Subject: [PATCH] [2.4] beyond_eof check in generic_direct_IO
From: Chris Mason <mason@suse.com>
To: marcelo@conectiva.com.br
Cc: linux-kernel@vger.kernel.org
Content-Type: text/plain
Organization: 
Message-Id: <1060197938.5877.442.camel@tiny.suse.com>
Mime-Version: 1.0
Date: 06 Aug 2003 15:25:38 -0400
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1232
Lines: 36

Hello all,

reiserfs has to do a few odd tail conversion steps during direct io. 
Picture a file whose size is currently 3500 bytes, and a direct io write
extending it to 4096 bytes.

reiserfs will have those 3500 bytes packed into a tail, and when it gets
a get_block call with create == 1, it will convert the tail into a full
block and return the newly created block.

The new generic_direct_IO code doesn't send create == 1 to get_block
unless the new block is entirely past eof.  I think it should instead
allow any part of the block to be past eof, which allows reiserfs to
properly convert the tail:

-chris

--- 1.89/fs/buffer.c	Tue Jun 24 17:11:29 2003
+++ edited/fs/buffer.c	Fri Jul 18 13:24:57 2003
@@ -2147,7 +2148,7 @@
 		bh.b_size = blocksize;
 		bh.b_page = NULL;
 
-		if (((loff_t) blocknr) * blocksize >= inode->i_size)
+		if (((loff_t) (blocknr + 1)) * blocksize > inode->i_size)
 			beyond_eof = 1;
 
 		/* Only allow get_block to create new blocks if we are safely


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/