random hangs in cc1 during bootstrap

[Phil Edwards <phil at jaj dot com>]

Just during the last 15 hours or so I've been seeing stage1's cc1 hang, just
sitting and slowly chewing cpu time, during a bootstrap.  The backtrace at
one point looks like this:

#0  0x4008d9bb in memset () from /lib/libc.so.6
#1  0x082fb4aa in sbitmap_zero (bmap=0x41c0db34)
    at /home/pme/src/trunk/gcc/sbitmap.c:118
#2  0x082fb533 in sbitmap_vector_zero (bmap=0x41b9a008, n_vecs=6275)
    at /home/pme/src/trunk/gcc/sbitmap.c:147
#3  0x08376f56 in make_edges (label_value_list=0x0, min=0x8a23d88,
    max=0x8a23d88, update_p=1) at /home/pme/src/trunk/gcc/cfgbuild.c:299
#4  0x08377ebf in find_sub_basic_blocks (bb=0x8a23d88)
    at /home/pme/src/trunk/gcc/cfgbuild.c:844
#5  0x080e3be9 in commit_one_edge_insertion (e=0x8b727c0, watch_calls=0)
    at /home/pme/src/trunk/gcc/cfgrtl.c:1452
#6  0x080e3c3b in commit_edge_insertions ()
    at /home/pme/src/trunk/gcc/cfgrtl.c:1474
#7  0x081a1b4b in pre_gcse () at /home/pme/src/trunk/gcc/gcse.c:5457
#8  0x081a1c25 in one_pre_gcse_pass (pass=1)
    at /home/pme/src/trunk/gcc/gcse.c:5493
#9  0x08199c86 in gcse_main (f=0x40aa3b84, file=0x0)
    at /home/pme/src/trunk/gcc/gcse.c:828
#10 0x08321e76 in rest_of_compilation (decl=0x4067f1c0)
    at /home/pme/src/trunk/gcc/toplev.c:2874
#11 0x08072222 in c_expand_body (fndecl=0x4067f1c0, nested_p=0, can_defer_p=1)
    at /home/pme/src/trunk/gcc/c-decl.c:6541
#12 0x08071e06 in finish_function (nested=0, can_defer_p=1)
    at /home/pme/src/trunk/gcc/c-decl.c:6412
#13 0x0804a5d1 in yyparse () at c-parse.y:406
#14 0x0805be58 in c_common_parse_file (set_yydebug=0)
    at /home/pme/src/trunk/gcc/c-lex.c:165
#15 0x083205b7 in compile_file () at /home/pme/src/trunk/gcc/toplev.c:2135
#16 0x0832638b in do_compile () at /home/pme/src/trunk/gcc/toplev.c:5396
#17 0x083263e5 in toplev_main (argc=40, argv=0xbfffde34)
    at /home/pme/src/trunk/gcc/toplev.c:5426
#18 0x080c3b1a in main (argc=40, argv=0xbfffde34)
    at /home/pme/src/trunk/gcc/main.c:37

I don't see anything obviously related in the changelog, but I am completely
unfamiliar with this part of the compiler (more's the pity).

Phil

-- 
I would therefore like to posit that computing's central challenge, viz. "How
not to make a mess of it," has /not/ been met.
                                                 - Edsger Dijkstra, 1930-2002