[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20140827213103.GV11317@thunk.org>
Date: Wed, 27 Aug 2014 17:31:03 -0400
From: Theodore Ts'o <tytso@....edu>
To: Dmitry Monakhov <dmonakhov@...nvz.org>
Cc: linux-ext4@...r.kernel.org, lczerner@...hat.com
Subject: Re: [PATCH 3/5] ext4_zero_range: fix incorect journal credits
reservation
On Mon, Aug 25, 2014 at 11:19:39AM +0400, Dmitry Monakhov wrote:
> Yes, but ext4_index_trans_blocks is statically defined in inode.c,
> but since zero_range works only for extent based
> ext4_ext_index_trans_blocks is sufficient. Also we must reserve 2 block
> for data. So final calculation looks like follows:
> credits = ext4_ext_index_trans_blocks(inode, 2) + 2
It turns out this wasn't enough. This was causing test failures in
the data=journal case for tests generic/008, generic/091, generic/127,
and generic/263, due to not reserving enough credits. After taking a
closer look at the code, and the requirements of ext4_zero_range, I
believe this is the correct fix.
- Ted
diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index 0d531d1..0e9de23 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -4831,8 +4831,13 @@ static long ext4_zero_range(struct file *file, loff_t offset,
if (ret)
goto out_dio;
}
- /* In worst case we have to writeout two nonadjacent unwritten blocks */
- credits = ext4_ext_index_trans_blocks(inode, 2) + 2;
+ /*
+ * In worst case we have to writeout two nonadjacent unwritten
+ * blocks and update the inode
+ */
+ credits = (2 * ext4_ext_index_trans_blocks(inode, 2)) + 1;
+ if (ext4_should_journal_data(inode))
+ credits += 2;
handle = ext4_journal_start(inode, EXT4_HT_MISC, credits);
if (IS_ERR(handle)) {
ret = PTR_ERR(handle);
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists