== Series Details ==
Series: series starting with [1/2] drm/i915/execlists: Listen to COMPLETE
context event not ACTIVE_IDLE
URL : https://patchwork.freedesktop.org/series/34062/
State : success
== Summary ==
Test kms_flip:
Subgroup wf_vblank-vs-modeset-interruptible:
On Sat, 18 Nov 2017 23:29:31 +0530
Kirti Wankhede wrote:
> Extremely sorry for the delay.
> This works for VFIO_GFX_PLANE_TYPE_REGION. Tested with local changes.
>
> Reviewed-by: Kirti Wankhede
With that,
Acked-by: Alex Williamson
== Series Details ==
Series: series starting with [1/2] drm/i915/execlists: Listen to COMPLETE
context event not ACTIVE_IDLE
URL : https://patchwork.freedesktop.org/series/34062/
State : success
== Summary ==
Series 34062v1 series starting with [1/2] drm/i915/execlists: Listen to
COMPLETE
Extremely sorry for the delay.
This works for VFIO_GFX_PLANE_TYPE_REGION. Tested with local changes.
Reviewed-by: Kirti Wankhede
Thanks,
Kirti
On 11/18/2017 9:00 PM, Alex Williamson wrote:
>
> Kirti?
>
> On Wed, 15 Nov 2017 21:11:42 -0700
> Alex Williamson
The hardware needs some time to process the information received in the
ExecList Submission Port, and expects us to don't write anything new until
it has 'acknowledged' this new execlist by sending an IDLE_ACTIVE or
PREEMPTED CSB event.
If we do not follow this, the driver could write new data
Since commit e1fee72c2ea2e9c0c6e6743d32a6832f21337d6c
Author: Oscar Mateo
Date: Thu Jul 24 17:04:40 2014 +0100
drm/i915/bdw: Avoid non-lite-restore preemptions
execlists has listened to (ACTIVE_IDLE | ELEMENT_SWITCH) for detecting
when one context completed and it
Quoting Chris Wilson (2017-11-18 11:25:32)
> The hardware needs some time to process the information received in the
> ExecList Submission Port, and expects us to don't write anything new until
> it has 'acknowledged' this new execlist by sending an IDLE_ACTIVE or
> PREEMPTED CSB event.
>
> If we
Kirti?
On Wed, 15 Nov 2017 21:11:42 -0700
Alex Williamson wrote:
> On Thu, 16 Nov 2017 11:21:56 +0800
> Zhenyu Wang wrote:
>
> > On 2017.11.15 11:48:42 -0700, Alex Williamson wrote:
> > > On Wed, 15 Nov 2017 17:11:54 +0800
> > > Tina
== Series Details ==
Series: series starting with [1/2] drm/i915/execlists: Listen to COMPLETE
context event not ACTIVE_IDLE
URL : https://patchwork.freedesktop.org/series/34061/
State : success
== Summary ==
Test kms_flip:
Subgroup dpms-vs-vblank-race:
pass ->
== Series Details ==
Series: series starting with [1/2] drm/i915/execlists: Listen to COMPLETE
context event not ACTIVE_IDLE
URL : https://patchwork.freedesktop.org/series/34061/
State : success
== Summary ==
Series 34061v1 series starting with [1/2] drm/i915/execlists: Listen to
COMPLETE
Since commit e1fee72c2ea2e9c0c6e6743d32a6832f21337d6c
Author: Oscar Mateo
Date: Thu Jul 24 17:04:40 2014 +0100
drm/i915/bdw: Avoid non-lite-restore preemptions
execlists has listened to (ACTIVE_IDLE | ELEMENT_SWITCH) for detecting
when one context completed and it
The hardware needs some time to process the information received in the
ExecList Submission Port, and expects us to don't write anything new until
it has 'acknowledged' this new execlist by sending an IDLE_ACTIVE or
PREEMPTED CSB event.
If we do not follow this, the driver could write new data
== Series Details ==
Series: series starting with [CI,1/9] drm/i915/execlists: Listen to COMPLETE
context event not ACTIVE_IDLE
URL : https://patchwork.freedesktop.org/series/34058/
State : warning
== Summary ==
Test gem_eio:
Subgroup in-flight-internal:
skip ->
Since commit e1fee72c2ea2e9c0c6e6743d32a6832f21337d6c
Author: Oscar Mateo
Date: Thu Jul 24 17:04:40 2014 +0100
drm/i915/bdw: Avoid non-lite-restore preemptions
execlists has listened to (ACTIVE_IDLE | ELEMENT_SWITCH) for detecting
when one context completed and it
== Series Details ==
Series: series starting with [CI,1/3] drm/i915/execlists: Listen to COMPLETE
context event not ACTIVE_IDLE
URL : https://patchwork.freedesktop.org/series/34057/
State : success
== Summary ==
Test perf:
Subgroup polling:
pass -> FAIL
Quoting Chris Wilson (2017-11-18 01:05:24)
> Since commit e1fee72c2ea2e9c0c6e6743d32a6832f21337d6c
> Author: Oscar Mateo
> Date: Thu Jul 24 17:04:40 2014 +0100
>
> drm/i915/bdw: Avoid non-lite-restore preemptions
>
> execlists has listened to (ACTIVE_IDLE |
== Series Details ==
Series: series starting with [CI,1/9] drm/i915/execlists: Listen to COMPLETE
context event not ACTIVE_IDLE
URL : https://patchwork.freedesktop.org/series/34058/
State : success
== Summary ==
Series 34058v1 series starting with [CI,1/9] drm/i915/execlists: Listen to
== Series Details ==
Series: series starting with [CI,1/3] drm/i915/execlists: Listen to COMPLETE
context event not ACTIVE_IDLE
URL : https://patchwork.freedesktop.org/series/34057/
State : success
== Summary ==
Series 34057v1 series starting with [CI,1/3] drm/i915/execlists: Listen to
== Series Details ==
Series: drm/i915/execlists: Listen to COMPLETE context event not ACTIVE_IDLE
(rev3)
URL : https://patchwork.freedesktop.org/series/34044/
State : failure
== Summary ==
Series 34044v3 drm/i915/execlists: Listen to COMPLETE context event not
ACTIVE_IDLE
Having disabled the broken semaphores on Sandybridge, there is no need
for a modparam any more, so remove it in favour of a simple
HAS_LEGACY_SEMAPHORES() guard.
Signed-off-by: Chris Wilson
Cc: Joonas Lahtinen
Reviewed-by: Maarten
As the semaphores is just part of the engine, include it with the
general pretty printer universally used for debugging.
Signed-off-by: Chris Wilson
Cc: Joonas Lahtinen
Reviewed-by: Joonas Lahtinen
---
I should have admitted defeat long ago as there has been a rare but
persistent error on Sandybridge where semaphore signaling did not
propagate to the waiter, leading to a GPU hang.
With the work on fence signaling for v4.9, the impact of using CPU driven
signaling was greatly reduced wrt to the
The legacy context switch for ringbuffer submission is multistaged,
where each of those stages may fail. However, we were updating global
state after some stages, and so we had to force the incomplete request
to be submitted because we could not unwind. Save the global state
before performing the
During request construction, after pinning the context we know whether
or not we have to emit a context switch. So move this common operation
from every caller into i915_gem_request_alloc() itself.
v2: Always submit the request if we emitted some commands during request
construction, as typically
Execlists and legacy ringbuffer submission are no longer feature
comparable (execlists now offer greater functionality that should
overcome their performance hit) and obsoletes the unsafe module
parameter, i.e. comparing the two modes of execution is no longer
useful, so remove the debug tool.
Since removing the module parameter to force selection of ringbuffer
emission for gen8, the code is defunct. Remove it.
To put the difference into perspective, a couple of microbenchmarks
(bdw i7-5557u, 20170324):
ring execlists
exec continuous
Since commit e1fee72c2ea2e9c0c6e6743d32a6832f21337d6c
Author: Oscar Mateo
Date: Thu Jul 24 17:04:40 2014 +0100
drm/i915/bdw: Avoid non-lite-restore preemptions
execlists has listened to (ACTIVE_IDLE | ELEMENT_SWITCH) for detecting
when one context completed and it
Since commit e1fee72c2ea2e9c0c6e6743d32a6832f21337d6c
Author: Oscar Mateo
Date: Thu Jul 24 17:04:40 2014 +0100
drm/i915/bdw: Avoid non-lite-restore preemptions
execlists has listened to (ACTIVE_IDLE | ELEMENT_SWITCH) for detecting
when one context completed and it
As the request now may implicitly invoke a context-switch, we should
follow that with a GPU TLB invalidation. Also even before using GGTT, we
should invalidate the TLBs for any updates (as well as the ppgtt
invalidates that are unconditionally applied by execbuf). Since we
almost always require
As the request now may implicitly invoke a context-switch, we should
follow that with a GPU TLB invalidation. Also even before using GGTT, we
should invalidate the TLBs for any updates (as well as the ppgtt
invalidates that are unconditionally applied by execbuf). Since we
almost always require
During request construction, after pinning the context we know whether
or not we have to emit a context switch. So move this common operation
from every caller into i915_gem_request_alloc() itself.
v2: Always submit the request if we emitted some commands during request
construction, as typically
Since commit e1fee72c2ea2e9c0c6e6743d32a6832f21337d6c
Author: Oscar Mateo
Date: Thu Jul 24 17:04:40 2014 +0100
drm/i915/bdw: Avoid non-lite-restore preemptions
execlists has listened to (ACTIVE_IDLE | ELEMENT_SWITCH) for detecting
when one context completed and it
32 matches
Mail list logo