[Intel-gfx] [PATCH] drm/i915/gt: Only delay the barrier pm

[Intel-gfx] [PATCH] drm/i915/gt: Only delay the barrier pm

[Chris Wilson]

It is strictly sufficient to only delay the intel_engine_pm_put from the
context barrier (and not from the context exit) in order to prevent the
gem_exec_nop contention. Adding the delay to the context exit incurs
noticably extra penalty for soft-rc6.

Fixes: edee52c927ef ("drm/i915/gt: Delay release of engine-pm after last retirement")
Testcase: igt/i915_pm_rc6_residency/rc6-idle
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
---
 drivers/gpu/drm/i915/gt/intel_context.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/gt/intel_context.c b/drivers/gpu/drm/i915/gt/intel_context.c
index 622ff425fce9..e4aece20bc80 100644
--- a/drivers/gpu/drm/i915/gt/intel_context.c
+++ b/drivers/gpu/drm/i915/gt/intel_context.c
@@ -350,7 +350,7 @@ void intel_context_enter_engine(struct intel_context *ce)
 void intel_context_exit_engine(struct intel_context *ce)
 {
 	intel_timeline_exit(ce->timeline);
-	intel_engine_pm_put_delay(ce->engine, 1);
+	intel_engine_pm_put(ce->engine);
 }
 
 int intel_context_prepare_remote_request(struct intel_context *ce,
-- 
2.20.1

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] ✗ Fi.CI.BAT: failure for drm/i915/gt: Only delay the barrier pm

[Patchwork]

== Series Details ==

Series: drm/i915/gt: Only delay the barrier pm
URL   : https://patchwork.freedesktop.org/series/74986/
State : failure

== Summary ==

CI Bug Log - changes from CI_DRM_8180 -> Patchwork_17060
====================================================

Summary
-------

  **FAILURE**

  Serious unknown changes coming with Patchwork_17060 absolutely need to be
  verified manually.
  
  If you think the reported changes have nothing to do with the changes
  introduced in Patchwork_17060, please notify your bug team to allow them
  to document this new failure mode, which will reduce false positives in CI.

  External URL: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17060/index.html

Possible new issues
-------------------

  Here are the unknown changes that may have been introduced in Patchwork_17060:

### IGT changes ###

#### Possible regressions ####

  * igt@gem_exec_gttfill@basic:
    - fi-skl-6770hq:      [PASS][1] -> [INCOMPLETE][2]
   [1]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8180/fi-skl-6770hq/igt@gem_exec_gttfill@basic.html
   [2]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17060/fi-skl-6770hq/igt@gem_exec_gttfill@basic.html

  
Known issues
------------

  Here are the changes found in Patchwork_17060 that come from known issues:

### IGT changes ###

#### Issues hit ####

  * igt@i915_selftest@live@execlists:
    - fi-icl-dsi:         [PASS][3] -> [DMESG-FAIL][4] ([fdo#108569])
   [3]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8180/fi-icl-dsi/igt@i915_selftest@live@execlists.html
   [4]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17060/fi-icl-dsi/igt@i915_selftest@live@execlists.html

  * igt@kms_chamelium@dp-crc-fast:
    - fi-cml-u2:          [PASS][5] -> [FAIL][6] ([i915#217])
   [5]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8180/fi-cml-u2/igt@kms_chamelium@dp-crc-fast.html
   [6]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17060/fi-cml-u2/igt@kms_chamelium@dp-crc-fast.html

  
#### Possible fixes ####

  * igt@i915_selftest@live@execlists:
    - fi-bxt-dsi:         [INCOMPLETE][7] ([fdo#103927] / [i915#656]) -> [PASS][8]
   [7]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_8180/fi-bxt-dsi/igt@i915_selftest@live@execlists.html
   [8]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17060/fi-bxt-dsi/igt@i915_selftest@live@execlists.html

  
  [fdo#103927]: https://bugs.freedesktop.org/show_bug.cgi?id=103927
  [fdo#108569]: https://bugs.freedesktop.org/show_bug.cgi?id=108569
  [i915#217]: https://gitlab.freedesktop.org/drm/intel/issues/217
  [i915#656]: https://gitlab.freedesktop.org/drm/intel/issues/656


Participating hosts (40 -> 40)
------------------------------

  Additional (8): fi-bdw-5557u fi-kbl-7500u fi-gdg-551 fi-elk-e7500 fi-skl-lmem fi-blb-e6850 fi-skl-6600u fi-snb-2600 
  Missing    (8): fi-byt-squawks fi-glk-dsi fi-bsw-cyan fi-ctg-p8600 fi-cfl-8109u fi-bsw-kefka fi-byt-clapper fi-bdw-samus 


Build changes
-------------

  * CI: CI-20190529 -> None
  * Linux: CI_DRM_8180 -> Patchwork_17060

  CI-20190529: 20190529
  CI_DRM_8180: 257af0ddcea3a234dcb79579600f971edd47353f @ git://anongit.freedesktop.org/gfx-ci/linux
  IGT_5530: 2020d743940f06294d06006bb737be43fcd2881e @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
  Patchwork_17060: ac635589f9e6de1215adcdc7776f40df010e160b @ git://anongit.freedesktop.org/gfx-ci/linux


== Linux commits ==

ac635589f9e6 drm/i915/gt: Only delay the barrier pm

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_17060/index.html
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [Intel-gfx] [PATCH] drm/i915/gt: Only delay the barrier pm

[Tvrtko Ursulin]

On 23/03/2020 19:20, Chris Wilson wrote:
> It is strictly sufficient to only delay the intel_engine_pm_put from the
> context barrier (and not from the context exit) in order to prevent the
> gem_exec_nop contention. Adding the delay to the context exit incurs
> noticably extra penalty for soft-rc6.
> 
> Fixes: edee52c927ef ("drm/i915/gt: Delay release of engine-pm after last retirement")
> Testcase: igt/i915_pm_rc6_residency/rc6-idle
> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> ---
>   drivers/gpu/drm/i915/gt/intel_context.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/i915/gt/intel_context.c b/drivers/gpu/drm/i915/gt/intel_context.c
> index 622ff425fce9..e4aece20bc80 100644
> --- a/drivers/gpu/drm/i915/gt/intel_context.c
> +++ b/drivers/gpu/drm/i915/gt/intel_context.c
> @@ -350,7 +350,7 @@ void intel_context_enter_engine(struct intel_context *ce)
>   void intel_context_exit_engine(struct intel_context *ce)
>   {
>   	intel_timeline_exit(ce->timeline);
> -	intel_engine_pm_put_delay(ce->engine, 1);
> +	intel_engine_pm_put(ce->engine);
>   }
>   
>   int intel_context_prepare_remote_request(struct intel_context *ce,
> 

I wanted to ask during review about the effect on soft rc6 but thought 
you looked at it.

Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

Regards,

Tvrtko
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx