PDA

View Full Version : Patch - 5.0.0.17-purple1



purple
02-18-2005, 08:11 PM
Ok, he's an updated patch. It is against the latest CVS. Please keep you compiling problems to the help forum. If you're seeing crashing or severe problems probably not caused by you doing something dumb and instead by me doing something dumb, please bring it up! Otherwise take it to the help forum.

Hell if I know if it fixes any of the 3 outstanding problems that have been reported because I can't reproduce any of them. I've spent some time with valgrind fixing problems it reported all over the place. It's a bit cleaner now. I did fix one large problem with cached packet memory not getting freed properly because it was double decompressing (once before putting into the cache and once after taking it out) and only freeing the 2nd one.

Changelog from my last patch:
- Fixed a problem from my last patch where decompressed packets weren't getting freed properly
- A lot of small memory cleanups, mostly with exiting
- Fixed groupUpdateStruct and groupFullUpdateStruct to match live
- Made the console messages for grouping look better
- Made tells not double up on the console messages by ignoring the client->server channel message
- Updated worldopcodes.xml (thanks Doodman)
- Reformatted All Packets log (global.log) so it's more useful
- Updated version to 5.0.0.17-purple1

And included again from my previous patch to cvs:
- Compression on the net protocol
- All opcodes changed again (Thanks ieatacid and Doodman for help with these)
- Struct changes
- Buffs now go up to 25

Currently outstanding issues (let me know if any of these are fixed - I can't reproduce any of them):
- Compression fails when things are compiled 64bit
- Stack clobber randomly causes sigsegv (backtrace shows EQPacketStream::vtable, vnsprintf, vsprintf crytically)
- Fragmentation buffer overflows during zoning

Attached:5.0.0.17-purple1.patch.gz.txt (gzipped patch, remove .txt)

THIS PATCH IS SUPERCEDED. See 5.0.0.17-purple2. If you already have this working, it's not a big deal to update really.

I<3EQ
02-18-2005, 09:02 PM
I had hunk 2 in src/main.cpp fail. Upon closer inspection, it looks like that change was already committed to cvs (the patched line was the same as the original line in cvs).. For the record, this *was* against a brand new pull from CVS. :p

Other than that, it's looking good for me so far! Thanks for all your hard work purple. If you're ever in the Boston or DC areas (depending on where I am at the time), I so owe you several beers. :)

tanner
02-18-2005, 11:08 PM
I had hunk 2 in src/main.cpp fail. Upon closer inspection, it looks like that change was already committed to cvs (the patched line was the same as the original line in cvs).. For the record, this *was* against a brand new pull from CVS. :p

Similar here.

patching file src/main.cpp
Hunk #2 FAILED at 47.
1 out of 3 hunks FAILED -- saving rejects to file src/main.cpp.rej

% cat main.cpp.rej


***************
*** 47
- static const char *id="@(#) $Id: main.cpp,v 1.33.6.5 2004/09/24 04:25:53 dohpaz Exp $ $Name: $";
--- 47 -----
+ static const char *id="@(#) $Id: main.cpp,v 1.33.6.5 2004/09/24 04:25:53 dohpaz Exp $ $Name: pre_5_0_beta $";

Stupid RCS expansions!

Patch attached.

purple
02-18-2005, 11:32 PM
Yeah, sorry. I normally hand edit the diff to take those out. I missed one.

purple
02-18-2005, 11:40 PM
Preliminary feedback is that the 64bit problem is fixed.

tanner
02-18-2005, 11:58 PM
Rolled purple1 patch into debian package.


Format: 1.7
Date: Fri, 18 Feb 2005 23:15:49 -0600
Source: showeq
Binary: showeq-dev showeq showeq-doc
Architecture: source i386 all
Version: 5.0.0.17-0.3
Distribution: unstable
Urgency: low
Maintainer: Bob Tanner <[email protected]>
Changed-By: Robert J. Tanner <[email protected]>
Description:
showeq - realtime packet analyzer for Everquest
showeq-dev - development tools for ShowEQ
showeq-doc - documentation for showeq
Changes:
showeq (5.0.0.17-0.3) unstable; urgency=low
.
* 18-Feb-2005 purple1 patch
* http://www.showeq.net/forums/showpost.php?p=38253&postcount=1
* 18-Feb-2005 purple1-tanner1 patch to fix rcs keyword expansions
* http://www.showeq.net/forums/showpost.php?p=38260&postcount=3
Files:
aff4c4f393ebb8f7c3ea75913cf7b811 728 net optional showeq_5.0.0.17-0.3.dsc
efa130899325823cea9c35eefe9ad9e4 1038297 net optional showeq_5.0.0.17-0.3.tar.gz
c39ffb81e996157a4d859ff261f18e0e 74268 doc optional showeq-doc_5.0.0.17-0.3_all.deb
fe7dea6728720befee55f92bf6d24189 6604928 net optional showeq_5.0.0.17-0.3_i386.deb
d029642536081c9410c6f6084024a21d 80338 net optional showeq-dev_5.0.0.17-0.3_i386.deb

icyman
02-19-2005, 12:04 AM
Still attempting to reproduce the previous seg fault, but since using --enable-debug I haven't been able to. Will keep you posted. I am now using this patch.