On Fri, Aug 10, 2018 at 9:23 AM Dave Jiang [off-list ref] wrote:

On 08/10/2018 11:31 AM, Eric Sandeen wrote:

quoted

On 8/8/18 12:31 PM, Dave Jiang wrote:

quoted

This patch is the duplicate of ross's fix for ext4 for xfs.

If the refcount of a page is lowered between the time that it is returned
by dax_busy_page() and when the refcount is again checked in
xfs_break_layouts() => ___wait_var_event(), the waiting function
xfs_wait_dax_page() will never be called.  This means that
xfs_break_layouts() will still have 'retry' set to false, so we'll stop
looping and never check the refcount of other pages in this inode.

Instead, always continue looping as long as dax_layout_busy_page() gives us
a page which it found with an elevated refcount.

Hi Dave, does this have a testcase?  Have you seen the issue using Ross's
xfstest generic/503 or is there some other test?  Apologies if I missed
prior discussion on a testcase or race frequency...

I do not have a testcase. I know Ross replicated it on ext4. And Jan
asked to create the same fix with XFS when he reviewed Ross's fix for ext4.

In my testing I couldn't get this race to hit with XFS.  I couldn't
even get a failure with generic/503 when testing XFS before Dan's
initial patches went in which added xfs_break_layouts() et al.  I
think that Dan had to manually insert timing delays to get the warning
to hit for XFS when testing his patches.

The race we're fixing happens consistently with ext4 and through code
inspection we can see that the race exists in XFS.