This is the mail archive of the
gcc-regression@gcc.gnu.org
mailing list for the GCC project.
A recent patch increased GCC's memory consumption in some cases!
- From: gcctest at suse dot de
- To: jh at suse dot cz, gcc-regression at gcc dot gnu dot org
- Date: Fri, 21 Jul 2006 15:53:44 +0000
- Subject: A recent patch increased GCC's memory consumption in some cases!
Hi,
I am a friendly script caring about memory consumption in GCC. Please
contact jh@suse.cz if something is going wrong.
Comparing memory consumption on compilation of combine.i, insn-attrtab.i,
and generate-3.4.ii I got:
comparing combine.c compilation at -O0 level:
Peak amount of GGC memory allocated before garbage collecting run decreased from 9503k to 9160k, overall -3.74%
Peak amount of GGC memory still allocated after garbage collecting decreased from 8890k to 8547k, overall -4.01%
Amount of memory still referenced at the end of compilation decreased from 6742k to 6052k, overall -11.41%
Overall memory needed: 25013k -> 24717k
Peak memory use before GGC: 9503k -> 9160k
Peak memory use after GGC: 8890k -> 8547k
Maximum of released memory in single GGC run: 2603k -> 2580k
Garbage: 38649k -> 38214k
Leak: 6742k -> 6052k
Overhead: 5888k -> 5326k
GGC runs: 302 -> 307
comparing combine.c compilation at -O1 level:
Overall memory allocated via mmap and sbrk increased from 26888k to 36357k, overall 35.22%
Amount of produced GGC garbage decreased from 58951k to 56532k, overall -4.28%
Amount of memory still referenced at the end of compilation decreased from 6860k to 6066k, overall -13.09%
Overall memory needed: 26888k -> 36357k
Peak memory use before GGC: 17424k -> 16999k
Peak memory use after GGC: 17245k -> 16828k
Maximum of released memory in single GGC run: 2284k -> 2261k
Garbage: 58951k -> 56532k
Leak: 6860k -> 6066k
Overhead: 7253k -> 6070k
GGC runs: 376 -> 378
comparing combine.c compilation at -O2 level:
Amount of produced GGC garbage decreased from 83040k to 80305k, overall -3.41%
Amount of memory still referenced at the end of compilation decreased from 6989k to 6170k, overall -13.28%
Overall memory needed: 26888k -> 26496k
Peak memory use before GGC: 17428k -> 16997k
Peak memory use after GGC: 17245k -> 16828k
Maximum of released memory in single GGC run: 2385k -> 2254k
Garbage: 83040k -> 80305k
Leak: 6989k -> 6170k
Overhead: 9950k -> 8600k
GGC runs: 442 -> 446
comparing combine.c compilation at -O3 level:
Amount of produced GGC garbage decreased from 113734k to 109841k, overall -3.54%
Amount of memory still referenced at the end of compilation decreased from 7077k to 6247k, overall -13.29%
Overall memory needed: 25988k -> 25600k
Peak memory use before GGC: 18565k -> 18087k
Peak memory use after GGC: 17879k -> 17424k
Maximum of released memory in single GGC run: 3521k -> 3346k
Garbage: 113734k -> 109841k
Leak: 7077k -> 6247k
Overhead: 13575k -> 11873k
GGC runs: 491 -> 496
comparing insn-attrtab.c compilation at -O0 level:
Amount of memory still referenced at the end of compilation decreased from 10119k to 9116k, overall -11.01%
Overall memory needed: 82160k -> 81508k
Peak memory use before GGC: 69029k -> 68248k
Peak memory use after GGC: 44784k -> 43996k
Maximum of released memory in single GGC run: 36008k -> 35708k
Garbage: 140138k -> 138286k
Leak: 10119k -> 9116k
Overhead: 20393k -> 18992k
GGC runs: 239 -> 241
comparing insn-attrtab.c compilation at -O1 level:
Amount of produced GGC garbage decreased from 284126k to 268543k, overall -5.80%
Amount of memory still referenced at the end of compilation decreased from 10055k to 8940k, overall -12.47%
Overall memory needed: 105924k -> 103676k
Peak memory use before GGC: 85564k -> 83566k
Peak memory use after GGC: 79517k -> 77686k
Maximum of released memory in single GGC run: 32268k -> 31934k
Garbage: 284126k -> 268543k
Leak: 10055k -> 8940k
Overhead: 34782k -> 29970k
GGC runs: 239 -> 235
comparing insn-attrtab.c compilation at -O2 level:
Ovarall memory allocated via mmap and sbrk decreased from 111844k to 103656k, overall -7.90%
Amount of produced GGC garbage decreased from 331999k to 314640k, overall -5.52%
Amount of memory still referenced at the end of compilation decreased from 10041k to 8935k, overall -12.38%
Overall memory needed: 111844k -> 103656k
Peak memory use before GGC: 89993k -> 87827k
Peak memory use after GGC: 81920k -> 79920k
Maximum of released memory in single GGC run: 30719k -> 30382k
Garbage: 331999k -> 314640k
Leak: 10041k -> 8935k
Overhead: 42237k -> 36950k
GGC runs: 265 -> 263
comparing insn-attrtab.c compilation at -O3 level:
Ovarall memory allocated via mmap and sbrk decreased from 111500k to 103660k, overall -7.56%
Amount of produced GGC garbage decreased from 332603k to 315231k, overall -5.51%
Amount of memory still referenced at the end of compilation decreased from 10045k to 8938k, overall -12.39%
Overall memory needed: 111500k -> 103660k
Peak memory use before GGC: 90019k -> 87853k
Peak memory use after GGC: 81946k -> 79946k
Maximum of released memory in single GGC run: 30916k -> 30573k
Garbage: 332603k -> 315231k
Leak: 10045k -> 8938k
Overhead: 42424k -> 37130k
GGC runs: 268 -> 267
comparing Gerald's testcase PR8361 compilation at -O0 level:
Amount of memory still referenced at the end of compilation decreased from 49496k to 47149k, overall -4.98%
Overall memory needed: 118244k -> 116456k
Peak memory use before GGC: 95013k -> 92791k
Peak memory use after GGC: 94065k -> 91863k
Maximum of released memory in single GGC run: 20358k -> 19688k
Garbage: 218015k -> 214074k
Leak: 49496k -> 47149k
Overhead: 37974k -> 22154k
GGC runs: 362 -> 416
comparing Gerald's testcase PR8361 compilation at -O1 level:
Peak amount of GGC memory allocated before garbage collecting run decreased from 97659k to 93832k, overall -4.08%
Peak amount of GGC memory still allocated after garbage collecting decreased from 96680k to 92903k, overall -4.07%
Amount of produced GGC garbage decreased from 465511k to 437074k, overall -6.51%
Amount of memory still referenced at the end of compilation decreased from 52775k to 48570k, overall -8.66%
Overall memory needed: 115820k -> 112988k
Peak memory use before GGC: 97659k -> 93832k
Peak memory use after GGC: 96680k -> 92903k
Maximum of released memory in single GGC run: 18987k -> 18471k
Garbage: 465511k -> 437074k
Leak: 52775k -> 48570k
Overhead: 53396k -> 31867k
GGC runs: 505 -> 557
comparing Gerald's testcase PR8361 compilation at -O2 level:
Peak amount of GGC memory allocated before garbage collecting run decreased from 97659k to 93832k, overall -4.08%
Peak amount of GGC memory still allocated after garbage collecting decreased from 96680k to 92903k, overall -4.07%
Amount of produced GGC garbage decreased from 555592k to 521176k, overall -6.60%
Amount of memory still referenced at the end of compilation decreased from 53672k to 49262k, overall -8.95%
Overall memory needed: 115820k -> 112952k
Peak memory use before GGC: 97659k -> 93832k
Peak memory use after GGC: 96680k -> 92903k
Maximum of released memory in single GGC run: 18987k -> 18470k
Garbage: 555592k -> 521176k
Leak: 53672k -> 49262k
Overhead: 62424k -> 39516k
GGC runs: 588 -> 635
comparing Gerald's testcase PR8361 compilation at -O3 level:
Peak amount of GGC memory still allocated after garbage collecting decreased from 97976k to 94524k, overall -3.65%
Amount of produced GGC garbage decreased from 575982k to 539157k, overall -6.83%
Amount of memory still referenced at the end of compilation decreased from 54028k to 49600k, overall -8.93%
Overall memory needed: 117116k -> 114920k
Peak memory use before GGC: 98956k -> 96315k
Peak memory use after GGC: 97976k -> 94524k
Maximum of released memory in single GGC run: 19240k -> 18769k
Garbage: 575982k -> 539157k
Leak: 54028k -> 49600k
Overhead: 63531k -> 40673k
GGC runs: 595 -> 643
Head of the ChangeLog is:
--- /usr/src/SpecTests/sandbox-britten-memory/x86_64/mem-result/ChangeLog 2006-05-25 22:44:07.000000000 +0000
+++ /usr/src/SpecTests/sandbox-britten-memory/gcc/gcc/ChangeLog 2006-07-20 03:01:06.000000000 +0000
@@ -1,8 +1,1779 @@
+2006-07-19 Jakub Jelinek <jakub@redhat.com>
+
+ * except.c (duplicate_eh_regions_1): Clear next_peer.
+
+2006-07-17 Nathan Sidwell <nathan@codesourcery.com>
+
+ * doc/invoke.texi (RS/6000 and PowerPC Options): Document that
+ -mlongcall does not guarantee all calls will be long.
+ * doc/extend.texi (long_call/short_call): Document that longcall
+ does not guarantee a long call will be used.
+ (longcall/shortcall): Likewise.
+
+2006-07-18 Roger Sayle <roger@eyesopen.com>
+
+ PR middle-end/28283
+ * expmed.c (expand_shift): Additionally check that the shift_cost
+ is not MAX_COST and that INTVAL(op1) is less than MAX_BITS_PER_WORD
+ before implementing a LSHIFT_EXPR as a sequence of additions.
+ * config/sh/sh.c (shift_costs): Return MAX_COST to inform the
+ middle-end that DImode shifts need to be synthesized by expand.
+
+2006-07-18 Diego Novillo <dnovillo@redhat.com>
+
+ PR 28410
+ * tree-ssa-operands.c (access_can_touch_variable): Update
+ comment.
+ Return true if ALIAS is .GLOBAL_VAR.
+
+2006-07-18 David Daney <ddaney@avtrex.com>
+
+ * gcc.c (display_help): Fix typo in help text.
+
+2006-07-18 Olivier Hainque <hainque@adacore.com>
+
+ * tree.h (categorize_ctor_elements): Adjust prototype and add
+ descriptive comment, both in accordance with the interface change
+ described below.
+ * varasm.c (constructor_static_from_elts_p): New function.
+ Whether a constructor node is a valid static constant initializer
+ if all its elements are.
+ (initializer_constant_valid_p) <CONSTRUCTOR value>: Use it.
+ * output.h: Declare it.
+ * expr.c (categorize_ctor_elements_1): Return whether the constructor
+ is a valid constant initializer instead of computing the number of
+ non-constant elements. Use constructor_static_from_elts_p for this
+ purpose. Replace the head comment with an indication that this is a
+ helper for categorize_ctor_elements.
+ (categorize_ctor_elements): Same interface change as for the _1
+ helper. Former head comment from this helper moved here, adjusted to
+ account for the interface changes.
+ (mostly_zeros_p): Adjust call to categorize_ctor_elements.
+ (all_zeros_p): Likewise.
+ * gimplify.c (gimplify_init_constructor): Decide whether we can make
+ static versions of the constructor from the categorize_ctor_elements
+ return value instead of the formerly computed number of non-constant
+ elements.
+
+2006-07-18 Paolo Bonzini <bonzini@gnu.org>
+
+ * configure: Regenerate.
+ * Makefile.in (build_libsubdir): New configure substitution
+ (build_libobjdir): New variable.
+ (BUILD_LIBIBERTY): Use it.
+
+2006-07-18 Volker Reichelt <reichelt@igpm.rwth-aachen.de>
+
+ PR c/28286
+ * c-pragma.c (handle_pragma_pack): Handle invalid constants.
+
+2006-07-18 Alexandre Oliva <aoliva@redhat.com>
+
+ PR c/26993
+ * c-common.c (handle_weakref_attribute): Ignore attribute in
+ the same conditions the alias attribute is ignored.
+
+2006-07-17 Carlos O'Donell <carlos@codesourcery.com>
+
+ * dbxout.c (dbxout_function_end): Do not increment scope_labelno.
+ (dbxout_begin_prologue): Increment scope_labelno.
+
+2006-07-17 Richard Henderson <rth@redhat.com>
+
+ PR 27889
+ * tree-nested.c (create_tmp_var_for): Set DECL_COMPLEX_GIMPLE_REG_P
+ if needed.
+
+2006-07-17 Roger Sayle <roger@eyesopen.com>
+ Rainer Orth <ro@TechFak.Uni-Bielefeld.DE>
+
+ * config/i386/gmon-sol2.c: Silence compilation warnings by partially
+ synchronizing code with config/sparc/gmon-sol2.c. Remove #if 0
+ blocks. #include "tconfig.h" and "tsystem.h" to prototype system
+ functions such as sbrk, write, etc.. Add function prototypes.
+ Convert function declarations from K&R to ISO C. Cast pointers
+ to "long" and "unsigned long" instead of "int" and "unsigned int".
+ Convert tokens following #endif into comments.
+
+2006-07-17 Toon Moene <toon@moene.indiv.nluug.nl>
+
+ * invoke.texi: Remove mention of f77; Reformat table for
+ inline parameter options correctly; mention that -malign-double
+ is default on x86-64 targets.
+
+2006-07-17 Zdenek Dvorak <dvorakz@suse.cz>
+
+ * tree-chrec.c (avoid_arithmetics_in_type_p): New.
+ (convert_affine_scev, chrec_convert_aggressive): Use
+ avoid_arithmetics_in_type_p. Do not check for the subtypes
+ separately.
+
+2006-07-17 Richard Sandiford <richard@codesourcery.com>
+
+ PR middle-end/28403
+ * optabs.c (expand_doubleword_shift): Wrap the call to
+ do_compare_rtx_and_jump with NO_DEFER_POP and OK_DEFER_POP.
+
+2006-07-17 Richard Sandiford <richard@codesourcery.com>
+
+ PR middle-end/28402
+ * optabs.c (expand_binop): Pass next_methods rather than methods
+ to expand_doubleword_shift.
+
+2006-07-17 J"orn Rennecke <joern.rennecke@st.com>
+
+ PR other/28251
+ * tree.h (dump_addr): Declare.
+ * print-tree.c (dump_addr): New function.
+ (print_node_brief, print_node): Use it.
+ * print-rtl.c (print_rtx): Likewise.
+ * common.opt (-fdump-noaddr): New option.
+ * doc/invoke.texi (-fdump-noaddr): Document.
+ * loop-unroll.c (si_info_hash): Make hash independent of addresses.
+ (ve_info_hash): Likewise.
+
+2006-07-13 Andrew Haley <aph@redhat.com>
+
+ PR tree-optimization/19505
+ * tree-cfgcleanup.c (tree_forwarder_block_p): If we have an EH
+ edge leaving this block, make sure that the destination of this
+ block has only one predecessor.
+
+2006-07-17 Richard Guenther <rguenther@suse.de>
+
+ PR tree-optimization/28238
+ * tree-inline.c (copy_bb): Check if we produced valid
+ gimple copying and substituting a stmt. If not, gimplify
+ it.
+
+2006-07-16 Eric Botcazou <ebotcazou@adacore.com>
+
+ * optabs.c (maybe_encapsulate_block): New function extracted from...
+ (emit_libcall_block): ...here. Invoke it on the block of insns to
+ maybe emit REG_LIBCALL/REG_RETVAL notes around the block.
+ (emit_no_conflict_block): Likewise.
+
+2006-07-16 Eric Botcazou <ebotcazou@libertysurf.fr>
+
+ * doc/install.texi (sparc-sun-solaris2*): Add GMP version number.
+
+2006-07-15 Roger Sayle <roger@eyesopen.com>
+
+ * tree-gimple.c (is_gimple_stmt): Sink test of IS_EMPTY_STMT into
+ a new NOP_EXPR case of the switch statement.
+
+2006-07-15 Kaz Kojima <kkojima@gcc.gnu.org>
+
+ PR middle-end/28160
+ * stor-layout.c (place_field): Take the bit field with
+ an excessive size into account in the ms-bitfiled case.
+
+ PR middle-end/28161
+ * stor-layout.c (place_field): Use DECL_BIT_FIELD_TYPE of
+ the previous bit field.
+
+2006-07-14 Eliot Dresselhaus <eliot@sonic.net>
+
+ PR target/27287
+ * config/rs6000/spe.md (frob_di_df_2): Add m->r alternative.
+
+2006-07-14 Ulrich Weigand <uweigand@de.ibm.com>
+
+ * config/s390/linux-unwind.h (s390_fallback_frame_state): Detect
+ signal frames correctly even when the signal was installed with
+ sa_restorer set.
+
+2006-07-14 Carlos O'Donell <carlos@codesourcery.com>
+
+ * dbxout.c (dbxout_begin_prologue): Emit LFBB label at function start.
+ (dbxout_function_end): Use local label LFBB. Remove hook
+ DBX_OUTPUT_NFUN.
+ (dbxout_source_line): Use local label LFBB.
+ (dbxout_block): Likewise.
+ (dbx_output_lbrac): Remove hook DBX_OUTPUT_LBRAC.
+ (dbx_output_rbrac): Remove hook DBX_OUTPUT_RBRAC.
+ * config/rs6000/linux64.h: Remove DBX_OUTPUT_LBRAC, DBX_OUTPUT_RBRAC,
+ and DBX_OUTPUT_NFUN.
+
+2006-07-13 Janis Johnson <janis187@us.ibm.com>
+
+ * doc/sourcebuild.texi (Test Directives): Document dg-shouldfail.
+
+2006-07-13 Jan Hubicka <jh@suse.cz>
+
+ * cgraphunit.c (cgraph_varpool_analyze_pending_decls): Call
+ align_variable.
+ * output.h (align_variable): Declare.
+ * varasm.c (align_variable): Export.
+ * value-prof.c (tree_value_profile_transformations): Recompute iterator
+ when basic block changed.
+
+2006-07-13 Nick Clifton <nickc@redhat.com>
+
+ * config/sh/sh.c (sh_reorg): Ignore deleted insns whilst
+ walking the LOG_LINKS chain.
+
+2006-07-12 Geoffrey Keating <geoffk@apple.com>
+
+ * doc/invoke.texi (C++ Dialect Options): Explain difference
+ between -fvisibility-inlines-hidden and setting hidden
+ visibility explicitly.
+
+2006-07-12 Eric Christopher <echristo@apple.com>
+
+ * config/t-slibgcc-darwin (SHLIB_LINK): Don't munge stmp-lipo.
+ (stmp-lipo): Depend on libgcc.a.
+
+2006-07-12 Kazu Hirata <kazu@codesourcery.com>
+
+ * tree.c: Fix a comment typo.
+
+2006-07-12 Naveen.H.S <naveenh@kpitcummins.com>
+
+ * config/m32c/muldiv.md (umulhisi3_c): Use only registers for
+ dest.
+ (umulhisi3_r): Likewise.
+
+2006-07-12 Paolo Carlini <pcarlini@suse.de>
+
+ PR libstdc++/27878
+ * doc/install.texi (mips-sgi-irix6): Add note about IRIX 6.5.x,
+ x < 19, vs wchar_t support.
+
+2006-07-12 Nick Clifton <nickc@redhat.com>
+
+ * config/sh/sh.h (SUBTARGET_ASM_RELAX_SPEC): Use -isa=sh4-up
+ instead of -isa=sh4.
+
+2006-07-12 Mike Stump <mrs@apple.com>
+
+ * protoize.c (edit_fn_definition): Add volatile to clean_text_p.
+
+2006-07-10 Eric Christopher <echristo@apple.com>
+
+ * config/t-slibgcc-darwin: Rewrite.
+
+2006-07-10 Laurynas Biveinis <laurynas.biveinis@gmail.com>
+
+ PR bootstrap/20437
+ * Makefile.in (configure): add missing aclocal.m4 and acinclude.m4
+ dependencies.
+
+2006-07-08 Matthias Klose <doko@debian.org>
+
+ * doc/sourcebuild.texi: Fix typo.
+
+2006-07-07 Mike Stump <mrs@apple.com>
+
+ * toplev.c (process_options): Don't give -ffunction-sections may
+ affect debugging warning for Mach-O.
+
+2006-07-07 James E Wilson <wilson@specifix.com>
+
+ * config/mips/mips.md (cpu): Add sb1a.
+ * config/mips/sb1.md: Modify almost all patterns to accept sb1a in
+ addition to sb1. Adjust comments to clarify SB-1A differences from
+ SB-1.
+ (ir_sb1a_load): New. Modify bypasses to use it.
+ (ir_sb1a_simple_alu): Likewise.
+ * config/mips/mips.c (mips_cpu_info_table): Add sb1a.
+ (mips_rtx_cost_data): Add PROCESSOR_SB1A support.
+ (mips_issue_rate): Add PROCESSOR_SB1A.
+ (mips_multipass_dfa_lookahead): Use TUNE_SB1 instead of PROCESSOR_SB1.
+ * config/mips/mips.h (processor_type): Add PROCESSOR_SB1A.
+ (TARGET_SB1): Add PROCESSOR_SB1A check.
+ (TUNE_SB1): Likewise.
+
+ * config/mips/mips-protos.h (mips_store_data_bypass_p): New.
+ * config/mips/mips.c (mips_store_data_bypass_p): New.
+ * config/mips/sb1.md: Use mips_store_data_bypass_p instead of
+ store_data_bypass_p.
+
+2006-07-08 Paul Brook <paul@codesourcery.com>
+
+ PR target/27991
+ * config/m68k/m68k.md: Add r-m patterns for 68881 movxf.
+
+2006-07-07 Richard Guenther <rguenther@suse.de>
+
+ PR middle-end/28268
+ * tree.h (build_one_cst): Declare.
+ * tree.c (build_one_cst): New function.
+ * tree-ssa-math-opts.c (get_constant_one): Remove.
+ (insert_reciprocals): Use build_one_cst.
+ * fold-const.c (fold_plusminus_mult): Likewise.
+
+2006-07-07 Roger Sayle <roger@eyesopen.com>
+
+ * pointer-set.c (pointer_set_destroy): Correct whitespace.
+ * cfgloopanal.c (dump_graph): Likewise.
+ * dojump.c (discard_pending_stack_adjust): Likewise.
+ * fix-header.c (TARGET_EXTRA_INCLUDE, TARGET_EXTRA_PRE_INCLUDES):
+ Likewise.
+ * alloc-pool.c (dump_alloc_pool_statistics): Likewise.
+ * function.c (finalize_block_changes): Likewise.
+ * gcc-common.c (ggc_free_overhead, dump_gcc_loc_statistics): Likewise.
+ * tree-ssa-dce.c (clear_control_dependence_bitmap): Likewise.
+ * rtl.c (dump_rtx_statistics): Likewise.
+ * tree-ssa-structalias.c (sort_fieldstack, init_alias_heapvars):
+ Likewise.
+ * varray.c (dump_varray_statistics): Likewise.
+ * dfp.c (encode_decimal32, decode_decimal32, encode_decimal64,
+ decode_decimal64, encode_decimal128, decode_decimal128,
+ decimal_real_to_decimal): Likewise.
+
+2006-07-07 Richard Guenther <rguenther@suse.de>
+
+ PR tree-optimization/28187
+ * tree-vrp.c (vrp_operand_equal_p): New function.
+ (vrp_bitmap_equal_p): Likewise.
+ (update_value_range): Use them to compare old and new
+ max and min values.
+
+2006-07-06 Roger Sayle <roger@eyesopen.com>
+
+ * c-parser.c (c_parser_skip_to_end_of_block_or_statement): Add
+ a default case to switch statement to silence compiler warnings.
+
+2006-07-06 Roger Sayle <roger@eyesopen.com>
+
+ PR target/27968
+ * configure.ac (HAVE_AS_IX86_FFREEP): On x86 targets check whether
+ the configured assembler supports the x87's ffreep mnemonic.
+ * configure: Regenerate.
+ * config.in: Regenerate.
+
+ * config/i386/i386.c (output_387_ffreep): New function.
+ (output_fp_compare): Use output_387_ffreep to emit ffreep insns.
+ (output_387_reg_move): Likewise.
+
+2006-07-06 Paul Eggert <eggert@cs.ucla.edu>
+
+ Port to hosts whose 'sort' and 'tail' implementations
+ treat operands with leading '+' as file names, as POSIX
+ has required since 2001. However, make sure the code still
+ works on pre-POSIX hosts.
+ * Makefile.in (slowcompare): Port to POSIX 1003.1-2001,
+ which says you should use "tail -c +N" rather than "tail +Nc".
+ Fix a bug: the old code incorrectly skipped 15 bytes, not 16.
+
+2006-07-06 Zdenek Dvorak <dvorakz@suse.cz>
+
+ * tree-ssa-loop-niter.c (scev_probably_wraps_p): Fix typo in
+ argument name.
+
+2006-07-06 David Edelsohn <edelsohn@gnu.org>
+
+ PR target/28150
+ * config/rs6000/rs6000.c (rs6000_legitimate_address): Do not allow
+ PRE_{INC,DEC} of TFmode.
+
+2006-07-06 David Edelsohn <edelsohn@gnu.org>
+ Alan Modra <amodra@bigpond.net.au>
+
+ PR target/28170
+ * config/rs6000/rs6000.c (insvdi_rshift_rlwimi_p): Correct shiftop
+ bounds. Simplify.
+
+2006-07-06 Rainer Orth <ro@TechFak.Uni-Bielefeld.DE>
+
+ PR target/28285
+ * config/alpha/predicates.md (alpha_swapped_comparison_operator):
+ Remove duplicate gtu.
+
+2006-07-05 Jason Merrill <jason@redhat.com>
+
+ PR c++/13983
+ PR c++/17519
+ * stor-layout.c (finish_record_layout): Copy TYPE_PACKED to variants.
+ * c-common.c (handle_packed_attribute): So don't copy it here.
+ * c-decl.c (finish_struct): Don't copy TYPE_ALIGN.
+
+2006-07-05 Mike Stump <mrs@apple.com>
+
+ * doc/invoke.texi (Invoking G++): Clarify prose for g++.
+
+2006-07-05 H.J. Lu <hongjiu.lu@intel.com>
+
+ PR target/26146
+ * config.gcc: Check with_cpu for i[34567]86--*-solaris2.1[0-9]*.
+
+2006-07-05 Richard Guenther <rguenther@suse.de>
+
+ PR target/28158
+ * config/i386/i386.md (*negdf2_1): Restrict pattern to
+ TARGET_SSE2 targets if using SSE math.
+ (*absdf2_1): Likewise.
+
+2006-07-05 Richard Guenther <rguenther@suse.de>
+
+ PR tree-optimization/28162
+ * fold-const.c (fold_binary): For (-A) * (-B) -> A * B
+ make sure to convert the operands to the correct type.
+
+2006-07-04 Paolo Bonzini <bonzini@gnu.org>
+
+ PR tree-optimization/28218
+
+ * tree-ssa-math-opts.c (execute_cse_reciprocals): Fix calls
+ to calculate and free the dominator information.
+
+2006-07-04 Peter O'Gorman <peter@pogma.com>
+
+ * mklibgcc.in: chmod 644 before ranlib during install.
+
+2006-07-04 Andrew Stubbs <andrew.stubbs@st.com>
+ J"orn Rennecke <joern.rennecke@st.com>
+
+ * configure.ac (.eh_frame data check): Use diff -b.
+
+2006-07-04 Eric Botcazou <ebotcazou@libertysurf.fr>
+
+ PR bootstrap/18058
+ * Makefile.in (BUILD_RTL): Add build/vec.o.
+ (build/gencondmd.o): Filter out -fkeep-inline-functions.
+ (build/genextract): Delete.
+ (build/genautomata): Likewise.
+
+2006-07-04 Alan Modra <amodra@bigpond.net.au>
+
+ PR target/28207
+ * config/rs6000/rs6000.c (function_arg_boundary): Double-word align
+ 128-bit IBM long doubles for ABI_V4.
+
+2006-07-03 Eric Botcazou <ebotcazou@adacore.com>
+
+ * tree.c (range_in_array_bounds_p): New predicate.
+ * tree.h (range_in_array_bounds_p): Declare it.
+ * tree-eh.c (tree_could_trap_p) <ARRAY_RANGE_REF>: Use it to
+ return a less conservative answer.
+ * tree-sra.c (struct sra_elt): Add new pointer field 'groups'
+ and flag 'is_group'.
+ (IS_ELEMENT_FOR_GROUP): New macro.
+ (FOR_EACH_ACTUAL_CHILD): Likewise.
+ (next_child_for_group): New helper function.
+ (can_completely_scalarize_p): Take into account groups.
+ (sra_hash_tree): Handle RANGE_EXPR.
+ (sra_elt_eq): Likewise.
+ (lookup_element): Be prepared for handling groups.
+ (is_valid_const_index): Delete.
+ (maybe_lookup_element_for_expr) <ARRAY_REF>: Use in_array_bounds_p
+ instead of is_valid_const_index.
+ <ARRAY_RANGE_REF>: New case.
+ (sra_walk_expr) <ARRAY_REF>: Use in_array_bounds_p instead of
+ is_valid_const_index.
+ <ARRAY_RANGE_REF>: Do not unconditionally punt.
+ (scan_dump): Dump info for groups too.
+ (decide_instantiation_1): Likewise.
+ (decide_block_copy): Assert that the element is not a group.
+ Propagate decision to groups.
+ (generate_one_element_ref): Handle RANGE_EXPR.
+ (mark_no_warning): Iterate over actual childs.
+ (generate_copy_inout): Likewise.
+ (generate_element_copy): Likewise.
+ (generate_element_zero): Likewise.
+ (generate_element_init_1): Likewise.
+ (dump_sra_elt_name): Handle RANGE_EXPR.
+
+2006-07-03 Roger Sayle <roger@eyesopen.com>
+
+ PR tree-optimization/26251
+ * tree-ssa-threadupdate.c (redirection_block_p): New function.
+ (thread_block): When optimizing for size refuse to thread jumps
+ that would require duplication of blocks other than redirection
+ blocks.
+
+2006-07-03 Paolo Bonzini <bonzini@gnu.org>
+
+ * configure.ac: Fix thinko in previous check-in.
+ * configure: Regenerate.
+
+2006-07-03 Paolo Bonzini <bonzini@gnu.org>
+
+ PR other/27063
+ * configure.ac (for lang...): Simplify nesting of conditionals.
+ Test subdir_requires for non-enabled languages. Move processing
+ of lang_opt_files and lang_tree_fiels below that test.
+ * configure: Regenerate.
+ * doc/sourcebuild.texi (Front End Config): Document subdir_requires.
+
+2006-07-01 Martin Michlmayr <tbm@cyrius.com>
+
+ * doc/gcov.texi (Invoking Gcov): Add a missing word.
+
+2006-06-30 Kazu Hirata <kazu@codesourcery.com>
+
+ * cfgexpand.c, config/i386/i386.c, genpreds.c, tree-cfg.c: Fix
+ comment typos.
+
+2006-06-29 Jason Merrill <jason@redhat.com>
+
+ PR c++/26905
+ PR c++/26612
+ PR c++/27000
+ PR c++/26984
+ PR c++/19134
+ * tree.c (build_decl_stat): Don't hande #pragma visibility here.
+ * c-common.c (c_determine_visibility): Handle it here.
+ * c-decl.c (finish_decl): Call c_determine_visibility for
+ functions, too.
+ * flags.h (enum symbol_visibility): Sort from most to least visibility.
+ * tree.h: Likewise.
+ * varasm.c (default_assemble_visibility): Likewise.
+ * c-common.c (handle_visibility_attribute): Complain about trying
+ to give visibility to an already defined class, or trying to change
+ declared visibility. Always attach the attribute.
+
+ PR c++/26905
+ PR c++/21675
+ PR c++/17470
+ * attribs.c (decl_attributes): Ignore type-in-place attributes
+ once the type has been defined.
+
+2006-06-29 Roger Sayle <roger@eyesopen.com>
+
+ PR middle-end/27428
+ * c-lex.c (c_lex_with_flags) <CPP_N_INVALID>: Increment errorcount
+ to indicate the cpplib has issued an error message for us.
+
+2006-06-29 Asher Langton <langton2@llnl.gov>
+
+ PR target/25765
+ * config/i386/i386.c (ix86_output_function_epilogue): Don't
+ insert a label at the end of an function under Mach-O.
+
+2006-06-29 Eric Christopher <echristo@apple.com>
+ Evan Cheng <evan.cheng@apple.com>
+
+ * predicates.md (vector_all_ones_operand): New.
+ (nonimmediate_or_sse_const_operand): Ditto.
+ * config/i386/i386-protos.h (standard_sse_constant_opcode): Declare.
+ * config/i386/i386.c (standard_sse_constant_opcode): Define.
+ (standard_sse_mode_p): Ditto.
+ (standard_sse_constant_p): Add case for -1 vector.
+ (ix86_expand_vector_move): Try to use.
+ * sse.md (*mov<mode>_internal): Use nonimmediate_or_sse_const_operand.
+ Call standard_sse_constant_opcode.
+ (*movv4sf_internal): Ditto.
+ (*movv2df_internal): Ditto.
+
+2006-06-29 Roger Sayle <roger@eyesopen.com>
+
+ * genpreds.c (write_match_code_switch): Correctly use XSTR instead
+ of XEXP to extract the operands of a MATCH_CODE rtx.
+
+2006-06-28 Andrew Pinski <pinskia@gmail.com>
+
+ * tree.c (tree_size): Do not waste tail padding in
+ struct tree_string and make the size be the same as
+ build_string will generate.
+
+2006-06-28 Jason Merrill <jason@redhat.com>
+
+ PR c++/27768
+ * tree-ssa-alias.c (compute_flow_insensitive_aliasing): Add
+ may_aliases already in the tag's annotations to the bitmap.
+
+2006-06-28 Roger Sayle <roger@eyesopen.com>
+
+ * genpreds.c (write_predicate_stmts) <IOR>: Add missing break.
+
+2006-06-28 Roger Sayle <roger@eyesopen.com>
+
+ * genpreds.c (generate_switch_p): New function.
+ (add_mode_tests): Push the new mode test down inside an AND expr
+ if this allows the switch-suitable MATCH_CODE to be near the root.
+ (write_match_code_switch): New function to write a MATCH_CODE as
+ a switch statement.
+ (write_predicate_stmts): New function to write a predicate RTX
+ expression as a sequence of statements.
+ (write_one_predicate_function): Use write_predicate_stmts.
+ (write_tm_constrs_h): Likewise.
+
+2006-06-26 Fred Fish <fnf@specifix.com>
+
+ * config/mips/t-sb1: New file.
+ * config.gcc (mipsisa64sb1-*-elf*): Add mips/t-sb1 to
+ tmake_file list.
+
+2006-06-26 DJ Delorie <dj@redhat.com>
+
+ * config/m32c/m32c.c (m32c_print_operand): Fix sign-merging logic.
+
+2006-06-26 Naveen H.S <naveenh@kpitcummins.com>
+ Jayant Sonar <jayants@kpitcummins.com>
+ Jaydeep Vipradas <jaydeepv@kpitcummins.com>
+
+ * config/m32c/addsub.md (addsi3, addsi3_1, addsi3_2): New.
+ (subsi3, subsi3_1, subsi3_2): New.
+ * config/m32c/bitops.md (andsi3, iorsi3, xorsi3): New.
+ * config/m32c/mov.md (SI mov peephole): New.
+ * config/m32c/m32.c (m32c_immd_dbl_mov): New.
+ * config/m32c/m32c-protos.h (m32c_immd_dbl_mov): New.
+
+2006-06-26 Olivier Hainque <hainque@adacore.com>
+
+ * function.c (aggregate_value_p): Honor DECL_BY_REFERENCE on
+ a CALL_EXPR target function declaration.
+
+2006-06-26 Richard Guenther <rguenther@suse.de>
+
+ * tree.c (build_string): Do not waste tail padding in
+ struct tree_string.
+
+2006-06-26 Richard Guenther <rguenther@suse.de>
+
+ * ggc-page.c (init_ggc): Add missing element to size_lookup
+ table.
+
+ Revert
+ 2006-06-23 Richard Guenther <rguenther@suse.de>
+
+ * ggc-page.c (init_ggc): Do not round up the extra_order_size_table
+ sizes to MAX_ALIGNMENT. Fix the size_lookup table to honour
+ alignment requests instead. Add verification code.
+ Add struct tree_function_decl and struct tree_binfo size to
+ extra_order_size_table. Add missing element to size_lookup
+ table.
+
+2006-06-25 Eric Botcazou <ebotcazou@adacore.com>
+
+ PR middle-end/28151
+ * fold-const.c (const_binop): Be prepared for self returning zero.
+ Simplify code handling complex values.
+
+2006-06-24 Olivier Hainque <hainque@adacore.com>
+
+ * gimplify.c (gimplify_scalar_mode_aggregate_compare): New function.
+ (gimplify_expr): Use it for tcc_comparison of operands of non BLKmode
+ aggregate types.
+
+2006-06-24 Olivier Hainque <hainque@adacore.com>
+
+ * tree-ssa-loop-im.c (for_each_index): Handle ARRAY_RANGE_REF as
+ ARRAY_REF, so have the callback called for the low bound expression.
+
+2006-06-23 Janis Johnson <janis187@us.ibm.com>
+
+ * tree.h (DECIMAL_FLOAT_TYPE_P): New.
+ * c-typeck.c (c_common_type): Disallow operations on decimal float
+ types and other float types.
+ * convert.c (convert_to_real): Don't ignore conversions involving
+ decimal float types.
+
+2006-06-23 Olivier Hainque <hainque@adacore.com>
+
+ * tree.c (max_int_size_in_bytes): New function, inspired from
+ code in function.c:assign_temp.
+ * tree.h (max_int_size_in_bytes): Declare.
+ * function.c (assign_temp): Use it.
+ * gimplify.c (create_tmp_var): Relax the assertions on the type
+ properties, not mandating constant size any more.
+ (force_constant_size): New static function.
+ (gimple_add_tmp_var): Use it, forcing variable size to a
+ constant upper bound if it is not constant on entry.
+
+2006-06-23 Danny Smith <dannysmith@users.sourceforge.net>
+
+ PR target/27789
+ * config/i386/winnt.c (ix86_handle_selectany_attribute): Move check
+ for initialization and setting of one_only flag to ...
+ (i386_pe_encode_section_info): ...here.
+ (i386_pe_dllimport_p): Check for DECL_DLLIMPORT_P also.
+ Recheck that the symbol has not been defined.
+
+2006-06-23 Richard Guenther <rguenther@suse.de>
+
+ * ggc-page.c (init_ggc): Do not round up the extra_order_size_table
+ sizes to MAX_ALIGNMENT. Fix the size_lookup table to honour
+ alignment requests instead. Add verification code.
+ Add struct tree_function_decl and struct tree_binfo size to
+ extra_order_size_table. Add missing element to size_lookup
+ table.
+
+2006-06-22 Roger Sayle <roger@eyesopen.com>
+ John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
+
+ PR middle-end/28131
+ * expr.c (expand_expr_real_1) <VECTOR_CST>: Check whether the
+ call to lang_hooks.types.type_for_mode returned NULL_TREE.
+
+2006-06-23 Ben Elliston <bje@au.ibm.com>
+
+ * dfp.c (decimal_to_decnumber): Do not use decNumberNegate to
+ negate dn -- manipulate it directly.
+
+2006-06-22 Roger Sayle <roger@eyesopen.com>
+ Steven Bosscher <stevenb.gcc@gmail.com>
+
+ PR target/27531
+ * reload1.c (gen_reload): Call mark_jump_label on the new insns
+ generated by gen_move_insn to add REG_LABEL notes if necessary.
+
+2006-06-22 Bob Wilson <bob.wilson@acm.org>
+
+ * config/xtensa/lib1funcs.asm (MIN_ESA): Delete.
+ (leaf_entry): Remove use of MIN_ESA.
+
+2006-06-22 Zdenek Dvorak <dvorakz@suse.cz>
+
+ PR rtl-optimization/28121
+ * cfgexpand.c (tree_expand_cfg): Clean EDGE_EXECUTABLE flag from
+ the entry edge.
+
+2006-06-21 DJ Delorie <dj@redhat.com>
+
+ * config/m32c/muldiv.md (mulhisi3_i): Only use registers for dest
+ (TN-16C-A156A/E).
+ (mulhisi3_r): Likewise.
+
+2006-06-21 Richard Sandiford <richard@codesourcery.com>
+
+ PR middle-end/28034
+ * coverage.c (coverage_counter_alloc): Leave the index type
+ unspecified.
+ (coverage_counter_alloc): Use null arguments for operands 2 and 3
+ of the ARRAY_REF.
+
+2006-06-21 Andrew Pinski <pinskia@physics.uc.edu>
+
+ * doc/invoke.texi (-fopenmp) Move to "Options Controlling C Dialect"
+ section from the "Options for Code Generation Conventions".
+
+2006-06-21 Richard Henderson <rth@redhat.com>
+
+ PR target/26347
+ PR target/27082
+ * config/alpha/predicates.md (small_symbolic_operand): Deny weak
+ symbols.
+ (global_symbolic_operand): Allow weak symbols, even if local_p.
+
+2006-06-21 Eric Botcazou <ebotcazou@adacore.com>
+
+ * tree-ssa-loop-ivopts.c (may_be_nonaddressable_p): New function.
+ (find_interesting_uses_address): Punt if above function returns true.
+
+2006-06-21 Richard Guenther <rguenther@suse.de>
+
+ * configure: Regenerated.
+
+2006-06-21 Joseph Myers <joseph@codesourcery.com>
+
+ * config/mips/mips.c (function_arg): Where one part of a
+ complex argument is in registers and the other part in the stack,
+ return a REG not a PARALLEL.
+
+2006-06-21 Mark Mitchell <mark@codesourcery.com>
+
+ * configure.ac: Set gcc_gxx_include_dir to $target/include/c++
+ when cross-compiling. Do not substitute libstdcxx_incdir.
+ * configure: Regenerated.
+
+2006-06-21 Jakub Jelinek <jakub@redhat.com>
+
+ * varasm.c (mergeable_string_section): Check for embedded NULs and
+ NUL termination in the first int_size_in_bytes (TREE_TYPE (decl))
+ rather than TREE_STRING_LENGTH bytes.
+
+2006-06-21 Richard Guenther <rguenther@suse.de>
+
+ * ggc-page.c (extra_order_size_table): Add var_ann_d,
+ tree_ssa_name, bitmap_element and phi_node with 4 arguments
+ sizes.
+
+2006-06-21 Richrad Guenther <rguenther@suse.de>
+
+ * tree-ssa-loop-niter.c (simplify_using_initial_conditions):
+ Limit iteration over the dominators.
+
+2006-06-20 Roger Sayle <roger@eyesopen.com>
+
+ * config/mips/iris6.h (LIB_SPEC): Add support for -pthread.
+ (SUBTARGET_CPP_SPEC): Define _REENTERANT if -pthread specified.
+
+2006-06-20 Richard Guenther <rguenther@suse.de>
+ Michael Matz <matz@suse.de>
+
+ * alloc-pool.h (free_alloc_pool_if_empty): Prototype new
+ function.
+ * alloc-pool.c (free_alloc_pool_if_empty): New function.
+ * et-forest.h (et_free_pools): Prototype new function.
+ * et-forest.c (et_free_tree_force): Free parent occurrence.
+ (et_free_pools): New function.
+ * dominance.c (free_dominance_info): Free et-forest alloc
+ pools.
+
+2006-06-20 Roger Sayle <roger@eyesopen.com>
+
+ * expr.c (expand_expr_real_1) <VECTOR_CST>: For vector constants with
+ integer modes, attempt to directly construct an integer constant.
+
+ * fold-const.c (native_encode_vector): Determine the size of each
+ element from the vector type instead of the first vector element.
+
+ * tree.c (build_constructor_single): Mark a CONSTRUCTOR as constant,
+ if all of its elements/components are constant.
+ (build_constructor_from_list): Likewise.
+
+2006-06-20 Zdenek Dvorak <dvorakz@suse.cz>
+
+ PR tree-optimization/27331
+ * tree-data-ref.c (free_data_ref): New function.
+ (create_data_ref): Fail if the data reference has unknown access
+ function.
+ (free_data_refs): Use free_data_ref.
+
+2006-06-19 Andrew Pinski <pinskia@gmail.com>
+
+ PR middle-end/28075
+ * tree-inline.c (setup_one_parameter): Strip useless
+ type conversion before adding it to the IR.
+ (declare_return_variable): Likewise.
+
+2006-06-19 Martin Michlmayr <tbm@cyrius.com>
+
+ PR c/27149
+ * c-common.c (c_common_truthvalue_conversion): Fix grammar in warning.
+
+2006-06-19 Mike Stump <mrs@apple.com>
+
+ * tree.c (variably_modified_type_p): VM arguments don't make the
+ type a VM type.
+
+2006-06-19 Richard Guenther <rguenther@suse.de>
+
+ * tree-ssa-pre.c (try_combine_conversion): Strip useless
+ type conversions after folding.
+
+2006-06-19 Richard Guenther <rguenther@suse.de>
+
+ * ggc-page.c (extra_order_size_tab): Add entries with
+ size of struct function and basic_block.
+ (size_lookup): Extend to cover sizes up to 511.
+ (ggc_alloc_stat): Adjust user.
+ (ggc_pch_count_object): Likewise.
+ (ggc_pch_alloc_object): Likewise.
+ (ggc_pch_write_object): Likewise.
+
+2006-06-19 Zdenek Dvorak <dvorakz@suse.cz>
+
+ * tree-ssa-loop-niter.c (implies_ge_p): New function.
+ (derive_constant_upper_bound): Handle OP0 - CST in unsigned types
+ correctly.
+
+2006-06-19 Denis Chertykov <denisc@overta.ru>
+
+ * config/avr/libgcc.S : Correct my wrong previous commit.
+
+2006-06-19 Roger Sayle <roger@eyesopen.com>
+
+ PR target/27861
+ * expmed.c (expand_shift): On SHIFT_COUNT_TRUNCATED targets, we may
+ have stripped a SUBREG from the shift count, so we may need to
+ convert_to_mode back to the type's mode before calling make_tree.
+ Use new_amount instead of amount to avoid expanding a tree twice.
+
+2006-06-19 Richard Guenther <rguenther@suse.de>
+
+ PR middle-end/28045
+ * fold-const.c (operand_equal_p): Check if the argument types
+ have the same precision before stripping NOPs.
+
+2006-06-19 Daniel Berlin <dberlin@dberlin.org>
+
+ Fix PR tree-optimization/27341
+ * tree-cfg.c (gimplify_val): Call mark_new_vars_to_rename on the
+ statement we get.
+ * tree-complex.c (pass_lower_complex): Update SMT usage.
+
+2006-06-19 Andreas Krebbel <krebbel1@de.ibm.com>
+
+ * config/s390/s390.md ("doloop_si64", "doloop_si31", "doloop_di"):
+ Add a new alternative to the constraint strings. Add move of
+ operand 1 to 3 to the splitter definition.
+ ("doloop_si_long"): Second contraint alternative removed.
+
+2006-06-08 Anatoly Sokolov <aesok@post.ru>
+
+ * config/avr/avr.h (avr_have_movw_lpmx_p): Declare.
+ (TARGET_CPU_CPP_BUILTINS): Add __AVR_HAVE_MOVW__ and
+ __AVR_HAVE_LPMX__.
+ (AVR_HAVE_MOVW) Define.
+ (ASSEMBLER_DIALECT): Use AVR_HAVE_MOVW.
+ (ASM_SPEC): Add avr25.
+ (LINK_SPEC): Add avr25.
+ * config/avr/avr.c (avr_have_movw_lpmx_p): Add.
+ (base_arch_s): Add new member 'have_movw_lpmx'.
+ (avr_arch_types): Init 'have_movw_lpmx'.
+ (avr_mcu_types): Add 'avr25'. Move attiny13, attiny2313, attiny24,
+ attiny44, attiny84, attiny25, attiny45, attiny85, attiny261,
+ attiny461, attiny861 and at86rf401 in 'avr25' arhitecture.
+ (avr_override_options): Init 'avr_have_movw_lpmx_p'.
+ (output_movhi, output_movsisf, ashlsi3_out, avr_rtx_costs): Use
+ AVR_HAVE_MOVW.
+ (avr_file_start): Do not output '.arh' derective.
+ * config/avr/libgcc.S (mov_l): Use __AVR_HAVE_MOVW__.
+ (__do_copy_data): Use __AVR_HAVE_LPMX__.
+ * config/avr/avr.md (mcu_enhanced): Delete.
+ (mcu_have_movw): Define.
+ (negsi2, extendhisi2, zero_extendhisi2): Use 'mcu_have_movw'.
+ (call_insn, call_value_insn): Use 'call_insn' and __AVR_HAVE_MOVW__.
+ * config/avr/t-avr(MULTILIB_OPTIONS, MULTILIB_DIRNAMES): Add avr25.
+ (MULTILIB_MATCHES): Add attiny13, attiny2313, attiny24, attiny44,
+ attiny84, attiny25, attiny45, attiny85, attiny261, attiny461,
+ attiny861 and at86rf401 devices.
+
+2006-06-18 Jie Zhang <jie.zhang@analog.com>
+
+ * config/bfin/bfin.c (MAX_LOOP_LENGTH): Define to be 2042 instead
+ of 4096.
+
+2006-06-17 Karl Berry <karl@gnu.org>
+
+ * doc/cpp.texi, doc/cppinternals.texi, doc/gcc.texi,
+ doc/gccint.texi, doc/install.texi (@dircategory): Use
+ "Software development" instead of "Programming", following
+ the Free Software Directory.
+
+2006-06-17 Alexandre Oliva <aoliva@redhat.com>
+
+ * config/i386/i386.c (legitimate_tls_address)
+ <TARGET_GNU2_TLS>: Add tp after DTPOFF.
+ * config/i386/i386.md (*tls_dynamic_gnu2_combine_32): Adjust.
+ (*tls_dynamic_gnu2_combine_64): Likewise.
+
+2006-06-16 Richard Guenther <rguenther@suse.de>
+
+ PR middle-end/27116
+ * fold-const.c (negate_expr_p): Do not introduce undefined
+ overflow in negating INTEGER_CSTs.
+ (fold_negate_expr): Rename from negate_expr. Revert last
+ change for folding BIT_NOT_EXPR. Change semantics to
+ return NULL_TREE for non-simplified negations. Do not
+ strip type conversions and unify type handling.
+ (negate_expr): New function, wrap around fold_negate_expr
+ but ensure building a tree always. Strip type conversions
+ here, fold to result type.
+ (fold_unary): Use fold_negate_expr for folding NEGATE_EXPR.
+
+2006-06-16 Roger Sayle <roger@eyesopen.com>
+
+ PR middle-end/27802
+ * reg-stack.c (subst_stack_regs): Handle noreturn function calls
+ that (would) return their results in stack registers.
+
+2006-06-16 Michael Matz <matz@suse.de>
+ Richard Guenther <rguenther@suse.de>
+
+ * fold-const.c (fold_truthop): Only return new tree node if
+ we canonicalized something.
+
+2006-06-16 Richard Guenther <rguenther@suse.de>
+
+ * function.h (enum function_frequency): Move declaration
+ out of struct function.
+ (struct function): Move bool fields and the enum to the bitfield
+ section. Reorder elements to avoid padding. Remove unused
+ fields left over from the RTL inliner removal.
+ * function.c (free_after_compilation): Do not zero removed fields.
+ * Makefile.in (ipa-reference.o): Add $(FUNCTION_H) dependency.
+ (ipa-type-escape.o): Likewise.
+
+2006-06-16 Richard Guenther <rguenther@suse.de>
+
+ * cgraph.h (struct cgraph_local_info): Turn bool elements
+ into a bitfield.
+ (struct cgraph_node): Likewise.
+ (struct cgraph_varpool_node): Likewise.
+
+2006-06-15 Richard Guenther <rguenther@suse.de>
+
+ * fold-const.c (fold_truth_not_expr): Rename from
+ invert_truthvalue. Give it fold_* semantics to avoid
+ generating garbage.
+ (invert_truthvalue): New function. Wrapper around
+ fold_truth_not_expr.
+ (optimize_minmax_comparison): Avoid creating garbage.
+ (fold_unary): Use fold_truth_not_expr for folding
The results can be reproduced by building a compiler with
--enable-gather-detailed-mem-stats targetting x86-64
and compiling preprocessed combine.c or testcase from PR8632 with:
-fmem-report --param=ggc-min-heapsize=1024 --param=ggc-min-expand=1 -Ox -Q
The memory consumption summary appears in the dump after detailed listing
of the places they are allocated in. Peak memory consumption is actually
computed by looking for maximal value in {GC XXXX -> YYYY} report.
Your testing script.