ACCH registers whenever they are available.
16657 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16659 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
16660 double regs fix when not used.
16662 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16664 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
16666 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
16667 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
16669 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16671 * config/arc/arc.c (arc_output_addsi): Check for h-register class
16672 when emitting short ADD instructions.
16674 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16676 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
16678 (cmpsi_cc_c_insn): Likewise.
16679 (cbranchsi4_scratch): Compute proper instruction length using
16680 compact_hreg_operand.
16681 * config/arc/predicates.md (compact_hreg_operand): New predicate.
16683 2017-04-25 Richard Biener <rguenther@suse.de>
16685 PR middle-end/80509
16686 * passes.c (pass_manager::pass_manager): Initialize
16687 m_name_to_pass_map.
16689 2017-04-25 Richard Biener <rguenther@suse.de>
16691 PR tree-optimization/79201
16692 * tree-ssa-sink.c (statement_sink_location): Handle calls.
16694 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16697 * config/s390/vector.md: Split MEM->GPR vector moves for
16698 non-s_operand addresses.
16700 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16703 * config/s390/predicates.md (reload_const_wide_int_operand): New
16705 * config/s390/s390.md ("movti"): Remove d/P alternative.
16706 ("movti_bigconst"): New pattern definition.
16708 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16711 * s390-protos.h (s390_expand_cs_hqi): Removed.
16712 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
16713 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
16714 modes as well as CCZ1mode and CCZmode.
16715 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
16716 signature of s390_emit_compare_and_swap.
16717 (s390_expand_cs_hqi): Likewise, make static.
16718 (s390_expand_cs_tdsi): Generate an explicit compare before trying
16719 compare-and-swap, in some cases.
16720 (s390_expand_cs): Wrapper function.
16721 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
16723 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
16724 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
16725 patterns for small and large integers. Forbid symref memory operands.
16726 Move expander to s390.c. Require cc register.
16727 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
16728 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
16729 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
16730 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
16731 symref memory operands. Remove CC mode and call s390_match_ccmode
16733 ("atomic_exchange<mode>"): Allow and implement all integer modes.
16735 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16737 * config/s390/s390.md (define_peephole2): New peephole to help
16738 combining the load-and-test pattern with volatile memory.
16740 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16742 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
16743 with CCZmode for TARGET_Z196.
16745 2017-04-25 Jakub Jelinek <jakub@redhat.com>
16747 PR rtl-optimization/80501
16748 * combine.c (make_compound_operation_int): Set subreg_code to SET
16749 even for AND with mask of the sign bit of mode.
16751 PR rtl-optimization/80500
16752 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
16753 sum's initial value.
16755 2017-04-25 Julian Brown <julian@codesourcery.com>
16756 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16758 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
16760 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
16762 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
16764 2017-04-25 Julian Brown <julian@codesourcery.com>
16765 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16767 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
16768 (thunderx2t99_sha): New Reservation.
16770 2017-04-25 Julian Brown <julian@codesourcery.com>
16771 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16773 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
16774 type for 1-element load.
16776 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
16778 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
16780 2017-04-24 Martin Jambor <mjambor@suse.cz>
16782 PR tree-optimization/80293
16783 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
16784 char arrays not totally scalarizable if it is false.
16785 (analyze_all_variable_accesses): Pass correct value in the new
16786 parameter. Add a statistics counter.
16788 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
16790 PR middle-end/79931
16791 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
16793 2017-04-24 Richard Biener <rguenther@suse.de>
16795 PR tree-optimization/80494
16796 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
16797 out for complex types.
16799 2017-04-24 Richard Biener <rguenther@suse.de>
16801 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
16802 * tree-ssa-sccvn.c (print_scc): Print SCC size.
16803 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
16804 (DFS): Adjust and never fail.
16805 (sccvn_dom_walker::fail): Remove.
16806 (sccvn_dom_walker::before_dom_children): Adjust.
16807 (run_scc_vn): Likewise and never fail.
16808 * tree-ssa-pre.c (pass_pre::execute): Adjust.
16809 (pass_fre::execute): Likewise.
16811 2017-04-24 Richard Biener <rguenther@suse.de>
16813 PR tree-optimization/79725
16814 * tree-ssa-sink.c (statement_sink_location): Return whether
16815 failure reason was zero uses. Move that check later.
16816 (sink_code_in_bb): Deal with zero uses by removing the stmt
16819 2017-04-24 Richard Biener <rguenther@suse.de>
16822 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
16823 pointer-based references.
16825 2017-04-24 Richard Biener <rguenther@suse.de>
16828 * pass_manager.h (pass_manager::operator new): Remove.
16829 (pass_manager::operator delete): Likewise.
16830 * passes.c (pass_manager::operator new): Remove.
16831 (pass_manager::operator delete): Likewise.
16832 (pass_manager::pass_manager): Zero individual pass members.
16834 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
16837 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
16838 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
16839 Check "XEXP (src, 1)" operand here.
16840 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
16841 Check "XEXP (src, 1)" operand here.
16842 (dimode_scalar_chain::make_vector_copies): Detect count register
16843 of a shift instruction. Zero extend count register from QImode
16844 to DImode to satisfy vector shift pattern count operand predicate.
16845 Substitute vector shift count operand with a DImode copy.
16846 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
16849 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
16851 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
16852 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
16853 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
16854 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
16855 (UNSPEC_NOREX_MEM): Remove definition.
16857 2017-04-21 Richard Biener <rguenther@suse.de>
16859 PR tree-optimization/79547
16860 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
16861 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
16862 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
16863 without any constraints.
16865 2017-04-21 Richard Biener <rguenther@suse.de>
16867 PR tree-optimization/78847
16868 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
16870 2017-04-21 Richard Biener <rguenther@suse.de>
16872 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
16873 (build_distinct_type_copy): Likewise.
16874 (build_variant_type_copy): Likewise.
16875 * tree.c (build_qualified_type): Pass down mem-stat info.
16876 (build_distinct_type_copy): Likewise.
16877 (build_variant_type_copy): Likewise.
16879 2017-04-21 Richard Biener <rguenther@suse.de>
16881 PR tree-optimization/80237
16882 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
16884 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
16885 for a simplified result.
16887 2016-04-21 Richard Biener <rguenther@suse.de>
16889 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
16890 sth as strict as a simple_iv but a chrec without symbols and an
16891 operand defined in the loop we are peeling (and not some subloop).
16892 (propagate_constants_for_unrolling): Propagate all constants.
16894 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16897 * config/i386/i386.c (print_reg): Remove assert for disalowed
16898 regno values, call output_operand_lossage instead.
16900 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16903 * config/i386/constraints.md (Yc): New register constraint.
16904 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
16905 Use Yc constraint for alternative 2 of operand 0. Remove
16906 preferred_for_speed attribute.
16908 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
16910 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
16911 lastprivate clauses in SIMT case.
16913 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
16915 * doc/invoke.texi (-Wextra-semi): Document new warning option.
16917 2017-04-20 Richard Biener <rguenther@suse.de>
16919 PR tree-optimization/57796
16920 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
16921 as N scalar stores.
16922 (vect_model_load_cost): Cost gathers as N scalar loads.
16924 2017-04-20 Richard Biener <rguenther@suse.de>
16926 * ggc-page.c (ggc_allocated_p): Rename to ...
16927 (safe_lookup_page_table_entry): ... this and return the lookup
16929 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
16931 2017-04-20 Richard Biener <rguenther@suse.de>
16933 PR tree-optimization/80453
16934 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
16935 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
16936 from the conditions.
16937 (vn_phi_eq): Pass them down.
16938 (vn_phi_lookup): Record them.
16939 (vn_phi_insert): Likewise.
16941 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
16943 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
16944 uninitialized variable warning to avoid buffer overrun.
16946 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
16949 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
16950 is suppressed for '{ 0 }' in C.
16952 2017-04-20 Jakub Jelinek <jakub@redhat.com>
16954 * BASE-VER: Set to 8.0.0.
16956 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
16958 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
16959 priority .init_array and .fini_array section with SECTION_NOTYPE
16962 2017-04-20 Jakub Jelinek <jakub@redhat.com>
16964 PR middle-end/80423
16965 * tree.h (build_array_type): Add typeless_storage default argument.
16966 * tree.c (type_cache_hasher::equal): Also compare
16967 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
16968 (build_array_type): Add typeless_storage argument, set
16969 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
16971 (build_nonshared_array_type): Adjust build_array_type_1 caller.
16972 (build_array_type): Likewise. Add typeless_storage argument.
16974 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
16975 Jakub Jelinek <jakub@redhat.com>
16977 PR tree-optimization/80426
16978 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
16979 operation on symbolic operands, also compute the overflow for the
16980 invariant part when the operation degenerates into a negation.
16982 2017-04-19 Jakub Jelinek <jakub@redhat.com>
16985 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
16986 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
16989 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
16991 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
16994 * config/avr/avr.c (tree.h): Include it.
16995 (cgraph.h): Include it.
16996 (avr_encode_section_info): Don't warn for uninitialized progmem
16997 variable if it's just an alias.
16999 2017-04-19 Richard Biener <rguenther@suse.de>
17002 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17003 when needed by AutoPGO.
17005 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17008 * doc/lto.texi: Remove an extra 'that'.
17010 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17012 PR rtl-optimization/80429
17013 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17014 are only used in debug insns.
17016 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17017 Vladimir Makarov <vmakarov@redhat.com>
17019 * config/sparc/predicates.md (input_operand): Add comment. Return
17020 true for any memory operand when LRA is in progress.
17021 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17023 2017-04-18 Jeff Law <law@redhat.com>
17026 * mips.md ({return,simple_return}_internal): Do not overwrite
17029 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17031 PR tree-optimization/80443
17032 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17033 instead of adding 1, subtract -1 and similarly instead of subtracting
17036 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17038 PR rtl-optimization/80357
17039 * haifa-sched.c (tmp_bitmap): New variable.
17040 (model_recompute): Handle duplicate use records.
17041 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17042 (free_global_sched_pressure_data): Free it.
17044 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17047 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17048 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17049 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17050 instead of SYSTEM_HEADER_DIR.
17052 2017-04-18 Jeff Law <law@redhat.com>
17054 PR middle-end/80422
17055 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17056 predecessors after walking up the insn chain.
17058 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17061 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17062 sizetype type into debug info.
17064 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17067 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17068 unneeded test for TARGET_UPPER_REGS_SF.
17069 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17071 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17074 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17075 instead of gsi_after_labels.
17077 2017-04-18 Jeff Law <law@redhat.com>
17079 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17083 2017-04-13 Jeff Law <law@redhat.com>
17084 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17085 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17087 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17090 * config/avr/avr.c (intl.h): Include it.
17091 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17093 2017-04-18 Martin Liska <mliska@suse.cz>
17095 PR gcov-profile/78783
17096 * gcov-tool.c (gcov_output_files): Validate that destination
17097 file is either removed by the tool or by a user.
17099 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17100 Guy Benyei <guybe@mellanox.com>
17102 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
17103 block, and do not negate it, the stored id is already negative.
17105 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17107 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
17109 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17112 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
17113 masks of options that should be turned off if the VSX vector
17114 options are turned off.
17115 (OTHER_P8_VECTOR_MASKS): Likewise.
17116 (OTHER_VSX_VECTOR_MASKS): Likewise.
17117 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
17118 rs6000_disable_incompatible_switches to validate no type switches
17120 (rs6000_incompatible_switch): New function to disallow turning on
17121 other vector options if -mno-vsx, -mno-power8-vector, or
17122 -mno-power9-vector are specified.
17124 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17126 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
17128 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17130 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
17131 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
17132 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
17133 (ARG_POINTER_CFA_OFFSET): Likewise.
17135 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17137 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
17138 conditions to take advantage of various optimizations.
17140 2017-04-13 Jeff Law <law@redhat.com>
17142 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17143 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17144 (zero_extendsidi2_dext): Likewise.
17146 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17149 * fold-const.c (fold_ternary_loc): Revert
17150 use op0 instead of fold_convert_loc (loc, type, arg0) part of
17153 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
17155 PR rtl-optimization/80343
17156 * lra-remat.c (update_scratch_ops): Assign original hard reg to
17157 new scratch pseudo.
17159 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17162 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
17163 to ubsan_encode_value.
17165 2017-04-13 Jeff Law <law@redhat.com>
17167 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
17168 appearing in DEBUG_INSNs.
17170 2017-04-13 Martin Liska <mliska@suse.cz>
17172 PR gcov-profile/80413
17173 * gcov-io.c (gcov_write_string): Copy to buffer just when
17174 allocated size is greater than zero.
17176 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17179 * dwarf2out.c (decls_for_scope): Ignore declarations of
17180 current_function_decl in BLOCK_NONLOCALIZED_VARS.
17182 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
17185 * ipa-visibility.c (non_local_p): Fix typos.
17186 (localize_node): When localizing symbol in same comdat group,
17187 dissolve the group only when we know external symbols are going
17189 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
17191 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17193 PR tree-optimization/79390
17194 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
17195 order does not result in usable sequence, retry with reversed operand
17201 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
17202 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
17203 op0 instead of fold_convert_loc (loc, type, arg0).
17205 2017-04-12 Jeff Law <law@redhat.com>
17207 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
17208 has a delay slot in the generated code.
17210 * config/cris/cris.md (cris_preferred_reload_class): Return
17211 GENNONACR_REGS rather than GENERAL_REGS.
17213 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17216 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
17217 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
17218 signedness of the result type.
17220 2017-04-12 Richard Biener <rguenther@suse.de>
17221 Jeff Law <law@redhat.com>
17223 PR tree-optimization/80359
17224 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
17225 trim stores to TARGET_MEM_REFs.
17227 2017-04-12 Richard Biener <rguenther@suse.de>
17229 PR tree-optimization/79390
17230 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
17231 threading case even more.
17233 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17236 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
17237 for quad_address_p for TImode, instead of just not indexed_address.
17239 2017-04-12 Richard Biener <rguenther@suse.de>
17240 Bernd Edlinger <bernd.edlinger@hotmail.de>
17242 PR middle-end/79671
17243 * alias.c (component_uses_parent_alias_set_from): Handle
17244 TYPE_TYPELESS_STORAGE.
17245 (get_alias_set): Likewise.
17246 * tree-core.h (tree_type_common): Add typeless_storage flag.
17247 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
17248 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
17249 for types containing members with TYPE_TYPELESS_STORAGE.
17250 (place_field): Likewise.
17251 (layout_type): Likewise for ARRAY_TYPE.
17252 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
17253 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
17254 TYPE_TYPELESS_STORAGE.
17255 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
17257 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17260 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17261 first argument to type.
17263 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17267 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
17268 CONST0_RTX (mode) rather than const0_rtx where appropriate.
17269 (rs6000_expand_binop_builtin): Likewise.
17270 (rs6000_expand_ternop_builtin): Likewise; also add missing
17271 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
17272 vshasigma built-ins.
17273 * doc/extend.texi: Document that vec_xxpermdi's third argument
17274 must be a constant.
17276 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
17278 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
17279 Use shift_const cost parameter when calculating gain of STV shifts.
17281 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
17283 PR rtl-optimization/70478
17284 * lra-constraints.c (process_alt_operands): Check memory for
17285 disfavoring memory insn operand.
17287 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17289 PR middle-end/80100
17290 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
17291 left shift in unsigned HOST_WIDE_INT type.
17293 PR rtl-optimization/80385
17294 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
17295 (not (neg X)) into (plus X -1) for complex or non-integral modes.
17298 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
17299 if they have any depend clauses.
17301 2017-04-11 Martin Liska <mliska@suse.cz>
17304 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
17305 * ipa-split.c (split_function): Create a local comdat symbol
17306 if caller is in a comdat group.
17308 2017-04-11 Martin Liska <mliska@suse.cz>
17311 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
17314 2017-04-11 Martin Sebor <msebor@redhat.com>
17316 PR middle-end/80364
17317 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
17318 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
17320 (directive::set_width, directive::set_precision, format_character):
17322 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
17325 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
17328 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
17329 conflict, set target->arch_name instead of target->cpu_name.
17331 2017-04-11 Richard Biener <rguenther@suse.de>
17333 PR tree-optimization/80374
17334 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
17335 build_zero_cst, remove fold_convertible_p check again.
17337 2017-04-11 Martin Liska <mliska@suse.cz>
17340 * ubsan.c (instrument_object_size): Do not instrument register
17343 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17346 * config/i386/i386-builtin-types.def
17347 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
17348 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
17349 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
17350 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
17351 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
17352 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
17353 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
17354 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
17355 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
17356 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
17357 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
17358 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
17359 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
17360 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
17361 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
17362 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
17363 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
17364 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
17365 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
17366 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
17367 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
17368 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
17369 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
17370 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
17371 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
17372 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
17373 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
17374 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
17375 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
17376 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
17377 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
17378 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
17379 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
17380 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
17381 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
17382 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
17383 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
17384 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
17385 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
17386 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
17387 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
17388 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
17389 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
17390 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
17391 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
17392 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
17394 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
17395 flag to second_arg_count, handle 4 argument function type _COUNT
17396 aliases, handle second_arg_count on second argument rather than last.
17398 2017-04-10 Jeff Law <law@redhat.com>
17400 PR tree-optimization/80374
17401 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
17402 record anything if we can not convert integer_zero_node to the
17405 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17408 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17409 Enhance special handling given to the TARGET_P9_MINMAX option in
17410 relation to certain other options.
17412 2017-04-10 Bin Cheng <bin.cheng@arm.com>
17414 PR tree-optimization/80153
17415 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
17416 remove POINTER_PLUS_EXPR's base part directly, rather than through
17419 2017-04-10 Richard Biener <rguenther@suse.de>
17420 Bin Cheng <bin.cheng@arm.com>
17422 PR tree-optimization/80153
17423 * tree-affine.c (aff_combination_to_tree): Get base pointer from
17424 the first element of pointer type aff_tree. Build result expr in
17426 (add_elt_to_tree): Convert to type unconditionally. Remove other
17427 fold_convert calls.
17428 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
17429 (rewrite_use_nonlinear_expr): Check invariant using iv information.
17431 2017-04-10 Richard Biener <rguenther@suse.de>
17433 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17436 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
17438 PR rtl-optimization/70478
17439 * lra-constraints.c (curr_small_class_check): New.
17440 (update_and_check_small_class_inputs): New.
17441 (process_alt_operands): Update curr_small_class_check. Disfavor
17442 alternative insn memory operands. Check available regs for small
17445 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17448 * config/mips/mips.opt (-mvirt): Update description.
17449 * doc/invoke.texi (-mvirt): Likewise.
17451 2017-04-10 Richard Biener <rguenther@suse.de>
17453 PR middle-end/80362
17454 * fold-const.c (fold_binary_loc): Look at unstripped ops when
17455 looking for NEGATE_EXPR in -A / -B to A / B folding.
17457 2017-04-10 Martin Liska <mliska@suse.cz>
17459 PR gcov-profile/80224
17460 * gcov.c (print_usage): Fix usage string.
17461 (get_gcov_intermediate_filename): Remove.
17462 (output_gcov_file): Use both for normal and intermediate format.
17463 (generate_results): Do not initialize special file for
17464 intermediate format.
17466 2017-04-10 Richard Biener <rguenther@suse.de>
17468 PR tree-optimization/80304
17469 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
17472 2017-04-10 Nathan Sidwell <nathan@acm.org>
17475 * config/rs6000/rs6000.c (rs6000_vector_type): New.
17476 (rs6000_init_builtins): Use it.
17478 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17480 * config/arm/arm.md (<mrc>): Add mode to SET source.
17481 (<mrrc>): Likewise.
17483 2017-04-10 Richard Biener <rguenther@suse.de>
17485 PR middle-end/80344
17486 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
17488 2017-04-10 Jakub Jelinek <jakub@redhat.com>
17491 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
17492 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
17493 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
17494 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
17495 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
17496 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
17497 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
17498 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
17499 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
17500 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
17501 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
17502 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
17503 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
17504 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
17505 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
17506 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
17507 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
17508 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
17509 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
17510 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
17511 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
17512 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
17513 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
17515 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
17517 PR rtl-optimization/70478
17518 * lra-constraints.c: Reverse the last patch.
17520 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
17522 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
17523 Add comment for WCHAR_T.
17525 2017-04-08 Martin Liska <mliska@suse.cz>
17528 2017-04-07 Martin Liska <mliska@suse.cz>
17531 * ipa-split.c (split_function): Add function part to a same comdat
17534 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17537 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
17539 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
17541 * rs6000/rs6000.c (vec_load_pendulum): Rename...
17542 (vec_pairing): ...to this.
17543 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
17544 (rs6000_sched_init): Adjust for name change.
17545 (struct rs6000_sched_context): Likewise.
17546 (rs6000_init_sched_context): Likewise.
17547 (rs6000_set_sched_context): Likewise.
17549 2017-04-07 Jakub Jelinek <jakub@redhat.com>
17555 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
17557 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
17558 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
17559 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
17561 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
17563 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
17565 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17567 PR rtl-optimization/70703
17568 * ira-color.c (update_conflict_hard_regno_costs): Use
17569 int64_t instead of HOST_WIDE_INT.
17571 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17573 PR rtl-optimization/70478
17574 * lra-constraints.c (process_alt_operands): Disfavor alternative
17575 insn memory operands.
17577 2017-04-07 Jeff Law <law@redhat.com>
17579 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
17580 CALL and NOTE_INSN_CALL_ARG_LOCATION.
17582 2017-04-07 Martin Liska <mliska@suse.cz>
17585 * config/aarch64/aarch64.c (aarch64_process_target_attr):
17586 Show error message instead of an ICE.
17588 2017-04-07 Martin Liska <mliska@suse.cz>
17591 * ipa-split.c (split_function): Add function part to a same comdat
17594 2017-04-07 Richard Biener <rguenther@suse.de>
17596 PR middle-end/80341
17597 * tree.c (get_unwidened): Also handle ! for_type case for
17599 * convert.c (do_narrow): Split out from ...
17600 (convert_to_integer_1): ... here. Do not pass final truncation
17601 type to get_unwidened for TRUNC_DIV_EXPR.
17603 2017-04-07 Richard Biener <rguenther@suse.de>
17605 * tree-affine.c (wide_int_ext_for_comb): Take type rather
17607 (aff_combination_const): Adjust.
17608 (aff_combination_scale): Likewise.
17609 (aff_combination_add_elt): Likewise.
17610 (aff_combination_add_cst): Likewise.
17611 (aff_combination_convert): Likewise.
17612 (add_elt_to_tree): Likewise. Remove unused argument.
17613 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
17615 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
17617 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
17619 * config/arm/arm.c (arm_default_short_enums): Use
17620 ARM_DEFAULT_SHORT_ENUMS.
17621 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
17623 2017-04-06 Jakub Jelinek <jakub@redhat.com>
17626 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
17627 members with redundant out-of-class redeclaration.
17629 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17632 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
17633 * config/i386/i386.md (*zero_extendsidi2):
17634 Add (?*x,*x) and (?*v,*v) alternatives.
17636 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17639 * config/i386/i386.c (ix86_expand_builtin)
17640 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
17641 mode from insn data. Convert operands to insn operand mode.
17642 Copy operands that don't satisfy insn predicate to a register.
17644 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
17646 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
17649 2017-04-06 Richard Biener <rguenther@suse.de>
17651 PR tree-optimization/80334
17652 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
17653 preserve alignment of accesses.
17655 2017-04-06 Richard Biener <rguenther@suse.de>
17657 PR tree-optimization/80262
17658 * tree-sra.c (build_ref_for_offset): Preserve address-space
17660 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
17661 Drop useless address-space information on MEM_REF offsets.
17663 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
17665 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
17667 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17669 PR rtl-optimization/70703
17670 * ira-color.c (update_conflict_hard_regno_costs): Use
17671 HOST_WIDE_INT instead of long.
17673 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
17676 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
17677 not defined for x86_64 target. Add -mmmx target option when __SSE2__
17679 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
17680 for x86_64 target. Handle -m3dnowa option.
17682 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17684 PR rtl-optimization/70703
17685 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
17686 (update_conflict_hard_regno_costs): Use long instead of unsigned
17687 arithmetic for cost calculation.
17689 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17690 Bernd Edlinger <bernd.edlinger@hotmail.de>
17693 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
17696 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
17699 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
17700 ptr_mode with Pmode throughout.
17701 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
17702 into probe_stack_range and use DImode.
17704 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
17707 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
17708 call_eh_return is true.
17710 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17712 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
17713 Initialize last_match_fntype_index.
17715 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17718 * tree-nvr.c: Include internal-fn.h.
17719 (pass_return_slot::execute): Ignore internal calls without
17722 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17723 Richard Biener <rguenther@suse.de>
17726 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
17727 captures used multiple times, except for the last use.
17728 * generic-match-head.c: Include gimplify.h.
17730 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17732 PR tree-optimization/79390
17733 * target.h (struct noce_if_info): Declare.
17734 * targhooks.h (default_noce_conversion_profitable_p): Declare.
17735 * target.def (noce_conversion_profitable_p): New target hook.
17736 * ifcvt.h (struct noce_if_info): New type, moved from ...
17737 * ifcvt.c (struct noce_if_info): ... here.
17738 (noce_conversion_profitable_p): Renamed to ...
17739 (default_noce_conversion_profitable_p): ... this. No longer
17741 (noce_try_store_flag_constants, noce_try_addcc,
17742 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
17743 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
17744 instead of noce_conversion_profitable_p.
17745 * config/i386/i386.c: Include ifcvt.h.
17746 (ix86_option_override_internal): Don't override
17747 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
17748 (ix86_noce_conversion_profitable_p): New function.
17749 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
17750 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
17751 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
17752 * doc/tm.texi: Regenerated.
17754 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17756 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
17759 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17762 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
17763 instructions for small multiply cores.
17765 2017-04-04 Jeff Law <law@redhat.com>
17767 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
17769 (mips_expand_vec_perm_const): Initialize elements in orig_perm
17770 that are not set by the loop over the elements.
17772 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17775 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
17776 int mode, convert_modes it to mode as unsigned, otherwise use
17777 lowpart_subreg to mode rather than SImode.
17778 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
17779 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
17780 Use DImode instead of SImode for the shift count operand.
17781 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
17784 2017-04-04 Richard Biener <rguenther@suse.de>
17786 PR middle-end/80281
17787 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
17788 arithmetic done for the negate or the plus. Simplify.
17789 (A - (-B) -> A + B): Likewise.
17790 * fold-const.c (split_tree): Make sure to not negate pointers.
17792 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
17794 PR rtl-optimization/60818
17795 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
17796 a compare of comparisons with the thing compared if this results
17797 in a different machine mode.
17799 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
17801 * alias.c (base_alias_check): Fix typo in comment.
17802 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
17803 * cgraphunit.c (symbol_table::compile): Likewise.
17804 * collect2.c (maybe_run_lto_and_relink): Likewise.
17805 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
17806 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
17807 * config/avr/avr.c (avr_map_op_t): Likewise.
17808 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
17809 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
17810 * config/epiphany/epiphany.md (movcc): Likewise.
17811 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
17812 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
17814 * config/mips/mips.c (mips_save_restore_reg): Likewise.
17815 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
17816 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
17817 * config/sh/sh.c (sh_rtx_costs): Likewise.
17818 * fold-const.c (fold_truth_andor): Likewise.
17819 * genautomata.c (collapse_flag): Likewise.
17820 * gengtype.h (struct type::u::s): Likewise.
17821 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
17822 * input.c (FORMAT_AMOUNT): Likewise.
17823 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
17824 (known_aggs_to_agg_replacement_list): Likewise.
17825 * ipa-inline-analysis.c: Likewise.
17826 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
17827 * ipa-polymorphic-call.c
17828 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
17829 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
17830 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
17832 * modulo-sched.c (apply_reg_moves): Likewise.
17833 * omp-expand.c (build_omp_regions_1): Likewise.
17834 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
17835 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
17836 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
17837 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
17838 * value-prof.c: Likewise.
17839 * var-tracking.c (val_reset): Likewise.
17841 2017-04-03 Richard Biener <rguenther@suse.de>
17843 PR tree-optimization/80275
17844 * fold-const.c (split_address_to_core_and_offset): Handle
17847 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
17849 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
17850 descriptors is at least equal to that of functions.
17852 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
17854 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
17856 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
17859 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
17860 (mov<IMOD4:mode>): New expander.
17861 (*mov<IMOD4:mode>_internal): New insn and split pattern.
17863 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
17865 PR rtl-optimization/79405
17866 * fwprop.c (propagations_left): New variable.
17867 (forward_propagate_into): Decrement it.
17868 (fwprop_init): Initialize it.
17869 (fw_prop): If the variable has reached zero, stop propagating.
17870 (fwprop_addr): Ditto.
17872 2017-03-31 Jakub Jelinek <jakub@redhat.com>
17875 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
17876 a FUNCTION_DECL, pass it as decl instead of origin to
17879 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
17881 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
17884 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
17887 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
17888 TARGET_VSX_SMALL_INTEGER.
17890 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17892 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17893 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
17895 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17897 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
17898 extraction from odd-numbered MSA register.
17900 2017-03-31 Jakub Jelinek <jakub@redhat.com>
17902 PR middle-end/80173
17903 * expmed.c (store_bit_field_1): Don't attempt to create
17904 a word subreg out of hard registers wider than word if they
17905 have HARD_REGNO_NREGS of 1 for their mode.
17907 PR middle-end/80163
17908 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
17909 conversions to integer types wider than word and pointer.
17912 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
17913 (rtx_equal_for_cselib_p): Pass 0 to it.
17914 * cselib.c (cselib_hasher::equal): Likewise.
17915 (rtx_equal_for_cselib_1): Add depth argument. If depth
17916 is 128, don't look up VALUE locs and punt. Increment
17917 depth in recursive calls when walking VALUE locs.
17919 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
17921 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
17922 (make_gcov_file_name): Use the canonical path name for generating
17924 (read_line): Fix handling of files with ascii null bytes.
17926 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
17928 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
17929 to initialise a vector register instead
17930 of using a const_int.
17932 2017-03-30 Jakub Jelinek <jakub@redhat.com>
17934 PR translation/80189
17935 * gimplify.c (omp_default_clause): Use %qs instead of %s in
17936 diagnostic messages.
17938 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
17941 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
17942 (dfp_diex_<mode>): Update mode of operand 1.
17943 * doc/extend.texi (dxex, dxexq): Document change to return type.
17944 (diex, diexq): Document change to argument type.
17946 2017-03-30 Martin Jambor <mjambor@suse.cz>
17949 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
17950 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
17951 it reflects the signature changes performed at the callee side.
17952 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
17953 to cgraph_build_function_type_skip_args.
17954 (build_function_decl_skip_args): Adjust call to the above function.
17956 2017-03-30 Jakub Jelinek <jakub@redhat.com>
17959 * config/i386/sse.md
17960 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
17961 register as dest whenever it is a MEM not rtx_equal_p to the
17962 corresponding dup operand, and when forcing into reg move the
17963 reg into the memory afterwards.
17964 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
17965 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
17966 for the force_reg mode.
17967 (avx512vl_vextractf128<mode>): Use register as dest either
17968 always when a MEM, or when it is a MEM not rtx_equal_p to the
17969 corresponding dup operand, or even not when it is a CONST_VECTOR
17970 depending on the mode and lo vs. hi.
17971 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
17973 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
17974 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
17975 Likewise. Require that operands[2] is even.
17976 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
17977 Remove extraneous parens. Require that operands[2] is a multiple
17979 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
17980 operands[0] is a MEM if <mask_applied>, the predicates/constraints
17981 disallow memory then.
17983 2017-03-30 Richard Biener <rguenther@suse.de>
17985 PR tree-optimization/77498
17986 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
17987 to non-constants over backedges.
17989 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
17991 PR rtl-optimization/80233
17992 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
17993 as last_combined_insn. Do not test for BARRIER_P separately.
17995 2017-03-29 Andreas Schwab <schwab@suse.de>
17998 * calls.c (prepare_call_address): Convert funexp to Pmode before
17999 copying to temp reg.
18001 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18003 PR tree-optimization/80158
18004 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18005 Handle possible future case of more than one alternate
18007 (replace_rhs_if_not_dup): Likewise.
18008 (replace_one_candidate): Likewise.
18010 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18012 PR rtl-optimization/80193
18013 * ira.c (ira): Do not check allocation for LRA.
18015 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18017 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18018 (nvptx_output_simt_exit): Declare.
18019 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18020 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18021 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18022 (nvptx_declare_function_name): ...here. Emit declaration of local
18023 memory space buffer for omp_simt_enter insn.
18024 (nvptx_output_unisimt_switch): New.
18025 (nvptx_output_softstack_switch): New.
18026 (nvptx_output_simt_enter): New.
18027 (nvptx_output_simt_exit): New.
18028 * config/nvptx/nvptx.h (struct machine_function): New fields
18029 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18030 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18031 (UNSPECV_SIMT_EXIT): Ditto.
18032 (omp_simt_enter_insn): New insn.
18033 (omp_simt_enter): New expansion.
18034 (omp_simt_exit): New insn.
18035 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18037 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18038 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18039 (expand_GOMP_SIMT_EXIT): New.
18040 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18041 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18042 (GOMP_SIMT_EXIT): Ditto.
18043 * target-insns.def (omp_simt_enter): New insn.
18044 (omp_simt_exit): Ditto.
18045 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18047 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18048 (lower_rec_input_clauses): Likewise.
18049 (lower_lastprivate_clauses): Handle SIMT privatization.
18051 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18052 (ompdevlow_adjust_simt_enter): New.
18053 (find_simtpriv_var_op): New.
18054 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18055 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18057 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18058 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18059 (copy_decl_for_dup_finish): Ditto.
18061 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18063 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18066 * config/i386/i386.c (ix86_option_override_internal): Always
18067 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18069 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18071 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18073 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18075 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18076 mark new edge's irreducible flag accordign to it.
18077 (vect_do_peeling): Check loop preheader edge's irreducible flag
18078 and pass it to function slpeel_add_loop_guard.
18080 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18082 PR tree-optimization/80218
18083 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18084 Update block frequencies and counts.
18086 2017-03-28 Richard Biener <rguenther@suse.de>
18088 PR tree-optimization/78644
18089 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18090 of a simplification result we may not use it at all.
18092 2017-03-28 Richard Biener <rguenther@suse.de>
18095 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18096 without arguments, generate default definition of a SSA name.
18098 2017-03-28 Richard Biener <rguenther@suse.de>
18100 PR middle-end/80222
18101 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
18102 TYPE_REF_CAN_ALIAS_ALL references.
18103 * fold-const.c (fold_indirect_ref_1): Likewise.
18105 2017-03-28 Martin Liska <mliska@suse.cz>
18108 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
18109 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
18111 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18112 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
18114 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
18115 (EXTRA_SPECS): Define.
18116 (SUBTARGET_EXTRA_SPECS): Likewise.
18117 (SUBTARGET_CPP_SPEC): Likewise.
18118 * config/arc/elf.h (EXTRA_SPECS): Renamed to
18119 SUBTARGET_EXTRA_SPECS.
18120 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
18122 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18124 * config/arc/simdext.md (vst64_insn): Update pattern.
18125 (vld32wh_insn): Likewise.
18126 (vld32wl_insn): Likewise.
18127 (vld64_insn): Likewise.
18128 (vld32_insn): Likewise.
18130 2017-03-28 Marek Polacek <polacek@redhat.com>
18133 * fold-const.c (fold_comparison): Use protected_set_expr_location
18134 instead of SET_EXPR_LOCATION.
18136 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
18138 * tree.c (add_expr): Avoid name lookup warning.
18140 2017-03-27 Jeff Law <law@redhat.com>
18142 PR tree-optimization/80216
18143 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
18144 function name. Limit recursion depth.
18145 (record_temporary_equivalences): Corresponding changes.
18147 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
18149 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
18152 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18155 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
18157 * cfgcleanup.c (reg_note_cfa_p): New array.
18158 (insns_have_identical_cfa_notes): New function.
18159 (old_insns_match_p): Don't cross-jump in between /f
18160 and non-/f instructions. If both i1 and i2 are frame related,
18161 verify all CFA notes, their order and content.
18163 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18166 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
18167 HImode and SImode with zero extend to DImode to one insn.
18168 (bswap<mode>2_extenddi): Likewise.
18169 (bswapsi2_extenddi): Likewise.
18170 (bswaphi2_extendsi): Likewise.
18171 (bswaphi2): Combine bswap HImode and SImode into one insn.
18172 Separate memory insns from swapping register.
18173 (bswapsi2): Likewise.
18174 (bswap<mode>2): Likewise.
18175 (bswaphi2_internal): Delete, no longer used.
18176 (bswapsi2_internal): Likewise.
18177 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
18178 store, and gpr<-gpr swap insns.
18179 (bswap<mode>2_store): Likewise.
18180 (bswaphi2_reg): Register only splitter, combine with the splitter.
18181 (bswaphi2 splitter): Likewise.
18182 (bswapsi2_reg): Likewise.
18183 (bswapsi2 splitter): Likewise.
18184 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
18185 the insns into load, store, and register/register insns.
18186 (bswapdi2_ldbrx): Likewise.
18187 (bswapdi2_load): Likewise.
18188 (bswapdi2_store): Likewise.
18189 (bswapdi2_reg): Likewise.
18191 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
18193 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
18194 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
18196 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
18199 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
18201 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18202 special handling for target option conflicts between dform
18203 options (-mpower9-dform, -mpower9-dform-vector,
18204 -mpower9-dform-scalar) and -mno-direct-move.
18206 2017-03-27 Richard Biener <rguenther@suse.de>
18208 PR tree-optimization/80181
18209 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
18211 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18213 * config/arc/predicates.md (move_double_src_operand): Replace the
18214 call to move_double_src_operand with a call to address_operand.
18216 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18218 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
18219 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
18220 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
18222 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18224 * config/arc/predicates.md (long_immediate_loadstore_operand):
18225 Consider scaled addresses cases.
18227 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18229 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
18230 restored when in interrupt.
18231 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
18232 doesn't have delay slot.
18234 2017-03-27 Richard Biener <rguenther@suse.de>
18237 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
18238 inlined thunk clones.
18240 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18243 * asan.c (instrument_derefs): Copy over last operand from
18244 original COMPONENT_REF to the new COMPONENT_REF with
18245 DECL_BIT_FIELD_REPRESENTATIVE.
18246 * ubsan.c (instrument_object_size): Likewise.
18248 2017-03-27 Richard Biener <rguenther@suse.de>
18250 PR tree-optimization/80170
18251 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
18252 sure DR/SCEV didnt fold in constants we do not see when looking
18253 at the reference base alignment.
18255 2017-03-27 Richard Biener <rguenther@suse.de>
18257 PR middle-end/80171
18258 * gimple-fold.c (fold_ctor_reference): Properly guard against
18259 NULL return value from canonicalize_constructor_val.
18261 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
18264 * config/i386/i386.c (ix86_expand_builtin)
18265 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
18266 flags reg setting and flags reg using instructions.
18267 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
18268 clobbering instructions to zero extend op2.
18270 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
18272 * doc/install.texi (Configuration) <--with-aix-soname>:
18273 Update link to AIX ld.
18275 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
18277 PR rtl-optimization/80160
18278 PR rtl-optimization/80159
18279 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
18280 reg_alternate_class into account.
18282 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
18285 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
18286 to consider in curr_insn_transform.
18288 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18290 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
18291 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
18292 and emit_mode_inner.
18294 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18296 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
18297 argument to the overloaded builtin variants. Use the new flag to
18298 deprecate certain builtin variants.
18299 * config/s390/s390-builtin-types.def: Add new builtin types.
18300 * config/s390/s390-builtins.h: Support new flags field for
18301 overloaded builtins.
18302 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
18303 (s390_macro_to_expand): Enable vector float data type.
18304 (s390_cpu_cpp_builtins_internal): Indicate support of the new
18305 builtins by incrementing the __VEC__ version number.
18306 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
18308 (s390_resolve_overloaded_builtin): Emit error messages depending
18309 on the builtin flags.
18310 * config/s390/s390.c (s390_expand_builtin): Support additional
18311 flags argument. Change error message to match the messages
18312 emitted in s390-c.c.
18313 * config/s390/s390.md: New UNSPEC_* constants.
18314 (op_type): Add new instruction types.
18315 * config/s390/vecintrin.h: Add new builtins and test data class
18317 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
18318 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
18319 (VEC_INEXACT, VEC_NOINEXACT): New constants.
18320 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
18321 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
18322 ("vec_mergel<mode>"): V_HW -> VEC_HW.
18324 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
18325 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
18326 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
18327 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
18329 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
18330 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
18331 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
18332 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
18334 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
18335 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
18336 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
18337 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
18338 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
18339 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
18340 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
18342 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
18343 ("vec_scatter_element<V_HW_4:mode>_DI")
18344 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
18345 ("vec_fpint<mode>", "vflls")
18346 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
18347 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
18348 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
18349 ("*vec_cmphe<mode>_cc"): ... these.
18351 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
18352 mode constant instead of magic value.
18354 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18356 * config/s390/s390.c (s390_expand_vec_compare): Support other
18357 vector floating point modes than just V2DF.
18358 (s390_expand_vcond): Likewise.
18359 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
18360 (s390_cannot_change_mode_class): Prevent mode changes between TF
18361 and V1TF in vector registers.
18362 * config/s390/s390.md (DF, SF): New mode attributes.
18363 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
18364 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
18365 SFmode support for VRs.
18366 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
18368 (VFT, VF_HW): New mode iterators.
18369 (vw, sdx): New mode attributes.
18370 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
18371 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
18372 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
18373 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
18374 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
18375 also the new vector floating point modes. Renaming to ...
18377 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
18378 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
18379 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
18380 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
18381 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
18382 ("vec_unordered<mode>"): ... these.
18384 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
18385 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
18386 ("*vec_extendv2df"): New insn definitions.
18388 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18390 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
18391 ("mulditi3_2", "*muldi3_sign"): New patterns.
18392 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
18393 rename the pattern definition.
18395 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18397 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
18399 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
18401 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18403 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
18404 instruction if possible.
18405 * config/s390/vector.md (vec_halfnumelts): New mode
18407 ("*vec_vllezlf<mode>"): New pattern.
18409 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18411 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
18412 ("popcountv4si2", "popcountv2di2"): Rename to ...
18413 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
18414 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
18416 ("popcount<mode>2_vxe"): New pattern.
18418 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18420 * common/config/s390/s390-common.c (processor_flags_table): Add
18422 * config.gcc: Add arch12.
18423 * config/s390/driver-native.c (s390_host_detect_local_cpu):
18424 Default to arch12 for unknown CPU model numbers.
18425 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
18426 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
18427 PROCESSOR_max sanity check.
18428 * config/s390/s390-opts.h (enum processor_type): Add
18430 * config/s390/s390.c (processor_table): Add arch12.
18431 (s390_expand_builtin): Add check for B_VXE flag.
18432 (s390_issue_rate): Add PROCESSOR_ARCH12.
18433 (s390_get_sched_attrmask): Likewise.
18434 (s390_get_unit_mask): Likewise.
18435 (s390_sched_score): Enable z13 scheduling for arch12.
18436 (s390_sched_reorder): Likewise.
18437 (s390_sched_variable_issue): Likewise.
18438 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
18440 (s390_tune_attr): Use z13 scheduling also for arch12.
18441 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
18442 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
18443 (TARGET_VXE_P): New macros.
18444 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
18445 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
18446 * config/s390/s390.opt: Add arch12 as processor_type.
18448 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18450 * config/s390/s390.md
18451 ("fixuns_truncdddi2", "fixuns_trunctddi2")
18452 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
18453 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
18455 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
18456 Rename expanders to ...
18458 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
18459 ("fixuns_truncdddi2_emu"): ... these.
18461 ("fixuns_trunc<mode>si2_emu"): New expander.
18463 ("*fixuns_truncdfdi2_z13"): Rename to ...
18464 ("*fixuns_truncdfdi2_vx"): ... this.
18466 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18468 * config/s390/2964.md: Remove the single element vector compare
18469 instructions which are no longer used.
18470 * config/s390/s390.c (s390_select_ccmode): Remove handling of
18472 (s390_canonicalize_comparison): Remove handling of DFmode
18474 (s390_expand_vec_compare_scalar): Remove function.
18475 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
18476 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
18478 ("*cmp<mode>_ccs"): Add wfcdb instruction.
18480 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18482 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
18484 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
18485 will anyway by matched by mov<mode>_64dfp.
18487 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18489 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
18490 vlef/vstef. Add missing operand to vleif.
18492 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18494 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
18495 pair for all vector types with 64 bit elements.
18496 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
18497 * config/s390/vector.md (V_HW_64): ... here.
18498 (V_128_NOSINGLE): New mode iterator.
18499 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
18500 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
18501 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
18502 ("*vec_load_pairv2di"): Change to ...
18503 ("*vec_load_pair<mode>"): ... this one.
18505 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18507 * config/s390/constraints.md: Add comments.
18508 (jKK): Reject element sizes > 8 bytes.
18509 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
18511 * config/s390/s390.md: Add the s_operand checks formerly in
18512 s390_split_ok_p to various splitters where they are still
18514 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
18515 for 128 bit vectors. Plus two splitters.
18517 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18519 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
18522 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18525 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
18526 error if the boundary argument is not constant.
18528 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18530 PR rtl-optimization/80112
18531 * loop-doloop.c (doloop_condition_get): Don't check condition
18532 if cmp isn't SET with IF_THEN_ELSE src.
18534 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18536 PR tree-optimization/80158
18537 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
18538 replacing a candidate statement, also replace it for the
18539 candidate's alternate interpretation.
18540 (replace_rhs_if_not_dup): Likewise.
18541 (replace_one_candidate): Likewise.
18543 2017-03-24 Richard Biener <rguenther@suse.de>
18545 PR tree-optimization/80167
18546 * graphite-isl-ast-to-gimple.c
18547 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
18549 (translate_isl_ast_to_gimple::get_rename): Likewise.
18551 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
18553 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
18554 handling of certain combinations of target options, including the
18555 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
18556 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
18558 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18561 * config/arm/arm.md (*load_multiple): Add reload_completed to
18562 matching condition.
18564 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18565 Richard Biener <rguenth@suse.de>
18567 PR tree-optimization/79908
18568 PR tree-optimization/80136
18569 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18570 been cast away, gimplify_and_add suffices.
18572 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
18574 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
18576 2017-03-23 Richard Biener <rguenther@suse.de>
18578 PR tree-optimization/80032
18579 * gimplify.c (gimple_push_cleanup): Forced unconditional
18580 cleanups still have to go to the conditional_cleanups
18583 2017-03-22 Jakub Jelinek <jakub@redhat.com>
18585 PR tree-optimization/80072
18586 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
18588 (next_operand_entry_id): Change type to unsigned int.
18589 (sort_by_operand_rank): Make sure to return the right return value
18590 even if unsigned fields are bigger than INT_MAX.
18591 (struct oecount): Change cnt and id type to unsigned int.
18592 (oecount_hasher::equal): Formatting fix.
18593 (oecount_cmp): Make sure to return the right return value
18594 even if unsigned fields are bigger than INT_MAX.
18595 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
18598 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
18599 TREE_READONLY on result if writing it more than once.
18602 * doc/invoke.texi (-fsanitize=thread): Document that with
18603 -fnon-call-exceptions atomics are not able to throw
18607 * tsan.c: Include tree-eh.h.
18608 (instrument_builtin_call): Call maybe_clean_eh_stmt or
18609 maybe_clean_or_replace_eh_stmt where needed.
18610 (instrument_memory_accesses): Add cfg_changed argument.
18611 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
18612 if it returned true.
18613 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
18615 PR rtl-optimization/63191
18616 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
18617 wrapper function, moved the whole old content into ...
18618 (ix86_delegitimize_address_1): ... this. New inline function.
18619 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
18620 true as last argument instead of ix86_delegitimize_address.
18622 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18624 * config/aarch64/aarch64.c (generic_branch_cost): Copy
18625 cortexa57_branch_cost.
18627 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18629 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
18631 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18634 * doc/md.texi (Constraints): Document wA constraint.
18635 * config/rs6000/constraints.md (wA): New.
18636 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
18637 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
18638 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
18639 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
18641 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
18644 * gimplify.c (is_oacc_declared): New function.
18645 (oacc_default_clause): Use it to set default flags for acc declared
18646 variables inside parallel regions.
18647 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
18648 declared variables.
18649 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
18650 declare attribute to any decl as necessary.
18652 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18655 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
18656 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
18657 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
18658 (arm_arch_lpae): This.
18659 * config/arm/arm.c (arm_arch7ve): Rename into ...
18660 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
18661 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
18664 2017-03-22 Martin Liska <mliska@suse.cz>
18667 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
18668 error message instead of an ICE.
18670 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18672 * doc/extend.texi (6.11 Additional Floating Types): Revise.
18674 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
18676 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18678 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18681 2017-03-21 Martin Sebor <msebor@redhat.com>
18683 * doc/extend.texi: Use "cannot" instead of "can't."
18684 * doc/hostconfig.texi: Same.
18685 * doc/install.texi: Same.
18686 * doc/invoke.texi: Same.
18687 * doc/loop.texi: Same.
18688 * doc/md.texi: Same.
18689 * doc/objc.texi: Same.
18690 * doc/rtl.texi: Same.
18691 * doc/tm.texi: Same.
18692 * doc/tm.texi.in: Same.
18693 * doc/trouble.texi: Same.
18695 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
18698 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
18699 (collect_checksum_attributes): Set it.
18700 (die_checksum_ordered): Use it.
18702 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18704 PR tree-optimization/79908
18705 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
18706 change: For a VA_ARG whose LHS has been cast away, use
18707 force_gimple_operand to construct the side effects.
18709 2017-03-21 David Malcolm <dmalcolm@redhat.com>
18711 PR translation/80001
18712 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
18713 more amenable to translation.
18714 (oacc_loop_auto_partitions): Likewise.
18716 2017-03-21 Marek Polacek <polacek@redhat.com>
18717 Martin Sebor <msebor@redhat.com>
18719 PR tree-optimization/80109
18720 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
18721 on INTEGRAL_TYPE_P.
18723 2017-03-21 Jakub Jelinek <jakub@redhat.com>
18724 Segher Boessenkool <segher@kernel.crashing.org>
18727 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
18728 check reg_used_between_p between insn and one of succ or succ2
18729 depending on if succ is artificial insn not inserted into insn
18732 2017-03-21 Martin Liska <mliska@suse.cz>
18734 PR gcov-profile/80081
18735 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
18736 * doc/gcc.texi: Include gcov-dump stuff.
18737 * doc/gcov-dump.texi: New file.
18739 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
18741 PR rtl-optimization/79150
18742 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
18743 conditional jump, if the jump is the last insn of the loop.
18745 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18746 Richard Biener <rguenth@suse.de>
18748 PR tree-optimization/79908
18749 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18750 been cast away, use force_gimple_operand to construct the side
18753 2017-03-21 Martin Liska <mliska@suse.cz>
18755 PR libfortran/79956
18756 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
18759 2017-03-21 Brad Spengler <spender@grsecurity.net>
18762 * plugin.c (htab_hash_plugin): New function.
18763 (add_new_plugin): Use it and adjust.
18764 (parse_plugin_arg_opt): Adjust.
18765 (init_one_plugin): Likewise.
18767 2017-03-21 Richard Biener <rguenther@suse.de>
18769 PR tree-optimization/80032
18770 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
18771 if set force the cleanup to happen unconditionally.
18772 (gimplify_target_expr): Push inserted clobbers with force_uncond
18773 to avoid them being removed by control-dependent DCE.
18775 2017-03-21 Richard Biener <rguenther@suse.de>
18777 PR tree-optimization/80122
18778 * tree-inline.c (copy_bb): Do not expans va-arg packs or
18779 va_arg_pack_len when the inlined call stmt requires pack
18781 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
18783 2017-03-21 Jakub Jelinek <jakub@redhat.com>
18786 * tsan.c (instrument_builtin_call): If the memory model argument
18787 is not a constant, assume it is valid.
18790 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
18793 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
18795 PR rtl-optimization/79910
18796 * combine.c (can_combine_p): Do not allow combining an I0 or I1
18797 if its dest is used by an insn before I2 (other than the combined
18798 insns themselves, which are properly handled already).
18800 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
18803 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
18805 * combine.c (record_used_regs): New static function.
18806 (try_combine): Handle situations where there is an additional
18807 instruction between I2 and I3 which needs to have a LOG_LINK
18811 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
18813 * combine.c (try_combine): Delete redundant i1 test. Call
18814 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
18816 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18819 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
18820 alternatives 13/14.
18822 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18824 PR tree-optimization/80054
18825 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
18826 the optimization if a PHI or any of its arguments is not dominated
18827 by the candidate's basis. Use gphi* rather than gimple* as
18829 (replace_profitable_candidates): Clean up a gimple* variable that
18830 should be a gphi* variable.
18832 2017-03-20 Martin Sebor <msebor@redhat.com>
18835 * doc/extend.texi (attribute constructor): Document present limitation.
18837 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
18840 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
18841 __POWER9_VECTOR__ #ifdef control, change template definition to
18842 use Power9-specific built-in function.
18843 (vec_any_eq): Likewise.
18844 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
18845 to control outcomes from this test.
18846 (vector_ae_<mode>p): For VEC_F modes, likewise.
18848 2017-03-20 Ian Lance Taylor <iant@google.com>
18850 * config/i386/i386.c (ix86_function_regparm): Save an extra
18851 register for -fsplit-stack with DECL_STATIC_CHAIN.
18853 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
18856 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
18857 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
18859 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
18861 * config/riscv/riscv.c (riscv_print_operand): Use "fence
18863 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
18866 2017-03-20 Marek Polacek <polacek@redhat.com>
18869 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
18871 2017-03-20 Richard Biener <rguenther@suse.de>
18873 PR tree-optimization/80113
18874 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
18875 allocate extra SSA name for PHI def.
18876 (add_close_phis_to_outer_loops): Likewise.
18877 (add_close_phis_to_merge_points): Likewise.
18878 (copy_loop_close_phi_args): Likewise.
18879 (copy_cond_phi_nodes): Likewise.
18881 2017-03-20 Martin Liska <mliska@suse.cz>
18883 PR middle-end/79753
18884 * tree-chkp.c (chkp_build_returned_bound): Do not build
18885 returned bounds for a LHS that's not a BOUNDED_P type.
18887 2017-03-20 Martin Liska <mliska@suse.cz>
18891 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
18892 COMPLEX_CST and VECTOR_CST.
18894 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18897 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
18898 target operand. A new splitter adds the clobber statement in case
18899 the target operand is dead anyway.
18901 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
18903 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
18904 to age-old versions of binutils and glibc.
18906 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
18908 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
18910 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
18912 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
18914 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
18916 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
18917 requirement for binutils 2.13.
18919 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
18921 * combine.c (try_combine): Delete redundant i1 test. Call
18922 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
18924 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
18926 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
18927 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
18929 <riscv64-*-elf>: Re-arrange section
18930 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
18931 <riscv32-*-linux>: Likewise.
18932 <riscv64-*-elf>: Likewise
18933 <riscv64-*-linux>: Likewise.
18935 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
18938 * aarch64.opt(verbose-cost-dump): Fix typo.
18940 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
18943 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
18944 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
18946 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
18948 * reload.c (find_reloads): When reloading a nonoffsettable address,
18949 use RELOAD_OTHER for it and its address reloads.
18951 PR rtl-optimization/79910
18952 * combine.c (record_used_regs): New static function.
18953 (try_combine): Handle situations where there is an additional
18954 instruction between I2 and I3 which needs to have a LOG_LINK
18957 2017-03-17 Jeff Law <law@redhat.com>
18959 PR tree-optimization/71437
18960 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
18961 conditional in the hash table first.
18962 (vrp_dom_walker::before_dom_children): Extract condition from
18963 ASSERT_EXPR. Record condition, its inverion and any implied
18964 conditions as well.
18966 2017-03-17 Marek Polacek <polacek@redhat.com>
18967 Markus Trippelsdorf <markus@trippelsdorf.de>
18969 PR tree-optimization/80079
18970 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
18973 2017-03-17 Richard Biener <rguenther@suse.de>
18975 PR middle-end/80075
18976 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
18977 Properly verify the LHS before the RHS possibly claims to be
18979 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
18982 2017-03-17 Martin Jambor <mjambor@suse.cz>
18984 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
18985 (List of -O2 options): Likewise.
18986 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
18989 2017-03-17 Tom de Vries <tom@codesourcery.com>
18991 * gcov-dump.c (print_usage): Print bug_report_url.
18993 2017-03-17 Richard Biener <rguenther@suse.de>
18995 PR middle-end/80050
18996 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
18997 (parser::peek): Likewise.
18999 2017-03-17 Richard Biener <rguenther@suse.de>
19001 PR tree-optimization/80048
19002 * sese.c (free_sese_info): Properly release rename_map and
19003 copied_bb_map elements.
19005 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19007 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19008 Add linked-list forward and backlinks. Insert on
19009 construction, remove on destruction.
19010 (class pass_store_merging): Add m_stores_head field.
19011 (pass_store_merging::terminate_and_process_all_chains):
19012 Iterate over m_stores_head list.
19013 (pass_store_merging::terminate_all_aliasing_chains):
19015 (pass_store_merging::execute): Check for debug stmts first.
19016 Push new chains onto the m_stores_head stack.
19018 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19021 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19022 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19023 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19025 2017-03-16 Jeff Law <law@redhat.com>
19027 PR tree-optimization/71437
19028 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19029 member function. Implementation moved into after_dom_children
19030 member function and into the threader's thread_outgoing_edges
19032 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19033 some code into new thread_outgoing_edges.
19034 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19035 definition. Simplify marker handling (do it here). Assume we always
19036 have the available expression and the const/copies tables.
19037 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19039 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19040 * tree-vrp.c (equiv_stack): No longer file scoped.
19041 (vrp_dom_walker): New class.
19042 (vrp_dom_walker::before_dom_children): New member function.
19043 (vrp_dom_walker::after_dom_children): Likewise.
19044 (identify_jump_threads): Setup domwalker. Use it rather than
19045 walking edges in a random order by hand. Simplify setup/finalization.
19046 (finalize_jump_threads): Remove.
19047 (vrp_finalize): Do not call identify_jump_threads here.
19048 (execute_vrp): Do it here instead and call thread_through_all_blocks
19051 PR tree-optimization/71437
19052 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19054 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19056 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19057 (dom_opt_dom_walker::thread_across_edge): Remove
19058 handle_dominating_asserts argument. All callers changed.
19059 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19060 changes. Remove calls to lhs_of_dominating_assert. Other
19061 uses of handle_dominating_asserts turn into unconditional code
19062 (simplify_control_stmt_condition_1): Likewise.
19063 (simplify_control_stmt_condition): Likewise.
19064 (thread_through_normal_block, thread_across_edge): Likewise.
19065 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19066 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19067 object if it is not an SSA_NAME.
19068 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19069 before calling into the VRP specific simplifiers.
19070 (identify_jump_threads): Remove handle_dominating_asserts
19073 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19076 * tree-diagnostic.c (default_tree_printer): No longer static.
19077 * tree-diagnostic.h (default_tree_printer): New prototype.
19079 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19081 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19082 Change ins into fmov.
19084 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19086 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19087 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19088 Use h_con constraint for operand 1.
19089 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19090 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19092 2017-03-15 Jeff Law <law@redhat.com>
19094 PR tree-optimization/71437
19095 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19096 (record_temporary_equivalences): Use it.
19098 PR tree-optimization/71437
19099 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
19100 tree-ssa-scopedtables.
19101 (lookup_avail_expr, build_and_record_new_cond): Likewise.
19102 (record_conditions, record_cond, vuse_eq): Likewise.
19103 (record_edge_info): Adjust to API tweak of record_conditions.
19104 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
19105 (record_temporary_equivalences, optimize_stmt): Likewise.
19106 (eliminate_redundant_computations): Likewise.
19107 (record_equivalences_from_stmt): Likewise.
19108 * tree-ssa-scopedtables.c: Include options.h and params.h.
19109 (vuse_eq): New function, moved from tree-ssa-dom.c
19110 (build_and_record_new_cond): Likewise.
19111 (record_conditions): Likewise. Accept vector of conditions rather
19112 than edge_equivalence structure for first argument.
19113 for the first argument.
19114 (avail_exprs_stack::lookup_avail_expr): New member function, moved
19115 from tree-ssa-dom.c.
19116 (avail_exprs_stack::record_cond): Likewise.
19117 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
19118 from tree-ssa-dom.c.
19119 (avail_exprs_stack): Add new member functions lookup_avail_expr
19121 (record_conditions): Declare.
19123 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
19126 * lra-constraints.c (process_alt_operands): Increase reject for
19127 reloading an input/output operand.
19129 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
19132 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
19133 insns to convert from signed/unsigned char/short to IEEE 128-bit
19135 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
19137 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
19140 * config/i386/i386.c (ix86_vector_duplicate_value): Create
19141 subreg of inner mode for values already in registers.
19143 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
19145 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
19146 iteration reg is used after the loop.
19148 2017-03-14 Martin Sebor <msebor@redhat.com>
19150 PR tree-optimization/79800
19151 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
19152 precision in negative-positive range.
19153 (format_floating): Call non-const overload with adjusted precision.
19155 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19158 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
19161 2017-03-14 Martin Sebor <msebor@redhat.com>
19163 PR middle-end/80020
19164 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
19165 * builtins.def (aligned_alloc): Use it.
19168 * Makefile.in (GTFILES): Add calls.c.
19169 * calls.c: Include "gt-calls.h".
19171 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
19173 PR rtl-optimization/79728
19174 * regs.h (struct target_regs): New field
19175 x_contains_allocatable_regs_of_mode.
19176 (contains_allocatable_regs_of_mode): New macro.
19177 * reginfo.c (init_reg_sets_1): Initialize it, and change
19178 contains_reg_of_mode so it includes global regs as well.
19179 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
19180 rather than contains_regs_of_mode.
19182 2017-03-14 Martin Liska <mliska@suse.cz>
19184 * doc/invoke.texi: Document options that can't be combined with
19185 -fcheck-pointer-bounds.
19187 2017-03-14 Martin Liska <mliska@suse.cz>
19189 PR middle-end/79831
19190 * doc/invoke.texi (-Wchkp): Document the option.
19192 2017-03-14 Martin Liska <mliska@suse.cz>
19194 * Makefile.in: Install gcov-dump.
19196 2017-03-14 Martin Liska <mliska@suse.cz>
19198 * multiple_target.c (expand_target_clones): Bail out for
19199 an invalid attribute.
19201 2017-03-14 Richard Biener <rguenther@suse.de>
19203 * alias.c (struct alias_set_entry): Pack properly.
19204 * cfgloop.h (struct loop): Likewise.
19205 * cse.c (struct set): Likewise.
19206 * ipa-utils.c (struct searchc_env): Likewise.
19207 * loop-invariant.c (struct invariant): Likewise.
19208 * lra-remat.c (struct cand): Likewise.
19209 * recog.c (struct change_t): Likewise.
19210 * rtl.h (struct address_info): Likewise.
19211 * symbol-summary.h (function_summary): Likewise.
19212 * tree-loop-distribution.c (struct partition): Likewise.
19213 * tree-object-size.c (struct object_size_info): Likewise.
19214 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
19215 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
19216 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
19217 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
19218 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19219 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
19220 (struct _stmt_vec_info): Likewise.
19222 2017-03-14 Martin Liska <mliska@suse.cz>
19225 * multiple_target.c (create_dispatcher_calls): Check that
19226 a target can create a function dispatcher.
19228 2017-03-14 Martin Liska <mliska@suse.cz>
19231 * multiple_target.c (expand_target_clones): Drop local.local
19232 flag for default implementation.
19234 2017-03-14 Richard Biener <rguenther@suse.de>
19236 PR tree-optimization/80030
19237 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
19239 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
19241 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
19242 gcc_fallthrough() instead of __attribute__((fallthrough));
19244 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19246 * doc/gcc.texi: Remove "up" link to (DIR).
19247 * doc/gccint.texi: Ditto.
19249 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19251 * doc/install.texi (Specific) <avr>: Remove reference to
19254 2017-03-13 Jeff Law <law@redhat.com>
19256 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
19257 attribute rather than comments.
19259 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
19260 match_scratch operand is highest.
19262 2017-03-13 Martin Liska <mliska@suse.cz>
19264 PR middle-end/78339
19265 * ipa-pure-const.c (warn_function_noreturn): If the declarations
19266 is a CHKP clone, use original declaration.
19268 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19270 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
19271 (arc_conditional_register_usage): Use a different allocation order
19272 when optimizing for size.
19273 * common/config/arc/arc-common.c (arc_option_optimization_table):
19274 Section anchors default on when optimizing for size.
19276 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19278 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
19280 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19282 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
19283 * config/arc/arc.md (cpu_facility): Add cd variant.
19284 (*movqi_insn): Add code density variant.
19285 (*movhi_insn): Likewise.
19286 (*movqi_insn): Likewise.
19287 (*addsi3_mixed): Likewise.
19288 (subsi3_insn): Likewise.
19290 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19292 * config/arc/arc.md (movsi_cond_exec): Update constraint.
19294 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19296 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
19297 expressions with MINUS and UNARY ops.
19299 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19302 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
19304 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
19305 between vec_select and vector argument.
19306 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
19307 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
19308 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
19309 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
19310 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
19311 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
19313 2017-03-13 Richard Biener <rguenther@suse.de>
19316 * params.def (vect-max-peeling-for-alignment): Fix typo.
19318 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19320 * doc/install.texi (Specific) <mips-*-*>: Remove description of
19321 issue that only occurred with binutils below 2.18.
19323 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19325 * doc/install.texi (Specific) <cris-axis-elf>: No longer
19326 refer to binutils 2.11/2.12 minimum.
19328 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19330 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
19331 ftp.kernel.org and simplify binutils requirement.
19333 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
19335 * doc/invoke.texi (Warning Options): Fix spelling of link-time
19337 (Optimize Options): Ditto. Also remove redundancy.
19339 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19341 PR translation/79848
19342 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
19344 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
19345 to G_ to avoid double translation.
19347 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19349 PR translation/79923
19350 * auto-profile.c (get_combined_location): Convert leading
19351 character of diagnostics to lower case and remove trailing period.
19352 (read_profile): Likewise for various diagnostics.
19353 * config/arm/arm.c (arm_option_override): Remove trailing period
19354 from various diagnostics.
19355 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
19356 (msp430_expand_delay_cycles): Likewise.
19358 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19361 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
19362 full command-line argument, rather than just "str".
19363 (aarch64_validate_march): Likewise.
19364 (aarch64_validate_mtune): Likewise.
19366 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
19368 PR rtl-optimization/78911
19369 * lra-assigns.c (must_not_spill_p): New function.
19370 (spill_for): Use it.
19372 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19374 PR tree-optimization/79981
19375 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
19376 ATOMIC_COMPARE_EXCHANGE ifn result.
19377 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
19378 IFN_ATOMIC_COMPARE_EXCHANGE.
19380 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19383 * opts.c (parse_sanitizer_options): Add missing question mark to
19384 "did you mean" message.
19386 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19388 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
19390 (VMULEUH_UNS): Likewise.
19391 (VMULOUB_UNS): Likewise.
19392 (VMULOUH_UNS): Likewise.
19393 * config/rs6000/rs6000.c (builtin_function_type): Remove
19394 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
19396 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19399 * read-rtl-function.c (function_reader::read_rtx_operand): Update
19400 x with result of extra_parsing_for_operand_code_0.
19401 (function_reader::extra_parsing_for_operand_code_0): Convert
19402 return type from void to rtx, returning x. When reading
19403 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
19404 larger size containing struct block_symbol.
19406 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
19408 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
19409 -mfloat128-hardware without -m64.
19411 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
19414 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
19415 entries to the case statement that marks unsigned arguments to
19416 overloaded functions.
19418 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19420 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
19421 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
19423 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
19426 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
19427 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
19429 2017-03-10 Martin Liska <mliska@suse.cz>
19433 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
19434 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
19437 2017-03-10 Olivier Hainque <hainque@adacore.com>
19439 * tree-switch-conversion (array_value_type): Start by resetting
19440 candidate type to it's main variant.
19442 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19444 PR rtl-optimization/79909
19445 * combine.c (try_combine): Use simplify_replace_rtx on individual
19446 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
19447 of the whole CALL_INSN_FUNCTION_USAGE.
19449 PR tree-optimization/79972
19450 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
19451 get_range_info on SSA_NAMEs. Formatting fixes.
19453 2017-03-10 Richard Biener <rguenther@suse.de>
19454 Jakub Jelinek <jakub@redhat.com>
19456 PR tree-optimization/77975
19457 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
19458 edge to be constant.
19459 (get_val_for): For constant x return it. Formatting fix.
19460 (loop_niter_by_eval): Avoid pointless looping if the next iteration
19461 would use the same bases as the current one.
19463 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19465 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
19466 instead of vec_select for V1TImode.
19467 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
19469 (VSX_LE_128): Add V1TI to this mode iterator.
19470 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
19471 (*vsx_le_perm_store_<mode>): Likewise.
19472 (pre-reload splitter for VSX stores): Likewise.
19473 (post-reload splitter for VSX stores): Likewise.
19474 (*vsx_xxpermdi2_le_<mode>): Likewise.
19475 (*vsx_lxvd2x2_le_<mode>): Likewise.
19476 (*vsx_stxvd2x2_le_<mode>): Likewise.
19478 2017-03-09 Michael Eager <eager@eagercon.com>
19480 Correct failures with --enable-checking=yes,rtl.
19482 * config/microblaze/microblaze.c (microblaze_expand_shift):
19483 Replace GET_CODE test with CONST_INT_P and INTVAL test with
19484 test for const0_rtx.
19485 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
19486 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
19488 2017-03-09 Richard Biener <rguenther@suse.de>
19490 PR tree-optimization/79977
19491 * graphite-scop-detection.c (scop_detection::merge_sese):
19492 Handle the case of extra exits to blocks dominating the entry.
19494 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
19496 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19499 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
19501 PR rtl-optimization/79949
19502 * lra-constraints.c (process_alt_operands): Check memory when
19503 trying to predict a cycle. Print about the overall increase.
19505 2017-03-09 Richard Biener <rguenther@suse.de>
19507 PR middle-end/79971
19508 * gimple-expr.c (useless_type_conversion_p): Preserve
19509 TYPE_SATURATING for fixed-point types.
19511 2017-03-09 Richard Biener <rguenther@suse.de>
19514 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
19515 alignment of BLKmode params.
19517 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19520 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
19521 (VALL_NO_V2Q): Likewise.
19523 * config/aarch64/aarch64-simd.md
19524 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
19526 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
19527 VALL_NO_V2Q mode iterator.
19528 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
19530 2017-03-09 Martin Liska <mliska@suse.cz>
19532 PR tree-optimization/79631
19533 * tree-chkp-opt.c (chkp_is_constant_addr): Call
19534 tree_int_cst_sign_bit just for INTEGER constants.
19536 2017-03-09 Martin Liska <mliska@suse.cz>
19540 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
19543 2017-03-09 Marek Polacek <polacek@redhat.com>
19546 * tree.c (inchash::add_expr): Handle TREE_VEC.
19548 2017-03-09 Martin Liska <mliska@suse.cz>
19551 (chkp_narrow_size_and_offset): New function.
19552 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
19553 (void chkp_parse_bit_field_ref): New function.
19554 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
19555 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
19557 2017-03-09 Martin Liska <mliska@suse.cz>
19560 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
19561 (chkp_find_bounds_1): Remove gcc_unreachable.
19563 2017-03-09 Jakub Jelinek <jakub@redhat.com>
19566 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
19567 BUILT_IN_SYNC*, determine the access type from the size suffix and
19568 always build a MEM_REF with that type. Handle forgotten
19569 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
19572 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
19573 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
19574 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
19575 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
19576 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
19577 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
19578 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
19579 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
19580 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
19581 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
19582 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
19583 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
19584 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
19585 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
19586 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
19587 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
19588 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
19589 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
19590 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
19591 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
19592 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
19593 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
19594 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
19595 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
19596 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
19597 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
19598 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
19599 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
19600 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
19601 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
19602 definitions outside of __OPTIMIZE__ guarded section.
19605 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
19606 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
19607 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
19608 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
19611 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19613 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
19614 ("vfenez<mode>"): Add missing constraints.
19616 2017-03-08 Martin Sebor <msebor@redhat.com>
19619 * config/nds32/nds32.c (nds32_option_override):
19620 Fix misspelled diagnostic.
19622 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19625 * gimplify.c (gimplify_omp_for): Replace index var in outer
19626 taskloop statement with an artificial variable and add
19627 OMP_CLAUSE_PRIVATE clause for it.
19629 2017-03-08 Richard Biener <rguenther@suse.de>
19631 PR tree-optimization/79955
19632 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
19633 for accesses that are completely outside of the variable.
19635 2017-03-08 Andrew Haley <aph@redhat.com>
19637 PR tree-optimization/79943
19638 * tree-ssa-loop-split.c (compute_new_first_bound): When
19639 calculating the new upper bound, (END-BEG) should be added, not
19642 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19644 * config/avr/avr.md (setmemhi): Make sure match_dup
19645 operand number comes before match_scratch.
19647 2017-03-08 Richard Biener <rguenther@suse.de>
19649 PR tree-optimization/79920
19650 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
19651 with ncopies == 1 to ...
19652 (vect_transform_slp_perm_load): ... here. Properly compute
19653 all element loads by iterating VF times over the group. Do
19654 not handle ncopies (computed in a broken way) in
19655 vect_create_mask_and_perm.
19657 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19660 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
19661 is a uniform vector, use uniform_vector_p return value instead of
19662 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
19664 2017-03-07 Marek Polacek <polacek@redhat.com>
19666 PR middle-end/79809
19667 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
19668 (alloca_call_type): Likewise.
19670 2017-03-07 Martin Liska <mliska@suse.cz>
19672 * gcov.c (process_args): Put comment to correct location.
19674 2017-03-07 Martin Liska <mliska@suse.cz>
19676 PR middle-end/68270
19677 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
19678 Use array_at_struct_end_p instead of DECL_CHAIN (field).
19679 (chkp_narrow_bounds_for_field): Likewise.
19680 (chkp_parse_array_and_component_ref): Pass one more argument to
19683 2017-03-07 Richard Biener <rguenther@suse.de>
19685 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
19688 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
19690 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
19691 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
19693 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19696 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
19697 to end of description.
19698 (PARAM_MAX_STORES_TO_MERGE): Likewise.
19700 2017-03-07 Jakub Jelinek <jakub@redhat.com>
19702 PR rtl-optimization/79901
19703 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
19705 (*avx512f_<code><mode>3<mask_name>): ... this.
19706 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
19707 iterator instead of VI8_AVX2_AVX512BW.
19709 PR rtl-optimization/79901
19710 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
19711 min/max expander, expand it using expand_vec_cond_expr.
19714 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
19717 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19720 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
19721 to void * for PCH reasons.
19722 * dwarf2out.c (output_loc_operands, output_die): Cast
19723 v.val_vec.array to unsigned char *.
19725 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
19728 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
19731 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
19733 PR rtl-optimization/79571
19734 * lra-constraints.c (process_alt_operands): Calculate static
19735 reject and subtract it from overall when only addresses will be
19738 2017-03-06 Julia Koval <julia.koval@intel.com>
19741 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
19742 incoming stack boundary to 128 for 64-bit targets.
19744 2017-03-06 Richard Biener <rguenther@suse.de>
19746 PR tree-optimization/79894
19747 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
19748 to NULL after folding it.
19750 2017-03-06 Richard Biener <rguenther@suse.de>
19752 PR tree-optimization/79824
19753 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
19754 check disabling peeling for gaps.
19756 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
19758 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19759 attributes): Document gettimeofday.
19761 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
19763 * config/s390/s390.c (s390_option_override_internal): Set
19764 PARAM_MIN_VECT_LOOP_BOUND
19766 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
19768 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
19769 * config/s390/s390.md: Likewise.
19771 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19774 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
19775 (<avx2_avx512>_perm<mode>): Rename to ...
19776 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
19778 (<avx512>_perm<mode>_mask): Rename to ...
19779 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
19781 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
19782 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
19783 instead of VI8F_256_512.
19784 (avx512f_perm<mode>): New define_expand.
19785 (avx512f_perm<mode>_mask): Likewise.
19786 (avx512f_perm<mode>_1<mask_name>): New define_insn.
19787 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
19789 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19791 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
19792 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
19794 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
19796 2017-03-06 Martin Liska <mliska@suse.cz>
19799 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
19800 when having a SSA NAME w/o VAR_DECL assigned to it.
19802 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19804 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
19805 msa_dpsub_<su>_d): Fix MODE for vec_select.
19807 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19809 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
19811 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
19813 2017-03-06 Richard Biener <rguenther@suse.de>
19815 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
19816 * plugin.c (register_plugin_info): Likewise.
19817 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
19819 2017-03-05 Jakub Jelinek <jakub@redhat.com>
19821 * config/i386/sse.md (sse_storehps, sse_storelps,
19822 avx_<castmode><avxsizesuffix>_<castmode>,
19823 avx512f_<castmode><avxsizesuffix>_<castmode>,
19824 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
19825 in condition that at least one operand is not a MEM.
19827 2017-03-03 Jakub Jelinek <jakub@redhat.com>
19829 PR middle-end/79805
19830 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
19831 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
19833 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
19834 gimple_call_nothrow_p flag based on whether original builtin can throw.
19835 If it can, emit following stmts on the fallthrough edge.
19836 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
19837 don't create new bb if inserting just debug stmts on the edge, try to
19838 insert them on the fallthru bb or just reset debug stmts.
19840 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
19843 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
19844 restore recog_data (including the operand rtxes inside it) around
19845 the call to get_insn_template.
19847 2017-03-03 Martin Sebor <msebor@redhat.com>
19849 PR tree-optimization/79699
19850 * context.c (context::~context): Free MPFR caches to avoid
19851 a memory leak on program exit.
19853 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19855 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
19856 Use wide_int::ulow () instead of .elt (0).
19858 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
19860 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
19861 (*pushxf): Limit oF constraint to 32bit targets and add oC
19862 constraint for 64bit targets.
19863 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
19864 (*pushdf): Change rmF constraint to rmC.
19866 2017-03-03 Martin Liska <mliska@suse.cz>
19868 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
19869 Remove unused variable.
19871 2017-03-03 Jakub Jelinek <jakub@redhat.com>
19874 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
19875 is a memory operand, increase num_memory.
19876 (ix86_expand_args_builtin): Likewise.
19878 2017-03-03 Jan Hubicka <jh@suse.cz>
19881 * ipa-devirt.c (maybe_record_node): Properly handle
19882 __cxa_pure_virtual visibility.
19884 2017-03-03 Martin Liska <mliska@suse.cz>
19886 PR tree-optimization/79803
19887 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
19889 (pass_loop_prefetch::execute): Disabled optimization if an
19890 assumption about L1 cache size is not met.
19892 2017-03-03 Martin Liska <mliska@suse.cz>
19894 PR rtl-optimization/79574
19895 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
19896 (hash_scan_set): Likewise.
19897 (dump_hash_table): Likewise.
19898 (hoist_code): Likewise.
19900 2017-03-03 Richard Biener <rguenther@suse.de>
19902 * fixed-value.c (fixed_from_string): Restore use of elt (1)
19903 in place of uhigh ().
19904 (fixed_convert_from_real): Likewise.
19906 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
19909 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
19911 2017-03-03 Richard Biener <rguenther@suse.de>
19913 PR middle-end/79818
19914 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
19915 TYPE_OVERFLOW_UNDEFINED check.
19917 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19919 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
19921 (vector_ae_<mode>_p): Likewise.
19922 (vector_nez_<mode>_p): Likewise.
19923 (vector_ne_v2di_p): Likewise.
19924 (vector_ae_v2di_p): Likewise.
19925 (vector_ne_<mode>_p): Likewise.
19926 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
19928 (vsx_tsqrt<mode>2_fe): Likewise.
19930 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
19933 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
19935 2017-03-02 Jakub Jelinek <jakub@redhat.com>
19937 PR rtl-optimization/79780
19938 * cprop.c (one_cprop_pass): When second and further conditional trap
19939 in a single basic block is turned into an unconditional trap, turn it
19940 into a deleted note to avoid RTL verification failures.
19942 2017-03-02 Richard Biener <rguenther@suse.de>
19944 * fold-const.c (const_binop): Use ulow () instead of elt (0).
19946 2017-03-02 Richard Biener <rguenther@suse.de>
19948 PR tree-optimization/79345
19950 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
19951 param and abort the walk, returning -1 if it is hit.
19952 (walk_aliased_vdefs): Take a limit param and pass it on.
19953 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
19954 defaulting to 0 and return a signed int.
19955 * tree-ssa-uninit.c (struct check_defs_data): New struct.
19956 (check_defs): New helper.
19957 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
19958 about uninitialized memory.
19959 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
19960 bogus uninitialized warning.
19961 (fixed_convert_from_real): Likewise.
19963 2017-03-02 Bin Cheng <bin.cheng@arm.com>
19965 PR tree-optimization/66768
19966 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
19967 iv_use if base object can't be determined.
19969 2017-03-02 Jakub Jelinek <jakub@redhat.com>
19971 PR tree-optimization/79345
19972 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
19973 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
19974 (get_pattern_stats): Initialize it.
19975 * genemit.c (gen_expand): Verify match_scratch numbers come after
19976 match_operand/match_dup numbers.
19977 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
19978 match_scratch numbers.
19979 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
19981 * config/s390/s390.md (trunctdsd2): Likewise.
19983 2017-03-02 Richard Biener <rguenther@suse.de>
19985 * wide-int.h (wide_int_storage::operator=): Implement in terms
19988 2017-03-02 Richard Biener <rguenther@suse.de>
19990 PR tree-optimization/79777
19991 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
19992 the to insert expression to sth existing.
19994 2017-03-01 Martin Sebor <msebor@redhat.com>
19996 PR middle-end/79692
19997 * gimple-ssa-sprintf.c
19998 (directive::known_width_and_precision): New function.
19999 (format_integer): Use it.
20000 (get_mpfr_format_length): Consider the full range of precision
20001 when computing %g output with the # flag. Set the likely byte
20002 count to 3 rather than 1 when precision is indeterminate.
20003 (format_floating): Correct the lower bound of precision.
20005 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20007 * doc/invoke.texi: Document default code model for 64-bit Linux.
20009 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20012 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20013 udiv rather than div since input pattern is unsigned.
20015 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20017 * config/i386/i386.c (print_reg): Warn for values of
20018 unsupported size in integer register.
20020 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20023 * config/rs6000/predicates.md (current_file_function_operand): Do
20024 not allow self calls to be local if the function is replaceable.
20026 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20029 * config/rs6000/altivec.h (vec_ctz and others): Change the
20030 preprocessor macro that controls conditional compilation from
20031 _ARCH_PWR9 to __POWER9_VECTOR__.
20032 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20033 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20034 control (instead of _ARCH_PWR9 control) so that template
20035 definition uses power9-specific function.
20036 (vec_any_eq): Likewise.
20037 (vec_all_ne): Change macro definition to use a power9-specific
20038 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20039 _ARCH_PWR9 control).
20040 (vec_any_eq) Likewise.
20041 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20042 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20043 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20044 support for xvcmpnedp instruction.
20045 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20046 macro expansion so that Power9 implementation of vec_all_ne does
20047 not use the AltiVec predicate framework.
20048 (VCMPNEH_P): Likewise.
20049 (VCMPNEW_P): Likewise.
20050 (VCMPNED_P): Likewise.
20051 (VCMPNEFP_P): Likewise.
20052 (VCMPNEDP_P): Likewise.
20053 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20054 implementation of vec_any_eq to not use AltiVec predicate
20056 (VCMPAEH_P): Likewise.
20057 (VCMPAEW_P): Likewise.
20058 (VCMPAED_P): Likewise.
20059 (VCMPAEFP_P): Likewise.
20060 (VCMPAEDP_P): Likewise.
20061 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20062 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20063 not use the AltiVec predicate framework.
20064 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20065 of vec_any_eq to not use AltiVec predicate framework.
20066 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20067 support for predefined __POWER9_VECTOR__ macro to indicate that
20068 Power9 instruction selection is enabled.
20069 (altivec_overloaded_builtins): Remove extraneous
20070 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20071 function argument types RS6000_BTI_bool_V16QI and
20072 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20073 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20074 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20075 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20076 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20077 Power9 for implementations of vec_cmpne. Change the signature for
20078 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20079 (representing vec_all_ne) to remove the previously described first
20080 argument of type RS6000_BTI_INTSI, as this was an artifact of
20081 reliance on the AltiVec predicate framework, which is no longer
20082 used in the implementation of these functions. Add
20083 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20084 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20085 since, unlike the AltiVec predicate framework implementation, we
20086 do not share function descriptors between vec_alle and vec_anyeq.
20087 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20088 set of modes that receive special treatment even when
20089 TARGET_P9_VECTOR is true. The special treatment emits code that
20090 does not depend on Power9 instructions.
20091 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20092 define_expand to not rely on AltiVec predicate framework.
20093 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20095 (vector_ne_v2di_p): Change this define_expand to not rely on
20096 AltiVec predicate framework.
20097 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20099 (vector_ne_<mode>_p): Change this define_expand to not rely on
20100 AltiVec predicate framework.
20101 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20103 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
20104 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
20105 define_insn pattern.
20106 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
20107 define_insn pattern because the xvcmpne<VSs>. instruction is not
20109 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
20110 instruction is not supported.
20112 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20114 * config/nvptx/nvptx.c: Include intl.h.
20116 2017-03-01 Martin Jambor <mjambor@suse.cz>
20119 * ipa-prop.h (ipa_bits): Removed field known.
20120 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
20121 to pointers. Adjusted their comments to warn about their sharing.
20122 (ipcp_transformation_summary): Change bits to a vector of pointers.
20123 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
20124 (ipa_get_ipa_bits_for_value): Declare.
20125 * tree-vrp.h (value_range): Mark as GTY((for_user)).
20126 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
20127 (ipa_bits_hash_table): Likewise.
20128 (ipa_vr_ggc_hash_traits): Likewise.
20129 (ipa_vr_hash_table): Likewise.
20130 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
20131 being pointers and vr_known being removed.
20132 (ipa_set_jf_unknown): Likewise.
20133 (ipa_get_ipa_bits_for_value): New function.
20134 (ipa_set_jfunc_bits): Likewise.
20135 (ipa_get_value_range): New overloaded functions.
20136 (ipa_set_jfunc_vr): Likewise.
20137 (ipa_compute_jump_functions_for_edge): Use the above functions to
20138 construct bits and vr parts of jump functions.
20139 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
20140 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20142 (ipcp_grow_transformations_if_necessary): Also allocate
20143 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20145 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
20146 them. Fix too long lines.
20147 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
20148 vr_known being removed.
20149 (ipa_read_jump_function): Use new setter functions to construct bits
20150 and vr parts of jump functions or set them to NULL.
20151 (write_ipcp_transformation_info): Adjust for bits being pointers.
20152 (read_ipcp_transformation_info): Likewise.
20153 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
20155 Include gt-ipa-prop.h.
20156 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
20158 (ipcp_store_bits_results): Likewise.
20159 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
20160 Do not write to existing jump functions but use a temporary instead.
20162 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20165 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
20166 attempt to use its first operand as BIT_FIELD_REF base.
20168 2017-03-01 Richard Biener <rguenther@suse.de>
20170 PR middle-end/79721
20171 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
20172 interpolating formula in wrapping arithmetic.
20173 (chrec_apply): Convert chrec_evaluate return value to wanted type.
20175 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20177 PR tree-optimization/79734
20178 * tree-vect-generic.c (expand_vector_condition): Optimize
20179 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
20180 Handle VEC_COND_EXPR where comparison has different inner width from
20181 type's inner width.
20183 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
20185 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
20186 markup, and similar issues. Remove @opindex entries for things
20187 that aren't options. Add missing -mmpy-option entries.
20189 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20191 PR tree-optimization/79737
20192 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
20193 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
20194 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
20195 instead of byte_size. Formatting fix.
20196 (shift_bytes_in_array_right): Formatting fix.
20198 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
20201 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
20202 condition on optimize for the leaf function test.
20204 2017-02-28 Martin Liska <mliska@suse.cz>
20207 * read-rtl-function.c (function_reader::handle_unknown_directive):
20208 Bail out when one uses -flto.
20210 2017-02-28 Martin Liska <mliska@suse.cz>
20212 * common.opt: Replace space with tabular for options of <number>
20214 * config/i386/i386.opt: Show <number> value for
20215 -mlarge-data-threshold.
20216 * opts.c (print_filtered_help): Do not display number in hexadecimal
20219 2017-02-28 Martin Liska <mliska@suse.cz>
20221 * common.opt: Fix --help=option -Q for options which are of
20224 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
20226 * config/i386/i386.c (print_reg): Error out for values
20227 of 8-bit size in invalid integer register.
20229 2017-02-28 Martin Sebor <msebor@redhat.com>
20231 PR tree-optimization/79691
20232 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
20234 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20237 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
20238 gcc_unreachable with output_operand_lossage.
20240 2017-02-28 Richard Biener <rguenther@suse.de>
20242 PR tree-optimization/79740
20243 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
20245 (visit_nary_op): Insert the nary into the hashtable if we
20246 pattern-matched sth.
20247 * tree-ssa-pre.c (eliminate_insert): Robustify.
20249 2017-02-28 Richard Biener <rguenther@suse.de>
20251 PR middle-end/79731
20252 * fold-const.c (decode_field_reference): Reject out-of-bound
20255 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20257 * config/i386/i386.c: Include intl.h.
20258 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
20259 instead of just cond ? "..." : "...".
20260 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
20261 * coverage.c (read_counts_file): Likewise.
20262 * omp-offload.c: Include intl.h.
20263 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
20264 of just cond ? "..." : "...".
20265 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
20266 of just cond ? "..." : "...".
20268 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
20271 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
20273 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
20274 'tune for' CPU name.
20275 * config/arm/arm-cpu-data.h: Regenerated.
20277 2017-02-28 Richard Biener <rguenther@suse.de>
20279 PR tree-optimization/79732
20280 * tree-inline.c (expand_call_inline): Do not shadow var.
20282 2017-02-28 Richard Biener <rguenther@suse.de>
20284 PR tree-optimization/79723
20285 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
20286 address-space properly.
20288 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
20290 * doc/optinfo.texi (Optimization groups): Fix option used for
20292 * doc/invoke.texi (-fopt-info): Document "omp".
20293 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
20294 (OPTGROUP_ALL): Add OPTGROUP_OMP.
20295 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
20296 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
20297 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
20299 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
20301 * dumpfile.c (optgroup_options): Instead of "openmp", associate
20302 OPTGROUP_OMP with "omp".
20304 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
20307 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
20308 for arithmetic shift of unsigned V2DI.
20310 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
20312 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
20313 arc/linux.h headers.
20314 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
20315 (LINK_SPEC): Likewise.
20316 (ARC_TLS_EXTRA_START_SPEC): Likewise.
20317 (EXTRA_SPECS): Likewise.
20318 (STARTFILE_SPEC): Likewise.
20319 (ENDFILE_SPEC): Likewise.
20320 (LIB_SPEC): Likewise.
20321 (TARGET_SDATA_DEFAULT): Likewise.
20322 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
20323 (MULTILIB_DEFAULTS): Likewise.
20324 (DWARF2_UNWIND_INFO): Likewise.
20325 * config/arc/big.h: New file.
20326 * config/arc/elf.h: Likewise.
20327 * config/arc/linux.h: Likewise.
20328 * config/arc/t-uClibc: Remove.
20330 2017-02-27 Bin Cheng <bin.cheng@arm.com>
20332 PR tree-optimization/77536
20333 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
20334 (tree_transform_and_unroll_loop): Use above function to compute the
20335 estimated niter of unrolled loop and use it when scaling profile.
20336 Also use count info rather than frequency if it's non-zero.
20337 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
20338 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
20339 (vect_transform_loop): Call above function.
20341 2017-02-27 Richard Biener <rguenther@suse.de>
20343 PR tree-optimization/45397
20344 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
20345 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
20346 (visit_nary_op): Add pattern matching for CSEing sign-changed
20347 or truncated operations with wider ones.
20349 2017-02-27 Richard Biener <rguenther@suse.de>
20351 PR tree-optimization/79690
20352 * tree-vect-stmts.c (vectorizable_store): Use vector type
20353 built from the DR with address-space.
20355 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
20357 * doc/invoke.texi (Optimize Options): Refine the description
20358 of asan-use-after-return.
20360 2017-02-25 Alan Modra <amodra@gmail.com>
20362 PR rtl-optimization/79584
20363 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
20364 base, not ad->base_term, the reg within base. Remove assertion
20365 that ad->base == ad->base_term. Replace gen_int_mode using
20366 bogus mode with const0_rtx.
20368 2017-02-25 Jakub Jelinek <jakub@redhat.com>
20370 PR middle-end/79396
20371 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
20372 FMA_EXPR like tcc_binary or tcc_unary.
20374 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
20377 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
20379 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
20380 (output_loc_operands): Handle DW_OP_call_ref and
20381 DW_OP_GNU_variable_value.
20382 (struct variable_value_struct): New type.
20383 (struct variable_value_hasher): Likewise.
20384 (variable_value_hash): New variable.
20385 (string_types): Remove.
20386 (copy_loc_descr): New function.
20387 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
20388 (prepend_loc_descr_to_each): New function.
20389 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
20390 instead of add_loc_descr_to_each if the first argument is single
20391 location list and the second has multiple.
20392 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
20393 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
20394 when looking for variable value which doesn't have other location info.
20395 (loc_list_from_tree): Formatting fix.
20396 (gen_array_type_die): Simplify DW_AT_string_length handling.
20397 (adjust_string_types): Remove.
20398 (gen_subprogram_die): Don't call adjust_string_types nor test/set
20399 string_types. Call resolve_variable_values.
20400 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
20401 (resolve_addr_in_expr): Likewise. Add A argument.
20402 (copy_deref_exprloc): Remove deref argument. Adjust for the
20403 original expression being DW_OP_GNU_variable_value with optionally
20404 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
20405 optionally after it.
20406 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
20407 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
20408 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
20409 (variable_value_hasher::hash, variable_value_hasher::equal): New
20411 (resolve_variable_value_in_expr, resolve_variable_value,
20412 resolve_variable_values, note_variable_value_in_expr,
20413 note_variable_value): New functions.
20414 (dwarf2out_early_finish): Call note_variable_value on all toplevel
20417 2017-02-24 Jakub Jelinek <jakub@redhat.com>
20420 * opts.h (handle_generated_option): Add GENERATED_P argument.
20421 * opts-common.c (handle_option): Adjust function comment.
20422 (handle_generated_option): Add GENERATED_P argument, pass it to
20424 (control_warning_option): Pass false to handle_generated_option
20426 * opts.c (maybe_default_option): Pass true to handle_generated_option
20428 * optc-gen.awk: Likewise.
20430 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20432 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
20433 a REG, look at the REG it is a SUBREG of.
20434 (splitter for cmpeqsi_t): Ditto.
20436 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20438 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
20439 the special USEs with the pattern of the insn, not the insn itself.
20441 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
20444 * doc/invoke.texi: Document -mload-store-pairs.
20446 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20447 Sandra Loosemore <sandra@codesourcery.com>
20449 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
20450 argument isn't a CONST_INT.
20451 (nios2_alternate_compare_const): Assert op is a CONST_INT.
20452 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
20453 (nios2_validate_compare): Bypass alternate compare logic if *op2
20454 is not a CONST_INT.
20455 (ldstwm_operation_p): Return false if first_base is not a REG or
20456 if first_offset is not a CONST_INT.
20458 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20460 * config/cris/cris.md: Use correct operand in a define_peephole2.
20462 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20464 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
20466 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20468 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
20469 this_insn if it is an INSN or JUMP_INSN.
20470 (force_offsettable): Look at base, not at addr.
20471 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
20472 on things that aren't necessarily CONST_INTs.
20474 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
20476 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
20477 -mfpmath=sse is the default also for x86-32 targets with SSE2
20478 instruction set when @option{-ffast-math} is enabled
20480 2017-02-24 Jeff Law <law@redhat.com>
20482 PR rtl-optimizatoin/79286
20483 * ira.c (update_equiv_regs): Drop may_trap_p exception to
20486 2017-02-24 Richard Biener <rguenther@suse.de>
20488 PR tree-optimization/79389
20489 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
20492 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
20494 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
20495 function comment to reflect reality.
20496 (loop_exits_before_overflow): Fix typo in function description.
20498 2017-02-24 Richard Biener <rguenther@suse.de>
20500 PR tree-optimization/79389
20501 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
20502 properly that a threading opportunity exists. Detect conditional
20503 copy/constant propagation opportunities.
20505 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
20507 * config/visium/visium.md (type): Add trap.
20508 (b): New mode attribute.
20509 (*btst): Rename into...
20510 (*btst<mode>): ...this and adjust.
20511 (*cbranchsi4_btst_insn): Rename into...
20512 (*cbranch<mode>4_btst_insn): ...this and adjust.
20513 (trap): New define_insn.
20515 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20517 PR tree-optimization/79389
20518 * ifcvt.c (struct noce_if_info): Add rev_cond field.
20519 (noce_reversed_cond_code): New function.
20520 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
20521 reversed_comparison_code. Formatting fix.
20522 (noce_try_store_flag): Test rev_cond != NULL in addition to
20523 reversed_comparison_code.
20524 (noce_try_store_flag_constants): Likewise.
20525 (noce_try_store_flag_mask): Likewise.
20526 (noce_try_addcc): Use rev_cond if non-NULL instead of
20527 reversed_comparison_code.
20528 (noce_try_cmove_arith): Likewise. Formatting fixes.
20529 (noce_try_minmax, noce_try_abs): Clear rev_cond.
20530 (noce_find_if_block): Initialize rev_cond.
20531 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
20532 instead of false as last argument never attempt to reverse it
20535 2017-02-23 Bin Cheng <bin.cheng@arm.com>
20537 PR tree-optimization/79663
20538 * tree-predcom.c (combine_chains): Process refs in reverse order
20539 only for ZERO length chains, and add explaining comment.
20541 2017-02-23 Jeff Law <law@redhat.com>
20543 PR tree-optimization/79578
20544 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
20545 in call to operand_equal_p.
20547 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
20550 * config/i386/cpuid.h: Fix another undefined behavior.
20552 2017-02-23 Richard Biener <rguenther@suse.de>
20554 PR tree-optimization/79683
20555 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
20556 vector types for data-refs.
20558 2017-02-23 Martin Liska <mliska@suse.cz>
20560 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
20562 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20564 PR middle-end/79665
20565 * internal-fn.c (get_range_pos_neg): Moved to ...
20566 * tree.c (get_range_pos_neg): ... here. No longer static.
20567 * tree.h (get_range_pos_neg): New prototype.
20568 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
20569 are known to be in between 0 and signed maximum inclusive, try to
20570 expand both unsigned and signed divmod and use the cheaper one from
20573 2017-02-22 Jeff Law <law@redhat.com>
20575 PR tree-optimization/79578
20576 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
20577 to compare base operands.
20579 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
20582 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
20583 gpc_reg_operand instead of fpr_reg_operand.
20585 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
20587 * config/mips/mips.c (mips_return_in_memory): Force FP
20588 vector types to be returned in memory for o32 ABI.
20590 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20592 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
20593 instead of DW_TAG_member for static data member declarations and don't
20594 set no_linkage_name for static inline data members.
20595 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
20598 2017-02-22 Martin Liska <mliska@suse.cz>
20600 * doc/invoke.texi: Replace inequality signs with square brackets
20603 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20605 PR tree-optimization/68644
20606 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
20608 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
20611 * lra-constraints.c (simplify_operand_subreg): Handle
20612 WORD_REGISTER_OPERATIONS targets.
20614 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20617 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
20618 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
20619 elimination by swapping fld*.
20621 2017-02-22 Richard Biener <rguenther@suse.de>
20623 PR tree-optimization/79673
20624 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
20625 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
20626 irrelevant address-space qualifiers and avoiding a
20627 ADDR_SPACE_CONVERT_EXPR from fold_convert.
20629 2017-02-22 Richard Biener <rguenther@suse.de>
20631 PR tree-optimization/79666
20632 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
20633 to not symbolically negate if that may introduce undefined
20636 2017-02-22 Martin Liska <mliska@suse.cz>
20639 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
20640 * data-streamer-out.c (streamer_write_gcov_count_stream):
20642 * value-prof.c (stream_out_histogram_value): Make assert more
20643 precise based on type of counter.
20645 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
20648 * config/i386/i386.md (standard_x87sse_constant_load splitter):
20649 Use nonimmediate_operand instead of memory_operand for operand 1.
20650 (float-extend standard_x87sse_constant_load splitter): Ditto.
20652 2017-02-21 Jeff Law <law@redhat.com>
20654 PR tree-optimization/79621
20655 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
20656 blocks with edges to themselves.
20658 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20661 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
20662 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
20663 Use gimple_call_builtin_p.
20666 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
20667 on temporarily removed DEBUG_INSNs.
20669 PR tree-optimization/79649
20670 * tree-loop-distribution.c (classify_partition): Give up on
20671 non-generic address space loads/stores.
20673 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
20675 * doc/loop.texi (Loop manipulation): Remove nonexistent
20676 tree_ssa_loop_version from the documentation.
20677 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
20679 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20682 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
20683 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
20684 * config/rs6000/rs6000.c: Include except.h.
20685 (rs6000_expand_split_stack_prologue): Call
20686 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
20688 2017-02-21 Martin Jambor <mjambor@suse.cz>
20691 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
20692 have been analyzed.
20694 2017-02-21 Martin Jambor <mjambor@suse.cz>
20696 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
20697 for backward compatibility only.
20698 * doc/invoke.texi (Option Summary): Remove all references to
20699 -fipa-cp-alignment.
20701 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
20705 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20707 * lra-constraints.c (curr_insn_transform): Handle
20708 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
20710 2017-02-21 Martin Liska <mliska@suse.cz>
20712 * config/i386/i386.opt: Replace -masm-dialect with -masm.
20714 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
20716 PR translation/79638
20717 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
20719 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
20722 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
20723 (arm_function_ok_for_sibcall): Return false for an indirect call by
20724 descriptor if all the argument registers are used.
20725 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
20726 alignment of the function.
20728 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20730 PR tree-optimization/61441
20731 * simplify-rtx.c (simplify_const_unary_operation): For
20732 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
20733 the sNaN unmodified.
20735 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20737 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20738 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20739 instead of SYSTEM_HEADER_DIR.
20741 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
20742 Martin Liška <mliska@suse.cz>
20744 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
20745 Fix typos and grammar, use active voice, and clarify.
20747 2017-02-20 Marek Polacek <polacek@redhat.com>
20749 PR middle-end/79537
20750 * gimplify.c (gimplify_expr): Handle unused *&&L;.
20753 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
20755 2017-02-20 Jakub Jelinek <jakub@redhat.com>
20758 * config/i386/i386.c (ix86_expand_builtin): Handle
20759 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
20760 ix86_builtins_isa[fcode].isa as a requirement of those
20761 flags and any other flag in the bitmask.
20762 (ix86_init_mmx_sse_builtins): Use 0 instead of
20763 ~OPTION_MASK_ISA_64BIT as mask.
20764 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
20765 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
20766 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
20767 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
20769 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20772 * lra-constraints.c (split_reg): Check requested split mode
20773 is supported by the register.
20775 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20777 * lra-constraints.c (simplify_operand_subreg): Remove early
20780 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20783 * lra-constraints.c (curr_insn_transform): Tighten condition
20784 for converting SUBREG reloads from OP_OUT to OP_INOUT.
20786 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20789 * lra-constraints.c (curr_insn_transform): Handle
20790 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
20792 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
20795 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
20797 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
20799 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
20802 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
20805 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20807 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
20810 2017-02-18 Jakub Jelinek <jakub@redhat.com>
20812 * final.c (last_columnnum, override_columnnum): New variables.
20813 (final_start_function): Set last_columnnum, pass it to begin_prologue
20814 hook and pass 0 to dwarf2out_begin_prologue.
20815 (final_scan_insn): Update override_columnnum. Pass last_columnnum
20816 to source_line debug hook.
20817 (notice_source_line): Compute last_columnnum and for debug_column_info
20818 return true on column changes.
20819 * debug.h (struct gcc_debug_hooks): Add column argument to
20820 source_line and begin_prologue hooks.
20821 (debug_nothing_int_charstar_int_bool): Remove prototype.
20822 (debug_nothing_int_int_charstar,
20823 debug_nothing_int_int_charstar_int_bool): New prototypes.
20824 (dwarf2out_begin_prologue): Add column argument.
20825 * debug.c (do_nothing_debug_hooks): Adjust source_line and
20826 begin_prologue hooks.
20827 (debug_nothing_int_charstar_int_bool): Remove.
20828 (debug_nothing_int_int_charstar,
20829 debug_nothing_int_int_charstar_int_bool): New functions.
20830 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
20831 through to dwarf2out_source_line.
20832 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
20833 (dwarf2out_source_line): Add column argument, emit it if requested.
20834 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
20836 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
20837 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
20838 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
20839 through to dwarf2out_begin_prologue.
20840 (vmsdbgout_source_line): Add column argument, pass it through to
20841 dwarf2out_source_line.
20842 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
20843 dbxout_source_line caller.
20844 (dbxout_source_line): Add column argument.
20846 * common.opt (gno-column-info, gcolumn-info): New options.
20847 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
20848 (check_die): Also test for multiple DW_AT_decl_column attributes.
20849 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
20850 DW_AT_decl_column if requested.
20851 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
20853 (gen_variable_die): Likewise.
20854 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
20855 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
20858 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
20859 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
20860 (ix86_handle_option): Handle OPT_m3dnowa.
20861 * doc/invoke.texi (-m3dnowa): Document.
20862 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
20863 -m3dnowa instead of -m3dnow -march=athlon.
20866 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
20867 instead of gcc_assert for K, r and R code checks. Formatting fixes.
20869 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20872 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
20873 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
20874 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
20875 generator for vsx_xxpermdi_<mode>_be.
20876 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
20877 force big-endian semantics.
20878 (vsx_xxpermdi_<mode>_be): New define_expand with same
20879 implementation as previous version of vsx_xxpermdi_<mode>.
20881 2017-02-17 Jakub Jelinek <jakub@redhat.com>
20883 PR tree-optimization/79327
20884 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
20885 variable, its initialization and use.
20887 2017-02-17 Julia Koval <julia.koval@intel.com>
20889 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
20890 (OPTION_MASK_ISA_PKU_UNSET): New.
20891 (ix86_handle_option): Handle -mrdpid.
20892 * config/i386/cpuid.h (bit_RDPID): New.
20893 * config/i386/driver-i386.c (host_detect_local_cpu):
20894 Detect RDPID feature.
20895 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
20896 * config/i386/i386-c.c (ix86_target_macros_internal):
20898 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
20899 (ix86_valid_target_attribute_inner_p): Add "rdpid".
20900 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
20901 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
20902 * config/i386/i386.md (define_insn "rdpid"): New.
20903 * config/i386/i386.opt Add -mrdpid.
20904 * config/i386/immintrin.h (_rdpid_u32): New.
20906 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
20908 PR rtl-optimization/79541
20909 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
20910 instead of transforming it into USE.
20912 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
20914 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
20915 If HONOR_SNANS (SFmode) force the input to a register.
20916 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
20917 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
20918 an frsp or similar insn.
20920 2017-02-17 Martin Liska <mliska@suse.cz>
20922 PR rtl-optimization/79577
20923 * params.def (selsched-max-sched-times): Increase minimum to 1.
20925 2017-02-17 Martin Liska <mliska@suse.cz>
20927 PR rtl-optimization/79574
20928 * gcse.c (want_to_gcse_p): Prevent integer overflow.
20930 2017-02-17 Martin Liska <mliska@suse.cz>
20932 PR tree-optimization/79529
20933 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
20934 ssa_defined_default_def_p to handle cases which are implicitly
20936 * tree-ssa.c (ssa_defined_default_def_p): New function.
20937 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
20938 which are implicitly defined.
20939 * tree-ssa.h (ssa_defined_default_def_p): Declare.
20941 2017-02-17 Richard Biener <rguenther@suse.de>
20943 PR middle-end/79576
20944 * params.def (max-ssa-name-query-depth): Limit to 10.
20946 2017-02-17 Richard Biener <rguenther@suse.de>
20948 PR tree-optimization/79552
20949 * tree-ssa-structalias.c (visit_loadstore): Properly verify
20952 2017-02-17 Richard Biener <rguenther@suse.de>
20955 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
20957 2017-02-17 Marek Polacek <polacek@redhat.com>
20959 PR middle-end/79536
20960 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
20961 (fold_negate_expr): New wrapper.
20963 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
20965 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
20966 Correct terminology and de-emphasize pre-standard behavior.
20968 2017-02-16 Alan Modra <amodra@gmail.com>
20970 PR rtl-optimization/79286
20971 * ira.c (def_dominates_uses): New function.
20972 (update_equiv_regs): Don't create an equivalence for insns that
20973 may trap where the register def does not dominate the use.
20975 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
20977 PR rtl-optimization/78127
20978 * lra.c (lra): Call lra_eliminate before finish the loop after
20981 2017-02-16 Richard Biener <rguenther@suse.de>
20983 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
20985 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
20986 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
20987 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
20988 (isl_val_int_from_wi): New function.
20989 (extract_affine_gmp): Rename to ...
20990 (extract_affine_wi): ... this, take a widest_int.
20991 (extract_affine_int): Just wrap extract_affine_wi.
20992 (add_param_constraints): Use isl_val_int_from_wi.
20993 (add_loop_constraints): Likewise, and extract_affine_wi.
20995 2017-02-15 Jeff Law <law@redhat.com>
20997 PR middle-end/79521
20998 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
20999 ira_init_register_move_cost_if_necessary.
21001 2017-02-15 Martin Sebor <msebor@redhat.com>
21003 PR middle-end/32003
21004 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21005 removed in a prior commit.
21007 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21009 PR tree-optimization/79347
21010 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21011 counters during peeling.
21013 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21015 * Makefile.in (site.exp): Remove "set ISLVER".
21017 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21020 * real.c (real_from_integer): Call real_convert even for decimal.
21022 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21025 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21027 2017-02-14 Andrew Pinski <apinski@cavium.com>
21029 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21030 cores and change the partno/implementer to be correct.
21031 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21032 the 'B" as the implementer.
21033 * config/aarch64/aarch64-tune.md: Regenerate.
21035 2017-02-14 Carl Love <cel@us.ibm.com>
21037 * config/rs6000/rs6000.c: Add case statement entry to make the
21038 xvcvuxdsp built-in argument unsigned.
21039 * config/rs6000/vsx.md: Fix the source and return operand types so they
21040 match the instruction definitions from the ISA document. Fix typo
21041 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21044 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21047 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21048 member early_clobber_alts.
21049 * lra-lives.c (reg_early_clobber_p): New.
21050 (process_bb_lives): Use it.
21051 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21052 (debug_operand_data): Initialize early_clobber_alts.
21053 (setup_operand_alternative): Set up early_clobber_alts.
21054 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21055 alternatives to new_insn_reg.
21056 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21058 (lra_update_insn_regno_info): Pass the new arg.
21060 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21062 PR middle-end/79505
21063 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21064 (new_oacc_loop_raw): Don't clear already cleared fields.
21067 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21068 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21069 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21071 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21074 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21076 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21079 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21080 the extra instruction to the right place to store 128-bit constant
21083 2017-02-14 Martin Sebor <msebor@redhat.com>
21085 PR middle-end/79448
21086 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21087 warning for strings of unknown length.
21089 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21091 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21093 2017-02-14 Jeff Law <law@redhat.com>
21096 * ira-costs.c (scan_one_insn): Initialize register move costs
21097 for pseudos seen in USE/CLOBBER insns.
21099 PR tree-optimization/79095
21100 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
21101 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
21102 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
21103 if the operands are known to be not equal, then the resulting range
21105 (intersect_ranges): If the new range is ~[0,0] and the old range is
21106 wide, then prefer ~[0,0].
21107 * tree-vrp.c (overflow_comparison_p_1): New function.
21108 (overflow_comparison_p): New function.
21109 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
21110 if NAME is used in an overflow test.
21111 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
21112 overflow check that can be expressed as an equality test, then adjust
21113 ops to be that equality test.
21115 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21117 * config/s390/s390-builtin-types.def: Remove flags argument.
21118 * config/s390/s390.c (s390_init_builtins): Likewise.
21120 2017-02-14 Martin Liska <mliska@suse.cz>
21122 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
21123 vector. Fix trailing white spaces.
21125 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
21127 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
21130 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21132 PR rtl-optimization/68664
21133 * config/arm/arm.c (arm_sched_can_speculate_insn):
21134 New function. Declare prototype.
21135 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21137 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21139 PR rtl-optimization/68664
21140 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
21142 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21144 2017-02-14 Amit Pawar <amit.pawar@amd.com>
21146 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
21147 max skip bytes for function, loop and jump.
21149 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21151 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
21152 ABS_EXPR for gimple dump.
21154 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21157 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
21159 PR tree-optimization/79408
21160 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
21161 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
21162 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
21163 also if rhs1 is INTEGER_CST.
21165 2017-02-14 Richard Biener <rguenther@suse.de>
21167 PR middle-end/79432
21168 * tree-into-ssa.c (insert_phi_nodes): When the function can
21169 have abnormal edges rewrite SSA names with broken use-def
21170 dominance out of SSA and register them for PHI insertion.
21172 2017-02-13 Martin Sebor <msebor@redhat.com>
21174 PR middle-end/79496
21175 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
21176 clearing info.nowrite flag when snprintf size argument is a range.
21178 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21180 * cprop.c (cprop_jump): Add missing space in string literal.
21181 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
21182 (get_constraint_for_component_ref): Likewise.
21183 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
21184 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
21185 * lra-constraints.c (process_alt_operands): Likewise.
21186 * ipa-inline.c (inline_small_functions): Likewise.
21187 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
21188 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21189 * trans-mem.c (diagnose_tm_1_op): Likewise.
21190 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
21191 (grid_parallel_clauses_gridifiable): Likewise.
21193 * config/nvptx/mkoffload.c (process): Add space in between
21196 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
21197 "MOD4_SSE_REGS" and "ALL_REGS".
21199 * spellcheck.c (test_data): Add , in between "foo" and "food".
21201 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21204 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
21205 boundary crossing check and subsequent code generation agree.
21207 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21209 * config/aarch64/aarch64.c (has_memory_op): Delete.
21210 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
21213 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21215 PR rtl-optimization/79388
21216 PR rtl-optimization/79450
21217 * combine.c (distribute_notes): When removing TEM_INSN for which
21218 corresponding dest has last value recorded, invalidate that last
21221 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21223 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
21224 of explicit '@'. Add missing assembly comment marker on branch costs
21227 2017-02-13 Nathan Sidwell <nathan@acm.org>
21229 * gengtype-lex.l (<in_struct>): Add '/'.
21231 2017-02-13 Martin Liska <mliska@suse.cz>
21234 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
21236 2017-02-13 Richard Biener <rguenther@suse.de>
21238 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
21240 * configure: Re-generate.
21241 * config.in: Likewise.
21242 * graphite-dependences.c: Simplify as if
21243 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
21244 * graphite-isl-ast-to-gimple.c: Likewise.
21245 * graphite-optimize-isl.c: Likewise.
21246 * graphite-poly.c: Likewise.
21247 * graphite-sese-to-poly.c: Likewise.
21248 * graphite.h: Likewise.
21249 * toplev.c: Include isl/version.h and use isl_version () for
21250 printing the ISL version.
21251 * doc/install.texi: Update ISL requirement.
21253 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21255 * doc/standards.texi (Standards): Update reference to
21258 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21260 * doc/extend.texi (Named Address Spaces): sourceware.org now
21262 * doc/install.texi (Binaries): Ditto.
21265 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21267 * doc/cpp.texi: Replace "stringify"/"stringification" with C
21268 standard terminology "stringize"/"stringizing" throughout.
21269 * doc/cppinternals.texi: Likewise.
21271 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21273 * doc/extend.texi: Fix some spelling mistakes and typos.
21274 * doc/invoke.texi: Likewise.
21276 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21279 * params.def (inline-min-speedup) Change from 10 to 8.
21281 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21283 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
21286 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21289 * ipa-inline-analysis.c (get_minimal_bb): New function.
21290 (record_modified): Use it.
21291 (remap_edge_change_prob): Handle also ancestor functions.
21293 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
21295 * doc/contrib.texi (Contributors): Remove broken link into
21296 the Mauve CVS repository.
21298 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21300 PR middle-end/79454
21301 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
21302 result computation whenever lhs doesn't have vector mode, not
21303 just when it has BLKmode.
21305 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21307 * doc/makefile.texi (profiledbootstrap): Refer to the
21308 installation instructions only in textual form.
21310 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21313 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
21315 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21317 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
21318 (Specific): Update mingw-w64 reference.
21320 (Specific): Remove broken link to Renesas RX processor.
21322 2017-02-10 Richard Biener <rguenther@suse.de>
21324 * toplev.c (process_options): Do not mention obsolete graphite
21325 options when printing sorry message about missing graphite support.
21326 Mention -floop-nest-optimize.
21328 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
21330 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
21331 (vtst_p16): Likewise.
21332 (vtstq_p8): Likewise.
21333 (vtstq_p16): Likewise.
21335 (vtstq_p64): Likewise.
21336 * config/arm/arm_neon.h (vgetq_lane_p64): New.
21337 (vset_lane_p64): New.
21338 (vsetq_lane_p64): New.
21340 2017-02-10 Jakub Jelinek <jakub@redhat.com>
21342 PR tree-optimization/79411
21343 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
21344 stmt operands are SSA_NAMEs used in abnormal phis.
21345 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
21348 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21351 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
21354 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21356 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
21358 2017-02-09 Jakub Jelinek <jakub@redhat.com>
21360 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
21364 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
21365 not arbitrary TREE_CONSTANT.
21368 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
21369 "omp declare target link" attribute unless is_global_var.
21370 * omp-offload.c (find_link_var_op): Likewise.
21372 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
21373 Chung-Lin Tang <cltang@codesourcery.com>
21375 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
21377 (gimplify_adjust_omp_clauses): Don't delete TILE.
21378 (gimplify_omp_for): Deal with TILE.
21379 * internal-fn.c (expand_GOACC_TILE): New function.
21380 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
21382 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
21383 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
21385 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
21386 avoid DIV for outermost collapse var.
21387 (expand_oacc_for): Insert tile element loop as needed. Adjust.
21388 Remove out of date comments, fix whitespace.
21389 * omp-general.c (omp_extract_for_data): Deal with tiling.
21390 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
21391 adjust OLF_DIM_BASE value.
21392 (struct omp_for_data): Add tiling field.
21393 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
21394 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
21395 for auto loops. Remove default auto determining, moved to
21396 oacc_loop_fixed_partitions.
21397 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
21398 stmts, add e_mask field.
21399 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
21400 (oacc_thread_numbers): Use oacc_dim_call.
21401 (oacc_xform_tile): New.
21402 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
21403 (finish_oacc_loop): Adjust for ifns vector.
21404 (oacc_loop_discover_walk): Append loop abstraction sites to list,
21405 add case for GOACC_TILE fns.
21406 (oacc_loop_xform_loop): Delete.
21407 (oacc_loop_process): Iterate over call list directly, and add
21408 handling for GOACC_TILE fns.
21409 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
21411 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
21412 vector partitioning to outer loops. Assign 2 partitions to loops
21413 when available. Add TILE handling.
21414 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
21415 (execite_oacc_device_lower): Process GOACC_TILE fns,
21416 ignore unknown specs.
21417 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
21418 * tree.c (omp_clause_num_ops): Adjust TILE ops.
21419 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
21421 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
21423 * configure.ac (ACX_BUGURL): Update.
21424 * configure: Regenerate.
21426 2017-02-09 Richard Biener <rguenther@suse.de>
21428 PR tree-optimization/69823
21429 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
21430 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
21432 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21434 * config/arc/arc-c.def: Add __NPS400__ definition.
21435 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
21436 (TARGET_NPS400): Define.
21438 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21440 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
21442 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
21443 pointer, arch_info.
21444 (arc_cpu_types): Fill the arch_info field with a pointer into the
21445 arc_arch_types table.
21446 (arc_selected_cpu): Declare.
21447 * config/arc/arc.c (arc_selected_cpu): Make global.
21448 (arc_selected_arch): Delete.
21449 (arc_base_cpu): Delete.
21450 (arc_override_options): Remove references to deleted variables,
21451 update access to arch information.
21452 (ARC_OPT): Update access to arch information.
21453 (ARC_OPTX): Likewise.
21454 * config/arc/arc.h (arc_base_cpu): Remove declaration.
21455 (TARGET_ARC600): Update access to arch information.
21456 (TARGET_ARC601): Likewise.
21457 (TARGET_ARC700): Likewise.
21458 (TARGET_EM): Likewise.
21459 (TARGET_HS): Likewise.
21460 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
21463 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
21466 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
21467 condition/operands for integer GE/LE/GEU/LEU operations.
21469 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
21471 PR translation/79397
21472 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
21475 2017-02-08 Martin Jambor <mjambor@suse.cz>
21478 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
21479 whether allocation happened.
21480 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
21481 nothing was allocated.
21483 2017-02-08 Jakub Jelinek <jakub@redhat.com>
21485 PR tree-optimization/79408
21486 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
21487 constant, but SSA_NAME with a known integer range, use the minimum
21488 of that range instead of op1 to determine if modulo can be replaced
21489 with its first operand.
21491 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21493 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
21495 2017-02-08 Richard Biener <rguenther@suse.de>
21497 PR tree-optimization/71824
21498 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
21499 Check all loops contained in the merged region.
21501 2017-02-07 Andrew Pinski <apinski@cavium.com>
21503 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
21505 2017-02-07 Andrew Pinski <apinski@cavium.com>
21507 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
21508 (thunderxt88): Likewise.
21509 (thunderxt81): Disable LSE and change v8.1 to v8.
21510 (thunderxt83): Likewise.
21512 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21513 Richard Biener <rguenther@suse.de>
21515 PR middle-end/79399
21516 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
21517 type from int to size_t.
21518 * ira-costs.c (struct_costs_size): Change type from int to size_t.
21520 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21522 PR rtl-optimization/79386
21523 * cprop.c (bypass_conditional_jumps): Initialize
21524 bypass_last_basic_block already before splitting bbs after
21525 unconditional traps...
21526 (bypass_conditional_jumps): ... rather than here.
21529 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
21530 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
21531 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
21532 fix -masm=intel patterns.
21534 2017-02-07 Richard Biener <rguenther@suse.de>
21536 PR tree-optimization/79256
21537 PR middle-end/79278
21538 * builtins.c (get_object_alignment_2): Use min_align_of_type
21539 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
21540 and ADJUST_FIELD_ALIGN.
21542 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
21544 * doc/tm.texi: Regenerate.
21545 * stor-layout.c (layout_decl): Adjust.
21546 (update_alignment_for_field): Likewise.
21547 (place_field): Likewise.
21548 (min_align_of_type): Likewise.
21549 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
21550 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
21551 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
21552 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
21553 * config/frv/frv.c (frv_adjust_field_align): Likewise.
21554 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
21555 * config/i386/i386.c (x86_field_alignment): Likewise.
21556 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
21557 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
21558 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
21559 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
21560 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
21561 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
21565 2017-01-30 Richard Biener <rguenther@suse.de>
21567 PR tree-optimization/79256
21568 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21569 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21572 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
21574 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
21575 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
21576 builtins to SImode and emit a zero-extend, if necessary.
21578 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21580 * docs/invoke.texi (RISC-V Options): Alphabetize.
21582 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21584 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
21587 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21589 * config/riscv/riscv.c: New file.
21590 * common/config/riscv/riscv-common.c: Likewise.
21591 * config.gcc: Likewise.
21592 * config/riscv/constraints.md: Likewise.
21593 * config/riscv/elf.h: Likewise.
21594 * config/riscv/generic.md: Likewise.
21595 * config/riscv/linux.h: Likewise.
21596 * config/riscv/multilib-generator: Likewise.
21597 * config/riscv/peephole.md: Likewise.
21598 * config/riscv/pic.md: Likewise.
21599 * config/riscv/predicates.md: Likewise.
21600 * config/riscv/riscv-builtins.c: Likewise.
21601 * config/riscv/riscv-c.c: Likewise.
21602 * config/riscv/riscv-ftypes.def: Likewise.
21603 * config/riscv/riscv-modes.def: Likewise.
21604 * config/riscv/riscv-opts.h: Likewise.
21605 * config/riscv/riscv-protos.h: Likewise.
21606 * config/riscv/riscv.h: Likewise.
21607 * config/riscv/riscv.md: Likewise.
21608 * config/riscv/riscv.opt: Likewise.
21609 * config/riscv/sync.md: Likewise.
21610 * config/riscv/t-elf-multilib: Likewise.
21611 * config/riscv/t-linux: Likewise.
21612 * config/riscv/t-linux-multilib: Likewise.
21613 * config/riscv/t-riscv: Likewise.
21614 * configure.ac: Likewise.
21615 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
21616 Waterman as RISC-V maintainers.
21617 * doc/install.texi: Add RISC-V entries.
21618 * doc/invoke.texi: Add RISC-V options section.
21619 * doc/md.texi: Add RISC-V constraints section.
21620 * configure: Regenerated.
21622 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
21625 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
21626 false values to be constant vectors with all 0 or all 1 bits set.
21627 (vcondu<mode><mode>): Likewise.
21628 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
21630 (fpmask_comparison_operator): Update comment.
21631 (vecint_comparison_operator): New predicate.
21632 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
21633 vector conditionals when the true and false values are constant
21634 vectors with all 0 bits or all 1 bits set.
21636 2017-02-06 Martin Sebor <msebor@redhat.com>
21638 PR tree-optimization/79376
21639 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
21641 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
21643 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
21644 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
21645 to simplify split condition.
21647 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21649 * omp-expand.c (oxpand_omp_atomic_fetch_op,
21650 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
21653 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
21655 PR rtl-optimization/68664
21656 * target.def (can_speculate_insn): New hook.
21657 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
21658 * doc/tm.texi: Regenerate.
21659 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
21660 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
21661 (rs6000_sched_can_speculate_insn): New function.
21663 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21665 PR tree-optimization/79284
21666 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
21667 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
21668 vectorizable_mask_load_store, vectorizable_operation,
21669 vect_is_simple_cond, get_same_sized_vectype): Use it instead
21670 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
21671 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
21672 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
21673 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
21674 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
21675 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
21676 is_gimple_assign (stmt). Replace another such test with
21677 is_gimple_assign (stmt).
21679 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21682 * config/avr/avr.c (rtl-iter.h): Include it.
21683 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
21684 (avr_legitimate_combined_insn): ...and implementation.
21686 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
21688 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
21689 * config/s390/s390.c (s390_const_operand_ok)
21690 (s390_canonicalize_comparison, s390_extract_part)
21691 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
21692 (s390_contiguous_bitmask_p, s390_rtx_costs)
21693 (legitimize_pic_address): Likewise.
21694 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
21695 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
21696 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
21697 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
21698 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
21700 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21702 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
21703 REGNO($0) == REGNO($1).
21705 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21707 * config/s390/linux.h(SIZE_TYPE): Add comment.
21709 2017-02-06 Julian Brown <julian@codesourcery.com>
21710 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21711 Virendra Pathak <virendra.pathak@broadcom.com>
21713 * config/aarch64/aarch64-cores.def: Change the scheduler
21715 * config/aarch64/aarch64.md: Include thunderx2t99.md.
21716 * config/aarch64/thunderx2t99.md: New file.
21718 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21720 * doc/standards.texi (Go Language): Update link to language
21723 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21725 * tree-eh.c (lower_resx): Sanitize profile.
21726 (cleanup_empty_eh_move_lp): Likewise.
21728 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21731 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
21733 * cfgloopmanip.h (loop_version): Update prototype.
21734 * modulo-sched.c (sms_schedule): Update call of loop_version.
21735 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
21736 * tree-parloops.c (gen_parallel_loop): Likewise.
21737 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
21738 * tree-ssa-loop-split.c (split_loop): Likewise.
21739 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
21740 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21742 2017-02-05 Martin Liska <mliska@suse.cz>
21745 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
21747 (print_operand_address): Initialize a struct to zero.
21749 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21751 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
21752 garbage collector only in textual form.
21754 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21756 * doc/extend.texi (x86 specific memory model extensions for
21757 transactional memory): Simplify a phrase.
21759 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
21762 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
21763 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
21764 (atomic_storedi_1): Likewise.
21766 2017-02-04 Jakub Jelinek <jakub@redhat.com>
21768 PR tree-optimization/79338
21769 * tree-parloops.c (gather_scalar_reductions): Don't call
21770 vect_analyze_loop_form for loop->inner before destroying loop's
21773 2017-02-03 Martin Sebor <msebor@redhat.com>
21775 PR tree-optimization/79327
21776 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
21777 when precision has resulted in leading zeros.
21778 (format_integer): Adjust the likely counter to assume an unknown
21779 argument that may be zero is non-zero.
21781 2017-02-03 Jason Merrill <jason@redhat.com>
21784 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
21785 avoid copying non-taken branch.
21787 2017-02-03 Jakub Jelinek <jakub@redhat.com>
21789 PR tree-optimization/79340
21790 * tree-vect-loop.c (vectorizable_reduction): Release
21791 vec_defs elements after safe_splicing them into other vectors.
21794 PR tree-optimization/79327
21795 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
21796 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
21798 (format_integer): Use wide_int_to_tree instead of build_int_cst
21799 + to_?hwi. If argmin is NULL, just set argmin and argmax to
21800 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
21801 of shortest and longest sequence.
21803 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
21805 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
21806 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
21808 2017-02-03 Walter Lee <walt@tilera.com>
21811 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
21812 after initial stackframe link reg save.
21813 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
21815 2017-02-03 Jakub Jelinek <jakub@redhat.com>
21818 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
21819 wu for stxssp alternative.
21821 2017-02-03 Martin Sebor <msebor@redhat.com>
21823 PR tree-optimization/79352
21824 * gimple-fold.c (get_range_strlen): Add argument.
21825 (get_range_strlen): Change return type to bool.
21826 (get_maxval_strlen): Pass in a dummy argument.
21827 * gimple-fold.h (get_range_strlen): Change return type to bool.
21828 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
21829 * tree.h (array_at_struct_end_p): Add argument.
21830 * tree.c (array_at_struct_end_p): Handle it.
21832 2017-02-03 Martin Liska <mliska@suse.cz>
21835 * multiple_target.c (create_dispatcher_calls): Redirect edge
21836 from a caller of a dispatcher.
21837 (expand_target_clones): Make the clones local.
21838 (ipa_target_clone): Do both target clones and resolvers.
21839 (ipa_dispatcher_calls): Remove the pass.
21840 (pass_dispatcher_calls::gate): Likewise.
21841 (make_pass_dispatcher_calls): Likewise.
21842 * passes.def (pass_target_clone): Put as very first IPA early
21845 2017-02-03 Martin Liska <mliska@suse.cz>
21847 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
21848 in case of a function with ifunc attribute.
21850 2017-02-03 Martin Liska <mliska@suse.cz>
21852 * cgraph.c (cgraph_node::dump): Dump function version info.
21853 * symtab.c (symtab_node::dump_base): Add missing new line.
21855 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
21857 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
21858 (ifcombine_ifandif): Use it.
21860 2017-02-03 Martin Liska <mliska@suse.cz>
21862 * doc/invoke.texi: Document default value for
21863 use-after-scope-direct-emission-threshold.
21865 2017-02-03 Martin Liska <mliska@suse.cz>
21867 PR tree-optimization/79339
21868 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
21869 (format_floating): Likewise.
21871 2017-02-03 Martin Liska <mliska@suse.cz>
21874 * ipa-prop.c (ipa_node_params_t::insert): Remove current
21876 (ipa_node_params_t::remove): Likewise.
21877 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
21878 initialization from removed ipa_node_params_t::insert.
21879 (ipa_node_params::~ipa_node_params): Move from removed
21880 ipa_node_params_t::release.
21881 * symbol-summary.h (symbol_summary::m_released): New member.
21882 Do not release a summary twice. Do not allow to call finalizer
21883 for types of a summary that live in GGC memory.
21885 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21887 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
21890 2017-02-02 Martin Sebor <msebor@redhat.com>
21892 PR middle-end/79275
21893 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
21894 (format_string): Tighten up the range of output for non-constant
21895 strings and correct the expected range for wide non-constant strings.
21897 2017-02-02 Martin Sebor <msebor@redhat.com>
21899 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
21901 PR middle-end/32003
21902 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
21904 (-fdump-tree-@var): Add to index and document how to come up
21905 with pass-specific option and dump file names.
21906 (-fdump-passes): Clarify where to look for output.
21908 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
21910 PR middle-end/77445
21911 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
21912 statistics of the analyzed path; allow threading for speed when
21913 any of BBs along the path are optimized for speed.
21915 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
21917 PR middle-end/78468
21918 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
21919 settings of the virtual registers.
21922 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
21924 * explow.c (get_dynamic_stack_size): Take known alignment of stack
21925 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
21928 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21930 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
21931 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
21933 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21935 * config/s390/s390.md: Add missing comments with the expanded
21937 * config/s390/vector.md: Likewise.
21938 * config/s390/vx-builtins.md: Likewise.
21940 2017-02-02 Jakub Jelinek <jakub@redhat.com>
21943 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
21944 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
21945 conditions on a single line.
21947 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21949 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
21950 __S390_VX__ to __VX__.
21952 2017-02-01 Andrew Pinski <apinski@cavium.com>
21954 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
21955 stmt_info to record_stmt_cost.
21956 (vect_get_known_peeling_cost): Pass stmt_info if known to
21958 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
21959 cpu_vector_cost field into
21960 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
21961 field into vec_int_stmt_cost and vec_fp_stmt_cost.
21962 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
21963 splitting of scalar_stmt_cost and vec_stmt_cost.
21964 (thunderx_vector_cost): Likewise.
21965 (cortexa57_vector_cost): LIkewise.
21966 (exynosm1_vector_cost): Likewise.
21967 (xgene1_vector_cost): Likewise.
21968 (thunderx2t99_vector_cost): Improve after the splitting of the two
21970 (aarch64_builtin_vectorization_cost): Update for the splitting of
21971 scalar_stmt_cost and vec_stmt_cost.
21973 2017-02-01 Torvald Riegel <triegel@redhat.com>
21974 Richard Henderson <rth@redhat.com>
21976 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
21977 conditional on existance of a fast atomic load.
21978 * optabs-query.c (can_atomic_load_p): New function.
21979 * optabs-query.h (can_atomic_load_p): Declare it.
21980 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
21981 no fast atomic load is available for the particular size of access.
21982 (expand_atomic_compare_and_swap): Likewise.
21983 (expand_atomic_load): Likewise.
21984 (expand_atomic_store): Likewise.
21985 (expand_atomic_fetch_op): Likewise.
21986 * testsuite/lib/target-supports.exp
21987 (check_effective_target_sync_int_128): Remove x86 because it provides
21988 no fast atomic load.
21989 (check_effective_target_sync_int_128_runtime): Likewise.
21991 2017-02-01 Richard Biener <rguenther@suse.de>
21993 * graphite.c: Include tree-vectorizer.h for find_loop_location.
21994 (graphite_transform_loops): Provide opt-info for optimized nests.
21995 * tree-parloop.c (parallelize_loops): Provide opt-info for
21996 parallelized loops.
21998 2017-02-01 Richard Biener <rguenther@suse.de>
22000 PR middle-end/79315
22001 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22002 was not set before.
22004 2017-02-01 Richard Biener <rguenther@suse.de>
22006 PR tree-optimization/71824
22007 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22008 Verify the loops are valid in the merged SESE region.
22009 (scop_detection::can_represent_loop_1): Check analyzing the
22010 evolution of the number of iterations in the region succeeds.
22012 2017-01-31 Ian Lance Taylor <iant@golang.org>
22014 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22015 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22017 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22019 PR preprocessor/79210
22020 * input.c (get_substring_ranges_for_loc): Replace line_width
22021 assertion with error-handling.
22023 2017-01-31 Richard Biener <rguenther@suse.de>
22025 PR tree-optimization/77318
22026 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22027 (create_pw_aff_from_tree): Take loop parameter.
22028 (add_condition_to_pbb): Pass loop of the condition to
22029 create_pw_aff_from_tree.
22031 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22033 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22034 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22036 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22040 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22042 (convert_int_to_float128): Likewise.
22043 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22044 (convert_int_to_float128): Likewise.
22045 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22046 (UNSPEC_IEEE128_CONVERT): Likewise.
22047 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22048 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22049 Use local variables for IBM extended format.
22050 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22051 (fix_trunc<mode>si2_fprs): Likewise.
22052 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22053 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22054 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22055 to know that we can now have integers of all sizes in vector
22057 (fix<uns>_<mode>di2_hw): Likewise.
22058 (float<uns>_<mode>si2_hw): Likewise.
22059 (fix_<mode>si2_hw): Likewise.
22060 (fixuns_<mode>si2_hw): Likewise.
22061 (float<uns>_<mode>di2_hw): Likewise.
22062 (float_<mode>di2_hw): Likewise.
22063 (float_<mode>si2_hw): Likewise.
22064 (floatuns_<mode>di2_hw): Likewise.
22065 (floatuns_<mode>si2_hw): Likewise.
22066 (xscvqp<su>wz_<mode>): Delete, no longer used.
22067 (xscvqp<su>dz_<mode>): Likewise.
22068 (xscv<su>dqp_<mode>): Likewise.
22069 (ieee128_mfvsrd_64bit): Likewise.
22070 (ieee128_mfvsrd_32bit): Likewise.
22071 (ieee128_mfvsrwz): Likewise.
22072 (ieee128_mtvsrw): Likewise.
22073 (ieee128_mtvsrd_64bit): Likewise.
22074 (ieee128_mtvsrd_32bit): Likewise.
22076 2017-01-31 Martin Liska <mliska@suse.cz>
22079 * ipa-prop.c (ipa_free_all_node_params): Call release method
22080 instead of ~sumbol_summary to not to trigger double times
22083 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22085 PR tree-optimization/71691
22086 * bitmap.h (class auto_bitmap): New.
22087 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22088 is_maybe_undefined instead of ssa_undefined_value_p.
22090 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22092 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22093 __S390_ARCH_LEVEL__ to __ARCH__.
22095 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22097 PR tree-optimization/79267
22098 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22099 if should_remove_lhs_p is true.
22101 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
22104 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
22105 (add_alignment_attribute): New.
22106 (base_type_die): Add alignment attribute.
22107 (subrange_type_die): Likewise.
22108 (modified_type_die): Likewise.
22109 (gen_array_type_die): Likewise.
22110 (gen_descr_array_type_die: Likewise.
22111 (gen_enumeration_type_die): Likewise.
22112 (gen_subprogram_die): Likewise.
22113 (gen_variable_die): Likewise.
22114 (gen_field_die): Likewise.
22115 (gen_ptr_to_mbr_type_die): Likewise.
22116 (gen_struct_or_union_type_die): Likewise.
22117 (gen_subroutine_type_die): Likewise.
22118 (gen_typedef_die): Likewise.
22119 (base_type_cmp): Compare alignment attribute.
22121 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22124 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
22125 (setb_unsigned) New pattern for setb with CCUNS.
22126 * config/rs6000/rs6000.c (expand_block_compare): Use a different
22127 subfc./subfe sequence to avoid overflow problems. Generate a
22128 shorter sequence with cmpld/setb for power9.
22129 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
22130 for generating subfc. instruction.
22131 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
22132 now uses this instruction.
22134 2017-01-30 Ian Lance Taylor <iant@google.com>
22137 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
22138 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
22140 2017-01-30 Martin Sebor <msebor@redhat.com>
22142 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
22143 Move constant to the right of a relational operator.
22144 (get_mpfr_format_length, format_character, format_string): Ditto.
22145 (should_warn_p, maybe_warn): Same.
22147 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
22149 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
22152 * asan.c (get_translation_unit_decl): Remove function.
22153 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
22155 2017-01-30 Martin Liska <mliska@suse.cz>
22157 PR gcov-profile/79259
22158 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
22159 -fprofile-generate.
22161 2017-01-30 Martin Liska <mliska@suse.cz>
22164 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
22165 Initialize variables with NULL value.
22167 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
22170 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
22172 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
22174 2017-01-30 Richard Biener <rguenther@suse.de>
22176 PR tree-optimization/79276
22177 * tree-vrp.c (process_assert_insertions): Properly adjust common
22178 when removing a duplicate.
22180 * gcc.dg/torture/pr79276.c: New testcase.
22182 2017-01-30 Richard Biener <rguenther@suse.de>
22184 PR tree-optimization/79256
22185 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22186 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22188 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
22190 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
22193 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
22194 ("*r<noxa>sbg_<mode>_sll_bitmask")
22195 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
22196 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
22197 Use contiguous_bitmask_nowrap_operand.
22199 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22202 * config/rs6000/altivec.h (vec_xl): Revise #define.
22203 (vec_xst): Likewise.
22205 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
22207 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
22209 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
22211 PR rtl-optimization/79194
22212 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
22213 traps before call to bypass_conditional_jumps.
22215 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22217 PR tree-optimization/71374
22218 * lra-constraints.c (check_conflict_input_operands): New.
22219 (match_reload): Use it.
22221 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22224 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
22225 account to calculate conflict_set.
22227 2017-01-27 Bin Cheng <bin.cheng@arm.com>
22229 PR rtl-optimization/78559
22230 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
22231 other_insn in combine.
22233 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22235 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
22236 uint16_type_node for BT_UINT16.
22238 2017-01-27 David Malcolm <dmalcolm@redhat.com>
22240 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
22241 "RTL Tests" to menu.
22242 (GIMPLE Tests): New node.
22243 (RTL Tests): New node.
22245 2017-01-27 Richard Biener <rguenther@suse.de>
22247 PR tree-optimization/79245
22248 * tree-loop-distribution.c (distribute_loop): Apply cost
22249 modeling also to detected patterns.
22251 2017-01-27 Richard Biener <rguenther@suse.de>
22253 PR tree-optimization/71433
22254 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
22255 (compare_assert_loc): New function.
22256 (process_assert_insertions): Sort and optimize assert locations
22257 to remove duplicates and push down identical assertions on
22258 edges to their destination block.
22260 2017-01-27 Richard Biener <rguenther@suse.de>
22262 PR tree-optimization/79244
22263 * tree-vrp.c (remove_range_assertions): Forcefully propagate
22264 out SSA names even if abnormal.
22266 2017-01-27 Jakub Jelinek <jakub@redhat.com>
22268 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
22269 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
22270 instead of MPFR_RNDN.
22272 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
22275 * arm.c (arm_option_override): Don't call build_target_option_node
22276 until after doing all option overrides.
22277 (arm_valid_target_attribute_tree): Likewise.
22279 2017-01-27 Martin Liska <mliska@suse.cz>
22281 * doc/invoke.texi (-fprofile-arcs): Document profiling support
22282 for {cd}tors and C++ {cd}tors.
22284 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
22286 * config/s390/s390.md ("*setmem_long_and")
22287 ("*setmem_long_and_31z"): Use zero_extend instead of and.
22289 2017-01-26 Martin Sebor <msebor@redhat.com>
22291 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
22294 2017-01-26 Martin Sebor <msebor@redhat.com>
22296 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
22297 HAVE_DFmode before using XFmode or DFmode.
22298 (parse_directive): Avoid using the z length modifier to avoid
22299 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
22301 PR middle-end/78703
22302 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
22303 to accept adjustment as an array.
22304 (get_int_range): New function.
22305 (struct directive): Make width and prec arrays.
22306 (directive::set_width, directive::set_precision): Call get_int_range.
22307 (format_integer, format_floating): Handle width and precision ranges.
22308 (format_string, parse_directive): Same.
22310 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22313 * dwarf2out.c (generate_skeleton_bottom_up): For children with
22314 comdat_type_p set, just clone them, but keep the children in the
22318 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
22319 which have direct callers with -fvar-tracking-assignments enabled
22321 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
22322 inside of type units.
22324 2017-01-26 Martin Sebor <msebor@redhat.com>
22326 PR middle-end/78703
22327 * gimple-ssa-sprintf.c (struct result_range): Add likely and
22329 (struct format_result): Replace number_chars, number_chars_min,
22330 and number_chars_max with a single member of struct result_range.
22332 (format_result::operator+=): Adjust.
22333 (struct fmtresult): Remove bounded. Handle likely and unlikely
22335 (fmtresult::adjust_for_width_or_precision): New function.
22336 (fmtresult:type_max_digits): New function.
22337 (bytes_remaining): Handle likely and unlikely counters.
22338 (min_bytes_remaining): Remove.
22339 (format_percent): Simplify.
22340 (format_integer, format_floating): Set likely and unlikely counters.
22341 (get_string_length, format_character, format_string): Same.
22342 (format_plain, should_warn_p): New function.
22343 (maybe_warn): Call should_warn_p. Update diagnostic messages
22344 and handle those for all directives, including plain strings.
22345 (format_directive): Handle likely and unlikely counters.
22346 Remove unnecessary quoting from diagnostics. Add an informational
22348 (add_bytes): Remove.
22349 (pass_sprintf_length::compute_format_length): Simplify.
22350 (try_substitute_return_value): Handle likely and unlikely counters.
22352 2017-01-26 Carl Love <cel@us.ibm.com>
22354 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
22355 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
22357 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
22360 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
22361 endianess for subregs into account.
22362 * lra-constraints.c (lra_constraints): Do risky transformations
22363 always on the first iteration.
22364 * lra-lives.c (check_pseudos_live_through_calls): Add arg
22365 last_call_used_reg_set.
22366 (process_bb_lives): Define and use last_call_used_reg_set.
22367 * lra.c (lra): Always continue after lra_constraints on the first
22370 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
22372 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
22374 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
22376 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22378 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
22379 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
22380 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
22381 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
22382 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
22383 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
22384 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
22385 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
22386 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
22388 2017-01-26 Marek Polacek <polacek@redhat.com>
22391 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
22392 for the third operand.
22394 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22396 PR middle-end/79236
22397 * omp-low.c (struct omp_context): Add simt_stmt field.
22398 (scan_omp_for): Return omp_context *.
22399 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
22400 context to the _simt_ SIMD stmt.
22401 (lower_omp_for): For combined SIMD with sibling _simt_
22402 SIMD, make sure to use the same decls in _looptemp_
22403 clauses as in the sibling.
22405 2017-01-26 David Sherwood <david.sherwood@arm.com>
22407 PR middle-end/79212
22408 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
22411 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22414 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
22415 emit fld b; fld a; if possible.
22417 * brig-builtins.def: Update copyright years.
22418 * config/arm/arm_acle_builtins.def: Update copyright years.
22420 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
22423 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
22424 constraint instead of o for the stxsd instruction.
22426 2017-01-25 Carl Love <cel@us.ibm.com>
22428 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
22429 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
22431 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
22433 * doc/invoke.texi (C++ Dialect Options): Fix typo.
22435 2017-01-25 Richard Biener <rguenther@suse.de>
22437 PR tree-optimization/69264
22438 * target.def (vector_alignment_reachable): Improve documentation.
22439 * doc/tm.texi: Regenerate.
22440 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
22442 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
22443 earlier changes with respect to TYPE_USER_ALIGN.
22444 (vector_alignment_reachable_p): Likewise. Improve dumping.
22446 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22449 * config/arm/arm.md (xordi3): Force constant operand into a register
22452 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22454 * doc/invoke.texi (-fstore-merging): Correct default optimization
22455 levels at which it is enabled.
22456 (-O): Move -fstore-merging from list to...
22459 2017-01-25 Richard Biener <rguenther@suse.de>
22462 * omp-expand.c: Include debug.h.
22463 (expand_omp_taskreg): Make sure to generate early debug before
22464 outlining anything from a function.
22465 (expand_omp_target): Likewise.
22466 (grid_expand_target_grid_body): Likewise.
22468 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
22471 * asan.c (get_translation_unit_decl): New function.
22472 (asan_add_global): Extract modules file name from globals
22473 TRANSLATION_UNIT_DECL name.
22475 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
22478 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
22479 for long calls with APCS frame and VFP.
22481 2017-01-24 David Malcolm <dmalcolm@redhat.com>
22483 * cfg.c (original_copy_tables_initialized_p): New function.
22484 * cfg.h (original_copy_tables_initialized_p): New decl.
22485 * cfgrtl.c (relink_block_chain): Guard the call to
22486 free_original_copy_tables with a call to
22487 original_copy_tables_initialized_p.
22488 * cgraph.h (symtab_node::native_rtl_p): New decl.
22489 * cgraphunit.c (symtab_node::native_rtl_p): New function.
22490 (symtab_node::needed_p): Don't assert for early assembly output
22491 for __RTL functions.
22492 (cgraph_node::finalize_function): Set "force_output" for __RTL
22494 (cgraph_node::analyze): Bail out early for __RTL functions.
22495 (analyze_functions): Update assertion to support __RTL functions.
22496 (cgraph_node::expand): Bail out early for __RTL functions.
22497 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
22499 * function.h (struct function): Update comment for field
22501 * gimple-expr.c: Include "tree-pass.h".
22502 (gimple_has_body_p): Return false for __RTL functions.
22503 * Makefile.in (OBJS): Add run-rtl-passes.o.
22504 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
22506 (gcc::pass_manager::get_clean_slate): New accessor.
22507 * passes.c: Include "insn-addr.h".
22508 (should_skip_pass_p): Add logging. Update logic for running
22509 "expand" to be compatible with both __GIMPLE and __RTL. Guard
22510 property-provider override so it is only done for gimple passes.
22512 (skip_pass): New function.
22513 (execute_one_pass): Call skip_pass when skipping passes.
22514 * read-md.c (md_reader::read_char): Support filtering
22515 the input to a subset of line numbers.
22516 (md_reader::md_reader): Initialize fields
22517 m_first_line and m_last_line.
22518 (md_reader::read_file_fragment): New function.
22519 * read-md.h (md_reader::read_file_fragment): New decl.
22520 (md_reader::m_first_line): New field.
22521 (md_reader::m_last_line): New field.
22522 * read-rtl-function.c (function_reader::create_function): Only
22523 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
22524 curr_properties. Set DECL_INITIAL to a dummy block.
22525 (read_rtl_function_body_from_file_range): New function.
22526 * read-rtl-function.h (read_rtl_function_body_from_file_range):
22528 * run-rtl-passes.c: New file.
22529 * run-rtl-passes.h: New file.
22531 2017-01-24 Jeff Law <law@redhat.com>
22533 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
22536 2017-01-24 Bin Cheng <bin.cheng@arm.com>
22538 PR tree-optimization/79159
22539 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
22540 (record_nonwrapping_iv): Improve boundary using above function if no
22541 value range information.
22543 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
22544 Martin Jambor <mjambor@suse.cz>
22546 * brig-builtins.def: New file.
22547 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
22548 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
22549 (DEF_HSAIL_SAT_BUILTIN): Likewise.
22550 (DEF_HSAIL_INTR_BUILTIN): Likewise.
22551 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
22552 * builtin-types.def (BT_INT8): New.
22553 (BT_INT16): Likewise.
22554 (BT_UINT8): Likewise.
22555 (BT_UINT16): Likewise.
22556 (BT_FN_ULONG): Likewise.
22557 (BT_FN_UINT_INT): Likewise.
22558 (BT_FN_UINT_ULONG): Likewise.
22559 (BT_FN_UINT_LONG): Likewise.
22560 (BT_FN_UINT_PTR): Likewise.
22561 (BT_FN_ULONG_PTR): Likewise.
22562 (BT_FN_INT8_FLOAT): Likewise.
22563 (BT_FN_INT16_FLOAT): Likewise.
22564 (BT_FN_UINT32_FLOAT): Likewise.
22565 (BT_FN_UINT16_FLOAT): Likewise.
22566 (BT_FN_UINT8_FLOAT): Likewise.
22567 (BT_FN_UINT64_FLOAT): Likewise.
22568 (BT_FN_UINT16_UINT32): Likewise.
22569 (BT_FN_UINT32_UINT16): Likewise.
22570 (BT_FN_UINT16_UINT16_UINT16): Likewise.
22571 (BT_FN_INT_PTR_INT): Likewise.
22572 (BT_FN_UINT_PTR_UINT): Likewise.
22573 (BT_FN_LONG_PTR_LONG): Likewise.
22574 (BT_FN_ULONG_PTR_ULONG): Likewise.
22575 (BT_FN_VOID_UINT64_UINT64): Likewise.
22576 (BT_FN_UINT8_UINT8_UINT8): Likewise.
22577 (BT_FN_INT8_INT8_INT8): Likewise.
22578 (BT_FN_INT16_INT16_INT16): Likewise.
22579 (BT_FN_INT_INT_INT): Likewise.
22580 (BT_FN_UINT_FLOAT_UINT): Likewise.
22581 (BT_FN_FLOAT_UINT_UINT): Likewise.
22582 (BT_FN_ULONG_UINT_UINT): Likewise.
22583 (BT_FN_ULONG_UINT_PTR): Likewise.
22584 (BT_FN_ULONG_ULONG_ULONG): Likewise.
22585 (BT_FN_UINT_UINT_UINT): Likewise.
22586 (BT_FN_VOID_UINT_PTR): Likewise.
22587 (BT_FN_UINT_UINT_PTR: Likewise.
22588 (BT_FN_UINT32_UINT64_PTR): Likewise.
22589 (BT_FN_INT_INT_UINT_UINT): Likewise.
22590 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
22591 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
22592 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
22593 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
22594 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
22595 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
22596 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
22597 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
22598 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
22599 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
22600 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
22601 * doc/frontends.texi: List BRIG FE.
22602 * doc/install.texi (Testing): Add BRIG tesring requirements.
22603 * doc/invoke.texi (Overall Options): Mention BRIG.
22604 * doc/standards.texi (Standards): Doucment BRIG HSA version.
22606 2017-01-24 Richard Biener <rguenther@suse.de>
22608 PR translation/79208
22609 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
22611 2017-01-24 Martin Jambor <mjambor@suse.cz>
22614 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
22615 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
22616 and known_contexts.
22618 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
22620 PR middle-end/79123
22621 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
22622 casts from signed to unsigned really don't have a range.
22624 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
22626 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
22627 GMP_RNDx for compatiblity.
22629 2017-01-24 Martin Liska <mliska@suse.cz>
22632 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
22633 that would prevent us to call alloca with -1 as argument.
22635 2017-01-24 Jakub Jelinek <jakub@redhat.com>
22637 * dwarf2out.c (output_compilation_unit_header, output_file_names):
22638 Avoid -Wformat-security warning.
22640 2017-01-23 Andrew Pinski <apinski@cavium.com>
22642 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
22645 2017-01-23 Martin Sebor <msebor@redhat.com>
22647 PR middle-end/78703
22648 * gimple-ssa-sprintf.c (warn_level): New global.
22649 (format_integer): Use it here and throughout the rest of the file.
22650 Use the same switch to compute sign as base.
22651 (maybe_warn): New function.
22652 (format_directive): Factor out warnings into maybe_warn.
22653 Add debugging output. Use warn_level.
22654 (add_bytes): Use warn_level.
22655 (pass_sprintf_length::compute_format_length): Add debugging output.
22656 (try_substitute_return_value): Same.
22657 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
22659 PR middle-end/78703
22660 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
22661 (struct fmtresult, format_integer, format_floating): Adjust.
22662 (fmtresult::fmtresult): Set max correctly in two argument ctor.
22663 (get_string_length, format_string,format_directive): Same.
22664 (pass_sprintf_length::compute_format_length): Same.
22665 (try_substitute_return_value): Simplify slightly.
22667 PR middle-end/78703
22668 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
22669 (fmtresult::operator+=): Outlined.
22670 (struct fmtresult): Add ctors.
22671 (struct conversion_spec): Rename...
22672 (struct directive): ...to this. Add and remove data members.
22673 (directive::set_width, directive::set_precision): New functions.
22674 (format_percent): Use fmtresult ctor.
22675 (get_width_and_precision): Remove.
22676 (format_integer): Make naming changes. Avoid computing width and
22678 (format_floating): Same. Adjust indentation.
22679 (format_character, format_none): New functions.
22680 (format_string): Moved character handling to format_character.
22681 (format_directive): Remove arguments, change return type.
22682 (parse_directive): New function.
22683 (pass_sprintf_length::compute_format_length): Move directive
22684 parsing to parse_directive.
22686 2017-01-23 Jakub Jelinek <jakub@redhat.com>
22688 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
22689 (assign_assembler_name_if_needed): ... this.
22690 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
22691 (assign_assembler_name_if_needed): ... this.
22692 (free_lang_data_in_cgraph): Adjust callers.
22693 * cgraphunit.c (cgraph_node::analyze): Likewise.
22694 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
22697 2017-01-23 Richard Biener <rguenther@suse.de>
22699 PR tree-optimization/79088
22700 PR tree-optimization/79188
22701 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
22702 resetting loop bounds after last path deletion. Reset loop
22703 bounds of the target loop, make code match the comments.
22704 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
22705 Make sure loops need no fixups.
22707 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22709 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
22710 exponent support with double type for first argument.
22711 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
22712 type returned by __builtin_vec_extract_sig,
22713 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
22714 functions from "vector int" to "vector unsigned int" or from
22715 "vector long long int" to "vector unsigned long long int".
22716 Changed type returned by __builtin_vec_extract_exp,
22717 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
22718 functions from "vector int" to "vector unsigned int" or from
22719 "vector long long int" to "vector unsigned long long int".
22720 Changed return type of __builtin_vec_test_data_class,
22721 __builtin_vec_test_data_class_sp, and
22722 __builtin_vec_test_data_class_dp from "vector int" to
22723 "vector bool int" or from "vector long long int" to "vector bool
22724 long long int" and changed second argument type from "unsigned
22725 int" to "int". Added new overloaded function forms "vector float
22726 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
22727 "vector float __builtin_vec_insert_exp_sp (vector float, vector
22728 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
22729 double, vector unsigned long long int)" and "vector double
22730 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
22731 long int)". Changed return type of
22732 __builtin_scalar_test_data_class and
22733 __builtin_scalar_test_data_class_sp and
22734 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
22735 int" and changed second argument from "unsigned int" to "int".
22736 Changed type returned by __builtin_scalar_test_neg,
22737 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
22738 from "int" to "bool int". Added new overloaded function form
22739 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
22740 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
22741 exponent double-precision with floating point first argument.
22742 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
22743 documentation of scalar_test_data_class, scalar_test_neg,
22744 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
22745 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
22746 vec_test_data_class built-in functions to reflect refinements in
22747 their type signatures.
22749 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
22751 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
22753 (aarch64_elf_asm_destructor): Likewise.
22755 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
22757 PR rtl-optimization/78634
22758 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
22759 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
22760 * ifcvt.c (noce_try_cmove): Add missing cost check.
22762 PR rtl-optimization/71724
22763 * combine.c (if_then_else_cond): Look for situations where it is
22764 beneficial to undo the work of one of the recursive calls.
22766 2017-01-23 Bin Cheng <bin.cheng@arm.com>
22768 PR tree-optimization/70754
22769 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
22770 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
22771 combined stmt before it if not NULL.
22772 (combine_chains): Process refs reversely and compute dominance point
22775 2017-01-23 Martin Liska <mliska@suse.cz>
22777 PR tree-optimization/79196
22778 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
22779 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
22781 (strlen_optimize_stmt): Call the renamed function.
22783 2017-01-23 Michael Matz <matz@suse.de>
22785 PR tree-optimization/78384
22786 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
22788 2017-01-23 Richard Biener <rguenther@suse.de>
22790 PR tree-optimization/79186
22791 * tree-vrp.c (register_new_assert_for): Make sure we've seen
22792 both incoming edges before moving an assert.
22794 2017-01-23 Martin Jambor <mjambor@suse.cz>
22796 * ipa-prop.c (load_from_param_1): Removed.
22797 (load_from_unmodified_param): Bits from load_from_param_1 put back
22799 (load_from_param): Removed.
22800 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
22801 with stmt. Reverted back to use of load_from_unmodified_param.
22803 2017-01-23 Martin Jambor <mjambor@suse.cz>
22806 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
22807 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
22808 field a pointer to garbage collected vector, mark lattices and
22809 ipcp_orig_node with GTY((skip)).
22810 (ipa_get_param_count): Adjust to descriptors being a pointer.
22811 (ipa_get_param): Likewise.
22812 (ipa_get_type): Likewise.
22813 (ipa_get_param_move_cost): Likewise.
22814 (ipa_set_param_used): Likewise.
22815 (ipa_get_controlled_uses): Likewise.
22816 (ipa_set_controlled_uses): Likewise.
22817 (ipa_is_param_used): Likewise.
22818 (ipa_node_params_t): Move into garbage collector. New methods insert
22820 (ipa_node_params_sum): Annotate wth GTY(()).
22821 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
22823 (ipa_load_from_parm_agg): Adjust declaration.
22824 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
22825 * ipa-profile.c (ipa_profile): Likewise.
22826 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
22827 (ipa_populate_param_decls): Make descriptors parameter garbage
22829 (ipa_dump_param): Adjust to descriptors being a pointer.
22830 (ipa_alloc_node_params): Likewise.
22831 (ipa_initialize_node_params): Likewise.
22832 (load_from_param_1): Make descriptors parameter garbage collected.
22833 (load_from_unmodified_param): Likewise.
22834 (load_from_param): Likewise.
22835 (ipa_load_from_parm_agg): Likewise.
22836 (ipa_node_params::~ipa_node_params): Removed.
22837 (ipa_free_all_node_params): Remove call to delete operator.
22838 (ipa_node_params_t::insert): New.
22839 (ipa_node_params_t::remove): Likewise.
22840 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
22841 copy known_csts and known_contexts vectors.
22842 (ipa_read_node_info): Adjust to descriptors being a pointer.
22843 (ipcp_modif_dom_walker): Make m_descriptors field garbage
22845 (ipcp_transform_function): Make descriptors variable garbage
22848 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
22850 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
22851 * config/i386/avx512dqintrin.h: Ditto.
22852 * config/i386/avx512fintrin.h: Ditto.
22853 * config/i386/i386.c: Handle new builtins.
22854 * config/i386/i386-builtin.def: Add new builtins.
22855 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
22856 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
22858 2017-01-23 Jakub Jelinek <jakub@redhat.com>
22859 Martin Liska <mliska@suse.cz>
22861 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
22862 * asan.c (asan_expand_poison_ifn): Support stores and use
22863 appropriate ASAN report function.
22864 * internal-fn.c (expand_ASAN_POISON_USE): New function.
22865 * internal-fn.def (ASAN_POISON_USE): Declare.
22866 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
22867 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
22868 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
22869 ASAN_POISON calls w/o LHS.
22870 * tree-ssa.c (execute_update_addresses_taken): Create clobber
22871 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
22872 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
22873 * gimplify.c (asan_poison_variables): Add attribute
22874 use_after_scope_memory to variables that really needs to live
22876 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
22877 having the attribute.
22879 2017-01-23 Martin Liska <mliska@suse.cz>
22881 * asan.c (create_asan_shadow_var): New function.
22882 (asan_expand_poison_ifn): Likewise.
22883 * asan.h (asan_expand_poison_ifn): New declaration.
22884 * internal-fn.c (expand_ASAN_POISON): Likewise.
22885 * internal-fn.def (ASAN_POISON): New builtin.
22886 * sanopt.c (pass_sanopt::execute): Expand
22887 asan_expand_poison_ifn.
22888 * tree-inline.c (copy_decl_for_dup_finish): Make function
22890 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
22891 * tree-ssa.c (is_asan_mark_p): New function.
22892 (execute_update_addresses_taken): Rewrite local variables
22893 (identified just by use-after-scope as addressable) into SSA.
22895 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
22897 * doc/install.texi (Specific): opensource.apple.com uses https
22898 now. Remove trailing slash.
22900 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
22902 * README.Portability: Remove note on an Irix compatibility issue.
22904 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
22906 * gcov.c (INCLUDE_ALGORITHM): Define.
22907 (INCLUDE_VECTOR): Define.
22908 No longer include <vector> and <algorithm> directly.
22910 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
22912 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
22914 * doc/invoke.texi (Code Gen Options): Ditto.
22916 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
22919 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
22921 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
22923 rtl-optimization/79125
22924 * cprop.c (local_cprop_pass): Handle cases where we make an
22925 unconditional trap.
22927 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
22931 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
22932 read from, for big endian.
22934 2017-01-20 Jiong Wang <jiong.wang@arm.com>
22936 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
22937 register pauth builtins for LP64 only.
22939 2017-01-20 Marek Polacek <polacek@redhat.com>
22942 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
22945 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
22947 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
22949 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
22950 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
22951 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
22953 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22956 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
22957 in big-endian mode when they are not a single duplicated value.
22959 2017-01-20 Richard Biener <rguenther@suse.de>
22961 * BASE-VER: Bump to 7.0.1.
22963 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
22965 * omp-low.c (omplow_simd_context): New struct. Use it...
22966 (lower_rec_simd_input_clauses): ...here and...
22967 (lower_rec_input_clauses): ...here to hold common data. Adjust all
22968 references to idx, lane, max_vf, is_simt.
22970 2017-01-20 Graham Markall <graham.markall@embecosm.com>
22972 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
22975 2017-01-20 Martin Jambor <mjambor@suse.cz>
22977 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
22978 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
22980 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
22981 (GTFILES): Rename hsa.c to hsa-common.c.
22982 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
22983 * hsa-dump.c: Likewise.
22984 * hsa-gen.c: Likewise.
22985 * hsa-regalloc.c: Likewise.
22986 * ipa-hsa.c: Likewise.
22987 * omp-expand.c: Likewise.
22988 * omp-low.c: Likewise.
22989 * toplev.c: Likewise.
22991 2017-01-20 Marek Polacek <polacek@redhat.com>
22994 * doc/invoke.texi: Document -Wduplicated-branches.
22995 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
22996 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
22997 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
22998 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
22999 return 0 only when not OEP_LEXICOGRAPHIC.
23000 (fold_build_cleanup_point_expr): Use the expression
23001 location when building CLEANUP_POINT_EXPR.
23002 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23003 * tree.c (add_expr): Handle error_mark_node.
23005 2017-01-20 Martin Liska <mliska@suse.cz>
23008 * tree-profile.c (init_ic_make_global_vars): Do not call
23010 (gimple_init_gcov_profiler): Likewise.
23012 2017-01-20 Martin Liska <mliska@suse.cz>
23015 * cgraph.h (maybe_create_reference): Remove argument and
23017 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23019 * ipa-cp.c (create_specialized_node): Likewise.
23020 * symtab.c (symtab_node::maybe_create_reference): Handle
23021 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23023 2017-01-20 Martin Liska <mliska@suse.cz>
23025 * read-rtl-function.c (function_reader::create_function): Use
23026 build_decl instread of build_decl_stat.
23028 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23030 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23031 * config/i386/avx512dqintrin.h: Ditto.
23032 * config/i386/avx512fintrin.h: Ditto.
23033 * config/i386/i386-builtin-types.def: Add new types.
23034 * config/i386/i386.c: Handle new types.
23035 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23036 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23037 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23038 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23039 (__builtin_ia32_kshiftridi): New.
23040 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23042 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23046 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23047 define to rs6000_init_stack_protect_guard.
23048 (rs6000_init_stack_protect_guard): New function.
23050 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23051 Yunqiang Su <yunqiang.su@imgtec.com>
23053 * config.gcc (supported_defaults): Add madd4.
23054 (with_madd4): Add validation.
23055 (all_defaults): Add madd4.
23056 * config/mips/mips.opt (mmadd4): New option.
23057 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23059 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23061 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23062 (ISA_HAS_FUSED_MADD4): Likewise.
23063 * doc/invoke.texi (-mmadd4): Document the new option.
23064 * doc/install.texi (--with-madd4): Document the new option.
23066 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23068 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23069 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23070 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23071 (aarch64_init_pauth_hint_builtins): New.
23072 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23073 (aarch64_expand_builtin): Expand new builtins.
23075 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23077 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23078 * combine-stack-adj.c (no_unhandled_cfa): Handle
23079 REG_CFA_TOGGLE_RA_MANGLE.
23080 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23081 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23082 info for return address signing.
23083 (aarch64_expand_epilogue): Likewise.
23085 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23087 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23088 * config/aarch64/aarch64-protos.h
23089 (aarch64_return_address_signing_enabled): New declaration.
23090 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23092 (aarch64_expand_prologue): Sign return address before it's pushed onto
23094 (aarch64_expand_epilogue): Authenticate return address fetched from
23096 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23097 (aarch64_attributes): New function attributes for "sign-return-address".
23098 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23099 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
23100 ("*do_return"): Generate combined instructions according to key index.
23101 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
23102 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
23104 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
23105 * config/aarch64/aarch64.opt (msign-return-address=): New.
23106 * doc/extend.texi (AArch64 Function Attributes): Documents
23107 "sign-return-address=".
23108 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
23110 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23112 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
23113 overall option summary.
23115 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23117 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
23118 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
23119 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
23120 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
23122 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
23124 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
23125 -mpower9-minmax by default for -mcpu=power9.
23126 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
23127 128-bit floating point.
23129 2017-01-20 Alan Modra <amodra@gmail.com>
23131 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
23132 optimizing for size.
23134 2017-01-20 Alan Modra <amodra@gmail.com>
23137 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
23138 for strcmp and strncmp from corresponding builtin decl.
23140 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23142 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
23143 instead of i386/rtems-64.h.
23144 * config/i386/rtems-64.h: Remove.
23146 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23150 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
23152 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
23154 2017-01-19 Tamar Christina <tamar.christina@arm.com>
23156 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
23157 Change int to HOST_WIDE_INT.
23158 * config/aarch64/aarch64-protos.h
23159 (aarch64_simd_gen_const_vector_dup): Likewise.
23160 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
23162 2017-01-19 David Malcolm <dmalcolm@redhat.com>
23164 * langhooks-def.h (lhd_type_for_size): New decl.
23165 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
23166 * langhooks.c (lhd_type_for_size): New function, taken from
23169 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
23171 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
23172 define_bypass for CR latency.
23173 (power9-cracked-alu): Update bypass latency and remove power9-branch.
23174 (power9-alu2): Add define_bypass for CR latency.
23176 (power9-mul): Update insn latency.
23177 (power9-mul-compare): Update insn latency, bypass latency and remove
23180 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23182 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
23184 * config/aarch64/aarch64.md
23185 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
23186 aarch64_nopcrelative_literal_loads.
23187 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
23189 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
23191 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
23192 TARGET_LOONGSON_3A.
23193 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
23195 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
23198 * config.gcc (supported_defaults): Add lxc1-sxc1.
23199 (with_lxc1_sxc1): Add validation.
23200 (all_defaults): Add lxc1-sxc1.
23201 * config/mips/mips.opt (mlxc1-sxc1): New option.
23202 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23204 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23205 __mips_no_lxc1_sxc1.
23206 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
23207 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
23208 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
23210 2017-01-19 Richard Biener <rguenther@suse.de>
23212 PR tree-optimization/72488
23213 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
23214 sure to restore SSA info.
23215 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
23217 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
23219 PR rtl-optimization/79121
23220 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
23221 of the inner type when shifting an extended value.
23223 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23226 * symtab.c (symtab_node::equal_address_to): Fix comparing of
23227 interposable aliases.
23229 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
23232 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
23233 Use the evmergelohi instruction.
23234 (mov_si<mode>_e500_subreg4_2_le): Likewise.
23235 (mov_sitf_e500_subreg8_2_be): Likewise.
23236 (mov_sitf_e500_subreg12_2_le): Likewise.
23237 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
23238 (mov_si<mode>_e500_subreg4_2_be): Likewise.
23239 (mov_sitf_e500_subreg8_2_le): Likewise.
23240 (mov_sitf_e500_subreg12_2_be): Likewise.
23242 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23244 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
23245 attribute from vecsimple to vecperm.
23246 (altivec_vbpermq2): Likewise.
23248 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23251 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
23253 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23254 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
23255 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
23256 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
23257 case where N arg is SIZE_MAX.
23258 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
23259 (cmpstrsi): Add pattern.
23261 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23263 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23264 __builtin_vec_revb builtins.
23265 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
23266 built-in functions to support generation of the ISA 3.0 XXBR<x>
23267 vector byte reverse instructions.
23268 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
23269 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
23270 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
23271 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
23272 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
23273 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
23274 (P9V_BUILTIN_VEC_REVB): Likewise.
23275 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
23276 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
23277 (p9_xxbrq_v16qi): Likewise.
23278 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
23279 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
23280 (p9_xxbrh_v8hi): Likewise.
23281 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
23282 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
23283 vec_revb built-in functions.
23285 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
23287 PR rtl-optimization/78952
23288 * config/i386/i386.md (any_extract): New code iterator.
23289 (*insvqi_2): Use any_extract for source operand.
23290 (*insvqi_3): Use any_shiftrt for source operand.
23292 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23294 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
23296 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
23298 2017-01-18 Matthias Klose <doko@ubuntu.com>
23300 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
23302 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23304 * config/rs6000/altivec.h (vec_bperm): Change #define.
23305 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
23306 (altivec_vbpermq2): New define_insn.
23307 (altivec_vbpermd): Likewise.
23308 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
23309 function interface.
23310 (VBPERMD): Likewise.
23311 (VBPERM): New polymorphic function interface.
23312 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
23313 Add entries for P9V_BUILTIN_VEC_VBPERM.
23314 * doc/extend.texi: Add interfaces for vec_bperm.
23316 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23318 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
23319 first letter of error messages.
23320 (s390_resolve_overloaded_builtin): Likewise.
23321 * config/s390/s390.c (s390_expand_builtin): Likewise.
23322 (s390_invalid_arg_for_unprototyped_fn): Likewise.
23323 (s390_valid_target_attribute_inner_p): Likewise.
23324 * config/s390/s390.md ("tabort"): Likewise.
23326 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
23328 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
23329 (ISA_AVOID_DIV_HILO): New macro.
23330 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
23331 (ISA_HAS_DDIV): Likewise.
23333 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23335 * doc/invoke.texi (fabi-version): Correct number of occurrences.
23337 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23339 * doc/invoke.texi (fabi-version): Spelling fix.
23341 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23344 * doc/invoke.texi (fabi-version): Mention mangling fix for
23347 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23350 * doc/invoke.texi (fabi-version): Document discriminator mangling.
23352 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
23355 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
23356 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
23358 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
23360 (stack_protect_test): Ditto.
23361 * config/rs6000/rs6000.opt (mstack-protector-guard=,
23362 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
23364 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
23365 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
23366 -mstack-protector-guard-offset=.
23367 (RS/6000 and PowerPC Options): Ditto.
23369 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23371 * config/i386/i386.h (MASK_CLASS_P): New define.
23372 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
23373 there are no registers from different register sets also when
23374 mask registers are used. Update function comment.
23375 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
23376 to (*k/*r) and (*k/*km) alternatives.
23378 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
23380 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
23381 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
23382 (EH_RETURN_HANDLER_RTX): New define.
23383 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
23384 Force frame pointer in EH return functions.
23385 (aarch64_expand_epilogue): Add barrier for eh_return.
23386 (aarch64_final_eh_return_addr): Remove.
23387 (aarch64_eh_return_handler_rtx): New function.
23388 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
23390 (aarch64_eh_return_handler_rtx): New prototype.
23392 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23394 * config/rs6000/altivec.h (vec_rlmi): New #define.
23395 (vec_vrlnm): Likewise.
23396 (vec_rlnm): Likewise.
23397 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
23398 (UNSPEC_VRLNM): Likewise.
23399 (VIlong): New mode iterator.
23400 (altivec_vrl<VI_char>mi): New define_insn.
23401 (altivec_vrl<VI_char>nm): Likewise.
23402 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
23404 (VRLDNM): Likewise.
23405 (RLNM): New polymorphic function entry.
23406 (VRLWMI): New monomorphic function entry.
23407 (VRLDMI): Likewise.
23408 (RLMI): New polymorphic function entry.
23409 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
23410 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
23411 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
23414 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23417 * dwarf2out.c (field_byte_offset): Restore the
23418 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
23419 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
23420 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
23423 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23426 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
23428 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23431 * dwarf2out.c (add_data_member_location_attribute): For constant
23432 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
23433 instead of DW_AT_data_member_location, DW_AT_bit_offset and
23434 DW_AT_byte_size attributes.
23436 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23438 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
23439 after forcing to constant memory when the code model is medium.
23441 2017-01-17 Julia Koval <julia.koval@intel.com>
23444 * config/i386/avx512fintrin.h
23445 (_mm512_i32gather_ps): Change __addr type to void const*.
23446 (_mm512_mask_i32gather_ps): Ditto.
23447 (_mm512_i32gather_pd): Ditto.
23448 (_mm512_mask_i32gather_pd): Ditto.
23449 (_mm512_i64gather_ps): Ditto.
23450 (_mm512_mask_i64gather_ps): Ditto.
23451 (_mm512_i64gather_pd): Ditto.
23452 (_mm512_mask_i64gather_pd): Ditto.
23453 (_mm512_i32gather_epi32): Ditto.
23454 (_mm512_mask_i32gather_epi32): Ditto.
23455 (_mm512_i32gather_epi64): Ditto.
23456 (_mm512_mask_i32gather_epi64): Ditto.
23457 (_mm512_i64gather_epi32): Ditto.
23458 (_mm512_mask_i64gather_epi32): Ditto.
23459 (_mm512_i64gather_epi64): Ditto.
23460 (_mm512_mask_i64gather_epi64): Ditto.
23461 (_mm512_i32scatter_ps): Change __addr type to void*.
23462 (_mm512_mask_i32scatter_ps): Ditto.
23463 (_mm512_i32scatter_pd): Ditto.
23464 (_mm512_mask_i32scatter_pd): Ditto.
23465 (_mm512_i64scatter_ps): Ditto.
23466 (_mm512_mask_i64scatter_ps): Ditto.
23467 (_mm512_i64scatter_pd): Ditto.
23468 (_mm512_mask_i64scatter_pd): Ditto.
23469 (_mm512_i32scatter_epi32): Ditto.
23470 (_mm512_mask_i32scatter_epi32): Ditto.
23471 (_mm512_i32scatter_epi64): Ditto.
23472 (_mm512_mask_i32scatter_epi64): Ditto.
23473 (_mm512_i64scatter_epi32): Ditto.
23474 (_mm512_mask_i64scatter_epi32): Ditto.
23475 (_mm512_i64scatter_epi64): Ditto.
23476 (_mm512_mask_i64scatter_epi64): Ditto.
23477 * config/i386/avx512pfintrin.h
23478 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
23479 (_mm512_mask_prefetch_i32gather_ps): Ditto.
23480 (_mm512_mask_prefetch_i64gather_pd): Ditto.
23481 (_mm512_mask_prefetch_i64gather_ps): Ditto.
23482 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
23483 (_mm512_prefetch_i32scatter_ps): Ditto.
23484 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
23485 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
23486 (_mm512_prefetch_i64scatter_pd): Ditto.
23487 (_mm512_prefetch_i64scatter_ps): Ditto.
23488 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
23489 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
23490 * config/i386/avx512vlintrin.h
23491 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
23492 (_mm_mmask_i32gather_ps): Ditto.
23493 (_mm256_mmask_i32gather_pd): Ditto.
23494 (_mm_mmask_i32gather_pd): Ditto.
23495 (_mm256_mmask_i64gather_ps): Ditto.
23496 (_mm_mmask_i64gather_ps): Ditto.
23497 (_mm256_mmask_i64gather_pd): Ditto.
23498 (_mm_mmask_i64gather_pd): Ditto.
23499 (_mm256_mmask_i32gather_epi32): Ditto.
23500 (_mm_mmask_i32gather_epi32): Ditto.
23501 (_mm256_mmask_i32gather_epi64): Ditto.
23502 (_mm_mmask_i32gather_epi64): Ditto.
23503 (_mm256_mmask_i64gather_epi32): Ditto.
23504 (_mm_mmask_i64gather_epi32): Ditto.
23505 (_mm256_mmask_i64gather_epi64): Ditto.
23506 (_mm_mmask_i64gather_epi64): Ditto.
23507 (_mm256_i32scatter_ps): Change __addr type to void*.
23508 (_mm256_mask_i32scatter_ps): Ditto.
23509 (_mm_i32scatter_ps): Ditto.
23510 (_mm_mask_i32scatter_ps): Ditto.
23511 (_mm256_i32scatter_pd): Ditto.
23512 (_mm256_mask_i32scatter_pd): Ditto.
23513 (_mm_i32scatter_pd): Ditto.
23514 (_mm_mask_i32scatter_pd): Ditto.
23515 (_mm256_i64scatter_ps): Ditto.
23516 (_mm256_mask_i64scatter_ps): Ditto.
23517 (_mm_i64scatter_ps): Ditto.
23518 (_mm_mask_i64scatter_ps): Ditto.
23519 (_mm256_i64scatter_pd): Ditto.
23520 (_mm256_mask_i64scatter_pd): Ditto.
23521 (_mm_i64scatter_pd): Ditto.
23522 (_mm_mask_i64scatter_pd): Ditto.
23523 (_mm256_i32scatter_epi32): Ditto.
23524 (_mm256_mask_i32scatter_epi32): Ditto.
23525 (_mm_i32scatter_epi32): Ditto.
23526 (_mm_mask_i32scatter_epi32): Ditto.
23527 (_mm256_i32scatter_epi64): Ditto.
23528 (_mm256_mask_i32scatter_epi64): Ditto.
23529 (_mm_i32scatter_epi64): Ditto.
23530 (_mm_mask_i32scatter_epi64): Ditto.
23531 (_mm256_i64scatter_epi32): Ditto.
23532 (_mm256_mask_i64scatter_epi32): Ditto.
23533 (_mm_i64scatter_epi32): Ditto.
23534 (_mm_mask_i64scatter_epi32): Ditto.
23535 (_mm256_i64scatter_epi64): Ditto.
23536 (_mm256_mask_i64scatter_epi64): Ditto.
23537 (_mm_i64scatter_epi64): Ditto.
23538 (_mm_mask_i64scatter_epi64): Ditto.
23539 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
23540 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
23541 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
23542 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
23543 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
23544 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
23545 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
23546 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
23547 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
23548 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
23549 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
23550 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
23551 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
23552 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
23553 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
23554 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
23555 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
23556 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
23557 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
23558 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
23559 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
23560 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
23561 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
23562 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
23563 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
23564 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
23565 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
23566 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
23567 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
23568 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
23569 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
23570 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
23571 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
23572 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
23573 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
23574 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
23575 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
23576 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
23577 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
23578 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
23579 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
23580 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
23581 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
23582 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
23583 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
23584 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
23585 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
23586 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
23587 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
23588 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
23589 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
23590 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
23591 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
23592 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
23593 definitions accordingly.
23595 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
23596 Kuan-Lin Chen <kuanlinchentw@gmail.com>
23599 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
23602 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
23605 * ira-conflicts.c (ira_build_conflicts): Update total conflict
23606 hard regs for inner regno.
23608 2017-01-17 Martin Liska <mliska@suse.cz>
23611 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
23612 assumption and add comment.
23614 2017-01-17 Nathan Sidwell <nathan@acm.org>
23616 * ipa-visibility.c (localize_node): New function, broken out of ...
23617 (function_and_variable_visibility): ... here. Call it.
23619 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23621 PR middle-end/77445
23622 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
23623 correctly set frequency of oudgoing edge.
23624 (duplicate_thread_path): Fix profile updating.
23626 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23629 * configure.ac: Add GCC_BASE_VER.
23630 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
23631 version from BASE-VER file.
23632 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
23633 (gcc.o): Depend on $(BASEVER).
23634 * common.opt (dumpfullversion): New option.
23635 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
23636 * doc/invoke.texi: Document -dumpfullversion.
23637 * doc/install.texi: Document --with-gcc-major-version-only.
23638 * configure: Regenerated.
23640 2017-01-17 Richard Biener <rguenther@suse.de>
23642 PR tree-optimization/71433
23643 * tree-vrp.c (register_new_assert_for): Merge same asserts
23644 on all incoming edges.
23645 (process_assert_insertions_for): Handle insertions at the
23648 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
23650 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
23651 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
23653 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
23656 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
23659 2017-01-17 Alan Modra <amodra@gmail.com>
23662 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
23663 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
23664 symbolic stack limit when pic.
23666 2017-01-16 Martin Sebor <msebor@redhat.com>
23668 PR tree-optimization/78608
23669 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
23671 2017-01-16 Jeff Law <law@redhat.com>
23674 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
23675 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
23676 for several include directories that may be relative to sysroot.
23677 * config/i386/x-mingw32 (gplus_includedir): Define.
23678 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
23679 (native_system_includedir): Likewise.
23680 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
23681 override if TARGET_SYSTEM_ROOT is defined.
23682 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
23684 PR tree-optimization/79090
23685 PR tree-optimization/33562
23686 PR tree-optimization/61912
23687 PR tree-optimization/77485
23688 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
23689 and computed trims into the dump file.
23691 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23693 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
23695 2017-01-16 Jakub Jelinek <jakub@redhat.com>
23698 * gimplify.c (gimplify_init_constructor): If want_value and
23699 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
23703 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
23704 sequence. Formatting fixes.
23705 (doloop_optimize): Formatting fixes.
23708 * gcc.c (debug_level_greater_than_spec_func): New function.
23709 (static_spec_functions): Add debug-level-gt spec function.
23710 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
23712 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23713 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23714 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
23715 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
23716 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
23717 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
23719 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
23721 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
23722 QImode fixups to general and mask registers only.
23724 2017-01-16 Carl Love <cel@us.ibm.com>
23726 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
23727 for built-in functions
23728 vector signed char vec_nabs (vector signed char)
23729 vector signed short vec_nabs (vector signed short)
23730 vector signed int vec_nabs (vector signed int)
23731 vector signed long long vec_nabs (vector signed long long)
23732 vector float vec_nabs (vector float)
23733 vector double vec_nabs (vector double)
23734 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
23736 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
23737 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
23738 * doc/extend.texi: Update the documentation file for the new built-in
23741 2017-01-16 Martin Sebor <msebor@redhat.com>
23743 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
23746 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23748 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
23749 UNSPEC_VSX__XXSPLTD to require special splat handling.
23751 2017-01-16 David Malcolm <dmalcolm@redhat.com>
23754 * system.h: Poison strndup.
23756 2017-01-16 Alan Modra <amodra@gmail.com>
23759 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
23762 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
23764 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
23766 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
23768 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
23769 call recog here. Assert that INSN_CODE (insn) is non-negative.
23771 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23774 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
23776 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
23777 in the currently scheduled RTL fragment.
23779 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23781 PR rtl-optimization/78751
23782 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
23785 2017-01-14 Jeff Law <law@redhat.com>
23787 PR tree-optimization/79090
23788 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
23789 variable length stores.
23790 (compute_trims): Delete dead assignment to *trim_tail.
23791 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
23794 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
23796 PR rtl-optimization/78626
23797 PR rtl-optimization/78727
23798 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
23799 of a block, and split such blocks after everything else is finished.
23801 2017-01-14 Alan Modra <amodra@gmail.com>
23804 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
23805 target legitimate_combined_insn.
23806 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
23807 (rs6000_legitimate_combined_insn): New function.
23808 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
23810 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
23811 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
23812 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
23814 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
23816 * doc/frontends.texi (G++ and GCC): Remove references to Java.
23818 2017-01-13 Jeff Law <law@redhat.com>
23820 PR tree-optimization/33562
23821 PR tree-optimization/61912
23822 PR tree-optimization/77485
23823 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
23825 (delete_dead_assignment): Likewise.
23826 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
23827 statement to delete_dead_call and delete_dead_assignment.
23829 2017-01-13 David Malcolm <dmalcolm@redhat.com>
23832 * substring-locations.c (format_warning_va): Strengthen case 1 so
23833 that both endpoints of the substring must be within the format
23834 range for just the substring to be printed.
23836 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
23838 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
23839 * config/i386/i386.c (ix86_target_string): Add missing options
23840 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
23841 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
23842 flags_other and ix86_target_other to flags2_other. Display unknown
23844 (ix86_valid_target_attribute_inner_p): Add missing options and
23845 reorder options by implied ISAs, as in ix86_target_string.
23847 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
23849 * hash-table.h (hash_table::too_empty_p): New function.
23850 (hash_table::expand): Use it.
23851 (hash_table::traverse): Likewise.
23852 (hash_table::empty_slot): Use sizeof (value_type) instead of
23853 sizeof (PTR) to convert bytes to elements. Shrink the table
23854 if the current size is excessive for the current number of
23857 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
23859 * ira-costs.c (record_reg_classes): Break from the inner loop
23860 early once alt_fail is known to be true. Update outer loop
23861 handling accordingly.
23863 2017-01-13 Jeff Law <law@redhat.com>
23865 * tree-ssa-dse.c (decrement_count): New function.
23866 (increment_start_addr, maybe_trim_memstar_call): Likewise.
23867 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
23868 when we know the partially dead statement is a mem* function.
23870 PR tree-optimization/61912
23871 PR tree-optimization/77485
23872 * tree-ssa-dse.c: Include expr.h.
23873 (maybe_trim_constructor_store): New function.
23874 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
23876 PR tree-optimization/33562
23877 PR tree-optimization/61912
23878 PR tree-optimization/77485
23879 * doc/invoke.texi: Document new dse-max-object-size param.
23880 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
23881 * tree-ssa-dse.c: Include params.h.
23882 (dse_store_status): New enum.
23883 (initialize_ao_ref_for_dse): New, partially extracted from
23885 (valid_ao_ref_for_dse, normalize_ref): New.
23886 (setup_live_bytes_from_ref, compute_trims): Likewise.
23887 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
23888 (maybe_trim_partially_dead_store): Likewise.
23889 (maybe_trim_complex_store): Likewise.
23890 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
23891 Track what bytes live from the original store. Return tri-state
23892 for dead, partially dead or live.
23893 (dse_dom_walker): Add constructor, destructor and new private members.
23894 (delete_dead_call, delete_dead_assignment): New extracted from
23896 (dse_optimize_stmt): Make a member of dse_dom_walker.
23897 Use initialize_ao_ref_for_dse.
23899 PR tree-optimization/33562
23900 PR tree-optimization/61912
23901 PR tree-optimization/77485
23902 * sbitmap.h (bitmap_count_bits): Prototype.
23903 (bitmap_clear_range, bitmap_set_range): Likewise.
23904 * sbitmap.c (bitmap_clear_range): New function.
23905 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
23907 2017-01-13 Martin Liska <mliska@suse.cz>
23910 * function.c (set_cfun): Add new argument force.
23911 * function.h (set_cfun): Likewise.
23912 * ipa-inline-transform.c (inline_call): Use the function when
23913 strict alising from is dropped for function we inline to.
23915 2017-01-13 Richard Biener <rguenther@suse.de>
23917 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
23918 for dumping GIMPLE INTEGER_CSTs.
23920 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23922 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
23923 to 201112L since C++17.
23925 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23928 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
23929 if -fsanitize=kernel-address is present.
23931 2017-01-13 Richard Biener <rguenther@suse.de>
23933 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
23934 as _Literal ( type ) number in case usual suffixes do not
23935 preserve all information.
23937 2017-01-13 Richard Biener <rguenther@suse.de>
23939 PR tree-optimization/77283
23940 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
23941 and ssa-iterators.h.
23942 (is_feasible_trace): Implement a cost model based on joiner
23945 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23948 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
23949 char or short to __float128/_Float128 directly.
23951 2017-01-12 Martin Sebor <msebor@redhat.com>
23953 to -Wformat-overflow.
23954 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
23955 (min_bytes_remaining): Same.
23956 (get_string_length): Same.
23957 (format_string): Same.
23958 (format_directive): Same.
23960 (pass_sprintf_length::handle_gimple_call): Same.
23962 2017-01-12 Jakub Jelinek <jakub@redhat.com>
23964 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
23965 info.nowrite calls with no lhs that can't throw. Return bool
23966 whether gsi_remove has been called or not.
23967 (pass_sprintf_length::handle_gimple_call): Return bool whether
23968 try_substitute_return_value called gsi_remove. Formatting fix.
23969 (pass_sprintf_length::execute): Don't use gsi_remove if
23970 handle_gimple_call returned true.
23973 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
23974 be removed due to side-effects, don't remove following barrier nor
23975 turn the successor edge into fallthru edge.
23977 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23980 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
23981 element-reversing loads and stores as not swappable.
23983 2017-01-12 Nathan Sidwell <nathan@acm.org>
23984 Nicolai Stange <nicstange@gmail.com>
23986 * combine.c (try_combine): Don't ignore result of overlap checking
23987 loop. Combine overlap & asm check into single loop.
23989 2017-01-12 Richard Biener <rguenther@suse.de>
23991 * tree-pretty-print.c (dump_generic_node): Provide -gimple
23992 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
23994 2017-01-12 Richard Biener <rguenther@suse.de>
23996 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
23997 and TS_TARGET_OPTION directly derive from TS_BASE.
23998 * tree-core.h (tree_optimization_option): Derive from tree_base.
23999 (tree_target_option): Likewise.
24001 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24003 * config/i386/i386.c (memory_address_length): Increase len
24004 only when rip_relative_addr_p returns false.
24006 2017-01-11 Julia Koval <julia.koval@intel.com>
24008 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24009 (OPTION_MASK_ISA_SGX_SET): New.
24010 (ix86_handle_option): Handle OPT_msgx.
24011 * config.gcc: Added sgxintrin.h.
24012 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24013 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24014 * config/i386/i386.c (ix86_target_string): Add -msgx.
24016 (ix86_option_override_internal): Handle new options.
24017 (ix86_valid_target_attribute_inner_p): Add sgx.
24018 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24019 * config/i386/i386.opt: Add msgx.
24020 * config/i386/sgxintrin.h: New file.
24021 * config/i386/x86intrin.h: Add sgxintrin.h.
24023 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24026 * fold-const.c (maybe_nonzero_address): Return 1 for function
24028 (tree_single_nonzero_warnv_p): Don't handle function local objects
24032 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24035 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24038 * opts.c: Include "spellcheck.h"
24039 (struct string_fragment): New struct.
24040 (struct edit_distance_traits<const string_fragment &>): New
24042 (get_closest_sanitizer_option): New function.
24043 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24045 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24047 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24049 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24050 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24051 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24052 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24053 for initial die_offset if dwarf_split_debug_info.
24054 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24055 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24057 (output_skeleton_debug_sections): Formatting fix. Use
24058 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24059 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24061 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24063 * config/arm/cortex-a53.md: Add bypasses for
24064 cortex_a53_r2f_cvt.
24065 (cortex_a53_r2f): Only use for transfers.
24066 (cortex_a53_f2r): Likewise.
24067 (cortex_a53_r2f_cvt): Add reservation for conversions.
24068 (cortex_a53_f2r_cvt): Likewise.
24070 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24072 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24073 to all inlined functions, change static to extern.
24075 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24078 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24080 (arm_assemble_integer): Likewise.
24082 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24084 * config.gcc: Use new awk script to check CPU, FPU and architecture
24085 parameters for --with-... options.
24086 * config/arm/parsecpu.awk: New file
24087 * config/arm/arm-cpus.in: New file.
24088 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24090 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24092 * config/arm/t-arm: Update dependency rules.
24093 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24094 of processing .def files.
24095 * config/arm/genopt.sh: Deleted.
24096 * config/arm/gentune.sh: Deleted.
24097 * config/arm/arm-cores.def: Deleted.
24098 * config/arm/arm-arches.def: Deleted.
24099 * config/arm/arm-fpus.def: Deleted.
24100 * config/arm/arm-tune.md: Regenerated.
24101 * config/arm/arm-tables.opt: Regenerated.
24102 * config/arm/arm-cpu.h: New generated file.
24103 * config/arm/arm-cpu-data.h: New generated file.
24104 * config/arm/arm-cpu-cdata.h: New generated file.
24106 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
24109 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
24111 (input_varpool_node): Unpack dynamically_initialized bit.
24113 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
24115 PR rtl-optimization/79032
24116 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
24117 the alignment of the adjusted memory reference against that of MODE,
24118 instead of the alignment of the original memory reference.
24120 2017-01-11 Martin Jambor <mjambor@suse.cz>
24122 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
24124 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
24125 decorated functions.
24127 2017-01-11 Richard Biener <rguenther@suse.de>
24129 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
24130 set range/nonnull info for PHI results. Do not set it on
24131 stmts marked for removal.
24133 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
24135 * expr.c (store_field): In the bitfield case, fetch the return value
24136 from the registers before applying a single big-endian adjustment.
24137 Always do a final load for a BLKmode value not larger than a word.
24139 2017-01-10 David Malcolm <dmalcolm@redhat.com>
24142 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24143 that we correctly handle column numbers greater than
24144 LINE_MAP_MAX_COLUMN_NUMBER.
24146 2017-01-10 Martin Sebor <msebor@redhat.com>
24148 PR middle-end/78245
24149 * gimple-ssa-sprintf.c (get_destination_size): Call
24150 {init,fini}object_sizes.
24151 * tree-object-size.c (addr_object_size): Adjust.
24152 (pass_through_call): Adjust.
24153 (pass_object_sizes::execute): Adjust.
24154 * tree-object-size.h (fini_object_sizes): Declare.
24156 2017-01-10 Martin Sebor <msebor@redhat.com>
24158 PR tree-optimization/78775
24159 * builtins.c (get_size_range): Move...
24160 * calls.c: ...to here.
24161 (alloc_max_size): Accept zero argument.
24162 (operand_signed_p): Remove.
24163 (maybe_warn_alloc_args_overflow): Call get_size_range.
24164 * calls.h (get_size_range): Declare.
24166 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
24168 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
24169 from TI's devices.csv file as of September 2016.
24170 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
24172 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
24174 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
24175 * doc/invoke.texi: Likewise.
24176 * doc/md.texi: Likewise.
24177 * doc/objc.texi: Likewise.
24179 2017-01-10 Joshua Conner <joshconner@google.com>
24181 * config/arm/fuchsia-elf.h: New file.
24182 * config/fuchsia.h: New file.
24183 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
24184 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
24186 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
24188 2016-01-10 Richard Biener <rguenther@suse.de>
24190 PR tree-optimization/79034
24191 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24192 Propagate out degenerate PHIs in the joiner.
24194 2017-01-10 Martin Liska <mliska@suse.cz>
24196 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
24197 (sort_congruence_classes_by_decl_uid): Likewise.
24198 (sort_congruence_class_groups_by_decl_uid): Likewise.
24199 (sem_item_optimizer::merge_classes): Sort class, groups in these
24200 classes and members in the groups by DECL_UID of declarations.
24201 This would make merge operations stable.
24203 2017-01-10 Martin Liska <mliska@suse.cz>
24205 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
24206 usage of m_classes_vec.
24207 (sem_item_optimizer::~sem_item_optimizer): Likewise.
24208 (sem_item_optimizer::get_group_by_hash): Likewise.
24209 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
24210 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
24211 (sem_item_optimizer::verify_classes): Likewise.
24212 (sem_item_optimizer::process_cong_reduction): Likewise.
24213 (sem_item_optimizer::dump_cong_classes): Likewise.
24214 (sem_item_optimizer::merge_classes): Likewise.
24215 * ipa-icf.h (congruence_class_hash): Rename from
24216 congruence_class_group_hash. Remove declaration of m_classes_vec.
24218 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
24220 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
24221 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
24222 * config.gcc: Add avx512vpopcntdqintrin.h.
24223 * config/i386/avx512vpopcntdqintrin.h: New.
24224 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
24225 * config/i386/i386-builtin-types.def: Add new types.
24226 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
24227 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
24228 __builtin_ia32_vpopcountq_v8di_mask): New.
24229 * config/i386/i386-c.c (ix86_target_macros_internal): Define
24230 __AVX512VPOPCNTDQ__.
24231 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
24232 (PTA_AVX512VPOPCNTDQ): Define.
24233 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
24234 TARGET_AVX512VPOPCNTDQ_P): Define.
24235 * config/i386/i386.opt: Add mavx512vpopcntdq.
24236 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
24237 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
24239 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24241 PR middle-end/77484
24242 * predict.def (PRED_CALL): Set to 67.
24244 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
24246 * expr.c (store_field): In the bitfield case, if the value comes from
24247 a function call and is of an aggregate type returned in registers, do
24248 not modify the field mode; extract the value in all cases if the mode
24249 is BLKmode and the size is not larger than a word.
24251 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
24254 * config/i386/cpuid.h: Fix undefined behavior.
24256 2017-01-04 Jeff Law <law@redhat.com>
24258 PR tree-optimization/79007
24259 PR tree-optimization/67955
24260 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
24261 conservative for pt.null when flag_non_call_exceptions is on.
24263 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24265 PR translation/79019
24266 PR translation/79020
24267 * params.def (PARAM_INLINE_MIN_SPEEDUP,
24268 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
24269 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
24271 * config/avr/avr.opt (maccumulate-args): Likewise.
24272 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
24273 * common.opt (freport-bug): Likewise.
24274 * cif-code.def (CIF_FINAL_ERROR): Likewise.
24275 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
24276 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
24277 translatable string.
24278 * config/i386/i386.c (function_value_32): Likewise.
24279 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
24280 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
24282 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
24283 * common/config/msp430/msp430-common.c (msp430_handle_option):
24285 * symtab.c (symtab_node::verify_base): Likewise.
24286 * opts.c (set_debug_level): Likewise.
24287 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
24288 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
24289 missing whitespace to translatable strings.
24290 * config/avr/avr.md (bswapsi2): Fix typo in comment.
24291 * config/sh/superh.h: Likewise.
24292 * config/i386/xopintrin.h: Likewise.
24293 * config/i386/znver1.md: Likewise.
24294 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
24295 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
24296 * double-int.h (struct double_int): Likewise.
24297 * double-int.c (div_and_round_double): Likewise.
24298 * wide-int.cc: Likewise.
24299 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
24300 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
24301 * cfgcleanup.c (crossjumps_occured): Renamed to ...
24302 (crossjumps_occurred): ... this.
24303 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
24306 PR tree-optimization/78899
24307 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
24308 returning bool return struct loop *, NULL for failure and the new
24310 (versionable_outer_loop_p): Don't version outer loop if it has
24311 dont_vectorized bit set.
24312 (tree_if_conversion): When versioning outer loop, ensure
24313 tree_if_conversion is performed also on the inner loop of the
24314 non-vectorizable outer loop copy.
24315 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
24316 LOOP_VECTORIZED in inner loop of the scalar outer loop and
24317 prevent vectorization of it.
24318 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
24319 the outer loop vectorization of the non-scalar version is attempted
24320 before vectorization of the inner loop in scalar version. If
24321 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
24322 vectorization of its inner loop.
24323 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
24324 has 2 inner loops, rename also on edges from bb whose single pred
24325 is outer_loop->header. Fix typo in function comment.
24327 2017-01-09 Martin Sebor <msebor@redhat.com>
24330 * asan.c (asan_emit_stack_protection): Increase local buffer size
24331 to avoid snprintf truncation warning.
24333 2017-01-09 Andrew Pinski <apinski@cavium.com>
24335 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
24336 to reference thunderx2t99 for the tuning structure
24337 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
24339 (thunderx2t99_extra_costs): This.
24340 * config/aarch64/aarch64-tune.md: Regenerate.
24341 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
24342 (vulcan_addrcost_table): This.
24343 (vulcan_regmove_cost): Rename to ...
24344 (thunderx2t99_regmove_cost): This.
24345 (vulcan_vector_cost): Rename to ...
24346 (thunderx2t99_vector_cost): this.
24347 (vulcan_branch_cost): Rename to ...
24348 (thunderx2t99_branch_cost): This.
24349 (vulcan_tunings): Rename to ...
24350 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
24351 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
24353 2017-01-09 Martin Jambor <mjambor@suse.cz>
24357 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
24358 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
24359 (propagate_vr_accross_jump_function): Use the above function for all
24360 value range computations for pass-through jump functions and type
24361 converasion from explicit value range values.
24362 (ipcp_propagate_stage): Do not attempt to deduce types of formal
24363 parameters from TYPE_ARG_TYPES.
24364 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
24365 (ipa_write_node_info): Stream type of the actual argument.
24366 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
24368 2017-01-09 Martin Liska <mliska@suse.cz>
24371 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
24372 (lookup_compiler): Do not show error message with have_E.
24374 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24376 PR tree-optimization/78938
24377 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
24378 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
24379 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
24380 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
24383 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24385 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
24388 2017-01-09 Richard Biener <rguenther@suse.de>
24390 PR tree-optimization/78997
24391 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
24392 name condition properly.
24394 2017-01-09 Richard Biener <rguenther@suse.de>
24397 * dwarf2out.c (is_cxx): New overload with context.
24398 (is_naming_typedef_decl): Use it.
24400 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24402 * invoke.texi (Option Summary): Correct spacing in option lists
24403 and add line breaks to fix over-long lines.
24405 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24407 PR middle-end/17660
24409 * extend.texi (Common Variable Attributes): Add xref to GCC
24410 Internals manual to explain mode attribute keywords.
24412 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24415 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
24416 and Preprocessor Options.
24417 (Options for Linking): Document -pthread here....
24418 (RS/6000 and PowerPC Options): ...not here.
24419 (Solaris 2 Options): ...or here.
24420 * doc/cppopts.texi: Document -pthread.
24422 2017-01-08 Martin Sebor <msebor@redhat.com>
24424 PR middle-end/77708
24425 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
24426 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
24427 New member functions.
24428 (format_directive): Used them.
24430 (pass_sprintf_length::handle_gimple_call): Same.
24431 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
24432 to avoid truncation for any argument.
24433 (extract_affine_mul): Same.
24434 * tree.c (get_file_function_name): Same.
24436 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24438 PR middle-end/77484
24439 * predict.def (PRED_INDIR_CALL): Set to 86.
24441 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24443 PR preprocessor/54124
24444 * doc/cppopts.texi: Reformat -d subtable to list the full name
24445 of the options. Add cross-reference to the docs for the general
24446 compiler -d options.
24447 * doc/invoke.texi (Developer Options): Add cross-reference to the
24448 preprocessor-specific -d option documentation.
24450 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24452 PR preprocessor/13498
24453 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
24454 redudant material, and reflect new command-line options.
24455 (System Headers): Likewise.
24457 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24459 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
24460 -isystem, and -idirafter. Copy-edit.
24461 * doc/cppopts.texi: Copy-edit. Remove contradiction about
24462 default for -ftrack-macro-expansion. Delete obsolete and
24463 badly-formatted implementation details about -fdebug-cpp output.
24464 * doc/cppwarnopts.texi: Copy-edit.
24466 2017-01-07 David Malcolm <dmalcolm@redhat.com>
24469 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24470 that the transition from a max line width >= 1<<10 to narrower
24471 lines works correctly.
24473 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
24475 * doc/options.texi (PerFunction): New.
24476 * opt-functions.awk (switch_flags): Map both Optimization and
24477 PerFunction to CL_OPTIMIZATION.
24478 * opth-gen.awk: Test for PerFunction flag along with
24480 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
24481 it only when the latter is present. Skip those that don't in
24482 the hash function generator.
24483 * common.opt (fvar-tracking): Mark as PerFunction instead of
24485 (fvar-tracking-assignments): Likewise.
24486 (fvar-tracking-assignments-toggle): Likewise.
24487 (fvar-tracking-uninit): Likewise.
24489 2017-01-07 Jakub Jelinek <jakub@redhat.com>
24491 PR translation/79018
24492 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
24495 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
24498 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
24499 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
24500 TARGET_LONG_JUMP_TABLE_OFFSETS.
24501 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
24502 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
24503 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
24504 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
24505 * config/m68k/m68k.md (tablejump expander): Likewise.
24506 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
24507 TARGET_LONG_JUMP_TABLE_OFFSETS.
24508 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
24509 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
24511 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
24512 David Holsgrove <david.holsgrove@xilinx.com>
24514 * common/config/microblaze/microblaze-common.c
24515 (TARGET_EXCEPT_UNWIND_INFO): Remove.
24516 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
24518 * config/microblaze/microblaze.c (microblaze_must_save_register)
24519 (microblaze_expand_epilogue, microblaze_return_addr): Handle
24521 (microblaze_eh_return): New function.
24522 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
24523 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
24524 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
24525 * config/microblaze/microblaze.md (eh_return): New pattern.
24527 2017-01-06 Jakub Jelinek <jakub@redhat.com>
24529 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
24530 GCC_DIAGNOSTIC_STRINGIFY): Define.
24532 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
24534 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24536 * config/arm/arm.md (<mcrr>): New.
24538 * config/arm/arm.c (arm_arch5te): New.
24539 (arm_option_override): Set arm_arch5te.
24540 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
24542 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
24543 (arm_mcrr_qualifiers): ... this. New.
24544 (MRRC_QUALIFIERS): Define to...
24545 (arm_mrrc_qualifiers): ... this. New.
24546 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
24548 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
24549 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
24550 (MRRCI, mrrc, MRRC): New.
24551 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
24552 VUNSPEC_MRRC2): New.
24554 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24556 * config/arm/arm.md (<mcr>): New.
24558 * config/arm/arm.c (arm_coproc_builtin_available): Add
24559 support for mcr, mrc, mcr2 and mrc2.
24560 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
24561 (arm_mcr_qualifiers): ... this. New.
24562 (MRC_QUALIFIERS): Define to ...
24563 (arm_mrc_qualifiers): ... this. New.
24564 (MCR_QUALIFIERS): Define to ...
24565 (arm_mcr_qualifiers): ... this. New.
24566 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
24568 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
24569 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
24570 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
24571 VUNSPEC_MRC2): New.
24573 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24575 * config/arm/arm.md (*ldc): New.
24579 * config/arm/arm.c (arm_coproc_builtin_available): Add
24580 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
24581 (arm_coproc_ldc_stc_legitimate_address): New.
24582 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
24583 'qualifier_const_pointer'.
24584 (LDC_QUALIFIERS): Define to...
24585 (arm_ldc_qualifiers): ... this. New.
24586 (STC_QUALIFIERS): Define to...
24587 (arm_stc_qualifiers): ... this. New.
24588 * config/arm/arm-protos.h
24589 (arm_coproc_ldc_stc_legitimate_address): New.
24590 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
24591 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
24592 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
24593 stc2, stcl, stc2l): New.
24594 * config/arm/constraints.md (Uz): New.
24595 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
24596 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
24597 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
24598 VUNSPEC_STC2L): New.
24600 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24602 * config/arm/arm.md (<cdp>): New.
24603 * config/arm/arm.c (neon_const_bounds): Rename this ...
24604 (arm_const_bounds): ... this.
24605 (arm_coproc_builtin_available): New.
24606 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
24607 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
24608 (CDP_QUALIFIERS): Define to...
24609 (arm_cdp_qualifiers): ... this. New.
24611 (arm_expand_builtin_args): Add case for 6 arguments.
24612 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
24613 (arm_const_bounds): ... this.
24614 (arm_coproc_builtin_available): New.
24615 * config/arm/arm_acle.h (__arm_cdp): New.
24617 * config/arm/arm_acle_builtins.def (cdp): New.
24619 * config/arm/iterators.md (CDPI,CDP,cdp): New.
24620 * config/arm/neon.md: Rename all 'neon_const_bounds' to
24621 'arm_const_bounds'.
24622 * config/arm/types.md (coproc): New.
24623 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
24624 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
24625 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
24626 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
24628 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24630 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
24631 (UBINOP_QUALIFIERS): New.
24633 (acle_builtin_data): New. Change comment.
24634 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
24635 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
24636 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
24637 arm_acle_builtins.def.
24638 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
24639 (arm_init_acle_builtins): New.
24640 (CRC32_BUILTIN): Remove.
24641 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
24642 crc32cb, crc32ch and crc32cw.
24643 (arm_init_crc32_builtins): Remove.
24644 (arm_init_builtins): Use arm_init_acle_builtins rather
24645 than arm_init_crc32_builtins.
24646 (arm_expand_acle_builtin): New.
24647 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
24648 * config/arm/arm_acle_builtins.def: New.
24650 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24652 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
24653 (arm_builtin_datum): ... this.
24654 (arm_init_neon_builtin): Rename to ...
24655 (arm_init_builtin): ... this. Add a new parameters PREFIX
24656 and USE_SIG_IN_NAME.
24657 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
24658 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
24659 'arm_builtin_datum'.
24660 (arm_init_vfp_builtins): Likewise.
24661 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
24662 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
24663 (arm_expand_neon_args): Rename to ...
24664 (arm_expand_builtin_args): ... this. Rename builtin_arg
24665 enum values and differentiate between ARG_BUILTIN_MEMORY
24666 and ARG_BUILTIN_NEON_MEMORY.
24667 (arm_expand_neon_builtin_1): Rename to ...
24668 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
24669 values, arm_expand_builtin_args and add bool parameter NEON.
24670 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
24671 (arm_expand_vfp_builtin): Likewise.
24672 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
24674 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24676 PR middle-end/77484
24677 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
24678 * predict.c (tree_estimate_probability_bb): Reverse direction of
24679 polymorphic call predictor.
24681 2017-01-06 David Malcolm <dmalcolm@redhat.com>
24683 * passes.c (execute_one_pass): Split out pass-skipping logic into...
24684 (determine_pass_name_match): ...this new function and...
24685 (should_skip_pass_p): ...this new function.
24687 2017-01-06 Nathan Sidwell <nathan@acm.org>
24689 * ipa-visibility.c (function_and_variable_visibility): Reformat
24690 comments and long lines. Remove extrneous if.
24691 * symtab.c (symtab_node::make_decl_local): Fix code format.
24692 (symtab_node::set_section_for_node): Fix comment typo.
24694 2017-01-06 Martin Liska <mliska@suse.cz>
24697 * lra-constraints.c: Rename invariant to lra_invariant.
24698 * predict.c (set_even_probabilities): Initialize e to NULL.
24700 2017-01-05 Martin Sebor <msebor@redhat.com>
24702 PR tree-optimization/78910
24703 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
24704 (format_integer): Correct off-by-one error in the handling
24705 of precision with negative numbers in signed conversions..
24707 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
24709 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
24711 2017-01-05 Jakub Jelinek <jakub@redhat.com>
24713 PR tree-optimization/71016
24714 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
24715 factor_out_conditional_conversion. Formatting fix.
24716 (factor_out_conditional_conversion): Add cond_stmt argument.
24717 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
24718 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
24721 2017-01-05 David Malcolm <dmalcolm@redhat.com>
24723 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
24724 read-rtl-function.o, and selftest-rtl.o.
24725 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
24726 (selftest::aarch64_test_loading_full_dump): New function.
24727 (selftest::aarch64_run_selftests): New function.
24728 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
24729 selftest::aarch64_run_selftests.
24730 * config/i386/i386.c
24731 (selftest::ix86_test_loading_dump_fragment_1): New function.
24732 (selftest::ix86_test_loading_call_insn): New function.
24733 (selftest::ix86_test_loading_full_dump): New function.
24734 (selftest::ix86_test_loading_unspec): New function.
24735 (selftest::ix86_run_selftests): Call the new functions.
24736 * emit-rtl.c (maybe_set_max_label_num): New function.
24737 * emit-rtl.h (maybe_set_max_label_num): New decl.
24738 * function.c (instantiate_decls): Guard call to
24739 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
24740 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
24742 * gensupport.c (gen_reader::gen_reader): Pass "false"
24743 for new "compact" param of rtx_reader.
24744 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
24745 rather than an empty string for NULL strings.
24746 * read-md.c: Potentially include config.h rather than bconfig.h.
24747 Wrap include of errors.h with #ifdef GENERATOR_FILE.
24748 (have_error): New global, copied from errors.c.
24749 (md_reader::read_name): Rename to...
24750 (md_reader::read_name_1): ...this, adding "out_loc" param,
24751 and converting "missing name or number" to returning false, rather
24753 (md_reader::read_name): Reimplement in terms of read_name_1.
24754 (md_reader::read_name_or_nil): New function.
24755 (md_reader::read_string): Handle "(nil)" by returning NULL.
24756 (md_reader::md_reader): Add new param "compact".
24757 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
24758 (md_reader::read_file): New method.
24759 * read-md.h (md_reader::md_reader): Add new param "compact".
24760 (md_reader::read_file): New method.
24761 (md_reader::is_compact): New accessor.
24762 (md_reader::read_name): Convert return type from void to file_location.
24763 (md_reader::read_name_or_nil): New decl.
24764 (md_reader::read_name_1): New decl.
24765 (md_reader::m_compact): New field.
24766 (noop_reader::noop_reader): Pass "false" for new "compact" param
24768 (rtx_reader::rtx_reader): Add new "compact" param.
24769 (rtx_reader::read_rtx_operand): Make virtual and convert return
24770 type from void to rtx.
24771 (rtx_reader::read_until): New decl.
24772 (rtx_reader::handle_any_trailing_information): New virtual function.
24773 (rtx_reader::postprocess): New virtual function.
24774 (rtx_reader::finalize_string): New virtual function.
24775 (rtx_reader::m_in_call_function_usage): New field.
24776 (rtx_reader::m_reuse_rtx_by_id): New field.
24777 * read-rtl-function.c: New file.
24778 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
24779 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
24780 (selftest::verify_three_block_rtl_cfg): New decl.
24781 * read-rtl-function.h: New file.
24782 * read-rtl.c: Potentially include config.h rather than bconfig.h.
24783 For host, include function.h, memmodel.h, and emit-rtl.h.
24784 (one_time_initialization): New function.
24785 (struct compact_insn_name): New struct.
24786 (compact_insn_names): New array.
24787 (find_code): Handle insn codes in compact dumps.
24788 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
24789 (bind_subst_iter_and_attr): Likewise.
24790 (add_condition_to_string): Likewise.
24791 (add_condition_to_rtx): Likewise.
24792 (apply_attribute_uses): Likewise.
24793 (add_current_iterators): Likewise.
24794 (apply_iterators): Likewise.
24795 (initialize_iterators): Guard usage of apply_subst_iterator with
24796 #ifdef GENERATOR_FILE.
24797 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
24798 (md_reader::read_mapping): Likewise.
24799 (add_define_attr_for_define_subst): Likewise.
24800 (add_define_subst_attr): Likewise.
24801 (read_subst_mapping): Likewise.
24802 (check_code_iterator): Likewise.
24803 (rtx_reader::read_rtx): Likewise. Move one-time initialization
24805 (one_time_initialization): New function.
24806 (rtx_reader::read_until): New method.
24807 (read_flags): New function.
24808 (parse_reg_note_name): New function.
24809 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
24810 Handle reuse_rtx ids.
24811 Wrap iterator lookup within #ifdef GENERATOR_FILE.
24812 Add parsing support for RTL dumps, mirroring the special-cases in
24813 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
24814 values, and calling handle_any_trailing_information.
24815 (rtx_reader::read_rtx_operand): Convert return type from void
24816 to rtx, returning return_rtx. Handle case 'e'. Call
24817 finalize_string on XSTR and XTMPL fields.
24818 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
24819 "(nil)" values were omitted. Call the postprocess vfunc on the
24821 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
24822 class ctor. Initialize m_in_call_function_usage. Call
24823 one_time_initialization.
24824 * rtl-tests.c (selftest::test_uncond_jump): Call
24825 set_new_first_and_last_insn.
24826 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
24827 * selftest-rtl.c: New file.
24828 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
24829 (selftest::get_insn_by_uid): New decl.
24830 * selftest-run-tests.c (selftest::run_tests): Call
24831 read_rtl_function_c_tests.
24832 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
24833 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
24836 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
24838 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
24839 operands in a special way. Assert that pos+len <= mode precision.
24841 2017-01-05 Jakub Jelinek <jakub@redhat.com>
24843 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
24844 3 argument Alias with unlimited for the negative form.
24845 (fno-vect-cost-model): Removed.
24847 2017-01-05 Martin Liska <mliska@suse.cz>
24849 * hsa-gen.c (gen_hsa_divmod): New function.
24850 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
24852 2017-01-05 Martin Liska <mliska@suse.cz>
24855 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
24858 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24860 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
24861 small constant length operands.
24863 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24865 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
24866 between loop iterations.
24868 2017-01-05 Martin Liska <mliska@suse.cz>
24871 * gimplify.c (gimplify_decl_expr): Compare to
24872 asan_poisoned_variables instread of checking flags.
24873 (gimplify_target_expr): Likewise.
24874 (gimplify_expr): Likewise.
24875 (gimplify_function_tree): Conditionally initialize
24876 asan_poisoned_variables.
24878 2017-01-04 Jeff Law <law@redhat.com>
24880 PR tree-optimizatin/78812
24881 * rtl.h (contains_mem_rtx_p): Prototype.
24882 * ifcvt.c (containts_mem_rtx_p): Move from here to...
24883 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
24884 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
24885 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
24886 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
24888 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24890 * input.c (assert_char_at_range): Default-initialize actual_range.
24892 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24894 * df-scan.c (df_ref_create_structure): Make regno unsigned,
24895 to match the caller.
24897 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24899 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
24900 insns after final jump in test to emit dummy move.
24902 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24904 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
24905 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
24907 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24909 * multiple_target.c (create_dispatcher_calls): Init e_next.
24910 * tree-ssa-loop-split.c (split_loop): Init border.
24911 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
24914 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
24919 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
24920 (altivec_register_operand): Do not return true if the operand
24921 contains a SUBREG mixing SImode and SFmode.
24922 (vsx_register_operand): Likewise.
24923 (vsx_reg_sfsubreg_ok): New predicate.
24924 (vfloat_operand): Do not return true if the operand contains a
24925 SUBREG mixing SImode and SFmode.
24926 (vint_operand): Likewise.
24927 (vlogical_operand): Likewise.
24928 (gpc_reg_operand): Likewise.
24929 (int_reg_operand): Likewise.
24930 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
24931 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
24932 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
24934 (rs6000_emit_move_si_sf_subreg): New helper function.
24935 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
24936 fixup SUBREGs involving SImode and SFmode.
24937 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
24938 numbers for the new peephole2 optimization.
24939 (peephole2 for SFmode unions): New peephole2 to optimize cases in
24940 the GLIBC math library that do AND/IOR/XOR operations on single
24941 precision floating point.
24942 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
24943 target macros to say whether we need to avoid SUBREGs mixing
24945 (TARGET_ALLOW_SF_SUBREG): Likewise.
24946 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
24947 (UNSPEC_SI_FROM_SF): Likewise.
24948 (iorxor): Change spacing.
24949 (and_ior_xor): New iterator for AND, IOR, and XOR.
24950 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
24951 (movdi_from_sf_zero_ext): Likewise.
24952 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
24953 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
24954 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
24955 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
24956 (fms<mode>4): Likewise.
24957 (fnma<mode>4): Likewise.
24958 (fnms<mode>4): Likewise.
24959 (nfma<mode>4): Likewise.
24960 (nfms<mode>4): Likewise.
24962 2017-01-04 Marek Polacek <polacek@redhat.com>
24965 * doc/invoke.texi: Document -Wpointer-compare.
24967 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24969 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
24972 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
24973 descriptions for -gdwarf-5 and emit them as uleb128 instead of
24976 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
24979 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
24980 documentation of the powerpc_popcntb_ok attribute.
24981 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24982 code to issue warning messages if a requested CPU configuration is
24983 not supported by the binary (assembler and loader) toolchain.
24984 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
24985 made to define a built-in function that has been disabled.
24986 (paired_init_builtins): Add assertion to prevent ICE if attempt is
24987 made to define a built-in function that has been disabled.
24988 (altivec_init_builtins): Add comment explaining why definition
24989 of the DST built-in functions is not preceded by an assertion
24990 check. Add assertions to prevent ICE if attempts are made to
24991 define an altivec predicate or an abs* built-in function that has
24993 (htm_init_builtins): Add comment explaining why definition of the
24994 htm built-in functions is not preceded by an assertion check.
24996 2017-01-04 Jeff Law <law@redhat.com>
24998 PR tree-optimizatin/67955
24999 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25000 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25001 the points-to solution does not include pt_null. Use DECL_PT_UID
25004 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25006 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25007 Use gen_int_mode instead of gen_lopwart for const_int operands.
25009 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25011 PR tree-optimization/71563
25012 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25013 out of range value - has low bits known to be zero.
25015 2017-01-04 Alan Modra <amodra@gmail.com>
25017 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25018 * configure: Regenerate.
25019 * config.in: Regenerate.
25021 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25024 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25025 a substring of the message, but strcmp with the whole message. Ifdef
25026 ENABLE_NLS, translate the message first using dgettext.
25028 2017-01-03 Jeff Law <law@redhat.com>
25030 PR tree-optimizatin/78856
25031 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25032 (mark_threaded_blocks): Remove code to truncate thread paths that
25033 cross multiple loop headers. Instead invalidate the cached loop
25034 iteration information and handle case of a thread path walking
25035 into an irreducible region.
25037 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25040 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25041 assertions. Add support for doing the signbit if the IEEE 128-bit
25042 floating point value is in a GPR.
25043 * config/rs6000/rs6000.md (Fsignbit): Delete.
25044 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25045 Update the length attribute if the value is in a GPR.
25046 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25047 the sign or zero extension instruction, since the value is always 0/1.
25048 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25051 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25052 extracting SImode to a GPR register so that we can generate a
25053 store, limit the vector to be in a traditional Altivec register
25054 for the vextuwrx instruction.
25056 2017-01-03 Ian Lance Taylor <iant@google.com>
25058 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25060 2017-01-03 Martin Sebor <msebor@redhat.com>
25062 PR tree-optimization/78696
25063 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25064 precision. Use MPFR for %f for greater fidelity. Correct handling
25066 (pass_sprintf_length::compute_format_length): Set width and precision
25067 specified by asrerisk to void_node for vararg functions.
25068 (try_substitute_return_value): Adjust dump output.
25070 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25072 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25074 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25076 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25077 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25078 -mlra/-mno-lra was passed to the compiler.
25080 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25082 PR rtl-optimization/65618
25083 * emit-rtl.c (try_split): Move initialization of "before" and
25084 "after" to just before the call to emit_insn_after_setloc.
25086 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25088 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25090 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25092 * dwarf2out.c (gen_enumeration_type_die): When
25093 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25095 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25097 PR tree-optimization/78965
25098 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25099 Change first argument from const call_info & to call_info &. For %n
25100 set info.nowrite to false.
25102 PR middle-end/78901
25103 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
25104 possibly throwing calls.
25106 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
25107 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
25108 and fns handling, rather than in a separate case SSA_NAME.
25110 2017-01-02 Jeff Law <law@redhat.com>
25112 * config/darwin-driver.c (darwin_driver_init): Const-correctness
25113 fixes for first_period and second_period variables.
25115 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
25118 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
25119 (*insvqi_1): New insn pattern.
25120 (*insvqi_1_mem_rex64): Ditto.
25121 (*insvqi_2): Ditto.
25122 (*insvqi_3): Rename from *insvqi.
25124 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
25126 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
25128 * doc/cfg.texi (Edges): Remove reference to Java.
25129 (Maintaining the CFG): Ditto.
25131 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25133 PR middle-end/77674
25134 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
25135 transparent aliases.
25137 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25139 PR middle-end/77484
25140 * predict.def (PRED_CALL): Update hitrate.
25141 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
25142 * predict.c (tree_estimate_probability_bb): Split CALL predictor
25143 into direct/indirect/polymorphic variants.
25145 2017-01-01 Jakub Jelinek <jakub@redhat.com>
25147 Update copyright years.
25149 * gcc.c (process_command): Update copyright notice dates.
25150 * gcov-dump.c (print_version): Ditto.
25151 * gcov.c (print_version): Ditto.
25152 * gcov-tool.c (print_version): Ditto.
25153 * gengtype.c (create_file): Ditto.
25154 * doc/cpp.texi: Bump @copying's copyright year.
25155 * doc/cppinternals.texi: Ditto.
25156 * doc/gcc.texi: Ditto.
25157 * doc/gccint.texi: Ditto.
25158 * doc/gcov.texi: Ditto.
25159 * doc/install.texi: Ditto.
25160 * doc/invoke.texi: Ditto.
25162 Copyright (C) 2017 Free Software Foundation, Inc.
25164 Copying and distribution of this file, with or without modification,
25165 are permitted in any medium without royalty provided the copyright
25166 notice and this notice are preserved.