== Series Details ==
Series: drm/i915/execlists: Listen to COMPLETE context event not ACTIVE_IDLE
(rev2)
URL : https://patchwork.freedesktop.org/series/34044/
State : success
== Summary ==
Warning: bzip CI_DRM_3359/shard-glkb6/results1.json.bz2 wasn't in correct JSON
format
Test kms_flip:
== Series Details ==
Series: drm/i915/lrc: Stop writing to ELSP until HW has processed the previous
write
URL : https://patchwork.freedesktop.org/series/34043/
State : warning
== Summary ==
Warning: bzip CI_DRM_3359/shard-glkb6/results1.json.bz2 wasn't in correct JSON
format
Test kms_flip:
== 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/34049/
State : warning
== Summary ==
Series 34049v1 series starting with [CI,1/9] drm/i915/execlists: Listen to
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
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
== Series Details ==
Series: drm/i915/execlists: Listen to COMPLETE context event not ACTIVE_IDLE
(rev2)
URL : https://patchwork.freedesktop.org/series/34044/
State : success
== Summary ==
Series 34044v2 drm/i915/execlists: Listen to COMPLETE context event not
ACTIVE_IDLE
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
---
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
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.
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
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
== Series Details ==
Series: drm/i915/execlists: Listen to COMPLETE context event not ACTIVE_IDLE
URL : https://patchwork.freedesktop.org/series/34044/
State : success
== Summary ==
Series 34044v1 drm/i915/execlists: Listen to COMPLETE context event not
ACTIVE_IDLE
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 Michel Thierry (2017-11-18 00:53:57)
> On 11/17/2017 4:31 PM, Chris Wilson wrote:
> > Since its inception, execlists has listened to (ACTIVE_IDLE |
> > ELEMENT_SWITCH) for detecting when one context completed and it either
> > continued onto the next (in port 1) or idled. We would always
On 11/17/2017 4:31 PM, Chris Wilson wrote:
Since its inception, execlists has listened to (ACTIVE_IDLE |
ELEMENT_SWITCH) for detecting when one context completed and it either
continued onto the next (in port 1) or idled. We would always see
COMPLETE | ACTIVE_IDLE on the final context-switch
== Series Details ==
Series: drm/i915/lrc: Stop writing to ELSP until HW has processed the previous
write
URL : https://patchwork.freedesktop.org/series/34043/
State : success
== Summary ==
Series 34043v1 drm/i915/lrc: Stop writing to ELSP until HW has processed the
previous write
Reviewed-by: Lyude Paul
On Thu, 2017-11-16 at 13:45 +0100, Maarten Lankhorst wrote:
> Signed-off-by: Maarten Lankhorst
> ---
> tests/Makefile.am | 6 +++---
> tests/intel-ci/fast-feedback.testlist | 18
Quoting Michel Thierry (2017-11-18 00:30:38)
> 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
From: John Harrison
The 'request pre-emption' GuC command seems to be slower than other
commands. It typically takes 20-30us on a GP-MRB system (BXT). That
means that the super-fast busy-spin wait in the GuC send action code
hits the 10us time out. It then drops
From: John Harrison
While working on a customer project, it was noticed that the time
taken to issue a pre-emption request to the GuC would vary quite
significantly. The correct case was low microseconds but the worst
case was tens of milliseconds. Two separate issues
From: John Harrison
There is a mutex_lock in the GuC send action code path to ensure
serialised access to the host-to-GuC mechanism. Acquiring the lock
apparently sees random stalls of around 6ms. That is even when the
lock is definitely not acquired by any other
Since its inception, execlists has listened to (ACTIVE_IDLE |
ELEMENT_SWITCH) for detecting when one context completed and it either
continued onto the next (in port 1) or idled. We would always see
COMPLETE | ACTIVE_IDLE on the final context-switch event, but on recent
gen it appears that we now
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
Reviewed-by: Lyude Paul
On Thu, 2017-11-16 at 13:45 +0100, Maarten Lankhorst wrote:
> igt_output_from_connector should be used for disconnected outputs
> too, this is useful for chamelium testing, where disconnected outputs
> may reappear.
>
> Signed-off-by: Maarten Lankhorst
Reviewed-by: Lyude Paul
On Thu, 2017-11-16 at 13:45 +0100, Maarten Lankhorst wrote:
> Instead of first calling kmstest_unset_all_crtcs, and calling
> igt_display_init for each test, use igt_display_reset to reset
> igt_display between tests, and use atomic commit to disable all
Reviewed-by: Lyude Paul
On Thu, 2017-11-16 at 13:45 +0100, Maarten Lankhorst wrote:
> A lot of code duplicates this, but it should be handled in the core.
> Add it and use it after igt_display_init(), the tests have to be
> converted one by one.
>
> Signed-off-by: Maarten
Hi all,
The following changes tagged drm-intel-testing-2017-11-17-2:
drm-intel-next-2017-11-17-1:
More change sets for 4.16:
- Many improvements for selftests and other igt tests (Chris)
- Forcewake with PUNIT->PMIC bus fixes and robustness (Hans)
- Define an engine class for uABI (Tvrtko)
-
On Fri, Nov 17, 2017 at 08:32:19PM +, Ville Syrjälä wrote:
> On Fri, Nov 17, 2017 at 12:12:15PM -0800, Rodrigo Vivi wrote:
> > On Fri, Nov 17, 2017 at 07:19:10PM +, Ville Syrjala wrote:
> > > From: Ville Syrjälä
> > >
> > > Rename enum plane to enum
== Series Details ==
Series: drm/i915: Plane assert/readout cleanups etc. (rev9)
URL : https://patchwork.freedesktop.org/series/31758/
State : warning
== Summary ==
Test kms_setmode:
Subgroup basic:
pass -> FAIL (shard-hsw) fdo#99912
Test drv_module_reload:
On Fri, Nov 17, 2017 at 12:12:15PM -0800, Rodrigo Vivi wrote:
> On Fri, Nov 17, 2017 at 07:19:10PM +, Ville Syrjala wrote:
> > From: Ville Syrjälä
> >
> > Rename enum plane to enum i9xx_plane_id to make it clear that it only
> > applies to pre-SKL platforms.
>
On Fri, Nov 17, 2017 at 07:19:10PM +, Ville Syrjala wrote:
> From: Ville Syrjälä
>
> Rename enum plane to enum i9xx_plane_id to make it clear that it only
> applies to pre-SKL platforms.
Oh! I should had read this before commenting on the cover letter...
On Fri, Nov 17, 2017 at 07:19:07PM +, Ville Syrjala wrote:
> From: Ville Syrjälä
>
> OK, one more time. This time with s/plane/i9xx_plane/ etc. all over.
why not intel_plane?!
> Maybe that will make everyone happy? Unlikely, but let's try.
:)
>
> Patch 3
== Series Details ==
Series: drm/i915: Plane assert/readout cleanups etc. (rev9)
URL : https://patchwork.freedesktop.org/series/31758/
State : success
== Summary ==
Series 31758v9 drm/i915: Plane assert/readout cleanups etc.
From: Ville Syrjälä
Check that the planes are in the state we expect them to be. For
now we can only check whether each plane is correctly enabled or
disabled. In the future we may want to expand the plane state
readout to support a more thorough verification.
v2:
From: Ville Syrjälä
Since we now have a ->get_hw_state() method for planes, let's use
that during the initial plane fb readout.
v2: s/plane/i9xx_plane/ etc. (James)
Cc: James Ausmus
Cc: Daniel Vetter
Suggested-by:
From: Ville Syrjälä
The only relevant difference between i9xx_get_initial_plane_config() and
ironlake_get_initial_plane_config() is the HSW/BDW TILEOFF handling.
Add that to i9xx_get_initial_plane_config() and nuke
ironlake_get_initial_plane_config().
v2:
From: Ville Syrjälä
OK, one more time. This time with s/plane/i9xx_plane/ etc. all over.
Maybe that will make everyone happy? Unlikely, but let's try.
Patch 3 is the only one missing r-b.
Ville Syrjälä (10):
drm/i915: Add .get_hw_state() method for planes
From: Ville Syrjälä
Stop using the old for_each_intel_plane_in_state() type iteration
macro and replace it with for_each_new_intel_plane_in_state().
And similarly replace drm_atomic_get_existing_crtc_state() with
intel_atomic_get_new_crtc_state(). Switch over to
From: Ville Syrjälä
Unify the plane disabling during state readout by pulling the code into
a new helper intel_plane_disable_noatomic(). We'll also read out the
state of all planes, so that we know which planes really need to be
diabled.
Additonally we change the
From: Ville Syrjälä
Eliminate crtc->plane since it's pretty much a layering violation.
We can always get the plane via crtc->primary if we actually need it.
The only ugly thing left is plane_to_crtc_mapping[], but that's
still needed by the pre-g4x watermark code.
From: Ville Syrjälä
Rename enum plane to enum i9xx_plane_id to make it clear that it only
applies to pre-SKL platforms.
enum i9xx_plane_id is a global identifier, whereas enum plane_id is
per-pipe. We need the old global identifier to index the primary plane
(and
From: Ville Syrjälä
Replace the 0 and 1 with PLANE_A and PLANE_B in the pre-g4x wm code.
v2: s/old_plane_id/i9xx_plane_id/ (Daniel)
v3: s/plane/i9xx_plane/ etc. (James)
Cc: James Ausmus
Cc: Daniel Vetter
From: Ville Syrjälä
Add a .get_hw_state() method for planes, returning true or false
depending on whether the plane is enabled. Use it to rewrite the
plane enabled/disabled asserts in platform agnostic fashion.
We do lose the pre-gen4 plane<->pipe mapping checks,
From: Ville Syrjälä
Use enum pipe, enum plane_id, and enum i9xx_plane_id consistently in the
initial framebuffe readout.
v2: Use old_plane_id in the ilk code
v3: s/old_plane_id/i9xx_plane_id/ (Daniel)
v4: Rebase due to GLK/CNL PLANE_COLOR_CTL alpha stuff
v5:
== Series Details ==
Series: drm/i915: Fix init_clock_gating for resume (rev2)
URL : https://patchwork.freedesktop.org/series/33718/
State : success
== Summary ==
Series 33718v2 drm/i915: Fix init_clock_gating for resume
https://patchwork.freedesktop.org/api/1.0/series/33718/revisions/2/mbox/
On Tue, Nov 07, 2017 at 06:40:08PM +, Ramalingam C wrote:
> From: "C, Ramalingam"
>
> Existing debugfs entry i915_drrs_status is updated with crtc id and
> if PSR is cause for DRRS disabled state.
>
> [v2]: Dropped the module parameter details as ctl moved from
On Tue, Nov 07, 2017 at 06:38:23PM +, Ramalingam C wrote:
> From: "C, Ramalingam"
>
> Debugfs called i915_drrs_ctl is added to enable and disable the
> eDP DRRS. Writing 0 will disable the feature, whereas non-zero
> will enable the feature.
>
> Possibility of
Quoting Patchwork (2017-11-17 18:37:16)
> == Series Details ==
>
> Series: drm/i915/selftests: Report ENOMEM clearly for an allocation failure
> (rev3)
> URL : https://patchwork.freedesktop.org/series/33994/
> State : success
>
> == Logs ==
>
> For more details see:
>
== Series Details ==
Series: drm/i915/selftests: Report ENOMEM clearly for an allocation failure
(rev3)
URL : https://patchwork.freedesktop.org/series/33994/
State : success
== Summary ==
Test kms_frontbuffer_tracking:
Subgroup fbc-1p-primscrn-shrfb-pgflip-blt:
pass
== Series Details ==
Series: drm/i915: Don't use GEN6_RC_VIDEO_FREQ on gen10+ (rev2)
URL : https://patchwork.freedesktop.org/series/33608/
State : success
== Summary ==
Series 33608v2 drm/i915: Don't use GEN6_RC_VIDEO_FREQ on gen10+
== Series Details ==
Series: drm/i915: Enable fastboot, v2!
URL : https://patchwork.freedesktop.org/series/34010/
State : success
== Summary ==
Test drv_selftest:
Subgroup mock_sanitycheck:
pass -> DMESG-WARN (shard-hsw) fdo#103719
Test kms_frontbuffer_tracking:
On Fri, Nov 17, 2017 at 03:37:53PM +, Maarten Lankhorst wrote:
> Small fixes for IPS, and then we flip the switch! :-)
>
> Maarten Lankhorst (3):
> drm/i915: Make ips_enabled a property depending on whether IPS is
> enabled.
> drm/i915: Enable IPS for sprite plane
> drm/i915:
On Fri, Nov 17, 2017 at 01:08:25AM +, Radhakrishna Sripada wrote:
> This reverts commit 8f067837c4b713ce2e69be95af7b2a5eb3bd7de8.
>
> HSD says "WA withdrawn. It was causing corruption with some images.
> WA is not strictly necessary since this bug just causes loss of FBC
> compression with
Quoting Rodrigo Vivi (2017-11-17 17:33:49)
> On Fri, Nov 17, 2017 at 11:11:28AM +, Jani Nikula wrote:
> > On Fri, 17 Nov 2017, Chris Wilson wrote:
> > > Rodrigo gave a persuasive argument for keeping workarounds: that they
> > > serve as a good guide for the bring up
On Fri, Nov 17, 2017 at 02:02:51AM +, Zhenyu Wang wrote:
> On 2017.11.16 12:20:14 -0800, Rodrigo Vivi wrote:
> > Hi Zhenyu,
> >
> > On Thu, Nov 16, 2017 at 09:20:07AM +, Zhenyu Wang wrote:
> > >
> > > Hi,
> > >
> > > As we missed 4.15 cycle, here's the bigger initial 4.16 gvt-next pull,
== Series Details ==
Series: series starting with [CI,1/2] drm/i915: Pull the unconditional GPU
cache invalidation into request construction
URL : https://patchwork.freedesktop.org/series/34016/
State : warning
== Summary ==
Series 34016v1 series starting with [CI,1/2] drm/i915: Pull the
On Fri, Nov 17, 2017 at 11:11:28AM +, Jani Nikula wrote:
> On Fri, 17 Nov 2017, Chris Wilson wrote:
> > Rodrigo gave a persuasive argument for keeping workarounds: that they
> > serve as a good guide for the bring up of the next generation. Not only
> > do
On Fri, Nov 17, 2017 at 01:21:35PM +, Ville Syrjälä wrote:
> On Thu, Nov 16, 2017 at 12:49:23PM -0800, Rodrigo Vivi wrote:
> > On Thu, Nov 16, 2017 at 07:14:47PM +, Ville Syrjala wrote:
> > > From: Ville Syrjälä
> > >
> > > The current code is trying to be
== Series Details ==
Series: drm/i915/selftests: Report ENOMEM clearly for an allocation failure
(rev3)
URL : https://patchwork.freedesktop.org/series/33994/
State : success
== Summary ==
Series 33994v3 drm/i915/selftests: Report ENOMEM clearly for an allocation
failure
== Series Details ==
Series: drm/i915/selftests: Report ENOMEM clearly for an allocation failure
(rev2)
URL : https://patchwork.freedesktop.org/series/33994/
State : success
== Summary ==
Test kms_busy:
Subgroup extended-modeset-hang-newfb-with-reset-render-a:
skip
== Series Details ==
Series: drm/i915: Pull the unconditional GPU cache invalidation into request
construction (rev2)
URL : https://patchwork.freedesktop.org/series/34007/
State : warning
== Summary ==
Series 34007v2 drm/i915: Pull the unconditional GPU cache invalidation into
request
== Series Details ==
Series: drm/i915: Enable fastboot, v2!
URL : https://patchwork.freedesktop.org/series/34010/
State : success
== Summary ==
Series 34010v1 drm/i915: Enable fastboot, v2!
https://patchwork.freedesktop.org/api/1.0/series/34010/revisions/1/mbox/
Test chamelium:
If we can not run the drunk_hole test because we couldn't allocate the
memory for the permutation array (even after we tried trimming the
size), report a clear ENOMEM. Similary, if we are asked to operate on a
hole too small for ourselves, make it skip quietly.
v2: Avoid malloc(0) since that
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
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
== Series Details ==
Series: drm/i915/selftests: Report ENOMEM clearly for an allocation failure
(rev2)
URL : https://patchwork.freedesktop.org/series/33994/
State : success
== Summary ==
Series 33994v2 drm/i915/selftests: Report ENOMEM clearly for an allocation
failure
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
== Series Details ==
Series: drm/i915: Expose more GPU properties through sysfs (rev2)
URL : https://patchwork.freedesktop.org/series/33950/
State : success
== Summary ==
Test kms_frontbuffer_tracking:
Subgroup fbc-1p-offscren-pri-shrfb-draw-blt:
fail -> PASS
Quoting Chris Wilson (2017-11-17 15:23:58)
> @@ -1818,8 +1814,7 @@ static int eb_move_to_gpu(struct i915_execbuffer *eb)
> /* Unconditionally flush any chipset caches (for streaming writes). */
> i915_gem_chipset_flush(eb->i915);
>
> - /* Unconditionally invalidate GPU
On 17 November 2017 at 15:31, Chris Wilson wrote:
> If we can not run the drunk_hole test because we couldn't allocate the
> memory for the permutation array (even after we tried trimming the
> size), report a clear ENOMEM. Similary, if we are asked to operate on a
>
== Series Details ==
Series: series starting with [CI,1/2] drm/i915: Pull the unconditional GPU
cache invalidation into request construction
URL : https://patchwork.freedesktop.org/series/34008/
State : failure
== Summary ==
Series 34008v1 series starting with [CI,1/2] drm/i915: Pull the
On 17 November 2017 at 15:31, Chris Wilson wrote:
> If we can not run the drunk_hole test because we couldn't allocate the
> memory for the permutation array (even after we tried trimming the
> size), report a clear ENOMEM. Similary, if we are asked to operate on a
>
On Fri, Nov 17, 2017 at 04:37:55PM +0100, Maarten Lankhorst wrote:
> Signed-off-by: Maarten Lankhorst
> ---
> drivers/gpu/drm/i915/intel_display.c | 13 ++---
> 1 file changed, 6 insertions(+), 7 deletions(-)
>
> diff --git
== Series Details ==
Series: drm/i915: Pull the unconditional GPU cache invalidation into request
construction
URL : https://patchwork.freedesktop.org/series/34007/
State : failure
== Summary ==
Series 34007v1 drm/i915: Pull the unconditional GPU cache invalidation into
request construction
On Fri, Nov 17, 2017 at 04:37:54PM +0100, Maarten Lankhorst wrote:
> ips_enabled was used as a variable of whether IPS can be enabled or not,
> but should be used to test whether IPS is actually enabled.
>
> Signed-off-by: Maarten Lankhorst
> ---
>
On Thu, Nov 16, 2017 at 03:21:32PM -0800, James Ausmus wrote:
> On Mon, Oct 23, 2017 at 05:50:32PM +0300, Ville Syrjala wrote:
> > From: Ville Syrjälä
> >
> > Rename enum plane to enum i9xx_plane_id to make it clear that it only
> > applies to pre-SKL platforms.
>
== Series Details ==
Series: drm/i915: Expose more GPU properties through sysfs (rev2)
URL : https://patchwork.freedesktop.org/series/33950/
State : success
== Summary ==
Series 33950v2 drm/i915: Expose more GPU properties through sysfs
Small fixes for IPS, and then we flip the switch! :-)
Maarten Lankhorst (3):
drm/i915: Make ips_enabled a property depending on whether IPS is
enabled.
drm/i915: Enable IPS for sprite plane
drm/i915: Re-enable fastboot by default
drivers/gpu/drm/i915/i915_params.h| 2 +-
ips_enabled was used as a variable of whether IPS can be enabled or not,
but should be used to test whether IPS is actually enabled.
Signed-off-by: Maarten Lankhorst
---
drivers/gpu/drm/i915/intel_display.c | 75 ---
Signed-off-by: Maarten Lankhorst
---
drivers/gpu/drm/i915/intel_display.c | 13 ++---
1 file changed, 6 insertions(+), 7 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_display.c
b/drivers/gpu/drm/i915/intel_display.c
index
This fix was originally reverted because it left a chromebook pixel
black, and no immediate fix was available. This has been fixed in the
meantime.
Rather than trying to remove the parameter, set it to default to true
for now, so we can always back out if required.
Signed-off-by: Maarten
If we can not run the drunk_hole test because we couldn't allocate the
memory for the permutation array (even after we tried trimming the
size), report a clear ENOMEM. Similary, if we are asked to operate on a
hole too small for ourselves, make it skip quietly.
v2: Avoid malloc(0) since that
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
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
On 17/11/17 10:53, Chris Wilson wrote:
Quoting Lionel Landwerlin (2017-11-16 16:00:03)
With the introduction of asymetric slices in CNL, we cannot rely on
the previous SUBSLICE_MASK getparam. Here we introduce a more detailed
way of querying the Gen's GPU topology that doesn't aggregate
With the introduction of asymetric slices in CNL, we cannot rely on
the previous SUBSLICE_MASK getparam. Here we introduce a more detailed
way of querying the Gen's GPU topology that doesn't aggregate numbers.
This is essential for monitoring parts of the GPU with the OA unit,
because signals
Now that we have that information in topology fields, let's just reused it.
Signed-off-by: Lionel Landwerlin
---
drivers/gpu/drm/i915/i915_debugfs.c | 26 ++
1 file changed, 10 insertions(+), 16 deletions(-)
diff --git
This enables userspace to discover the engines available on the GPU.
Here is the layout on a Skylake GT4:
/sys/devices/pci:00/:00:02.0/drm/card0/gt
├── bcs
│ └── 0
│ ├── capabilities
│ ├── class
│ └── id
├── rcs
│ └── 0
│ ├── capabilities
│ ├── class
│
Up to now, subslice mask was assumed to be uniform across slices. But
starting with Cannonlake, slices can be asymetric (for example slice0
has different number of subslices as slice1+). This change stores all
subslices masks for all slices rather than having a single mask that
applies to all
Hi,
An update based on Chris & Tvrtko's feedback.
Cheers,
Lionel Landwerlin (4):
drm/i915: store all subslice masks
drm/i915/debugfs: reuse max slice/subslices already stored in sseu
drm/i915: expose engine availability through sysfs
drm/i915: expose EU topology through sysfs
== Series Details ==
Series: series starting with [CI,1/2] drm/i915: Automatic i915_switch_context
for legacy
URL : https://patchwork.freedesktop.org/series/34005/
State : failure
== Summary ==
Series 34005v1 series starting with [CI,1/2] drm/i915: Automatic
i915_switch_context for legacy
Op 17-11-17 om 15:53 schreef Ville Syrjälä:
> On Fri, Nov 17, 2017 at 03:47:58PM +0100, Maarten Lankhorst wrote:
>> Op 17-11-17 om 14:31 schreef Ville Syrjälä:
>>> On Wed, Nov 15, 2017 at 05:31:56PM +0100, Maarten Lankhorst wrote:
The watermarks it should calculate against are the old optimal
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
On Fri, Nov 17, 2017 at 03:47:58PM +0100, Maarten Lankhorst wrote:
> Op 17-11-17 om 14:31 schreef Ville Syrjälä:
> > On Wed, Nov 15, 2017 at 05:31:56PM +0100, Maarten Lankhorst wrote:
> >> The watermarks it should calculate against are the old optimal watermarks.
> >> The currently active crtc
Op 17-11-17 om 14:31 schreef Ville Syrjälä:
> On Wed, Nov 15, 2017 at 05:31:56PM +0100, Maarten Lankhorst wrote:
>> The watermarks it should calculate against are the old optimal watermarks.
>> The currently active crtc watermarks are pure fiction, and are invalid in
>> case of a nonblocking
== Series Details ==
Series: drm/i915: Automatic i915_switch_context for legacy
URL : https://patchwork.freedesktop.org/series/34002/
State : failure
== Summary ==
Test gem_exec_reloc:
Subgroup basic-range:
pass -> INCOMPLETE (shard-snb)
Subgroup
1 - 100 of 162 matches
Mail list logo