ceph: check that the new inode size is within limits in ceph_fallocate()

Currently the ceph client doesn't respect the rlimit in fallocate.  This
means that a user can allocate a file with size > RLIMIT_FSIZE.  This
patch adds the call to inode_newsize_ok() to verify filesystem limits and
ulimits.  This should make ceph successfully run xfstest generic/228.

Signed-off-by: Luis Henriques <lhenriques@suse.com>
Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index 3fdde0b283c9b86f7fa6aaa9585593699be29fca..29308a80d66ffa118b9aec996874d65a952c29aa 100644 (file)
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -1671,8 +1671,12 @@ static long ceph_fallocate(struct file *file, int mode,
        }
 
        size = i_size_read(inode);
-       if (!(mode & FALLOC_FL_KEEP_SIZE))
+       if (!(mode & FALLOC_FL_KEEP_SIZE)) {
                endoff = offset + length;
+               ret = inode_newsize_ok(inode, endoff);
+               if (ret)
+                       goto unlock;
+       }
 
        if (fi->fmode & CEPH_FILE_MODE_LAZY)
                want = CEPH_CAP_FILE_BUFFER | CEPH_CAP_FILE_LAZYIO;