From 5a3d82dc8aced6e6a9d3f18b2211031d841ec41d Mon Sep 17 00:00:00 2001 From: Zenghui Yu Date: Tue, 1 Jun 2021 22:07:38 +0800 Subject: [PATCH] KVM: arm64: Resolve all pending PC updates before immediate exit commit e3e880bb1518eb10a4b4bb4344ed614d6856f190 upstream. Commit 26778aaa134a ("KVM: arm64: Commit pending PC adjustemnts before returning to userspace") fixed the PC updating issue by forcing an explicit synchronisation of the exception state on vcpu exit to userspace. However, we forgot to take into account the case where immediate_exit is set by userspace and KVM_RUN will exit immediately. Fix it by resolving all pending PC updates before returning to userspace. Since __kvm_adjust_pc() relies on a loaded vcpu context, I moved the immediate_exit checking right after vcpu_load(). We will get some overhead if immediate_exit is true (which should hopefully be rare). Fixes: 26778aaa134a ("KVM: arm64: Commit pending PC adjustemnts before returning to userspace") Signed-off-by: Zenghui Yu Signed-off-by: Marc Zyngier Link: https://lore.kernel.org/r/20210526141831.1662-1-yuzenghui@huawei.com Cc: stable@vger.kernel.org # 5.11 [yuz: stable-5.12.y backport] Signed-off-by: Zenghui Yu Reviewed-by: Marc Zyngier Signed-off-by: Greg Kroah-Hartman --- arch/arm64/kvm/arm.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c index c18740a1e5418..7730b81aad6d1 100644 --- a/arch/arm64/kvm/arm.c +++ b/arch/arm64/kvm/arm.c @@ -715,11 +715,13 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu) return ret; } - if (run->immediate_exit) - return -EINTR; - vcpu_load(vcpu); + if (run->immediate_exit) { + ret = -EINTR; + goto out; + } + kvm_sigset_activate(vcpu); ret = 1; @@ -892,6 +894,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu) kvm_sigset_deactivate(vcpu); +out: /* * In the unlikely event that we are returning to userspace * with pending exceptions or PC adjustment, commit these