[PATCH 2/2] drm/amdgpu: use irq-safe lock for adev->ring_lru_list_lock

Christian König ckoenig.leichtzumerken at gmail.com
Tue Nov 7 10:36:08 UTC 2017


Am 07.11.2017 um 08:26 schrieb Pixel Ding:
> From: pding <Pixel.Ding at amd.com>
>
> This lock is used during register accessing in SRIOV guest
> since KIQ uses general ring submission (amdgpu_ring_commit).
> The register accessing could happen both in irq enabled and
> irq disabled cases. Always use irq-safe lock.

I would rather completely skip calling amdgpu_ring_lru_touch() for the 
KIQ ring.

That is completely unused in this case anyway and just additional overhead.

Regards,
Christian.

>
> Signed-off-by: pding <Pixel.Ding at amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c | 11 +++++++----
>   1 file changed, 7 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
> index e5ece1f..d7997b5 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
> @@ -401,11 +401,12 @@ int amdgpu_ring_lru_get(struct amdgpu_device *adev, int type,
>   			bool lru_pipe_order, struct amdgpu_ring **ring)
>   {
>   	struct amdgpu_ring *entry;
> +	unsigned long flags;
>   
>   	/* List is sorted in LRU order, find first entry corresponding
>   	 * to the desired HW IP */
>   	*ring = NULL;
> -	spin_lock(&adev->ring_lru_list_lock);
> +	spin_lock_irqsave(&adev->ring_lru_list_lock, flags);
>   	list_for_each_entry(entry, &adev->ring_lru_list, lru_list) {
>   		if (entry->funcs->type != type)
>   			continue;
> @@ -430,7 +431,7 @@ int amdgpu_ring_lru_get(struct amdgpu_device *adev, int type,
>   	if (*ring)
>   		amdgpu_ring_lru_touch_locked(adev, *ring);
>   
> -	spin_unlock(&adev->ring_lru_list_lock);
> +	spin_unlock_irqrestore(&adev->ring_lru_list_lock, flags);
>   
>   	if (!*ring) {
>   		DRM_ERROR("Ring LRU contains no entries for ring type:%d\n", type);
> @@ -450,9 +451,11 @@ int amdgpu_ring_lru_get(struct amdgpu_device *adev, int type,
>    */
>   void amdgpu_ring_lru_touch(struct amdgpu_device *adev, struct amdgpu_ring *ring)
>   {
> -	spin_lock(&adev->ring_lru_list_lock);
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&adev->ring_lru_list_lock, flags);
>   	amdgpu_ring_lru_touch_locked(adev, ring);
> -	spin_unlock(&adev->ring_lru_list_lock);
> +	spin_unlock_irqrestore(&adev->ring_lru_list_lock, flags);
>   }
>   
>   /*




More information about the amd-gfx mailing list