btrfs: fallback if compressed IO fails for ENOSPC

In commit b4ccace878f4 ("btrfs: refactor submit_compressed_extents()"), if
an async extent compressed but failed to find enough space, we changed
from falling back to an uncompressed write to just failing the write
altogether. The principle was that if there's not enough space to write
the compressed version of the data, there can't possibly be enough space
to write the larger, uncompressed version of the data.

However, this isn't necessarily true: due to fragmentation, there could
be enough discontiguous free blocks to write the uncompressed version,
but not enough contiguous free blocks to write the smaller but
unsplittable compressed version.

This has occurred to an internal workload which relied on write()'s
return value indicating there was space. While rare, it has happened a
few times.

Thus, in order to prevent early ENOSPC, re-add a fallback to
uncompressed writing.

Fixes: b4ccace878f4 ("btrfs: refactor submit_compressed_extents()")
CC: [email protected] # 6.1+
Reviewed-by: Qu Wenruo <[email protected]>
Co-developed-by: Neal Gompa <[email protected]>
Signed-off-by: Neal Gompa <[email protected]>
Signed-off-by: Sweet Tea Dorminy <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Signed-off-by: David Sterba <[email protected]>

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index c65fe5de40220d3b51003bb73b3e6414eaefba08..7fed887e700c4e8e07b6ff7932434a384790b8da 100644 (file)
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -1145,13 +1145,13 @@ static void submit_one_async_extent(struct async_chunk *async_chunk,
                                   0, *alloc_hint, &ins, 1, 1);
        if (ret) {
                /*
-                * Here we used to try again by going back to non-compressed
-                * path for ENOSPC.  But we can't reserve space even for
-                * compressed size, how could it work for uncompressed size
-                * which requires larger size?  So here we directly go error
-                * path.
+                * We can't reserve contiguous space for the compressed size.
+                * Unlikely, but it's possible that we could have enough
+                * non-contiguous space for the uncompressed size instead.  So
+                * fall back to uncompressed.
                 */
-               goto out_free;
+               submit_uncompressed_range(inode, async_extent, locked_page);
+               goto done;
        }
 
        /* Here we're doing allocation and writeback of the compressed pages */
@@ -1203,7 +1203,6 @@ done:
 out_free_reserve:
        btrfs_dec_block_group_reservations(fs_info, ins.objectid);
        btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 1);
-out_free:
        mapping_set_error(inode->vfs_inode.i_mapping, -EIO);
        extent_clear_unlock_delalloc(inode, start, end,
                                     NULL, EXTENT_LOCKED | EXTENT_DELALLOC |