[Intel-gfx] [PATCH] drm/i915/selftests: Wait only on the expected barrier

[Tvrtko Ursulin]

On 29/11/2019 10:34, Chris Wilson wrote:
> Wait on only the last request on the kernel_context after emitting a
> barrier so that we do not wait for everything in general and by doing so
> cause an accidental emission of the barrier!
> 
> Bugzilla; https://bugs.freedesktop.org/show_bug.cgi?id=112405
> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> ---
>   .../drm/i915/gt/selftest_engine_heartbeat.c   | 24 ++++++++++++++++++-
>   1 file changed, 23 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/i915/gt/selftest_engine_heartbeat.c b/drivers/gpu/drm/i915/gt/selftest_engine_heartbeat.c
> index 0b1148cf3f61..5227e79204a5 100644
> --- a/drivers/gpu/drm/i915/gt/selftest_engine_heartbeat.c
> +++ b/drivers/gpu/drm/i915/gt/selftest_engine_heartbeat.c
> @@ -11,6 +11,28 @@
>   #include "intel_gt_requests.h"
>   #include "i915_selftest.h"
>   
> +static int timeline_sync(struct intel_timeline *tl)
> +{
> +	struct dma_fence *fence;
> +	long timeout;
> +
> +	fence = i915_active_fence_get(&tl->last_request);
> +	if (!fence)
> +		return 0;
> +
> +	timeout = dma_fence_wait_timeout(fence, true, HZ / 2);
> +	dma_fence_put(fence);
> +	if (timeout < 0)
> +		return timeout;
> +
> +	return 0;
> +}
> +
> +static int engine_sync_barrier(struct intel_engine_cs *engine)
> +{
> +	return timeline_sync(engine->kernel_context->timeline);
> +}
> +
>   struct pulse {
>   	struct i915_active active;
>   	struct kref kref;
> @@ -90,7 +112,7 @@ static int __live_idle_pulse(struct intel_engine_cs *engine,
>   
>   	GEM_BUG_ON(!llist_empty(&engine->barrier_tasks));
>   
> -	if (intel_gt_retire_requests_timeout(engine->gt, HZ / 5)) {
> +	if (engine_sync_barrier(engine)) {
>   		struct drm_printer m = drm_err_printer("pulse");
>   
>   		pr_err("%s: no heartbeat pulse?\n", engine->name);
> 

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

Regards,

Tvrtko