Fix nfs4.2 return -EINVAL when do dedupe operation
authorXiaoli Feng <fengxiaoli0714@gmail.com>
Sat, 16 Mar 2019 01:43:30 +0000 (09:43 +0800)
committerAnna Schumaker <Anna.Schumaker@Netapp.com>
Thu, 25 Apr 2019 18:18:15 +0000 (14:18 -0400)
dedupe_file_range operations is combiled into remap_file_range.
But in nfs42_remap_file_range, it's skiped for dedupe operations.
Before this patch:
  # dd if=/dev/zero of=nfs/file bs=1M count=1
  # xfs_io -c "dedupe nfs/file 4k 64k 4k" nfs/file
  XFS_IOC_FILE_EXTENT_SAME: Invalid argument
After this patch:
  # dd if=/dev/zero of=nfs/file bs=1M count=1
  # xfs_io -c "dedupe nfs/file 4k 64k 4k" nfs/file
  deduped 4096/4096 bytes at offset 65536
  4 KiB, 1 ops; 0.0046 sec (865.988 KiB/sec and 216.4971 ops/sec)

Signed-off-by: Xiaoli Feng <fengxiaoli0714@gmail.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
fs/nfs/nfs4file.c

index d2846f164997cd5127a711b1c6b7274ae4d6812f..cf42a8b939e3ed0c4e866af3905b1d4ee8ff1dff 100644 (file)
@@ -187,7 +187,7 @@ static loff_t nfs42_remap_file_range(struct file *src_file, loff_t src_off,
        bool same_inode = false;
        int ret;
 
-       if (remap_flags & ~REMAP_FILE_ADVISORY)
+       if (remap_flags & ~(REMAP_FILE_DEDUP | REMAP_FILE_ADVISORY))
                return -EINVAL;
 
        /* check alignment w.r.t. clone_blksize */