On Mon, Jan 04, 2021 at 03:03:07PM +0100, Jan Beulich wrote: > On 29.12.2020 11:54, Roger Pau Monné wrote: > > My preference however would be for this to use vmap. Could the mapping > > be established in acpi_fadt_parse_sleep_info instead of having to map > > and unmap every time in acpi_sleep_prepare? > > Establishing a permanent mapping would seem like a waste of resources > to me, no matter that it's just a single page. In particular if the > system would never really make any attempt to enter S3.
IMO I would rather have a simpler path going into suspension (ie: not having to create a mapping to get the vector at all) with the cost of using one page of virtual address space permanently. Thanks, Roger.
