1 files changed, 26 insertions, 155 deletions

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 18d05323ca53..77b32415d9f2 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -50,6 +50,7 @@
 #include "sysfs.h"
 #include "qgroup.h"
 #include "compression.h"
+#include "tree-checker.h"
 
 #ifdef CONFIG_X86
 #include <asm/cpufeature.h>
@@ -452,9 +453,9 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root,
 	int mirror_num = 0;
 	int failed_mirror = 0;
 
-	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
 	io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
 	while (1) {
+		clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
 		ret = read_extent_buffer_pages(io_tree, eb, WAIT_COMPLETE,
 					       btree_get_extent, mirror_num);
 		if (!ret) {
@@ -465,14 +466,6 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root,
 				ret = -EIO;
 		}
 
-		/*
-		 * This buffer's crc is fine, but its contents are corrupted, so
-		 * there is no reason to read the other copies, they won't be
-		 * any less wrong.
-		 */
-		if (test_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags))
-			break;
-
 		num_copies = btrfs_num_copies(root->fs_info,
 					      eb->start, eb->len);
 		if (num_copies == 1)
@@ -546,145 +539,6 @@ static int check_tree_block_fsid(struct btrfs_fs_info *fs_info,
 	return ret;
 }
 
-#define CORRUPT(reason, eb, root, slot)				\
-	btrfs_crit(root->fs_info, "corrupt %s, %s: block=%llu,"	\
-		   " root=%llu, slot=%d",			\
-		   btrfs_header_level(eb) == 0 ? "leaf" : "node",\
-		   reason, btrfs_header_bytenr(eb), root->objectid, slot)
-
-static noinline int check_leaf(struct btrfs_root *root,
-			       struct extent_buffer *leaf)
-{
-	struct btrfs_key key;
-	struct btrfs_key leaf_key;
-	u32 nritems = btrfs_header_nritems(leaf);
-	int slot;
-
-	/*
-	 * Extent buffers from a relocation tree have a owner field that
-	 * corresponds to the subvolume tree they are based on. So just from an
-	 * extent buffer alone we can not find out what is the id of the
-	 * corresponding subvolume tree, so we can not figure out if the extent
-	 * buffer corresponds to the root of the relocation tree or not. So skip
-	 * this check for relocation trees.
-	 */
-	if (nritems == 0 && !btrfs_header_flag(leaf, BTRFS_HEADER_FLAG_RELOC)) {
-		struct btrfs_root *check_root;
-
-		key.objectid = btrfs_header_owner(leaf);
-		key.type = BTRFS_ROOT_ITEM_KEY;
-		key.offset = (u64)-1;
-
-		check_root = btrfs_get_fs_root(root->fs_info, &key, false);
-		/*
-		 * The only reason we also check NULL here is that during
-		 * open_ctree() some roots has not yet been set up.
-		 */
-		if (!IS_ERR_OR_NULL(check_root)) {
-			struct extent_buffer *eb;
-
-			eb = btrfs_root_node(check_root);
-			/* if leaf is the root, then it's fine */
-			if (leaf != eb) {
-				CORRUPT("non-root leaf's nritems is 0",
-					leaf, check_root, 0);
-				free_extent_buffer(eb);
-				return -EIO;
-			}
-			free_extent_buffer(eb);
-		}
-		return 0;
-	}
-
-	if (nritems == 0)
-		return 0;
-
-	/* Check the 0 item */
-	if (btrfs_item_offset_nr(leaf, 0) + btrfs_item_size_nr(leaf, 0) !=
-	    BTRFS_LEAF_DATA_SIZE(root)) {
-		CORRUPT("invalid item offset size pair", leaf, root, 0);
-		return -EIO;
-	}
-
-	/*
-	 * Check to make sure each items keys are in the correct order and their
-	 * offsets make sense.  We only have to loop through nritems-1 because
-	 * we check the current slot against the next slot, which verifies the
-	 * next slot's offset+size makes sense and that the current's slot
-	 * offset is correct.
-	 */
-	for (slot = 0; slot < nritems - 1; slot++) {
-		btrfs_item_key_to_cpu(leaf, &leaf_key, slot);
-		btrfs_item_key_to_cpu(leaf, &key, slot + 1);
-
-		/* Make sure the keys are in the right order */
-		if (btrfs_comp_cpu_keys(&leaf_key, &key) >= 0) {
-			CORRUPT("bad key order", leaf, root, slot);
-			return -EIO;
-		}
-
-		/*
-		 * Make sure the offset and ends are right, remember that the
-		 * item data starts at the end of the leaf and grows towards the
-		 * front.
-		 */
-		if (btrfs_item_offset_nr(leaf, slot) !=
-			btrfs_item_end_nr(leaf, slot + 1)) {
-			CORRUPT("slot offset bad", leaf, root, slot);
-			return -EIO;
-		}
-
-		/*
-		 * Check to make sure that we don't point outside of the leaf,
-		 * just in case all the items are consistent to each other, but
-		 * all point outside of the leaf.
-		 */
-		if (btrfs_item_end_nr(leaf, slot) >
-		    BTRFS_LEAF_DATA_SIZE(root)) {
-			CORRUPT("slot end outside of leaf", leaf, root, slot);
-			return -EIO;
-		}
-	}
-
-	return 0;
-}
-
-static int check_node(struct btrfs_root *root, struct extent_buffer *node)
-{
-	unsigned long nr = btrfs_header_nritems(node);
-	struct btrfs_key key, next_key;
-	int slot;
-	u64 bytenr;
-	int ret = 0;
-
-	if (nr == 0 || nr > BTRFS_NODEPTRS_PER_BLOCK(root)) {
-		btrfs_crit(root->fs_info,
-			   "corrupt node: block %llu root %llu nritems %lu",
-			   node->start, root->objectid, nr);
-		return -EIO;
-	}
-
-	for (slot = 0; slot < nr - 1; slot++) {
-		bytenr = btrfs_node_blockptr(node, slot);
-		btrfs_node_key_to_cpu(node, &key, slot);
-		btrfs_node_key_to_cpu(node, &next_key, slot + 1);
-
-		if (!bytenr) {
-			CORRUPT("invalid item slot", node, root, slot);
-			ret = -EIO;
-			goto out;
-		}
-
-		if (btrfs_comp_cpu_keys(&key, &next_key) >= 0) {
-			CORRUPT("bad key order", node, root, slot);
-			ret = -EIO;
-			goto out;
-		}
-	}
-out:
-	return ret;
-}
-
 static int btree_readpage_end_io_hook(struct btrfs_io_bio *io_bio,
 				      u64 phy_offset, struct page *page,
 				      u64 start, u64 end, int mirror)
@@ -750,12 +604,12 @@ static int btree_readpage_end_io_hook(struct btrfs_io_bio *io_bio,
 	 * that we don't try and read the other copies of this block, just
 	 * return -EIO.
 	 */
-	if (found_level == 0 && check_leaf(root, eb)) {
+	if (found_level == 0 && btrfs_check_leaf_full(root, eb)) {
 		set_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
 		ret = -EIO;
 	}
 
-	if (found_level > 0 && check_node(root, eb))
+	if (found_level > 0 && btrfs_check_node(root, eb))
 		ret = -EIO;
 
 	if (!ret)
@@ -4086,7 +3940,13 @@ void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
 				     buf->len,
 				     root->fs_info->dirty_metadata_batch);
 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
-	if (btrfs_header_level(buf) == 0 && check_leaf(root, buf)) {
+	/*
+	 * Since btrfs_mark_buffer_dirty() can be called with item pointer set
+	 * but item data not updated.
+	 * So here we should only check item pointers, not item data.
+	 */
+	if (btrfs_header_level(buf) == 0 &&
+	    btrfs_check_leaf_relaxed(root, buf)) {
 		btrfs_print_leaf(root, buf);
 		ASSERT(0);
 	}
@@ -4491,6 +4351,7 @@ static int btrfs_destroy_marked_extents(struct btrfs_root *root,
 static int btrfs_destroy_pinned_extent(struct btrfs_root *root,
 				       struct extent_io_tree *pinned_extents)
 {
+	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct extent_io_tree *unpin;
 	u64 start;
 	u64 end;
@@ -4500,21 +4361,31 @@ static int btrfs_destroy_pinned_extent(struct btrfs_root *root,
 	unpin = pinned_extents;
 again:
 	while (1) {
+		/*
+		 * The btrfs_finish_extent_commit() may get the same range as
+		 * ours between find_first_extent_bit and clear_extent_dirty.
+		 * Hence, hold the unused_bg_unpin_mutex to avoid double unpin
+		 * the same extent range.
+		 */
+		mutex_lock(&fs_info->unused_bg_unpin_mutex);
 		ret = find_first_extent_bit(unpin, 0, &start, &end,
 					    EXTENT_DIRTY, NULL);
-		if (ret)
+		if (ret) {
+			mutex_unlock(&fs_info->unused_bg_unpin_mutex);
 			break;
+		}
 
 		clear_extent_dirty(unpin, start, end);
 		btrfs_error_unpin_extent_range(root, start, end);
+		mutex_unlock(&fs_info->unused_bg_unpin_mutex);
 		cond_resched();
 	}
 
 	if (loop) {
-		if (unpin == &root->fs_info->freed_extents[0])
-			unpin = &root->fs_info->freed_extents[1];
+		if (unpin == &fs_info->freed_extents[0])
+			unpin = &fs_info->freed_extents[1];
 		else
-			unpin = &root->fs_info->freed_extents[0];
+			unpin = &fs_info->freed_extents[0];
 		loop = false;
 		goto again;
 	}