Re: [PATCH] Mutilated form of Andi Kleen's AMD prefetch errata patch

[Andi Kleen]

On Wed, Oct 01, 2003 at 12:00:15AM -0700, Andrew Morton wrote:
> Andi Kleen <ak@xxxxxxx> wrote:
> >
> > Andrew Morton <akpm@xxxxxxxx> writes:
> > 
> > > Looking at Andi's patch, it is also a dead box if the fault happens inside
> > > down_write(mmap_sem).  That should be fixed, methinks.
> > 
> > The only way to fix all that would be to move the instruction checks early
> > into the fast path.
> 
> Well the deadlock avoidance only needs to happen if the fault occured in
> kernel mode.  Presumably most faults are in userspace, so most of the
> overhead can be avoided.

Hmm.   I guess that's possible, but will be somewhat intrusive in 
do_page_fault()

Also you have to be very careful to avoid recursive faults (EIP unmapped) 
recursing further. In the original patch I did that for kernel mode by always
checking the exception table first to catch the __get_user in __is_prefetch
early.

Maybe it would be better to just use a down_read_timeout().

-Andi
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/