On Wed, Jan 08, 2025 at 10:24:00AM +1100, Alistair Popple wrote: > On Mon, Jan 06, 2025 at 04:43:13PM -0800, Dan Williams wrote: > > Matthew Wilcox (Oracle) wrote: > > > This looks like a complete mess (why are we setting page->index at page > > > fault time?) > > > > Full story in Alistair's patches, but this a side effect of bypassing > > the page allocator for instantiating file-backed mappings. > > > > > but I no longer care about DAX, and there's no reason to > > > let DAX hold us back from removing page->index. > > > > Question is whether to move ahead with this now and have Alistair > > rebase, or push ahead with getting Alistair's series into -next? I am > > hoping that Alistair's series can move ahead this cycle, but still > > catching up on the latest after the holiday break. > > The rebase probably isn't that hard, but if we push ahead with my series it's > largely unnecessary as it moves this over to the folio anyway. I've just posted > a respin on top of next-20241216 - > https://lore.kernel.org/linux-mm/cover.425da7c4e76c2749d0ad1734f972b06114e02d52.1736221254.git-series.apopple@xxxxxxxxxx/ Looking at what's in linux-next today, there's no changes to dax_set_mapping(), so this patch still applies cleanly (patch 2/2 is obviated). Can you look at this patch (1/2) again and apply it if it seems good?
