[PATCH] KVM: arm64: Flush HCR_EL2.VSE to deliver SErrors to pKVM guests

[PATCH] KVM: arm64: Flush HCR_EL2.VSE to deliver SErrors to pKVM guests

[tabba]

With pKVM enabled, the host injects a virtual SError by setting
HCR_EL2.VSE on its vCPU copy, but flush_hyp_vcpu() only flows TWI/TWE
into the hyp vCPU that runs, so VSE never reaches it and a deferred
(masked) SError is never delivered. VSE is a host-owned injection
control, not a trap-configuration bit, so restricting the host's
trap-register values should not have dropped it. Flow it on entry;
sync_hyp_vcpu() already copies hcr_el2 back, so delivery is reflected
to the host.

Fixes: b56680de9c648 ("KVM: arm64: Initialize trap register values in hyp in pKVM")
Reported-by: Sashiko (local):gemini-3.1-pro
Signed-off-by: Fuad Tabba <tabba@google.com>
---
Not an urgent fix. I should stop running Sashiko...
---
 arch/arm64/kvm/hyp/nvhe/hyp-main.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
index 06db299c37a8..9e4a20df6409 100644
--- a/arch/arm64/kvm/hyp/nvhe/hyp-main.c
+++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
@@ -129,9 +129,14 @@ static void flush_hyp_vcpu(struct pkvm_hyp_vcpu *hyp_vcpu)
 	hyp_vcpu->vcpu.arch.ctxt	= host_vcpu->arch.ctxt;
 
 	hyp_vcpu->vcpu.arch.mdcr_el2	= host_vcpu->arch.mdcr_el2;
-	hyp_vcpu->vcpu.arch.hcr_el2 &= ~(HCR_TWI | HCR_TWE);
+	/*
+	 * HCR_EL2.VSE is host-owned (a pending virtual SError to inject), not a
+	 * trap-control bit, so it must flow to the hyp vCPU alongside TWI/TWE
+	 * for the vSError to be delivered. sync_hyp_vcpu() reflects it back.
+	 */
+	hyp_vcpu->vcpu.arch.hcr_el2 &= ~(HCR_TWI | HCR_TWE | HCR_VSE);
 	hyp_vcpu->vcpu.arch.hcr_el2 |= READ_ONCE(host_vcpu->arch.hcr_el2) &
-						 (HCR_TWI | HCR_TWE);
+						 (HCR_TWI | HCR_TWE | HCR_VSE);
 
 	hyp_vcpu->vcpu.arch.iflags	= host_vcpu->arch.iflags;
 
-- 
2.54.0.929.g9b7fa37559-goog

Re: [PATCH] KVM: arm64: Flush HCR_EL2.VSE to deliver SErrors to pKVM guests

[Oliver Upton]

On Sun, May 31, 2026 at 04:45:48PM +0100, tabba@google.com wrote:
> With pKVM enabled, the host injects a virtual SError by setting
> HCR_EL2.VSE on its vCPU copy, but flush_hyp_vcpu() only flows TWI/TWE
> into the hyp vCPU that runs, so VSE never reaches it and a deferred
> (masked) SError is never delivered. VSE is a host-owned injection
> control, not a trap-configuration bit, so restricting the host's
> trap-register values should not have dropped it. Flow it on entry;
> sync_hyp_vcpu() already copies hcr_el2 back, so delivery is reflected
> to the host.

Might be worth mentioning that flush_hyp_vcpu() also forwards VSESR_EL2
from the host since that bit is _just_ out of context of this diff :)

> Fixes: b56680de9c648 ("KVM: arm64: Initialize trap register values in hyp in pKVM")
> Reported-by: Sashiko (local):gemini-3.1-pro
> Signed-off-by: Fuad Tabba <tabba@google.com>

Reviewed-by: Oliver Upton <oupton@kernel.org>

Thanks,
Oliver

> ---
> Not an urgent fix. I should stop running Sashiko...
> ---
>  arch/arm64/kvm/hyp/nvhe/hyp-main.c | 9 +++++++--
>  1 file changed, 7 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
> index 06db299c37a8..9e4a20df6409 100644
> --- a/arch/arm64/kvm/hyp/nvhe/hyp-main.c
> +++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
> @@ -129,9 +129,14 @@ static void flush_hyp_vcpu(struct pkvm_hyp_vcpu *hyp_vcpu)
>  	hyp_vcpu->vcpu.arch.ctxt	= host_vcpu->arch.ctxt;
>  
>  	hyp_vcpu->vcpu.arch.mdcr_el2	= host_vcpu->arch.mdcr_el2;
> -	hyp_vcpu->vcpu.arch.hcr_el2 &= ~(HCR_TWI | HCR_TWE);
> +	/*
> +	 * HCR_EL2.VSE is host-owned (a pending virtual SError to inject), not a
> +	 * trap-control bit, so it must flow to the hyp vCPU alongside TWI/TWE
> +	 * for the vSError to be delivered. sync_hyp_vcpu() reflects it back.
> +	 */
> +	hyp_vcpu->vcpu.arch.hcr_el2 &= ~(HCR_TWI | HCR_TWE | HCR_VSE);
>  	hyp_vcpu->vcpu.arch.hcr_el2 |= READ_ONCE(host_vcpu->arch.hcr_el2) &
> -						 (HCR_TWI | HCR_TWE);
> +						 (HCR_TWI | HCR_TWE | HCR_VSE);
>  
>  	hyp_vcpu->vcpu.arch.iflags	= host_vcpu->arch.iflags;
>  
> -- 
> 2.54.0.929.g9b7fa37559-goog
> 

Re: [PATCH] KVM: arm64: Flush HCR_EL2.VSE to deliver SErrors to pKVM guests

[Fuad Tabba]

On Mon, 1 Jun 2026 at 20:16, Oliver Upton <oupton@kernel.org> wrote:
>
> On Sun, May 31, 2026 at 04:45:48PM +0100, tabba@google.com wrote:
> > With pKVM enabled, the host injects a virtual SError by setting
> > HCR_EL2.VSE on its vCPU copy, but flush_hyp_vcpu() only flows TWI/TWE
> > into the hyp vCPU that runs, so VSE never reaches it and a deferred
> > (masked) SError is never delivered. VSE is a host-owned injection
> > control, not a trap-configuration bit, so restricting the host's
> > trap-register values should not have dropped it. Flow it on entry;
> > sync_hyp_vcpu() already copies hcr_el2 back, so delivery is reflected
> > to the host.
>
> Might be worth mentioning that flush_hyp_vcpu() also forwards VSESR_EL2
> from the host since that bit is _just_ out of context of this diff :)

Sure, would you like me to respin? Or would one of the maintainers
kindly add that? :)

Cheers,
/fuad

> > Fixes: b56680de9c648 ("KVM: arm64: Initialize trap register values in hyp in pKVM")
> > Reported-by: Sashiko (local):gemini-3.1-pro
> > Signed-off-by: Fuad Tabba <tabba@google.com>
>
> Reviewed-by: Oliver Upton <oupton@kernel.org>
>
> Thanks,
> Oliver
>
> > ---
> > Not an urgent fix. I should stop running Sashiko...
> > ---
> >  arch/arm64/kvm/hyp/nvhe/hyp-main.c | 9 +++++++--
> >  1 file changed, 7 insertions(+), 2 deletions(-)
> >
> > diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
> > index 06db299c37a8..9e4a20df6409 100644
> > --- a/arch/arm64/kvm/hyp/nvhe/hyp-main.c
> > +++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
> > @@ -129,9 +129,14 @@ static void flush_hyp_vcpu(struct pkvm_hyp_vcpu *hyp_vcpu)
> >       hyp_vcpu->vcpu.arch.ctxt        = host_vcpu->arch.ctxt;
> >
> >       hyp_vcpu->vcpu.arch.mdcr_el2    = host_vcpu->arch.mdcr_el2;
> > -     hyp_vcpu->vcpu.arch.hcr_el2 &= ~(HCR_TWI | HCR_TWE);
> > +     /*
> > +      * HCR_EL2.VSE is host-owned (a pending virtual SError to inject), not a
> > +      * trap-control bit, so it must flow to the hyp vCPU alongside TWI/TWE
> > +      * for the vSError to be delivered. sync_hyp_vcpu() reflects it back.
> > +      */
> > +     hyp_vcpu->vcpu.arch.hcr_el2 &= ~(HCR_TWI | HCR_TWE | HCR_VSE);
> >       hyp_vcpu->vcpu.arch.hcr_el2 |= READ_ONCE(host_vcpu->arch.hcr_el2) &
> > -                                              (HCR_TWI | HCR_TWE);
> > +                                              (HCR_TWI | HCR_TWE | HCR_VSE);
> >
> >       hyp_vcpu->vcpu.arch.iflags      = host_vcpu->arch.iflags;
> >
> > --
> > 2.54.0.929.g9b7fa37559-goog
> >

Re: [PATCH] KVM: arm64: Flush HCR_EL2.VSE to deliver SErrors to pKVM guests

[Oliver Upton]

On Mon, Jun 01, 2026 at 08:17:18PM +0100, Fuad Tabba wrote:
> On Mon, 1 Jun 2026 at 20:16, Oliver Upton <oupton@kernel.org> wrote:
> >
> > On Sun, May 31, 2026 at 04:45:48PM +0100, tabba@google.com wrote:
> > > With pKVM enabled, the host injects a virtual SError by setting
> > > HCR_EL2.VSE on its vCPU copy, but flush_hyp_vcpu() only flows TWI/TWE
> > > into the hyp vCPU that runs, so VSE never reaches it and a deferred
> > > (masked) SError is never delivered. VSE is a host-owned injection
> > > control, not a trap-configuration bit, so restricting the host's
> > > trap-register values should not have dropped it. Flow it on entry;
> > > sync_hyp_vcpu() already copies hcr_el2 back, so delivery is reflected
> > > to the host.
> >
> > Might be worth mentioning that flush_hyp_vcpu() also forwards VSESR_EL2
> > from the host since that bit is _just_ out of context of this diff :)
> 
> Sure, would you like me to respin? Or would one of the maintainers
> kindly add that? :)

Marc is kindly dealing with the tree for the time being so hopefully
he can throw it in.

Thanks,
Oliver

> > > Fixes: b56680de9c648 ("KVM: arm64: Initialize trap register values in hyp in pKVM")
> > > Reported-by: Sashiko (local):gemini-3.1-pro
> > > Signed-off-by: Fuad Tabba <tabba@google.com>
> >
> > Reviewed-by: Oliver Upton <oupton@kernel.org>
> >
> > Thanks,
> > Oliver
> >
> > > ---
> > > Not an urgent fix. I should stop running Sashiko...
> > > ---
> > >  arch/arm64/kvm/hyp/nvhe/hyp-main.c | 9 +++++++--
> > >  1 file changed, 7 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
> > > index 06db299c37a8..9e4a20df6409 100644
> > > --- a/arch/arm64/kvm/hyp/nvhe/hyp-main.c
> > > +++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
> > > @@ -129,9 +129,14 @@ static void flush_hyp_vcpu(struct pkvm_hyp_vcpu *hyp_vcpu)
> > >       hyp_vcpu->vcpu.arch.ctxt        = host_vcpu->arch.ctxt;
> > >
> > >       hyp_vcpu->vcpu.arch.mdcr_el2    = host_vcpu->arch.mdcr_el2;
> > > -     hyp_vcpu->vcpu.arch.hcr_el2 &= ~(HCR_TWI | HCR_TWE);
> > > +     /*
> > > +      * HCR_EL2.VSE is host-owned (a pending virtual SError to inject), not a
> > > +      * trap-control bit, so it must flow to the hyp vCPU alongside TWI/TWE
> > > +      * for the vSError to be delivered. sync_hyp_vcpu() reflects it back.
> > > +      */
> > > +     hyp_vcpu->vcpu.arch.hcr_el2 &= ~(HCR_TWI | HCR_TWE | HCR_VSE);
> > >       hyp_vcpu->vcpu.arch.hcr_el2 |= READ_ONCE(host_vcpu->arch.hcr_el2) &
> > > -                                              (HCR_TWI | HCR_TWE);
> > > +                                              (HCR_TWI | HCR_TWE | HCR_VSE);
> > >
> > >       hyp_vcpu->vcpu.arch.iflags      = host_vcpu->arch.iflags;
> > >
> > > --
> > > 2.54.0.929.g9b7fa37559-goog
> > >