Browse Source

GFS2: Make atime checks more efficient

We do not need to start a transaction unless the atime
check has proved positive. Also if we are going to flush
the complete ail list anyway, we might as well skip the
writeback for this specific inode's metadata, since that
will be done as part of the ail writeback process in an
order offering potentially more efficient I/O.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Steven Whitehouse 14 years ago
parent
commit
1d4ec642d9
1 changed files with 8 additions and 8 deletions
  1. 8 8
      fs/gfs2/super.c

+ 8 - 8
fs/gfs2/super.c

@@ -768,30 +768,30 @@ static int gfs2_write_inode(struct inode *inode, struct writeback_control *wbc)
 			goto do_flush;
 		unlock_required = 1;
 	}
-	ret = gfs2_trans_begin(sdp, RES_DINODE, 0);
-	if (ret)
-		goto do_unlock;
 	ret = gfs2_meta_inode_buffer(ip, &bh);
 	if (ret == 0) {
 		di = (struct gfs2_dinode *)bh->b_data;
 		atime.tv_sec = be64_to_cpu(di->di_atime);
 		atime.tv_nsec = be32_to_cpu(di->di_atime_nsec);
 		if (timespec_compare(&inode->i_atime, &atime) > 0) {
-			gfs2_trans_add_bh(ip->i_gl, bh, 1);
-			gfs2_dinode_out(ip, bh->b_data);
+			ret = gfs2_trans_begin(sdp, RES_DINODE, 0);
+			if (ret == 0) {
+				gfs2_trans_add_bh(ip->i_gl, bh, 1);
+				gfs2_dinode_out(ip, bh->b_data);
+				gfs2_trans_end(sdp);
+			}
 		}
 		brelse(bh);
 	}
-	gfs2_trans_end(sdp);
-do_unlock:
 	if (unlock_required)
 		gfs2_glock_dq_uninit(&gh);
 do_flush:
 	if (wbc->sync_mode == WB_SYNC_ALL)
 		gfs2_log_flush(GFS2_SB(inode), ip->i_gl);
-	filemap_fdatawrite(metamapping);
 	if (bdi->dirty_exceeded)
 		gfs2_ail1_flush(sdp, wbc);
+	else
+		filemap_fdatawrite(metamapping);
 	if (!ret && (wbc->sync_mode == WB_SYNC_ALL))
 		ret = filemap_fdatawait(metamapping);
 	if (ret)