Vulnerability Development mailing list archives

Re: procmail again


From: ned <nd () felinemenace org>
Date: Sat, 18 Oct 2003 22:34:14 -0700 (PDT)

libd.so.1 is the sharefuzz getenv() hooker which just returns big buffers.
i no longer have a redhat 7.1 machine and that information is little over 
12 months old therefore someone with a rh 7.1 system please send in your 
results!
-nd

On Sat, 18 Oct 2003 Valdis.Kletnieks () vt edu wrote:

On Sat, 18 Oct 2003 17:31:35 PDT, ned <nd () felinemenace org>  said:
can't reproduce this on anything other than redhat 7.1 although procmail 
will segfault when a ctrl-c is passed on mandrake.
procmail has a very obvious signal handling problem.

RedHat 7.1 shipped procmail 3.14.

Current is 3.22.  I don't have a 3.14 handy, but there's at least 3 things I've
spotted in the HISTORY file that may be applicable and explain why you can't
reproduce it on other systems.

1) Does the system have a /etc/procmailrc, and if so, what's in it?  There's a
fix for LASTFOLDER getting set on '?' tests incorrectly, but triggering that bug
would require at least one procmailrc file being read (and $HOME/.procmailrc isn't
available according to the output).

2) What is this 'LD_PRELOAD=./libd.so.1', and are you positive that it's not
the cause of the problem?  Is the problem reproducible without preloading your
own libraries?

3) It's a *LOT* easier to comment if the gdb output includes a 'where', as that
will do a stack backtrace and give us a HUGE hint as to where the problem lies.





-- 
http://felinemenace.org/~nd


Current thread: