Re: [PATCH] drm/ttm: fix busy memory to fail other user v5

["Christian König" <ckoenig.leichtzumerken@gmail.com>]

Am 30.04.19 um 09:01 schrieb Chunming Zhou:
> heavy gpu job could occupy memory long time, which lead other user fail to get memory.
>
> basically pick up Christian idea:
>
> 1. Reserve the BO in DC using a ww_mutex ticket (trivial).
> 2. If we then run into this EBUSY condition in TTM check if the BO we need memory for (or rather the ww_mutex of its reservation object) has a ticket assigned.
> 3. If we have a ticket we grab a reference to the first BO on the LRU, drop the LRU lock and try to grab the reservation lock with the ticket.
> 4. If getting the reservation lock with the ticket succeeded we check if the BO is still the first one on the LRU in question (the BO could have moved).
> 5. If the BO is still the first one on the LRU in question we try to evict it as we would evict any other BO.
> 6. If any of the "If's" above fail we just back off and return -EBUSY.
>
> v2: fix some minor check
> v3: address Christian v2 comments.
> v4: fix some missing
> v5: handle first_bo unlock and bo_get/put
>
> Change-Id: I21423fb922f885465f13833c41df1e134364a8e7
> Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_object.c    |  7 +-
>   .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 22 +++--
>   drivers/gpu/drm/ttm/ttm_bo.c                  | 81 +++++++++++++++++--
>   3 files changed, 99 insertions(+), 11 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
> index affde72b44db..523773e85284 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
> @@ -811,7 +811,12 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
>   			     u64 min_offset, u64 max_offset)
>   {
>   	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
> -	struct ttm_operation_ctx ctx = { false, false };
> +	struct ttm_operation_ctx ctx = {
> +		.interruptible = false,
> +		.no_wait_gpu = false,
> +		.resv = bo->tbo.resv,
> +		.flags = 0
> +	};
>   	int r, i;
>   
>   	if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm))
> diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
> index a5cacf846e1b..cc3677c4a4c2 100644
> --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
> +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
> @@ -4101,6 +4101,9 @@ static int dm_plane_helper_prepare_fb(struct drm_plane *plane,
>   	struct amdgpu_device *adev;
>   	struct amdgpu_bo *rbo;
>   	struct dm_plane_state *dm_plane_state_new, *dm_plane_state_old;
> +	struct list_head list, duplicates;
> +	struct ttm_validate_buffer tv;
> +	struct ww_acquire_ctx ticket;
>   	uint64_t tiling_flags;
>   	uint32_t domain;
>   	int r;
> @@ -4117,9 +4120,18 @@ static int dm_plane_helper_prepare_fb(struct drm_plane *plane,
>   	obj = new_state->fb->obj[0];
>   	rbo = gem_to_amdgpu_bo(obj);
>   	adev = amdgpu_ttm_adev(rbo->tbo.bdev);
> -	r = amdgpu_bo_reserve(rbo, false);
> -	if (unlikely(r != 0))
> +	INIT_LIST_HEAD(&list);
> +	INIT_LIST_HEAD(&duplicates);
> +
> +	tv.bo = &rbo->tbo;
> +	tv.num_shared = 1;
> +	list_add(&tv.head, &list);
> +
> +	r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates);
> +	if (r) {
> +		dev_err(adev->dev, "fail to reserve bo (%d)\n", r);
>   		return r;
> +	}
>   
>   	if (plane->type != DRM_PLANE_TYPE_CURSOR)
>   		domain = amdgpu_display_supported_domains(adev);
> @@ -4130,21 +4142,21 @@ static int dm_plane_helper_prepare_fb(struct drm_plane *plane,
>   	if (unlikely(r != 0)) {
>   		if (r != -ERESTARTSYS)
>   			DRM_ERROR("Failed to pin framebuffer with error %d\n", r);
> -		amdgpu_bo_unreserve(rbo);
> +		ttm_eu_backoff_reservation(&ticket, &list);
>   		return r;
>   	}
>   
>   	r = amdgpu_ttm_alloc_gart(&rbo->tbo);
>   	if (unlikely(r != 0)) {
>   		amdgpu_bo_unpin(rbo);
> -		amdgpu_bo_unreserve(rbo);
> +		ttm_eu_backoff_reservation(&ticket, &list);
>   		DRM_ERROR("%p bind failed\n", rbo);
>   		return r;
>   	}
>   
>   	amdgpu_bo_get_tiling_flags(rbo, &tiling_flags);
>   
> -	amdgpu_bo_unreserve(rbo);
> +	ttm_eu_backoff_reservation(&ticket, &list);

Well I can only repeat myself, please put that into a separate patch!

>   
>   	afb->address = amdgpu_bo_gpu_offset(rbo);
>   
> diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
> index 8502b3ed2d88..2c4963e105d9 100644
> --- a/drivers/gpu/drm/ttm/ttm_bo.c
> +++ b/drivers/gpu/drm/ttm/ttm_bo.c
> @@ -766,11 +766,13 @@ EXPORT_SYMBOL(ttm_bo_eviction_valuable);
>    * b. Otherwise, trylock it.
>    */
>   static bool ttm_bo_evict_swapout_allowable(struct ttm_buffer_object *bo,
> -			struct ttm_operation_ctx *ctx, bool *locked)
> +			struct ttm_operation_ctx *ctx, bool *locked, bool *busy)
>   {
>   	bool ret = false;
>   
>   	*locked = false;
> +	if (busy)
> +		*busy = false;
>   	if (bo->resv == ctx->resv) {
>   		reservation_object_assert_held(bo->resv);
>   		if (ctx->flags & TTM_OPT_FLAG_ALLOW_RES_EVICT
> @@ -779,6 +781,8 @@ static bool ttm_bo_evict_swapout_allowable(struct ttm_buffer_object *bo,
>   	} else {
>   		*locked = reservation_object_trylock(bo->resv);
>   		ret = *locked;
> +		if (!ret && busy)
> +			*busy = true;
>   	}
>   
>   	return ret;
> @@ -791,7 +795,7 @@ static int ttm_mem_evict_first(struct ttm_bo_device *bdev,
>   {
>   	struct ttm_bo_global *glob = bdev->glob;
>   	struct ttm_mem_type_manager *man = &bdev->man[mem_type];
> -	struct ttm_buffer_object *bo = NULL;
> +	struct ttm_buffer_object *bo = NULL, *first_bo = NULL;
>   	bool locked = false;
>   	unsigned i;
>   	int ret;
> @@ -799,8 +803,15 @@ static int ttm_mem_evict_first(struct ttm_bo_device *bdev,
>   	spin_lock(&glob->lru_lock);
>   	for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) {
>   		list_for_each_entry(bo, &man->lru[i], lru) {
> -			if (!ttm_bo_evict_swapout_allowable(bo, ctx, &locked))
> +			bool busy = false;
> +			if (!ttm_bo_evict_swapout_allowable(bo, ctx, &locked,
> +							    &busy)) {
> +				if (!first_bo && busy) {
> +					ttm_bo_get(bo);
> +					first_bo = bo;
> +				}
>   				continue;
> +			}
>   
>   			if (place && !bdev->driver->eviction_valuable(bo,
>   								      place)) {
> @@ -808,6 +819,7 @@ static int ttm_mem_evict_first(struct ttm_bo_device *bdev,
>   					reservation_object_unlock(bo->resv);
>   				continue;
>   			}
> +
>   			break;
>   		}
>   
> @@ -820,7 +832,65 @@ static int ttm_mem_evict_first(struct ttm_bo_device *bdev,
>   
>   	if (!bo) {
>   		spin_unlock(&glob->lru_lock);
> -		return -EBUSY;
> +		/* check if other user occupy memory too long time */
> +		if (!first_bo || !ctx || !ctx->resv || !ctx->resv->lock.ctx) {
> +			if (first_bo)
> +				ttm_bo_put(first_bo);
> +			return -EBUSY;
> +		}
> +		if (ctx->interruptible)
> +			ret = ww_mutex_lock_interruptible(&first_bo->resv->lock,
> +							  ctx->resv->lock.ctx);
> +		else
> +			ret = ww_mutex_lock(&first_bo->resv->lock, ctx->resv->lock.ctx);
> +		if (ret) {
> +			ttm_bo_put(first_bo);
> +			return ret;
> +		}
> +		spin_lock(&glob->lru_lock);
> +		for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) {
> +			/* previous busy resv lock is held by above, idle now,
> +			 * so let them evictable.
> +			 */
> +			struct ttm_operation_ctx busy_ctx = {
> +				.interruptible = ctx->interruptible,
> +				.no_wait_gpu   = ctx->no_wait_gpu,
> +				.resv	       = first_bo->resv,
> +				.flags	       = TTM_OPT_FLAG_ALLOW_RES_EVICT
> +			};
> +			list_for_each_entry(bo, &man->lru[i], lru) {
> +				if (!ttm_bo_evict_swapout_allowable(bo,
> +								    &busy_ctx,
> +								    &locked,
> +								    NULL))
> +					continue;
> +
> +				if (place && !bdev->driver->eviction_valuable(bo,
> +								      place)) {
> +					if (locked)
> +						reservation_object_unlock(bo->resv);
> +					continue;
> +				}
> +				break;
> +			}
> +			/* If the inner loop terminated early, we have our candidate */
> +			if (&bo->lru != &man->lru[i])
> +				break;
> +			bo = NULL;
> +		}

So we are now searching for any BO again which matches the criteria? 
Good idea.

Can we factor those two loops out into a separate function? I mean they 
are mostly identical now to the one above, aren't they?

> +		if (bo && (bo->resv == first_bo->resv))
> +			locked = true;
> +		else if (bo)
> +			ww_mutex_unlock(&first_bo->resv->lock);
> +		ttm_bo_put(first_bo);
> +		first_bo = NULL;
> +		if (!bo) {
> +			spin_unlock(&glob->lru_lock);
> +			return -EBUSY;
> +		}
> +	} else {
> +		if (first_bo)
> +			ttm_bo_put(first_bo);

This must come a bit later, since we can't call ttm_bo_put() while 
holding the spinlock.

Regards,
Christian.

>   	}
>   
>   	kref_get(&bo->list_kref);
> @@ -1784,7 +1854,8 @@ int ttm_bo_swapout(struct ttm_bo_global *glob, struct ttm_operation_ctx *ctx)
>   	spin_lock(&glob->lru_lock);
>   	for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) {
>   		list_for_each_entry(bo, &glob->swap_lru[i], swap) {
> -			if (ttm_bo_evict_swapout_allowable(bo, ctx, &locked)) {
> +			if (ttm_bo_evict_swapout_allowable(bo, ctx, &locked,
> +							   NULL)) {
>   				ret = 0;
>   				break;
>   			}

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel