1 2007-09-11 Steve Kenton <skenton@ou.edu
3 * pa/linux-unwind.h: Guard with inhibit_libc.
4 * pa/hpux-unwind.h: Likewise.
6 2007-09-11 David Daney <ddaney@avtrex.com>
8 * doc/invoke.texi: Document new MIPS -mllsc and -mno-llsc options.
9 * doc/install.texi: Document new --with-llsc and --without-llsc
11 * config.gcc: Handle --with-llsc and --without-llsc configure options.
12 * config/mips/mips.md (sync, memory_barrier): Wrap sync instrunction
13 in %| and %- operand codes. Depend on GENERATE_SYNC instead of
15 (sync_compare_and_swap<mode>, sync_add<mode>, sync_sub<mode>,
16 sync_old_add<mode>, sync_old_sub<mode>, sync_new_add<mode>,
17 sync_new_sub<mode>, sync_<optab><mode>, sync_old_<optab><mode>,
18 sync_new_<optab><mode>, sync_nand<mode>, sync_old_nand<mode>,
19 sync_new_nand<mode>, sync_lock_test_and_set<mode>): Depend on
20 GENERATE_LL_SC instead of ISA_HAS_LL_SC.
21 * config/mips/mips.opt (mllsc): New option.
22 * config/mips/mips.c (mips_llsc): Define variable.
23 (mips_handle_option): Handle mllsc option.
24 (override_options): Set mips_print_operand_punct for '|' and '-'.
25 (print_operand): Add new %| and %- operand codes.
26 * config/mips/mips.h (mips_llsc_setting): New enum type.
28 (OPTION_DEFAULT_SPECS): Add llsc handling.
29 (GENERATE_SYNC): New macro.
30 (GENERATE_LL_SC): New macro.
31 (MIPS_COMPARE_AND_SWAP, MIPS_SYNC_OP, MIPS_SYNC_OLD_OP,
32 MIPS_SYNC_NEW_OP, MIPS_SYNC_NAND, MIPS_SYNC_OLD_NAND,
33 MIPS_SYNC_NEW_NAND, MIPS_SYNC_EXCHANGE): Wrap instructions
34 in %| and %- operand codes.
36 2007-09-11 Eric Botcazou <ebotcazou@adacore.com>
38 * tree-ssa-structalias.c (push_fields_onto_fieldstack): Deal with
39 TYPE_NONALIASED_COMPONENT like with DECL_NONADDRESSABLE_P.
41 2007-09-11 Jason Merrill <jason@redhat.com>
44 * stor-layout.c (layout_decl): Do pack variable size fields.
46 2007-09-11 Maxim Kuvyrkov <maxim@codesourcery.com>
48 * config/m68k/predicates.md (movsi_const0_operand,
49 non_symbolic_call_operand): New predicates.
51 * config/m68k/constraints.md: (Cs, Ci, C0, Cj, CQ, CW, CZ, CS, Ap, Ac):
53 * doc/md.texi (Constraints for Particular Machines: Motorola 680x0):
54 Document constraints N, O, P, R, S, T, Q, U, W, Cs, Ci, C0, Cj, CQ,
55 CW, CZ, CS, Ap and Ac.
57 * config/m68k/m68k.md (UNSPEC_IB): New constant.
58 (constraints.md): New include.
59 (cpu, type, type1, opx, opy, opx_type, opy_type, size, opx_access,
60 opx_mem, opy_mem, op_mem, guess, split): New attributes.
61 (movdf_internal): Name pattern. Fix to use alternatives. Add split.
64 (tstsi_internal): Name pattern. Fix to use alternatives. Specify
65 attributes. Split tstsi_internal_68020_cf from it.
66 (tstsi_internal_68020_cf): New pattern.
67 (tsthi_internal, tstqi_internal): Name pattern. Specify attributes.
68 (tst<mode>_cf): Specify attributea.
69 (cmpsi_cf): Name pattern. Specify attributes.
70 (cmp<mode>_68881, cmp<mode>_cf): Specify type attribute.
71 (pushexthisi_const): Fix to use alternatives. Specify
73 (movsi_const0): Split movsi_const0_68000_10 and movsi_const0_68040_60
74 from it. Fix to use alternatives. Specify attributes.
75 (movsi_const0_68040_10, movsi_const0_68040_60): New patterns.
76 (movsi_cf, movstrictqi_cf): Fix to use alternatives. Specify
78 (movsf_cf_soft): Specify attributes.
79 (movdf_cf_soft): Add split.
80 (pushasi, zero_extendhisi2_cf, zero_extendqisi2_cfv4,
81 cfv4_extendhisi2, 68k_extendhisi2, extendqihi2, cfv4_extendqisi2,
82 68k_extendqisi2, truncdfsf2_cf): Specify attributes.
83 (truncdfsf2_68881): Name pattern. Specify attributes.
84 (floatsi<mode>2_cf, floathi<mode>2_68881, floathi<mode>2_cf,
85 floatqi<mode>2_68881, floatqi<mode>2_cf, ftrunc<mode>2_cf,
86 fix<mode>qi2_cf, fix<mode>hi2_cf, fix<mode>si2_cf, adddi_dishl32):
88 (addsi3_5200): Fix to use alternatives. Specify attributes.
90 (add<mode>3_cf, subdi_dishl32): Specify attributes.
91 (subsi3): Add alternative for subq.l. Specify attributes.
92 (sub<mode>3_cf, mulhi3, mulhisi3): Specify attributes.
93 (mulhisisi3_s, mulsi3_68020, mulsi3_cf): Name pattern. Specify
95 (umulhisi3): Specify attributes.
96 (mulhisisi3_z): Name pattern. Specify attributes.
97 (fmul<mode>3_cf, div<mode>3_cf, negsi2_internal, negsi2_5200,
98 sqrt<mode>2_68881, clzsi2, one_cmplsi2_5200, subreghi1ashrdi_const32,
99 subregsi1ashrdi_const32, ashrsi3, subreg1lshrdi_const32, lshrsi3,
100 bsetmemqi): Specify attributes.
101 (bsetmemqi_ext): Name pattern. Specify attributes.
102 (bclrmemqi): Specify attributes.
103 (bclrmemqi_ext, scc, sls): Name pattern. Specify attributes.
104 (beq, bne, bgt, bgtu, blt, bltu, bge, bgeu, ble, bleu): Specify
106 (beq2, bne2, bgt2, bgtu2, blt2, bltu2, bge2, bgeu2, ble2, bleu2): Name
107 pattern. Specify attributes.
108 (jump): Specify attributes.
109 (tablejump_internal): Name pattern. Specify attributes.
110 (call_value): Split into non_symbolic_call_value,
111 symbolic_call_value_jsr, symbolic_call_value_bsr. Fix to use
112 alternatives. Specify attributes.
113 (non_symbolic_call_value, symbolic_call_value_jsr,
114 symbolic_call_value_bsr): New patterns.
115 (nop, return, unlink, indirect_jump): Specify attributes.
116 (trap): Fix condition. Specify attributes.
119 * config/m68k/m68k.c (m68k_symbolic_call_var): New variable.
120 (override_options): Initialize it. Initialize m68k_sched_cpu.
121 (CONST_METHOD): Rename to M68K_CONST_METHOD, move to m68k.h.
122 (const_method): Make global, rename to m68k_const_method.
123 (const_int_cost, output_move_const_into_data_reg): Update.
124 (output_move_double): Parametrize to emit rtl code, rename to
126 (output_reg_adjust, emit_reg_adjust, output_compadr, output_movsi,
127 emit_movsi): New static functions.
128 (output_move_double): New function with semantics of old
130 (m68k_emit_move_double): New function.
131 (m68k_sched_cpu): New variable.
132 (attr_op_type): New enum.
133 (sched_guess_p): New variable.
134 (sched_address_type, sched_operand_type, sched_attr_op_type):
135 New static functions.
136 (m68k_sched_attr_opx_type, m68k_sched_attr_opy_type,
137 m68k_sched_attr_size, m68k_sched_attr_op_mem): New functions.
138 (sched_branch_type): New static variable.
139 (m68k_sched_branch_type): New function.
140 * config/m68k/m68k.h (M68K_SYMBOLIC_CALL): New enum.
141 (m68k_symbolic_call_var): Declare.
142 (M68K_CONST_METHOD): Rename from CONST_METHOD. Move here from m68k.c.
143 (m68k_const_method, m68k_emit_move_double, m68k_sched_cpu,
144 m68k_sched_attr_opx_type, m68k_sched_attr_opy_type,
145 m68k_sched_attr_size, m68k_sched_attr_op_mem, m68k_sched_branch_type):
148 2007-09-11 Jakub Jelinek <jakub@redhat.com>
150 * builtins.def (BUILT_IN_VA_ARG_PACK_LEN): New builtin.
151 * builtins.c (expand_builtin) <case BUILT_IN_VA_ARG_PACK_LEN>: Issue
152 error if __builtin_va_arg_pack_len () wasn't optimized out during
154 * tree-inline.c (copy_bb): Replace __builtin_va_arg_pack_len ()
155 with the number of inline's anonymous arguments.
156 * doc/extend.texi: Document __builtin_va_arg_pack_len ().
158 2007-09-11 Zdenek Dvorak <ook@ucw.cz>
160 * fold-const.c (extract_muldiv_1): Do not simplify
163 2007-09-11 Jan Hubicka <jh@suse.cz>
165 * i386.h (ix86_tune_indices): Add X86_TUNE_INTER_UNIT_CONVERSIONS.
166 (TARGET_INTER_UNIT_CONVERSIONS): New.
167 * i386.md (floatsi expanders): Remove redundant check for SImode
168 source; offload to memory when asked for.
169 (floatsisf2_mixed, floatsisf2_sse, floatsidf2_mixed, floatsidf2_sse
170 floatdisf2_mixed, floatsisf2_sse, floatsidf2_mixed, floatsidf2_sse):
172 (floatsisf2_mixed_memory, floatsisf2_sse_memory,
173 floatsidf2_mixed_memory, floatsidf2_sse_memory
174 floatdisf2_mixed_memory, floatsisf2_sse_memory,
175 floatsidf2_mixed_memory, floatsidf2_sse_memory): New.
177 2007-09-11 Jan Hubicka <jh@suse.cz>
179 * toplev.c (process_options): all frontends now do unit-at-a-time.
180 * cgraphunit.c: update comments.
181 (cgraph_expand_function): call passmanager dirrectly; emit thunks.
182 * c-decl.c (finish_function): use cgraph_add_new_function.
183 * function.c (expand_function_end): We are always unit-at-a-time.
185 2007-09-11 Richard Sandiford <richard@codesourcery.com>
187 * config/mips/mips.c (mips_set_mips16_mode): Use separate anchor
189 (mips_use_anchors_for_symbol_p): Use default_use_anchors_for_symbol_p.
191 2007-09-11 Richard Sandiford <richard@codesourcery.com>
193 * config/mips/mips.c (mips_symbol_insns_1): Allow LEAs of
194 SYMBOL_FORCE_TO_MEM constants.
195 (mips_rtx_costs): Give a cost of 1 to force_to_mem_operands.
196 (mips16_rewrite_pool_refs_info): New structure.
197 (mips16_rewrite_pool_constant): New function, split out from...
198 (mips16_rewrite_pool_refs): ...here. Take a pointer to a
199 mips16_rewrite_pool_refs_info structure rather than a pointer
200 to a constant pool. Force force_to_mem_operands into memory.
201 (mips16_lay_out_constants): Update call to mips16_rewrite_pool_refs.
202 * config/mips/predicates.md (force_to_mem_operand): New predicate.
203 * config/mips/constraints.md (kf): New constraint.
204 * config/mips/mips.md (*movdi_64bit_mips16): Add a d <- kf alternative.
205 (*movsi_mips16): Likewise.
207 2007-09-11 Richard Sandiford <richard@codesourcery.com>
208 Nigel Stephens <nigel@mips.com>
209 David Ung <davidu@mips.com>
211 * config/mips/mips.h (CONSTANT_POOL_COST): Move to...
212 * config/mips/mips.c: ...here and set to 4 for TARGET_MIPS16.
213 (mips16_constant_cost, mips_immediate_operand_p, mips_binary_cost)
214 (mips_fp_mult_cost, mips_fp_div_cost, mips_sign_extend_cost)
215 (mips_zero_extend_cost): New functions.
216 (mips_rtx_costs): Treat COMPARE constants as having zero cost.
217 Use the new functions. Tweak many cost estimates, both here
218 and in the new subroutines. Return false when the cost of the
219 operands has not been calculated. Check for *clear_upper32.
220 Check for floating-point multiply-add, reciprocal and rsqrt
221 patterns. Handle comparison and rotation codes.
223 2007-09-11 Danny Smith <dannysmith@users.sourceforge.net>
225 * config/i386/cygming.h (TARGET_STRIP_NAME_ENCODING): Don't
227 * config/i386/i386.c (get_dllimport_decl): Don't strip
230 2007-09-10 Janis Johnson <janis187@us.ibm.com>
233 * config/dfp-bit.c: Don't skip TFmode conversions; move strto*
235 (DFP_TO_BFP): Use for either XFmode or TFmode.
236 (BFP_TO_DFP): Use for either XFmode or TFmode; always use cast
238 * config/dfp-bit.h: Include float.h.
239 (LONG_DOUBLE_HAS_XF_MODE, LONG_DOUBLE_HAS_TF_MODE): Define if long
240 double is one of these modes, rather than using LIBGCC_HAS_*F_MODE
241 which doesn't mean the same thing.
242 (BFP_KIND): Use 4 to mean TFmode.
243 (BFP_FMT): Specify the number of decimal digits based on the
244 number of mantissa digits.
245 (BFP_VIA_TYPE): Binary float type to use as cast for sprintf.
246 (BFP_TO_DFP, DFP_TO_BFP): Define names for TFmode variants.
247 (STR_TO_BFP): Use strtold for XFmode or TFmode.
248 (TFtype): Define if TFmode is supported.
249 * doc/libgcc.texi (Decimal float library routines): Document
250 TF conversion functions.
252 2007-09-10 Chao-ying Fu <fu@mips.com>
254 * config/mips/mips.c (mips_scalar_mode_supported_p): Declare.
255 (TARGET_SCALAR_MODE_SUPPORTED_P): Define.
256 (mips_emit_compare): Process fixed-point modes.
257 (mips_pad_arg_upward): Support fixed-point types.
258 (override_options): Allow fixed-point modes in accumulators.
259 (mips_pass_by_reference): Pass DQ, UDQ, DA, and UDA modes in registers.
260 (mips_vector_mode_supported_p): Support V2HQmode, V2UHQmode, V2HAmode,
261 V2UHAmode, V4QQmode, and V4UQQmode when TARGET_DSP.
262 (mips_scalar_mode_supported_p): New function to accept fixed-point
263 modes if the width is not greater than two BITS_PER_WORD.
264 * config/mips/mips.h (SHORT_FRACT_TYPE_SIZE, FRACT_TYPE_SIZE,
265 LONG_FRACT_TYPE_SIZE, LONG_LONG_FRACT_TYPE_SIZE,
266 SHORT_ACCUM_TYPE_SIZE, ACCUM_TYPE_SIZE, LONG_ACCUM_TYPE_SIZE,
267 LONG_LONG_ACCUM_TYPE_SIZE): Define.
268 * config/mips/mips.md ("d"): Update mode attribute for fixed-point
270 ("IMODE"): New mode attribute.
271 (mips-fixed.md): Include.
272 * config/mips/mips-modes.def: Create VECTOR_MODES for FRACT, UFRACT,
274 * config/mips/mips-fixed.md: New file.
276 2007-09-11 Ben Elliston <bje@au.ibm.com>
278 * config/spu/spu.md: Formatting fixes.
280 2007-09-10 Janis Johnson <janis187@us.ibm.com>
282 * config/dfp-bit.c (dfp_unary_func): Delete.
283 (dfp_unary_op): Delete.
284 (dfp_binary_op): Use decFloat functions instead of decNumber
285 functions for binary operations.
286 (d32_binary_op): Convert 32-bit operands to 64 bits for evaluation.
287 (dnn_binary_op): Call dfp_binary_op with decFloat rather than
289 (dfp_compare_op): Use decFloat functions insteadof decNumber
290 functions for comparisons.
291 (d32_compare_op): Convert 32-bit operands to 64 bits for evaluation.
292 (dnn_binary_op): Call dfp_compare_op with decFloat rather than
294 (DFP_ADD, DFP_SUB, DFP_MULTIPLE, DFP_DIVIDE): Use macros for
295 call to dxx_binary_op and decFloat function.
296 (DFP_EQ, DFP_NE, DFP_LT, DFP_GT, DFP_LE, DFP_GE): Use macros for
297 calls to dxx_binary_op and decFloat function.
298 * config/dfp-bit.h: Include decFloat header files.
299 (decFloat, DFP_BINARY_OP, DFP_COMPARE_OP, DEC_FLOAT_ADD,
300 DEC_FLOAT_SUBTRACT, DEC_FLOAT_MULTIPLY, DEC_FLOAT_DIVIDE,
301 DEC_FLOAT_COMPARE, DEC_FLOAT_IS_ZERO, DEC_FLOAT_IS_NAN,
302 DEC_FLOAT_IS_SIGNED: Define for each of 3 operand widths.
304 2007-09-10 Harsha Jagasia <harsha.jagasia@amd.com>
305 Jan Sjodin <jan.sjodin@amd.com>
307 * tree-vect-analyze.c (vect_analyze_operations): Change
308 comparison of loop iterations with threshold to less than
309 or equal to instead of less than. Reduce
310 min_scalar_loop_bound by one.
311 * tree-vect-transform.c (vect_estimate_min_profitable_iters):
312 Change prologue and epilogue iterations estimate to vf/2,
313 when unknown at compile-time. Change versioning guard
314 cost to taken_branch_cost. If peeling for alignment is
315 unknown at compile-time, change peel guard costs to one
316 taken branch and one not-taken branch per peeled loop.
317 If peeling for alignment is known but number of scalar loop
318 iterations is unknown at compile-time, change peel guard
319 costs to one taken branch per peeled loop. Change the cost
320 model equation to consider vector iterations as the loop
321 iterations less the prologue and epilogue iterations.
322 Change outside vector cost check to less than or equal to
323 zero instead of equal to zero.
324 (vect_do_peeling_for_loop_bound): Reduce
325 min_scalar_loop_bound by one.
326 * tree-vectorizer.h: Add TARG_COND_TAKEN_BRANCH_COST and
327 TARG_COND_NOT_TAKEN_BRANCH_COST.
328 * config/i386/i386.h (processor_costs): Add
329 scalar_stmt_cost, scalar_load_cost, scalar_store_cost,
330 vec_stmt_cost, vec_to_scalar_cost, scalar_to_vec_cost,
331 vec_align_load_cost, vect_unalign_load_cost,
332 vec_store_cost, cond_taken_branch_cost,
333 cond_not_taken_branch_cost.
334 Define macros for x86 costs.
335 * config/i386/i386.c:
336 (size_cost): Set scalar_stmt_cost, scalar_load_cost,
337 scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
338 scalar_to_vec_cost, vec_align_load_cost,
339 vect_unalign_load_cost, vec_store_cost,
340 cond_taken_branch_cost, cond_not_taken_branch_cost to one.
341 (i386_cost, i486_cost, pentium_cost, pentiumpro_cost,
342 geode_cost, k6_cost, athlon_cost, pentium4_cost, nocona_cost,
343 core2_cost, generic64_cost, generic32_cost): Set to default
345 (k8_cost, amdfam10_cost): Costs for vectorization tuned.
346 (x86_builtin_vectorization_cost): New.
348 2007-09-10 Janis Johnson <janis187@us.ibm.com>
349 Ben Elliston <bje@au.ibm.com>
351 * dfp.c: Include decimal128Local.h;
352 (dfp_byte_swap): Remove.
353 (encode_decimal32, decode_decimal32): Don't handle endianness.
354 (encode_decimal64, decode_decimal64): Ditto.
355 (encode_decimal128, decode_decimal128): Ditto.
356 * config/dfp-bit.c (host_to_ieee32, ieee_to_host_32): Ditto.
358 (host_to_ieee_64, ieee_to_host_64): Don't handle endianness.
360 (host_to_ieee_128, ieee_to_host_128): Don't handle endianness.
361 * Makefile.in (DECNUM_H): Add decimal128Local.h.
363 2007-09-10 David Daney <ddaney@avtrex.com>
365 * config/mips/mips.md (UNSPEC_MEMORY_BARRIER): New entry in
367 (memory_barrier): Rewrote as an insn that clobbers memory.
369 2007-09-10 Richard Sandiford <richard@codesourcery.com>
371 * config/mips/mips.c (mips_global_pointer): Check
372 call_really_used_regs instead of call_used_regs.
373 (mips_save_reg_p): Likewise. Save all call-saved registers
374 if current_function_saves_all_registers. Fix indentation.
375 No longer treat $18 as a special case.
376 (compute_frame_size): Guard FPR loop with TARGET_HARD_FLOAT.
378 2007-09-10 Richard Sandiford <richard@codesourcery.com>
380 * config/mips/mips.h (MIPS_ARCH_FLOAT_SPEC): New macro.
381 * config/mips/mips.c (mips_cpu_info_table): Mention it in the
382 the introductory comment.
383 (MIPS_MARCH_CONTROLS_SOFT_FLOAT): Delete.
384 (override_options): Don't test for it.
385 * config/mips/sde.h (MIPS_MARCH_CONTROLS_SOFT_FLOAT): Delete.
386 (DRIVER_SELF_SPECS): Add MIPS_ARCH_FLOAT_SPEC.
387 * config/mips/vr.h: As for config/mips/sde.h.
389 2007-09-10 Trevor Smigiel <trevor_smigiel@playstation.sony.com>
390 Revital Eres <eres@il.ibm.com>
392 * target.h (struct gcc_target.sched): New field: sms_res_mii.
393 (struct ddg): Define.
394 * target-def.h (TARGET_SCHED_SMS_RES_MII): Define.
395 (TARGET_SCHED): Add TARGET_SCHED_SMS_RES_MII.
396 * config/spu/spu.c: Include ddg.h.
397 (TARGET_SCHED_SMS_RES_MII): Define.
398 (spu_sms_res_mii): New function to calculate mii.
399 * modulo-sched (res_MII): Use it.
400 * doc/tm.texi: Document TARGET_SCHED_SMS_RES_MII.
402 2007-09-10 Andreas Krebbel <krebbel1@de.ibm.com>
404 * config/s390/s390.c (s390_dump_pool): Create copy of constant
405 pool entries since they might hold values that must not be shared.
407 2007-09-10 Uros Bizjak <ubizjak@gmail.com>
410 * gcc/config/i386/sse.md (ashr<mode>3): Change op2 mode to SImode.
411 Use 'N' operand constraint for op2.
412 (lshr<mode>3): Ditto.
413 (ashl<mode>3): Ditto.
414 (vec_shl_<mode>): Use const_0_to_255_mul_8_operand predicate for op2.
415 (vec_shr_<mode>): Use const_0_to_255_mul_8_operand predicate for op2.
417 * gcc/config/i386/i386.c (ix86_expand_builtin) [IX86_BUILTIN_PSLL?128,
418 IX86_BUILTIN_PSRA*?128, IX86_BUILTIN_PSRL?128]: Convert op1 to SImode.
420 2007-09-10 Andreas Krebbel <krebbel1@de.ibm.com>
422 * config/s390/s390.md ("fixuns_trunc<BFP:mode><GPR:mode>2"):
423 Change mode macro in the last real_2expN parameter to uppercase.
425 2007-09-10 Michael Matz <matz@suse.de>
427 * tree-pass.h (pass_cselim): Declare new pass.
428 * passes.c (init_optimization_passes): Link in pass_cselim.
429 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Renamed from
430 tree_ssa_phiopt; add do_store_elim parameter, handle it by calling
431 cond_store_replacement.
432 (condstoretemp): New static variable.
433 (cond_store_replacement): New function.
434 (tree_ssa_phiopt, tree_ssa_cs_elim): New wrappers around
435 tree_ssa_phiopt_worker.
436 (struct name_to_bb): New.
437 (get_non_trapping, name_to_bb_hash, name_to_bb_eq, add_or_mark_expr,
438 nt_init_block, nt_fini_block): New static functions.
439 (seen_ssa_names, nontrap_set): New static variables.
440 (gate_cselim, pass_cselim): Define new pass.
441 * common.opt (ftree-cselim): New flag.
442 * toplev.c (process_options): Set flag_tree_cselim if required.
444 2007-09-10 Hans-Peter Nilsson <hp@axis.com>
446 * simplify-rtx.c (simplify_relational_operation_1): For recent
447 canonicalization, don't recurse if op1 equals both PLUS arguments.
449 2007-09-09 David Daney <ddaney@avtrex.com>
451 * optabs.c (expand_sync_operation): Use plus insn if minus
453 (expand_sync_fetch_operation): Ditto.
455 2007-09-09 H.J. Lu <hongjiu.lu@intel.com>
457 * i386.md (*floatsisf2_mixed_vector): Use cvtdq2ps instead
459 (*floatsisf2_sse_vector): Likewise.
461 2007-09-09 Krister Walfridsson <cato@df.lth.se>
463 * config/netbsd.h (HANDLE_PRAGMA_PACK_PUSH_POP): Define to 1.
465 2007-09-09 H.J. Lu <hongjiu.lu@intel.com>
467 * config/i386/i386.h (ix86_tune_indices): Rename
468 X86_USE_VECTOR_CONVERTS to X86_TUNE_USE_VECTOR_CONVERTS.
469 (TARGET_USE_VECTOR_CONVERTS): Updated.
470 * config/i386/i386.c: Likewise.
472 2007-09-09 Sandra Loosemore <sandra@codesourcery.com>
473 Nigel Stephens <nigel@mips.com>
475 * doc/invoke.texi (Overall Options): Add .sx file extension
477 * cppspec.c (known_suffixes): Likewise.
478 * gcc.c (default_compilers): Likewise.
480 2007-09-09 Rask Ingemann Lambertsen <rask@sygehus.dk>
483 * config/i386/i386.h (CANONICALIZE_COMPARISON): Delete.
484 * simplify-rtx.c (simplify_relational_operation_1): Add the
485 canonicalization from i386.h.
486 * doc/md.texi (Canonicalization of Instructions): Document it.
488 2007-09-09 Jan Hubicka <jh@suse.cz>
489 Dwarakanath Rajagopal <dwarak.rajagopal@amd.com>
491 * i386.h (ix86_tune_indices): Add X86_USE_VECTOR_CONVERTS.
492 (TARGET_USE_VECTOR_CONVERTS): New.
493 * i386.md: New post-reload splitters for converting SF to DF and DF to
495 (floatsi* expander): Special case vector conversions.
496 (floatsisf2_mixed_vector, floatsisf2_sse_vector_nointernunit,
497 floatsisf2_sse_vector_internunit, floatsisf2_sse_vector,
498 floatsidf2_mixed_vector, floatsidf2_sse_vector): New.
499 (floatsisf2_mixed, floatsisf2_sse, floatsidf2_mixed, floatsidf2_sse):
500 Disable when doing vector converts.
501 (floatsi<mode>2_i387): Disable when
502 * sse.md (vec_dupv2df): Export.
503 * i386.c (ix86_tune_features): Enable SSE conversions.
505 2007-09-09 Richard Guenther <rguenther@suse.de>
507 * tree-ssa-operands.c (add_virtual_operand): Only mark
508 stores as has_volatile_ops if alias information is not available.
510 2007-09-09 Revital Eres <eres@il.ibm.com>
512 * doc/contrib.texi: Add myself.
514 2007-09-09 Ira Rosen <irar@il.ibm.com>
516 * tree-vectorizer.h (stmt_vinfo_set_inside_of_loop_cost,
517 stmt_vinfo_set_outside_of_loop_cost): New functions.
518 * tree-vect-transform.c (vect_get_cost_fields): Remove.
519 (vect_model_simple_cost): Call
520 stmt_vinfo_set_inside/outside_of_loop_cost to set the relevant cost
521 field instead of calling vect_get_cost_fields.
522 (vect_model_store_cost, vect_model_load_cost): Likewise.
524 2007-09-09 Revital Eres <eres@il.ibm.com>
526 * config/rs6000/rs6000.c (paired_init_builtins): Add const
527 declaration to bdesc_paired_preds variable.
528 (paired_expand_builtin): Likewise.
530 2007-09-09 Revital Eres <eres@il.ibm.com>
532 * dbgcnt.def (sms_sched_loop): New counter.
533 * modulo-sched.c: Use sms_sched_loop instead of
534 MAX_SMS_LOOP_NUMBER to determine the maximum number of loops to
535 perform swing modulo scheduling on. Include dbgcnt.h.
536 * Makefile.in: Add DBGCNT_H to modulo-sched.o.
537 * params.def: Remove PARAM_MAX_SMS_LOOP_NUMBER.
539 2007-09-09 Uros Bizjak <ubizjak@gmail.com>
541 * config/i386/i386.md (X87MODEF12, SSEMODEF): Remove mode iterators.
542 Substitute all uses with ...
543 (MODEF): New mode iterator.
545 (fix_trunc<mode>_fisttp_i387_1): Remove operand constraints
546 from pre-regalloc define_insn_and_split splitter pattern.
547 (*fix_trunc<mode>_i387_1): Ditto.
549 (*fist<mode>2_1): Ditto.
550 (frndintxf2_floor): Ditto.
551 (*fist<mode>2_floor_1): Ditto.
552 (frndintxf2_ceil): Ditto.
553 (*fist<mode>2_ceil_1): Ditto.
554 (frndintxf2_trunc): Ditto.
555 (frndintxf2_mask_pm): Ditto.
557 (prologue): Use (const_int 0) as never generated filler insn.
559 (sibcall_epilogue): Ditto.
560 (eh_return_si): Ditto.
561 (eh_return_di): Ditto.
563 (add<mode>3): Rename from adddf3 and addsf3. Macroize expander
564 using MODEF mode iterator.
565 (sub<mode>3): Rename from subdf3 and subsf3. Macroize expander
566 using MODEF mode iterator.
567 (mul<mode>3): Rename from muldf3 and mulsf3. Macroize expander
568 using MODEF mode iterator.
569 (nearbyint<mode>2): Rename from nearbyintdf2 and nearbyintsf2.
570 Macroize expander using MODEF mode iterator.
572 (zero_extendsidi2): Remove operand constraints from expander.
573 (smuldi3_highpart): Ditto.
574 (indirect_jump): Ditto.
577 * config/i386/sse.md (storentv4sf): Ditto.
578 (storentv2df): Ditto.
579 (storentv2di): Ditto.
581 (sse2_cvtpd2ps): Ditto.
582 (vec_interleave_highv16qi): Ditto.
583 (vec_interleave_lowv16qi): Ditto.
584 (vec_interleave_highv8hi): Ditto.
585 (vec_interleave_lowv8hi): Ditto.
586 (vec_interleave_highv4si): Ditto.
587 (vec_interleave_lowv4si): Ditto.
588 (vec_interleave_highv2di): Ditto.
589 (vec_interleave_lowv2di): Ditto.
590 (sse2_maskmovdqu): Ditto.
591 * config/i386/mmx.md (mmx_maskmovq): Ditto.
593 2007-09-09 Ira Rosen <irar@il.ibm.com>
595 * tree-vectorizer.h (enum vect_def_type): Start enumeration from 1.
596 (struct _slp_tree, struct _slp_instance): Define new data structures
597 along macros for their access.
598 (struct _loop_vec_info): Define new fields: strided_stores,
599 slp_instances, and slp_unrolling_factor along macros for their access.
600 (enum slp_vect_type): New.
601 (struct _stmt_vec_info): Define new field, slp_type, and macros for its
603 (STMT_VINFO_STRIDED_ACCESS): New macro.
604 (vect_free_slp_tree): Declare.
605 (vectorizable_load): Add an argument of type slp_tree.
606 (vectorizable_store, vectorizable_operation, vectorizable_conversion,
607 vectorizable_assignment): Likewise.
608 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
609 Declare (make extern).
610 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
611 (new_loop_vec_info): Likewise.
612 (destroy_loop_vec_info): Free memory allocated for SLP structures.
613 * tree-vect-analyze.c: Include recog.h.
614 (vect_update_slp_costs_according_to_vf): New.
615 (vect_analyze_operations): Add argument for calls to vectorizable_ ()
616 functions. For not pure SLP stmts with strided access check that the
617 group size is power of 2. Update the vectorization factor according to
618 SLP. Call vect_update_slp_costs_according_to_vf.
619 (vect_analyze_group_access): New.
620 (vect_analyze_data_ref_access): Call vect_analyze_group_access.
621 (vect_free_slp_tree): New functions.
622 (vect_get_and_check_slp_defs, vect_build_slp_tree, vect_print_slp_tree,
623 vect_mark_slp_stmts, vect_analyze_slp_instance, vect_analyze_slp,
624 vect_make_slp_decision, vect_detect_hybrid_slp_stmts,
625 vect_detect_hybrid_slp): Likewise.
626 (vect_analyze_loop): Call vect_analyze_slp, vect_make_slp_decision
627 and vect_detect_hybrid_slp.
628 * tree-vect-transform.c (vect_estimate_min_profitable_iters): Take
629 SLP costs into account.
630 (vect_get_cost_fields): New function.
631 (vect_model_simple_cost): Make extern, add SLP parameter and handle
633 (vect_model_store_cost, vect_model_load_cost): Likewise.
634 (vect_get_constant_vectors): New function.
635 (vect_get_slp_vect_defs, vect_get_slp_defs,
636 vect_get_vec_defs_for_stmt_copy, vect_get_vec_defs_for_stmt_copy,
637 vect_get_vec_defs): Likewise.
638 (vectorizable_reduction): Don't handle SLP for now.
639 (vectorizable_call): Don't handle SLP for now. Add argument to
640 vect_model_simple_cost.
641 (vectorizable_conversion): Handle SLP (call vect_get_vec_defs to
642 get SLPed and vectorized defs). Fix indentation and spacing.
643 (vectorizable_assignment): Handle SLP.
644 (vectorizable_induction): Don't handle SLP for now.
645 (vectorizable_operation): Likewise.
646 (vectorizable_type_demotion): Add argument to
647 vect_model_simple_cost.
648 (vectorizable_type_promotion): Likewise.
649 (vectorizable_store, vectorizable_load): Handle SLP.
650 (vectorizable_condition): Don't handle SLP for now.
651 (vect_transform_stmt): Add a new argument for SLP. Check that there is
652 no SLP transformation required for unsupported cases. Add SLP
653 argument for supported cases.
654 (vect_remove_stores): New function.
655 (vect_schedule_slp_instance, vect_schedule_slp): Likewise.
656 (vect_transform_loop): Schedule SLP instances.
657 * Makefile.in: (tree-vect-analyze.o): Depend on recog.h.
659 2007-09-09 Andrew Haley <aph@redhat.com>
661 * optabs.c (sign_expand_binop): Set libcall_gen = NULL in the
664 2007-09-09 Hans-Peter Nilsson <hp@axis.com>
666 Divide REG_LABEL notes into REG_LABEL_OPERAND and REG_LABEL_TARGET.
667 * doc/rtl.texi (Insns): Specify when a label_ref makes a jump_insn.
668 Similar for what label_refs can go in the JUMP_TARGET field. Split
669 REG_LABEL documentation into REG_LABEL_TARGET and REG_LABEL_OPERAND.
670 * reload.c (find_reloads): Generate REG_LABEL_OPERAND, not
671 REG_LABEL when replacing an operand with a LABEL_REF for a
673 (subst_reloads): When replacing a LABEL_REG with a register,
674 instead of generating a REG_LABEL note, assert that there already
675 is one or that the label is a known target for the insn.
676 * rtlanal.c (computed_jump_p): Instead of looking for a REG_LABEL
677 note, check the JUMP_LABEL field. Remove "else" after return.
678 * reorg.c (emit_delay_sequence): Replace case for REG_LABEL with
679 cases for REG_LABEL_OPERAND and REG_LABEL_TARGET.
680 (fill_slots_from_thread): Handle both REG_LABEL_OPERAND and
681 REG_LABEL_TARGET notes, including the JUMP_TARGET field on JUMP_P
682 insns. Iterate over all notes; don't assume there's only one.
683 * cse.c (recorded_label_ref): Adjust comment to refer to
685 (cse_extended_basic_block): Do LABEL_REF check for all INSN_P
686 insns, not just NONJUMP_INSN_P.
687 (check_for_label_ref): For JUMP_P insns, check that the LABEL_REF
689 * jump.c (rebuild_jump_labels): Adjust head comment.
690 (init_label_info): Ditto. Remove REG_LABEL_OPERAND notes only;
691 don't reset REG_LABEL_TARGET notes, including the JUMP_LABEL field.
692 (mark_all_labels): For JUMP_P insns without a target, check if the
693 the target is noted on the previous nonjump insn.
694 (mark_jump_label_1): New function, guts from mark_jump_label.
695 <case IF_THEN_ELSE>: Handle first operand as a non-target when
696 marking jump target labels.
697 <case LABEL_REF>: Adjust for whether to generate a
698 REG_LABEL_TARGET or a REG_LABEL_OPERAND note.
699 For 'E' format rtl, iterate in descending element order.
700 (delete_related_insns): Handle both REG_LABEL_TARGET and
701 REG_LABEL_OPERAND notes. For JUMP_P insns with labels with zero
702 reference count, delete and fallthrough. Move finding-next-
703 non-deleted insn last in the function. Look at all INSN_P insns
704 for REG_LABEL_OPERAND notes.
705 (redirect_jump_2): Assert that OLABEL equals the old JUMP_LABEL of
707 * print-rtl.c (print_rtx): For JUMP_P insns and a non-NULL
708 JUMP_LABEL, output the INSN_UID of it.
709 * gcse.c: Adjust comments as appropriate to say REG_LABEL_OPERAND
710 and/or REG_LABEL_TARGET.
711 (add_label_notes): Only add REG_LABEL_OPERAND notes. Put in line
712 with jump.c copy by only adding notes for labels actually
713 referenced in the insn.
714 * emit-rtl.c (try_split): Don't assume only NONJUMP_INSN_P need
715 usage count increment; handle all INSN_P trial insns.
716 (emit_copy_of_insn_after): Change to not copy REG_LABEL_OPERAND
718 * rtl.h (struct rtx_def) <volatil>: Adjust to mention
719 REG_LABEL_TARGET and REG_LABEL_OPERAND.
720 (LABEL_REF_NONLOCAL_P): Allow REG_LABEL_TARGET and
722 * combine.c (distribute_notes): Adjust for REG_LABEL_TARGET on
723 JUMP_P insns and REG_LABEL_OPERAND everywhere.
724 * sched-rgn.c (is_cfg_nonregular): Check for REG_LABEL_OPERANDS
726 * reg-notes.def (LABEL_TARGET, LABEL_OPERAND): Split from LABEL.
727 * cfgrtl.c (delete_insn): Adjust to handle REG_LABEL_TARGET and
728 REG_LABEL_OPERAND notes.
729 * reload1.c (calculate_needs_all_insns): Adjust comments.
730 (set_label_offsets): Adjust to look for REG_LABEL_OPERAND notes.
731 * config/alpha/alpha.md (split for load of an address into a
732 four-insn sequence on Unicos/Mk): Adjust to use
734 * config/sh/sh.md (sh_reorg, final_prescan_insn): Ditto.
736 2007-09-09 Laurynas Biveinis <laurynas.biveinis@gmail.com>
739 2007-09-05 Laurynas Biveinis <laurynas.biveinis@gmail.com>
740 * regrename.c (copyprop_hardreg_forward_1): New variable next. Use
741 FOR_BB_INSNS_SAFE instead of for loop.
742 * cse.c (cse_extended_basic_block): Likewise.
743 * postreload.c (reload_cse_regs_1): New variable next. Make sure
744 that the for loop does not invoke NEXT_INSN on a deleted insn.
745 * function.c (instantiate_virtual_regs): Likewise.
746 * lower-subreg.c (remove_retval_note): Likewise.
747 (decompose_multiword_subregs): Use FOR_BB_INSNS_SAFE instead of
749 * emit-rtl.c (remove_insn): Set NEXT_INSN and PREV_INSN to NULL on
751 * cfgrtl.c (delete_insn): Set JUMP_LABEL to NULL on a deleted
752 insn, if it's a jump.
753 (try_redirect_by_replacing_jump): New variable jump_p. Call
754 tablejump_p before delete_insn_chain.
755 * reload1.c (reload): New variable next. Make sure that the for
756 loop does not invoke NEXT_INSN on a deleted insn.
757 (fixup_eh_region_note): Make the loop terminate if i becomes NULL.
758 (delete_output_reload): New variable prev. Make sure the the for
759 loops do not invoke PREV_INSN on a deleted insn.
761 2007-09-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
763 * pa/constraints.md: Remove 'X' from unused letters comment.
765 2007-09-08 Richard Guenther <rguenther@suse.de>
767 * tree-tailcall.c (find_tail_calls): If we don't have aliases
768 computed check stmt_ann->references_memory instead of counting
771 2007-09-08 Segher Boessenkool <segher@kernel.crashing.org>
773 * cse.c (fold_rtx): Use validate_unshare_change() instead of
774 validate_change() in one more case.
776 2007-09-08 Zdenek Dvorak <ook@ucw.cz>
778 PR tree-optimization/32283
779 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Use
780 estimated_loop_iterations.
781 (determine_use_iv_cost_condition): Decrease cost of expressions
782 used in iv elimination.
784 2007-09-08 Richard Guenther <rguenther@suse.de>
786 * tree-cfg.c (verify_gimple_expr): Avoid building new
787 pointer types, use TYPE_POINTER_TO if available instead.
789 2007-09-08 Uros Bizjak <ubizjak@gmail.com>
792 PR rtl-optimization/26449
793 * config/i386/sse.md (mulv4si3): Do not expand sse2 sequence.
794 (*sse2_mulv4si3): New define_insn_and_split pattern. Split insn in
796 (mulv16qi3): Implement as define_insn_and_split pattern instead of
797 define_expand. Split insn in split1 pass.
800 2007-09-08 Dorit Nuzman <dorit@il.ibm.com>
802 PR tree-optimization/33301
803 * tree-vect-analyze (analyze_operations): Look at the type of the rhs
806 2007-09-07 Zdenek Dvorak <ook@ucw.cz>
808 PR tree-optimization/32183
809 * Makefile.in (tree-ssa-reassoc.o): Also depend on $(CFGLOOP_H).
811 * tree-ssa-reassoc.c: Include cfgloop.h.
812 (is_reassociable_op): Add a loop argument and return true only
814 (linearize_expr): Updated.
815 (should_break_up_subtract): Likewise.
816 (linearize_expr_tree): Likewise.
817 (init_reassoc): Call loop_optimizer_init with
818 AVOID_CFG_MODIFICATIONS. Remove calculate_dominance_info call
820 (fini_reassoc): Call loop_optimizer_finalize.
822 2007-09-07 Sterling Augustine <sterling@tensilica.com>
824 * config/xtensa/lib2funcs.S (__xtensa_sync_caches): Use an ISYNC even
825 if there is no i-cache.
827 2007-09-07 Richard Guenther <rguenther@suse.de>
830 2007-09-06 Richard Guenther <rguenther@suse.de>
832 PR tree-optimization/32586
833 * tree-ssa-sccvn.c (simplify_binary_expression): Avoid
834 folding if nothing changed.
835 (simplify_unary_expression): New function. Do tree combining
836 on conversion like codes.
837 (try_to_simplify): Call it.
839 2007-09-07 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
842 * gthr-posix.h (__gthread_active_p): Add implementation for hppa-hpux.
843 (__gthread_active,__gthread_start, __gthread_active_init): New.
844 * gthr-posix95.h: Likewise.
846 2007-09-07 Roman Zippel <zippel@linux-m68k.org>
848 * function.h (struct function): Rename calls_unwind_init
849 to saves_all_registers.
850 (current_function_saves_all_registers): Renamed from
851 current_function_calls_unwind_init.
852 * reload1.c (reload): Set current_function_saves_all_registers.
853 * except.c (expand_builtin_unwind_init): Likewise.
854 * config/m68k/m68k.c (m68k_save_reg): Use
855 current_function_saves_all_registers to save pic register.
857 2007-09-07 Janis Johnson <janis187@us.ibm.com>
859 config/m68k/m68k.c (floating_exact_log2): Update call to real_2expN.
860 config/s390/s390.md (fixuns_trunc<BFP:mode><GPR:mode>2): Ditto.
862 Reapply reverted changes:
864 2007-09-06 Jan Hubicka <jh@suse.cz>
866 * config/i386.c (ix86_expand_lround, ix86_expand_round): Update call of
869 2007-09-06 Richard Sandiford <richard@codesourcery.com>
871 * config/mips/mips.md (fixuns_truncdfsi2, fixuns_truncdfdi2)
872 (fixuns_truncsfsi2, fixuns_truncsfdi2): Update calls to real_2expN.
874 2007-09-05 Janis Johnson <janis187@us.ibm.com>
876 * optabs.c (expand_float): Convert unsigned integer as signed only
877 if it provides sufficient accuracy; add mode argument to real_2expN.
878 (expand_fix): Fix comment typos; extend binary float into mode
879 wider than destination for converion to unsigned integer; add mode
880 argument to real_2expN.
881 * real.c (real_2expN): Add mode argument to special-case decimal
883 * real.h (real_2expN): Ditto.
884 * fixed-value.c (check_real_for_fixed_mode): Add mode argument to
886 (fixed_from_string): Ditto.
887 (fixed_to_decimal): Ditto.
888 (fixed_convert_from_real): Ditto.
889 (real_convert_from_fixed): Ditto.
890 * config/rs6000/rs6000.md (FP): Include DD and TD modes.
891 * config/rs6000/dfp.md (extendddtd2, adddd3, addtd3, subdd3, subtd3,
892 muldd3, multd3, divdd3, divtd3, cmpdd_internal1, cmptd_internal1,
893 floatditd2, ftruncdd2, fixdddi2, ftrunctd2, fixddi2): New.
895 2007-09-07 Diego Novillo <dnovillo@google.com>
897 * tree-flow.h (const_block_stmt_iterator): Remove.
898 Update all users to use block_stmt_iterator.
899 * tree-iterator.h (const_tree_stmt_iterator): Remove.
900 Update all users to use tree_stmt_iterator.
902 2007-09-07 Sandra Loosemore <sandra@codesourcery.com>
904 * config/mips/mips.c (mips_set_current_function): Temporarily
905 make this a no-op to fix bootstrap errors, pending rewrite.
907 2007-09-07 Jan Hubicka <jh@suse.cz>
909 * reorg.c (dbr_schedule): Move code removing placeholder USEs later
912 2007-09-07 Dorit Nuzman <dorit@il.ibm.com>
914 PR tree-optimization/33299
915 * tree-vect-transform.c (vect_create_epilog_for_reduction): Update
916 uses for all relevant loop-exit phis, not just the first.
918 2007-09-07 Richard Guenther <rguenther@suse.de>
921 * tree-ssa-operands.c (access_can_touch_variable): An access
922 of the form (*p)[0] can touch a variable of same size.
924 2007-09-07 Jan Hubicka <jh@suse.cz>
926 * passes.c (init_optimization_passes): Add simple dce and addressable
928 * tree-ssa.c (execute_update_addresses_taken): New function.
929 (pass_update_address_taken): New.
930 * tree-ssa-dse.c (execute_simple_dse): New function.
931 (pass_simple_dse): New.
932 * tree-pass.h (pass_simple_dse, pass_update_address_taken): Declare.
934 2007-09-07 Tobias Burnus <burnus@net-b.de>
937 * doc/invoke.texi: Fix -Wstrict-overflow= table.
939 2007-09-07 Richard Guenther <rguenther@suse.de>
942 2007-09-06 Richard Guenther <rguenther@suse.de>
944 PR tree-optimization/32586
945 * tree-ssa-sccvn.c (simplify_binary_expression): Avoid
946 folding if nothing changed.
947 (simplify_unary_expression): New function. Do tree combining
948 on conversion like codes.
949 (try_to_simplify): Call it.
951 2007-09-07 Richard Guenther <rguenther@suse.de>
952 Uros Bizjak <ubizjak@gmail.com>
954 PR tree-optimization/32821
955 * tree_if_conv.c (combine_blocks): Use alloc_stmt_list instead of
956 NULL in the call to set_bb_stmt_list.
958 2007-09-07 Richard Sandiford <richard@codesourcery.com>
960 * config/mips/mips.c (build_mips16_call_stub): Emit all direct
961 float calls here, rather than leaving some to the caller.
962 Use call_internal_direct and call_value_internal_direct.
963 * config/mips/mips.md (call_internal_direct): New pattern.
964 (call_value_internal_direct): Likewise.
966 2007-09-07 Richard Sandiford <richard@codesourcery.com>
968 * config/mips/mips.c (mips_base_move_loop_invariants): New variable.
969 (mips_set_mips16_mode): Restore flag_move_loop_invariants, then set
971 (override_options): Set mips_base_move_loop_invariants.
973 2007-09-07 Richard Sandiford <richard@codesourcery.com>
975 * config/mips/mips.opt (mcode-readable=): Move to keep list
978 2007-09-07 Richard Sandiford <richard@codesourcery.com>
980 * doc/invoke.texi: Document the MIPS -mlocal-sdata, -mextern-sdata
981 and -mgpopt options. Adjust the -G documentation to match.
982 * config/mips/mips.opt (mextern-data, mgpopt, mlocal-data): New.
983 * config/mips/mips.c (mips_rtx_constant_in_small_data_p): New
984 function, split out from mips_classify_symbol. Return false for
986 (mips_classify_symbol): Call mips_rtx_constant_in_small_data_p.
987 Only use GP-relative accesses for non-pool symbols if TARGET_GPOPT.
988 (override_options): Check whether the -mgpopt setting is consistent
989 with the other flags.
990 (symbolic_expression_p): Delete.
991 (mips_select_rtx_section): Use mips_rtx_constant_in_small_data_p
992 and default_elf_select_rtx_section.
993 (mips_in_small_data_p): Honor section attributes for MIPS16 too.
994 Return false for local data unless TARGET_LOCAL_SDATA. Likewise
995 external data and TARGET_EXTERN_SDATA.
997 2007-09-07 Tobias Burnus <burnus@net-b.de>
1000 * doc/cpp.texi (Common Predefined Macros): Add __GFORTRAN__.
1002 2007-09-05 Laurynas Biveinis <laurynas.biveinis@gmail.com>
1004 * regrename.c (copyprop_hardreg_forward_1): New variable next. Use
1005 FOR_BB_INSNS_SAFE instead of for loop.
1006 * cse.c (cse_extended_basic_block): Likewise.
1007 * postreload.c (reload_cse_regs_1): New variable next. Make sure
1008 that the for loop does not invoke NEXT_INSN on a deleted insn.
1009 * function.c (instantiate_virtual_regs): Likewise.
1010 * lower-subreg.c (remove_retval_note): Likewise.
1011 (decompose_multiword_subregs): Use FOR_BB_INSNS_SAFE instead of
1013 * emit-rtl.c (remove_insn): Set NEXT_INSN and PREV_INSN to NULL on
1015 * cfgrtl.c (delete_insn): Set JUMP_LABEL to NULL on a deleted
1016 insn, if it's a jump.
1017 (try_redirect_by_replacing_jump): New variable jump_p. Call
1018 tablejump_p before delete_insn_chain.
1019 * reload1.c (reload): New variable next. Make sure that the for
1020 loop does not invoke NEXT_INSN on a deleted insn.
1021 (fixup_eh_region_note): Make the loop terminate if i becomes NULL.
1022 (delete_output_reload): New variable prev. Make sure the the for
1023 loops do not invoke PREV_INSN on a deleted insn.
1025 2007-09-06 Zdenek Dvorak <ook@ucw.cz>
1027 * cgraphbuild.c (rebuild_cgraph_edges): Export.
1028 * cgraph.h (rebuild_cgraph_edges): Declare.
1029 * tree-pass.h (pass_expand_omp_ssa): New.
1030 * omp-low.c (find_omp_clause): Export.
1031 (copy_var_decl): Split from omp_copy_decl_2.
1032 (build_omp_barrier): Return the call to emit instead of emitting
1034 (lower_rec_input_clauses, expand_omp_single): Gimplify the result of
1036 (extract_omp_for_data, expand_parallel_call, expand_omp_parallel,
1037 expand_omp_for_generic, expand_omp_for_static_nochunk,
1038 expand_omp_for_static_chunk, expand_omp_for, expand_omp_sections):
1039 Adapted to work on SSA form.
1040 (execute_expand_omp): Do not invalidate dominance information.
1041 (gate_expand_omp): Do not run with -fopenmp-ssa flag.
1042 (gate_expand_omp_ssa, pass_expand_omp_ssa): New.
1043 * gimplify.c (gimplify_omp_for): Ensure that the control variable is
1045 (force_gimple_operand): Allow gimplifying code expressions without
1047 * tree-predcom.c (mark_virtual_ops_for_renaming): Handle phi nodes.
1048 * common.opt (fopenmp-ssa): New.
1049 * tree-flow.h (find_omp_clause, copy_var_decl): Declare.
1050 * Makefile.in (tree-cfg.o): Add TREE_INLINE_H dependency.
1051 * tree-cfg.c: Include tree-inline.h.
1052 (struct move_stmt_d): Replace vars_to_remove by vars_map field.
1053 (replace_by_duplicate_decl, replace_ssa_name,
1054 mark_virtual_ops_in_region): New functions.
1055 (move_stmt_r, move_block_to_fn, move_sese_region_to_fn): Adapted
1056 to work on SSA form.
1057 * passes.c (init_optimization_passes): Add pass_expand_omp_ssa pass.
1058 * tree-ssa-operands.c (get_expr_operands): Handle operands of OMP
1061 2007-09-06 Laurynas Biveinis <laurynas.biveinis@gmail.com>
1063 * tree-loop-linear.c: Include obstack.h.
1064 (linear_transform_loops): New obstack lambda_obstack.
1065 Initialize it, pass it to gcc_loopnest_to_lambda_loopnest,
1066 lambda_loopnest_transform, lambda_loopnest_to_gcc_loopnest calls
1067 and free afterwards.
1068 * lambda.h (struct obstack): New forward declaration.
1069 (lambda_linear_expression_new): New parameter of type struct
1071 (lambda_loopnest_new): Likewise.
1072 (lambda_loopnest_transform): Likewise.
1073 (lambda_body_vector_new): Likewise.
1074 (lambda_body_vector_compute_new): Likewise.
1075 (gcc_loopnest_to_lambda_loopnest): Likewise.
1076 (lambda_loopnest_to_gcc_loopnest): Likewise.
1077 * lambda-code.c: Include obstack.h.
1078 (lambda_lattice_new): New parameter lambda_obstack. Use it for
1080 (lambda_body_vector_new): Likewise.
1081 (lambda_linear_expression_new): Likewise.
1082 (lambda_lattice_new): Likewise.
1083 (lambda_loopnest_new): Likewise. Additionally use obstack to
1084 allocate LN_LOOPS(ret).
1085 (lambda_lattice_compute_base): New parameter lambda_obstack. Pass
1086 it to lambda_lattice_new.
1087 (lambda_body_vector_compute_new): New parameter lambda_obstack.
1088 Pass it to lambda_body_vector_new.
1089 (lambda_lattice_compute_base): New paramater lambda_obstack. Pass
1090 it to lambda_lattice_new.
1091 (compute_nest_using_fourier_motzkin): New parameter lambda_obstack.
1092 Pass it to lambda_loopnest_new, lambda_linear_expression_new.
1093 (lambda_compute_target_space): Likewise.
1094 (lambda_compute_auxillary_space): New parameter lambda_obstack.
1095 Pass it to lambda_lattice_compute_base and
1096 compute_nest_using_fourieer_motzkin.
1097 (lambda_loopnest_transform): New parameter lambda_obstack. Pass
1098 it to lambda_lattice_compute_base, lambda_lattice_auxillary_space
1099 and lambda_lattice_compute_target_space.
1100 (gcc_tree_to_linear_expression): Nex parameter lambda_obstack.
1101 Pass it to lambda_linear_expression_new.
1102 (gcc_loop_to_lambda_loop): New parameter lambda_obstack. Pass it
1103 to gcc_tree_to_linear_expression.
1104 (gcc_loopnest_to_lambda_loopnest): New parameter lambda_obstack.
1105 Pass it to gcc_loop_to_lambda_loop and lambda_loopnest_new.
1106 (lambda_loopnest_to_gcc_loopnest): New parameter lambda_obstack.
1107 Pass it to lambda_body_vector_new and
1108 lambda_body_vector_compute_new.
1109 * Makefile.in (tree-loop-linear.o): Add $(OBSTACK_H) dependency.
1110 (lambda-code.o): Likewise.
1112 2007-09-06 Chao-ying Fu <fu@mips.com>
1114 * ginclude/stdfix.h: New file.
1115 * Makefile.in (USER_H): Add $(srcdir)/ginclude/stdfix.h.
1116 (convert.o): Add dependence on fixed-value.h.
1117 * c-convert.c (convert): Support FIXED_POINT_TYPE.
1118 * c-cppbuiltin.c (builtin_define_fixed_point_constants): New function
1119 to define fixed-point constants.
1120 (c_cpp_builtins): Define fixed-point constants.
1121 * convert.c (fixed-value.h): New include.
1122 (convert_to_real): Update comment to include fixed-point.
1123 Support FIXED_POINT_TYPE.
1124 (convert_to_integer): Update comment to include fixed-point.
1125 Support FIXED_POINT_TYPE.
1126 (convert_to_complex): Support FIXED_POINT_TYPE.
1127 (convert_to_fixed): New function.
1128 * convert.h (convert_to_fixed): Declare.
1129 * genopinit.c: Add comment about $Q for only fixed-point modes.
1130 (optabs): Add fract_optab, fractuns_optab, satfract_optab,
1131 satfractuns_optab, add_optab, ssadd_optab, usadd_optab, sub_optab,
1132 sssub_optab, ussub_optab, smul_optab, ssmul_optab, usmul_optab,
1133 ssmadd_widen_optab, usmadd_widen_optab, ssdiv_optab, udiv_optab,
1134 usdiv_optab, ssashl_optab, usashl_optab, neg_optab, ssneg_optab,
1135 usneg_optab for fixed-point modes.
1136 (gen_insn): Add force_fixed to track the $Q format for all fixed-point
1138 * optabs.c (optab_for_tree_code): For *DIV_EXPR, LSHIFT_EXPR,
1139 PLUS_EXPR, MINUS_EXPR, MULT_EXPR, NEGATE_EXPR, return signed or
1140 unsigned saturation optabs, when type is saturating.
1141 (shift_optab_p): Return true for SS_ASHIFT or US_ASHIFT.
1142 (expand_fixed_convert): New function.
1143 (gen_fixed_libfunc, gen_signed_fixed_libfunc,
1144 gen_unsigned_fixed_libfunc, gen_int_fp_fixed_libfunc,
1145 gen_int_fp_signed_fixed_libfunc, gen_int_fixed_libfunc,
1146 gen_int_signed_fixed_libfunc, gen_int_unsigned_fixed_libfunc,
1147 gen_fract_conv_libfunc, gen_fractuns_conv_libfunc,
1148 gen_satfract_conv_libfunc, gen_satfractuns_conv_libfunc): New
1150 (init_optabs): Initialize ssadd_optab, usadd_optab, sssub_optab,
1151 ussub_optab, ssmul_optab, usmul_optab, ssmadd_widen_optab,
1152 usmadd_widen_optab, ssmsub_widen_optab, usmsub_widen_optab,
1153 ssdiv_optab, usdiv_optab, ssashl_optab, usashl_optab, ssneg_optab,
1154 usneg_optab, fract_optab, fractuns_optab, satfract_optab,
1156 Initialize fixed-point libraries, including add, ssadd, usadd, sub,
1157 sssub, ussub, mul, ssmul, usmul, div, ssdiv, udiv, usdiv, ashl,
1158 ssashl, usashl, ashr, lshr, neg, ssneg, usneg, cmp, fract, satfract,
1159 fractuns, satfractuns.
1160 * optabs.h (enum optab_index): Add OTI_ssadd, OTI_usadd, OTI_sssub,
1161 OTI_ussub, OTI_ssmul, OTI_usmul, OTI_ssdiv, OTI_usdiv, OTI_ssneg,
1162 OTI_usneg, OTI_ssashl, OTI_usashl, OTI_ssmadd_widen, OTI_usmadd_widen,
1163 OTI_ssmsub_widen, OTI_usmsub_widen.
1164 (ssadd_optab, usadd_optab, sssub_optab, ussub_optab, ssmul_optab,
1165 usmul_optab, ssdiv_optab, usdiv_optab, ssneg_optab, usneg_optab,
1166 ssashl_optab, usashl_optab, ssmadd_widen_optab, usmadd_widen_optab,
1167 umsub_widen_optab, usmsub_widen_optab): Define.
1168 (enum convert_optab_index): Add COI_fract, COI_fractuns, COI_satfract,
1170 (fract_optab, fractuns_optab, satfract_optab, satfractuns_optab):
1172 (expand_fixed_convert): Declare.
1173 * expr.c (convert_move): Support the move of fixed-point modes.
1174 (emit_move_insn_1): Handle fixed-point mode to move via integer.
1175 (categorize_ctor_elements_1): Handle FIXED_CST.
1176 (count_type_elements): Handle FIXED_POINT_TYPE.
1177 (expand_expr_real_1): For VECTOR_CST, check MODE_VECTOR_FRACT,
1178 MODE_VECTOR_UFRACT, MODE_VECTOR_ACCUM, MODE_VECTOR_UACCUM.
1180 For PLUS_EXPR and MINUS_EXPR, support saturating and non-saturating
1181 multiply and add/subtract for fixed-point types.
1182 For MULT_EXPR, *DIV_EXPR, *SHIFT_EXPR, if the mode if a fixed-point
1183 mode, we jump to binop directly.
1184 Support FIXED_CONVERT_EXPR.
1185 (do_store_flag): Check FIXED_CST to put a constant second.
1186 (vector_mode_valid_p): Handle MODE_VECTOR_FRACT,
1187 MODE_VECTOR_UFRACT, MODE_VECTOR_ACCUM, MODE_VECTOR_UACCUM.
1188 (const_vector_from_tree): Support FIXED_CST.
1189 * doc/extend.texi (Fixed-Point): New node.
1190 * doc/md.texi (ssadd, usadd, sssub, ussub, ssmul, usmul, ssdiv, usdiv,
1191 ssmadd, usmadd, ssmsub, usmsub, ssashl, usashl, ssneg, usneg, fract,
1192 satfract, fractuns, satfractuns): Document them.
1194 2007-09-07 Bernd Schmidt <bernd.schmidt@analog.com>
1196 * config/bfin/bfin.h (PREFERRED_RELOAD_CLASS): Don't reload autoinc
1197 addresses into I registers.
1199 2007-09-06 Jan Hubicka <jh@suse.cz>
1200 Andreas Tobler <a.tobler@schweiz.org>
1202 * config/darwin.c (machopic_indirect_data_reference): Avoid invalid
1204 (machopic_legitimize_pic_address): Likewise.
1206 2007-09-06 Andrew Pinski <andrew_pinski@playstation.sony.com>
1207 Jan Hubicka <jh@suse.cz>
1209 * config/spu/spu.md (floatsidf2): Use convert_optab_libfunc
1210 instead of ufloat_optab->handlers directly.
1211 (floatdidf2): Likewise.
1213 2007-09-06 Sandra Loosemore <sandra@codesourcery.com>
1215 * config/mips/mips.c: Include diagnostic.h.
1216 (mips_set_current_function): Check errorcount and sorrycount
1217 before generating RTL.
1219 2007-09-06 Richard Sandiford <richard@codesourcery.com>
1222 * config/mips/mips.c (mips_classify_symbolic_expression): New function.
1223 (mips_classify_address): Use it instead of mips_symbolic_constant_p.
1224 (print_operand_reloc): Likewise.
1226 2007-09-06 Janis Johnson <janis187@us.ibm.com>
1230 2007-09-06 Jan Hubicka <jh@suse.cz>
1232 * i386.c (ix86_expand_lround, ix86_expand_round): Update call of
1235 2007-09-06 Richard Sandiford <richard@codesourcery.com>
1237 * config/mips/mips.md (fixuns_truncdfsi2, fixuns_truncdfdi2)
1238 (fixuns_truncsfsi2, fixuns_truncsfdi2): Update calls to real_2expN.
1240 2007-09-05 Janis Johnson <janis187@us.ibm.com>
1242 * optabs.c (expand_float): Convert unsigned integer as signed only
1243 if it provides sufficient accuracy; add mode argument to real_2expN.
1244 (expand_fix): Fix comment typos; extend binary float into mode
1245 wider than destination for converion to unsigned integer; add mode
1246 argument to real_2expN.
1247 * real.c (real_2expN): Add mode argument to special-case decimal
1249 * real.h (real_2expN): Ditto.
1250 * fixed-value.c (check_real_for_fixed_mode): Add mode argument to
1252 (fixed_from_string): Ditto.
1253 (fixed_to_decimal): Ditto.
1254 (fixed_convert_from_real): Ditto.
1255 (real_convert_from_fixed): Ditto.
1256 * config/rs6000/rs6000.md (FP): Include DD and TD modes.
1257 * config/rs6000/dfp.md (extendddtd2, adddd3, addtd3, subdd3, subtd3,
1258 muldd3, multd3, divdd3, divtd3, cmpdd_internal1, cmptd_internal1,
1259 floatditd2, ftruncdd2, fixdddi2, ftrunctd2, fixddi2): New.
1261 2007-09-06 Tom Tromey <tromey@redhat.com>
1263 * tree-cfg.c (remove_bb): Only warn if line is non-zero.
1264 * c-pch.c (c_common_read_pch): Restore current location after
1266 * tree.c (expand_location): Update.
1267 (expr_filename): Changed return type. Unified the two cases.
1268 (expr_lineno): Likewise.
1269 (annotate_with_file_line): Don't use EXPR_LINENO and EXPR_FILENAME
1271 * toplev.c (line_table): Changed type.
1272 (general_init): Update.
1273 (realloc_for_line_map): New function.
1274 (general_init): Allocate line_table using GC.
1275 * fix-header.c (line_table): Changed type.
1276 (read_scan_file): Update.
1277 (read_scan_file): Update.
1278 * c-ppoutput.c (maybe_print_line): Update.
1279 (print_line): Update.
1280 (cb_line_change): Update.
1281 (cb_define): Update.
1282 (pp_file_change): Update.
1283 * c-opts.c (c_common_init_options): Update.
1284 (finish_options): Update.
1285 (push_command_line_include): Update.
1286 * c-lex.c (cb_line_change): Update.
1287 (cb_def_pragma): Update.
1288 (cb_define): Update.
1290 (c_lex_with_flags): Use cpp_get_token_with_location.
1291 * input.h (line_table): Changed type.
1292 (location_from_locus): New macro.
1293 * tree.h (EXPR_FILENAME): No longer an lvalue.
1294 (EXPR_LINENO): Likewise.
1295 (expr_locus, set_expr_locus): Declare separately for
1296 USE_MAPPED_LOCATION.
1297 (expr_filename, expr_lineno): Changed return type.
1298 * gimplify.c (tree_to_gimple_tuple): Use SET_EXPR_LOCUS.
1299 * cfgexpand.c (expand_gimple_cond_expr): Use location_from_locus.
1300 (expand_gimple_basic_block): Likewise.
1301 * final.c (final_scan_insn): Use expanded_location.
1303 2007-09-06 Richard Guenther <rguenther@suse.de>
1305 PR tree-optimization/32586
1306 * tree-ssa-sccvn.c (simplify_binary_expression): Avoid
1307 folding if nothing changed.
1308 (simplify_unary_expression): New function. Do tree combining
1309 on conversion like codes.
1310 (try_to_simplify): Call it.
1311 * builtins.c (fold_builtin_cexp): Fold the built expressions.
1312 * fold-const.c (fold_unary): Test result of get_callee_fndecl().
1314 2007-09-06 Jan Hubicka <jh@suse.cz>
1317 * cse.c (fold_rtx): Avoid invalid sharing.
1319 2007-09-06 Richard Sandiford <richard@codesourcery.com>
1321 * config/mips/mips.md (fixuns_truncdfsi2, fixuns_truncdfdi2)
1322 (fixuns_truncsfsi2, fixuns_truncsfdi2): Update calls to real_2expN.
1324 2007-09-06 Revital Eres <eres@il.ibm.com>
1326 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
1327 Fix offsettable memory reference for 750CL.
1329 2007-09-06 Pat Haugen <pthaugen@us.ibm.com>
1331 * reload.c (find_reloads_address_1): Try to preserve original
1332 base/index regclass of operands.
1333 * regrename.c (scan_rtx_address): Likewise.
1334 (replace_oldest_value_addr): Likewise.
1336 2007-09-06 Richard Guenther <rguenther@suse.de>
1338 PR tree-optimization/33302
1339 * domwalk.c (walk_dominator_tree): The exit block is
1340 interesting even if it is not reachable.
1342 2007-09-06 Richard Sandiford <richard@codesourcery.com>
1345 * optabs.c (avoid_expensive_constant): Do nothing if MODE is VOIDmode.
1347 2007-09-06 Basile Starynkevitch <basile@starynkevitch.net>
1349 * cfg.c (dump_bb_info, dump_edge_info): Added cfun test for
1352 2007-09-05 Ian Lance Taylor <iant@google.com>
1354 * tree-pretty-print.c (dump_decl_name): Cast LABEL_DECL_UID to int
1356 (dump_generic_node): Likewise.
1357 * print-rtl.c (print_decl_name): Likewise.
1358 * print-tree.c (print_node_brief): Likewise.
1359 (print_node): Likewise.
1360 * Makefile.in (RTL_BASE_H): Add alias.h.
1363 2007-09-06 Jesper Nilsson <jesper.nilsson@axis.com>
1365 * longlong.h [__CRIS_arch_version >= 8] (count_trailing_zeros):
1367 * config/cris/cris.md (ctzsi2, cris_swap_bits): Implemented.
1368 * config/cris/cris.h (CTZ_DEFINED_VALUE_AT_ZERO): Defined.
1370 2007-09-06 Jie Zhang <jie.zhang@analog.com>
1372 * config.gcc (tm_file): Add linux.h for bfin*-uclinux*.
1373 (tm_defines): Define UCLIBC_DEFAULT to 1.
1374 (extra_options): Add linux.opt.
1375 * config/bfin/linux.h (CPLUSPLUS_CPP_SPEC): Don't define.
1376 (CRT_CALL_STATIC_FUNCTION): Likewise.
1377 (NO_IMPLICIT_EXTERN_C): Likewise.
1378 (TARGET_OS_CPP_BUILTINS): Define as LINUX_TARGET_OS_CPP_BUILTINS.
1379 * config/bfin/elf.h (OBJECT_FORMAT_ELF): Don't define.
1380 * config/bfin/uclinux.h (CPLUSPLUS_CPP_SPEC): Don't define.
1381 (ENDFILE_SPEC): Don't define.
1382 (LIB_SPEC): Likewise.
1383 (CRT_CALL_STATIC_FUNCTION): Likewise.
1384 (NO_IMPLICIT_EXTERN_C): Likewise.
1385 (LINUX_TARGET_OS_CPP_BUILTINS): Likewise.
1386 (TARGET_OS_CPP_BUILTINS): Define as LINUX_TARGET_OS_CPP_BUILTINS.
1388 2007-09-06 Jan Hubicka <jh@suse.cz>
1390 * i386.c (ix86_expand_lround, ix86_expand_round): Update call of
1393 2007-09-06 Jan Hubicka <jh@suse.cz>
1395 * opts.c (common_handle_option): Enable inlining functions for
1398 2007-09-06 Jakub Jelinek <jakub@redhat.com>
1400 * builtin-attrs.def (ATTR_NONNULL_3): New.
1401 (DEF_FORMAT_ATTRIBUTE): Use just ATTR_NONNULL_##FA instead of
1402 ATTR_NOTHROW_NONNULL_##FA.
1403 (DEF_FORMAT_ATTRIBUTE_NOTHROW, DEF_FORMAT_ATTRIBUTE_BOTH): New macros.
1404 (ATTR_FORMAT_PRINTF_NOTHROW_2_0, ATTR_PRINTF_NOTHROW_2_3,
1405 ATTR_FORMAT_PRINTF_NOTHROW_3_0, ATTR_FORMAT_PRINTF_NOTHROW_3_4,
1406 ATTR_FORMAT_PRINTF_NOTHROW_4_0, ATTR_PRINTF_NOTHROW_4_5,
1407 ATTR_FORMAT_PRINTF_NOTHROW_5_0, ATTR_FORMAT_PRINTF_NOTHROW_5_6,
1408 ATTR_FORMAT_SCANF_NOTHROW_2_0, ATTR_FORMAT_SCANF_NOTHROW_2_3,
1409 ATTR_FORMAT_STRFTIME_NOTHROW_3_0, ATTR_FORMAT_STRFMON_NOTHROW_3_4):
1411 (ATTR_FORMAT_PRINTF_4_0, ATTR_PRINTF_4_5, ATTR_FORMAT_PRINTF_5_0,
1412 ATTR_FORMAT_PRINTF_5_6, ATTR_FORMAT_STRFTIME_3_0,
1413 ATTR_FORMAT_NOTHROW_3_4): Remove.
1414 * builtins.def (snprintf, sprintf, sscanf, vsnprintf, vsprintf,
1415 vsscanf, strfmon, strftime, __snprintf_chk, __sprintf_chk,
1416 __vsnprintf_chk, __vsprintf_chk): Use ATTR_FORMAT_*_NOTHROW_*
1417 instead of ATTR_FORMAT_*_*.
1419 2007-09-06 Hans-Peter Nilsson <hp@axis.com>
1421 * config/cris/cris.c (cris_emit_movem_store): Call copy_rtx on
1422 parts re-used for REG_FRAME_RELATED_EXPR.
1424 2007-09-05 Janis Johnson <janis187@us.ibm.com>
1426 * optabs.c (expand_float): Convert unsigned integer as signed only
1427 if it provides sufficient accuracy; add mode argument to real_2expN.
1428 (expand_fix): Fix comment typos; extend binary float into mode
1429 wider than destination for converion to unsigned integer; add mode
1430 argument to real_2expN.
1431 * real.c (real_2expN): Add mode argument to special-case decimal
1433 * real.h (real_2expN): Ditto.
1434 * fixed-value.c (check_real_for_fixed_mode): Add mode argument to
1436 (fixed_from_string): Ditto.
1437 (fixed_to_decimal): Ditto.
1438 (fixed_convert_from_real): Ditto.
1439 (real_convert_from_fixed): Ditto.
1440 * config/rs6000/rs6000.md (FP): Include DD and TD modes.
1441 * config/rs6000/dfp.md (extendddtd2, adddd3, addtd3, subdd3, subtd3,
1442 muldd3, multd3, divdd3, divtd3, cmpdd_internal1, cmptd_internal1,
1443 floatditd2, ftruncdd2, fixdddi2, ftrunctd2, fixddi2): New.
1445 2007-09-05 Ian Lance Taylor <iant@google.com>
1447 * init-regs.c (initialize_uninitialized_regs): Call
1448 emit_insn_before rather than add_insn_before.
1450 2007-09-05 Jakub Jelinek <jakub@redhat.com>
1452 * builtins.def (BUILT_IN_VA_ARG_PACK): New built-in.
1453 * tree.h (CALL_EXPR_VA_ARG_PACK): Define.
1454 * tree-inline.h (copy_body_data): Add call_expr field.
1455 * tree-inline.c (expand_call_inline): Initialize call_expr.
1456 (copy_bb): Append anonymous inline fn arguments to arguments
1457 when inlining a CALL_EXPR_VA_ARG_PACK call.
1458 * builtins.c (expand_builtin): Issue an error if
1459 BUILT_IN_VA_ARG_PACK is seen during expand.
1460 (fold_call_expr, fold_builtin_call_array): Don't fold
1461 CALL_EXPR_VA_ARG_PACK CALL_EXPRs or calls with
1462 __builtin_va_arg_pack () call as last argument.
1463 * gimplify.c (gimplify_call_expr): If last argument to a vararg
1464 function is __builtin_va_arg_pack (), decrease number of call
1465 arguments and instead set CALL_EXPR_VA_ARG_PACK on the CALL_EXPR.
1466 * expr.c (expand_expr_real_1): Issue an error if
1467 CALL_EXPR_VA_ARG_PACK CALL_EXPR is seen during expand.
1468 * tree-pretty-print.c (dump_generic_node): Handle printing
1469 CALL_EXPR_VA_ARG_PACK bit on CALL_EXPRs.
1470 * doc/extend.texi (__builtin_va_arg_pack): Document.
1472 2007-09-05 Adam Nemet <anemet@caviumnetworks.com>
1474 PR tree-optimization/21513
1475 * builtins.c (build_builtin_expect_predicate): New function.
1476 (fold_builtin_expect): Add argument for expected value.
1477 Distribute __builtin_expect over short-circuiting operations.
1478 Fold nested builtin_expects.
1479 (fold_builtin_2): Adjust call to fold_builtin_expect.
1481 2007-09-05 Rask Ingemann Lambertsen <rask@sygehus.dk>
1484 PR tree-optimization/13756
1485 * doc/invoke.texi (Options That Control Optimization): Document
1488 2007-09-05 Uros Bizjak <ubizjak@gmail.com>
1490 * gcc/config/i386/cpuid.h: New file.
1491 * gcc/config/i386/driver-i386.c: Include cpuid.h.
1492 (describe_cache): Shrink size and line strings to 100 bytes.
1493 (detect_caches_amd): Return "" for unsupported max_ext_level.
1494 Use __cpuid function.
1495 (detect_caches_intel): Return "" for unsupported max_level.
1496 Use __cpuid function.
1497 (host_detect_local_cpu): Change feature flag variables to
1498 unsigned int. Initialize only extended feature flag variables.
1499 Use __get_cpuid_max to determine max supported cpuid level.
1500 Use __cpuid function to determine supported features. Fix
1501 calculation of family id. Remove is_amd and check signature
1502 directly. Check for Geode signature. Handle family 4 id.
1503 [PROCESSOR_GENERIC32]: New default for unknown family id. Move
1504 cpu discovery code to lower part of the function.
1505 [PROCESSOR_PENTIUM, PROCESSOR_K6, PROCESSOR_ATHLON]: Do not tune
1506 for sub-architecture.
1507 [PROCESSOR_PENTIUMPRO]: Simplify cpu discovery code.
1508 [PROCESSOR_K8]: Add k8-sse3 architecture.
1509 [PROCESSOR_NOCONA]: Remove.
1510 [PROCESSOR_GENERIC64]: Ditto.
1511 * gcc/config/i386/x-i386 (driver-i386.o): Depend on cpuid.h.
1512 * gcc/config/i386/crtfastmath.c: Include cpuid.h. Use __get_cpuid
1513 to check for SSE and FXSAVE support.
1514 * gcc/config/i386/t-crtfm (crtfastmath.o): Depend on cpuid.h.
1515 Add -minline-all-stringops.
1516 * gcc/config.gcc (i[34567]86-*-*): Add cpuid.h to extra_headers.
1517 (x86_64-*-*): Ditto.
1519 2007-09-05 Jie Zhang <jie.zhang@analog.com>
1521 * config/bfin/linux-unwind.h: New file.
1522 * config/bfin/linux.h (MD_UNWIND_SUPPORT): Define.
1523 * config/bfin/uclinux.h (MD_UNWIND_SUPPORT): Define.
1525 2007-09-05 Laurynas Biveinis <laurynas.biveinis@gmail.com>
1527 * Makefile.in (stmt.o): Add alloc-pool.h dependency.
1528 * stmt.c: Include alloc-pool.h.
1529 (struct case_node): Remove GTY marker.
1530 (add_case_node): New parameter case_node_pool. Use it for
1532 (expand_case): New allocation pool case_node_pool. Initialize it,
1533 pass to add_case_node and free it.
1535 2007-09-05 Sandra Loosemore <sandra@codesourcery.com>
1536 David Ung <davidu@mips.com>
1537 Nigel Stephens <nigel@mips.com>
1539 Add mips16/nomips16 function attributes and -mflip-mips16 option
1540 for testing mixed-mode compilation.
1542 * config/mips/mips.opt (mflip-mips16): New.
1544 * config/mips/mips.h (SYMBOL_FLAG_MIPS16_FUNC): Define.
1545 (SYMBOL_FLAG_MIPS16_FUNC_P): Define.
1547 * config/mips/mips.c (mips_base_target_flags): New.
1548 (mips_base_mips16): New.
1549 (mips_base_schedule_insns): New.
1550 (mips_base_reorder_blocks_and_partition): New.
1551 (mips_base_align_loops): New.
1552 (mips_base_align_jumps): New.
1553 (mips_base_align_functions): New.
1554 (mips16_flipper): New.
1555 (mips_attribute_table): Add "mips16" and "nomips16" entries.
1556 (TARGET_SET_CURRENT_FUNCTION): Define.
1557 (mips_mips16_type_p, mips_nomips16_type_p): New.
1558 (mips_comp_type_attributes): Check mips16/nomips16 attributes.
1559 (mips_function_ok_for_sibcall): Make it deal with functions with
1561 (mips_init_split_addresses): New, split out from override_options.
1562 (mips_init_relocs): New, split out from override_options.
1563 (was_mips16_p): New.
1564 (mips_set_mips16_mode): New, split out from override_options.
1565 (mips_set_current_function): New.
1566 (override_options): Add sorry for unsupported mips16/pic
1567 combination. Remove error for mips16/dsp combination. Save
1568 base option settings.
1569 (mips_file_start): Move mips16 mode setting output from here....
1570 (mips_output_function_prologue): ....to here.
1571 (mips_output_mi_thunk): Check for mips16 function.
1572 (build_mips16_function_stub): Don't set .mips16 here.
1573 (build_mips16_call_stub): Likewise.
1574 (mips_expand_builtin): Error in mips16 mode.
1575 (mips_use_mips16_mode_p): New.
1576 (mips_encode_section_info): Check for mips16 function, and set
1577 SYMBOL_REF_FLAGS accordingly.
1579 * doc/extend.texi (Function Attributes): Document new
1580 mips16/nomips16 attributes.
1581 * doc/invoke.texi (Option Summary): Add -mflip-mips16.
1582 (MIPS Options): Document -mflip-mips16.
1584 2007-09-05 Sandra Loosemore <sandra@codesourcery.com>
1586 * config/mips/mips.c (mips_legitimize_tls_address): Call sorry
1587 if we encounter TLS address in MIPS16 mode.
1588 (mips_legitimize_const_move): Check cannot_force_const_mem for
1589 the (const (plus symbol offset)) case; this forces invalid TLS
1590 address in MIPS16 mode to be caught by the above call to sorry.
1591 (override_options): Don't reset targetm.have_tls in MIPS16 mode,
1592 because that now enables emutls, which is not ABI compatible
1593 with native TLS in non-MIPS16 mode.
1595 2007-09-05 Sandra Loosemore <sandra@codesourcery.com>
1597 Add target hook invoked when cfun changes.
1599 * doc/tm.texi (TARGET_SET_CURRENT_FUNCTION): Document.
1600 * target.h (struct gcc_target): Add set_current_function.
1601 * target-def.h (TARGET_SET_CURRENT_FUNCTION): Define.
1602 (TARGET_INITIALIZER): Add initializer for set_current_function.
1603 * tree.h (push_struct_function): New.
1604 * tree-inline.h (push_cfun, pop_cfun): Move declarations to...
1606 (set_cfun): Declare.
1607 * tree-inline.c (cfun_stack, push_cfun, pop_cfun): Moved to...
1609 (push_function_context_to): Use allocate_struct_function
1610 to create null context, not init_dummy_function_start. Use set_cfun.
1611 (pop_function_context_from): Use set_cfun.
1612 (in_dummy_function): New.
1613 (invoke_set_current_function_hook): New.
1615 (push_cfun, pop_cfun): Use set_cfun.
1616 (push_struct_function): New.
1617 (allocate_struct_function): Call invoke_set_current_function_hook
1619 (prepare_function_start): Don't set cfun here. Remove unused
1620 argument; fix all callers.
1621 (init_dummy_function_start): Fiddle with in_dummy_function. Call
1622 push_struct_function.
1623 (init_function_start): Set cfun here.
1624 (expand_dummy_function_end): Fiddle with in_dummy_function. Pop cfun.
1625 * omp-low.c (create_omp_child_function): Use push_struct_function
1626 and pop_cfun to save/restore state.
1627 (expand_omp_parallel): Remove unused saved_cfun variable.
1628 * cgraphunit.c (ipa_passes): Use set_cfun.
1629 * gimple-low.c (record_vars_into): Use push_cfun/pop_cfun here.
1630 * dwarf2out.c (dwarf2out_abstract_function): Likewise.
1631 * matrix-reorg.c (transform_allocation_sites): Likewise.
1632 (matrix_reorg): Use set_cfun.
1633 * gimplify.c (gimplify_function_tree): Use push_cfun/pop_cfun here.
1634 * tree-optimize.c (tree_rest_of_compilation): Remove one redundant
1635 assignment to cfun; use set_cfun for the other.
1636 * tree-cfg.c (move_sese_region_to_fn): Use set_cfun.
1637 (dump_function_to_file): Use push_cfun/pop_cfun here.
1638 * c-decl.c (finish_function): Use set_cfun.
1640 2007-09-05 Kenneth Zadeck <zadeck@naturalbridge.com>
1642 * regrename.c (rerename_optimize): Use deferred rescanning and
1643 insert explicit calls to rescan insns when changed.
1645 2007-09-05 Jan Hubicka <jh@suse.cz>
1647 * optabs.c (libfunc_entry): Change optab to integer.
1648 (hash_libfunc, convert_optab_libfunc, optab_libfunc,
1649 set_optab_libfunc, set_conv_libfunc): Use optable indexes instead
1650 of pointers for hashing.
1652 2007-09-05 Jan Hubicka <jh@suse.cz>
1654 * regrename.c (pass_regrename, pass_cprop_hardreg): Add RTL sharing
1656 * fwprop.c (pass_rtl_fwprop, pass_rtl_fwprop_add): Likewise.
1657 * see.c (pass_see): Likewise.
1658 * tracer.c (pass_tracer): Likewise.
1659 * postreload-gcse.c (pass_gcse2): Likewise.
1660 * postreload.c (pass_postreload_cse): Likewise.
1661 * mode-switching.c (pass_mode_switching): Likewise.
1662 * modulo-sched.c (pass_sms): Likewise.
1663 * cse.c (cse_insn): Likewise.
1664 * web.c (pass_web): Likweise.
1665 * combine-stack-adj.c (pass_stack_adjustments): Likewise.
1666 * dce.c (pass_ud_rtl_dce, pass_fast_rtl_dce): Likewise.
1667 * loop-init.c (pass_rtl_loop_init): Likewise.
1668 (pass_rtl_loop_done, pass_rtl_move_loop_, pass_rtl_unswitch,
1669 pass_rtl_unroll_and, pass_rtl_doloop): Likewise.
1670 * global.c (pass_global_alloc): Likewise.
1671 * ifcvt.c (pass_rtl_ifcvt, pass_if_after_combine,
1672 pass_if_after_reload): Likewise.
1673 * reload.c (pass_peephole2, pass_split_for_shoren_branches): Likewise.
1674 * dse.c (pass_rtl_dse1, pass_rtl_dse2): Likewise.
1675 * regmove.c (pass_regmove): Likewise.
1676 * function.c (pass_thread_prologugues_epilogues): Likewise.
1677 * gcse.c (pass_gcse): Likewise.
1678 * rtl-factoring.c (pass_rtl_seqabstr): Likewise.
1679 * lower-subreg.c (pass_lower_subreg2): Likewise.
1680 * bt-load.c (pass_branch_target_load): Likewise.
1681 * emit-rtl.c (pass_unshare_all_rtl): Likewise.
1682 * cfgcleanup.c (pass_jump, pass_jump2): Likewise.
1683 * combine.c (pass_combine): Likewise.
1684 * bb-reorder.c (pass_duplicate_comp, pass_reorder_blocks): Likewise.
1685 (pass_partition_blocks): Likewise.
1686 * var-tracking.c (pass_variable_track): Likewise.
1687 * reg-stack.c (pass_stack_regs_run): Likewise.
1688 * sched-rgn.c (pass_sched, pass_sched2): Likewise.
1689 * passes.c (pass_postreload): Likewise.
1690 (execute_function_todo): Add TODO_verify_rtl_sharing handling code.
1691 * tree-pass.h (TODO_verify_rtl_sharing): New.
1692 (TODO_update_ssa, TODO_update_ssa_no_phi, TODO_update_ssa_full_phi,
1693 TODO_update_ssa_only_virtuals, TODO_remove_unused_locals,
1694 TODO_set_props, TODO_df_finish, TODO_df_verify,
1695 TODO_mark_first_instance, TODO_rebuild_alias): Renumber.
1697 2007-09-05 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1699 * fold-const.c (all_ones_mask_p, sign_bit_p, simple_operand_p,
1700 fold_real_zero_addition_p, reorder_operands_p,
1701 div_if_zero_remainder, fold_undefer_overflow_warnings,
1702 int_binop_types_match_p, fold_convert_const_int_from_int,
1703 fold_convert_const_int_from_real,
1704 fold_convert_const_int_from_fixed,
1705 fold_convert_const_real_from_real,
1706 fold_convert_const_real_from_fixed,
1707 fold_convert_const_fixed_from_fixed,
1708 fold_convert_const_fixed_from_int,
1709 fold_convert_const_fixed_from_real, maybe_lvalue_p,
1710 fold_checksum_tree, fold_check_failed, print_fold_checksum,
1711 debug_fold_checksum, multiple_of_p): Constify.
1712 * tree-flow-inline.h (get_lineno): Likewise.
1713 * tree-flow.h (get_lineno): Likewise.
1714 * tree-object-size.c (compute_object_offset, addr_object_size,
1715 alloc_object_size, pass_through_call): Likewise.
1716 * tree-pretty-print.c (op_symbol, print_call_name,
1717 print_struct_decl, do_niy): Likewise.
1718 * tree.h (fold_undefer_overflow_warnings, multiple_of_p,
1719 debug_fold_checksum): Likewise.
1721 2007-09-04 Steven Bosscher <steven@gcc.gnu.org>
1724 * lower-subreg.c (resolve_clobber): If we remove a REG_LIBCALL
1725 note, remove the associated REG_RETVAL note.
1727 2007-09-04 Andrew Pinski <andrew_pinski@playstation.sony.com>
1729 * config.gcc (powerpc*-*-*): Install
1730 spu2vmx.h, vec_types.h, and si2vmx.h headers.
1731 * config/rs6000/spu2vmx.h: New header.
1732 * config/rs6000/si2vmx.h: New header.
1733 * config/rs6000/vec_types.h: New header.
1735 2007-09-05 Ben Elliston <bje@au.ibm.com>
1737 * varasm.c (initializer_constant_valid_p): Fix comment typo.
1739 2007-09-05 Ben Elliston <bje@au.ibm.com>
1741 * config/rs6000/ppu_intrinsics.h (__ldarx): Use `Z' constraint,
1743 (__stdcx): Make asm volatile.
1744 (__stwcx, __stdcx): Return only 0 or 1.
1746 2007-09-05 Jan Hubicka <jh@suse.cz>
1750 2007-09-04 Jan Hubicka <jh@suse.cz>
1751 * tree-tailcall.c (eliminate_tail_call): Expect unrenamed return value.
1753 2007-09-04 Richard Guenther <rguenther@suse.de>
1754 * tree-ssa-operands.c (add_virtual_operand): Only mark
1755 stores as has_volatile_ops if alias information is not available.
1757 2007-09-05 Jakub Jelinek <jakub@redhat.com>
1759 * config/rs6000/tramp.asm: Include config.h.
1760 Check __PIC__ or __pic__ macro instead of SHARED.
1762 PR rtl-optimization/32300
1763 * see.c (see_copy_insn): New function.
1764 (see_def_extension_not_merged, see_merge_one_use_extension,
1765 see_merge_one_def_extension): Use it. Avoid changing
1766 PREV_INSN/NEXT_INSN chains directly, insted emit insns
1767 into sequences. Call df_insn_delete on temporary insns
1768 that won't be emitted into the insn stream.
1769 (rest_of_handle_see): Turn off DF_DEFER_INSN_RESCAN
1770 and run df_process_deferred_rescans () before run_fast_dce.
1772 PR tree-optimization/33017
1773 * tree-data-ref.c (split_constant_offset) <case SSA_NAME>: Don't
1774 recurse for pure or const function calls.
1776 2007-09-04 Laurynas Biveinis <laurynas.biveinis@gmail.com>
1778 * c-format.c: Include alloc-pool.h.
1779 (check_format_info_main): New argument fwt_alloc. Use allocation
1780 pool instead of GC. Remove GC deallocation code.
1781 (check_format_arg): Create allocation pool, pass it to
1782 check_format_info_main and free it afterwards.
1783 * Makefile.in (c-format.o): Add alloc-pool.h dependency.
1785 2007-09-05 Ben Elliston <bje@au.ibm.com>
1787 * config/rs6000/ppu_intrinsics.h (__protected_stream_count):
1788 Rename count macro arugment for clarity.
1789 (__protected_unlimited_stream_set): Pass UG=1 to __dcbt_TH1000.
1790 (__protected_stream_set): Likewise, pass 0.
1792 2007-09-04 Jan Hubicka <jh@suse.cz>
1794 * tree-tailcall.c (eliminate_tail_call): Expect unrenamed return value.
1796 2007-09-04 Jan Hubicka <jh@suse.cz>
1798 * invoke.texi (-finline-small-functions): Document.
1799 * ipa-inline.c (cgraph_default_inline_p): Do not use DECL_INLINE
1800 when deciding what is inlinable.
1801 (cgraph_decide_recursive_inlining): Handle flag_inline_functions.
1802 (cgraph_decide_inlining_of_small_function): Handle new flags.
1803 (cgraph_decide_inlining_incrementally): Likewise.
1804 * opts.c (decode_options): Enable flag_inline_small_functions at -O2
1805 * common.opt (finline-small-functions): New.
1806 * Makefile.in (build/gengtype.o-warn): Work around PR29478
1808 2007-09-04 Richard Guenther <rguenther@suse.de>
1810 * tree-ssa-operands.c (add_virtual_operand): Only mark
1811 stores as has_volatile_ops if alias information is not available.
1813 2007-09-04 Andrew Haley <aph@redhat.com>
1815 * config/arm/libgcc-bpabi.ver: Add _Unwind_Backtrace as GCC_4.3.0.
1817 2007-09-04 Andrew Haley <aph@redhat.com>
1819 * config/arm/libunwind.S (UNWIND_WRAPPER _Unwind_Backtrace): New.
1820 * config/arm/unwind-arm.h (__gnu_Unwind_Backtrace): New.
1821 * config/arm/unwind-arm.c (__gnu_Unwind_Backtrace): New.
1823 2007-09-04 Emmanuel Thome <Emmanuel.Thome@inria.fr>
1826 * common.opt: Mention ABI version 2 in comment.
1828 2007-09-04 Jan Hubicka <jh@suse.cz>
1830 * optabs.c (debug_optab_libfunc): Update; make available to gdb.
1831 (libfunc_entry): New structure.
1832 (libfunc_hash): New hashtable.
1833 (hash_libfunc): New function.
1834 (eq_libfunc): New function.
1835 (convert_optab_libfunc): New function.
1836 (optab_libfunc): New function.
1837 (expand_binop, sign_expand_binop, expand_twoval_binop_libfunc,
1838 expand_unop, prepare_cmp_insn, prepare_float_insn, gen_add2_insn,
1839 expand_float, expand_fix, new_optab, new_convert_optab):
1840 Update for new libfunc API.
1841 (init_libfunc, init_integral_libfuncs,
1842 init_floating_libfuncs, init_interclass_conv_libfuncs
1843 init_intraclass_conv_libfuncs): Remove; reorganize all logic to:
1844 (gen_libfunc, gen_int_libfunc, gen_fp_libfunc, gen_int_fp_libfunc,
1845 gen_intv_fp_libfunc, gen_interclass_conv_libfunc,
1846 gen_int_to_fp_conv_libfunc, gen_ufloat_conv_libfunc,
1847 gen_int_to_fp_nondecimal_conv_libfunc, gen_fp_to_int_conv_libfunc,
1848 gen_intraclass_conv_libfunc, gen_trunc_conv_libfunc,
1849 gen_extend_conv_libfunc): New.
1850 (init_one_libfunc): Revamp for hashtables.
1851 (set_conv_libfunc): Likewise.
1852 (init_optabs): Initialize hashtable; use lazy initialization where
1854 * optabs.h (optab_handlers): Move out of GGC.
1855 (optab, convert_optab): Move out of GGC; add lazy gen info.
1856 (code_to_optab, convert_optab_table, optab_table): Move out of GGC.
1857 (optab_libfunc, convert_optab_libfunc): New.
1858 * builtins.c (expand_builtin_powi): Update for new API.
1859 * expr.c (convert_move): Likewise.
1860 * expmed.c (expand_divmod): Likewise.
1862 2007-09-04 Daniel Jacobowitz <dan@codesourcery.com>
1864 * config/rs6000/rs6000.c (rs6000_stack_info): Allocate space for the
1865 GOT pointer only if there is a constant pool. Use the allocated space
1868 2007-09-04 Daniel Jacobowitz <dan@codesourcery.com>
1870 * config/rs6000/linux-unwind.h (ppc_fallback_frame_state): Correct
1873 2007-09-04 Revital Eres <eres@il.ibm.com>
1874 Richard Guenther <rguenther@suse.de>
1875 R. Clint Whaley <whaley@cs.utsa.edu>
1877 * doc/invoke.texi (-fassociative-math, -freciprocal-math):
1879 * tree-tailcall.c (process_assignment): Use -fassociative-math
1880 when reodering operands of floating-point type.
1881 * fold-const.c (fold_comparison, fold_binary): Use
1882 -fassociative-math and -freciprocal-math instead of
1883 -funsafe-math-optimization flag.
1884 * toplev.h (set_unsafe_math_optimizations_flags): Declare function.
1885 * tree-ssa-math-opts.c (gate_cse_reciprocals): Use
1886 -freciprocal-math instead of -funsafe-math-optimizations.
1887 * opts.c (set_fast_math_flags): Set -freciprocal-math and
1888 -fassociative-math when -ffast-math is set.
1889 (set_unsafe_math_optimizations_flags): New Function
1890 to set -freciprocal-math and -fassociative-math when
1891 -funsafe-math-optimizations is set.
1892 (common_handle_option): Call it.
1893 * tree-vectorizer.c (vect_is_simple_reduction): Use
1894 -fassociative-math when doing reduction on floats.
1895 * loop-unroll.c (analyze_insn_to_expand_var): Use
1896 -fassociative-math when expanding an accumulator of type float.
1897 * simplify-rtx.c (simplify_binary_operation_1): Use
1898 -fassociative-math and -freciprocal-math when reordeing operands
1899 of floating-point type.
1900 * combine.c (combine_simplify_rtx): Likewise.
1901 * tree-ssa-reassoc.c (break_up_subtract_bb, reassociate_bb):
1903 * common.opt (-fassociative-math, -freciprocal-math): New flags.
1905 2007-09-04 Paolo Carlini <pcarlini@suse.de>
1908 * doc/install.texi (--enable-__cxa_atexit): Fix typo.
1910 2007-09-04 Jan Hubicka <jh@suse.cz>
1912 * regrename.c(copyprop_hardreg_forward_1): Unshare when doing copyprop.
1914 2007-09-04 Jan Hubicka <jh@suse.cz>
1916 * regrename.c (build_def_use): Don't share RTL between MATCH_OPERATOR
1917 and corresponding MATCH_DUP.
1919 2007-09-04 Uros Bizjak <ubizjak@gmail.com>
1922 * combine.c (subst): Do not try to simplify X if it represents load
1923 of FP constant from the constant pool via float extension.
1925 2007-09-04 Ben Elliston <bje@au.ibm.com>
1927 * c-opts.c: Include "tm_p.h".
1928 * Makefile.in (c-opts.o): Add $(TM_P_H) as a dependency.
1930 2007-09-04 Richard Guenther <rguenther@suse.de>
1932 PR tree-optimization/33291
1933 * tree-pretty-print.c (dump_generic_node): Dump all
1934 qualifiers for pointer types, not only first. Dump
1935 qualifiers for aggregate types as well.
1936 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Always
1937 use the canonical type for building ARRAY_REFs.
1938 * gimplify.c (canonicalize_addr_expr): Clean up. The
1939 correct validness check is compatibility of the pointer
1940 types. Always use the canonical type for building
1941 ARRAY_REFs and ADDR_EXPRs.
1942 * tree-ssa-forwprop.c (forward_propagate_addr_expr): Revert
1943 change that disabled propagation of ADDR_EXPRs into statements
1946 2007-09-03 Zack Weinberg <zack@codesourcery.com>
1948 * config/arm/arm.md (mulsidi3, umulsidi3, mulsi3_highpart)
1949 (umulsi3_highpart): Make into expanders; existing insns are
1950 now named *[insn]_nov6. Add stars to existing [insn]_v6
1951 counterparts' names.
1953 2007-09-03 Daniel Jacobowitz <dan@codesourcery.com>
1955 * Makefile.in (DRIVER_DEFINES, gcc.o, gccspec.o, protoize.o)
1956 (unprotoize.o): Remove SHLIB_MULTILIB.
1958 2007-09-03 Eric Weddington <eweddington@cso.atmel.com>
1960 * doc/extend.texi: Document progmem attribute for AVR.
1962 2007-09-03 Anatoly Sokolov <aesok@post.ru>
1965 * config/avr/avr.h (TARGET_VTABLE_ENTRY_ALIGN): Define.
1967 2007-09-03 Zack Weinberg <zack@codesourcery.com>
1969 * optabs.c (expand_ffs): Initialize val and defined_at_zero
1972 * optabs.c: Remove unnecessary forward declarations.
1973 (expand_unop_direct): New, broken out of expand_unop.
1974 (expand_doubleword_clz): New.
1975 (expand_ctz): Move above expand_ffs. Use
1976 start_sequence, end_sequence, add_equal_note, and
1977 expand_unop_direct. Add more commentary.
1978 (expand_ffs): Try both ctz optab and expand_ctz.
1979 Generate a test and branch if the hardware doesn't give us
1980 a useful value for input zero. Style improvements similar to
1983 * config/arm/arm.md (ffssi2, ctzsi2): Delete.
1985 2007-09-03 Zack Weinberg <zack@codesourcery.com>
1987 * config.gcc: Delete stanza for arm-semi-aof and
1988 armel-semi-aof targets.
1989 * config/arm/arm-protos.h
1991 * config/arm/arm.h: Delete all #ifdef AOF_ASSEMBLER blocks;
1992 make all #ifndef AOF_ASSEMBLER blocks unconditional. Also
1993 delete aof_pic_label and remove mention of AOF in comments.
1994 * config/arm/arm.md: Delete patterns used only for AOF assembly.
1996 * config/arm/semiaof.h
1997 * config/arm/t-semi: Delete file.
1999 2007-09-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2001 * system.h (CONST_CAST2, CONST_CAST_TREE, CONST_CAST_RTX,
2002 CONST_CAST_BB): New macros for type-specific casts.
2003 (CONST_CAST): Add a TYPE parameter and define it in terms of
2006 * c-decl.c (c_make_fname_decl): Supply a TYPE for CONST_CAST.
2007 * c-lex.c (cb_ident, lex_string): Likewise,
2008 * c-typeck.c (free_all_tagged_tu_seen_up_to): Likewise,
2009 * config/i386/i386.c (ix86_function_regparm,
2010 ix86_function_sseregparm): Use CONST_CAST_TREE.
2011 * config/mmix/mmix.c (mmix_encode_section_info): Supply a TYPE for
2013 * gcc.c (set_spec, read_specs, for_each_path, execute, do_spec_1,
2014 give_switch, set_multilib_dir): Likewise,
2015 * gengtype-parse.c (string_seq, typedef_name): Likewise,
2016 * gimple-low.c (block_may_fallthru): Use CONST_CAST_TREE.
2017 * jump.c (reversed_comparison_code_parts): Use CONST_CAST_RTX.
2018 * passes.c (execute_one_pass): Supply a TYPE for CONST_CAST.
2019 * prefix.c (update_path): Likewise,
2020 * pretty-print.c (pp_base_destroy_prefix): Likewise,
2021 * rtlanal.c (keep_with_call_p): Use CONST_CAST_RTX.
2022 * tree-cfg.c (tree_can_merge_blocks_p,
2023 tree_block_ends_with_condjump_p): Use CONST_CAST_BB.
2024 * tree-eh.c (lookup_stmt_eh_region_fn): USe CONST_CAST_TREE.
2025 * tree.c (build_string): Supply a TYPE for CONST_CAST.
2026 (attribute_list_contained): Use CONST_CAST_TREE.
2028 2007-09-03 Krister Walfridsson <cato@df.lth.se>
2030 * config/netbsd-elf.h (USE_LD_AS_NEEDED): Define.
2032 2007-09-03 Richard Sandiford <richard@codesourcery.com>
2035 * optabs.c (avoid_expensive_constant): Canonicalize CONST_INTs
2036 before forcing them into a register.
2038 2007-09-03 Richard Sandiford <richard@codesourcery.com>
2040 * config/mips/mips.md (fetchop_bit): Use define_code_iterator
2041 rather than define_code_macro.
2043 2007-09-03 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
2045 * gcc.c (xputenv): New function.
2046 (putenv_from_prefixes): Use xputenv instead of putenv.
2047 (process_command): Use xputenv instead of putenv.
2048 (set_collect_gcc_options): Use xputenv instead of putenv.
2049 (main): Use xputenv instead of putenv.
2051 2007-09-03 Jesper Nilsson <jesper.nilsson@axis.com>
2053 * config/cris/cris.h (TARGET_HAS_SWAP): Defined to describe
2054 availability of swap instruction.
2055 * config/cris/cris.md (bswapsi2): Implement using swap instruction.
2057 2007-09-03 Revital Eres <eres@il.ibm.com>
2059 * doc/invoke.texi (-fmodulo-sched-allow-regmoves): Update
2062 2007-09-03 Vladimir Yanovsky <yanov@il.ibm.com>
2063 Ayal Zaks <zaks@il.ibm.com>
2064 Revital Eres <eres@il.ibm.com>
2066 * modulo-sched.c (ps_insert_empty_row, verify_partial_schedule,
2067 compute_split_row): New functions.
2068 (ps_unschedule_node): Remove.
2069 (normalize_sched_times): Iterate over the already scheduled
2070 insns instead of the number of nodes.
2071 (MAX_SPLIT_NUM): New definition.
2072 (sms_schedule_by_order): Change the scheduling heuristic to
2073 avoid useless increases of initiation interval ii.
2074 (get_sched_window): Add dump printouts.
2076 2007-09-02 David Daney <ddaney@avtrex.com>
2078 * config/mips/mips.md (UNSPEC_COMPARE_AND_SWAP, UNSPEC_SYNC_OLD_OP,
2079 UNSPEC_SYNC_NEW_OP, UNSPEC_SYNC_EXCHANGE): New define_constants.
2080 (optab, insn): Add more attributes.
2081 (fetchop_bit): New code macro.
2082 (immediate_insn): New code macro attribute.
2083 (sync): Change condition to ISA_HAS_SYNC.
2084 (rdhwr): Change predicate for operand 0 to register_operand.
2085 (memory_barrier): New expand.
2086 (sync_compare_and_swap<mode>, sync_add<mode>, sync_sub<mode>,
2087 sync_old_add<mode>, sync_old_sub<mode>, sync_new_add<mode>,
2088 sync_new_sub<mode>, sync_<optab><mode>, sync_old_<optab><mode>,
2089 sync_new_<optab><mode>, sync_nand<mode>, sync_old_nand<mode>,
2090 sync_new_nand<mode>, sync_lock_test_and_set<mode>): New insns.
2091 * config/mips/mips.h (ISA_HAS_SYNC, ISA_HAS_LL_SC): New ISA predicates.
2092 (MIPS_COMPARE_AND_SWAP, MIPS_SYNC_OP, MIPS_SYNC_OLD_OP,
2093 MIPS_SYNC_NEW_OP, MIPS_SYNC_NAND, MIPS_SYNC_OLD_NAND,
2094 MIPS_SYNC_NEW_NAND, MIPS_SYNC_EXCHANGE): New Macros.
2096 2007-09-03 Jesper Nilsson <jesper.nilsson@axis.com>
2097 Hans-Peter Nilsson <hp@axis.com>
2099 * config/cris/cris.h (TARGET_HAS_LZ, CLZ_DEFINED_VALUE_AT_ZERO):
2100 Defined to describe availability and behavior of CLZ.
2101 * config/cris/cris.md (clzsi2): Implement using lz instruction.
2102 * config/cris/cris.opt: Tweak comment for "-metrax4".
2103 * config/cris/arit.c (LZ): When defined, define as __builtin_clz.
2104 * longlong.h [__CRIS__ && __CRIS_arch_version >= 3]
2105 (count_leading_zeros): Define.
2107 2007-09-01 Laurynas Biveinis <laurynas.biveinis@gmail.com>
2109 * ggc-page.c (ggc_pch_read): Call validate_free_objects.
2111 2007-09-02 Joseph Myers <joseph@codesourcery.com>
2114 * c-decl.c (grokdeclarator): Apply qualifiers to type of parameter
2117 2007-09-01 Kazu Hirata <kazu@codesourcery.com>
2119 * config/arm/arm.c, config/rs6000/ppu_intrinsics.h,
2120 config/spu/spu.c, df-scan.c, fixed-value.c, fold-const.c,
2121 ginclude/tgmath.h, haifa-sched.c, optabs.c, recog.c,
2122 sched-deps.c, sched-int.h, system.h, target.h,
2123 tree-ssa-live.c, tree-vect-transform.c, tree-vectorizer.c,
2124 tree.def: Fix comment typos.
2126 2007-09-01 Kazu Hirata <kazu@codesourcery.com>
2128 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove.
2129 (override_options): Turn on -mstrict-align on non-ColdFire
2131 * config/m68k/m68k.h (BIGGEST_ALIGNMENT): Update a comment.
2133 2007-09-01 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2135 * c-common.c (const_strip_array_types): Delete.
2136 * c-common.h (const_strip_array_types): Likewise.
2137 * emit-rtl.c (const_next_insn, const_previous_insn,
2138 const_next_nonnote_insn, const_prev_nonnote_insn,
2139 const_next_real_insn, const_prev_real_insn,
2140 const_next_active_insn, const_prev_active_insn, const_next_label,
2141 const_prev_label): Delete.
2142 * gimple-low.c (block_may_fallthru): Use expr_last.
2143 * jump.c (reversed_comparison_code_parts): Use prev_nonnote_insn.
2144 * rtl.h (const_previous_insn, const_next_insn,
2145 const_prev_nonnote_insn, const_next_nonnote_insn,
2146 const_prev_real_insn, const_next_real_insn,
2147 const_prev_active_insn, const_next_active_insn, const_prev_label,
2148 const_next_label): Delete.
2149 * rtlanal.c (keep_with_call_p): Use next_nonnote_insn.
2150 * system.h (CONST_CAST): Update comment.
2151 * tree-cfg.c (tree_can_merge_blocks_p): Use last_stmt.
2152 (const_first_stmt, const_last_stmt): Delete.
2153 (tree_block_ends_with_condjump_p): Use last_stmt.
2154 * tree-flow.h (const_first_stmt, const_last_stmt): Delete.
2155 * tree-iterator.c (const_expr_first, const_expr_last,
2156 const_expr_only): Delete.
2157 * tree.c (const_lookup_attribute): Likewise.
2158 (attribute_list_contained): Use lookup_attribute.
2159 * tree.h (const_lookup_attribute, const_expr_first,
2160 const_expr_last, const_expr_only): Delete.
2162 2007-09-01 Richard Guenther <rguenther@suse.de>
2164 * tree-ssa-alias.c (create_name_tags): Use TYPE_VOLATILE
2165 instead of TREE_THIS_VOLATILE.
2166 * tree-ssa-operands.c (add_virtual_operand): Do not set
2167 has_volatile_ops. Do not check s_ann for NULL.
2168 (get_indirect_ref_operands): Likewise.
2169 (get_expr_operands): Likewise. For all reference trees
2170 set has_volatile_ops based on TREE_THIS_VOLATILE. Also
2171 check base of reference for TREE_THIS_VOLATILE.
2173 2007-09-01 Richard Guenther <rguenther@suse.de>
2175 * tree-cfg.c (verify_gimple_expr): Fix check for conversions
2176 between integral types. Also allow conversions between
2179 2007-08-31 Andrew Pinski <andrew_pinski@playstation.sony.com>
2181 * optabs.c (prepare_float_lib_cmp): Use
2182 libgcc_cmp_return_mode instead of word_mode.
2183 * config/dfp-bit.h: Likewise.
2184 * config/fp-bit.h: Likewise.
2185 * config/spu/spu.c (spu_libgcc_cmp_return_mode):
2187 (spu_libgcc_shift_count_mode): New function.
2188 (TARGET_LIBGCC_CMP_RETURN_MODE): Define.
2189 (TARGET_LIBGCC_SHIFT_COUNT_MODE): Define.
2191 2007-08-31 Joseph Myers <joseph@codesourcery.com>
2193 * tree-pretty-print.c: Include output.h.
2194 (dump_decl_name, dump_generic_node): Use %wd with pp_printf, not
2195 HOST_WIDE_INT_PRINT_DEC.
2196 * Makefile.in (tree-pretty-print.o): Add dependency on output.h.
2198 2007-08-31 Zdenek Dvorak <ook@ucw.cz>
2200 PR rtl-optimization/33224
2201 * loop-iv.c (latch_dominating_def): Check that the definition belongs
2202 to the analysed region.
2204 2007-08-31 Richard Guenther <rguenther@suse.de>
2206 * function.c (allocate_struct_function): Do not set
2207 current_function_returns_pointer.
2208 * function.h (struct var_refs_queue): Remove.
2209 (struct simple_obstack_stack): Likewise.
2210 (struct function): Remove fixup_var_refs_queue member.
2211 Remove returns_pointer flag.
2212 (current_function_returns_pointer): Remove define.
2214 2007-08-31 Nick Clifton <nickc@redhat.com>
2216 * tree-ssa-coalesce.c (compare_pairs): Use the elements as
2217 secondary keys in order to obtain a stable sort.
2219 2007-08-31 Nick Clifton <nickc@redhat.com>
2222 * config/m32r/constraints.md: Add W constraint for integer zero.
2223 * config/m32r/m32r.md (get_pc): Use W and i constraints.
2224 Fix length calculation by using alternatives.
2226 2007-08-31 Richard Sandiford <richard@codesourcery.com>
2228 * optabs.c (shift_optab_p, commutative_optab_p): New functions,
2229 split out from expand_binop.
2230 (avoid_expensive_constant): New function.
2231 (expand_binop_directly): Remove commutative_op argument and
2232 (expand_binop_directly): Remove commutative_op argument and
2233 call cummutative_optab_p instead. Do not change op0 or op1
2234 when swapping xop0 and xop1. Apply avoid_expensive_constant
2235 to each argument after potential swapping. Enforce the
2236 canonical order of commutative operands.
2237 (expand_binop): Use shift_optab_p and commutative_optab_p.
2238 Update the calls to expand_binop_directly. Only force constants
2239 into registers when widening an operation. Only swap operands
2240 once a direct expansion has been rejected.
2241 (expand_twoval_binop): Only force constants into registers when
2242 using a direct expansion.
2244 2007-08-31 Maxim Kuvyrkov <maxim@codesourcery.com>
2246 * sched-deps.c (update_dep): Mark arguments with ATTRIBUTE_UNUSED.
2247 (sched_analyze_insn): Don't postprocess speculative dependencies when
2248 target has no scheduling.
2249 * tree-vect-transform.c (vect_estimate_min_profitable_iters): Fix
2250 'uninitialized' warning.
2252 2007-08-31 Richard Guenther <rguenther@suse.de>
2254 * tree.c (build_complex_type): Always set TYPE_NAME for
2257 2007-08-31 Olivier Hainque <hainque@adacore.com>
2259 * config/rs6000/xcoff.h (XCOFF_CSECT_DEFAULT_ALIGNMENT_STR): New
2260 internal macro. Default alignment factor for csect directives, chosen
2261 to match what BIGGEST_ALIGNMENT allows.
2262 (DATA_SECTION_ASM_OP): Use it.
2263 * config/rs6000/rs6000.c
2264 (rs6000_xcoff_output_readonly_section_asm_op):
2265 Use XCOFF_CSECT_DEFAULT_ALIGNMENT_STR.
2266 (rs6000_xcoff_output_readwrite_section_asm_op): Likewise.
2268 2007-08-31 Olivier Hainque <hainque@adacore.com>
2270 * dbxout.c (dbxout_block): Move declaration within the first
2271 #if defined (DBX_DEBUGGING_INFO) block. Rearrange another such
2272 block to include the definition, together with the associated
2275 2007-08-31 David Edelsohn <edelsohn@gnu.org>
2276 Revital Eres <eres@il.ibm.com>
2278 * doc/invoke.texi (-mpaired): Document flag.
2279 * config.gcc: Include paired.h in powerpc extra_headers and
2280 750cl.h in powerpc-*-linux*paired*.
2281 * config/rs6000/rs6000.opt (-mpaired): New flag.
2282 * config/rs6000/rs6000.c (paired_init_builtins,
2283 paired_expand_builtin, paired_expand_lv_builtin,
2284 paired_expand_stv_builtin, paired_expand_predicate_builtin):
2285 New functions to support the paired single builtin functions.
2286 (rs6000_hard_regno_mode_ok): Handle PAIRED_VECTOR_MODE.
2287 (def_builtin, bdesc_3arg, bdesc_2arg, bdesc_1arg): Add paired
2289 (bdesc_paired_preds): New structure for paired predicate
2291 (rs6000_expand_builtin): Expand paired single builtins.
2292 (rs6000_init_builtins): Init paired single builtins.
2293 (rs6000_common_init_builtins): Add v2sf_ftype_v2sf_v2sf_v2sf.
2294 Rename v2sf_ftype_v2sf_v2sf to v2sf_ftype_v2sf_v2sf_spe
2295 and v2sf_ftype_v2sf_spe to v2sf_ftype_v2sf. Add new types
2296 v2sf_ftype_v2sf_v2sf and v2sf_ftype_v2sf.
2297 (rs6000_vector_mode_supported_p): Support paired vector mode.
2298 * config/rs6000/rs6000.h: (UNITS_PER_PAIRED_WORD,
2299 PAIRED_VECTOR_MODE, PAIRED_SIMD_REGNO_P, TARGET_PAIRED_FLOAT):
2301 (LOCAL_ALIGNMENT): Handle PAIRED_VECTOR_MODE.
2302 (DATA_ALIGNMENT): Likewise.
2303 (UNITS_PER_SIMD_WORD): Handle PAIRED_VECTOR_MODE.
2304 (rs6000_builtins): Add PAIRED builtins.
2305 * config/rs6000/rs6000.md: Include paired.md.
2306 * config/rs6000/paired.h: New.
2307 * config/rs6000/paired.md: New.
2308 * config/rs6000/750cl.h: New.
2309 * config/rs6000/spe.md: Recognize movv2sf instruction for 750cl.
2311 2007-08-30 Ollie Wild <aaw@google.com>
2313 * c-opts.c (c_common_handle_option): Support -fno-directives-only.
2314 * gcc.c (default_compilers): Add -fno-directives-only to
2315 @assembler-with-cpp.
2317 2007-08-30 Sandra Loosemore <sandra@codesourcery.com>
2321 * regclass.c (initial_fixed_regs): Revert previous change and make
2323 (initial_call_used_regs): Likewise.
2324 (initial_call_really_used_regs): Delete, reverting previous addition.
2325 (initial_reg_names): Likewise.
2326 (init_reg_sets): Revert previous change.
2327 (saved_fixed_regs): New.
2328 (saved_call_used_regs): New.
2329 (saved_call_really_used_regs): New.
2330 (saved_reg_names): New.
2331 (save_register_info): New.
2332 (restore_register_info): New.
2333 (init_reg_sets_1): Replace reset of register info with call to
2334 restore_register_info.
2335 * rtl.h (save_register_info): Declare.
2336 * toplev.c (backend_init): Call save_register_info.
2338 2007-08-30 Laurynas Biveinis <laurynas.biveinis@gmail.com>
2340 * tree-ssa-propagate.c (set_rhs): Remove the copied annotation
2341 from the original statement.
2343 2007-08-30 Chao-ying Fu <fu@mips.com>
2345 * c-lex.c (interpret_fixed): Declare.
2346 (interpret_float): Process _Fract and _Accum.
2347 (interpret_fixed): New function.
2348 * final.c (output_addr_const): Process CONST_FIXED.
2349 * simplify-rtx.c (simplify_const_unary_operation): Handle US_NEG.
2350 (simplify_binary_operation_1): Handle US_ASHIFT, SS_MULT, US_MULT,
2352 (simplify_const_binary_operation): Handle SS_MULT, US_MULT, SS_DIV,
2354 (simplify_immed_subreg): Support CONST_FIXED.
2355 Process MODE_FRACT, MODE_UFRACT, MODE_ACCUM, MODE_UACCUM.
2356 (simplify_subreg): Support CONST_FIXED.
2358 2007-08-30 Andrew Pinski <andrew_pinski@playstation.sony.com>
2360 * config/rs6000/ppu_intrinsics.h: New file.
2361 * config.gcc (powerpc*-*-* <extra_headers>): Install
2364 2007-08-30 Kenneth Zadeck <zadeck@naturalbridge.com>
2366 * cfg.c (dump_flow_info): Change to also print entry and exit
2368 * print-rtl.c (print_rtl_single): Allow to print rtl with
2371 2007-08-30 Kenneth Zadeck <zadeck@naturalbridge.com>
2373 * df-core.h (df_dump_region): New function.
2374 * df.h (df_dump_region): New function.
2375 * loop-invariant.c (find_defs): Add call to df_dump_region.
2376 * loop-iv.c (iv_analysis_loop_init): Changed call from df_dump to
2379 2007-08-30 Jakub Jelinek <jakub@redhat.com>
2382 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Return
2383 true if any of the compare_section_name calls returned true,
2384 rather than if any returned false.
2386 2007-08-30 Richard Guenther <rguenther@suse.de>
2388 PR tree-optimization/33199
2389 * tree-ssa-structalias.c (handle_lhs_call): New function.
2390 (find_func_aliases): In non-IPA mode make sure that for
2391 calls that return a pointer we add a constraint for the
2392 result to point to anything.
2394 2007-08-30 Richard Guenther <rguenther@suse.de>
2396 * doc/invoke.texi (-mveclibabi): Document new target option.
2397 * config/i386/i386.opt (-mveclibabi): New target option.
2398 * config/i386/i386.c (ix86_veclib_handler): Handler for
2399 vectorization library support.
2400 (override_options): Handle the -mveclibabi option, initialize
2401 the vectorization library handler.
2402 (ix86_builtin_vectorized_function): As fallback call the
2403 vectorization library handler, if set.
2404 (ix86_veclibabi_acml): New static function for ACML ABI style
2405 vectorization support.
2407 2007-08-30 Jakub Jelinek <jakub@redhat.com>
2409 * config/rs6000/rs6000.c (rs6000_emit_sync): For QI or HI mode
2410 used_m, even if it is 32-bit aligned, adjust used_m MEM to have
2411 SImode and update m. Don't run gen_lowpart_common on arbitrary
2412 memory address, force it to register first.
2415 * dce.c (dce_process_block): Don't delete setters of
2416 artificially used registers.
2418 2007-08-30 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2420 * config/s390/s390.md ("*add<mode>3_alc_carry1_cc",
2421 "*add<mode>3_alc_carry1_cconly", "*add<mode>3_alc_carry2_cc",
2422 "*add<mode>3_alc_carry2_cconly"): New insn definitions.
2424 2007-08-30 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2426 * config/s390/s390.c (s390_expand_addcc): Emit canonical alc
2428 * config/s390/s390.md ("*add<mode>3_alc_cc", "*add<mode>3_alc",
2429 "addti3", "*adddi3_31z", "*scond<mode>"): Make alc pattern canonical.
2431 2007-08-29 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2433 * system.h: Activate -Wcast-qual as warning-only.
2435 2007-08-29 Paolo Bonzini <bonzini@gnu.org>
2438 * config/rs6000/rs6000.c (compare_section_name): New function.
2439 (rs6000_elf_in_small_data_p): Compare section prefixes instead
2442 2007-08-29 Olivier Hainque <hainque@adacore.com>
2444 * xcoffout.c: #include debug.h.
2445 * Makefile.in (xcoffout.o): Add debug.h dependency.
2447 2007-08-29 Uros Bizjak <ubizjak@gmail.com>
2449 * simplify-rtx.c (simplify_binary_operation_1) [VEC_SELECT]:
2450 Change CONST_VECTOR_ELT to XVECEXP in elem calculation.
2452 2007-08-29 Richard Guenther <rguenther@suse.de>
2454 * tree-pretty-print.c (dump_generic_node): Print
2455 label DECL_UID as D.%u.
2456 * tree-dfa.c (dump_variable): Print DECL_UID as D.%u.
2458 2007-08-29 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
2460 * gcc/builtin-types.def (BT_FN_PTR_PTR_SIZE): New type.
2461 * gcc/builtins.def (BUILT_IN_REALLOC): New builtin.
2463 2007-08-29 Douglas Gregor <doug.gregor@gmail.com>
2466 * tree.c (build_type_attribute_qual_variant): Set canonical types
2467 on the final, unqualified attribute variant before building the
2470 2007-08-29 Jie Zhang <jie.zhang@analog.com>
2472 * config/bfin/bfin.c (bfin_expand_builtin): Fix the argument
2473 order of __builtin_bfin_cmplx_mac and __builtin_bfin_cmplx_msu.
2475 2007-08-29 Jie Zhang <jie.zhang@analog.com>
2478 2007-08-29 Jie Zhang <jie.zhang@analog.com>
2479 * config/bfin/bfin.md (composev2hi): Put operands into vector
2482 2007-08-29 Jie Zhang <jie.zhang@analog.com>
2484 * config/bfin/bfin.md (composev2hi): Put operands into vector
2487 2007-08-29 Jie Zhang <jie.zhang@analog.com>
2489 * config/bfin/bfin.c (bfin_expand_call): Inline PLT with l1_text
2490 attribute when appropriate.
2491 (bfin_handle_l1_text_attribute): New.
2492 (bfin_handle_l1_data_attribute): New.
2493 (bfin_attribute_table): Add attributes: l1_text, l1_data,
2494 l1_data_A and l1_data_B.
2495 * doc/extend.texi (node Function Attributes): Document l1_text
2497 (Variable Attributes): Add Blackfin subsection. Document l1_data,
2498 l1_data_A and l1_data_B variable attributes.
2500 2007-08-28 Jie Zhang <jie.zhang@analog.com>
2502 * config/bfin/bfin.opt (minline-plt): Add.
2503 * config/bfin/bfin.c (bfin_expand_call): Inline PLT when emit
2504 call to global functions.
2505 * doc/invoke.texi (Option Summary): Mention -minline-plt.
2506 (Blackfin Options): Document -minline-plt.
2508 2007-08-28 Nathan Sidwell <nathan@codesourcery.com>
2510 * config/m68k/m68k.c (m68k_get_function_kind): Assert we're never
2511 given a non-function.
2512 (m68k_ok_for_sibcall_p): Only sibcall functions of the same kind.
2514 2007-08-28 DJ Delorie <dj@redhat.com>
2516 * config/sh/sh.c (sh_gimplify_va_arg_expr): Fix sh2a support.
2518 2007-08-28 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2520 * system.h (CONST_CAST): Avoid union for gcc-4.0.x.
2522 2007-08-28 Richard Guenther <rguenther@suse.de>
2524 * tree.h (struct tree_function_decl): Increase size of
2525 function_code bitfield.
2527 2007-08-28 Nathan Sidwell <nathan@codesourcery.com>
2528 Kazu Hirata <kazu@codesourcery.com>
2530 * gcc/config/m68k/linux.h
2531 (M68K_HONOR_TARGET_STRICT_ALIGNMENT): Redefine as 0.
2532 * config/m68k/m68k.c (TARGET_RETURN_IN_MEMORY): New.
2533 (m68k_return_in_memory): New.
2534 * gcc/config/m68k/m68k.h (M68K_HONOR_TARGET_STRICT_ALIGNMENT):
2537 2007-08-28 Uros Bizjak <ubizjak@gmail.com>
2540 * simplify-rtx.c (simplify_binary_operation_1) [VEC_SELECT]:
2541 Simplify nested VEC_SELECT (with optional VEC_CONCAT operator as
2542 operand) when top VEC_SELECT extracts scalar element.
2543 * config/i386/sse.md (*vec_extract_v4si_mem): New pattern.
2544 (*vec_extract_v4sf_mem): Ditto.
2546 2007-08-28 Jakub Jelinek <jakub@redhat.com>
2549 * passes.c (init_optimization_passes): Move pass_df_finish
2550 after pass_postreload sublist.
2552 PR rtl-optimization/33148
2553 * simplify-rtx.c (simplify_unary_operation_1): Only optimize
2554 (neg (lt X 0)) if X has scalar int mode.
2557 * dwarf2out.c (rtl_for_decl_init): If vector decl has CONSTRUCTOR
2558 initializer, use build_vector_from_ctor if possible to create
2559 VECTOR_CST out of it. If vector initializer is not VECTOR_CST
2560 even after this, return NULL.
2562 2007-08-28 Gabriel Dos Reis <gdr@integrable-solutions.net>
2564 * c-pretty-print.c (pp_c_constant): Handle COMPLEX_CST.
2566 2007-08-28 Richard Guenther <rguenther@suse.de>
2568 * tree.h (DECL_DISREGARD_INLINE_LIMITS): New.
2569 (struct tree_function_decl): Make function_code a bitfield.
2570 Add disregard_inline_limits flag.
2571 * cgraphunit.c (cgraph_process_new_functions): Check
2572 DECL_DISREGARD_INLINE_LIMITS instead of disregard_inline_limits_p.
2573 (cgraph_preserve_function_body_p): Likewise.
2574 * ipa-inline.c (compute_inline_parameters): Likewise.
2575 * c-decl.c (finish_function): Set DECL_DISREGARD_INLINE_LIMITS
2576 for GNU C extern inline functions.
2577 (merge_decls): Merge DECL_DISREGARD_INLINE_LIMITS.
2578 * tree-inline.c (disregard_inline_limits_p): Remove.
2579 * tree-inline.h (disregard_inline_limits_p): Likewise.
2580 * c-common.c (handle_always_inline_attribute): Set
2581 DECL_DISREGARD_INLINE_LIMITS.
2582 * langhooks.c (add_builtin_function): Verify the function code
2583 fits in the bitfield.
2585 2007-08-28 Mircea Namolaru <namolaru@il.ibm.com>
2586 Vladimir Yanovsky <yanov@il.ibm.com>
2587 Revital Eres <eres@il.ibm.com>
2588 Andrey Belevantsev <abel@ispras.ru>
2590 * config/spu/spu.md: Recognize doloop pattern when -fmodulo-sched
2592 * modulo-sched.c: Add documentation regarding do-loop.
2593 (doloop_register_get): Change number of arguments to support
2594 the new do-loop pattern and check whether COUNT_REG has no other
2595 occurences in the loop besides in the control part.
2596 (duplicate_insns_of_cycles): Do not duplicate the insn which
2597 changes count_reg as it is already adjusted.
2598 (generate_prolog_epilog): New argument to support the new
2599 do-loop pattern. Change the subtract instruction to use
2600 expand_simple_binop. Call duplicate_insns_of_cycles with new
2602 (sms_schedule): Call doloop_register_get and
2603 generate_prolog_epilog with new argument. Do not handle loops
2604 with single sets insns with subreg in their lhs.
2605 * loop-doloop.c (doloop_optimize): Support for another do-loop
2607 (doloop_condition_get): Gets an instruction instead of a pattern
2608 and change the return condition when the do-loop pattern is
2610 * ddg.c (create_ddg_dep_from_intra_loop_link): Handle only reg
2611 deps when considering to not create edges.
2613 2007-08-27 Alexandre Oliva <aoliva@redhat.com>
2615 * doc/extend.texi (gnu_inline funtion attribute): Document C++
2618 2007-08-27 Jason Merrill <jason@redhat.com>
2621 * gimplify.c (gimplify_modify_expr): Discard the assignment of
2622 zero-sized types after calling gimplify_modify_expr_rhs.
2624 2007-08-27 Sandra Loosemore <sandra@codesourcery.com>
2626 * regclass.c (init_reg_autoinc): Fix typo.
2628 2007-08-27 Daniel Berlin <dberlin@dberlin.org>
2630 Fix PR tree-optimization/33173
2631 * tree-ssa-alias.c (find_used_portions): Fix reversed test.
2633 2007-08-27 H.J. Lu <hongjiu.lu@intel.com>
2636 * config/i386/i386.h (VALID_DFP_MODE_P): New.
2637 * config/i386/i386.c (ix86_hard_regno_mode_ok): Allow DFP in
2640 2007-08-27 Sandra Loosemore <sandra@codesourcery.com>
2641 David Ung <davidu@mips.com>
2642 Nigel Stephens <nigel@mips.com>
2644 Separate target-specific initialization from general
2645 back-end initialization.
2647 * toplev.c (init_alignments): New, split out from...
2648 (process_options): ...here.
2649 (backend_init_target): New, split out from...
2650 (backend_init): ...here.
2651 (lang_dependent_init_target): New, split out from...
2652 (lang_dependent_init): ...here.
2653 (target_reinit): New.
2654 * toplev.h (target_reinit): Declare.
2655 * expr.c (init_expr_target): Renamed from init_expr_once, since it
2656 now can be called more than once. Update comments.
2657 * expr.h (init_expr_target): Likewise.
2658 * alias.c (init_alias_target): Renamed from init_alias_once, since it
2659 now can be called more than once. Explicitly zero
2660 static_reg_base_value.
2661 * emit-rtl.c (init_emit_regs): New, split out from...
2662 (init_emit_once): Here.
2663 * regclass.c (initial_fixed_regs, initial_call_used_regs): Make
2664 non-const, so that changes from command-line arguments can overwrite
2665 values provided by the static initializers.
2666 (initial_call_really_used_regs): New, used similarly to the above.
2667 (initial_reg_names): Likewise.
2668 (last_mode_for_init_move_cost): Promoted function-local static to
2669 file-scope static to make it accessible outside init_move_cost.
2670 (init_reg_sets): Do not initialize fixed_regs and call_used_regs yet.
2671 Do not initialize inv_reg_alloc_order yet, either. Do initialize
2672 reg_names since it is needed for parsing command-line options.
2673 (init_move_cost): Use last_mode_for_init_move_cost instead of
2674 function-local static.
2675 (init_reg_sets_1): Initialize fixed_regs, call_used_regs, and
2676 call_really_used_regs now. Reinitialize reg_names. Also
2677 initialize inv_reg_alloc_order. Zero reg_class_subunion and
2678 reg_class_superunion. Clear losing_caller_save_reg_set.
2679 Preserve global registers if called more than once. Reset
2680 move cost, may_move_in_cost, may_move_out_cost, and
2681 last_mode_for_init_move_cost.
2682 (init_reg_modes_target): Renamed from init_reg_modes_once, since it
2683 can now be invoked more than once. Update comments.
2684 (init_regs): Update comments.
2685 (fix_register): Update initial_fixed_regs, initial_call_used_regs,
2686 and initial_call_really_used_regs, instead of the non-initial
2687 variables. This allows us to save the command-line register settings
2688 after target reinitialization.
2689 (init_reg_autoinc): Zero forbidden_inc_dec_classes.
2690 * rtl.h (init_emit_regs): Declare.
2691 (init_reg_modes_target, init_alias_target): Renamed as described
2693 * reload1.c (init_reload): Update comments.
2694 * optabs.c (init_optabs): Likewise.
2695 * cfgloopanal.c (init_set_costs): Explicitly zero target_avail_regs.
2697 2007-08-26 Chao-ying Fu <fu@mips.com>
2699 * rtl.h (XCNMPFV): Preserve const-ness of parameters through use of
2702 2007-08-26 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2704 * alias.c (memory_modified_1, memory_modified_in_insn_p):
2706 * basic-block.h (tree_predicted_by_p, rtl_predicted_by_p):
2708 * bb-reorder.c (copy_bb_p): Likewise.
2709 * cfghooks.c (can_remove_branch_p, can_merge_blocks_p,
2710 predicted_by_p, can_duplicate_block_p, block_ends_with_call_p,
2711 block_ends_with_condjump_p): Likewise.
2712 * cfghooks.h (can_remove_branch_p, can_merge_blocks_p,
2713 predicted_by_p, can_duplicate_block_p, block_ends_with_call_p,
2714 block_ends_with_condjump_p): Likewise.
2715 * cfglayout.c (cfg_layout_can_duplicate_bb_p): Likewise.
2716 * cfgrtl.c (can_delete_note_p, can_delete_label_p,
2717 rtl_can_merge_blocks, cfg_layout_can_merge_blocks_p,
2718 rtl_block_ends_with_call_p, rtl_block_ends_with_condjump_p,
2719 need_fake_edge_p, rtl_can_remove_branch_p): Likewise.
2720 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Use
2721 debug_true_const_tree instad of debug_true_tree.
2722 * debug.c (do_nothing_debug_hooks): Likewise.
2723 * debug.h (ignore_block, debug_true_const_tree): Constify.
2724 * dwarf2out.c (stack_adjust_offset, clobbers_queued_reg_save,
2725 dwarf2out_ignore_block, is_pseudo_reg, is_tagged_type,
2726 decl_ultimate_origin, block_ultimate_origin, lookup_decl_loc,
2727 same_dw_val_p, add_ranges, is_subrange_type, type_is_enum,
2728 dbx_reg_number, is_based_loc, field_type,
2729 simple_type_align_in_bits, simple_decl_align_in_bits,
2730 simple_type_size_in_bits, field_byte_offset, insert_float,
2731 type_tag, member_declared_type, is_redundant_typedef,
2732 secname_for_decl, is_inlined_entry_point): Likewise.
2733 * emit-rtl.c (const_next_insn, const_previous_insn,
2734 const_next_nonnote_insn, const_prev_nonnote_insn,
2735 const_next_real_insn, const_prev_real_insn,
2736 const_next_active_insn, const_prev_active_insn, const_next_label,
2737 const_prev_label): Likewise.
2738 * except.h (lookup_stmt_eh_region_fn, lookup_stmt_eh_region):
2740 * haifa-sched.c (may_trap_exp, haifa_classify_insn,
2741 find_set_reg_weight, no_real_insns_p, find_set_reg_weight,
2742 sched_insn_is_legitimate_for_speculation_p): Likewise.
2743 * jump.c (reversed_comparison_code_parts,
2744 reversed_comparison_code, reversed_comparison, condjump_label):
2746 * predict.c (rtl_predicted_by_p, tree_predicted_by_p): Likewise.
2747 * reg-stack.c (stack_regs_mentioned_p, stack_regs_mentioned):
2749 * rtl.h (const_previous_insn, const_next_insn,
2750 const_prev_nonnote_insn, const_next_nonnote_insn,
2751 const_prev_real_insn, const_next_real_insn,
2752 const_prev_active_insn, const_next_active_insn, const_prev_label,
2753 const_next_label, modified_between_p, modified_in_p,
2754 const_note_storeskeep_with_call_p, condjump_label,
2755 reversed_comparison, reversed_comparison_code,
2756 reversed_comparison_code_parts, memory_modified_in_insn_p,
2757 stack_regs_mentioned): Likewise.
2758 * rtlanal.c (modified_between_p, modified_in_p, const_note_stores,
2759 keep_with_call_p): Likewise.
2760 * sched-deps.c (deps_may_trap_p, sched_get_condition,
2761 conditions_mutex_p, sched_insns_conditions_mutex_psd_next_list,
2762 sd_lists_size, sd_lists_empty_p): Likewise.
2763 * sched-int.h (sched_insns_conditions_mutex_p,
2764 haifa_classify_insn, no_real_insns_p,
2765 sched_insn_is_legitimate_for_speculation_p, sd_next_list,
2766 sd_lists_size, sd_lists_empty_p): Likewise.
2767 * sdbout.c (sdb_debug_hooks): Likewise.
2768 * tree-cfg.c (tree_can_merge_blocks_p, is_ctrl_altering_stmt,
2769 stmt_ends_bb_p, tree_can_remove_branch_p, tree_can_duplicate_bb_p,
2770 -tree_block_ends_with_call_p, tree_block_ends_with_condjump_p):
2772 * tree-eh.c (lookup_stmt_eh_region_fn, lookup_stmt_eh_region,
2773 tree_can_throw_internal): Likewise.
2774 * tree-flow-inline.h (phi_nodes): Likewise.
2775 * tree-flow.h (phi_nodesstmt_ends_bb_p, is_ctrl_altering_stmt,
2776 tree_can_throw_internal, lookup_stmt_eh_region): Likewise.
2777 * tree-gimple.c (const_get_call_expr_in): Likewise.
2778 * tree-gimple.h (const_get_call_expr_in): Likewise.
2779 * tree.c (const_lookup_attribute, attribute_list_equal,
2780 attribute_list_contained): Likewise.
2781 * tree.h (attribute_list_equal, attribute_list_contained,
2782 const_lookup_attribute): Likewise.
2783 * vmsdbgout.c (vmsdbgout_ignore_block): Likewise.
2785 2007-08-26 Bernhard Fischer <aldot@gcc.gnu.org>
2787 * tree-ssa-sccvn.c, tree-ssa-sccvn.h, tree-vn.c,
2788 tree-ssa-pre.c: Remove unnecessary trailing whitespace.
2790 2007-08-26 H.J. Lu <hongjiu.lu@intel.com>
2793 * config/alpha/constraints.md: Delete "svn:mime-type" property
2794 with svn propdel "svn:mime-type".
2796 2007-08-26 H.J. Lu <hongjiu.lu@intel.com>
2799 * ifcvt.c (noce_get_alt_condition): Make sure that the previous
2800 non NOTE insn doesn't cross basic block.
2801 (noce_try_abs): Likewise.
2802 (noce_process_if_block): Likewise.
2804 2007-08-26 David Edelsohn <edelsohn@gnu.org>
2807 * config/rs6000/predicates.md (offsettable_mem_operand): Memory
2808 operand without auto-inc-dec.
2809 * config/rs6000/rs6000.md (floatsidf2_internal): Use
2810 offsettable_mem_operand.
2811 (floatunssidf2_internal): Same.
2812 (fix_truncdfsi2_internal): Same.
2813 (floatsidf_ppc64): Same.
2814 (floatunssidf_ppc64): Same.
2815 (fix_trunctfsi2_internal): Same.
2817 2007-08-26 Bernhard Fischer <aldot@gcc.gnu.org>
2820 * Makefile.in (libbackend.o): Add gcov-iov.h dependency and
2821 pass defines needed for version.c.
2823 2007-08-25 Joseph Myers <joseph@codesourcery.com>
2825 * configure.ac (--debug-prefix-map check): Change assembler
2826 version presumed to have feature to 2.18.
2827 * configure: Regenerate.
2829 2007-08-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2831 * c-common.c (vector_types_convertible_p,
2832 decl_with_nonnull_addr_p, c_promoting_integer_type_p,
2833 self_promoting_args_p): Constify.
2834 * c-common.h (has_c_linkage, decl_with_nonnull_addr_p,
2835 c_promoting_integer_type_p, self_promoting_args_p,
2836 anon_aggr_type_p, vector_types_convertible_p): Likewise.
2837 * c-decl.c (anon_aggr_type_p): Likewise.
2838 * * c-dump.c (dump_stmt): Likewise.
2839 * c-objc-common.c (has_c_linkage): Likewise.
2840 * c-tree.h (same_translation_unit_p): Likewise.
2841 * c-typeck.c (null_pointer_constant_p,
2842 tagged_types_tu_compatible_p, function_types_compatible_p,
2843 type_lists_compatible_p, lvalue_or_else, lvalue_p,
2844 comptypes_internal, struct tagged_tu_seen_cache,
2845 same_translation_unit_p, alloc_tagged_tu_seen_cache,
2846 c_size_in_bytes): Likewise.
2847 * ggc-common.c (compare_ptr_data, hash_descriptor, eq_descriptor,
2848 hash_ptr, eq_ptr): Likewise.
2849 * langhooks-def.h (lhd_decl_ok_for_sibcall,
2850 LANG_HOOKS_OMP_PRIVATIZE_BY_REFERENCE): Likewise.
2851 * langhooks.c (lhd_decl_ok_for_sibcall): Likewise.
2852 * langhooks.h (ok_for_sibcall, omp_privatize_by_reference):
2854 * optabs.c (optab_for_tree_code): Likewise.
2855 * optabs.h (optab_for_tree_code): Likewise.
2856 * rtl.h (simplify_rtx): Likewise.
2857 * simplify-rtx.c (simplify_rtx): Likewise.
2858 * tree-dump.c (queue, queue_and_dump_index, queue_and_dump_type,
2859 dump_flag, dump_node): Likewise.
2860 * tree-dump.h (struct dump_info, dump_stmt, queue_and_dump_index,
2861 queue_and_dump_type, dump_flag): Likewise.
2862 * tree-flow.h (vect_can_force_dr_alignment_p): Likewise.
2863 * tree-pass.h (dump_node): Likewise.
2864 * tree-vectorizer.c (vect_can_force_dr_alignment_p,
2865 supportable_narrowing_operation): Likewise.
2866 * tree-vectorizer.h (vect_can_force_dr_alignment_p,
2867 supportable_narrowing_operation): Likewise.
2868 * tree-vrp.c (needs_overflow_infinity, supports_overflow_infinity,
2869 is_negative_overflow_infinity, is_positive_overflow_infinity,
2870 is_overflow_infinity, vrp_val_is_max, vrp_val_is_min,
2871 nonnull_arg_p, get_value_range, vrp_operand_equal_p,
2872 update_value_range, add_equivalence, ssa_name_nonnegative_p,
2873 ssa_name_nonzero_p, fp_predicate): Likewise.
2874 * tree.c (auto_var_in_fn_p, empty_body_p): Likewise.
2875 * tree.h (empty_body_p, auto_var_in_fn_p, ssa_name_nonzero_p,
2876 ssa_name_nonnegative_p): Likewise.
2878 2007-08-25 Hans Kester <hans.kester@ellips.nl>
2880 * config.gcc : Add x86_64-elf target.
2882 2007-08-25 Jakub Jelinek <jakub@redhat.com>
2884 * expr.c (store_expr): Fix order of store_by_pieces arguments.
2886 2007-08-24 Sandra Loosemore <sandra@codesourcery.com>
2887 Nigel Stephens <nigel@mips.com>
2891 * doc/tm.texi (SET_RATIO, SET_BY_PIECES_P): Document new macros.
2892 (STORE_BY_PIECES_P): No longer applies to __builtin_memset.
2893 * expr.c (SET_BY_PIECES_P): Define.
2894 (can_store_by_pieces, store_by_pieces): Add MEMSETP argument; use
2895 it to decide whether to use SET_BY_PIECES_P or STORE_BY_PIECES_P.
2896 (store_expr): Pass MEMSETP argument to can_store_by_pieces and
2898 * expr.h (SET_RATIO): Define.
2899 (can_store_by_pieces, store_by_pieces): Update prototypes.
2900 * builtins.c (expand_builtin_memcpy): Pass MEMSETP argument to
2901 can_store_by_pieces/store_by_pieces.
2902 (expand_builtin_memcpy_args): Likewise.
2903 (expand_builtin_strncpy): Likewise.
2904 (expand_builtin_memset_args): Likewise. Also remove special case
2905 for optimize_size so that can_store_by_pieces/SET_BY_PIECES_P can
2906 decide what to do instead.
2907 * value-prof.c (tree_stringops_transform): Pass MEMSETP argument
2908 to can_store_by_pieces.
2910 * config/sh/sh.h (SET_BY_PIECES_P): Clone from STORE_BY_PIECES_P.
2911 * config/s390/s390.h (SET_BY_PIECES_P): Likewise.
2913 * config/mips/mips.opt (mmemcpy): Change from Var to Mask.
2914 * config/mips/mips.c (override_options): Make -Os default to -mmemcpy.
2915 * config/mips/mips.h (MIPS_CALL_RATIO): Define.
2916 (MOVE_RATIO, CLEAR_RATIO, SET_RATIO): Define.
2917 (STORE_BY_PIECES_P): Define.
2919 2007-08-24 Tom Tromey <tromey@redhat.com>
2921 * varpool.c (varpool_last_needed_node): Fix comment typo.
2922 * c-decl.c (duplicate_decls): Fix comment typo.
2923 (clone_underlying_type): Update comment.
2925 2007-08-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2927 * config/alpha/alpha.c (alpha_mangle_type, decl_has_samegp,
2928 alpha_in_small_data_p, alpha_split_complex_arg,
2929 alpha_stdarg_optimize_hook, TARGET_ASM_CAN_OUTPUT_MI_THUNK):
2931 * config/arm/arm-protos.h (arm_mangle_type): Likewise.
2932 * config/arm/arm.c (arm_comp_type_attributes, arm_mangle_type):
2934 * config/bfin/bfin.c (funkind, bfin_comp_type_attributes,
2935 TARGET_ASM_CAN_OUTPUT_MI_THUNK): Likewise.
2936 * config/darwin-protos.h (darwin_binds_local_p): Likewise.
2937 * config/darwin.c (darwin_binds_local_p): Likewise.
2938 * config/frv/frv.c (frv_string_begins_with, frv_in_small_data_p):
2940 * config/i386/i386-protos.h (i386_pe_binds_local_p,
2941 i386_pe_valid_dllimport_attribute_p): Likewise.
2942 * config/i386/i386.c (ix86_function_regparm,
2943 ix86_comp_type_attributes, ix86_ms_bitfield_layout_p,
2944 x86_can_output_mi_thunk, ix86_mangle_type,
2945 TARGET_USE_BLOCKS_FOR_CONSTANT_P): Likewise.
2946 * config/i386/winnt.c (i386_pe_valid_dllimport_attribute_p,
2947 i386_pe_binds_local_p): Likewise.
2948 * config/ia64/ia64.c
2949 (ia64_first_cycle_multipass_dfa_lookahead_guard_spec,
2950 ia64_needs_block_p, ia64_in_small_data_p, ia64_mangle_type,
2951 ia64_invalid_conversion, ia64_invalid_unary_op,
2952 ia64_invalid_binary_op, TARGET_ASM_CAN_OUTPUT_MI_THUNK):
2954 * config/m32c/m32c.c (m32c_comp_type_attributes): Likewise.
2955 * config/m32r/m32r.c (m32r_in_small_data_p): Likewise.
2956 * config/m68k/m68k.c (TARGET_ASM_CAN_OUTPUT_MI_THUNK): Likewise.
2957 * config/mips/mips.c (mips_use_blocks_for_constant_p,
2958 mips_in_small_data_p, mips_use_anchors_for_symbol_p,
2959 mips_comp_type_attributes, TARGET_ASM_CAN_OUTPUT_MI_THUNK,
2960 mips_near_type_p, mips_far_type_p, mips_global_symbol_p,
2961 mips_symbol_binds_local_p, mips_classify_symbol): Likewise.
2962 * config/pa/pa.c (pa_commutative_p): Likewise.
2963 * config/rs6000/rs6000-protos.h (rs6000_elf_in_small_data_p):
2965 * config/rs6000/rs6000.c (rs6000_invalid_within_doloop,
2966 rs6000_ms_bitfield_layout_p, rs6000_mangle_type,
2967 rs6000_use_blocks_for_constant_p,
2968 rs6000_vector_alignment_reachable, rs6000_is_opaque_type,
2969 invalid_arg_for_unprototyped_fn, TARGET_ASM_CAN_OUTPUT_MI_THUNK,
2970 TARGET_SPLIT_COMPLEX_ARG, rs6000_elf_in_small_data_p): Likewise.
2971 * config/s390/s390.c (s390_mangle_type,
2972 TARGET_ASM_CAN_OUTPUT_MI_THUNK, TARGET_INVALID_WITHIN_DOLOOP):
2974 * config/score/score.c (TARGET_ASM_CAN_OUTPUT_MI_THUNK,
2975 th_in_small_data_p): Likewise.
2976 * config/sh/sh.c (sh_ms_bitfield_layout_p,
2977 sh_dwarf_calling_convention, TARGET_ASM_CAN_OUTPUT_MI_THUNK):
2979 * config/sparc/sparc.c (sparc_can_output_mi_thunk,
2980 sparc_mangle_type): Likewise.
2981 * config/spu/spu.c (spu_vector_alignment_reachable): Likewise.
2982 * config/xtensa/xtensa.c (TARGET_SPLIT_COMPLEX_ARG): Likewise.
2983 * emit-rtl.c (const_fixed_htab_hash, const_fixed_htab_eq):
2985 * hooks.c (hook_bool_mode_const_rtx_false,
2986 hook_bool_mode_const_rtx_true,
2987 hook_bool_const_tree_hwi_hwi_const_tree_false,
2988 hook_bool_const_tree_hwi_hwi_const_tree_true,
2989 hook_int_const_tree_0, hook_int_const_tree_const_tree_1,
2990 hook_constcharptr_const_tree_null,
2991 hook_constcharptr_const_rtx_null,
2992 hook_constcharptr_const_tree_const_tree_null,
2993 hook_constcharptr_int_const_tree_null,
2994 hook_constcharptr_int_const_tree_const_tree_null): New.
2995 (hook_bool_mode_rtx_false, hook_bool_mode_rtx_true,
2996 hook_bool_tree_hwi_hwi_tree_false,
2997 hook_bool_tree_hwi_hwi_tree_true, hook_int_tree_0,
2998 hook_int_tree_tree_1, hook_constcharptr_tree_null,
2999 hook_constcharptr_rtx_null, hook_constcharptr_tree_tree_null,
3000 hook_constcharptr_int_tree_null,
3001 hook_constcharptr_int_tree_tree_null): Delete.
3002 (default_can_output_mi_thunk_no_vcall): Constify.
3003 * hooks.h (hook_bool_mode_rtx_false, hook_bool_mode_rtx_true,
3004 hook_bool_tree_hwi_hwi_tree_false,
3005 hook_bool_tree_hwi_hwi_tree_true, hook_int_tree_0,
3006 hook_int_tree_tree_1, hook_constcharptr_tree_null,
3007 hook_constcharptr_rtx_null, hook_constcharptr_tree_tree_null,
3008 hook_constcharptr_int_tree_null,
3009 hook_constcharptr_int_tree_tree_null): Delete.
3010 (hook_bool_mode_const_rtx_false, hook_bool_mode_const_rtx_true,
3011 hook_bool_const_tree_hwi_hwi_const_tree_false,
3012 hook_bool_const_tree_hwi_hwi_const_tree_true,
3013 hook_int_const_tree_0, hook_int_const_tree_const_tree_1,
3014 hook_constcharptr_const_tree_null,
3015 hook_constcharptr_const_rtx_null,
3016 hook_constcharptr_const_tree_const_tree_null,
3017 hook_constcharptr_int_const_tree_null,
3018 hook_constcharptr_int_const_tree_const_tree_null): New.
3019 (default_can_output_mi_thunk_no_vcall): Constify.
3020 * integrate.c (function_attribute_inlinable_p): Likewise.
3021 * integrate.h (function_attribute_inlinable_p): Likewise.
3022 * jump.c (rtx_renumbered_equal_p): Likewise.
3023 * output.h (decl_readonly_section, categorize_decl_for_section,
3024 default_use_anchors_for_symbol_p, default_binds_local_p,
3025 default_binds_local_p_1): Likewise.
3026 * rtl.h (rtx_renumbered_equal_p, decl_default_tls_model):
3028 * target-def.h (TARGET_ASM_CAN_OUTPUT_MI_THUNK,
3029 TARGET_VALID_DLLIMPORT_ATTRIBUTE_P, TARGET_VECTOR_OPAQUE_P,
3030 TARGET_COMMUTATIVE_P, TARGET_USE_BLOCKS_FOR_CONSTANT_P,
3031 TARGET_COMP_TYPE_ATTRIBUTES,
3032 TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P,
3033 TARGET_MS_BITFIELD_LAYOUT_P, TARGET_MANGLE_TYPE,
3034 TARGET_IN_SMALL_DATA_P, TARGET_INVALID_CONVERSION,
3035 TARGET_INVALID_UNARY_OP, TARGET_INVALID_BINARY_OP,
3036 TARGET_DWARF_CALLING_CONVENTION): Likewise.
3037 * target.h (can_output_mi_thunk, needs_block_p,
3038 first_cycle_multipass_dfa_lookahead_guard_spec,
3039 vector_alignment_reachable, comp_type_attributes,
3040 function_attribute_inlinable_p, ms_bitfield_layout_p, mangle_type,
3041 commutative_p, use_blocks_for_constant_p,
3042 use_anchors_for_symbol_p, in_small_data_p, binds_local_p,
3043 vector_opaque_p, dwarf_calling_convention, stdarg_optimize_hook,
3044 invalid_within_doloop, valid_dllimport_attribute_p,
3045 split_complex_arg, invalid_arg_for_unprototyped_fn,
3046 invalid_conversion, invalid_unary_op, invalid_binary_op):
3048 * targhooks.c (default_invalid_within_doloop,
3049 hook_invalid_arg_for_unprototyped_fn,
3050 default_builtin_vector_alignment_reachable): Likewise.
3051 (hook_bool_rtx_commutative_p): Delete.
3052 (hook_bool_const_rtx_commutative_p): New.
3053 * targhooks.h (default_invalid_within_doloop,
3054 default_builtin_vector_alignment_reachable,
3055 hook_invalid_arg_for_unprototyped_fn): Constify.
3056 (hook_bool_rtx_commutative_p): Delete.
3057 (hook_bool_const_rtx_commutative_p): New.
3058 * varasm.c (bss_initializer_p, decl_default_tls_model,
3059 categorize_decl_for_section, decl_readonly_section,
3060 default_use_anchors_for_symbol_p, default_binds_local_p,
3061 default_binds_local_p_1): Constify.
3063 2007-08-24 Jie Zhang <jie.zhang@analog.com>
3065 * config/bfin/bfin.h (TARGET_CPU_CPP_BUILTINS): Define
3066 __NO_BUILTIN if -fno-builtin.
3068 2007-08-24 Jie Zhang <jie.zhang@analog.com>
3070 * config/bfin/bfin.c (print_operand): Report error instead of
3071 ICE for wrong operand.
3073 2007-08-24 Michael Matz <matz@suse.de>
3075 * Makefile.in (GTFILES_H): Use $(patsubst) instead of $(subst).
3077 2007-08-24 Richard Guenther <rguenther@suse.de>
3080 * tree-ssa.c (useless_type_conversion_p): Split into a
3081 recursive and a non-recursive part.
3082 (useless_type_conversion_p_1): New function.
3083 * tree-ssa-ccp.c (fold_stmt_r): Make sure that the result
3084 from maybe_fold_offset_to_reference is trivially convertible
3085 to the desired type before doing the simplification.
3087 2007-08-24 Jakub Jelinek <jakub@redhat.com>
3089 * expr.c (store_expr): Optimize initialization of an array
3091 * expr.h (builtin_strncpy_read_str): New prototype.
3092 * builtins.c (builtin_strncpy_read_str): Remove prototype.
3095 2007-08-24 Uros Bizjak <ubizjak@gmail.com>
3098 * ifcvt.c (noce_find_if_block): Do not clear if_info.cond_earliest
3099 field when initializing if_info structure.
3101 2007-08-24 Richard Guenther <rguenther@suse.de>
3103 * Makefile.in (tree-inline.o): Add $(TARGET_H) and $(INTEGRATE_H)
3105 * c-objc-common.c (c_cannot_inline_tree_fn): Remove.
3106 * langhooks.c (lhd_tree_inlining_cannot_inline_tree_fn): Likewise.
3107 * tree-inline.c (inlinable_function_p): Fold in common parts of
3108 the cannot_inline_tree_fn langhook.
3109 * langhooks-def.h (lhd_tree_inlining_cannot_inline_tree_fn): Remove.
3110 (LANG_HOOKS_TREE_INLINING_CANNOT_INLINE_TREE_FN): Likewise.
3111 (LANG_HOOKS_TREE_INLINING_INITIALIZER): Remove initializer for
3112 cannot_inline_tree_fn langhook.
3113 * langhooks.h (struct lang_hooks_for_tree_inlining): Remove
3114 cannot_inline_tree_fn member.
3116 2007-08-24 Richard Guenther <rguenther@suse.de>
3118 * expr.c (get_inner_reference): Do computation of bitoffset
3119 from offset in a way we can detect overflow reliably.
3121 2007-08-24 Jie Zhang <jie.zhang@analog.com>
3123 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define for uClibc.
3125 2007-08-24 Jie Zhang <jie.zhang@analog.com>
3127 * config/bfin/bfin.opt (mfast-fp): Add.
3128 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Let
3129 libbffastfp override libgcc if -mfast-fp.
3130 * config/bfin/bfin.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3131 * doc/invoke.texi (Option Summary): Mention -mfast-fp.
3132 (Blackfin Options): Document -mfast-fp.
3134 2007-08-24 Jakub Jelinek <jakub@redhat.com>
3136 PR tree-optimization/32573
3138 * tree-data-ref.c (initialize_data_dependence_relation): Clear
3139 DDR_SUBSCRIPTS, DDR_DIR_VECTS and DDR_DIST_VECTS at the beginning.
3140 (finalize_ddr_dependent): Clear DDR_SUBSCRIPTS after freeing it.
3141 (build_classic_dist_vector): Return false rather than true if
3142 DDR_ARE_DEPENDENT is non-NULL at the beginning. Return false
3143 if either subscript_dependence_tester_1 or build_classic_dist_vector_1
3144 returned false. Don't call save_dist_v before calling
3145 build_classic_dist_vector_1.
3146 (free_dependence_relation): Don't guard freeing DDR_SUBSCRIPTS
3147 with NULL DDR_ARE_DEPENDENT. Also free DDR_DIST_VECTS and/or
3148 DDR_DIR_VECTS vectors.
3150 2007-08-23 Brian Sidebotham <brian.sidebotham@gmail.com>
3152 * configure.ac (leb128): Modify sed statement to work with any
3153 binutils version string.
3154 * configure: Regenerate
3156 2007-08-23 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
3158 * config/arc/arc-protos.h arc_select_cc_mode, gen_compare_reg):
3159 Wrap in RTX_CODE macro guard.
3160 * config/arm/pe.c (arm_dllexport_p, arm_dllimport_p,
3161 arm_dllexport_name_p, arm_dllimport_name_p, arm_mark_dllexport,
3162 arm_mark_dllimport, arm_pe_encode_section_info,
3163 arm_pe_unique_section): Use ISO-C function declarations.
3164 * config/c4x/c4x-c.c (c4x_parse_pragma, c4x_pr_CODE_SECTION,
3165 c4x_pr_DATA_SECTION, c4x_pr_FUNC_IS_PURE,
3166 c4x_pr_FUNC_NEVER_RETURNS, c4x_pr_INTERRUPT, c4x_pr_ignored):
3168 * config/iq2000/iq2000.h (ASM_OUTPUT_SKIP): Fix format warning.
3169 * config/m68hc11/m68hc11.h (ASM_OUTPUT_EXTERNAL): Undef before
3171 * config/mips/mips.h (ASM_DECLARE_OBJECT_NAME): Fix format
3173 * config/mn10300/mn10300.h (OUTPUT_ADDR_CONST_EXTRA): Likewise.
3174 * config/pdp11/pdp11.c (pdp11_output_function_epilogue): Likewise.
3175 (register_move_cost): Use ISO-C function declarations.
3176 * config/pdp11/pdp11.h (PRINT_OPERAND): Fix format warnings.
3177 * config/score/score-protos.h (score_declare_object): Add
3179 * config/score/score.h (ASM_DECLARE_OBJECT_NAME): Fix format
3181 * final.c (profile_function): Avoid empty if-bodies.
3184 * calls.c (must_pass_in_stack_var_size,
3185 must_pass_in_stack_var_size_or_pad): Constify.
3186 * config/alpha/alpha-protos.h (function_value): Likewise.
3187 * config/alpha/alpha.c (alpha_return_in_memory,
3188 alpha_pass_by_reference, function_value,
3189 unicosmk_must_pass_in_stack, TARGET_PROMOTE_FUNCTION_ARGS,
3190 TARGET_PROMOTE_FUNCTION_RETURN, TARGET_PROMOTE_PROTOTYPES):
3192 * config/arc/arc.c (arc_return_in_memory, arc_pass_by_reference,
3193 TARGET_PROMOTE_FUNCTION_ARGS, TARGET_PROMOTE_FUNCTION_RETURN,
3194 TARGET_PROMOTE_PROTOTYPES): Likewise.
3195 * config/arm/arm-protos.h (arm_return_in_memory,
3196 arm_pad_arg_upward, arm_function_value): Likewise.
3197 * config/arm/arm.c (arm_pass_by_reference,
3198 arm_promote_prototypes, arm_return_in_msb, arm_must_pass_in_stack,
3199 TARGET_PROMOTE_FUNCTION_ARGS, TARGET_PROMOTE_FUNCTION_RETURN,
3200 arm_function_value, arm_return_in_memory, arm_pad_arg_upward):
3202 * config/arm/arm.h (ARM_DECLARE_FUNCTION_NAME): Likewise.
3203 * config/avr/avr-protos.h (avr_function_value): Likewise.
3204 * config/avr/avr.c (avr_return_in_memory,
3205 gas_output_limited_string, gas_output_ascii, avr_function_value,
3206 avr_return_in_memory): Likewise.
3207 * config/bfin/bfin-protos.h (bfin_return_in_memory): Likewise.
3208 * config/bfin/bfin.c (bfin_pass_by_reference,
3209 bfin_return_in_memory, TARGET_PROMOTE_PROTOTYPES,
3210 TARGET_PROMOTE_FUNCTION_ARGS, TARGET_PROMOTE_FUNCTION_RETURN):
3212 * config/cris/cris.c (cris_pass_by_reference,
3213 TARGET_PROMOTE_FUNCTION_ARGS): Likewise.
3214 * config/crx/crx.c (crx_return_in_memory): Likewise.
3215 * config/darwin.c (function_base, machopic_function_base_name):
3217 * config/fr30/fr30.c (fr30_must_pass_in_stack,
3218 TARGET_PROMOTE_PROTOTYPES): Likewise.
3219 * config/frv/frv.c (frv_must_pass_in_stack): Likewise.
3220 * config/h8300/h8300.c (h8300_return_in_memory): Likewise.
3221 * config/i386/i386-protos.h (ix86_return_in_memory,
3222 ix86_sol10_return_in_memory): Likewise.
3223 * config/i386/i386.c (ix86_function_value,
3224 ix86_function_sseregparm, ix86_must_pass_in_stack,
3225 type_natural_mode, classify_argument, examine_argument,
3226 construct_container, ix86_pass_by_reference, function_value_32,
3227 function_value_64, ix86_function_value_1, return_in_memory_32,
3228 return_in_memory_64, return_in_memory_ms_64,
3229 ix86_return_in_memory, ix86_sol10_return_in_memory,
3230 TARGET_PROMOTE_PROTOTYPES): Likewise.
3231 * config/ia64/ia64-protos.h (ia64_function_value,
3232 ia64_hpux_function_arg_padding): Likewise.
3233 * config/ia64/ia64.c (hfa_element_mode, ia64_return_in_memory,
3234 ia64_function_value, bundle_state_hash, bundle_state_eq_p,
3235 ia64_hpux_function_arg_padding): Likewise.
3236 * config/iq2000/iq2000-protos.h (function_arg,
3237 iq2000_function_value): Likewise.
3238 * config/iq2000/iq2000.c (iq2000_return_in_memory,
3239 iq2000_pass_by_reference, TARGET_PROMOTE_FUNCTION_ARGS,
3240 TARGET_PROMOTE_FUNCTION_RETURN, TARGET_PROMOTE_PROTOTYPES,
3241 function_arg, iq2000_function_value): Likewise.
3242 * config/m32c/m32c-protos.h (m32c_function_value,
3243 m32c_promote_function_return): Likewise.
3244 * config/m32c/m32c.c (m32c_pass_by_reference,
3245 m32c_promote_prototypes, m32c_promote_function_return,
3246 m32c_function_value): Likewise.
3247 * config/m32r/m32r.c (m32r_return_in_memory,
3248 m32r_pass_by_reference, TARGET_PROMOTE_PROTOTYPES,
3249 m32r_in_small_data_p): Likewise.
3250 * config/m68hc11/m68hc11-protos.h (m68hc11_function_arg_padding):
3252 * config/m68hc11/m68hc11.c (m68hc11_return_in_memory,
3253 m68hc11_function_arg_padding): Likewise.
3254 * config/m68k/m68k-protos.h (m68k_function_value): Likewise.
3255 * config/m68k/m68k.c (TARGET_PROMOTE_PROTOTYPES,
3256 m68k_function_value): Likewise.
3257 * config/mcore/mcore-protos.h (mcore_num_arg_regs,
3258 mcore_function_value): Likewise.
3259 * config/mcore/mcore.c (handle_structs_in_regs,
3260 mcore_return_in_memory, TARGET_PROMOTE_FUNCTION_ARGS,
3261 TARGET_PROMOTE_FUNCTION_RETURN, TARGET_PROMOTE_PROTOTYPES,
3262 mcore_num_arg_regs, mcore_function_value): Likewise.
3263 * config/mips/mips-protos.h (mips_pad_arg_upward,
3264 mips_function_value): Likewise.
3265 * config/mips/mips.c (mips_fpr_return_fields, mips_return_in_msb,
3266 mips_return_in_memory, mips_pass_by_reference, mips_callee_copies,
3267 TARGET_PROMOTE_FUNCTION_ARGS, TARGET_PROMOTE_FUNCTION_RETURN,
3268 TARGET_PROMOTE_PROTOTYPES, mips_pad_arg_upward,
3269 mips_function_value): Likewise.
3270 * config/mmix/mmix-protos.h (mmix_function_outgoing_value):
3272 * config/mmix/mmix.c (mmix_pass_by_reference,
3273 TARGET_PROMOTE_FUNCTION_ARGS, mmix_function_outgoing_value,
3274 mmix_encode_section_info): Likewise.
3275 * config/mn10300/mn10300-protos.h (mn10300_function_value):
3277 * config/mn10300/mn10300.c (mn10300_return_in_memory,
3278 mn10300_pass_by_reference, TARGET_PROMOTE_PROTOTYPES,
3279 mn10300_function_value): Likewise.
3280 * config/mt/mt-protos.h (mt_function_value): Likewise.
3281 * config/mt/mt.c (mt_pass_by_reference, mt_function_value,
3282 mt_pass_in_stack, TARGET_PROMOTE_PROTOTYPES): Likewise.
3283 * config/pa/pa-protos.h (function_arg_padding, function_value,
3284 pa_return_in_memory): Likewise.
3285 * config/pa/pa.c (pa_pass_by_reference,
3286 TARGET_PROMOTE_FUNCTION_RETURN, TARGET_PROMOTE_PROTOTYPES,
3287 function_arg_padding, function_value, pa_return_in_memory):
3289 * config/pdp11/pdp11.c (pdp11_return_in_memory): Likewise.
3290 * config/rs6000/rs6000-protos.h (rs6000_function_value,
3291 function_arg_padding): Likewise.
3292 * config/rs6000/rs6000.c (rs6000_return_in_memory,
3293 rs6000_darwin64_record_arg_recurse, rs6000_darwin64_record_arg,
3294 rs6000_pass_by_reference, rs6000_must_pass_in_stack,
3295 TARGET_PROMOTE_FUNCTION_ARGS, TARGET_PROMOTE_FUNCTION_RETURN,
3296 function_arg_padding, altivec_expand_dst_builtin,
3297 altivec_expand_builtin, rs6000_expand_builtin, spe_init_builtins,
3298 altivec_init_builtins, rs6000_common_init_builtins,
3299 rs6000_function_value): Likewise.
3300 * s390/s390-protos.h (s390_function_value): Likewise.
3301 * config/s390/s390.c (s390_function_arg_size,
3302 s390_pass_by_reference, s390_return_in_memory,
3303 s390_function_value, TARGET_PROMOTE_FUNCTION_ARGS,
3304 TARGET_PROMOTE_FUNCTION_RETURN): Likewise.
3305 * config/score/score-protos.h (score_function_value): Likewise.
3306 * config/score/score.c (score_arg_partial_bytes,
3307 TARGET_PROMOTE_FUNCTION_ARGS, TARGET_PROMOTE_FUNCTION_RETURN,
3308 TARGET_PROMOTE_PROTOTYPES, score_return_in_memory,
3309 score_pass_by_reference, score_add_offset, score_function_value):
3311 * config/sh/sh-protos.h (sh_attr_renesas_p,
3312 sh_promote_prototypes): Likewise.
3313 * config/sh/sh.c (sh_return_in_memory, sh_pass_by_reference,
3314 sh_callee_copies, sh_promote_prototypes, shcompact_byref,
3315 sh_attr_renesas_p): Likewise.
3316 * config/sparc/sparc-protos.h (function_value,
3317 function_arg_padding): Likewise.
3318 * config/sparc/sparc.c (sparc_promote_prototypes,
3319 sparc_return_in_memory, sparc_pass_by_reference,
3320 TARGET_PROMOTE_FUNCTION_ARGS, TARGET_PROMOTE_FUNCTION_RETURN,
3321 function_arg_record_value_2, function_arg_record_value_1,
3322 function_arg_record_value, function_arg_record_value,
3323 function_arg_padding, function_value): Likewise.
3324 * config/spu/spu-protos.h (spu_function_value): Likewise.
3325 * config/spu/spu.c (spu_pass_by_reference, spu_return_in_memory,
3326 spu_function_value): Likewise.
3327 * config/stormy16/stormy16-protos.h (xstormy16_function_value):
3329 * config/stormy16/stormy16.c (xstormy16_return_in_memory,
3330 xstormy16_function_value, TARGET_PROMOTE_FUNCTION_ARGS,
3331 TARGET_PROMOTE_FUNCTION_RETURN, TARGET_PROMOTE_PROTOTYPES):
3333 * config/v850/v850.c (v850_return_in_memory,
3334 v850_pass_by_reference, TARGET_PROMOTE_PROTOTYPES): Likewise.
3335 * config/vax/vax.c (TARGET_PROMOTE_PROTOTYPES): Likewise.
3336 * config/xtensa/xtensa.c (xtensa_return_in_msb,
3337 xtensa_return_in_memory, TARGET_PROMOTE_FUNCTION_ARGS,
3338 TARGET_PROMOTE_FUNCTION_RETURN, TARGET_PROMOTE_PROTOTYPES):
3340 * explow.c (promote_mode, hard_function_value): Likewise.
3341 * expr.h (hard_function_value, promote_mode): Likewise.
3342 * function.c (aggregate_value_p): Likewise.
3343 * hooks.c (hook_bool_const_tree_true): New.
3344 * hooks.h (hook_bool_const_tree_true): New.
3345 * sdbout.c (SET_KNOWN_TYPE_TAG, plain_type_1): Constify.
3346 * target-def.h (TARGET_PROMOTE_FUNCTION_ARGS,
3347 TARGET_PROMOTE_FUNCTION_RETURN, TARGET_PROMOTE_PROTOTYPES,
3348 TARGET_RETURN_IN_MSB): Likewise.
3349 * target.h (promote_function_args, promote_function_return,
3350 promote_prototypes, return_in_memory, return_in_msb,
3351 pass_by_reference, must_pass_in_stack, callee_copies,
3352 function_value): Likewise.
3353 * targhooks.c (default_return_in_memory,
3354 hook_pass_by_reference_must_pass_in_stack,
3355 hook_callee_copies_named,
3356 hook_bool_CUMULATIVE_ARGS_mode_tree_bool_false,
3357 hook_bool_CUMULATIVE_ARGS_mode_tree_bool_true,
3358 default_function_value): Likewise.
3359 * targhooks.h (default_return_in_memory,
3360 hook_pass_by_reference_must_pass_in_stack,
OSDN Git Service
