apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jeff Walter <j...@404ster.com>
Subject Re: apr_global_mutex_create never returns
Date Fri, 07 Jan 2005 16:47:34 GMT
Jeff Trawick wrote:
> what is the backtrace?  (attach with debugger)

Funny... I didn't notice it before, my apr_shm_create causes a segfault
(when I remove the file that actually gets created the run before).  A
trace...

open("/var/lib/mod_straw/mod_straw.shm", O_WRONLY|O_CREAT|O_EXCL, 0666) = 35
stat64("/var/lib/mod_straw/mod_straw.shm", {st_mode=S_IFREG|0644,
st_size=0, ...}) = 0
shmget(17008977, 20800, IPC_CREAT|IPC_EXCL|0600) = 4030469
shmat(4030469, 0, 0)                    = 0x411e4000
shmctl(4030469, IPC_64|IPC_STAT, 0xbffff9d0) = 0
getuid32()                              = 0
getgid32()                              = 0
shmctl(4030469, IPC_64|IPC_SET, 0xbffff9d0) = 0
write(35, "@Q\0\0", 4)                  = 4
close(35)                               = 0
--- SIGSEGV (Segmentation fault) @ 0 (0) ---


I took a gander at the APR SHM stuff, the only thing happening after the
file is closed is:

apr_pool_cleanup_register(new_m->pool, new_m, shm_cleanup_owner,
     apr_pool_cleanup_null);
*m = new_m;
return APR_SUCCESS;

If all it is doing is registering a cleanup handler, what could possibly
cause it to segfault?

Thanks again,

Jeff Walter

Mime
View raw message