On Wed, 19 May 2021 23:07:50 +0200 Uladzislau Rezki [off-list ref] wrote:

quoted

Basically a single-page allocator is more permissive so it is a higher
chance to success. Therefore a fallback to it makes sense.

Hello, Christoph.

See below the patch. Does it sound good for you? It is about moving
page allocation part into separate function:

Please just send over a patch which addresses this and Christoph's
other review comments.

I don't think the discussion with Mel against this patch identified any
needed changes, so I'll send this series to Linus.