Re: [PATCH 2/2 V7] powerpc/85xx: Add machine check handler to fix PCIe erratum on mpc85xx

[Scott Wood <scottwood@freescale.com>]

On 04/26/2013 09:26:26 PM, Jia Hongtao-B38951 wrote:
> > -----Original Message-----
> > From: Wood Scott-B07421
> > Sent: Friday, April 26, 2013 12:58 AM
> > To: Segher Boessenkool
> > Cc: Jia Hongtao-B38951; linuxppc-dev@lists.ozlabs.org;
> > galak@kernel.crashing.org; Wood Scott-B07421
> > Subject: Re: [PATCH 2/2 V7] powerpc/85xx: Add machine check handler =20
> to
> > fix PCIe erratum on mpc85xx
> >
> > On 04/25/2013 10:31:51 AM, Segher Boessenkool wrote:
> > >> * Remove A variant of load instruction emulation
> > >
> > > Why is this?  You handle all other simple load insns, there is =20
> nothing
> > > special about LHA.  (I reviewed the V4 email thread, no reason =20
> for the
> > > chance is given there).
> >
> > The LHA implementation in V5 was incorrect (didn't sign-extend).
> >
> > -Scott
>=20
> In former email you doubt whether we need A variant or not.
> Any particular reason for that?
> If not should I emulate all the A ARX AU AUX and AX variant?

I was just noting that the variants you left out from the earlier =20
revisions (e.g. BRX) were much more likely to be used for I/O than some =20
of the ones you included (e.g. "A").  Implementing all the normal =20
load/store instructions would be better, if they're done correctly.

-Scott=