The branch main has been updated by markj:

URL: 
https://cgit.FreeBSD.org/src/commit/?id=ef26f64c11220ea303de16d088c585f9d2908b0f

commit ef26f64c11220ea303de16d088c585f9d2908b0f
Author:     ShengYi Hung <[email protected]>
AuthorDate: 2026-09-11 08:22:39 +0000
Commit:     Mark Johnston <[email protected]>
CommitDate: 2026-09-11 13:01:44 +0000

    amd64/pmap: Return la57 in amd nested page
    
    AMD doesn't encode page depth in eptp. As a result, the page level is
    decided by the host la57 value. Without this, it uses 4 level page and
    therefore cause machine enable la57 have garbage page translation.
    
    PR:             291768
    Reviewed by:    kib
    Tested by:      Antranig Vartanian <[email protected]>
    MFC after:      1 week
    Differential Revision:  https://reviews.freebsd.org/D57978
---
 sys/amd64/amd64/pmap.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/sys/amd64/amd64/pmap.c b/sys/amd64/amd64/pmap.c
index eee310cc44b8..8f8d91c768b7 100644
--- a/sys/amd64/amd64/pmap.c
+++ b/sys/amd64/amd64/pmap.c
@@ -443,7 +443,19 @@ pmap_is_la57(pmap_t pmap)
 {
        if (pmap->pm_type == PT_X86)
                return (la57);
-       return (false);         /* XXXKIB handle EPT */
+       if (pmap->pm_type == PT_RVI) {
+               /*
+                * AMD nested paging has no field to specify the nested
+                * page table walk length; the hardware derives it from
+                * the host CR4.LA57 setting (see svm.c, where nCR3 is
+                * loaded directly from pm_pmltop with no level encoding).
+                * The NPT must therefore have the same number of levels
+                * as the host page tables, otherwise guest-physical
+                * translations are walked at the wrong depth.
+                */
+               return (la57);
+       }
+       return (false);         /* Intel EPT encodes its own walk length. */
 }
 
 #define        PAT_INDEX_SIZE  8

Reply via email to