On Thu, Sep 08, 2022 at 12:27:06PM -0700, Dan Williams wrote: > This thread is mainly about DAX slowly reinventing _mapcount that gets > managed in all the right places for a normal_page. Longer term I think > we either need to get back to the page-less DAX experiment and find a > way to unwind some of this page usage creep, or cut over to finally > making DAX-pages be normal pages and delete most of the special case > handling in fs/dax.c. I am open to discussing the former, but I think > the latter is more likely.
I'm still very much in favour of the former. I've started on replacing scatterlist with phyr, but haven't got very far yet.
