Re: VIA C7 / VIA PC-1 (PC2500) anyone?

[Linux-kernel <linux-kernel@vger.kernel.org>]

Michael Tokarev wrote:
> I bought a VIA PC2500 board a few days ago - this
> new series of their mobos,
> 
> This beast looks nice - after replacing their cooling
> system (that had a small fan on it) with larger but
> fanless, -- it becomes a almost real PC (1500MHz CPU),
> equipped with quite nice crypto and multimedia abilities,
> but with very low power consumption and very quiet.
> 
> But the thing is - it doesn't quite work.
> 
> It works generally - it boots, I can run my usual apps
> etc.  But on a random (yet frequent) basis it segfaults
> here and there.  For example:
> 
> $ man man
> Reformatting man(1), please wait...
> $ man man
> Segmentation fault
> $ man man
> Segmentation fault
> $ man man
> Segmentation fault
> $ man man
> Segmentation fault
> $ man man
> Reformatting man(1), please wait...
> $ _
> 
> (this is 100% idle machine, just booted).
> 
> (There are other - simple and comples - applications which
> inhibits this problem.  For example, it 99% reliable segfaults
> on compiling aic79xx_core.c file in kernel, while all the rest
> (in my configuration anyway) compiles, at least after second
> attempt).
> 
> It's definitely NOT memory issue - I tried several different
> memory modules (and different combinations) - the same results;
> I ran memtest86 for several days - no single error.
> 
> I've seen a thread here on LKML about C7 and C3 CPUs back in
> March this year - tried with patch from Andi titled
> "i386: Enable CX8/PGE CPUID bits early on VIA C3" - it didn't
> change anything (this board does not lock up - not when booting
> nor when doing something, -- just random applications are
> crashing randomly, and the crash is always SIGSEGV; there's
> _nothing_ in dmesg about that, too).
> 
> From all the above it seems like something's broke on the
> motherboard (I've no idea what it can be however - because
> memory testing - which also tests for CPU cache for exampe -
> shows no errors; testing disk controller/disk using md5 does
> not show errors either, except of occasional SIGSEGVs)..
> 
> However, being very curious about this, I tried installing
> 'doze on this machine - winXP.  And that one went just fine
> without any error so far -- i tried stress-testing it as far
> as I can imagine, running various applications and workloads, --
> no errors.
> 
> So I'm kinda.. stuck about what to do next.
> 
> Any.. idea, anyone? :)
> 
> Thanks!
> 
> /mjt

To me it looks like a wrong choice of gcc switches to user-mode programs. What
distribution are you using? try compiling failing programs from source with
conservative command line switches to gcc. See if things change.

Boaz