btrfs: Moved repair code from inode.c to extent_io.c

The raid-retry code in inode.c can be generalized so that it works for
metadata as well. Thus, this patch moves it to extent_io.c and makes the
raid-retry code a raid-repair code.

Repair works that way: Whenever a read error occurs and we have more
mirrors to try, note the failed mirror, and retry another. If we find a
good one, check if we did note a failure earlier and if so, do not allow
the read to complete until after the bad sector was written with the good
data we just fetched. As we have the extent locked while reading, no one
can change the data in between.
Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>

fs/btrfs/extent_io.h

@@ -68,7 +68,7 @@ struct extent_io_ops {
 			      unsigned long bio_flags);
 	int (*readpage_io_hook)(struct page *page, u64 start, u64 end);
 	int (*readpage_io_failed_hook)(struct bio *bio, struct page *page,
-				       u64 start, u64 end,
+				       u64 start, u64 end, u64 failed_mirror,
 				       struct extent_state *state);
 	int (*writepage_io_failed_hook)(struct bio *bio, struct page *page,
 					u64 start, u64 end,
@@ -252,6 +252,8 @@ void free_extent_buffer(struct extent_buffer *eb);
 int read_extent_buffer_pages(struct extent_io_tree *tree,
 			     struct extent_buffer *eb, u64 start, int wait,
 			     get_extent_t *get_extent, int mirror_num);
+unsigned long num_extent_pages(u64 start, u64 len);
+struct page *extent_buffer_page(struct extent_buffer *eb, unsigned long i);
 
 static inline void extent_buffer_get(struct extent_buffer *eb)
 {
@@ -301,4 +303,10 @@ int extent_clear_unlock_delalloc(struct inode *inode,
 struct bio *
 btrfs_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs,
 		gfp_t gfp_flags);
+
+struct btrfs_mapping_tree;
+
+int repair_io_failure(struct btrfs_mapping_tree *map_tree, u64 start,
+			u64 length, u64 logical, struct page *page,
+			int mirror_num);
 #endif

fs/btrfs/inode.c

@@ -45,10 +45,10 @@
 #include "btrfs_inode.h"
 #include "ioctl.h"
 #include "print-tree.h"
-#include "volumes.h"
 #include "ordered-data.h"
 #include "xattr.h"
 #include "tree-log.h"
+#include "volumes.h"
 #include "compression.h"
 #include "locking.h"
 #include "free-space-cache.h"
@@ -1818,154 +1818,10 @@ static int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end,
 	return btrfs_finish_ordered_io(page->mapping->host, start, end);
 }
 
-/*
- * When IO fails, either with EIO or csum verification fails, we
- * try other mirrors that might have a good copy of the data.  This
- * io_failure_record is used to record state as we go through all the
- * mirrors.  If another mirror has good data, the page is set up to date
- * and things continue.  If a good mirror can't be found, the original
- * bio end_io callback is called to indicate things have failed.
- */
-struct io_failure_record {
-	struct page *page;
-	u64 start;
-	u64 len;
-	u64 logical;
-	unsigned long bio_flags;
-	int last_mirror;
-};
-
-static int btrfs_io_failed_hook(struct bio *failed_bio,
-			 struct page *page, u64 start, u64 end,
-			 struct extent_state *state)
-{
-	struct io_failure_record *failrec = NULL;
-	u64 private;
-	struct extent_map *em;
-	struct inode *inode = page->mapping->host;
-	struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;
-	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
-	struct bio *bio;
-	int num_copies;
-	int ret;
-	int rw;
-	u64 logical;
-
-	ret = get_state_private(failure_tree, start, &private);
-	if (ret) {
-		failrec = kmalloc(sizeof(*failrec), GFP_NOFS);
-		if (!failrec)
-			return -ENOMEM;
-		failrec->start = start;
-		failrec->len = end - start + 1;
-		failrec->last_mirror = 0;
-		failrec->bio_flags = 0;
-
-		read_lock(&em_tree->lock);
-		em = lookup_extent_mapping(em_tree, start, failrec->len);
-		if (em->start > start || em->start + em->len < start) {
-			free_extent_map(em);
-			em = NULL;
-		}
-		read_unlock(&em_tree->lock);
-
-		if (IS_ERR_OR_NULL(em)) {
-			kfree(failrec);
-			return -EIO;
-		}
-		logical = start - em->start;
-		logical = em->block_start + logical;
-		if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
-			logical = em->block_start;
-			failrec->bio_flags = EXTENT_BIO_COMPRESSED;
-			extent_set_compress_type(&failrec->bio_flags,
-						 em->compress_type);
-		}
-		failrec->logical = logical;
-		free_extent_map(em);
-		set_extent_bits(failure_tree, start, end, EXTENT_LOCKED |
-				EXTENT_DIRTY, GFP_NOFS);
-		set_state_private(failure_tree, start,
-				 (u64)(unsigned long)failrec);
-	} else {
-		failrec = (struct io_failure_record *)(unsigned long)private;
-	}
-	num_copies = btrfs_num_copies(
-			      &BTRFS_I(inode)->root->fs_info->mapping_tree,
-			      failrec->logical, failrec->len);
-	failrec->last_mirror++;
-	if (!state) {
-		spin_lock(&BTRFS_I(inode)->io_tree.lock);
-		state = find_first_extent_bit_state(&BTRFS_I(inode)->io_tree,
-						    failrec->start,
-						    EXTENT_LOCKED);
-		if (state && state->start != failrec->start)
-			state = NULL;
-		spin_unlock(&BTRFS_I(inode)->io_tree.lock);
-	}
-	if (!state || failrec->last_mirror > num_copies) {
-		set_state_private(failure_tree, failrec->start, 0);
-		clear_extent_bits(failure_tree, failrec->start,
-				  failrec->start + failrec->len - 1,
-				  EXTENT_LOCKED | EXTENT_DIRTY, GFP_NOFS);
-		kfree(failrec);
-		return -EIO;
-	}
-	bio = bio_alloc(GFP_NOFS, 1);
-	bio->bi_private = state;
-	bio->bi_end_io = failed_bio->bi_end_io;
-	bio->bi_sector = failrec->logical >> 9;
-	bio->bi_bdev = BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
-	bio->bi_size = 0;
-
-	bio_add_page(bio, page, failrec->len, start - page_offset(page));
-	if (failed_bio->bi_rw & REQ_WRITE)
-		rw = WRITE;
-	else
-		rw = READ;
-
-	ret = BTRFS_I(inode)->io_tree.ops->submit_bio_hook(inode, rw, bio,
-						      failrec->last_mirror,
-						      failrec->bio_flags, 0);
-	return ret;
-}
-
-/*
- * each time an IO finishes, we do a fast check in the IO failure tree
- * to see if we need to process or clean up an io_failure_record
- */
-static int btrfs_clean_io_failures(struct inode *inode, u64 start)
-{
-	u64 private;
-	u64 private_failure;
-	struct io_failure_record *failure;
-	int ret;
-
-	private = 0;
-	if (count_range_bits(&BTRFS_I(inode)->io_failure_tree, &private,
-			     (u64)-1, 1, EXTENT_DIRTY, 0)) {
-		ret = get_state_private(&BTRFS_I(inode)->io_failure_tree,
-					start, &private_failure);
-		if (ret == 0) {
-			failure = (struct io_failure_record *)(unsigned long)
-				   private_failure;
-			set_state_private(&BTRFS_I(inode)->io_failure_tree,
-					  failure->start, 0);
-			clear_extent_bits(&BTRFS_I(inode)->io_failure_tree,
-					  failure->start,
-					  failure->start + failure->len - 1,
-					  EXTENT_DIRTY | EXTENT_LOCKED,
-					  GFP_NOFS);
-			kfree(failure);
-		}
-	}
-	return 0;
-}
-
 /*
  * when reads are done, we need to check csums to verify the data is correct
- * if there's a match, we allow the bio to finish.  If not, we go through
- * the io_failure_record routines to find good copies
+ * if there's a match, we allow the bio to finish.  If not, the code in
+ * extent_io.c will try to find good copies for us.
  */
 static int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 			       struct extent_state *state)
@@ -2011,10 +1867,6 @@ static int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 
 	kunmap_atomic(kaddr, KM_USER0);
 good:
-	/* if the io failure tree for this inode is non-empty,
-	 * check to see if we've recovered from a failed IO
-	 */
-	btrfs_clean_io_failures(inode, start);
 	return 0;
 
 zeroit:
@@ -7420,7 +7272,6 @@ static struct extent_io_ops btrfs_extent_io_ops = {
 	.readpage_end_io_hook = btrfs_readpage_end_io_hook,
 	.writepage_end_io_hook = btrfs_writepage_end_io_hook,
 	.writepage_start_hook = btrfs_writepage_start_hook,
-	.readpage_io_failed_hook = btrfs_io_failed_hook,
 	.set_bit_hook = btrfs_set_bit_hook,
 	.clear_bit_hook = btrfs_clear_bit_hook,
 	.merge_extent_hook = btrfs_merge_extent_hook,