On Thu, Mar 04, 2021 at 06:44:19PM +0100, David Sterba wrote:

On Tue, Feb 23, 2021 at 12:08:46PM +0000, fdmanana@kernel.org wrote:

quoted

From: Filipe Manana <redacted>

The first patch fixes a race between fsync and memory mapped writes, which
can result in corruptions. The second one fixes a different race that in
practice should be "impossible" to happen, but in case it's triggered
somehow, results in not logging an inode when it has new extents. The last
patch just removes some no longer needed code.

The first patch, as mentioned in its changelog, depends on Josef's patchset
which has the subject:

   "Introduce a mmap sem to deal with some mmap issues"

The patchset is now in for-next so I'll add your fixes on top of that.

Moved to misc-next, thanks.