I've also been experiencing segmentation faults. I reported it,
yesterday, some, not fully detailed, (ie: no stack trace since no
debugging symbols...) to the mod_perl mailing list.
What I did was run "gdb", then used the "ps wauxf" output to get the
PID of an Apache child process. I used GDB's "attach PID" to debug
that process, and then "cont" to let it run, since attaching stops
it.
I then hit the page over and over (whacked it with a netscape) until
that process segv'd, and used "where" to find out where it had
crashed. It listed an address, but ?? rather than a subroutine name,
since everything has been stripped.
I started up "gtop", and got a memory map of that process, then
looked to see what library the crash address resides in. It was in
"Cookie.so", which is part of Apache::Cookie. I do not know what
routine it's in; I've not built that with "-ggdb" yet to try and
track it down. Perhaps I'll take some time for that next week if I
don't hear anything re a cure / patch before then.
James? Please follow the similar process, and see if you can
determine what .so library the segv is occuring in. Will you also
please report to both lists?
--
mailto: (Karl M. Hegbloom) [EMAIL PROTECTED]
http://www.microsharp.com
phone://USA/WA/360-260-2066
