Commit cc8a4d5a authored by Rob Clark's avatar Rob Clark
Browse files

drm/msm: Avoid mutex in shrinker_count()



When the system is under heavy memory pressure, we can end up with lots
of concurrent calls into the shrinker.  Keeping a running tab on what we
can shrink avoids grabbing a lock in shrinker->count(), and avoids
shrinker->scan() getting called when not profitable.

Also, we can keep purged objects in their own list to avoid re-traversing
them to help cut down time in the critical section further.

Signed-off-by: default avatarRob Clark <robdclark@chromium.org>
Tested-by: default avatarDouglas Anderson <dianders@chromium.org>
Reviewed-by: default avatarDouglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20210401012722.527712-3-robdclark@gmail.com


Signed-off-by: default avatarRob Clark <robdclark@chromium.org>
parent bc90dc33
Loading
Loading
Loading
Loading
+1 −0
Original line number Original line Diff line number Diff line
@@ -448,6 +448,7 @@ static int msm_drm_init(struct device *dev, const struct drm_driver *drv)


	INIT_LIST_HEAD(&priv->inactive_willneed);
	INIT_LIST_HEAD(&priv->inactive_willneed);
	INIT_LIST_HEAD(&priv->inactive_dontneed);
	INIT_LIST_HEAD(&priv->inactive_dontneed);
	INIT_LIST_HEAD(&priv->inactive_purged);
	mutex_init(&priv->mm_lock);
	mutex_init(&priv->mm_lock);


	/* Teach lockdep about lock ordering wrt. shrinker: */
	/* Teach lockdep about lock ordering wrt. shrinker: */
+4 −2
Original line number Original line Diff line number Diff line
@@ -179,8 +179,8 @@ struct msm_drm_private {
	 * inactive lists (depending on whether or not it is shrinkable) or
	 * inactive lists (depending on whether or not it is shrinkable) or
	 * gpu->active_list (for the gpu it is active on[1])
	 * gpu->active_list (for the gpu it is active on[1])
	 *
	 *
	 * These lists are protected by mm_lock.  If struct_mutex is involved, it
	 * These lists are protected by mm_lock (which should be acquired
	 * should be aquired prior to mm_lock.  One should *not* hold mm_lock in
	 * before per GEM object lock).  One should *not* hold mm_lock in
	 * get_pages()/vmap()/etc paths, as they can trigger the shrinker.
	 * get_pages()/vmap()/etc paths, as they can trigger the shrinker.
	 *
	 *
	 * [1] if someone ever added support for the old 2d cores, there could be
	 * [1] if someone ever added support for the old 2d cores, there could be
@@ -188,6 +188,8 @@ struct msm_drm_private {
	 */
	 */
	struct list_head inactive_willneed;  /* inactive + !shrinkable */
	struct list_head inactive_willneed;  /* inactive + !shrinkable */
	struct list_head inactive_dontneed;  /* inactive +  shrinkable */
	struct list_head inactive_dontneed;  /* inactive +  shrinkable */
	struct list_head inactive_purged;    /* inactive +  purged */
	long shrinkable_count;               /* write access under mm_lock */
	struct mutex mm_lock;
	struct mutex mm_lock;


	struct workqueue_struct *wq;
	struct workqueue_struct *wq;
+16 −4
Original line number Original line Diff line number Diff line
@@ -719,6 +719,7 @@ void msm_gem_purge(struct drm_gem_object *obj)
	put_iova_vmas(obj);
	put_iova_vmas(obj);


	msm_obj->madv = __MSM_MADV_PURGED;
	msm_obj->madv = __MSM_MADV_PURGED;
	mark_unpurgable(msm_obj);


	drm_vma_node_unmap(&obj->vma_node, dev->anon_inode->i_mapping);
	drm_vma_node_unmap(&obj->vma_node, dev->anon_inode->i_mapping);
	drm_gem_free_mmap_offset(obj);
	drm_gem_free_mmap_offset(obj);
@@ -790,10 +791,11 @@ void msm_gem_active_get(struct drm_gem_object *obj, struct msm_gpu *gpu)
	might_sleep();
	might_sleep();
	WARN_ON(!msm_gem_is_locked(obj));
	WARN_ON(!msm_gem_is_locked(obj));
	WARN_ON(msm_obj->madv != MSM_MADV_WILLNEED);
	WARN_ON(msm_obj->madv != MSM_MADV_WILLNEED);
	WARN_ON(msm_obj->dontneed);


	if (msm_obj->active_count++ == 0) {
	if (msm_obj->active_count++ == 0) {
		mutex_lock(&priv->mm_lock);
		mutex_lock(&priv->mm_lock);
		list_del_init(&msm_obj->mm_list);
		list_del(&msm_obj->mm_list);
		list_add_tail(&msm_obj->mm_list, &gpu->active_list);
		list_add_tail(&msm_obj->mm_list, &gpu->active_list);
		mutex_unlock(&priv->mm_lock);
		mutex_unlock(&priv->mm_lock);
	}
	}
@@ -818,11 +820,19 @@ static void update_inactive(struct msm_gem_object *msm_obj)
	mutex_lock(&priv->mm_lock);
	mutex_lock(&priv->mm_lock);
	WARN_ON(msm_obj->active_count != 0);
	WARN_ON(msm_obj->active_count != 0);


	list_del_init(&msm_obj->mm_list);
	if (msm_obj->dontneed)
	if (msm_obj->madv == MSM_MADV_WILLNEED)
		mark_unpurgable(msm_obj);

	list_del(&msm_obj->mm_list);
	if (msm_obj->madv == MSM_MADV_WILLNEED) {
		list_add_tail(&msm_obj->mm_list, &priv->inactive_willneed);
		list_add_tail(&msm_obj->mm_list, &priv->inactive_willneed);
	else
	} else if (msm_obj->madv == MSM_MADV_DONTNEED) {
		list_add_tail(&msm_obj->mm_list, &priv->inactive_dontneed);
		list_add_tail(&msm_obj->mm_list, &priv->inactive_dontneed);
		mark_purgable(msm_obj);
	} else {
		WARN_ON(msm_obj->madv != __MSM_MADV_PURGED);
		list_add_tail(&msm_obj->mm_list, &priv->inactive_purged);
	}


	mutex_unlock(&priv->mm_lock);
	mutex_unlock(&priv->mm_lock);
}
}
@@ -971,6 +981,8 @@ void msm_gem_free_object(struct drm_gem_object *obj)
	struct msm_drm_private *priv = dev->dev_private;
	struct msm_drm_private *priv = dev->dev_private;


	mutex_lock(&priv->mm_lock);
	mutex_lock(&priv->mm_lock);
	if (msm_obj->dontneed)
		mark_unpurgable(msm_obj);
	list_del(&msm_obj->mm_list);
	list_del(&msm_obj->mm_list);
	mutex_unlock(&priv->mm_lock);
	mutex_unlock(&priv->mm_lock);


+49 −4
Original line number Original line Diff line number Diff line
@@ -50,18 +50,24 @@ struct msm_gem_object {
	 */
	 */
	uint8_t madv;
	uint8_t madv;


	/**
	 * Is object on inactive_dontneed list (ie. counted in priv->shrinkable_count)?
	 */
	bool dontneed : 1;

	/**
	/**
	 * count of active vmap'ing
	 * count of active vmap'ing
	 */
	 */
	uint8_t vmap_count;
	uint8_t vmap_count;


	/* And object is either:
	/**
	 *  inactive - on priv->inactive_list
	 * An object is either:
	 *  inactive - on priv->inactive_dontneed/willneed/purged depending
	 *     on status
	 *  active   - on one one of the gpu's active_list..  well, at
	 *  active   - on one one of the gpu's active_list..  well, at
	 *     least for now we don't have (I don't think) hw sync between
	 *     least for now we don't have (I don't think) hw sync between
	 *     2d and 3d one devices which have both, meaning we need to
	 *     2d and 3d one devices which have both, meaning we need to
	 *     block on submit if a bo is already on other ring
	 *     block on submit if a bo is already on other ring
	 *
	 */
	 */
	struct list_head mm_list;
	struct list_head mm_list;


@@ -186,10 +192,16 @@ static inline bool is_active(struct msm_gem_object *msm_obj)
	return msm_obj->active_count;
	return msm_obj->active_count;
}
}


/* imported/exported objects are not purgable: */
static inline bool is_unpurgable(struct msm_gem_object *msm_obj)
{
	return msm_obj->base.dma_buf && msm_obj->base.import_attach;
}

static inline bool is_purgeable(struct msm_gem_object *msm_obj)
static inline bool is_purgeable(struct msm_gem_object *msm_obj)
{
{
	return (msm_obj->madv == MSM_MADV_DONTNEED) && msm_obj->sgt &&
	return (msm_obj->madv == MSM_MADV_DONTNEED) && msm_obj->sgt &&
			!msm_obj->base.dma_buf && !msm_obj->base.import_attach;
			!is_unpurgable(msm_obj);
}
}


static inline bool is_vunmapable(struct msm_gem_object *msm_obj)
static inline bool is_vunmapable(struct msm_gem_object *msm_obj)
@@ -198,6 +210,39 @@ static inline bool is_vunmapable(struct msm_gem_object *msm_obj)
	return (msm_obj->vmap_count == 0) && msm_obj->vaddr;
	return (msm_obj->vmap_count == 0) && msm_obj->vaddr;
}
}


static inline void mark_purgable(struct msm_gem_object *msm_obj)
{
	struct msm_drm_private *priv = msm_obj->base.dev->dev_private;

	WARN_ON(!mutex_is_locked(&priv->mm_lock));

	if (is_unpurgable(msm_obj))
		return;

	if (WARN_ON(msm_obj->dontneed))
		return;

	priv->shrinkable_count += msm_obj->base.size >> PAGE_SHIFT;
	msm_obj->dontneed = true;
}

static inline void mark_unpurgable(struct msm_gem_object *msm_obj)
{
	struct msm_drm_private *priv = msm_obj->base.dev->dev_private;

	WARN_ON(!mutex_is_locked(&priv->mm_lock));

	if (is_unpurgable(msm_obj))
		return;

	if (WARN_ON(!msm_obj->dontneed))
		return;

	priv->shrinkable_count -= msm_obj->base.size >> PAGE_SHIFT;
	WARN_ON(priv->shrinkable_count < 0);
	msm_obj->dontneed = false;
}

void msm_gem_purge(struct drm_gem_object *obj);
void msm_gem_purge(struct drm_gem_object *obj);
void msm_gem_vunmap(struct drm_gem_object *obj);
void msm_gem_vunmap(struct drm_gem_object *obj);


+11 −17
Original line number Original line Diff line number Diff line
@@ -14,22 +14,7 @@ msm_gem_shrinker_count(struct shrinker *shrinker, struct shrink_control *sc)
{
{
	struct msm_drm_private *priv =
	struct msm_drm_private *priv =
		container_of(shrinker, struct msm_drm_private, shrinker);
		container_of(shrinker, struct msm_drm_private, shrinker);
	struct msm_gem_object *msm_obj;
	return priv->shrinkable_count;
	unsigned long count = 0;

	mutex_lock(&priv->mm_lock);

	list_for_each_entry(msm_obj, &priv->inactive_dontneed, mm_list) {
		if (!msm_gem_trylock(&msm_obj->base))
			continue;
		if (is_purgeable(msm_obj))
			count += msm_obj->base.size >> PAGE_SHIFT;
		msm_gem_unlock(&msm_obj->base);
	}

	mutex_unlock(&priv->mm_lock);

	return count;
}
}


static unsigned long
static unsigned long
@@ -45,6 +30,9 @@ msm_gem_shrinker_scan(struct shrinker *shrinker, struct shrink_control *sc)
	list_for_each_entry(msm_obj, &priv->inactive_dontneed, mm_list) {
	list_for_each_entry(msm_obj, &priv->inactive_dontneed, mm_list) {
		if (freed >= sc->nr_to_scan)
		if (freed >= sc->nr_to_scan)
			break;
			break;
		/* Use trylock, because we cannot block on a obj that
		 * might be trying to acquire mm_lock
		 */
		if (!msm_gem_trylock(&msm_obj->base))
		if (!msm_gem_trylock(&msm_obj->base))
			continue;
			continue;
		if (is_purgeable(msm_obj)) {
		if (is_purgeable(msm_obj)) {
@@ -56,8 +44,11 @@ msm_gem_shrinker_scan(struct shrinker *shrinker, struct shrink_control *sc)


	mutex_unlock(&priv->mm_lock);
	mutex_unlock(&priv->mm_lock);


	if (freed > 0)
	if (freed > 0) {
		trace_msm_gem_purge(freed << PAGE_SHIFT);
		trace_msm_gem_purge(freed << PAGE_SHIFT);
	} else {
		return SHRINK_STOP;
	}


	return freed;
	return freed;
}
}
@@ -75,6 +66,9 @@ vmap_shrink(struct list_head *mm_list)
	unsigned unmapped = 0;
	unsigned unmapped = 0;


	list_for_each_entry(msm_obj, mm_list, mm_list) {
	list_for_each_entry(msm_obj, mm_list, mm_list) {
		/* Use trylock, because we cannot block on a obj that
		 * might be trying to acquire mm_lock
		 */
		if (!msm_gem_trylock(&msm_obj->base))
		if (!msm_gem_trylock(&msm_obj->base))
			continue;
			continue;
		if (is_vunmapable(msm_obj)) {
		if (is_vunmapable(msm_obj)) {