[privoxy-devel] [ ijbswa-Bugs-1621173 ] crash with banners-by-size

[SourceForge.net <no...@so...>]

Bugs item #1621173, was opened at 2006-12-23 09:33
Message generated for change (Comment added) made by roro
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=111118&aid=1621173&group_id=11118

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: version 3.0
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Roland Rosenfeld (roro)
Assigned to: Nobody/Anonymous (nobody)
Summary: crash with banners-by-size 

Initial Comment:
http://bugs.debian.org/404284 reports that privoxy segfaults 
on downloading http://www.cs.rutgers.edu/~ccshan/debbugs/privoxy/

I can reproduce the problem here and found out that disabling 
the banners-by-size filter works around this problem.

When I run privoxy in gdb I get the following backtrace on this:

#0  0x00002b2f761a857e in pcre_exec () from /usr/lib/libpcre.so.3
#1  0x0000000000422fbb in pcrs_execute (job=0x588580, subject=0x2aaaaae66010 <Address 0x2aaaaae66010 out of bounds>, subject_length=542797, result=0x407fecf8, 
    result_length=0x407fed00) at pcrs.c:848
#2  0x000000000041688e in pcrs_filter_response (csp=0x53a460) at filters.c:1763
#3  0x0000000000419daf in serve (csp=0x53a460) at jcc.c:1676
#4  0x00002b2f763bcf1a in start_thread () from /lib/libpthread.so.0
#5  0x00002b2f76595612 in clone () from /lib/libc.so.6
#6  0x0000000000000000 in ?? ()

This uses Debian libpcre3 package which is pcre version 6.7.

No idea, what's going wrong here, I tried this with Debian 3.0.6-1
(including some Debian patches), with a vanilla 3.0.6 and with 
the CVS version as of 2006-12-23 and the problem is always reproducible.
It seems not to be architecture dependant, because I noticed it on
AMD64, while the original submitter users i686. 

Tschoeeee
Roland

----------------------------------------------------------------------

>Comment By: Roland Rosenfeld (roro)
Date: 2006-12-23 09:53

Message:
Logged In: YES 
user_id=43129
Originator: YES

I just found out, that compiling with --disable-dynamic-pcre 
seems to solve the problem.  It still takes a lot of time and CPU
until the page is received but compiled this way, no segmentation
fault appears (neither on 3.0.6 nor on 3.0.7 CVS).

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=111118&aid=1621173&group_id=11118