ceph: allow object copies across different filesystems in the same cluster
authorLuis Henriques <lhenriques@suse.com>
Mon, 9 Sep 2019 15:48:54 +0000 (16:48 +0100)
committerIlya Dryomov <idryomov@gmail.com>
Mon, 16 Sep 2019 10:06:25 +0000 (12:06 +0200)
OSDs are able to perform object copies across different pools.  Thus,
there's no need to prevent copy_file_range from doing remote copies if the
source and destination superblocks are different.  Only return -EXDEV if
they have different fsid (the cluster ID).

Signed-off-by: Luis Henriques <lhenriques@suse.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
fs/ceph/file.c

index ff17a81bf2e2a03d899c55b15b5a6760eee750c4..d277f71abe0b8e653ef0e439782c2ee930388bbf 100644 (file)
@@ -1922,6 +1922,7 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
        struct ceph_inode_info *src_ci = ceph_inode(src_inode);
        struct ceph_inode_info *dst_ci = ceph_inode(dst_inode);
        struct ceph_cap_flush *prealloc_cf;
+       struct ceph_fs_client *src_fsc = ceph_inode_to_client(src_inode);
        struct ceph_object_locator src_oloc, dst_oloc;
        struct ceph_object_id src_oid, dst_oid;
        loff_t endoff = 0, size;
@@ -1931,8 +1932,16 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
        int src_got = 0, dst_got = 0, err, dirty;
        bool do_final_copy = false;
 
-       if (src_inode->i_sb != dst_inode->i_sb)
-               return -EXDEV;
+       if (src_inode->i_sb != dst_inode->i_sb) {
+               struct ceph_fs_client *dst_fsc = ceph_inode_to_client(dst_inode);
+
+               if (ceph_fsid_compare(&src_fsc->client->fsid,
+                                     &dst_fsc->client->fsid)) {
+                       dout("Copying files across clusters: src: %pU dst: %pU\n",
+                            &src_fsc->client->fsid, &dst_fsc->client->fsid);
+                       return -EXDEV;
+               }
+       }
        if (ceph_snap(dst_inode) != CEPH_NOSNAP)
                return -EROFS;
 
@@ -1944,7 +1953,7 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
         * efficient).
         */
 
-       if (ceph_test_mount_opt(ceph_inode_to_client(src_inode), NOCOPYFROM))
+       if (ceph_test_mount_opt(src_fsc, NOCOPYFROM))
                return -EOPNOTSUPP;
 
        if ((src_ci->i_layout.stripe_unit != dst_ci->i_layout.stripe_unit) ||
@@ -2059,7 +2068,7 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
                                dst_ci->i_vino.ino, dst_objnum);
                /* Do an object remote copy */
                err = ceph_osdc_copy_from(
-                       &ceph_inode_to_client(src_inode)->client->osdc,
+                       &src_fsc->client->osdc,
                        src_ci->i_vino.snap, 0,
                        &src_oid, &src_oloc,
                        CEPH_OSD_OP_FLAG_FADVISE_SEQUENTIAL |