[Intel-gfx] [PATCH 05/22] drm/i915: Lock the gem_context->active_list while dropping the link
Tvrtko Ursulin
tvrtko.ursulin at linux.intel.com
Mon Mar 18 10:50:42 UTC 2019
On 18/03/2019 10:45, Chris Wilson wrote:
> Quoting Tvrtko Ursulin (2019-03-18 10:39:44)
>>
>> On 18/03/2019 09:51, Chris Wilson wrote:
>>> On unpinning the intel_context, we remove it from the active list
>>> inside the GEM context. This list is supposed to be guarded by the GEM
>>> context mutex, so remember to take it!
>>
>> Fixes: ?
>
> It is not broken yet :)
>
>>> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
>>> ---
>>> drivers/gpu/drm/i915/intel_context.c | 15 +++++++++++----
>>> drivers/gpu/drm/i915/intel_lrc.c | 3 ---
>>> drivers/gpu/drm/i915/intel_ringbuffer.c | 3 ---
>>> drivers/gpu/drm/i915/selftests/mock_engine.c | 2 --
>>> 4 files changed, 11 insertions(+), 12 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/i915/intel_context.c b/drivers/gpu/drm/i915/intel_context.c
>>> index 5a16c9bb2778..0ab894a058f6 100644
>>> --- a/drivers/gpu/drm/i915/intel_context.c
>>> +++ b/drivers/gpu/drm/i915/intel_context.c
>>> @@ -165,13 +165,13 @@ intel_context_pin(struct i915_gem_context *ctx,
>>> if (err)
>>> goto err;
>>>
>>> + i915_gem_context_get(ctx);
>>> + GEM_BUG_ON(ce->gem_context != ctx);
>>> +
>>> mutex_lock(&ctx->mutex);
>>> list_add(&ce->active_link, &ctx->active_engines);
>>> mutex_unlock(&ctx->mutex);
>>>
>>> - i915_gem_context_get(ctx);
>>> - GEM_BUG_ON(ce->gem_context != ctx);
>>> -
>>> smp_mb__before_atomic(); /* flush pin before it is visible */
>>> }
>>>
>>> @@ -194,9 +194,16 @@ void intel_context_unpin(struct intel_context *ce)
>>> /* We may be called from inside intel_context_pin() to evict another */
>>> mutex_lock_nested(&ce->pin_mutex, SINGLE_DEPTH_NESTING);
>>
>> Hm is the nested annotation and comment correct?
>
> Yes. pin -> vma bind -> evict -> retire requests -> unpin someone else.
Yep, the ops->pin.. I missed it.
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
Regards,
Tvrtko
More information about the Intel-gfx
mailing list