On Sun, Sep 11, 2016 at 05:47:41AM -0700, Christoph Hellwig wrote:
On Sat, Sep 10, 2016 at 07:52:53AM +0000, Matthew Wilcox wrote:
> DAX code over to using iomap requires converting all of ext2 away from
> buffer_head; are you saying he's wrong?
Not sure if he's really saying that, but it's wrong for sure. Just
to prove that I came up with a working ext2 iomap DAX implementation
in a few hours today. I'll take a stab at ext4 and the block device
as well and will post the updated series early next week - I'll need
to take care of a few high priority todo list items first.
Yay! Sorry if I was unclear, I wasn't trying to say that we had to change all
of ext2 over to using struct iomap. If we can (and apparently we can) just
switch over the DAX interfaces, that's good enough to me. I understand that
this will mean that we may have overlapping DAX paths for a while (an iomap
version and a buffer_head version). I just wanted to figure out whether this
overlap would need to be permanent - sounds like not, which is ideal.