On 2/26/25 11:02, Rik van Riel wrote: > On Wed, 2025-02-26 at 11:00 -0600, Tom Lendacky wrote: >> On 2/25/25 15:03, Borislav Petkov wrote: >>> On Tue, Feb 25, 2025 at 09:38:03PM +0100, Borislav Petkov wrote: >>>> On Tue, Feb 25, 2025 at 02:17:20PM -0500, Rik van Riel wrote: >>>>> Who do we need to ask to confirm that reading? >>>> >>>> Lemme figure it out. >>> >>> Confirmed - that really is the case. >> >> Hmmm... since this is for host/hypervisor TLB handling, this makes me >> think that we should always be setting the ASID valid bit with 0 for >> the >> ASID value in EDX[15:0] on all INVLPGB instructions in this series. >> Otherwise we will also be flushing any guest ASID TLB entries that >> match >> the conditions, when that isn't intended. > > The ASID is always zero in this series. > > We only ever use the PCID for address space identification. Right, but the ASID valid bit is not set, so the flushing may match more than just host/hypervisor TLB entries. Thanks, Tom >
