libpciaccess: Changes to 'master'

[Daniel Stone]

On Wed, Apr 16, 2008 at 12:12:06PM -0700, Keith Packard wrote:
> commit b30d458202bc0304c705eb081b12ead860584bea
> Author: Keith Packard <keithp at keithp.com>
> Date:   Wed Apr 16 12:10:52 2008 -0700
> 
>     Kludge around linux bug and turn off write-through and cache-disable bits
>     
>     When mmaping the PCI device, the kernel turns on the write-through and
>     cache-disable bits in the allocated PTEs. This disables write-combining mode
>     and dramatically reduces write bandwidth to the frame buffer. While that
>     should be fixed in the kernel, we'll kludge around it here by using mprotect
>     to rewrite the PTEs and get those bits turned off.

Wow, it's a déja vu.

On Thu, Apr 17, 2008 at 10:05:03AM -0700, Keith Packard wrote:
> commit c3fb62df4e60b63295f94c99b3c5de70dbf94e1c
> Author: Keith Packard <keithp at keithp.com>
> Date:   Thu Apr 17 10:04:55 2008 -0700
> 
>     Add a kludge-around to fix cd/wt bits in fb ptes on linux.
>     
>     Mmap from /sys/devices/pci* on linux forces the cache-disable and
>     write-through bits, which turns our write-combining map into an
>     uncached-map, seriously impacting performance. It turns out that a bug in
>     mprotect allows us to fix this by disabling access to those pages and then
>     immediately re-enabling them.

Wow, it's a déja vu.

Cheers,
Daniel
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.x.org/archives/xorg/attachments/20080417/ecece034/attachment.pgp>