Possible bug in pthread_sighandler?

[Matt Snelham]

Hello all,

I have a small tcl program (run perhaps 10,000 times a day) which randomly (an 
average of once every
~2 days or so) looses it, grows out of control and forces the machine into an 
OOM, and unuseable,
condition. It is running on ~100 workstations (Dual P4 Xeon, 2GB RAM, 2.4.3 and 
2.4.9 kernels, RH 7.1
with glibc 2.2.4-19), and so far seems entirely unpredictable.

Below is a trace, where it's looks pretty clear that pthreads_sighandler gets 
caught in an
infinitely recursive loop, filling up memory with the frames it's pushing onto 
the stack.

Is this an application bug, or something in pthreads?

Thanks for help,
--Matthew
address@hidden

--------------------------------------------------------------------------------

(gdb) where
#1 pthread_sighandler (signo=11, ctx={gs = 0, __gsh = 0, fs = 0, __fsh = 0, es 
= 43, __esh = 0, ds = 43,
__dsh = 0, edi = 1098342152, esi = 0, ebp = 3221218700, esp = 3221218676, ebx = 
1098352600, edx = 11,
ecx = 1, eax = 0, trapno = 13, err = 0, eip = 1098308137, cs = 35, __csh = 0, 
eflags = 66118,
esp_at_signal = 3221218676, ss = 43, __ssh = 0, fpstate = 0x0, oldmask = 
2147483648, cr2 = 0}) at signals.c:87
#2 <signal handler called>

<Thousands of alternating pthread_sighandler/signal handler calls deleted here>

#22054 pthread_sighandler (signo=11, ctx={gs = 0, __gsh = 0, fs = 0, __fsh = 0, 
es = 43, __esh = 0, ds = 43,
      __dsh = 0, edi = 1098342152, esi = 0, ebp = 3221218700, esp = 3221218676, 
ebx = 1098352600, edx = 11,
      ecx = 1, eax = 0, trapno = 13, err = 0, eip = 1098308137, cs = 35, __csh 
= 0, eflags = 66118,
      esp_at_signal = 3221218676, ss = 43, __ssh = 0, fpstate = 0x0, oldmask = 
2147483648, cr2 = 0}) at signals.c:87
#22055 <signal handler called>
#22056 pthread_sighandler (signo=11, ctx={gs = 0, __gsh = 0, fs = 0, __fsh = 0, 
es = 43, __esh = 0, ds = 43,
      __dsh = 0, edi = 1098342152, esi = 0, ebp = 3221219480, esp = 3221219440, 
ebx = 1098352600, edx = 1, ecx = 1,
      eax = 0, trapno = 13, err = 0, eip = 1098303892, cs = 35, __csh = 0, 
eflags = 66070,
      esp_at_signal = 3221219440, ss = 43, __ssh = 0, fpstate = 0xbfffe5f0, 
oldmask = 2147483648, cr2 = 0})
    at signals.c:87
#22057 <signal handler called>
#22058 __pthread_reset_main_thread () at internals.h:377
#22059 0x4176c0a5 in __vfork () at ptfork.c:92
#22060 0x415b8a8c in TclpCreateProcess () at eval.c:41
#22061 0x415a9557 in TclCreatePipeline () at eval.c:41
#22062 0x415a9885 in Tcl_OpenCommandChannel () at eval.c:41
#22063 0x4159f08a in Tcl_ExecCmd () at eval.c:41
#22064 0x4156e9e6 in TclInvokeStringCommand () at eval.c:41
#22065 0x4158b815 in TclExecuteByteCode () at eval.c:41
#22066 0x4156f3b4 in Tcl_EvalObj () at eval.c:41
#22067 0x415a0358 in Tcl_EvalFile () at eval.c:41
#22068 0x415a288a in Tcl_Main () at eval.c:41
#22069 0x8055994 in main () at eval.c:41
#22070 0x41853627 in __libc_start_main (main=0x8055920 <main>, argc=3, 
ubp_av=0xbffff884, init=0x80547f0 <_init>,
    fini=0x806fd40 <_fini>, rtld_fini=0x4000dcd4 <_dl_fini>, 
stack_end=0xbffff87c)
    at ../sysdeps/generic/libc-start.c:129
--------------------------------------------------------------------------------