[peruser] segfaults with patch v7 on freebsd 7/amd64
marcelo at tpn.com.br
Thu Jan 6 11:23:00 MST 2011
When you used RC2 patch, did you receive sig11's?
On Jan 6, 2011, at 3:01 PM, Dave Steinberg wrote:
> Hello list, I'm getting signal 11 messages in my error log, and I've managed to get core dumps from my httpd proc. Here's a snippit of what I'm seeing in gdb:
> (gdb) bt
> #0 apr_palloc (pool=0x80f712028, in_size=Variable "in_size" is not available.
> ) at memory/unix/apr_pools.c:252
> #1 0x000000080102ba60 in apr_pstrcat (a=Variable "a" is not available.
> ) at strings/apr_strings.c:149
> #2 0x000000000044902c in pass_request (r=0x80f7120a0, child=0x8005a8598) at peruser.c:1510
> #3 0x00000000004494a9 in peruser_post_read (r=0x80f7120a0) at peruser.c:3407
> #4 0x000000000042916a in ap_run_post_read_request (r=0x80f7120a0) at protocol.c:1671
> #5 0x000000000042b4ce in ap_read_request (conn=0x80f706298) at protocol.c:1011
> #6 0x00000000004407d3 in ap_process_http_connection (c=0x80f706298) at http_core.c:183
> #7 0x000000000043d0d2 in ap_run_process_connection (c=0x80f706298) at connection.c:43
> #8 0x000000000044a174 in process_socket (p=0x80f706028, sock=0x80f7060a0, conn_id=16, bucket_alloc=0x80f710028, pool=0x80f704028)
> at peruser.c:1364
> #9 0x000000000044b32b in child_main (child_num_arg=Variable "child_num_arg" is not available.
> ) at peruser.c:2211
> #10 0x000000000044bae5 in make_child (s=0x801617868, slot=16) at peruser.c:2532
> #11 0x000000000044c481 in ap_mpm_run (_pconf=Variable "_pconf" is not available.
> ) at peruser.c:2888
> #12 0x0000000000423afa in main (argc=3, argv=0x7fffffffeab8) at main.c:739
> I'm not sure if this is sufficient to help debug the problem, and I can provide the full gdb session if desired. Apache is running here with mod_php, mod_perl, and mod_jk ... just to make things extra fun.
> Is there more info I can provide that would be helpful?
> In terms of reproducing this, it seems to happen fairly consistently on a few child procs when I do an 'apachectl graceful', but beyond that it isn't clear what triggers it.
> I'll try to collect more core files and see if they all share a common backtrace.
> PS - Before switching to mpm-peruser, I almost never saw any sig11's.
> Thanks for your time,
> Dave Steinberg
> Peruser mailing list
> Peruser at telana.com
More information about the Peruser