[Bug 70461] [sna suse] ddx assertion hits

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=70461

--- Comment #128 from Chris Wilson <chris at chris-wilson.co.uk> ---
As expected it says the kernel failed to idle the bo when we tried to move it
to the CPU write domain:

[  2836.327] kgem_bo_sync__cpu: handle=21
[  2836.327] __kgem_busy: handle=21, busy=0, wedged=0
[  2836.327] kgem_bo_sync__cpu: SYNC: handle=21, needs_flush? 1, domain? 3,
busy
? 0
[  2836.327] __kgem_busy: handle=21, busy=0, wedged=0
[  2836.327] kgem_bo_retire: retiring bo handle=21 (needed flush? 1), rq? 1
[bus
y?=0]
[  2836.327] __kgem_retire_requests_upto(handle=21)
[  2836.327] __kgem_retire_rq: request 1 complete
[  2836.327] __kgem_bo_flush: handle=21, busy=65536, wedged=0
[  2836.327] __kgem_bo_flush: moving 21 to flushing
[  2836.327] __kgem_retire_rq: movied 21 to flushing


Hmm, at the time it should not have been accessible by any other client, but at
some point in its past the handle was exported.

This will prevent this assertion from triggering (by papering over the bug):

commit 3656bcf01e78554aaeafc11ecebb0caf7dd12713
Author: Chris Wilson <chris at chris-wilson.co.uk>
Date:   Mon Oct 19 09:34:43 2015 +0100

    sna: Manually clear the GPU bo needs_flush flag after doing an explicit
flush

    The kernel often gets confused and fails to flush the bo itself, so lets
    just pretend it worked.

    Reported-by: Jiri Slaby <jirislaby at gmail.com>
    References: https://bugs.freedesktop.org/show_bug.cgi?id=70461#c127
    Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20151019/0dddf072/attachment.html>