cc1plus 3.4 segfault.

[Karel Gardas <kgardas at objectsecurity dot com>]

Hello,

>From time to time I try gcc3.4 to find out if it works well with MICO. It
seems that recent sources (including from yesterday) segfaults.

(gdb) set args -quiet -I../include -D_GNU_SOURCE -DFAST_PCH -DPIC ir.cc
-quiet -dumpbase ir.cc -mtune=pentiumpro -auxbase-strip ir.pic.o -Wall
-fpermissive -fPIC -o /tmp/ccLF2GS9.s

^ I obtained this command-line from ps ux|grep cc1 while compiling
problematic source... I hope it's correct to enter it in such form into
gdb and start the cc1plus compiler...


(gdb) r
Starting program:
/home/karel/usr/local/gcc-main/lib/gcc-lib/i686-pc-linux-gnu/3.4/cc1plus
-quiet -I../include -D_GNU_SOURCE -DFAST_PCH -DPIC ir.cc -quiet -dumpbase
ir.cc -mtune=pentiumpro -auxbase-strip ir.pic.o -Wall -fpermissive -fPIC
-o /tmp/ccLF2GS9.s

Program received signal SIGSEGV, Segmentation fault.
ggc_set_mark (p=0x1) at /home/karel/cvs/gcc/gcc-main/gcc/ggc-page.c:1184
1184      if (entry->in_use_p[word] & mask)
(gdb) bt
#0  ggc_set_mark (p=0x1) at /home/karel/cvs/gcc/gcc-main/gcc/ggc-page.c:1184
#1  0x0806b7ba in gt_ggc_mx_lang_tree_node (x_p=0x85011a9) at gtype-cp.h:97
#2  0x0806bce3 in gt_ggc_mx_lang_tree_node (x_p=0x0) at gtype-cp.h:253
#3  0x0806bbcb in gt_ggc_mx_lang_tree_node (x_p=0x4c93c114) at gtype-cp.h:207
#4  0x08189492 in ggc_mark_roots () at /home/karel/cvs/gcc/gcc-main/gcc/ggc-common.c:105
#5  0x082dc562 in ggc_collect () at /home/karel/cvs/gcc/gcc-main/gcc/ggc-page.c:1756
#6  0x082bfec1 in rest_of_compilation (decl=0x4b4ad760) at /home/karel/cvs/gcc/gcc-main/gcc/toplev.c:3684
#7  0x080c49ff in genrtl_finish_function (fn=0x4b4ad760) at /home/karel/cvs/gcc/gcc-main/gcc/cp/semantics.c:2672
#8  0x080c471c in expand_body (fn=0x4b4ad760) at /home/karel/cvs/gcc/gcc-main/gcc/cp/semantics.c:2496
#9  0x080c99c4 in maybe_clone_body (fn=0x4b4ac320) at /home/karel/cvs/gcc/gcc-main/gcc/cp/optimize.c:268
#10 0x080c458f in expand_body (fn=0x4b4ac320) at /home/karel/cvs/gcc/gcc-main/gcc/cp/semantics.c:2414
#11 0x080a4d3a in cp_parser_function_definition_after_declarator (parser=0x40223580, inline_p=32)
    at /home/karel/cvs/gcc/gcc-main/gcc/cp/parser.c:13853
#12 0x080a4cb2 in cp_parser_function_definition_from_specifiers_and_declarator (parser=0x40223580, decl_specifiers=0x10000000,
    attributes=0x10000000, declarator=0x10000000) at /home/karel/cvs/gcc/gcc-main/gcc/cp/parser.c:13792
#13 0x080a10d0 in cp_parser_init_declarator (parser=0x40223580, decl_specifiers=0x0, prefix_attributes=0x0,
    function_definition_allowed_p=true, member_p=false, function_definition_p=0xbffff4d7)
    at /home/karel/cvs/gcc/gcc-main/gcc/cp/parser.c:9610
#14 0x0809e5de in cp_parser_simple_declaration (parser=0x40223580, function_definition_allowed_p=true)
    at /home/karel/cvs/gcc/gcc-main/gcc/cp/parser.c:6582
#15 0x0809e4b4 in cp_parser_block_declaration (parser=0x40223580, statement_p=false)
    at /home/karel/cvs/gcc/gcc-main/gcc/cp/parser.c:6508
#16 0x0809e36d in cp_parser_declaration (parser=0x0) at /home/karel/cvs/gcc/gcc-main/gcc/cp/parser.c:6428
#17 0x0809e218 in cp_parser_declaration_seq_opt (parser=0x0) at /home/karel/cvs/gcc/gcc-main/gcc/cp/parser.c:6337
#18 0x0809a4d2 in cp_parser_translation_unit (parser=0x40223580) at /home/karel/cvs/gcc/gcc-main/gcc/cp/parser.c:2256
#19 0x080a5c12 in yyparse () at /home/karel/cvs/gcc/gcc-main/gcc/cp/parser.c:14671
#20 0x080e75e6 in c_common_parse_file (set_yydebug=268435456) at /home/karel/cvs/gcc/gcc-main/gcc/c-opts.c:1394
#21 0x082bda2a in compile_file () at /home/karel/cvs/gcc/gcc-main/gcc/toplev.c:2223
#22 0x082c35f8 in do_compile () at /home/karel/cvs/gcc/gcc-main/gcc/toplev.c:5753
#23 0x082c3669 in toplev_main (argc=268435456, argv=0x0) at /home/karel/cvs/gcc/gcc-main/gcc/toplev.c:5782
#24 0x080eb40b in main (argc=268435456, argv=0x10000000) at /home/karel/cvs/gcc/gcc-main/gcc/main.c:37
(gdb)


Is this well known bug, or should I enter bugzilla report including
preprocessed sources?

Thanks,

Karel
--
Karel Gardas                  kgardas@objectsecurity.com
ObjectSecurity Ltd.           http://www.objectsecurity.com