public inbox for intel-gfx@lists.freedesktop.org
 help / color / mirror / Atom feed
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: Chris Wilson <chris@chris-wilson.co.uk>, intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH 17/21] drm/i915: Drop struct_mutex from around i915_retire_requests()
Date: Wed, 25 Sep 2019 09:49:48 +0100	[thread overview]
Message-ID: <ef7aaff6-a350-c2d9-df0b-b4e1856c5153@linux.intel.com> (raw)
In-Reply-To: <156940100781.4979.16490963525085766366@skylake-alporthouse-com>


On 25/09/2019 09:43, Chris Wilson wrote:
> Quoting Tvrtko Ursulin (2019-09-24 16:25:29)
>>
>> On 02/09/2019 05:02, Chris Wilson wrote:
>>> @@ -449,8 +447,6 @@ tl_write(struct intel_timeline *tl, struct intel_engine_cs *engine, u32 value)
>>>        struct i915_request *rq;
>>>        int err;
>>>    
>>> -     lockdep_assert_held(&tl->gt->i915->drm.struct_mutex); /* lazy rq refs */
>>> -
>>>        err = intel_timeline_pin(tl);
>>>        if (err) {
>>>                rq = ERR_PTR(err);
>>> @@ -461,10 +457,14 @@ tl_write(struct intel_timeline *tl, struct intel_engine_cs *engine, u32 value)
>>>        if (IS_ERR(rq))
>>>                goto out_unpin;
>>>    
>>> +     i915_request_get(rq);
>>> +
>>>        err = emit_ggtt_store_dw(rq, tl->hwsp_offset, value);
>>>        i915_request_add(rq);
>>> -     if (err)
>>> +     if (err) {
>>> +             i915_request_put(rq);
>>>                rq = ERR_PTR(err);
>>> +     }
>>>    
>>>    out_unpin:
>>>        intel_timeline_unpin(tl);
>>> @@ -500,7 +500,6 @@ static int live_hwsp_engine(void *arg)
>>>        struct intel_timeline **timelines;
>>>        struct intel_engine_cs *engine;
>>>        enum intel_engine_id id;
>>> -     intel_wakeref_t wakeref;
>>>        unsigned long count, n;
>>>        int err = 0;
>>>    
>>> @@ -515,14 +514,13 @@ static int live_hwsp_engine(void *arg)
>>>        if (!timelines)
>>>                return -ENOMEM;
>>>    
>>> -     mutex_lock(&i915->drm.struct_mutex);
>>> -     wakeref = intel_runtime_pm_get(&i915->runtime_pm);
>>> -
>>>        count = 0;
>>>        for_each_engine(engine, i915, id) {
>>>                if (!intel_engine_can_store_dword(engine))
>>>                        continue;
>>>    
>>> +             intel_engine_pm_get(engine);
>>> +
>>>                for (n = 0; n < NUM_TIMELINES; n++) {
>>>                        struct intel_timeline *tl;
>>>                        struct i915_request *rq;
>>> @@ -530,22 +528,26 @@ static int live_hwsp_engine(void *arg)
>>>                        tl = checked_intel_timeline_create(i915);
>>>                        if (IS_ERR(tl)) {
>>>                                err = PTR_ERR(tl);
>>> -                             goto out;
>>> +                             break;
>>>                        }
>>>    
>>>                        rq = tl_write(tl, engine, count);
>>>                        if (IS_ERR(rq)) {
>>>                                intel_timeline_put(tl);
>>>                                err = PTR_ERR(rq);
>>> -                             goto out;
>>> +                             break;
>>>                        }
>>>    
>>>                        timelines[count++] = tl;
>>> +                     i915_request_put(rq);
>>
>> This was a leak until now?
> 
> No, we added a get into tl_write() so that we own a reference to the
> request, just so that ownership is clear across the waits (and it can't
> accidentally be retired).

/me scrolls up, scrolls down.. rubs eyes.. yep, I was blind.

[snip]

>> Essentially looks fine. Provisional, meaning keep it if you do some
>> small tweaks:
> 
> Nothing seemed to be required.

Ack.

Regards,

Tvrtko


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

  reply	other threads:[~2019-09-25  8:49 UTC|newest]

Thread overview: 53+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-09-02  4:02 [PATCH 01/21] drm/i915: Restrict the aliasing-ppgtt to the size of the ggtt Chris Wilson
2019-09-02  4:02 ` [PATCH 02/21] drm/i915: Report aliasing ppgtt size as ggtt size Chris Wilson
2019-09-02  8:55   ` Matthew Auld
2019-09-02  4:02 ` [PATCH 03/21] drm/i915/execlists: Ignore lost completion events Chris Wilson
2019-09-02  4:02 ` [PATCH 04/21] drm/i915: Refresh the errno to vmf_fault translations Chris Wilson
2019-09-03 15:34   ` Abdiel Janulgue
2019-09-02  4:02 ` [PATCH 05/21] drm/i915: Replace obj->pin_global with obj->frontbuffer Chris Wilson
2019-09-02  4:02 ` [PATCH 06/21] dma-fence: Serialise signal enabling (dma_fence_enable_sw_signaling) Chris Wilson
2019-09-02  4:02 ` [PATCH 07/21] drm/mm: Pack allocated/scanned boolean into a bitfield Chris Wilson
2019-09-02  4:02 ` [PATCH 08/21] drm/i915: Make shrink/unshrink be atomic Chris Wilson
2019-09-10 19:54   ` Matthew Auld
2019-09-02  4:02 ` [PATCH 09/21] drm/i915: Only track bound elements of the GTT Chris Wilson
2019-09-02  4:02 ` [PATCH 10/21] drm/i915: Make i915_vma.flags atomic_t for mutex reduction Chris Wilson
2019-09-10 20:06   ` Matthew Auld
2019-09-02  4:02 ` [PATCH 11/21] drm/i915/gtt: Make sure the gen6 ppgtt is bound before first use Chris Wilson
2019-09-10 20:17   ` Matthew Auld
2019-09-02  4:02 ` [PATCH 12/21] drm/i915: Mark up address spaces that may need to allocate Chris Wilson
2019-09-20 16:22   ` Tvrtko Ursulin
2019-09-20 16:35     ` Chris Wilson
2019-09-23  8:10       ` Tvrtko Ursulin
2019-09-25  8:23         ` Chris Wilson
2019-09-25 15:59           ` Tvrtko Ursulin
2019-09-27 17:03             ` Chris Wilson
2019-09-02  4:02 ` [PATCH 13/21] drm/i915: Pull i915_vma_pin under the vm->mutex Chris Wilson
2019-09-16 10:13   ` Tvrtko Ursulin
2019-09-16 11:10     ` Chris Wilson
2019-09-17 12:37   ` Tvrtko Ursulin
2019-09-17 18:56     ` Chris Wilson
2019-09-19 13:37       ` Tvrtko Ursulin
2019-09-19 14:05         ` Chris Wilson
2019-09-02  4:02 ` [PATCH 14/21] drm/i915: Push the i915_active.retire into a worker Chris Wilson
2019-09-02  4:02 ` [PATCH 15/21] drm/i915: Coordinate i915_active with its own mutex Chris Wilson
2019-09-20 16:14   ` Tvrtko Ursulin
2019-09-20 16:32     ` Chris Wilson
2019-09-02  4:02 ` [PATCH 16/21] drm/i915: Move idle barrier cleanup into engine-pm Chris Wilson
2019-09-20 16:18   ` Tvrtko Ursulin
2019-09-02  4:02 ` [PATCH 17/21] drm/i915: Drop struct_mutex from around i915_retire_requests() Chris Wilson
2019-09-24 15:25   ` Tvrtko Ursulin
2019-09-25  8:43     ` Chris Wilson
2019-09-25  8:49       ` Tvrtko Ursulin [this message]
2019-09-02  4:03 ` [PATCH 18/21] drm/i915: Remove the GEM idle worker Chris Wilson
2019-09-24 15:26   ` Tvrtko Ursulin
2019-09-02  4:03 ` [PATCH 19/21] drm/i915: Merge wait_for_timelines with retire_request Chris Wilson
2019-09-24 15:57   ` Tvrtko Ursulin
2019-09-25  8:54     ` Chris Wilson
2019-09-02  4:03 ` [PATCH 20/21] drm/i915: Move request runtime management onto gt Chris Wilson
2019-09-02  4:03 ` [PATCH 21/21] drm/i915: Move global activity tracking from GEM to GT Chris Wilson
2019-09-25  9:55   ` Tvrtko Ursulin
2019-09-02  4:54 ` ✗ Fi.CI.CHECKPATCH: warning for series starting with [01/21] drm/i915: Restrict the aliasing-ppgtt to the size of the ggtt Patchwork
2019-09-02  5:20 ` ✓ Fi.CI.BAT: success " Patchwork
2019-09-02  8:00 ` ✗ Fi.CI.IGT: failure " Patchwork
2019-09-02  8:52 ` [PATCH 01/21] " Matthew Auld
  -- strict thread matches above, loose matches on Subject: below --
2019-08-30  6:11 [PATCH 01/21] drm/i915/gtt: Downgrade Baytrail back to aliasing-ppgtt Chris Wilson
2019-08-30  6:12 ` [PATCH 17/21] drm/i915: Drop struct_mutex from around i915_retire_requests() Chris Wilson

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=ef7aaff6-a350-c2d9-df0b-b4e1856c5153@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox