]> pilppa.com Git - linux-2.6-omap-h63xx.git/commitdiff
vfs: Don't call attach_nobh_buffers() with an empty list
authorDave Kleikamp <shaggy@linux.vnet.ibm.com>
Fri, 6 Feb 2009 20:59:26 +0000 (14:59 -0600)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 6 Feb 2009 21:34:22 +0000 (13:34 -0800)
This is a modification of a patch by Bill Pemberton <wfp5p@virginia.edu>

nobh_write_end() could call attach_nobh_buffers() with head == NULL.
This would result in a trap when attach_nobh_buffers() attempted to
access bh->b_this_page.

This can be illustrated by running the writev01 testcase from LTP on jfs.

This error was introduced by commit 5b41e74a "vfs: fix data leak in
nobh_write_end()".  That patch did not take into account that if
PageMappedToDisk() is true upon entry to nobh_write_begin(), then no
buffers will be allocated for the page.  In that case, we won't have to
worry about a failed write leaving unitialized data in the page.

Of course, head != NULL implies !page_has_buffers(page), so no need to
test both.

Signed-off-by: Dave Kleikamp <shaggy@linux.vnet.ibm.com>
Cc: Bill Pemberton <wfp5p@virginia.edu>
Cc: Dmitri Monakhov <dmonakhov@openvz.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/buffer.c

index b58208f1640a00eecbb0c760469a15b8fe31f6c3..665d446b25bc034241ef54c3c6b1d239c0ccf0f9 100644 (file)
@@ -2688,7 +2688,7 @@ int nobh_write_end(struct file *file, struct address_space *mapping,
        struct buffer_head *bh;
        BUG_ON(fsdata != NULL && page_has_buffers(page));
 
-       if (unlikely(copied < len) && !page_has_buffers(page))
+       if (unlikely(copied < len) && head)
                attach_nobh_buffers(page, head);
        if (page_has_buffers(page))
                return generic_write_end(file, mapping, pos, len,