[Citadel Development] Re: Unusual SSL behavior, likely bug

[scianos]

Sorry about the delay in getting these logs... Had my workstation decide to melt down (or at least the CPU, literally... thankfully Intel has a 3 year warranty...)

Logs are attached. At the end of foolog_IMAPS.3 and foolog_IMAPS.4 it appears that we might (and I stress might, because I have no idea and haven't looked at the code) be catching a buffer underrun when SSL is enabled.

I used mailing list traffic from full disclosure as the source messages to populate the development system, henceforth the strange headers or titles in some of the messages.

- Stu

Wed Aug 18 2010 02:05:28 AM EDT from dothebart @ Uncensored Subject: Re: Unusual SSL behavior, likely bug

if you compile citadel with

'-D BIGBAD_IODBG'

as one of your CFLAGS, it will write a transscript of what its doing on al IO chanels to /tmp/modulename_fd_random

this could give us a chance to catch your incident.

do you have some test system where you could try that out? (on a production system /tmp/ would run full verry quick)

 

citadel_ssl_debug_logs_1.tar.gz
Description: GNU Zip compressed data