pipe crash

Tue Feb 24 00:06:07 PST 2004

On Tue, Feb 17, 2004 at 05:37:29PM -0500, Brian Fundakowski Feldman wrote:
> This is semi-repeatable.  While doing a ports build, during the lib-dependency
> -check phase, ldconfig -r is called and hangs when the port is trying to 
> check for a library, e.g. I see the output:
> ===>   xmms-imms-1.0 depends on shared library: id3 - found
> ===>   xmms-imms-1.0 depends on shared library: vorbis.3 <hang>
> at which point that process is stuck in pipdwt.  I gdb'ed my kernel at this 
> point and got a backtrace:
> 
> (kgdb) bt
> #0  sched_switch (td=0xc2bbebd0) at ../../../kern/sched_4bsd.c:671
> #1  0xc04a76ab in mi_switch (flags=0xc2bbebd0) at ../../../kern/kern_synch.c:518
> #2  0xc04a6e27 in msleep (ident=0xc2803480, mtx=0xc28035d8, priority=0x14c, wmesg=0xc06306b5 "pipdwt", timo=0x0)
>     at ../../../kern/kern_synch.c:254
> #3  0xc04cc35a in pipe_direct_write (wpipe=0xc2803480, uio=0x0) at ../../../kern/sys_pipe.c:899
> #4  0xc04cc7db in pipe_write (fp=0xc2c97d8c, uio=0xd85cec80, active_cred=0xc316a200, flags=0x0, td=0xc2bbebd0)
>     at ../../../kern/sys_pipe.c:1009
> #5  0xc04c929c in dofilewrite (td=0xc2bbebd0, fp=0xc2c97d8c, fd=0x0, buf=0x804f000, nbyte=0x0, offset=0x0, flags=0x0)
>     at ../../../sys/file.h:249
> #6  0xc04c90ce in write (td=0xc2bbebd0, uap=0xd85ced14) at ../../../kern/sys_generic.c:334
> #7  0xc0603b62 in syscall (frame=
>       {tf_fs = 0x2f, tf_es = 0x2f, tf_ds = 0x2f, tf_edi = 0x804f000, tf_esi = 0x4000, tf_ebp = 0xbfbfe548, tf_isp = 0xd85ced74, tf_ebx = 0x281430cc, tf_edx = 0x28142f38, tf_ecx = 0x28142f38, tf_eax = 0x4, tf_trapno = 0x0, tf_err = 0x2, tf_eip = 0x280c980f, tf_cs = 0x1f, tf_eflags = 0x216, tf_esp = 0xbfbfe52c, tf_ss = 0x2f}) at ../../../i386/i386/trap.c:1008
> 
> Now, here's the weird part:
> 
> (kgdb) p *wpipe
> $1 = {pipe_buffer = {cnt = 0x0, in = 0x0, out = 0x0, size = 0x4000, buffer = 0x0}, pipe_map = {kva = 0xd8760000, 
>     cnt = 0x4000, pos = 0x0, npages = 0x4, ms = {0xc11b5030, 0xc1311c78, 0xc125dcc0, 0xc12504c0, 
>       0x0 <repeats 13 times>}}, pipe_sel = {si_thrlist = {tqe_next = 0x0, tqe_prev = 0x0}, si_thread = 0x0, si_note = {
>       slh_first = 0x0}, si_flags = 0x0}, pipe_atime = {tv_sec = 0x40328142, tv_nsec = 0x0}, pipe_mtime = {
>     tv_sec = 0x40328142, tv_nsec = 0x0}, pipe_ctime = {tv_sec = 0x40328142, tv_nsec = 0x0}, pipe_sigio = 0x0, 
>   pipe_peer = 0xc280352c, pipe_pair = 0xc2803480, pipe_state = 0xc00, pipe_busy = 0x1, pipe_present = 0x0}
I've got similar crash, when I tried to kill today sort that was part of
periodic.weekly script, and which was still hanging around. Afrter kill I got
page fault, and very similar backtrace:

(kgdb) bt
#0  doadump () at /usr/src/sys/kern/kern_shutdown.c:240
#1  0xc0492d13 in boot (howto=256) at /usr/src/sys/kern/kern_shutdown.c:374
#2  0xc04930d7 in __panic () at /usr/src/sys/kern/kern_shutdown.c:552
#3  0xc05a3d56 in trap_fatal (frame=0xced68b6c, eva=0) at /usr/src/sys/i386/i386/trap.c:821
#4  0xc05a39e2 in trap_pfault (frame=0xced68b6c, usermode=0, eva=0) at /usr/src/sys/i386/i386/trap.c:735
#5  0xc05a35a3 in trap (frame=
      {tf_fs = 24, tf_es = 16, tf_ds = 16, tf_edi = 0, tf_esi = -823435264, tf_ebp = -824800300, tf_isp = -824800360, tf_ebx = -1029943936, tf_edx = 4, tf_ecx = 2666, tf_eax = 823435264, tf_trapno = 12, tf_err = 2, tf_eip = -1067836450, tf_cs = 8, tf_eflags = 66070, tf_esp = -823435264, tf_ss = 10664}) at /usr/src/sys/i386/i386/trap.c:420
#6  0xc05a1bde in generic_bcopy () at {standard input}:221
#7  0xc04c179c in pipe_direct_write (wpipe=0xc29c4d80, uio=0x0) at /usr/src/sys/kern/sys_pipe.c:907
#8  0xc04c1b69 in pipe_write (fp=0xc2b1f550, uio=0xced68c80, active_cred=0xc29ed880, flags=0, td=0xc2b3f930)
    at /usr/src/sys/kern/sys_pipe.c:1007
#9  0xc04be4bb in dofilewrite (td=0xc2b3f930, fp=0xc2b1f550, fd=0, buf=0x885d000, nbyte=0, offset=0, flags=0) at /usr/src/sys/sys/file.h:249
#10 0xc04be2de in write (td=0xc2b3f930, uap=0xced68d14) at /usr/src/sys/kern/sys_generic.c:334
#11 0xc05a4090 in syscall (frame=
      {tf_fs = 47, tf_es = 47, tf_ds = 47, tf_edi = 142987264, tf_esi = 10664, tf_ebp = -1077941448, tf_isp = -824799884, tf_ebx = 1209327852, tf_edx = 1209327448, tf_ecx = 1209327448, tf_eax = 4, tf_trapno = 0, tf_err = 2, tf_eip = 1208820031, tf_cs = 31, tf_eflags = 530, tf_esp = -1077941476, tf_ss = 47}) at /usr/src/sys/i386/i386/trap.c:1010
#12 0x480d213f in ?? ()
---Can't read userspace from dump, or kernel process---

And:
(kgdb) p *wpipe
$1 = {pipe_buffer = {cnt = 10664, in = 10664, out = 0, size = 16384, buffer = 0x0}, pipe_map = {kva = 3471532032, cnt = 10664, pos = 0,
    npages = 3, ms = {0xc0fc45b0, 0xc0ff5df8, 0xc0f70dc0, 0xc0e9a008, 0x0 <repeats 13 times>}}, pipe_sel = {si_thrlist = {tqe_next = 0x0,
      tqe_prev = 0x0}, si_thread = 0x0, si_note = {slh_first = 0x0}, si_flags = 0}, pipe_atime = {tv_sec = 1077333432, tv_nsec = 0},
  pipe_mtime = {tv_sec = 1077333432, tv_nsec = 0}, pipe_ctime = {tv_sec = 1077333431, tv_nsec = 0}, pipe_sigio = 0x0,
  pipe_peer = 0xc29c4e2c, pipe_pair = 0xc29c4d80, pipe_state = 2304, pipe_busy = 1, pipe_present = 0}

Hope that helps anyone.

Cheers,

Wiktor Niesiobedzki