Marco Pietrobono wrote:
Jacob Kolding wrote:
>
> Lately I've been complaining random crashes in StarCraft, but haven't
> received any
> replies. The reason I suspect is the fact that starcraft runs
> perfectly as long as
> don't run on a SMP system. Is this a known problem, starcraft and
> possibly other
> apps suffering random crashes when running on SMP Linux?

  I don't know if it is a known problem, but I'm experiencing the very
same problem with Baldur's Gate on a dual celeron. If I play with an SMP
enabled kernel I'm getting several crashes too often at random times...

  I've tried to debug those crashed but I was getting nowhere...

I tried but it's my first encounter with wine debugging and debugging a threaded
app running in DGA mode on a SMP system doesn't seem like the right way
to start :)
I did however find while tracing SEH that the same exception is raised very often just
before the crash:

trace:seh:EXC_RtlRaiseException code=c000013a flags=0
wineserver: thread.c:454: sleep_on: Assertion `!current->wait' failed.
Client protocol error:0x805e0d8: Client protocol error:0x805dd48: read: Connection reset by peer

In the above situation starcraft would hang and the music loop, and the wineserver part
would appear when I killed the starcraft process.

Is it possible that it is a kernel problem?
Is just seems wrong to me that a program written for pthreads will work properly
on UP and not on SMP. Maybe I should post a message to the kernel list?
Or could it be that windows threads aren't designed to run on SMP?

>
> Anyway the kernels I've used were 2.2.14 and 2.3.99pre3 both with SMP
> and 2.2.14 UP. With only a single CPU active starcraft runs like a
> dream
> (I have played for several hours ), but with SMP it will crash with in
> 20 min sometimes
> even at the startup screen.

  I can confirm this for the 2.2.14 one... currently I switch back to
the uniprocessor kernel every time I find an animation, since it seems
that otherwise BG deadlocks at the end of each animation.

> I haven't really made any scientific test but it seems that it will
> crash more frequently
> under 2.3.99pre3.
>
> As mentioned many times the crashes are random and most of the time
> there is no error message,
> but I have registered the following errors:

  I've got way too many errors, all of them seems completely unrelated,
but most often it stops to work without any advice... BTW, there is a
strange sound's behaviour that makes me think of something
sound-related, but I have been not able to find any clue, yet...

  bye,

    MP.

--
Stud. Marco Pietrobono            | Murphy's Law: if something could
v. del Calice, 39 - 00178 ROMA    |     go wrong, it does.
Tel. +39.6.7186329  0339.7410893  | Legge di Murphy: se qualcosa può
http://www.pietrobo.com           |    andar male, lo farà.
------------------------------------------------------------------------
Some men see things as they are and say, why;
I dream things that never were and say, why not. ( George Bernard Shaw )

-- 
        -Jacob Kolding, mailto:[EMAIL PROTECTED]
 


Reply via email to