Some aspect of GCC memory consumption increased by recent patch

[gcctest at suse dot de]

Hi,
Comparing memory consumption on compilation of combine.i and generate-3.4.ii I got:


comparing combine.c compilation at -O0 level:
Amount of memory still referenced at the end of compilation decreased from 7372k to 6107k, overall -20.70%
Overall memory needed: 0k -> 2666k
Peak memory use before GGC: 9500k -> 9294k
Peak memory use after GGC: 8812k -> 8606k
Maximum of released memory in single GGC run: 2867k
Garbage: 42489k -> 42475k
Leak: 7372k -> 6107k
Overhead: 5816k -> 5590k
GGC runs: 360 -> 363

comparing combine.c compilation at -O1 level:
Amount of memory still referenced at the end of compilation decreased from 7181k to 6483k, overall -10.76%
Overall memory needed: 0k -> 2666k
Peak memory use before GGC: 9702k -> 9573k
Peak memory use after GGC: 8823k -> 8663k
Maximum of released memory in single GGC run: 2069k -> 2067k
Garbage: 79893k -> 78747k
Leak: 7181k -> 6483k
Overhead: 14001k -> 13868k
GGC runs: 588 -> 589

comparing combine.c compilation at -O2 level:
Peak amount of GGC memory allocated before garbage collecting run decreased from 13312k to 12756k, overall -4.36%
Peak amount of GGC memory still allocated after garbage collecting decreased from 13165k to 12610k, overall -4.40%
Amount of memory still referenced at the end of compilation decreased from 6999k to 6304k, overall -11.02%
Overall memory needed: 0k -> 2666k
Peak memory use before GGC: 13312k -> 12756k
Peak memory use after GGC: 13165k -> 12610k
Maximum of released memory in single GGC run: 2577k -> 2576k
Garbage: 95880k -> 94713k
Leak: 6999k -> 6304k
Overhead: 18830k -> 18777k
GGC runs: 575 -> 580

comparing combine.c compilation at -O3 level:
Peak amount of GGC memory allocated before garbage collecting run decreased from 13850k to 13246k, overall -4.56%
Peak amount of GGC memory still allocated after garbage collecting decreased from 13165k to 12610k, overall -4.40%
Amount of memory still referenced at the end of compilation decreased from 7565k to 6852k, overall -10.42%
Overall memory needed: 0k -> 2666k
Peak memory use before GGC: 13850k -> 13246k
Peak memory use after GGC: 13165k -> 12610k
Maximum of released memory in single GGC run: 3484k -> 3483k
Garbage: 127651k -> 126026k
Leak: 7565k -> 6852k
Overhead: 24742k -> 24652k
GGC runs: 642 -> 646
comparing insn-attrtab.c compilation at -O0 level:
Reference run miss the results so I can't compare, sadly
Overall memory needed: 0k -> 2666k
Peak memory use before GGC: 0k -> 76388k
Peak memory use after GGC: 0k -> 45185k
Maximum of released memory in single GGC run: 0k -> 41417k
Garbage: 0k -> 157790k
Leak: 0k -> 10620k
Overhead: 0k -> 19800k
GGC runs: 0 -> 310

comparing insn-attrtab.c compilation at -O1 level:
Reference run miss the results so I can't compare, sadly
Overall memory needed: 0k -> 2666k
Peak memory use before GGC: 0k -> 94307k
Peak memory use after GGC: 0k -> 71401k
Maximum of released memory in single GGC run: 0k -> 40513k
Garbage: 0k -> 474239k
Leak: 0k -> 10968k
Overhead: 0k -> 84931k
GGC runs: 0 -> 461

comparing insn-attrtab.c compilation at -O2 level:
Reference run miss the results so I can't compare, sadly
Overall memory needed: 0k -> 2666k
Peak memory use before GGC: 0k -> 109880k
Peak memory use after GGC: 0k -> 86974k
Maximum of released memory in single GGC run: 0k -> 35488k
Garbage: 0k -> 525180k
Leak: 0k -> 11150k
Overhead: 0k -> 95096k
GGC runs: 0 -> 383

comparing insn-attrtab.c compilation at -O3 level:
Reference run miss the results so I can't compare, sadly
Overall memory needed: 0k -> 2666k
Peak memory use before GGC: 0k -> 109882k
Peak memory use after GGC: 0k -> 86975k
Maximum of released memory in single GGC run: 0k -> 35488k
Garbage: 0k -> 527525k
Leak: 0k -> 11223k
Overhead: 0k -> 95873k
GGC runs: 0 -> 392

comparing Gerald's testcase PR8361 compilation at -O0 level:
Peak amount of GGC memory still allocated after garbage collecting decreased from 91694k to 88883k, overall -3.16%
Amount of memory still referenced at the end of compilation decreased from 61513k to 57972k, overall -6.11%
Overall memory needed: 0k -> 2666k
Peak memory use before GGC: 93183k -> 90409k
Peak memory use after GGC: 91694k -> 88883k
Maximum of released memory in single GGC run: 20289k -> 20325k
Garbage: 267691k -> 267696k
Leak: 61513k -> 57972k
Overhead: 35037k -> 34596k
GGC runs: 548 -> 553

comparing Gerald's testcase PR8361 compilation at -O1 level:
Peak amount of GGC memory allocated before garbage collecting run decreased from 97174k to 88921k, overall -9.28%
Amount of produced GGC garbage increased from 545290k to 671734k, overall 23.19%
Amount of memory still referenced at the end of compilation decreased from 63318k to 60153k, overall -5.26%
Overall memory needed: 0k -> 2666k
Peak memory use before GGC: 97174k -> 88921k
Peak memory use after GGC: 90561k -> 88039k
Maximum of released memory in single GGC run: 19488k -> 19486k
Garbage: 545290k -> 671734k
Leak: 63318k -> 60153k
Overhead: 75450k -> 144644k
GGC runs: 763 -> 841

comparing Gerald's testcase PR8361 compilation at -O2 level:
Peak amount of GGC memory allocated before garbage collecting run decreased from 97175k to 88922k, overall -9.28%
Amount of produced GGC garbage increased from 600955k to 726419k, overall 20.88%
Amount of memory still referenced at the end of compilation decreased from 63940k to 60762k, overall -5.23%
Overall memory needed: 0k -> 2666k
Peak memory use before GGC: 97175k -> 88922k
Peak memory use after GGC: 90562k -> 88039k
Maximum of released memory in single GGC run: 19489k -> 19487k
Garbage: 600955k -> 726419k
Leak: 63940k -> 60762k
Overhead: 93911k -> 166904k
GGC runs: 821 -> 876

comparing Gerald's testcase PR8361 compilation at -O3 level:
Amount of produced GGC garbage increased from 616982k to 770533k, overall 24.89%
Amount of memory still referenced at the end of compilation decreased from 64593k to 61369k, overall -5.25%
Overall memory needed: 0k -> 2666k
Peak memory use before GGC: 92864k -> 90120k
Peak memory use after GGC: 91396k -> 89108k
Maximum of released memory in single GGC run: 20194k -> 19982k
Garbage: 616982k -> 770533k
Leak: 64593k -> 61369k
Overhead: 95868k -> 181353k
GGC runs: 803 -> 866

Head of changelog is:

--- /usr/src/SpecTests/sandbox-britten-memory/x86_64/mem-result/ChangeLog	2004-09-10 08:34:00.000000000 +0000
+++ /usr/src/SpecTests/sandbox-britten-memory/gcc/gcc/ChangeLog	2004-09-11 22:14:13.000000000 +0000
@@ -1,3 +1,318 @@
+2004-09-11  Geoffrey Keating  <geoffk@apple.com>
+
+	* config/darwin.c (machopic_validate_stub_or_non_lazy_ptr): Mark
+	the thing referenced as used, not the stub itself.
+
+2004-09-11  Joseph S. Myers  <jsm@polyomino.org.uk>
+
+	* c-tree.h (enum c_storage_class): New.
+	(struct c_declspecs): Add storage_class, inline_p and thread_p.
+	* c-decl.c (shadow_tag_warned): Give errors for "inline" in empty
+	declarations and "auto" or "register" in file scope empty
+	declarations.  Give more specific warnings for other cases of
+	storage class specifiers in empty declarations.
+	(grokdeclarator): Update for new structures.  Don't check for
+	multiple storage classes.  Diagnose file-scope "register" if
+	pedantic.
+	(build_null_declspecs): Update.
+	(declspecs_add_scspec): Update.  Diagnose multiple storage class
+	specifiers and invalid uses of "__thread".
+
+2004-09-11  Zack Weinberg  <zack@codesourcery.com>
+
+	* tree.c (tree_code_size): New function, bulk of code from tree_size.
+	(tree_size, make_node): Use it.
+	* tree-ssa-pre.c (init_pre): Use it.
+	* tree.h: Prototype it.
+
+2004-09-11  Zdenek Dvorak  <rakdver@atrey.karlin.mff.cuni.cz>
+
+	* tree-cfg.c (set_bb_for_stmt): Use PHI_BB.
+	* tree-dfa.c (compute_immediate_uses, add_immediate_use,
+	redirect_immediate_use): Use PHI_DF.
+	* tree-flow-inline.h (stmt_ann): Abort on phi nodes.
+	(bb_for_stmt): Use PHI_BB.
+	(get_immediate_uses): Use PHI_DF.
+	* tree-ssa-dse.c (max_stmt_uid): New variable.
+	(get_stmt_uid): New function.
+	(dse_optimize_stmt, dse_record_phis, tree_ssa_dse): Do not use phi
+	node annotations.
+	* tree-ssa-loop-im.c (LIM_DATA): Do not use phi statement annotations.
+	(max_uid): Renamed to max_stmt_uid.
+	(get_stmt_uid): New function.
+	(maybe_queue_var, single_reachable_address, determine_lsm): Do not use
+	phi node annotations.
+	* tree-ssa.c (replace_immediate_uses): Do not use phi node annotations.
+	* tree.h (PHI_BB, PHI_DF): New accessor functions.
+	(struct tree_phi_node): Add bb and df fields.
+
+2004-09-11  Richard Henderson  <rth@redhat.com>
+
+	PR middle-end/17416
+	* cfgexpand.c (expand_used_vars_for_block): Fix off-by-one bug
+	in conflict loop.
+
+2004-09-11  Jan Hubicka  <jh@suse.cz>
+
+	* i386.h (MODE_NEEDED): Calls need UNINITIALIZED mode; add comment.
+	(EMIT_MODE_SET): UNINITIALIZED mode needs no set.
+	* i386.md (i387_cw): Add 'uninitialized' mode.
+
+	* profile.c (BB_TO_GCOV_INDEX): move more to the front.
+	(output_location): Break out from ....
+	(branch_prob): ... here; handle gcov output on trees.
+	* toplev.c (process_options): No longer be sorry about coverage for
+	trees.
+
+2004-09-11  Steven Bosscher  <stevenb@suse.de>
+
+	PR c++/17412
+	* fold-const.c (fold): Do not try to fold the operand of a
+	CLEANUP_POINT_EXPR if that operand does itself not have any
+	operands.
+
+	PR middle-end/17417
+	* langhooks.c (lhd_decl_printable_name): Make sure that this
+	function is called with is a decl node that has an identifier.
+	* tree-pretty-print.c (dump_function_name): New function to
+	wrap PRINT_FUNCTION_NAME and dump_decl_name.
+	(print_call_name): Use it.
+
+2004-09-11  Mohan Embar  <gnustuff@thisiscool.com>
+
+	* ggc-none.c: Include "bconfig.h" if -DGENERATOR_FILE,
+	"config.h" if not.
+
+2004-09-11  Jakub Jelinek  <jakub@redhat.com>
+
+	* tree-ssa-alias.c (collect_points_to_info_r): Handle
+	pointing to PTR + OFFSET as PTR, not as pointing to PTR or pointing
+	to OFFSET.
+
+2004-09-11  Richard Henderson  <rth@redhat.com>
+
+	* expr.c (expand_expr_addr_expr_1) Rename from expand_expr_addr_expr.
+	(expand_expr_addr_expr): New function.  Take ADDR_EXPR instead of the
+	operand of the ADDR_EXPR.
+	(expand_expr_real_1): Update to match.
+
+2004-09-11  David Edelsohn  <edelsohn@gnu.org>
+
+	PR target/17277
+	* config/rs6000/aix.h (MD_FROB_UPDATE_CONTEXT): Use __64BIT__ to
+	choose 64-bit version.
+
+2004-09-11  Richard Henderson  <rth@redhat.com>
+
+	PR c/17396
+	* c-typeck.c (build_unary_op): Add legacy offsetof hack.
+
+2004-09-11  Kazu Hirata  <kazu@cs.umass.edu>
+
+	* doc/cfg.texi, doc/hostconfig.texi, doc/install.texi,
+	doc/invoke.texi, doc/md.texi, doc/tm.texi, doc/tree-ssa.texi:
+	Insert @: after i.e. or e.g.
+
+2004-09-10  Geoffrey Keating  <geoffk@apple.com>
+
+	* tree-pretty-print.c (dump_decl_name): Print unnamed decls as
+	D.xxx rather than <Dxxx>.
+
+2004-09-10  Adam Nemet  <anemet@lnxw.com>
+
+	* configure.ac (AC_MSG_CHECKING(what linker to use)):
+	in_tree_ld_is_elf should be true for LynxOS.
+	* configure: Regenerate.
+
+2004-09-10  Adam Nemet  <anemet@lnxw.com>
+
+	* Makefile.in (all.cross): Add srcextra.
+
+2004-09-10  Roger Sayle  <roger@eyesopen.com>
+
+	PR middle-end/17024
+	* fold-const.c (fold_cond_expr_with_comparison): Handle unordered
+	floating point comparison operators.  Change aborts to gcc_asserts.
+
+2004-09-10  Geoffrey Keating  <geoffk@apple.com>
+
+	* final.c (output_asm_insn): Correct problem with -fverbose-asm.
+
+2004-09-10  James E Wilson  <wilson@specifixinc.com>
+
+	* Makefile.in (LIBGCC_DEPS): Delete duplicate stmp-int-hdrs reference.
+	(stmp-fixproto): Change stmp-int-headers to stmp-int-hdrs in comment.
+
+	* convert.c (convert_to_vector): Replace recursive call with
+	error_mark_node.
+
+	* emit-rtl.c (gen_const_vector): Renamed from gen_const_vector_0.
+	Add integer argument named constant.  Use const_tiny_rtx instead of
+	CONST0_RTX.
+	(gen_rtx_CONST_VECTOR): Rewrite to handle checks for both CONST0_RTX
+	and CONST1_RTX.
+	(init_emit_once): Fix users of gen_const_vector.  Set CONST1_RTX for
+	vector types.
+	* expr.c (const_vector_from_tree): Call gen_rtx_CONST_VECTOR instead
+	of gen_rtx_raw_CONST_VECTOR.
+
+2004-09-10  Eric Christopher  <echristo@redhat.com>
+
+	PR c/16046
+	* c-parse.in (attribute): Add NULL_TREE for error case
+	for error recovery.
+
+2004-09-10  Zack Weinberg  <zack@codesourcery.com>
+
+	* c-common.h (struct c_fileinfo): Comment moved here from cp-tree.h.
+	(extract_interface_info): Delete prototype.
+	* c-decl.c (extract_interface_info): Delete stub.
+	* c-lex.c (get_fileinfo): Initialize file_info_tree here, if
+	it's not already been done ...
+	(init_c_lex): ... not here.
+	(fe_file_change): Don't call extract_interface_info.
+
+2004-09-11  Joseph S. Myers  <jsm@polyomino.org.uk>
+
+	* c-tree.h (struct c_declspecs): Remove typedef_decl.  Add
+	typedef_p and typedef_signed_p.
+	* c-decl.c (shadow_tag_warned): Check typedef_p, not typedef_decl.
+	(grokdeclarator): Don't use typedef_decl for warn_deprecated_use.
+	Check typedef_p and typedef_signed_p, not typedef_decl.
+	(grokfield): Check typedef_p, not typedef_decl.
+	(build_null_declspecs): Initialize typedef_p and typedef_signed_p,
+	not typedef_decl.
+	(declspecs_add_type): Set typedef_p and typedef_signed_p, not
+	typedef_decl.
+
+2004-09-10  Kazu Hirata  <kazu@cs.umass.edu>
+
+	* doc/tm.texi, doc/tree-ssa.texi: Fix typos.
+
+2004-09-10  Geoffrey Keating  <geoffk@apple.com>
+
+	* gimplify.c (create_tmp_var_raw): Don't name otherwise-nameless
+	temporary variables.
+
+	* dbxout.c (dbxout_block): Update for
+	current_function_func_begin_label a string.
+	* dwarf2out.c (dwarf2out_begin_prologue): Likewise.
+	* except.c (dw2_output_call_site_table): Likewise.
+	* toplev.c (current_function_func_begin_label): Likewise.
+	* tree.h (current_function_func_begin_label): Likewise.
+	* config/rs6000/linux64.h (DBX_OUTPUT_BRAC): Likewise.
+
+	* final.c (output_asm_insn): Use strtoul instead of atoi, save a
+	loop.
+
+	* config/darwin.c: Include target.h.
+	(struct machopic_indirection): Make ptr_name a string.
+	(machopic_indirection_hash): Update for ptr_name a string.
+	(machopic_indirection_eq): Likewise.
+	(machopic_indirection_name): Likewise.
+	(machopic_output_indirection): Likewise.
+	(machopic_validate_stub_or_non_lazy_ptr): Update for ptr_name a
+	string.  Don't expect stub names to be in the identifier hash table.
+	Do call strip_name_encoding before looking up entry in ID hash table.
+	* config/t-darwin (darwin.o): Add target.h to dependencies.
+
+	* gcc/config/rs6000/rs6000.c (print_operand): Use fputs instead
+	of fprintf for register names.
+
+	* output.h (unlikely_section_label): Delete declaration.
+	(unlikely_text_section_name): Likewise.
+	* varasm.c (unlikely_section_label_printed): Make static.
+	(unlikely_section_label): Likewise.
+	(unlikely_text_section_name): Likewise.
+	(assemble_start_function): Use reconcat.
+
+	* rtl.def (REG): Add comment to describe third field.
+
+2004-09-10  Andrew Pinski  <apinski@apple.com>
+
+	* config/darwin.c (machopic_mcount_stub_name): Call
+	machopic_indirection_name instead of making the string
+	ourselves.
+
+2004-09-10  Diego Novillo  <dnovillo@redhat.com>
+
+	Revert
+
+	2004-09-08  Jie Zhang  <zhangjie@magima.com.cn>
+
+	    * tree-ssa-alias.c (compute_flow_insensitive_aliasing): If type
+	    memory tag is call clobbered, so are its aliases.
+	    (group_aliases): When two memory tags being grouped, if one is
+	    call clobbered, so are the other and its aliases.
+	    (add_may_alias): Remove call-clobbering stuff.
+	    (replace_may_alias): Likewise.
+	    (merge_pointed_to_info): Merge pt_global_mem
+
+2004-09-10  Kazu Hirata  <kazu@cs.umass.edu>
+
+	* bb-reorder.c, c-common.c, c-incpath.c, c-typeck.c,
+	genrecog.c, lambda-code.c, mips-tdump.c, mips-tfile.c,
+	passes.c, tree-data-ref.c, tree-data-ref.h, tree-mudflap.c,
+	tree-scalar-evolution.c, tree-ssa-copyrename.c,
+	tree-ssa-live.c, tree-ssa-live.h: Fix comment typos.
+
+2004-09-10  Richard Sandiford  <rsandifo@redhat.com>
+
+	* config/frv/frv.md (UNSPEC_MASACCS, UNSPEC_MDASACCS): New constants.
+	(maddaccs, msubaccs, *maddacc): Turn operand 0 into a single ACC/ACCG
+	pair.  Remove match_dup from inputs.  Remove masaccs handling.
+	(mdaddaccs, mdsubaccs, *mdaddacc): Likewise a dual ACC/ACCG pair.
+	(masaccs, mdasaccs): Turn into define_insns.
+
+2004-09-10  Richard Sandiford  <rsandifo@redhat.com>
+
+	* tree-tailcall.c (process_assignment): Only do accumulator transforms
+	for floating-point types if flag_unsafe_math_optimizations.
+
+2004-09-10  Kazu Hirata  <kazu@cs.umass.edu>
+
+	* config/darwin.c, config/alpha/alpha.h, config/arm/arm.c,
+	config/arm/arm.h, config/arm/arm.md, config/arm/bpabi.h,
+	config/arm/predicates.md, config/frv/frv.c, config/frv/frv.md,
+	config/h8300/h8300.md, config/i386/gmm_malloc.h,
+	config/ia64/ia64.md, config/ip2k/libgcc.S,
+	config/mips/mips-ps-3d.md, config/mips/mips.c,
+	config/rs6000/rs6000.c, config/s390/s390.c,
+	config/sh/symbian.c: Fix comment typos.
+
+2004-09-10  Andreas Krebbel  <krebbel1@de.ibm.com>
+
+	* config/s390/s390.c (s390_emit_prologue): Added missing
+	"{}" brackets.
+
+2004-09-10  Kazu Hirata  <kazu@cs.umass.edu>
+
+	* doc/extend.texi, doc/invoke.texi, doc/passes.texi: Fix
+	typos.  Follow spelling conventions.
+
+2004-09-10  Joseph S. Myers  <jsm@polyomino.org.uk>
+
+	* toplev.c (warn_deprecated_use): Correct logic for saying "type"
+	in diagnostic.  Don't dereference NULL TYPE_NAME.
+
+2004-09-10  Kazu Hirata  <kazu@cs.umass.edu>
+
+	* c-common.c, c-pch.c, defaults.h, lambda-code.c, passes.c,
+	tree-data-ref.c, tree-flow.h, tree-ssa-operands.c,
+	tree-ssa-threadupdate.c, tree-vectorizer.c, tree-vectorizer.h,
+	value-prof.c: Fix comment typos.  Follow spelling conventions.
+
+2004-09-10  Jan Hubicka  <jh@suse.cz>
+
+	* tree-ssa-dce.c (remove_dead_stmt): Update profile.
+	* tree-split_edge.c (tree_split_edge): Likewise.
+	(thread_jumps): Likewise.
+
+2004-09-10  Kazu Hirata  <kazu@cs.umass.edu>
+
+	* cgraphunit.c, predict.c, tree-ssa-loop-ivopts.c: Fix comment
+	typos.
+
 2004-09-10  Paolo Bonzini  <bonzini@gnu.org>
 
 	* cgraph.h (cgraph_dump_file): Do not declare.
--- /usr/src/SpecTests/sandbox-britten-memory/x86_64/mem-result/ChangeLog.cp	2004-09-10 08:34:01.000000000 +0000
+++ /usr/src/SpecTests/sandbox-britten-memory/gcc/gcc/cp/ChangeLog	2004-09-11 22:14:42.000000000 +0000
@@ -1,3 +1,56 @@
+2004-09-10  Zack Weinberg  <zack@codesourcery.com>
+
+	* cp-tree.h (interface_only, interface_unknown): Delete declarations;
+	comment explaining them moved to c-common.h.
+	* lex.c (interface_only, interface_unknown, extract_interface_info):
+	Delete definitions.
+	(cxx_finish): Don't reset interface_unknown.
+	(handle_pragma_interface): Don't set interface_only and
+	interface_unknown; just the like-named fields in finfo.
+	(handle_pragma_implementation): Adjust comment.
+	* decl2.c (cp_finish_file): Don't reset interface_only and
+	interface_unknown.
+	* method.c (synthesize_method): Don't reset interface_unknown or
+	call extract_interface_info.
+	* pt.c (pop_tinst_level): Don't call extract_interface_info.
+	* decl.c (start_cleanup_fn): Don't save or restore interface_only
+	and interface_unknown.
+	(make_rtl_for_nonlocal_decl): Call get_fileinfo on input_filename
+	and use the result instead of the interface_only/interface_unknown
+	globals.
+	(start_preparsed_function): Likewise.
+	* lex.c (cxx_make_type): Likewise.
+	* semantics.c (begin_class_definition): Likewise.
+	(expand_body): Don't call extract_interface_info.
+
+2004-09-10  Ziemowit Laski  <zlaski@apple.com>
+
+	* decl.c (objc_mark_locals_volatile): Make description of
+	routine more descriptive; only mark VAR_DECLs at each
+	binding level.
+
+2004-09-10  Richard Henderson  <rth@redhat.com>
+
+	PR c++/17386
+	* call.c (build_vfield_ref): Move...
+	* class.c (build_vfield_ref): ... here.  Convert datum to the
+	primary base containing the vptr.
+	(make_new_vtable): Simplify build_primary_vtable arguments.
+	(finish_struct_1): Do not duplicate TYPE_VFIELD.
+	* typeck.c (build_class_member_access_expr): Don't warn for
+	null object access to base fields.
+
+2004-09-10  Ziemowit Laski  <zlaski@apple.com>
+
+	* decl.c (objc_get_current_scope, objc_mark_locals_volatile):
+	New functions, to be called from ObjC++.
+
+2004-09-10  Kazu Hirata  <kazu@cs.umass.edu>
+
+	* class.c, cp-tree.h, decl.c, decl2.c, mangle.c,
+	name-lookup.h, parser.c, search.c, semantics.c, typeck2.c: Fix
+	comment typos.
+
 2004-09-09  Ziemowit Laski  <zlaski@apple.com>
 
 	* typeck.c (build_c_cast): Preserve the cast if casting

I am friendly script caring about memory consumption in GCC.  Please contact
jh@suse.cz if something is going wrong.

The results can be reproduced by building 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.

Yours testing script.