Jeff Trawick wrote:
Run under valgrind from the shell, and if that doesn't work create a wrapper script to be the CGI and use the wrapper script to invoke the real CGI under valgrind in that environment.
Tried that already: (executable is /.../request; paths shortened and valgrind line wrapped for brevity) #!/bin/sh valgrind --leak-check=full --track-origins=yes \ --log-file=/.../logs/valgrind.request.log /.../request CGI script runs to completion.
El 03/12/12 15:37, J.Lance Wilkinson escribió:
Title pretty much says it all.
HTTPD is 2.2.15 on RHEL6.
Executable aborts, log files show glibc detected memory issues
(messages vary).
Run same executable outside of the CGI environment, on the same
host and under an unprivileged user (can't run as "apache"
because that's not a real ID, my sysadmin says), duplicating all
input, under a debugger and not under a debugger, and it runs to
completion with no errors.
What could HTTPD as the parent environment be imposing that
would cause the aborts when they don't occur outside of Apache?
I speculated some ulimit kind of thing resulting in less memory
actually being available for the thousands of malloc() calls the
executable performs.
--
J.Lance Wilkinson ("Lance") InterNet: Lance.Wilkinson@xxxxxxx
Systems Design Specialist - Lead Phone: (814) 865-4870
Digital Library Technologies FAX: (814) 863-3560
E3 Paterno Library
Penn State University
University Park, PA 16802
http://ucs.psu.edu/home/jlw12@xxxxxxx?fmt=freebusy
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@xxxxxxxxxxxxxxxx
For additional commands, e-mail: users-help@xxxxxxxxxxxxxxxx
|