dma-buf: add dma_fence_chain_alloc/free v3

Add a common allocation helper. Cleaning up the mix of kzalloc/kmalloc
and some unused code in the selftest.

v2: polish kernel doc a bit
v3: polish kernel doc even a bit more
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210611120301.10595-3-christian.koenig@amd.com

drivers/dma-buf/st-dma-fence-chain.c

@@ -58,28 +58,20 @@ static struct dma_fence *mock_fence(void)
 	return &f->base;
 }
 
-static inline struct mock_chain {
-	struct dma_fence_chain base;
-} *to_mock_chain(struct dma_fence *f) {
-	return container_of(f, struct mock_chain, base.base);
-}
-
 static struct dma_fence *mock_chain(struct dma_fence *prev,
 				    struct dma_fence *fence,
 				    u64 seqno)
 {
-	struct mock_chain *f;
+	struct dma_fence_chain *f;
 
-	f = kmalloc(sizeof(*f), GFP_KERNEL);
+	f = dma_fence_chain_alloc();
 	if (!f)
 		return NULL;
 
-	dma_fence_chain_init(&f->base,
-			     dma_fence_get(prev),
-			     dma_fence_get(fence),
+	dma_fence_chain_init(f, dma_fence_get(prev), dma_fence_get(fence),
 			     seqno);
 
-	return &f->base.base;
+	return &f->base;
 }
 
 static int sanitycheck(void *arg)

drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c

@@ -1109,7 +1109,7 @@ static int amdgpu_cs_process_syncobj_timeline_out_dep(struct amdgpu_cs_parser *p
 
 		dep->chain = NULL;
 		if (syncobj_deps[i].point) {
-			dep->chain = kmalloc(sizeof(*dep->chain), GFP_KERNEL);
+			dep->chain = dma_fence_chain_alloc();
 			if (!dep->chain)
 				return -ENOMEM;
 		}
@@ -1117,7 +1117,7 @@ static int amdgpu_cs_process_syncobj_timeline_out_dep(struct amdgpu_cs_parser *p
 		dep->syncobj = drm_syncobj_find(p->filp,
 						syncobj_deps[i].handle);
 		if (!dep->syncobj) {
-			kfree(dep->chain);
+			dma_fence_chain_free(dep->chain);
 			return -EINVAL;
 		}
 		dep->point = syncobj_deps[i].point;

drivers/gpu/drm/drm_syncobj.c

@@ -861,7 +861,7 @@ static int drm_syncobj_transfer_to_timeline(struct drm_file *file_private,
 				     &fence);
 	if (ret)
 		goto err;
-	chain = kzalloc(sizeof(struct dma_fence_chain), GFP_KERNEL);
+	chain = dma_fence_chain_alloc();
 	if (!chain) {
 		ret = -ENOMEM;
 		goto err1;
@@ -1402,10 +1402,10 @@ drm_syncobj_timeline_signal_ioctl(struct drm_device *dev, void *data,
 		goto err_points;
 	}
 	for (i = 0; i < args->count_handles; i++) {
-		chains[i] = kzalloc(sizeof(struct dma_fence_chain), GFP_KERNEL);
+		chains[i] = dma_fence_chain_alloc();
 		if (!chains[i]) {
 			for (j = 0; j < i; j++)
-				kfree(chains[j]);
+				dma_fence_chain_free(chains[j]);
 			ret = -ENOMEM;
 			goto err_chains;
 		}

drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c

@@ -2983,7 +2983,7 @@ __free_fence_array(struct eb_fence *fences, unsigned int n)
 	while (n--) {
 		drm_syncobj_put(ptr_mask_bits(fences[n].syncobj, 2));
 		dma_fence_put(fences[n].dma_fence);
-		kfree(fences[n].chain_fence);
+		dma_fence_chain_free(fences[n].chain_fence);
 	}
 	kvfree(fences);
 }
@@ -3097,9 +3097,7 @@ add_timeline_fence_array(struct i915_execbuffer *eb,
 				return -EINVAL;
 			}
 
-			f->chain_fence =
-				kmalloc(sizeof(*f->chain_fence),
-					GFP_KERNEL);
+			f->chain_fence = dma_fence_chain_alloc();
 			if (!f->chain_fence) {
 				drm_syncobj_put(syncobj);
 				dma_fence_put(fence);

drivers/gpu/drm/msm/msm_gem_submit.c

@@ -586,9 +586,7 @@ static struct msm_submit_post_dep *msm_parse_post_deps(struct drm_device *dev,
 				break;
 			}
 
-			post_deps[i].chain =
-				kmalloc(sizeof(*post_deps[i].chain),
-				        GFP_KERNEL);
+			post_deps[i].chain = dma_fence_chain_alloc();
 			if (!post_deps[i].chain) {
 				ret = -ENOMEM;
 				break;
@@ -605,7 +603,7 @@ static struct msm_submit_post_dep *msm_parse_post_deps(struct drm_device *dev,
 
 	if (ret) {
 		for (j = 0; j <= i; ++j) {
-			kfree(post_deps[j].chain);
+			dma_fence_chain_free(post_deps[j].chain);
 			if (post_deps[j].syncobj)
 				drm_syncobj_put(post_deps[j].syncobj);
 		}

include/linux/dma-fence-chain.h

@@ -12,6 +12,7 @@
 
 #include <linux/dma-fence.h>
 #include <linux/irq_work.h>
+#include <linux/slab.h>
 
 /**
  * struct dma_fence_chain - fence to represent an node of a fence chain
@@ -66,6 +67,30 @@ to_dma_fence_chain(struct dma_fence *fence)
 	return container_of(fence, struct dma_fence_chain, base);
 }
 
+/**
+ * dma_fence_chain_alloc
+ *
+ * Returns a new struct dma_fence_chain object or NULL on failure.
+ */
+static inline struct dma_fence_chain *dma_fence_chain_alloc(void)
+{
+	return kmalloc(sizeof(struct dma_fence_chain), GFP_KERNEL);
+};
+
+/**
+ * dma_fence_chain_free
+ * @chain: chain node to free
+ *
+ * Frees up an allocated but not used struct dma_fence_chain object. This
+ * doesn't need an RCU grace period since the fence was never initialized nor
+ * published. After dma_fence_chain_init() has been called the fence must be
+ * released by calling dma_fence_put(), and not through this function.
+ */
+static inline void dma_fence_chain_free(struct dma_fence_chain *chain)
+{
+	kfree(chain);
+};
+
 /**
  * dma_fence_chain_for_each - iterate over all fences in chain
  * @iter: current fence