Re: SMP Error in VIA 694 based bord

Terry Lambert wrote:
> ?? wrote:
> > 
> > thanks to Terry, I added option DISABLE_PSE,then the smp works fine,I
> > already boot over 30hrs,and running stable, thank
> > you for your help.
> > May I ask you? what is PSE,it seens like a cpu function?
> 
> Thanks for posting to the list.  People generally don't
> believe the FreeBSD has a PSE problem, without a confirming
> success report in the DISABLE_PSE case, like yours.  Actually,
> Satoshi Asami and the people still at ClickArray should be
> able to conrim the problem as well as the workaround.

Is this the one you posted about regarding interactions with PG_G and
overlapping PG_PS (page size) pages and the timing of setting the
CR4_PSE flag?

> It is the page size extension for the use of 4M pages.
> 
> There is a subtle bug in both Intel and AMD processors with
> regard to the PSE.
> 
> FreeBSD used to accidently work around it.  Apparently, it
> no longer does.
> 
> By disabling the PSE, you prevent the use of 4M pages, which
> avoids triggering the bug in the first place.
> 
> If someone who cares about 4M pages wants to narrow down the
> code change that resulted in FreeBSD no longer working around
> the problem, it night be useful to seperate the use of the PSE
> for the kernel from the use of the PSE for memory mapped
> devices.

Terry, if you're not going to give details about what the problem is, then
please stop ranting on and on about this.  The whole "There is a bug, and
I'm not going to tell you what it is" thing is getting rather tiring.

Cheers,
-Peter
--
Peter Wemm - [e-mail]; [e-mail]; [e-mail]
"All of this is for nothing if we don't go to the stars" - JMS/B5


To Unsubscribe: send mail to [e-mail]
with "unsubscribe freebsd-smp" in the body of the message