lkml.org 
[lkml]   [2001]   [Apr]   [26]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectInquiry: RE: [BUG] threaded processes get stuck in rt_sigsuspend /fillonedir/exit_notify thread.
Date


Afternoon All-

Basically, I am sending out a pseudo 'help' and bug check. I found
this thread ([BUG] threaded processes get stuck in
rt_sigsuspend/fillonedir/exit_notify) from 9/11/2000. I helping write some
software, and I think I have ran into a bug similiar to this, first, a quick
description of what the program does:

the program runs, creates a pipe and then forks.

The parent
- immediately starts reading on the pipe, waiting for the child to say "OK"
- When it gets the OK, the parent exits cleanly.

The child
- calls setsid() to become a process group leader
- loads a shared library (ascf.so)
- calls pthread_sigmask(SIG_BLOCK, ...) for SIGINT, SIGTERM, SIGHUP
- starts a monitor thread which
* Proceeds to fork and exec the other server processes
* After it does this successfully, it writes "OK" to the pipe
* Proceeds to call wait() waiting to any of its children to die

The main thread continues on to:
- Starts up two more worker threads
- calls sigaction(..) to establish a signal handler function for SIGINT,
SIGTERM & SIGHUP
- calls pthread_sigmask(SIG_UNBLOCK, ...) for SIGINT, SIGTERM, SIGHUP
- calls pause() to wait for any incoming signals


Now, the problem is that the child threads/processes do not seem to be
returning the OK, instead, they seem to simply hang, here is some sample
strace output [snipped for brevity]:

main process:

read(4, "1860", 4096) = 4
read(4, "", 4096) = 0
close(4) = 0
munmap(0x4012e000, 4096) = 0
kill(1860, SIG_0) = -1 ESRCH (No such process)
pipe([4, 6]) = 0
fork() = 2048
close(6) = 0
read(4,

[it seems to be waiting for the response]

then:

first process:

old_mmap(NULL, 89056, PROT_READ|PROT_EXEC, MAP_PRIVATE, 4, 0) = 0x40314000
mprotect(0x40326000, 15328, PROT_NONE) = 0
old_mmap(0x40326000, 16384, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED, 4,
0x11000) = 0x40326000
close(4) = 0
mprotect(0x40314000, 73728, PROT_READ|PROT_WRITE) = 0
mprotect(0x40314000, 73728, PROT_READ|PROT_EXEC) = 0
rt_sigprocmask(SIG_BLOCK, [HUP INT TERM], NULL, 8) = 0
getrlimit(RLIMIT_STACK, {rlim_cur=8192*1024, rlim_max=2147483647}) = 0
setrlimit(RLIMIT_STACK, {rlim_cur=2044*1024, rlim_max=2147483647}) = 0
pipe([4, 7]) = 0
clone(child_stack=0x81a9234,
flags=CLONE_VM|CLONE_FS|CLONE_FILES|CLONE_SIGHAND|CLONE_PTRACE) = 2049
write(7, "\364\336\0@\5\0\0\0\0\0\0\0\0\0\0\0\0\336\0@\0\0\0\0\220"..., 148)
= 148
rt_sigprocmask(SIG_SETMASK, NULL, [HUP INT TERM CHLD RT_0 RT_5], 8) = 0
write(7, "`\177\24@\0\0\0\0\0\0\0\0\330&\6\10\0\0\0\0\3@\1\200\20"..., 148)
= 148
rt_sigprocmask(SIG_SETMASK, NULL, [HUP INT TERM CHLD RT_0 RT_5], 8) = 0
rt_sigsuspend([HUP INT TERM CHLD RT_5]


So I guess my question is twofold, Has anyone else seen this on the
most recent kernels (specifically, 2.4.2, and 2.2.18). The 2 kernels are the
only place I see this. Even more specifically, I see it with the compiled
kernel versions shipped with Red Hat 7.1 and SuSE 7.1.

The second half of the question, am I completely bonkers? If so,
just tell me.

Jesse Noller
Linux Systems Lead / Caffeine Distiller
Department of Shiny Happy People
Macromedia, Inc.
jnoller@macromedia.com

"This place is giving me the Fear. "
-Hunter S. Thompson
Fear and Loathing in Las Vegas

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 12:52    [W:0.286 / U:0.168 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site