On 23.12.2023 22:35, Christopher Clark wrote: > On Sat, Dec 23, 2023 at 12:47 PM Christopher Clark > <[email protected]> wrote: >> >> On Sun, Dec 17, 2023 at 11:55 PM Jan Beulich <[email protected]> wrote: >>> >>> Christopher, >>> >>> On 27.11.2023 14:55, Jan Beulich wrote: >>>> Holding a valid struct page_info * in hands already means the referenced >>>> MFN is valid; there's no need to check that again. Convert the checking >>>> logic to a switch(), to help keeping the extra (and questionable) x86- >>>> only check in somewhat tidy shape. >>>> >>>> Signed-off-by: Jan Beulich <[email protected]> >> >> Reviewed-by: Christopher Clark <[email protected]> > > I'd like to hold off on this just yet, sorry -- the change does look > ok as far as a transform being applied the prior logic and the > necessity of the check, but with it applied, it's not obvious that it > handles all the page types as best that it could there, so I'd like to > look at this (and the previously submitted patch again) please.
I'm puzzled: This patch merely removes a pointless check. Whatever is lacking with it in place will have been lacking before. Also can you please give a rough estimate towards when you'll be getting back on this, or ideally on both patches? Jan
