[PATCH 0/5] Add reclaim to the dmem cgroup controller

[PATCH 0/5] Add reclaim to the dmem cgroup controller

[Thomas Hellström]

When writing a "max" limit lower than the current usage, the
existing code silently failed. This series aims to improve
on that by returning -EBUSY on failure and also attempt
to synchronously reclaim device memory to push the usage
under the new max limit to avoid the error.

Patch 1 implements error propagation.
Patch 2 implements and documents a reclaim callback interface
      for the dmem controller.
Patch 3 implements a TTM reclaim callback.
Patch 4-5 hooks up the reclaim callback to the dmem cgroups-
      aware drivers xe and amdgpu.

Thomas Hellström (5):
  cgroup/dmem: Return error when setting max below current usage
  cgroup/dmem: Add reclaim callback for lowering max below current usage
  drm/ttm: Hook up a cgroup-aware reclaim callback for the dmem
    controller
  drm/xe: Wire up dmem cgroup reclaim for VRAM manager
  drm/amdgpu: Wire up dmem cgroup reclaim for VRAM manager

 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c      |   2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c |  10 +-
 drivers/gpu/drm/ttm/ttm_bo.c                 |  95 ++++++++++++++++-
 drivers/gpu/drm/ttm/ttm_bo_util.c            |   3 +-
 drivers/gpu/drm/ttm/ttm_resource.c           |  36 +++++++
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c         |  19 ++--
 include/drm/ttm/ttm_bo.h                     |  10 ++
 include/drm/ttm/ttm_resource.h               |   4 +
 include/linux/cgroup_dmem.h                  |  11 ++
 kernel/cgroup/dmem.c                         | 102 ++++++++++++++++---
 10 files changed, 265 insertions(+), 27 deletions(-)

-- 
2.53.0

[PATCH 1/5] cgroup/dmem: Return error when setting max below current usage

[Thomas Hellström]

Return -EBUSY to userspace when writing dmem.max below the cgroup's
current device memory usage, rather than silently leaving the limit
unchanged.

Assisted-by: GitHub Copilot:claude-sonnet-4.6
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
---
 kernel/cgroup/dmem.c | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/kernel/cgroup/dmem.c b/kernel/cgroup/dmem.c
index 9d95824dc6fa..3e6d4c0b26a1 100644
--- a/kernel/cgroup/dmem.c
+++ b/kernel/cgroup/dmem.c
@@ -144,22 +144,24 @@ static void free_cg_pool(struct dmem_cgroup_pool_state *pool)
 	dmemcg_pool_put(pool);
 }
 
-static void
+static int
 set_resource_min(struct dmem_cgroup_pool_state *pool, u64 val)
 {
 	page_counter_set_min(&pool->cnt, val);
+	return 0;
 }
 
-static void
+static int
 set_resource_low(struct dmem_cgroup_pool_state *pool, u64 val)
 {
 	page_counter_set_low(&pool->cnt, val);
+	return 0;
 }
 
-static void
+static int
 set_resource_max(struct dmem_cgroup_pool_state *pool, u64 val)
 {
-	page_counter_set_max(&pool->cnt, val);
+	return page_counter_set_max(&pool->cnt, val);
 }
 
 static u64 get_resource_low(struct dmem_cgroup_pool_state *pool)
@@ -726,7 +728,7 @@ static int dmemcg_parse_limit(char *options, struct dmem_cgroup_region *region,
 
 static ssize_t dmemcg_limit_write(struct kernfs_open_file *of,
 				 char *buf, size_t nbytes, loff_t off,
-				 void (*apply)(struct dmem_cgroup_pool_state *, u64))
+				 int (*apply)(struct dmem_cgroup_pool_state *, u64))
 {
 	struct dmemcg_state *dmemcs = css_to_dmemcs(of_css(of));
 	int err = 0;
@@ -773,7 +775,7 @@ static ssize_t dmemcg_limit_write(struct kernfs_open_file *of,
 		}
 
 		/* And commit */
-		apply(pool, new_limit);
+		err = apply(pool, new_limit);
 		dmemcg_pool_put(pool);
 
 out_put:
-- 
2.53.0

[PATCH 2/5] cgroup/dmem: Add reclaim callback for lowering max below current usage

[Thomas Hellström]

Add an optional reclaim callback to struct dmem_cgroup_region.  When
dmem.max is set below current usage, invoke the callback to evict memory
and retry setting the limit rather than failing immediately.  Signal
interruptions propagate back to the write() caller.

RFC:
Due to us updating the max limit _after_ the usage has been
sufficiently lowered, this should be prone to failures if there are
aggressive allocators running in parallel to the reclaim.
So can we somehow enforce the new limit while the eviction is
happening?

Assisted-by: GitHub Copilot:claude-sonnet-4.6
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
---
 include/linux/cgroup_dmem.h | 11 +++++
 kernel/cgroup/dmem.c        | 94 +++++++++++++++++++++++++++++++++----
 2 files changed, 96 insertions(+), 9 deletions(-)

diff --git a/include/linux/cgroup_dmem.h b/include/linux/cgroup_dmem.h
index dd4869f1d736..61520a431740 100644
--- a/include/linux/cgroup_dmem.h
+++ b/include/linux/cgroup_dmem.h
@@ -26,6 +26,10 @@ bool dmem_cgroup_state_evict_valuable(struct dmem_cgroup_pool_state *limit_pool,
 				      bool ignore_low, bool *ret_hit_low);
 
 void dmem_cgroup_pool_state_put(struct dmem_cgroup_pool_state *pool);
+void dmem_cgroup_region_set_reclaim(struct dmem_cgroup_region *region,
+				    int (*reclaim)(struct dmem_cgroup_pool_state *pool,
+						   u64 target_bytes, void *priv),
+				    void *priv);
 #else
 static inline __printf(2,3) struct dmem_cgroup_region *
 dmem_cgroup_register_region(u64 size, const char *name_fmt, ...)
@@ -62,5 +66,12 @@ bool dmem_cgroup_state_evict_valuable(struct dmem_cgroup_pool_state *limit_pool,
 static inline void dmem_cgroup_pool_state_put(struct dmem_cgroup_pool_state *pool)
 { }
 
+static inline void
+dmem_cgroup_region_set_reclaim(struct dmem_cgroup_region *region,
+			       int (*reclaim)(struct dmem_cgroup_pool_state *pool,
+					      u64 target_bytes, void *priv),
+			       void *priv)
+{ }
+
 #endif
 #endif	/* _CGROUP_DMEM_H */
diff --git a/kernel/cgroup/dmem.c b/kernel/cgroup/dmem.c
index 3e6d4c0b26a1..f993fb058b74 100644
--- a/kernel/cgroup/dmem.c
+++ b/kernel/cgroup/dmem.c
@@ -51,6 +51,18 @@ struct dmem_cgroup_region {
 	 * No new pools should be added to the region afterwards.
 	 */
 	bool unregistered;
+
+	/**
+	 * @reclaim: Optional callback invoked when dmem.max is set below the
+	 * current usage of a pool. The driver should attempt to free at least
+	 * @target_bytes from @pool. May be called multiple times if usage
+	 * remains above the limit after returning.
+	 */
+	int (*reclaim)(struct dmem_cgroup_pool_state *pool, u64 target_bytes,
+		       void *priv);
+
+	/** @reclaim_priv: Private data passed to @reclaim. */
+	void *reclaim_priv;
 };
 
 struct dmemcg_state {
@@ -145,23 +157,59 @@ static void free_cg_pool(struct dmem_cgroup_pool_state *pool)
 }
 
 static int
-set_resource_min(struct dmem_cgroup_pool_state *pool, u64 val)
+set_resource_min(struct dmem_cgroup_pool_state *pool, u64 val,
+		 struct dmem_cgroup_region *region)
 {
 	page_counter_set_min(&pool->cnt, val);
 	return 0;
 }
 
 static int
-set_resource_low(struct dmem_cgroup_pool_state *pool, u64 val)
+set_resource_low(struct dmem_cgroup_pool_state *pool, u64 val,
+		 struct dmem_cgroup_region *region)
 {
 	page_counter_set_low(&pool->cnt, val);
 	return 0;
 }
 
 static int
-set_resource_max(struct dmem_cgroup_pool_state *pool, u64 val)
+set_resource_max(struct dmem_cgroup_pool_state *pool, u64 val,
+		 struct dmem_cgroup_region *region)
 {
-	return page_counter_set_max(&pool->cnt, val);
+	int err = page_counter_set_max(&pool->cnt, val);
+
+	if (err != -EBUSY || !region || !region->reclaim)
+		return err;
+
+	/*
+	 * The new max is below current usage.  Ask the driver to evict memory
+	 * and retry, up to a bounded number of times.  Signal interruptions are
+	 * propagated back to the write() caller; other reclaim failures leave
+	 * -EBUSY as the result.
+	 */
+	for (int retries = 5; retries > 0; retries--) {
+		u64 usage = page_counter_read(&pool->cnt);
+		u64 target = usage > val ? usage - val : 0;
+		int reclaim_err;
+
+		if (!target) {
+			err = page_counter_set_max(&pool->cnt, val);
+			break;
+		}
+
+		reclaim_err = region->reclaim(pool, target, region->reclaim_priv);
+		if (reclaim_err) {
+			if (reclaim_err == -EINTR || reclaim_err == -ERESTARTSYS)
+				err = reclaim_err;
+			break;
+		}
+
+		err = page_counter_set_max(&pool->cnt, val);
+		if (err != -EBUSY)
+			break;
+	}
+
+	return err;
 }
 
 static u64 get_resource_low(struct dmem_cgroup_pool_state *pool)
@@ -186,9 +234,9 @@ static u64 get_resource_current(struct dmem_cgroup_pool_state *pool)
 
 static void reset_all_resource_limits(struct dmem_cgroup_pool_state *rpool)
 {
-	set_resource_min(rpool, 0);
-	set_resource_low(rpool, 0);
-	set_resource_max(rpool, PAGE_COUNTER_MAX);
+	set_resource_min(rpool, 0, NULL);
+	set_resource_low(rpool, 0, NULL);
+	set_resource_max(rpool, PAGE_COUNTER_MAX, NULL);
 }
 
 static void dmemcs_offline(struct cgroup_subsys_state *css)
@@ -570,6 +618,32 @@ void dmem_cgroup_pool_state_put(struct dmem_cgroup_pool_state *pool)
 }
 EXPORT_SYMBOL_GPL(dmem_cgroup_pool_state_put);
 
+/**
+ * dmem_cgroup_region_set_reclaim - Register a reclaim callback on a region.
+ * @region: The region to register the callback for.
+ * @reclaim: Callback to invoke when dmem.max is set below current usage.
+ *           Called with the pool that needs reclaiming and the number of
+ *           bytes to free. Returns 0 on progress, negative on failure.
+ * @priv: Opaque pointer passed back to @reclaim.
+ *
+ * When dmem.max is lowered below the current usage of a cgroup pool, the
+ * dmem controller will call @reclaim with a target number of bytes to free.
+ * After @reclaim returns the controller retries setting the limit; if usage
+ * is still too high it calls @reclaim again, up to a bounded retry count.
+ */
+void dmem_cgroup_region_set_reclaim(struct dmem_cgroup_region *region,
+				    int (*reclaim)(struct dmem_cgroup_pool_state *pool,
+						   u64 target_bytes, void *priv),
+				    void *priv)
+{
+	if (!region)
+		return;
+
+	region->reclaim = reclaim;
+	region->reclaim_priv = priv;
+}
+EXPORT_SYMBOL_GPL(dmem_cgroup_region_set_reclaim);
+
 static struct dmem_cgroup_pool_state *
 get_cg_pool_unlocked(struct dmemcg_state *cg, struct dmem_cgroup_region *region)
 {
@@ -728,7 +802,8 @@ static int dmemcg_parse_limit(char *options, struct dmem_cgroup_region *region,
 
 static ssize_t dmemcg_limit_write(struct kernfs_open_file *of,
 				 char *buf, size_t nbytes, loff_t off,
-				 int (*apply)(struct dmem_cgroup_pool_state *, u64))
+				 int (*apply)(struct dmem_cgroup_pool_state *, u64,
+					      struct dmem_cgroup_region *))
 {
 	struct dmemcg_state *dmemcs = css_to_dmemcs(of_css(of));
 	int err = 0;
@@ -775,7 +850,8 @@ static ssize_t dmemcg_limit_write(struct kernfs_open_file *of,
 		}
 
 		/* And commit */
-		err = apply(pool, new_limit);
+		err = apply(pool, new_limit, region);
+
 		dmemcg_pool_put(pool);
 
 out_put:
-- 
2.53.0

[PATCH 3/5] drm/ttm: Hook up a cgroup-aware reclaim callback for the dmem controller

[Thomas Hellström]

Add ttm_bo_evict_cgroup() to evict buffer objects charged to a specific
dmem cgroup pool from a resource manager's LRU until a byte target is
met.  Add ttm_resource_manager_set_dmem_region() to register the TTM
eviction path as the reclaim callback for a dmem cgroup region.

The eviction context is interruptible; signals abort the operation and
propagate back through the write() syscall.

Introduce a new mode for the bo LRU walker so that sleeping locks
can be taken. This can be used when the caller doesn't hold any
previous dma_resv locks, and where it intends to hold at most
one lock at a time.

Like the rest of the TTM eviction this should sooner than later
be converted to full WW transactions.

Assisted-by: GitHub Copilot:claude-sonnet-4.6
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
---
 drivers/gpu/drm/ttm/ttm_bo.c       | 95 +++++++++++++++++++++++++++++-
 drivers/gpu/drm/ttm/ttm_bo_util.c  |  3 +-
 drivers/gpu/drm/ttm/ttm_resource.c | 36 +++++++++++
 include/drm/ttm/ttm_bo.h           | 10 ++++
 include/drm/ttm/ttm_resource.h     |  4 ++
 5 files changed, 144 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
index d85f0a37ac35..1745557c184c 100644
--- a/drivers/gpu/drm/ttm/ttm_bo.c
+++ b/drivers/gpu/drm/ttm/ttm_bo.c
@@ -515,12 +515,20 @@ static s64 ttm_bo_evict_cb(struct ttm_lru_walk *walk, struct ttm_buffer_object *
 {
 	struct ttm_bo_evict_walk *evict_walk =
 		container_of(walk, typeof(*evict_walk), walk);
+	/* Capture size before eviction in case res is cleared. */
+	s64 bo_size = bo->base.size;
 	s64 lret;
 
 	if (!dmem_cgroup_state_evict_valuable(evict_walk->limit_pool, bo->resource->css,
 					      evict_walk->try_low, &evict_walk->hit_low))
 		return 0;
 
+	/*
+	 * evict_walk->place is NULL in cgroup drain mode.  Drivers'
+	 * eviction_valuable() callbacks must handle a NULL place, treating it
+	 * as "any placement": the TTM base implementation already does so via
+	 * ttm_resource_intersects().
+	 */
 	if (bo->pin_count || !bo->bdev->funcs->eviction_valuable(bo, evict_walk->place))
 		return 0;
 
@@ -536,11 +544,15 @@ static s64 ttm_bo_evict_cb(struct ttm_lru_walk *walk, struct ttm_buffer_object *
 		goto out;
 
 	evict_walk->evicted++;
-	if (evict_walk->res)
+	if (evict_walk->res) {
 		lret = ttm_resource_alloc(evict_walk->evictor, evict_walk->place,
 					  evict_walk->res, NULL);
-	if (lret == 0)
-		return 1;
+		if (lret == 0)
+			return 1;
+	} else {
+		/* Cgroup drain: return bytes freed for byte-denominated progress. */
+		return bo_size;
+	}
 out:
 	/* Errors that should terminate the walk. */
 	if (lret == -ENOSPC)
@@ -614,6 +626,83 @@ static int ttm_bo_evict_alloc(struct ttm_device *bdev,
 	return 0;
 }
 
+/**
+ * ttm_bo_evict_cgroup - Evict buffer objects charged to a specific cgroup.
+ * @bdev: The TTM device.
+ * @man: The resource manager whose LRU to walk.
+ * @limit_pool: The cgroup pool state whose members should be evicted.
+ * @target_bytes: Number of bytes to free.
+ * @ctx: The TTM operation context.
+ *
+ * Walk the LRU of @man and evict buffer objects that are charged to the
+ * cgroup identified by @limit_pool, until at least @target_bytes have been
+ * freed.  Mirrors the two-pass (trylock -> sleeping-lock, low-watermark)
+ * strategy used by ttm_bo_evict_alloc().
+ *
+ * Return: >= @target_bytes on full success, 0..target_bytes-1 if partial,
+ *         negative error code on fatal error.
+ */
+s64 ttm_bo_evict_cgroup(struct ttm_device *bdev,
+			struct ttm_resource_manager *man,
+			struct dmem_cgroup_pool_state *limit_pool,
+			s64 target_bytes,
+			struct ttm_operation_ctx *ctx)
+{
+	struct ttm_bo_evict_walk evict_walk = {
+		.walk = {
+			.ops = &ttm_evict_walk_ops,
+			.arg = { .ctx = ctx },
+		},
+		.limit_pool = limit_pool,
+		/* place, evictor, res left NULL: selects cgroup drain mode */
+	};
+	s64 lret, pass;
+
+	evict_walk.walk.arg.trylock_only = true;
+	lret = ttm_lru_walk_for_evict(&evict_walk.walk, bdev, man, target_bytes);
+	if (lret < 0 || lret >= target_bytes)
+		return lret;
+
+	/* Second pass: also evict BOs at the low watermark. */
+	if (evict_walk.hit_low) {
+		evict_walk.try_low = true;
+		pass = ttm_lru_walk_for_evict(&evict_walk.walk, bdev, man,
+					      target_bytes - lret);
+		if (pass < 0)
+			return pass;
+		lret += pass;
+		if (lret >= target_bytes)
+			return lret;
+	}
+
+	/* Full sleeping-lock pass for remaining target. */
+	evict_walk.try_low = evict_walk.hit_low = false;
+	evict_walk.walk.arg.trylock_only = false;
+
+retry:
+	evict_walk.walk.arg.sleeping_lock = true;
+	do {
+		evict_walk.evicted = 0;
+		pass = ttm_lru_walk_for_evict(&evict_walk.walk, bdev, man,
+					      target_bytes - lret);
+		if (pass < 0) {
+			lret = pass;
+			goto out;
+		}
+		lret += pass;
+	} while (lret < target_bytes && evict_walk.evicted);
+
+	/* One more attempt if we hit the low limit during sleeping-lock pass. */
+	if (lret < target_bytes && evict_walk.hit_low && !evict_walk.try_low) {
+		evict_walk.try_low = true;
+		goto retry;
+	}
+
+out:
+	return lret;
+}
+EXPORT_SYMBOL(ttm_bo_evict_cgroup);
+
 /**
  * ttm_bo_pin - Pin the buffer object.
  * @bo: The buffer object to pin
diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c
index f83b7d5ec6c6..81c6a674c462 100644
--- a/drivers/gpu/drm/ttm/ttm_bo_util.c
+++ b/drivers/gpu/drm/ttm/ttm_bo_util.c
@@ -999,7 +999,8 @@ __ttm_bo_lru_cursor_next(struct ttm_bo_lru_cursor *curs)
 		bo = res->bo;
 		if (ttm_lru_walk_trylock(curs, bo))
 			bo_locked = true;
-		else if (!arg->ticket || arg->ctx->no_wait_gpu || arg->trylock_only)
+		else if ((!arg->ticket && !arg->sleeping_lock) || arg->ctx->no_wait_gpu ||
+			 arg->trylock_only)
 			continue;
 
 		if (!ttm_bo_get_unless_zero(bo)) {
diff --git a/drivers/gpu/drm/ttm/ttm_resource.c b/drivers/gpu/drm/ttm/ttm_resource.c
index 9f36631d48b6..936552f426a7 100644
--- a/drivers/gpu/drm/ttm/ttm_resource.c
+++ b/drivers/gpu/drm/ttm/ttm_resource.c
@@ -937,3 +937,39 @@ void ttm_resource_manager_create_debugfs(struct ttm_resource_manager *man,
 #endif
 }
 EXPORT_SYMBOL(ttm_resource_manager_create_debugfs);
+
+static int ttm_resource_manager_dmem_reclaim(struct dmem_cgroup_pool_state *pool,
+					     u64 target_bytes, void *priv)
+{
+	struct ttm_resource_manager *man = priv;
+	struct ttm_operation_ctx ctx = { .interruptible = true };
+	s64 freed;
+
+	freed = ttm_bo_evict_cgroup(man->bdev, man, pool, target_bytes, &ctx);
+	if (freed < 0)
+		return freed;
+
+	return freed >= (s64)target_bytes ? 0 : -ENOSPC;
+}
+
+/**
+ * ttm_resource_manager_set_dmem_region - Associate a dmem cgroup region with a
+ *                                        resource manager and register a reclaim
+ *                                        callback.
+ * @man: The resource manager.
+ * @region: The dmem cgroup region to associate, may be NULL or IS_ERR().
+ *
+ * Sets @man->cg and registers ttm_resource_manager_dmem_reclaim() so that
+ * writing to dmem.max below current usage triggers TTM eviction rather than
+ * returning -EBUSY to userspace.
+ */
+void ttm_resource_manager_set_dmem_region(struct ttm_resource_manager *man,
+					  struct dmem_cgroup_region *region)
+{
+	man->cg = region;
+	if (!IS_ERR_OR_NULL(region))
+		dmem_cgroup_region_set_reclaim(region,
+					       ttm_resource_manager_dmem_reclaim,
+					       man);
+}
+EXPORT_SYMBOL(ttm_resource_manager_set_dmem_region);
diff --git a/include/drm/ttm/ttm_bo.h b/include/drm/ttm/ttm_bo.h
index 8310bc3d55f9..32791c4db2a9 100644
--- a/include/drm/ttm/ttm_bo.h
+++ b/include/drm/ttm/ttm_bo.h
@@ -226,6 +226,11 @@ struct ttm_lru_walk_arg {
 	struct ww_acquire_ctx *ticket;
 	/** @trylock_only: Only use trylock for locking. */
 	bool trylock_only;
+	/**
+	 * @sleeping_lock: Use sleeping locks even with %NULL @ticket.
+	 * @trylock_only has precedence over this field.
+	 */
+	bool sleeping_lock;
 };
 
 /**
@@ -431,6 +436,11 @@ void ttm_bo_unpin(struct ttm_buffer_object *bo);
 int ttm_bo_evict_first(struct ttm_device *bdev,
 		       struct ttm_resource_manager *man,
 		       struct ttm_operation_ctx *ctx);
+s64 ttm_bo_evict_cgroup(struct ttm_device *bdev,
+			struct ttm_resource_manager *man,
+			struct dmem_cgroup_pool_state *limit_pool,
+			s64 target_bytes,
+			struct ttm_operation_ctx *ctx);
 int ttm_bo_access(struct ttm_buffer_object *bo, unsigned long offset,
 		  void *buf, int len, int write);
 vm_fault_t ttm_bo_vm_reserve(struct ttm_buffer_object *bo,
diff --git a/include/drm/ttm/ttm_resource.h b/include/drm/ttm/ttm_resource.h
index 33e80f30b8b8..c187e6c8b871 100644
--- a/include/drm/ttm/ttm_resource.h
+++ b/include/drm/ttm/ttm_resource.h
@@ -39,6 +39,7 @@
 
 struct dentry;
 struct dmem_cgroup_device;
+struct dmem_cgroup_region;
 struct drm_printer;
 struct ttm_device;
 struct ttm_resource_manager;
@@ -475,6 +476,9 @@ void ttm_resource_manager_init(struct ttm_resource_manager *man,
 			       struct ttm_device *bdev,
 			       uint64_t size);
 
+void ttm_resource_manager_set_dmem_region(struct ttm_resource_manager *man,
+					  struct dmem_cgroup_region *region);
+
 int ttm_resource_manager_evict_all(struct ttm_device *bdev,
 				   struct ttm_resource_manager *man);
 
-- 
2.53.0

[PATCH 4/5] drm/xe: Wire up dmem cgroup reclaim for VRAM manager

[Thomas Hellström]

Register the VRAM manager with the dmem cgroup reclaim infrastructure
so that lowering dmem.max below current VRAM usage triggers TTM
eviction rather than failing with -EBUSY.

Assisted-by: GitHub Copilot:claude-sonnet-4.6
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
---
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/xe/xe_ttm_vram_mgr.c b/drivers/gpu/drm/xe/xe_ttm_vram_mgr.c
index 5fd0d5506a7e..1bdcb3fee901 100644
--- a/drivers/gpu/drm/xe/xe_ttm_vram_mgr.c
+++ b/drivers/gpu/drm/xe/xe_ttm_vram_mgr.c
@@ -303,13 +303,6 @@ int __xe_ttm_vram_mgr_init(struct xe_device *xe, struct xe_ttm_vram_mgr *mgr,
 	struct ttm_resource_manager *man = &mgr->manager;
 	int err;
 
-	if (mem_type != XE_PL_STOLEN) {
-		const char *name = mem_type == XE_PL_VRAM0 ? "vram0" : "vram1";
-		man->cg = drmm_cgroup_register_region(&xe->drm, name, size);
-		if (IS_ERR(man->cg))
-			return PTR_ERR(man->cg);
-	}
-
 	man->func = &xe_ttm_vram_mgr_func;
 	mgr->mem_type = mem_type;
 	mutex_init(&mgr->lock);
@@ -318,6 +311,18 @@ int __xe_ttm_vram_mgr_init(struct xe_device *xe, struct xe_ttm_vram_mgr *mgr,
 	mgr->visible_avail = io_size;
 
 	ttm_resource_manager_init(man, &xe->ttm, size);
+
+	if (mem_type != XE_PL_STOLEN) {
+		const char *name = mem_type == XE_PL_VRAM0 ? "vram0" : "vram1";
+		struct dmem_cgroup_region *cg =
+			drmm_cgroup_register_region(&xe->drm, name, size);
+
+		if (IS_ERR(cg))
+			return PTR_ERR(cg);
+
+		ttm_resource_manager_set_dmem_region(man, cg);
+	}
+
 	err = gpu_buddy_init(&mgr->mm, man->size, default_page_size);
 	if (err)
 		return err;
-- 
2.53.0

[PATCH 5/5] drm/amdgpu: Wire up dmem cgroup reclaim for VRAM manager

[Thomas Hellström]

Register the VRAM manager with the dmem cgroup reclaim infrastructure
so that lowering dmem.max below current VRAM usage triggers TTM
eviction rather than failing with -EBUSY.

Guard place->flags in amdgpu_ttm_bo_eviction_valuable() against NULL,
as the TTM reclaim path passes a NULL place in cgroup drain mode.

Assisted-by: GitHub Copilot:claude-sonnet-4.6
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c      |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c | 10 +++++++---
 2 files changed, 8 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index b4ab309bf08a..cd83f30a30f7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -1491,7 +1491,7 @@ static bool amdgpu_ttm_bo_eviction_valuable(struct ttm_buffer_object *bo,
 	dma_resv_for_each_fence(&resv_cursor, bo->base.resv,
 				DMA_RESV_USAGE_BOOKKEEP, f) {
 		if (amdkfd_fence_check_mm(f, current->mm) &&
-		    !(place->flags & TTM_PL_FLAG_CONTIGUOUS))
+		    !(place && (place->flags & TTM_PL_FLAG_CONTIGUOUS)))
 			return false;
 	}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
index 2a241a5b12c4..5987b1b9ec09 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
@@ -916,14 +916,18 @@ int amdgpu_vram_mgr_init(struct amdgpu_device *adev)
 {
 	struct amdgpu_vram_mgr *mgr = &adev->mman.vram_mgr;
 	struct ttm_resource_manager *man = &mgr->manager;
+	struct dmem_cgroup_region *cg;
 	int err;
 
-	man->cg = drmm_cgroup_register_region(adev_to_drm(adev), "vram", adev->gmc.real_vram_size);
-	if (IS_ERR(man->cg))
-		return PTR_ERR(man->cg);
 	ttm_resource_manager_init(man, &adev->mman.bdev,
 				  adev->gmc.real_vram_size);
 
+	cg = drmm_cgroup_register_region(adev_to_drm(adev), "vram",
+					 adev->gmc.real_vram_size);
+	if (IS_ERR(cg))
+		return PTR_ERR(cg);
+	ttm_resource_manager_set_dmem_region(man, cg);
+
 	mutex_init(&mgr->lock);
 	INIT_LIST_HEAD(&mgr->reservations_pending);
 	INIT_LIST_HEAD(&mgr->reserved_pages);
-- 
2.53.0

Re: [PATCH 0/5] Add reclaim to the dmem cgroup controller

[Thomas Hellström]

On Fri, 2026-03-27 at 09:15 +0100, Thomas Hellström wrote:
> When writing a "max" limit lower than the current usage, the
> existing code silently failed. This series aims to improve
> on that by returning -EBUSY on failure and also attempt
> to synchronously reclaim device memory to push the usage
> under the new max limit to avoid the error.
> 
> Patch 1 implements error propagation.
> Patch 2 implements and documents a reclaim callback interface
>       for the dmem controller.
> Patch 3 implements a TTM reclaim callback.
> Patch 4-5 hooks up the reclaim callback to the dmem cgroups-
>       aware drivers xe and amdgpu.
> 
> Thomas Hellström (5):
>   cgroup/dmem: Return error when setting max below current usage
>   cgroup/dmem: Add reclaim callback for lowering max below current
> usage
>   drm/ttm: Hook up a cgroup-aware reclaim callback for the dmem
>     controller
>   drm/xe: Wire up dmem cgroup reclaim for VRAM manager
>   drm/amdgpu: Wire up dmem cgroup reclaim for VRAM manager
> 
>  drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c      |   2 +-
>  drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c |  10 +-
>  drivers/gpu/drm/ttm/ttm_bo.c                 |  95 ++++++++++++++++-
>  drivers/gpu/drm/ttm/ttm_bo_util.c            |   3 +-
>  drivers/gpu/drm/ttm/ttm_resource.c           |  36 +++++++
>  drivers/gpu/drm/xe/xe_ttm_vram_mgr.c         |  19 ++--
>  include/drm/ttm/ttm_bo.h                     |  10 ++
>  include/drm/ttm/ttm_resource.h               |   4 +
>  include/linux/cgroup_dmem.h                  |  11 ++
>  kernel/cgroup/dmem.c                         | 102 ++++++++++++++++-
> --
>  10 files changed, 265 insertions(+), 27 deletions(-)

For reference, Series introducing igt test is here:
https://patchwork.freedesktop.org/series/163935/

/Thomas