[bf1942] Another new build...

[th?nder]

It crashed on a map change giving me a lot of:
STUB: menu player chat thingey {BFMultiPlayer/GameServer.cpp:1916}. 
and ended with
Segmentation fault

CPU usage was a bit high as well (compared to windows version) bout 50%
on 1500MHz AMD, 512RAM, slackware 8.1 with 20 players on it

 
-----Original Message-----
From: Steve Beck [mailto:spbeck at 2laurelbank.nildram.co.uk] 
Sent: Sunday, December 15, 2002 6:12 PM
To: bf1942 at icculus.org
Subject: Re: [bf1942] Another new build...

We set up a test server at Jolt and had 50% cpu usage with 8 players on


----- Original Message -----
From: "Ryan C. Gordon" <icculus at clutteredmind.org>
To: <bf1942 at icculus.org>
Sent: Sunday, December 15, 2002 7:26 AM
Subject: [bf1942] Another new build...


>
> Try this one on for size:
>
http://icculus.org/betas/bf1942/bf1942-lnxded-betaupdate-build-103993352
5.ta
r.bz2
>
> This applies over the complete install, with or without the previous
> build. Just untar it so it overwrites the binaries and README.
>
> I fixed a metric ton of invalid memory accesses, three deferences of
bogus
> pointers, and one serious memory corruption (which I'm hoping was the
> cause of all our woes).
>
> Please try it out. Usual disclaimer applies: this works great here
(but it
> always has, so this is not at all a comfort by this point, I'm sure),
but
> until I get some confirmation from the field, I can't promise this
won't
> crash on connect, etc.
>
> And, for the record, valgrind is a _fantastic_ tool for tracking down
> memory bugs. If you are a programmer and aren't using this on your
code,
> you really really should. It totally blows away ElectricFence (which,
> coincidentally, runs out of mmap()able address space before the server
is
> finished starting...I've had that problem with efence on every major
game
> I've used it with), and dmalloc, etc.
>
> --ryan.
>
>
>
>