[Intel-gfx] [PATCH 2/2] drm/i915: Disable page-faults around the fast pwrite/pread paths

[Keith Packard]

On Sat, 09 Jul 2011 22:23:28 +0100, Chris Wilson <chris at chris-wilson.co.uk> wrote:
> On Sat, 09 Jul 2011 14:06:23 -0700, Keith Packard <keithp at keithp.com> wrote:
> > On Sat, 09 Jul 2011 21:50:26 +0100, Chris Wilson <chris at chris-wilson.co.uk> wrote:
> > Sure, it's permitted, so ideally we'd detect this abuse and fall back to
> > the slow path, but we need a cheap check which takes the slow path,
> > perhaps pessimistically.
> 
> If we prefault every page, then we only hit the slow path under system
> load combined or severe memory pressure. I don't think that is too bad a
> compromise.
> 
> I can stick some counters in there and find out what the impact actually
> is.

Cool. I've separately posted a query to lkml asking if the existing
fault_in_pages_writeable/fault_in_pages_readable should be fixed - that
looks like it will provide a performance boost for filesystem reads
larger than two pages.

-- 
keith.packard at intel.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20110709/493f6318/attachment.sig>