Enable FEAT_NV on the 'max' CPU, and stop filtering it out for the
Neoverse N2 and Neoverse V1 CPUs.  We continue to downgrade FEAT_NV2
support to FEAT_NV for the latter two CPU types.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 docs/system/arm/emulation.rst | 1 +
 target/arm/cpu.c              | 8 +++++---
 target/arm/tcg/cpu64.c        | 1 +
 3 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/docs/system/arm/emulation.rst b/docs/system/arm/emulation.rst
index 0b604f90059..d827b42de79 100644
--- a/docs/system/arm/emulation.rst
+++ b/docs/system/arm/emulation.rst
@@ -63,6 +63,7 @@ the following architecture extensions:
 - FEAT_MTE (Memory Tagging Extension)
 - FEAT_MTE2 (Memory Tagging Extension)
 - FEAT_MTE3 (MTE Asymmetric Fault Handling)
+- FEAT_NV (Nested Virtualization)
 - FEAT_PACIMP (Pointer authentication - IMPLEMENTATION DEFINED algorithm)
 - FEAT_PACQARMA3 (Pointer authentication - QARMA3 algorithm)
 - FEAT_PACQARMA5 (Pointer authentication - QARMA5 algorithm)
diff --git a/target/arm/cpu.c b/target/arm/cpu.c
index efb22a87f9e..da0c02f850b 100644
--- a/target/arm/cpu.c
+++ b/target/arm/cpu.c
@@ -2238,9 +2238,11 @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp)
         /* FEAT_MPAM (Memory Partitioning and Monitoring Extension) */
         cpu->isar.id_aa64pfr0 =
             FIELD_DP64(cpu->isar.id_aa64pfr0, ID_AA64PFR0, MPAM, 0);
-        /* FEAT_NV (Nested Virtualization) */
-        cpu->isar.id_aa64mmfr2 =
-            FIELD_DP64(cpu->isar.id_aa64mmfr2, ID_AA64MMFR2, NV, 0);
+        /* FEAT_NV2 (Enhanced Nested Virtualization support) */
+        if (FIELD_EX64(cpu->isar.id_aa64mmfr2, ID_AA64MMFR2, NV) > 1) {
+            cpu->isar.id_aa64mmfr2 =
+                FIELD_DP64(cpu->isar.id_aa64mmfr2, ID_AA64MMFR2, NV, 1);
+        }
     }
 
     /* MPU can be configured out of a PMSA CPU either by setting has-mpu
diff --git a/target/arm/tcg/cpu64.c b/target/arm/tcg/cpu64.c
index 40e7a45166f..93f040e6e96 100644
--- a/target/arm/tcg/cpu64.c
+++ b/target/arm/tcg/cpu64.c
@@ -1204,6 +1204,7 @@ void aarch64_max_tcg_initfn(Object *obj)
     t = FIELD_DP64(t, ID_AA64MMFR2, UAO, 1);      /* FEAT_UAO */
     t = FIELD_DP64(t, ID_AA64MMFR2, IESB, 1);     /* FEAT_IESB */
     t = FIELD_DP64(t, ID_AA64MMFR2, VARANGE, 1);  /* FEAT_LVA */
+    t = FIELD_DP64(t, ID_AA64MMFR2, NV, 1);       /* FEAT_NV */
     t = FIELD_DP64(t, ID_AA64MMFR2, ST, 1);       /* FEAT_TTST */
     t = FIELD_DP64(t, ID_AA64MMFR2, AT, 1);       /* FEAT_LSE2 */
     t = FIELD_DP64(t, ID_AA64MMFR2, IDS, 1);      /* FEAT_IDST */
-- 
2.34.1

W dniu 18.12.2023 o 12:32, Peter Maydell pisze:

> Enable FEAT_NV on the 'max' CPU, and stop filtering it out for the 
> Neoverse N2 and Neoverse V1 CPUs.  We continue to downgrade FEAT_NV2 
> support to FEAT_NV for the latter two CPU types.

According to Neoverse-V1 TRM r1p2 it has FEAT_NV2. Similar with Neoverse-N2.

You wrote already:

> in practice hypervisors such as KVM are going to require FEAT_NV2 
> and not bother to support the FEAT_NV-only case, so I have
> implemented them one after the other in this single patchset.
So maybe they both should be FEAT_NV2 and FEAT_NV will be left unused. 
Or enable FEAT_NV for V1 (as being older) and FEAT_NV2 on N2.

This way if someone wants to test nested virtualization then both 
versions will be available without use of 'max' cpu.

On Fri, 29 Dec 2023 at 11:37, Marcin Juszkiewicz
<marcin.juszkiewicz@linaro.org> wrote:
>
> W dniu 18.12.2023 o 12:32, Peter Maydell pisze:
>
> > Enable FEAT_NV on the 'max' CPU, and stop filtering it out for the
> > Neoverse N2 and Neoverse V1 CPUs.  We continue to downgrade FEAT_NV2
> > support to FEAT_NV for the latter two CPU types.
>
> According to Neoverse-V1 TRM r1p2 it has FEAT_NV2. Similar with Neoverse-N2.

I'm not sure what you're trying to say here ? Yes, the V1 has
NV2, that's why we have to downgrade it to FEAT_NV for the moment
when we're emulating, because at this point in the series we
don't have FEAT_NV2 emulation. At the end of the series this downgrade
goes away, because at that point we can emulate FEAT_NV2.

>
> You wrote already:
>
> > in practice hypervisors such as KVM are going to require FEAT_NV2
> > and not bother to support the FEAT_NV-only case, so I have
> > implemented them one after the other in this single patchset.
> So maybe they both should be FEAT_NV2 and FEAT_NV will be left unused.
> Or enable FEAT_NV for V1 (as being older) and FEAT_NV2 on N2.

As usual, we emulate what the actual hardware has, to the extent
that we are able to. For both of these CPUs that's both FEAT_NV
and FEAT_NV2, and we enable emulation of each at the point in the
series when we've implemented it.

> This way if someone wants to test nested virtualization then both
> versions will be available without use of 'max' cpu.

In theory it might be nice to be able to emulate a pure FEAT_NV
CPU, but in practice I don't expect anybody to want to do that.
If anybody has a real use case for it we can add a -cpu suboption
later.

thanks
-- PMM

On 12/18/23 22:32, Peter Maydell wrote:
> Enable FEAT_NV on the 'max' CPU, and stop filtering it out for the
> Neoverse N2 and Neoverse V1 CPUs.  We continue to downgrade FEAT_NV2
> support to FEAT_NV for the latter two CPU types.
> 
> Signed-off-by: Peter Maydell<peter.maydell@linaro.org>
> ---
>   docs/system/arm/emulation.rst | 1 +
>   target/arm/cpu.c              | 8 +++++---
>   target/arm/tcg/cpu64.c        | 1 +
>   3 files changed, 7 insertions(+), 3 deletions(-)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~