[Valgrind-users] valgrind hangs my app on oldselect call

[Sergio I. <ser...@ya...>]

Hi all, 

first I'd like to say that valgrind is one of most
amazing tools I already used. Thanks for making it
such a great tool.

I got some problems when my system was upgraded to red
hat 8.0 and we migrated to gcc 3.2.2 + valgrind-1.9.5.
I have a multithread app and it hangs somewhere
calling oldselect... strace gave this:

fcntl(13, F_GETFL)                      = 0x801 (flags
O_WRONLY|O_NONBLOCK)
write(13, "MTclChannel FOPEN_MAX=16\nMTclCha"..., 57)
= 57
fcntl(13, F_GETFL)                      = 0x801 (flags
O_WRONLY|O_NONBLOCK)
fcntl(13, F_SETFL, O_WRONLY)            = 0
oldselect(19, [17 18], NULL, NULL, NULL


and, pstack generated this:

0x4018ccde: vgPlain_post_known_blocking_syscall +
0x20f (12, 4528d4c0, 0, 0, 0)
0x40176563: __select + 0x2d (12, 4528d4c0, 0, 0, 0, 0)
+ 1030
0x09b6a02f: _ZN11MTclChannel18shellCmdEventLoop_Ev +
0x12f (4528afa8, 9b69ee0)
0x09b69ef1: _ZN11MTclChannel21runShellCmdEventLoop_EPv
+ 0x11 (4528afa8, 0, 0, 0, 0, 0) + 18
0x4026b822: thread_wrapper + 0xad (4528b040, 0, 0, 0,
0, 0) + 789cc224
0x4016583e: do__apply_in_new_thread_bogusRA (4528afd0,
bfffc240, b0b0c60, 45287808, 452877d4, 45287798) + 40
0x09b69d23:
_ZN11MTclChannel4initEP10Tcl_InterpRN4_STL13basic_ostreamIcNS2_11char_traitsIcEEEES7_
+ 0x1c3 (b0b0c20, b0b0c60, b0b0c60, 4527c31c,
45269e44, 1) + 10

valgrind-1.0.4 on red hat 7.2 was working properly. 

Any ideas??

Sergio




__________________________________________________
Do you Yahoo!?
The New Yahoo! Search - Faster. Easier. Bingo
http://search.yahoo.com