[PATCH] powerpc/8xx: fix regression introduced by cache coherency rewrite

[Joakim Tjernlund]

>
> Thus spake Benjamin Herrenschmidt (benh at kernel.crashing.org):
>
> > On Thu, 2009-09-24 at 18:35 -0700, Rex Feany wrote:
> > >
> > > Then I can boot and get to a shell, but userspace is slow. 8 seconds
> > > to mount
> > > /proc (vs. less then a second using my old kernel)! Maybe this is an
> > > unrelated issue?  I'm pretty clueless about the details, I'm sorry.
> > > PG_arch_1 is used to prevent a cache flush unless it is actually
> > > needed?
> > > Then why would changing the location of the tlbil_va() make a
> > > difference?
> >
> > I think there's more finishyness to 8xx than we thought. IE. That
> > tlbil_va might have more reasons to be there than what the comment
> > seems to advertize. Can you try to move it even higher up ? IE.
> > Unconditionally at the beginning of set_pte_filter ?
> >
> > Also, if that doesn't help, can you try putting one in
> > set_access_flags_filter() just below ?
> >
> > (Beware that there's two different versions of both functions, only the
> > first one is compiled/used on 8xx).
> >
> > It's going to be hard for me to get that "right" since I don't really
> > know what's going on with the core here, but I suppose if we get it
> > moving along with extra tlb invalidations, that should be "good enough"
> > until somebody who really knows what's going on comes up with possibly
> > a better fix.
>
> I've tried sticking tlbil_va() in those places, nothing seems to help.
> In some cases userspace is slow, in other cases userspace is faster and
> unstable: sometimes commands hang, sometimes I am able to ctrl-c and
> and kill it, sometimes I get other strange crashes or falures (so far no
> kernel oopses though).

This is exactly what you get when the "cache insn does not update DAR" bug hits
you.