1 files changed, 30 insertions, 10 deletions

diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index c953297aa89a..b21cb72835cc 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -111,6 +111,24 @@ struct btrfs_bio_ctrl {
 	 */
 	unsigned long submit_bitmap;
 	struct readahead_control *ractl;
+
+	/*
+	 * The start offset of the last used extent map by a read operation.
+	 *
+	 * This is for proper compressed read merge.
+	 * U64_MAX means we are starting the read and have made no progress yet.
+	 *
+	 * The current btrfs_bio_is_contig() only uses disk_bytenr as
+	 * the condition to check if the read can be merged with previous
+	 * bio, which is not correct. E.g. two file extents pointing to the
+	 * same extent but with different offset.
+	 *
+	 * So here we need to do extra checks to only merge reads that are
+	 * covered by the same extent map.
+	 * Just extent_map::start will be enough, as they are unique
+	 * inside the same inode.
+	 */
+	u64 last_em_start;
 };
 
 static void submit_one_bio(struct btrfs_bio_ctrl *bio_ctrl)
@@ -909,7 +927,7 @@ static void btrfs_readahead_expand(struct readahead_control *ractl,
  * return 0 on success, otherwise return error
  */
 static int btrfs_do_readpage(struct folio *folio, struct extent_map **em_cached,
-		      struct btrfs_bio_ctrl *bio_ctrl, u64 *prev_em_start)
+			     struct btrfs_bio_ctrl *bio_ctrl)
 {
 	struct inode *inode = folio->mapping->host;
 	struct btrfs_fs_info *fs_info = inode_to_fs_info(inode);
@@ -1019,12 +1037,11 @@ static int btrfs_do_readpage(struct folio *folio, struct extent_map **em_cached,
 		 * non-optimal behavior (submitting 2 bios for the same extent).
 		 */
 		if (compress_type != BTRFS_COMPRESS_NONE &&
-		    prev_em_start && *prev_em_start != (u64)-1 &&
-		    *prev_em_start != em->start)
+		    bio_ctrl->last_em_start != U64_MAX &&
+		    bio_ctrl->last_em_start != em->start)
 			force_bio_submit = true;
 
-		if (prev_em_start)
-			*prev_em_start = em->start;
+		bio_ctrl->last_em_start = em->start;
 
 		btrfs_free_extent_map(em);
 		em = NULL;
@@ -1238,12 +1255,15 @@ int btrfs_read_folio(struct file *file, struct folio *folio)
 	const u64 start = folio_pos(folio);
 	const u64 end = start + folio_size(folio) - 1;
 	struct extent_state *cached_state = NULL;
-	struct btrfs_bio_ctrl bio_ctrl = { .opf = REQ_OP_READ };
+	struct btrfs_bio_ctrl bio_ctrl = {
+		.opf = REQ_OP_READ,
+		.last_em_start = U64_MAX,
+	};
 	struct extent_map *em_cached = NULL;
 	int ret;
 
 	lock_extents_for_read(inode, start, end, &cached_state);
-	ret = btrfs_do_readpage(folio, &em_cached, &bio_ctrl, NULL);
+	ret = btrfs_do_readpage(folio, &em_cached, &bio_ctrl);
 	btrfs_unlock_extent(&inode->io_tree, start, end, &cached_state);
 
 	btrfs_free_extent_map(em_cached);
@@ -2583,7 +2603,8 @@ void btrfs_readahead(struct readahead_control *rac)
 {
 	struct btrfs_bio_ctrl bio_ctrl = {
 		.opf = REQ_OP_READ | REQ_RAHEAD,
-		.ractl = rac
+		.ractl = rac,
+		.last_em_start = U64_MAX,
 	};
 	struct folio *folio;
 	struct btrfs_inode *inode = BTRFS_I(rac->mapping->host);
@@ -2591,12 +2612,11 @@ void btrfs_readahead(struct readahead_control *rac)
 	const u64 end = start + readahead_length(rac) - 1;
 	struct extent_state *cached_state = NULL;
 	struct extent_map *em_cached = NULL;
-	u64 prev_em_start = (u64)-1;
 
 	lock_extents_for_read(inode, start, end, &cached_state);
 
 	while ((folio = readahead_folio(rac)) != NULL)
-		btrfs_do_readpage(folio, &em_cached, &bio_ctrl, &prev_em_start);
+		btrfs_do_readpage(folio, &em_cached, &bio_ctrl);
 
 	btrfs_unlock_extent(&inode->io_tree, start, end, &cached_state);