Btrfs: proper -ENOSPC handling

At the start of a transaction we do a btrfs_reserve_metadata_space() and
specify how many items we plan on modifying.  Then once we've done our
modifications and such, just call btrfs_unreserve_metadata_space() for
the same number of items we reserved.

For keeping track of metadata needed for data I've had to add an extent_io op
for when we merge extents.  This lets us track space properly when we are doing
sequential writes, so we don't end up reserving way more metadata space than
what we need.

The only place where the metadata space accounting is not done is in the
relocation code.  This is because Yan is going to be reworking that code in the
near future, so running btrfs-vol -b could still possibly result in a ENOSPC
related panic.  This patch also turns off the metadata_ratio stuff in order to
allow users to more efficiently use their disk space.

This patch makes it so we track how much metadata we need for an inode's
delayed allocation extents by tracking how many extents are currently
waiting for allocation.  It introduces two new callbacks for the
extent_io tree's, merge_extent_hook and split_extent_hook.  These help
us keep track of when we merge delalloc extents together and split them
up.  Reservations are handled prior to any actually dirty'ing occurs,
and then we unreserve after we dirty.

btrfs_unreserve_metadata_for_delalloc() will make the appropriate
unreservations as needed based on the number of reservations we
currently have and the number of extents we currently have.  Doing the
reservation outside of doing any of the actual dirty'ing lets us do
things like filemap_flush() the inode to try and force delalloc to
happen, or as a last resort actually start allocation on all delalloc
inodes in the fs.  This has survived dbench, fs_mark and an fsx torture
test.

Signed-off-by: Josef Bacik <jbacik@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

1 files changed, 69 insertions, 23 deletions

diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 0cb88f8146ea..de1793ba004a 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -280,6 +280,14 @@ static struct extent_buffer *buffer_search(struct extent_io_tree *tree,
 	return NULL;
 }
 
+static void merge_cb(struct extent_io_tree *tree, struct extent_state *new,
+		     struct extent_state *other)
+{
+	if (tree->ops && tree->ops->merge_extent_hook)
+		tree->ops->merge_extent_hook(tree->mapping->host, new,
+					     other);
+}
+
 /*
  * utility function to look for merge candidates inside a given range.
  * Any extents with matching state are merged together into a single
@@ -303,6 +311,7 @@ static int merge_state(struct extent_io_tree *tree,
 		other = rb_entry(other_node, struct extent_state, rb_node);
 		if (other->end == state->start - 1 &&
 		    other->state == state->state) {
+			merge_cb(tree, state, other);
 			state->start = other->start;
 			other->tree = NULL;
 			rb_erase(&other->rb_node, &tree->state);
@@ -314,33 +323,37 @@ static int merge_state(struct extent_io_tree *tree,
 		other = rb_entry(other_node, struct extent_state, rb_node);
 		if (other->start == state->end + 1 &&
 		    other->state == state->state) {
+			merge_cb(tree, state, other);
 			other->start = state->start;
 			state->tree = NULL;
 			rb_erase(&state->rb_node, &tree->state);
 			free_extent_state(state);
+			state = NULL;
 		}
 	}
+
 	return 0;
 }
 
-static void set_state_cb(struct extent_io_tree *tree,
+static int set_state_cb(struct extent_io_tree *tree,
 			 struct extent_state *state,
 			 unsigned long bits)
 {
 	if (tree->ops && tree->ops->set_bit_hook) {
-		tree->ops->set_bit_hook(tree->mapping->host, state->start,
-					state->end, state->state, bits);
+		return tree->ops->set_bit_hook(tree->mapping->host,
+					       state->start, state->end,
+					       state->state, bits);
 	}
+
+	return 0;
 }
 
 static void clear_state_cb(struct extent_io_tree *tree,
 			   struct extent_state *state,
 			   unsigned long bits)
 {
-	if (tree->ops && tree->ops->clear_bit_hook) {
-		tree->ops->clear_bit_hook(tree->mapping->host, state->start,
-					  state->end, state->state, bits);
-	}
+	if (tree->ops && tree->ops->clear_bit_hook)
+		tree->ops->clear_bit_hook(tree->mapping->host, state, bits);
 }
 
 /*
@@ -358,6 +371,7 @@ static int insert_state(struct extent_io_tree *tree,
 			int bits)
 {
 	struct rb_node *node;
+	int ret;
 
 	if (end < start) {
 		printk(KERN_ERR "btrfs end < start %llu %llu\n",
@@ -365,11 +379,14 @@ static int insert_state(struct extent_io_tree *tree,
 		       (unsigned long long)start);
 		WARN_ON(1);
 	}
-	if (bits & EXTENT_DIRTY)
-		tree->dirty_bytes += end - start + 1;
 	state->start = start;
 	state->end = end;
-	set_state_cb(tree, state, bits);
+	ret = set_state_cb(tree, state, bits);
+	if (ret)
+		return ret;
+
+	if (bits & EXTENT_DIRTY)
+		tree->dirty_bytes += end - start + 1;
 	state->state |= bits;
 	node = tree_insert(&tree->state, end, &state->rb_node);
 	if (node) {
@@ -387,6 +404,15 @@ static int insert_state(struct extent_io_tree *tree,
 	return 0;
 }
 
+static int split_cb(struct extent_io_tree *tree, struct extent_state *orig,
+		     u64 split)
+{
+	if (tree->ops && tree->ops->split_extent_hook)
+		return tree->ops->split_extent_hook(tree->mapping->host,
+						    orig, split);
+	return 0;
+}
+
 /*
  * split a given extent state struct in two, inserting the preallocated
  * struct 'prealloc' as the newly created second half.  'split' indicates an
@@ -405,6 +431,9 @@ static int split_state(struct extent_io_tree *tree, struct extent_state *orig,
 		       struct extent_state *prealloc, u64 split)
 {
 	struct rb_node *node;
+
+	split_cb(tree, orig, split);
+
 	prealloc->start = orig->start;
 	prealloc->end = split - 1;
 	prealloc->state = orig->state;
@@ -542,8 +571,8 @@ hit_next:
 		if (err)
 			goto out;
 		if (state->end <= end) {
-			set |= clear_state_bit(tree, state, bits,
-					wake, delete);
+			set |= clear_state_bit(tree, state, bits, wake,
+					       delete);
 			if (last_end == (u64)-1)
 				goto out;
 			start = last_end + 1;
@@ -561,12 +590,11 @@ hit_next:
 			prealloc = alloc_extent_state(GFP_ATOMIC);
 		err = split_state(tree, state, prealloc, end + 1);
 		BUG_ON(err == -EEXIST);
-
 		if (wake)
 			wake_up(&state->wq);
 
-		set |= clear_state_bit(tree, prealloc, bits,
-				       wake, delete);
+		set |= clear_state_bit(tree, prealloc, bits, wake, delete);
+
 		prealloc = NULL;
 		goto out;
 	}
@@ -667,16 +695,23 @@ out:
 	return 0;
 }
 
-static void set_state_bits(struct extent_io_tree *tree,
+static int set_state_bits(struct extent_io_tree *tree,
 			   struct extent_state *state,
 			   int bits)
 {
+	int ret;
+
+	ret = set_state_cb(tree, state, bits);
+	if (ret)
+		return ret;
+
 	if ((bits & EXTENT_DIRTY) && !(state->state & EXTENT_DIRTY)) {
 		u64 range = state->end - state->start + 1;
 		tree->dirty_bytes += range;
 	}
-	set_state_cb(tree, state, bits);
 	state->state |= bits;
+
+	return 0;
 }
 
 static void cache_state(struct extent_state *state,
@@ -758,7 +793,10 @@ hit_next:
 			goto out;
 		}
 
-		set_state_bits(tree, state, bits);
+		err = set_state_bits(tree, state, bits);
+		if (err)
+			goto out;
+
 		cache_state(state, cached_state);
 		merge_state(tree, state);
 		if (last_end == (u64)-1)
@@ -805,7 +843,9 @@ hit_next:
 		if (err)
 			goto out;
 		if (state->end <= end) {
-			set_state_bits(tree, state, bits);
+			err = set_state_bits(tree, state, bits);
+			if (err)
+				goto out;
 			cache_state(state, cached_state);
 			merge_state(tree, state);
 			if (last_end == (u64)-1)
@@ -829,11 +869,13 @@ hit_next:
 			this_end = last_start - 1;
 		err = insert_state(tree, prealloc, start, this_end,
 				   bits);
-		cache_state(prealloc, cached_state);
-		prealloc = NULL;
 		BUG_ON(err == -EEXIST);
-		if (err)
+		if (err) {
+			prealloc = NULL;
 			goto out;
+		}
+		cache_state(prealloc, cached_state);
+		prealloc = NULL;
 		start = this_end + 1;
 		goto search_again;
 	}
@@ -852,7 +894,11 @@ hit_next:
 		err = split_state(tree, state, prealloc, end + 1);
 		BUG_ON(err == -EEXIST);
 
-		set_state_bits(tree, prealloc, bits);
+		err = set_state_bits(tree, prealloc, bits);
+		if (err) {
+			prealloc = NULL;
+			goto out;
+		}
 		cache_state(prealloc, cached_state);
 		merge_state(tree, prealloc);
 		prealloc = NULL;