[GIT PULL] KVM/arm64 fixes for 7.1, take #5

[GIT PULL] KVM/arm64 fixes for 7.1, take #5

[Marc Zyngier]

Paolo,

Yes, I said that last week's pull request was the last. I really meant
it, honest. But people (and their AI best mate) keep finding ugly
stuff in dark corners, and I can't sit on this stuff much longer.

So here we have an assorted mess of bad races, broken userspace, and
architectural bugs. What's not to like? As usual, details in the tag
below.

Please pull,

	M.

The following changes since commit 83726330748981372bde86ed5411d7b306612991:

  KVM: arm64: Correctly cap ZCR_EL2 provided by a guest hypervisor (2026-05-29 10:04:00 +0100)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/kvmarm/kvmarm.git tags/kvmarm-fixes-7.1-5

for you to fetch changes up to 17f073f78fc43280891ecde8f8ec3f84f98bb37c:

  KVM: arm64: Correctly identify executable PTEs at stage-2 (2026-06-05 14:07:57 +0100)

----------------------------------------------------------------
KVM/arm64 fixes for 7.1, take #5

- Correctly drop the ITS translation cache reference when it actually
  gets invalidated

- Take the SRCU lock for SW page table walks

- Restore POR_EL0 access to host EL0, avoiding POR_EL0 becoming
  inaccessible from EL0 after running a guest

- Reassign nested_mmus array behind mmu_lock, ensuring that vcpu init
  and MMU notifiers are mutually exclusive

- Correctly handle FEAT_XNX at stage-2

----------------------------------------------------------------
Hyunwoo Kim (3):
      KVM: arm64: vgic-its: Drop the translation cache reference only for the erased entry
      KVM: arm64: Take the SRCU lock for page table walks in fault injection and AT emulation
      KVM: arm64: Reassign nested_mmus array behind mmu_lock

Joey Gouly (1):
      KVM: arm64: Restore POR_EL0 access to host EL0

Oliver Upton (2):
      KVM: arm64: nv: Fix handling of XN[0] when !FEAT_XNX
      KVM: arm64: Correctly identify executable PTEs at stage-2

 arch/arm64/include/asm/kvm_nested.h     |  4 ++--
 arch/arm64/kvm/at.c                     |  6 ++++--
 arch/arm64/kvm/hyp/include/hyp/switch.h |  2 ++
 arch/arm64/kvm/hyp/pgtable.c            |  4 +++-
 arch/arm64/kvm/nested.c                 | 33 ++++++++++++++++++++-------------
 arch/arm64/kvm/vgic/vgic-its.c          |  6 ++++--
 6 files changed, 35 insertions(+), 20 deletions(-)

Re: [GIT PULL] KVM/arm64 fixes for 7.1, take #5

[Paolo Bonzini]

On Fri, Jun 5, 2026 at 6:45 PM Marc Zyngier <maz@kernel.org> wrote:
>
> Paolo,
>
> Yes, I said that last week's pull request was the last. I really meant
> it, honest. But people (and their AI best mate) keep finding ugly
> stuff in dark corners, and I can't sit on this stuff much longer.
>
> So here we have an assorted mess of bad races, broken userspace, and
> architectural bugs. What's not to like? As usual, details in the tag
> below.

x86 isn't doing any better, so who am I to judge.

Pulled, thanks.

Paolo

>
> Please pull,
>
>         M.
>
> The following changes since commit 83726330748981372bde86ed5411d7b306612991:
>
>   KVM: arm64: Correctly cap ZCR_EL2 provided by a guest hypervisor (2026-05-29 10:04:00 +0100)
>
> are available in the Git repository at:
>
>   git://git.kernel.org/pub/scm/linux/kernel/git/kvmarm/kvmarm.git tags/kvmarm-fixes-7.1-5
>
> for you to fetch changes up to 17f073f78fc43280891ecde8f8ec3f84f98bb37c:
>
>   KVM: arm64: Correctly identify executable PTEs at stage-2 (2026-06-05 14:07:57 +0100)
>
> ----------------------------------------------------------------
> KVM/arm64 fixes for 7.1, take #5
>
> - Correctly drop the ITS translation cache reference when it actually
>   gets invalidated
>
> - Take the SRCU lock for SW page table walks
>
> - Restore POR_EL0 access to host EL0, avoiding POR_EL0 becoming
>   inaccessible from EL0 after running a guest
>
> - Reassign nested_mmus array behind mmu_lock, ensuring that vcpu init
>   and MMU notifiers are mutually exclusive
>
> - Correctly handle FEAT_XNX at stage-2
>
> ----------------------------------------------------------------
> Hyunwoo Kim (3):
>       KVM: arm64: vgic-its: Drop the translation cache reference only for the erased entry
>       KVM: arm64: Take the SRCU lock for page table walks in fault injection and AT emulation
>       KVM: arm64: Reassign nested_mmus array behind mmu_lock
>
> Joey Gouly (1):
>       KVM: arm64: Restore POR_EL0 access to host EL0
>
> Oliver Upton (2):
>       KVM: arm64: nv: Fix handling of XN[0] when !FEAT_XNX
>       KVM: arm64: Correctly identify executable PTEs at stage-2
>
>  arch/arm64/include/asm/kvm_nested.h     |  4 ++--
>  arch/arm64/kvm/at.c                     |  6 ++++--
>  arch/arm64/kvm/hyp/include/hyp/switch.h |  2 ++
>  arch/arm64/kvm/hyp/pgtable.c            |  4 +++-
>  arch/arm64/kvm/nested.c                 | 33 ++++++++++++++++++++-------------
>  arch/arm64/kvm/vgic/vgic-its.c          |  6 ++++--
>  6 files changed, 35 insertions(+), 20 deletions(-)
>