From: Dave Chinner <david@fromorbit.com>
Date: Thu, 16 Apr 2015 12:13:18 +0000 (+1000)
Subject: Merge branch 'xfs-dio-extend-fix' into for-next
X-Git-Tag: firefly_0821_release~176^2~1877^2
X-Git-Url: http://demsky.eecs.uci.edu/git/?a=commitdiff_plain;h=542c311813d5cb2e6f0dfa9557f41c829b8fb6a0;p=firefly-linux-kernel-4.4.55.git

Merge branch 'xfs-dio-extend-fix' into for-next

Conflicts:
	fs/xfs/xfs_file.c
---

542c311813d5cb2e6f0dfa9557f41c829b8fb6a0
diff --cc fs/xfs/xfs_file.c
index c203839cd5be,79ffb3e74f49..3a5d305e60c9
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@@ -569,20 -565,39 +569,41 @@@ restart
  	 * write.  If zeroing is needed and we are currently holding the
  	 * iolock shared, we need to update it to exclusive which implies
  	 * having to redo all checks before.
+ 	 *
+ 	 * We need to serialise against EOF updates that occur in IO
+ 	 * completions here. We want to make sure that nobody is changing the
+ 	 * size while we do this check until we have placed an IO barrier (i.e.
+ 	 * hold the XFS_IOLOCK_EXCL) that prevents new IO from being dispatched.
+ 	 * The spinlock effectively forms a memory barrier once we have the
+ 	 * XFS_IOLOCK_EXCL so we are guaranteed to see the latest EOF value
+ 	 * and hence be able to correctly determine if we need to run zeroing.
  	 */
+ 	spin_lock(&ip->i_flags_lock);
  	if (*pos > i_size_read(inode)) {
 +		bool	zero = false;
 +
+ 		spin_unlock(&ip->i_flags_lock);
  		if (*iolock == XFS_IOLOCK_SHARED) {
  			xfs_rw_iunlock(ip, *iolock);
  			*iolock = XFS_IOLOCK_EXCL;
  			xfs_rw_ilock(ip, *iolock);
+ 
+ 			/*
+ 			 * We now have an IO submission barrier in place, but
+ 			 * AIO can do EOF updates during IO completion and hence
+ 			 * we now need to wait for all of them to drain. Non-AIO
+ 			 * DIO will have drained before we are given the
+ 			 * XFS_IOLOCK_EXCL, and so for most cases this wait is a
+ 			 * no-op.
+ 			 */
+ 			inode_dio_wait(inode);
  			goto restart;
  		}
 -		error = xfs_zero_eof(ip, *pos, i_size_read(inode));
 +		error = xfs_zero_eof(ip, *pos, i_size_read(inode), &zero);
  		if (error)
  			return error;
- 	}
+ 	} else
+ 		spin_unlock(&ip->i_flags_lock);
  
  	/*
  	 * Updating the timestamps will grab the ilock again from