Btrfs: adjust btrfs_discard_extent() return errors and trimmed bytes
authorLi Dongyang <lidongyang@novell.com>
Thu, 24 Mar 2011 10:24:27 +0000 (10:24 +0000)
committerroot <Chris Mason chris.mason@oracle.com>
Mon, 28 Mar 2011 09:37:46 +0000 (05:37 -0400)
Callers of btrfs_discard_extent() should check if we are mounted with -o discard,
as we want to make fitrim to work even the fs is not mounted with -o discard.
Also we should use REQ_DISCARD to map the free extent to get a full mapping,
last we only return errors if
1. the error is not a EOPNOTSUPP
2. no device supports discard

Signed-off-by: Li Dongyang <lidongyang@novell.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
fs/btrfs/ctree.h
fs/btrfs/disk-io.c
fs/btrfs/extent-tree.c

index 9e21176cdf575a07e675551809b95c7f55a49037..a18b7bc2b22ce1554aa295bab3dbccbc828bc013 100644 (file)
@@ -2229,7 +2229,7 @@ u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo);
 int btrfs_error_unpin_extent_range(struct btrfs_root *root,
                                   u64 start, u64 end);
 int btrfs_error_discard_extent(struct btrfs_root *root, u64 bytenr,
-                              u64 num_bytes);
+                              u64 num_bytes, u64 *actual_bytes);
 int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans,
                            struct btrfs_root *root, u64 type);
 
index 125639ddaffe9ed325d691d6ae493272c80f24ed..8ecc5419d8b6c3d64ccac45d5ae5c279a24457ee 100644 (file)
@@ -3054,7 +3054,10 @@ static int btrfs_destroy_pinned_extent(struct btrfs_root *root,
                        break;
 
                /* opt_discard */
-               ret = btrfs_error_discard_extent(root, start, end + 1 - start);
+               if (btrfs_test_opt(root, DISCARD))
+                       ret = btrfs_error_discard_extent(root, start,
+                                                        end + 1 - start,
+                                                        NULL);
 
                clear_extent_dirty(unpin, start, end, GFP_NOFS);
                btrfs_error_unpin_extent_range(root, start, end);
index 0671f5b77eb86573cce906d28923b90aaad45f03..e990d2d1ba4ade0fc40660d4d4114a792564988a 100644 (file)
@@ -1738,39 +1738,45 @@ static int remove_extent_backref(struct btrfs_trans_handle *trans,
        return ret;
 }
 
-static void btrfs_issue_discard(struct block_device *bdev,
+static int btrfs_issue_discard(struct block_device *bdev,
                                u64 start, u64 len)
 {
-       blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_NOFS, 0);
+       return blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_NOFS, 0);
 }
 
 static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
-                               u64 num_bytes)
+                               u64 num_bytes, u64 *actual_bytes)
 {
        int ret;
-       u64 map_length = num_bytes;
+       u64 discarded_bytes = 0;
        struct btrfs_multi_bio *multi = NULL;
 
-       if (!btrfs_test_opt(root, DISCARD))
-               return 0;
 
        /* Tell the block device(s) that the sectors can be discarded */
-       ret = btrfs_map_block(&root->fs_info->mapping_tree, READ,
-                             bytenr, &map_length, &multi, 0);
+       ret = btrfs_map_block(&root->fs_info->mapping_tree, REQ_DISCARD,
+                             bytenr, &num_bytes, &multi, 0);
        if (!ret) {
                struct btrfs_bio_stripe *stripe = multi->stripes;
                int i;
 
-               if (map_length > num_bytes)
-                       map_length = num_bytes;
 
                for (i = 0; i < multi->num_stripes; i++, stripe++) {
-                       btrfs_issue_discard(stripe->dev->bdev,
-                                           stripe->physical,
-                                           map_length);
+                       ret = btrfs_issue_discard(stripe->dev->bdev,
+                                                 stripe->physical,
+                                                 stripe->length);
+                       if (!ret)
+                               discarded_bytes += stripe->length;
+                       else if (ret != -EOPNOTSUPP)
+                               break;
                }
                kfree(multi);
        }
+       if (discarded_bytes && ret == -EOPNOTSUPP)
+               ret = 0;
+
+       if (actual_bytes)
+               *actual_bytes = discarded_bytes;
+
 
        return ret;
 }
@@ -4371,7 +4377,9 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
                if (ret)
                        break;
 
-               ret = btrfs_discard_extent(root, start, end + 1 - start);
+               if (btrfs_test_opt(root, DISCARD))
+                       ret = btrfs_discard_extent(root, start,
+                                                  end + 1 - start, NULL);
 
                clear_extent_dirty(unpin, start, end, GFP_NOFS);
                unpin_extent_range(root, start, end);
@@ -5427,7 +5435,8 @@ int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len)
                return -ENOSPC;
        }
 
-       ret = btrfs_discard_extent(root, start, len);
+       if (btrfs_test_opt(root, DISCARD))
+               ret = btrfs_discard_extent(root, start, len, NULL);
 
        btrfs_add_free_space(cache, start, len);
        btrfs_update_reserved_bytes(cache, len, 0, 1);
@@ -8765,7 +8774,7 @@ int btrfs_error_unpin_extent_range(struct btrfs_root *root, u64 start, u64 end)
 }
 
 int btrfs_error_discard_extent(struct btrfs_root *root, u64 bytenr,
-                              u64 num_bytes)
+                              u64 num_bytes, u64 *actual_bytes)
 {
-       return btrfs_discard_extent(root, bytenr, num_bytes);
+       return btrfs_discard_extent(root, bytenr, num_bytes, actual_bytes);
 }