From mboxrd@z Thu Jan  1 00:00:00 1970
From: wcheng@sourceware.org <wcheng@sourceware.org>
Date: 17 Nov 2006 05:00:57 -0000
Subject: [Cluster-devel] cluster/gfs-kernel/src/gfs ops_file.c
Message-ID: <20061117050057.27532.qmail@sourceware.org>
List-Id: <cluster-devel.redhat.com>
To: cluster-devel.redhat.com
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit

CVSROOT:	/cvs/cluster
Module name:	cluster
Changes by:	wcheng at sourceware.org	2006-11-17 05:00:57

Modified files:
	gfs-kernel/src/gfs: ops_file.c 

Log message:
	Bugzilla 214274: GFS has been splitting large writes into smaller atomic
	transactions. This would generate multiple aio completion calls (one
	for each transaction) that falsely notify application about data
	completion. Problem is reported by QA team as data corruption.

Patches:
http://sourceware.org/cgi-bin/cvsweb.cgi/cluster/gfs-kernel/src/gfs/ops_file.c.diff?cvsroot=cluster&r1=1.28&r2=1.29

--- cluster/gfs-kernel/src/gfs/ops_file.c	2006/10/27 16:22:17	1.28
+++ cluster/gfs-kernel/src/gfs/ops_file.c	2006/11/17 05:00:57	1.29
@@ -702,10 +702,13 @@
 
 		/* split large writes into smaller atomic transactions */
 		while (size) {
-			s = gfs_tune_get(sdp, gt_max_atomic_write);
-			if (s > size)
+			if (iocb)
 				s = size;
-
+			else {
+				s = gfs_tune_get(sdp, gt_max_atomic_write);
+				if (s > size) 
+					s = size;
+			}
 			error = do_write_direct_alloc(file, buf, s, offset, iocb);
 			if (error < 0)
 				goto out_gunlock;
@@ -974,9 +977,13 @@
 
 	/* split large writes into smaller atomic transactions */
 	while (size) {
-		s = gfs_tune_get(sdp, gt_max_atomic_write);
-		if (s > size)
+		if (iocb)
 			s = size;
+		else {
+			s = gfs_tune_get(sdp, gt_max_atomic_write);
+			if (s > size)
+				s = size;
+		}
 
 		error = do_do_write_buf(file, buf, s, offset, iocb);
 		if (error < 0)