[Intel-gfx] [PATCH] Drop the mutex whilst waiting for a request

[Eric Anholt]

On Wed, 2009-02-04 at 13:58 +0000, Chris Wilson wrote:
> Hello all,
> 
> I'm seem to be able to hang the chip at will, though it seems quite
> sensitive to timing. For example if I sync the batch buffer after every
> exec, the bug disappears and rendering continues.
> 
> During the course of my travails I found that i915_wait_request() holds
> the struct_mutex whilst sleeping which, among other effects, prevents me
> from accessing the proc/debug interfaces. So I've attached a rudimentary
> patch to drop the mutex for the duration of the sleep. Review would be
> appreciated.

NAK -- this allows other ioctls to change object state in the midst of
what are supposed to be atomic requests (think about two execbuffers()
happening at the same time).

-- 
Eric Anholt
eric at anholt.net                         eric.anholt at intel.com


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20090204/d1f30c5d/attachment.sig>