Thread (53 messages) 53 messages, 3 authors, 2026-02-02

Re: [PATCH 05/20] KVM: arm64: Extend unified RESx handling to runtime sanitisation

From: Fuad Tabba <tabba@google.com>
Date: 2026-01-26 19:17:09
Also in: kvm, kvmarm

Hi Marc,

On Mon, 26 Jan 2026 at 12:17, Marc Zyngier [off-list ref] wrote:
quoted hunk ↗ jump to hunk
Add a new helper to retrieve the RESx values for a given system
register, and use it for the runtime sanitisation.

This results in slightly better code generation for a fairly hot
path in the hypervisor.

Signed-off-by: Marc Zyngier <maz@kernel.org>
---
 arch/arm64/include/asm/kvm_host.h | 13 +++++++++++++
 arch/arm64/kvm/emulate-nested.c   | 10 +---------
 arch/arm64/kvm/nested.c           | 13 ++++---------
 3 files changed, 18 insertions(+), 18 deletions(-)
diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
index a7e4cd8ebf56f..9dca94e4361f0 100644
--- a/arch/arm64/include/asm/kvm_host.h
+++ b/arch/arm64/include/asm/kvm_host.h
@@ -635,6 +635,19 @@ struct kvm_sysreg_masks {
        struct resx mask[NR_SYS_REGS - __SANITISED_REG_START__];
 };

+#define kvm_get_sysreg_resx(k, sr)                                     \
+       ({                                                              \
+               struct kvm_sysreg_masks *__masks;                       \
+               struct resx __resx = {};                                \
+                                                                       \
+               __masks = (k)->arch.sysreg_masks;                       \
+               if (likely(__masks &&                                   \
+                          sr >= __SANITISED_REG_START__ &&             \
+                          sr < NR_SYS_REGS))                           \
+                       __resx = __masks->mask[sr - __SANITISED_REG_START__]; \
+               __resx;                                                 \
+       })
+
This now covers all registers that need to be sanitized, not just
VNCR-backed ones now.

nit: wouldn't it be better to capture sr in a local variable rather
than reuse it? It is an enum, but it would make checkpatch feel
slightly better :)

Reviewed-by: Fuad Tabba <tabba@google.com>

Cheers,
/fuad




quoted hunk ↗ jump to hunk
 #define kvm_set_sysreg_resx(k, sr, resx)               \
        do {                                            \
                (k)->arch.sysreg_masks->mask[sr - __SANITISED_REG_START__] = resx; \
diff --git a/arch/arm64/kvm/emulate-nested.c b/arch/arm64/kvm/emulate-nested.c
index 774cfbf5b43ba..43334cd2db9e5 100644
--- a/arch/arm64/kvm/emulate-nested.c
+++ b/arch/arm64/kvm/emulate-nested.c
@@ -2427,15 +2427,7 @@ static enum trap_behaviour compute_trap_behaviour(struct kvm_vcpu *vcpu,

 static u64 kvm_get_sysreg_res0(struct kvm *kvm, enum vcpu_sysreg sr)
 {
-       struct kvm_sysreg_masks *masks;
-
-       /* Only handle the VNCR-backed regs for now */
-       if (sr < __VNCR_START__)
-               return 0;
-
-       masks = kvm->arch.sysreg_masks;
-
-       return masks->mask[sr - __SANITISED_REG_START__].res0;
+       return kvm_get_sysreg_resx(kvm, sr).res0;
 }

 static bool check_fgt_bit(struct kvm_vcpu *vcpu, enum vcpu_sysreg sr,
diff --git a/arch/arm64/kvm/nested.c b/arch/arm64/kvm/nested.c
index c5a45bc62153e..75a23f1c56d13 100644
--- a/arch/arm64/kvm/nested.c
+++ b/arch/arm64/kvm/nested.c
@@ -1669,16 +1669,11 @@ u64 limit_nv_id_reg(struct kvm *kvm, u32 reg, u64 val)
 u64 kvm_vcpu_apply_reg_masks(const struct kvm_vcpu *vcpu,
                             enum vcpu_sysreg sr, u64 v)
 {
-       struct kvm_sysreg_masks *masks;
-
-       masks = vcpu->kvm->arch.sysreg_masks;
-
-       if (masks) {
-               sr -= __SANITISED_REG_START__;
+       struct resx resx;

-               v &= ~masks->mask[sr].res0;
-               v |= masks->mask[sr].res1;
-       }
+       resx = kvm_get_sysreg_resx(vcpu->kvm, sr);
+       v &= ~resx.res0;
+       v |= resx.res1;

        return v;
 }
--
2.47.3
  
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help