From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: Chris Wilson <chris@chris-wilson.co.uk>, intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH 6/9] drm/i915/selftests: Force bonded submission to overlap
Date: Wed, 20 Nov 2019 13:18:27 +0000 [thread overview]
Message-ID: <5eda82de-a7a3-d4d3-13ec-6ed4b70986b7@linux.intel.com> (raw)
In-Reply-To: <157425477060.13839.10577920977636430794@skylake-alporthouse-com>
On 20/11/2019 12:59, Chris Wilson wrote:
> Quoting Tvrtko Ursulin (2019-11-20 12:55:42)
>>
>> On 20/11/2019 09:32, Chris Wilson wrote:
>>> Bonded request submission is designed to allow requests to execute in
>>> parallel as laid out by the user. If the master request is already
>>> finished before its bonded pair is submitted, the pair were not destined
>>> to run in parallel and we lose the information about the master engine
>>> to dictate selection of the secondary. If the second request was
>>> required to be run on a particular engine in a virtual set, that should
>>> have been specified, rather than left to the whims of a random
>>> unconnected requests!
>>>
>>> In the selftest, I made the mistake of not ensuring the master would
>>> overlap with its bonded pairs, meaning that it could indeed complete
>>> before we submitted the bonds. Those bonds were then free to select any
>>> available engine in their virtual set, and not the one expected by the
>>> test.
>>
>> There is a submit await which ensures master is not runnable before
>> bonded pairs are submitted. Why was that not enough? Are the sporadic
>> test failures?
>
> One test is using the submit_await, the other does not. It takes the
> background retire worker to run as we are submitting the secondaries...
> But I have not noticed this failure before hooking up retirement to
> process_csb. However, the issue is definitely present in the current
> test.
So what happens? Is this another issue limited to selftests? Because I
don't see that uAPI itself can't be used in this way.
Regards,
Tvrtko
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
WARNING: multiple messages have this Message-ID (diff)
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: Chris Wilson <chris@chris-wilson.co.uk>, intel-gfx@lists.freedesktop.org
Subject: Re: [Intel-gfx] [PATCH 6/9] drm/i915/selftests: Force bonded submission to overlap
Date: Wed, 20 Nov 2019 13:18:27 +0000 [thread overview]
Message-ID: <5eda82de-a7a3-d4d3-13ec-6ed4b70986b7@linux.intel.com> (raw)
Message-ID: <20191120131827.5EsOBq2W6FXYZolGMolDkhkXA4XLjY3JOvvJsTVzy_U@z> (raw)
In-Reply-To: <157425477060.13839.10577920977636430794@skylake-alporthouse-com>
On 20/11/2019 12:59, Chris Wilson wrote:
> Quoting Tvrtko Ursulin (2019-11-20 12:55:42)
>>
>> On 20/11/2019 09:32, Chris Wilson wrote:
>>> Bonded request submission is designed to allow requests to execute in
>>> parallel as laid out by the user. If the master request is already
>>> finished before its bonded pair is submitted, the pair were not destined
>>> to run in parallel and we lose the information about the master engine
>>> to dictate selection of the secondary. If the second request was
>>> required to be run on a particular engine in a virtual set, that should
>>> have been specified, rather than left to the whims of a random
>>> unconnected requests!
>>>
>>> In the selftest, I made the mistake of not ensuring the master would
>>> overlap with its bonded pairs, meaning that it could indeed complete
>>> before we submitted the bonds. Those bonds were then free to select any
>>> available engine in their virtual set, and not the one expected by the
>>> test.
>>
>> There is a submit await which ensures master is not runnable before
>> bonded pairs are submitted. Why was that not enough? Are the sporadic
>> test failures?
>
> One test is using the submit_await, the other does not. It takes the
> background retire worker to run as we are submitting the secondaries...
> But I have not noticed this failure before hooking up retirement to
> process_csb. However, the issue is definitely present in the current
> test.
So what happens? Is this another issue limited to selftests? Because I
don't see that uAPI itself can't be used in this way.
Regards,
Tvrtko
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
next prev parent reply other threads:[~2019-11-20 13:18 UTC|newest]
Thread overview: 72+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-11-20 9:32 [PATCH 1/9] drm/i915/selftests: Take a ref to the request we wait upon Chris Wilson
2019-11-20 9:32 ` [Intel-gfx] " Chris Wilson
2019-11-20 9:32 ` [PATCH 2/9] drm/i915/gt: Close race between engine_park and intel_gt_retire_requests Chris Wilson
2019-11-20 9:32 ` [Intel-gfx] " Chris Wilson
2019-11-20 13:19 ` Tvrtko Ursulin
2019-11-20 13:19 ` [Intel-gfx] " Tvrtko Ursulin
2019-11-20 9:32 ` [PATCH 3/9] drm/i915/gt: Unlock engine-pm after queuing the kernel context switch Chris Wilson
2019-11-20 9:32 ` [Intel-gfx] " Chris Wilson
2019-11-20 11:58 ` Tvrtko Ursulin
2019-11-20 11:58 ` [Intel-gfx] " Tvrtko Ursulin
2019-11-20 12:07 ` Chris Wilson
2019-11-20 12:07 ` [Intel-gfx] " Chris Wilson
2019-11-20 12:40 ` Tvrtko Ursulin
2019-11-20 12:40 ` [Intel-gfx] " Tvrtko Ursulin
2019-11-20 12:44 ` Chris Wilson
2019-11-20 12:44 ` [Intel-gfx] " Chris Wilson
2019-11-20 13:19 ` Tvrtko Ursulin
2019-11-20 13:19 ` [Intel-gfx] " Tvrtko Ursulin
2019-11-20 9:32 ` [PATCH 4/9] drm/i915: Mark up the calling context for intel_wakeref_put() Chris Wilson
2019-11-20 9:32 ` [Intel-gfx] " Chris Wilson
2019-11-20 12:46 ` Tvrtko Ursulin
2019-11-20 12:46 ` [Intel-gfx] " Tvrtko Ursulin
2019-11-20 9:32 ` [PATCH 5/9] drm/i915/gt: Declare timeline.lock to be irq-free Chris Wilson
2019-11-20 9:32 ` [Intel-gfx] " Chris Wilson
2019-11-20 9:32 ` [PATCH 6/9] drm/i915/selftests: Force bonded submission to overlap Chris Wilson
2019-11-20 9:32 ` [Intel-gfx] " Chris Wilson
2019-11-20 12:55 ` Tvrtko Ursulin
2019-11-20 12:55 ` [Intel-gfx] " Tvrtko Ursulin
2019-11-20 12:59 ` Chris Wilson
2019-11-20 12:59 ` [Intel-gfx] " Chris Wilson
2019-11-20 13:18 ` Tvrtko Ursulin [this message]
2019-11-20 13:18 ` Tvrtko Ursulin
2019-11-20 13:29 ` Chris Wilson
2019-11-20 13:29 ` [Intel-gfx] " Chris Wilson
2019-11-22 9:34 ` Tvrtko Ursulin
2019-11-22 9:34 ` [Intel-gfx] " Tvrtko Ursulin
2019-11-22 10:03 ` Chris Wilson
2019-11-22 10:03 ` [Intel-gfx] " Chris Wilson
2019-11-22 10:43 ` [PATCH] " Chris Wilson
2019-11-22 10:43 ` [Intel-gfx] " Chris Wilson
2019-11-20 9:33 ` [PATCH 7/9] drm/i915/selftests: Flush the active callbacks Chris Wilson
2019-11-20 9:33 ` [Intel-gfx] " Chris Wilson
2019-11-20 9:33 ` [PATCH 8/9] drm/i915/selftests: Be explicit in ERR_PTR handling Chris Wilson
2019-11-20 9:33 ` [Intel-gfx] " Chris Wilson
2019-11-20 10:23 ` Matthew Auld
2019-11-20 10:23 ` [Intel-gfx] " Matthew Auld
2019-11-20 9:33 ` [PATCH 9/9] drm/i915/gt: Schedule request retirement when timeline idles Chris Wilson
2019-11-20 9:33 ` [Intel-gfx] " Chris Wilson
2019-11-20 13:16 ` Tvrtko Ursulin
2019-11-20 13:16 ` [Intel-gfx] " Tvrtko Ursulin
2019-11-20 13:39 ` Chris Wilson
2019-11-20 13:39 ` [Intel-gfx] " Chris Wilson
2019-11-20 14:16 ` Tvrtko Ursulin
2019-11-20 14:16 ` [Intel-gfx] " Tvrtko Ursulin
2019-11-20 14:25 ` Chris Wilson
2019-11-20 14:25 ` [Intel-gfx] " Chris Wilson
2019-11-20 10:19 ` [PATCH 1/9] drm/i915/selftests: Take a ref to the request we wait upon Matthew Auld
2019-11-20 10:19 ` [Intel-gfx] " Matthew Auld
2019-11-20 10:25 ` Chris Wilson
2019-11-20 10:25 ` [Intel-gfx] " Chris Wilson
2019-11-20 10:27 ` [PATCH] " Chris Wilson
2019-11-20 10:27 ` [Intel-gfx] " Chris Wilson
2019-11-20 10:34 ` Matthew Auld
2019-11-20 10:34 ` [Intel-gfx] " Matthew Auld
2019-11-20 13:51 ` ✗ Fi.CI.CHECKPATCH: warning for series starting with drm/i915/selftests: Take a ref to the request we wait upon (rev2) Patchwork
2019-11-20 13:51 ` [Intel-gfx] " Patchwork
2019-11-20 14:19 ` ✗ Fi.CI.BAT: failure " Patchwork
2019-11-20 14:19 ` [Intel-gfx] " Patchwork
2019-11-20 14:20 ` Chris Wilson
2019-11-20 14:20 ` [Intel-gfx] " Chris Wilson
2019-11-22 12:33 ` ✗ Fi.CI.BUILD: failure for series starting with drm/i915/selftests: Take a ref to the request we wait upon (rev3) Patchwork
2019-11-22 12:33 ` [Intel-gfx] " Patchwork
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5eda82de-a7a3-d4d3-13ec-6ed4b70986b7@linux.intel.com \
--to=tvrtko.ursulin@linux.intel.com \
--cc=chris@chris-wilson.co.uk \
--cc=intel-gfx@lists.freedesktop.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.