[PATCH] drm/ttm: move cpu_writers handling into vmwgfx

Thomas Hellstrom thellstrom at vmware.com
Wed Jun 12 15:27:46 UTC 2019


Hi, Christian,

This looks OK, although there are a couple of minor alterations needed
in the vmwgfx driver:

- We should operate on vmw_buffer_objects rather than on
user_buffer_objects.
- vmw_user_bo_verify_synccpu should move to the validate code.

I can take care of that if it's ok with you.

Thanks,
Thomas


On Fri, 2019-06-07 at 16:47 +0200, Christian König wrote:
> This feature is only used by vmwgfx and superflous for everybody
> else.
> 
> Signed-off-by: Christian König <christian.koenig at amd.com>
> ---
>  drivers/gpu/drm/ttm/ttm_bo.c             | 27 ------------------
>  drivers/gpu/drm/ttm/ttm_bo_util.c        |  1 -
>  drivers/gpu/drm/ttm/ttm_execbuf_util.c   |  7 +----
>  drivers/gpu/drm/vmwgfx/vmwgfx_bo.c       | 35 ++++++++++++++++++++
> ----
>  drivers/gpu/drm/vmwgfx/vmwgfx_drv.h      |  2 ++
>  drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c  |  8 ++++++
>  drivers/gpu/drm/vmwgfx/vmwgfx_resource.c |  4 +++
>  include/drm/ttm/ttm_bo_api.h             | 31 ---------------------
>  8 files changed, 45 insertions(+), 70 deletions(-)
> 
> diff --git a/drivers/gpu/drm/ttm/ttm_bo.c
> b/drivers/gpu/drm/ttm/ttm_bo.c
> index c7de667d482a..4ec055ffd6a7 100644
> --- a/drivers/gpu/drm/ttm/ttm_bo.c
> +++ b/drivers/gpu/drm/ttm/ttm_bo.c
> @@ -153,7 +153,6 @@ static void ttm_bo_release_list(struct kref
> *list_kref)
>  
>  	BUG_ON(kref_read(&bo->list_kref));
>  	BUG_ON(kref_read(&bo->kref));
> -	BUG_ON(atomic_read(&bo->cpu_writers));
>  	BUG_ON(bo->mem.mm_node != NULL);
>  	BUG_ON(!list_empty(&bo->lru));
>  	BUG_ON(!list_empty(&bo->ddestroy));
> @@ -1308,7 +1307,6 @@ int ttm_bo_init_reserved(struct ttm_bo_device
> *bdev,
>  
>  	kref_init(&bo->kref);
>  	kref_init(&bo->list_kref);
> -	atomic_set(&bo->cpu_writers, 0);
>  	INIT_LIST_HEAD(&bo->lru);
>  	INIT_LIST_HEAD(&bo->ddestroy);
>  	INIT_LIST_HEAD(&bo->swap);
> @@ -1814,31 +1812,6 @@ int ttm_bo_wait(struct ttm_buffer_object *bo,
>  }
>  EXPORT_SYMBOL(ttm_bo_wait);
>  
> -int ttm_bo_synccpu_write_grab(struct ttm_buffer_object *bo, bool
> no_wait)
> -{
> -	int ret = 0;
> -
> -	/*
> -	 * Using ttm_bo_reserve makes sure the lru lists are updated.
> -	 */
> -
> -	ret = ttm_bo_reserve(bo, true, no_wait, NULL);
> -	if (unlikely(ret != 0))
> -		return ret;
> -	ret = ttm_bo_wait(bo, true, no_wait);
> -	if (likely(ret == 0))
> -		atomic_inc(&bo->cpu_writers);
> -	ttm_bo_unreserve(bo);
> -	return ret;
> -}
> -EXPORT_SYMBOL(ttm_bo_synccpu_write_grab);
> -
> -void ttm_bo_synccpu_write_release(struct ttm_buffer_object *bo)
> -{
> -	atomic_dec(&bo->cpu_writers);
> -}
> -EXPORT_SYMBOL(ttm_bo_synccpu_write_release);
> -
>  /**
>   * A buffer object shrink method that tries to swap out the first
>   * buffer object on the bo_global::swap_lru list.
> diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c
> b/drivers/gpu/drm/ttm/ttm_bo_util.c
> index 895d77d799e4..6f43f1f0de7c 100644
> --- a/drivers/gpu/drm/ttm/ttm_bo_util.c
> +++ b/drivers/gpu/drm/ttm/ttm_bo_util.c
> @@ -511,7 +511,6 @@ static int ttm_buffer_object_transfer(struct
> ttm_buffer_object *bo,
>  	mutex_init(&fbo->base.wu_mutex);
>  	fbo->base.moving = NULL;
>  	drm_vma_node_reset(&fbo->base.vma_node);
> -	atomic_set(&fbo->base.cpu_writers, 0);
>  
>  	kref_init(&fbo->base.list_kref);
>  	kref_init(&fbo->base.kref);
> diff --git a/drivers/gpu/drm/ttm/ttm_execbuf_util.c
> b/drivers/gpu/drm/ttm/ttm_execbuf_util.c
> index 957ec375a4ba..80fa52b36d5c 100644
> --- a/drivers/gpu/drm/ttm/ttm_execbuf_util.c
> +++ b/drivers/gpu/drm/ttm/ttm_execbuf_util.c
> @@ -113,12 +113,7 @@ int ttm_eu_reserve_buffers(struct ww_acquire_ctx
> *ticket,
>  		struct ttm_buffer_object *bo = entry->bo;
>  
>  		ret = __ttm_bo_reserve(bo, intr, (ticket == NULL),
> ticket);
> -		if (!ret && unlikely(atomic_read(&bo->cpu_writers) >
> 0)) {
> -			reservation_object_unlock(bo->resv);
> -
> -			ret = -EBUSY;
> -
> -		} else if (ret == -EALREADY && dups) {
> +		if (ret == -EALREADY && dups) {
>  			struct ttm_validate_buffer *safe = entry;
>  			entry = list_prev_entry(entry, head);
>  			list_del(&safe->head);
> diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c
> b/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c
> index 5d5c2bce01f3..457861c5047f 100644
> --- a/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c
> +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c
> @@ -565,7 +565,7 @@ static void vmw_user_bo_ref_obj_release(struct
> ttm_base_object *base,
>  
>  	switch (ref_type) {
>  	case TTM_REF_SYNCCPU_WRITE:
> -		ttm_bo_synccpu_write_release(&user_bo->vbo.base);
> +		atomic_dec(&user_bo->vbo.cpu_writers);
>  		break;
>  	default:
>  		WARN_ONCE(true, "Undefined buffer object reference
> release.\n");
> @@ -681,12 +681,12 @@ static int vmw_user_bo_synccpu_grab(struct
> vmw_user_buffer_object *user_bo,
>  				    struct ttm_object_file *tfile,
>  				    uint32_t flags)
>  {
> +	bool nonblock = !!(flags & drm_vmw_synccpu_dontblock);
>  	struct ttm_buffer_object *bo = &user_bo->vbo.base;
>  	bool existed;
>  	int ret;
>  
>  	if (flags & drm_vmw_synccpu_allow_cs) {
> -		bool nonblock = !!(flags & drm_vmw_synccpu_dontblock);
>  		long lret;
>  
>  		lret = reservation_object_wait_timeout_rcu
> @@ -699,15 +699,20 @@ static int vmw_user_bo_synccpu_grab(struct
> vmw_user_buffer_object *user_bo,
>  		return 0;
>  	}
>  
> -	ret = ttm_bo_synccpu_write_grab
> -		(bo, !!(flags & drm_vmw_synccpu_dontblock));
> +	ret = ttm_bo_reserve(bo, true, nonblock, NULL);
> +	if (unlikely(ret != 0))
> +		return ret;
> +	ret = ttm_bo_wait(bo, true, nonblock);
> +	if (likely(ret == 0))
> +		atomic_inc(&user_bo->vbo.cpu_writers);
> +	ttm_bo_unreserve(bo);
>  	if (unlikely(ret != 0))
>  		return ret;
>  
>  	ret = ttm_ref_object_add(tfile, &user_bo->prime.base,
>  				 TTM_REF_SYNCCPU_WRITE, &existed,
> false);
>  	if (ret != 0 || existed)
> -		ttm_bo_synccpu_write_release(&user_bo->vbo.base);
> +		atomic_dec(&user_bo->vbo.cpu_writers);
>  
>  	return ret;
>  }
> @@ -731,6 +736,26 @@ static int vmw_user_bo_synccpu_release(uint32_t
> handle,
>  	return 0;
>  }
>  
> +/**
> + * vmw_user_bo_verify_synccpu - Verify if grab for CPU access exists
> + *
> + * @list: list of ttm_validate_buffer objects
> + *
> + * Return:
> + * -EBUSY if a CPU grab is found, 0 otherwise.
> + */
> +int vmw_user_bo_verify_synccpu(struct list_head *list)
> +{
> +	struct ttm_validate_buffer *entry;
> +
> +	list_for_each_entry(entry, list, head) {
> +		struct vmw_buffer_object *bo = vmw_buffer_object(entry-
> >bo);
> +
> +                if (unlikely(atomic_read(&bo->cpu_writers) > 0))
> +			return -EBUSY;
> +	}
> +	return 0;
> +}
>  
>  /**
>   * vmw_user_bo_synccpu_ioctl - ioctl function implementing the
> synccpu
> diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
> b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
> index 96983c47fb40..b3988c957f0e 100644
> --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
> +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
> @@ -90,6 +90,7 @@ struct vmw_buffer_object {
>  	struct ttm_buffer_object base;
>  	struct list_head res_list;
>  	s32 pin_count;
> +	atomic_t cpu_writers;
>  	/* Not ref-counted.  Protected by binding_mutex */
>  	struct vmw_resource *dx_query_ctx;
>  	/* Protected by reservation */
> @@ -749,6 +750,7 @@ extern int vmw_bo_init(struct vmw_private
> *dev_priv,
>  		       void (*bo_free)(struct ttm_buffer_object *bo));
>  extern int vmw_user_bo_verify_access(struct ttm_buffer_object *bo,
>  				     struct ttm_object_file *tfile);
> +extern int vmw_user_bo_verify_synccpu(struct list_head *list);
>  extern int vmw_user_bo_alloc(struct vmw_private *dev_priv,
>  			     struct ttm_object_file *tfile,
>  			     uint32_t size,
> diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c
> b/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c
> index 2ff7ba04d8c8..271217f6e3a3 100644
> --- a/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c
> +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c
> @@ -3719,6 +3719,10 @@ int vmw_execbuf_process(struct drm_file
> *file_priv,
>  	if (unlikely(ret != 0))
>  		goto out_err_nores;
>  
> +	ret = vmw_user_bo_verify_synccpu(&val_ctx.bo_list);
> +	if (unlikely(ret != 0))
> +		goto out_err;
> +
>  	ret = vmw_validation_bo_validate(&val_ctx, true);
>  	if (unlikely(ret != 0))
>  		goto out_err;
> @@ -3918,6 +3922,10 @@ void __vmw_execbuf_release_pinned_bo(struct
> vmw_private *dev_priv,
>  	if (ret)
>  		goto out_no_reserve;
>  
> +	ret = vmw_user_bo_verify_synccpu(&val_ctx.bo_list);
> +	if (unlikely(ret != 0))
> +		goto out_no_emit;
> +
>  	if (dev_priv->query_cid_valid) {
>  		BUG_ON(fence != NULL);
>  		ret = vmw_fifo_emit_dummy_query(dev_priv, dev_priv-
> >query_cid);
> diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
> b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
> index 1d38a8b2f2ec..c55da6f0ff03 100644
> --- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
> +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
> @@ -469,6 +469,10 @@ vmw_resource_check_buffer(struct ww_acquire_ctx
> *ticket,
>  	if (unlikely(ret != 0))
>  		goto out_no_reserve;
>  
> +	ret = vmw_user_bo_verify_synccpu(&val_list);
> +	if (unlikely(ret != 0))
> +		goto out_no_validate;
> +
>  	if (res->func->needs_backup && list_empty(&res->mob_head))
>  		return 0;
>  
> diff --git a/include/drm/ttm/ttm_bo_api.h
> b/include/drm/ttm/ttm_bo_api.h
> index 49d9cdfc58f2..e358e965d4c0 100644
> --- a/include/drm/ttm/ttm_bo_api.h
> +++ b/include/drm/ttm/ttm_bo_api.h
> @@ -145,7 +145,6 @@ struct ttm_tt;
>   * holds a pointer to a persistent shmem object.
>   * @ttm: TTM structure holding system pages.
>   * @evicted: Whether the object was evicted without user-space
> knowing.
> - * @cpu_writes: For synchronization. Number of cpu writers.
>   * @lru: List head for the lru list.
>   * @ddestroy: List head for the delayed destroy list.
>   * @swap: List head for swap LRU list.
> @@ -195,11 +194,6 @@ struct ttm_buffer_object {
>  	struct ttm_tt *ttm;
>  	bool evicted;
>  
> -	/**
> -	 * Members protected by the bo::reserved lock only when written
> to.
> -	 */
> -
> -	atomic_t cpu_writers;
>  
>  	/**
>  	 * Members protected by the bdev::lru_lock.
> @@ -443,31 +437,6 @@ void ttm_bo_unlock_delayed_workqueue(struct
> ttm_bo_device *bdev, int resched);
>  bool ttm_bo_eviction_valuable(struct ttm_buffer_object *bo,
>  			      const struct ttm_place *place);
>  
> -/**
> - * ttm_bo_synccpu_write_grab
> - *
> - * @bo: The buffer object:
> - * @no_wait: Return immediately if buffer is busy.
> - *
> - * Synchronizes a buffer object for CPU RW access. This means
> - * command submission that affects the buffer will return -EBUSY
> - * until ttm_bo_synccpu_write_release is called.
> - *
> - * Returns
> - * -EBUSY if the buffer is busy and no_wait is true.
> - * -ERESTARTSYS if interrupted by a signal.
> - */
> -int ttm_bo_synccpu_write_grab(struct ttm_buffer_object *bo, bool
> no_wait);
> -
> -/**
> - * ttm_bo_synccpu_write_release:
> - *
> - * @bo : The buffer object.
> - *
> - * Releases a synccpu lock.
> - */
> -void ttm_bo_synccpu_write_release(struct ttm_buffer_object *bo);
> -
>  /**
>   * ttm_bo_acc_size
>   *


More information about the dri-devel mailing list