[PATCH 18/18] orangefs: implement xattr cache

[Martin Brandenburg]

This uses the same timeout as the getattr cache.  This substantially
increases performance when writing files with smaller buffer sizes.

When writing, the size is (often) changed, which causes a call to
notify_change which calls security_inode_need_killpriv which needs a
getxattr.  Caching it reduces traffic to the server.

Without:

$ time (dd if=/dev/zero of=/orangefs/foo bs=256 count=32768; sync)
32768+0 records in
32768+0 records out
8388608 bytes (8.4 MB, 8.0 MiB) copied, 11.0343 s, 760 kB/s

real    0m11.788s
user    0m0.013s
sys     0m0.703s

With:

$ time (dd if=/dev/zero of=/orangefs/foo bs=256 count=32768; sync)
32768+0 records in
32768+0 records out
8388608 bytes (8.4 MB, 8.0 MiB) copied, 0.0438278 s, 191 MB/s

real    0m2.181s
user    0m0.002s
sys     0m0.048s

Signed-off-by: Martin Brandenburg <martin@xxxxxxxxxxxxxxxxxxxxx>
---
 fs/orangefs/inode.c           |  1 +
 fs/orangefs/orangefs-kernel.h | 10 +++++++
 fs/orangefs/super.c           |  9 +++++++
 fs/orangefs/xattr.c           | 62 +++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 82 insertions(+)

diff --git a/fs/orangefs/inode.c b/fs/orangefs/inode.c
index 66c40ac7c4ac..31d655239f09 100644
--- a/fs/orangefs/inode.c
+++ b/fs/orangefs/inode.c
@@ -164,6 +164,7 @@ static int orangefs_set_inode(struct inode *inode, void *data)
 	struct orangefs_object_kref *ref = (struct orangefs_object_kref *) data;
 	ORANGEFS_I(inode)->refn.fs_id = ref->fs_id;
 	ORANGEFS_I(inode)->refn.khandle = ref->khandle;
+	hash_init(ORANGEFS_I(inode)->xattr_cache);
 	return 0;
 }
 
diff --git a/fs/orangefs/orangefs-kernel.h b/fs/orangefs/orangefs-kernel.h
index 4cabde0a5e9f..ff9a874440a5 100644
--- a/fs/orangefs/orangefs-kernel.h
+++ b/fs/orangefs/orangefs-kernel.h
@@ -211,6 +211,8 @@ struct orangefs_inode_s {
 
 	unsigned long getattr_time;
 	u32 getattr_mask;
+
+	DECLARE_HASHTABLE(xattr_cache, 4);
 };
 
 /* per superblock private orangefs info */
@@ -268,6 +270,14 @@ struct orangefs_stats {
 	unsigned long writes;
 };
 
+struct orangefs_cached_xattr {
+	struct hlist_node node;
+	char key[ORANGEFS_MAX_XATTR_NAMELEN];
+	char val[ORANGEFS_MAX_XATTR_VALUELEN];
+	ssize_t length;
+	unsigned long timeout;
+};
+
 extern struct orangefs_stats orangefs_stats;
 
 /*
diff --git a/fs/orangefs/super.c b/fs/orangefs/super.c
index 5c1a343ba026..dd1a768e84ff 100644
--- a/fs/orangefs/super.c
+++ b/fs/orangefs/super.c
@@ -128,6 +128,15 @@ static void orangefs_i_callback(struct rcu_head *head)
 {
 	struct inode *inode = container_of(head, struct inode, i_rcu);
 	struct orangefs_inode_s *orangefs_inode = ORANGEFS_I(inode);
+	struct orangefs_cached_xattr *cx;
+	struct hlist_node *tmp;
+	int i;
+
+	hash_for_each_safe(orangefs_inode->xattr_cache, i, tmp, cx, node) {
+		hlist_del(&cx->node);
+		kfree(cx);
+	}
+
 	kmem_cache_free(orangefs_inode_cache, orangefs_inode);
 }
 
diff --git a/fs/orangefs/xattr.c b/fs/orangefs/xattr.c
index 03bcb871544d..dac297870d40 100644
--- a/fs/orangefs/xattr.c
+++ b/fs/orangefs/xattr.c
@@ -50,6 +50,35 @@ static inline int convert_to_internal_xattr_flags(int setxattr_flags)
 	return internal_flag;
 }
 
+static unsigned int xattr_key(const char *key)
+{
+	unsigned int i = 0;
+	while (key)
+		i += *key++;
+	return i % 16;
+}
+
+static struct orangefs_cached_xattr *find_cached_xattr(struct inode *inode,
+    const char *key)
+{
+	struct orangefs_inode_s *orangefs_inode = ORANGEFS_I(inode);
+	struct orangefs_cached_xattr *cx;
+	struct hlist_head *h;
+	struct hlist_node *tmp;
+	h = &orangefs_inode->xattr_cache[xattr_key(key)];
+	if (hlist_empty(h))
+		return NULL;
+	hlist_for_each_entry_safe(cx, tmp, h, node) {
+		if (!time_before(jiffies, cx->timeout)) {
+			hlist_del(&cx->node);
+			kfree(cx);
+			continue;
+		}
+		if (!strcmp(cx->key, key))
+			return cx;
+	}
+	return NULL;
+}
 
 /*
  * Tries to get a specified key's attributes of a given
@@ -65,6 +94,7 @@ ssize_t orangefs_inode_getxattr(struct inode *inode, const char *name,
 {
 	struct orangefs_inode_s *orangefs_inode = ORANGEFS_I(inode);
 	struct orangefs_kernel_op_s *new_op = NULL;
+	struct orangefs_cached_xattr *cx;
 	ssize_t ret = -ENOMEM;
 	ssize_t length = 0;
 	int fsuid;
@@ -93,6 +123,19 @@ ssize_t orangefs_inode_getxattr(struct inode *inode, const char *name,
 
 	down_read(&orangefs_inode->xattr_sem);
 
+	cx = find_cached_xattr(inode, name);
+	if (cx) {
+		if (cx->length == -1) {
+			ret = -ENODATA;
+			goto out_unlock;
+		} else {
+			memcpy(buffer, cx->val, cx->length);
+			memset(buffer + cx->length, 0, size - cx->length);
+			ret = cx->length;
+			goto out_unlock;
+		}
+	}
+
 	new_op = op_alloc(ORANGEFS_VFS_OP_GETXATTR);
 	if (!new_op)
 		goto out_unlock;
@@ -117,6 +160,15 @@ ssize_t orangefs_inode_getxattr(struct inode *inode, const char *name,
 				     " does not exist!\n",
 				     get_khandle_from_ino(inode),
 				     (char *)new_op->upcall.req.getxattr.key);
+			cx = kmalloc(sizeof *cx, GFP_KERNEL);
+			if (cx) {
+				strcpy(cx->key, name);
+				cx->length = -1;
+				cx->timeout = jiffies +
+				    orangefs_getattr_timeout_msecs*HZ/1000;
+				hash_add(orangefs_inode->xattr_cache, &cx->node,
+				    xattr_key(cx->key));
+			}
 		}
 		goto out_release_op;
 	}
@@ -156,6 +208,16 @@ ssize_t orangefs_inode_getxattr(struct inode *inode, const char *name,
 
 	ret = length;
 
+	cx = kmalloc(sizeof *cx, GFP_KERNEL);
+	if (cx) {
+		strcpy(cx->key, name);
+		memcpy(cx->val, name, length);
+		cx->length = length;
+		cx->timeout = jiffies + HZ;
+		hash_add(orangefs_inode->xattr_cache, &cx->node,
+		    xattr_key(cx->key));
+	}
+
 out_release_op:
 	op_release(new_op);
 out_unlock:
-- 
2.15.1