[dropping linux-kernel of the cc:]
On Fri, May 13, 2011 at 11:02:27PM -0700, Kevin Cernekee wrote:
> On Fri, May 13, 2011 at 10:13 PM, Jayachandran C.
> <jayachandranc@netlogicmicro.com> wrote:
> > Can you send me the patchset which works on top of queue with any
> > debugging you want enabled? I can try that and send you the results.
> >
> > It is also possible that something is broken with the XLR platform code,
> > it is currently almost straight r4k...
>
> Well, David suggested adding "#define DEBUG 1" at the very top of
> tlbex.c, then booting with "debug" and posting the TLB refill handler
> to make sure the RI/XI code isn't getting enabled. That seems like a
> reasonable start. Even if there's no smoking gun, we'd still be able
> to compare our TLB handlers side-by-side.
>
> Personally I don't have any other leads or patches to try. These
> changes work fine for me in every configuration I am able to test:
>
> 32-bit MIPS32 system, 32-bit kernel (non-RIXI)
> 32-bit MIPS32 system, 32-bit kernel (XI)
> 64-bit R5000 system, 32-bit kernel (non-RIXI, with 64-bit physical addresses)
> 64-bit R5000 system, 64-bit kernel (non-RIXI)
>
> So it's really best for somebody to debug the problem hands-on, on the
> system that showed the issue.
I will try this patch again on top of 'queue' with debugging and send you
an update. Hope to do it today or tomorrow.
JC.
|