[Python-Dev] test_fork1 on SMP? (was Re: [Python Dev] test_fork1 failing --with-threads (for some people)...) (original) (raw)

Neil Schemenauer nascheme@enme.ucalgary.ca
Wed, 9 Aug 2000 14:11:04 -0600

On Mon, Jul 31, 2000 at 04:42:50AM -0400, Tim Peters wrote:

It's a baffler! AFAIK, nobody yet has thought of a way that a fork can screw up the state of the locks in the parent process (it must be easy to see how they can get screwed up in a child, because two of us already did ).

If I add Py_BEGIN_ALLOW_THREADS and Py_END_ALLOW_THREADS around fork() in posixmodule then the child is the process which always seems to hang. The child is hanging at:

#0 0x4006d58b in __sigsuspend (set=0xbf7ffac4) at ../sysdeps/unix/sysv/linux/sigsuspend.c:48 #1 0x4001f1a0 in pthread_cond_wait (cond=0x8264e1c, mutex=0x8264e28) at restart.h:49 #2 0x806f3c3 in PyThread_acquire_lock (lock=0x8264e18, waitflag=1) at thread_pthread.h:311 #3 0x80564a8 in PyEval_RestoreThread (tstate=0x8265a78) at ceval.c:178 #4 0x80bf274 in posix_fork (self=0x0, args=0x8226ccc) at ./posixmodule.c:1659 #5 0x8059460 in call_builtin (func=0x82380e0, arg=0x8226ccc, kw=0x0) at ceval.c:2376 #6 0x8059378 in PyEval_CallObjectWithKeywords (func=0x82380e0, arg=0x8226ccc, kw=0x0) at ceval.c:2344 #7 0x80584f2 in eval_code2 (co=0x8265e98, globals=0x822755c, locals=0x0, args=0x8226cd8, argcount=0, kws=0x0, kwcount=0, defs=0x0, defcount=0, owner=0x0) at ceval.c:1682 #8 0x805974b in call_function (func=0x8264ddc, arg=0x8226ccc, kw=0x0) at ceval.c:2498 #9 0x805936b in PyEval_CallObjectWithKeywords (func=0x8264ddc, arg=0x8226ccc, kw=0x0) at ceval.c:2342 #10 0x80af26a in t_bootstrap (boot_raw=0x8264e00) at ./threadmodule.c:199 #11 0x4001feca in pthread_start_thread (arg=0xbf7ffe60) at manager.c:213

Since there is only one thread in the child this should not be happening. Can someone explain this? I have tested this both a SMP Linux machine and a UP Linux machine.

Neil