fix ufs write vs readpage race when writing into a hole

Followup to the UFS series - with the way we clear the new blocks (via
buffer cache, possibly on more than a page worth of file) we really
should not insert a reference to new block into inode block tree until
after we'd cleared it.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/fs/ufs/balloc.c b/fs/ufs/balloc.c
index fb8b54e..dc5fae6 100644 (file)
--- a/fs/ufs/balloc.c
+++ b/fs/ufs/balloc.c
@@ -417,14 +417,14 @@ u64 ufs_new_fragments(struct inode *inode, void *p, u64 fragment,
        if (oldcount == 0) {
                result = ufs_alloc_fragments (inode, cgno, goal, count, err);
                if (result) {
+                       ufs_clear_frags(inode, result + oldcount,
+                                       newcount - oldcount, locked_page != NULL);
                        write_seqlock(&UFS_I(inode)->meta_lock);
                        ufs_cpu_to_data_ptr(sb, p, result);
                        write_sequnlock(&UFS_I(inode)->meta_lock);
                        *err = 0;
                        UFS_I(inode)->i_lastfrag =
                                max(UFS_I(inode)->i_lastfrag, fragment + count);
-                       ufs_clear_frags(inode, result + oldcount,
-                                       newcount - oldcount, locked_page != NULL);
                }
                mutex_unlock(&UFS_SB(sb)->s_lock);
                UFSD("EXIT, result %llu\n", (unsigned long long)result);