Re: [PATCH 1/2] LoongArch: KVM: Fix PC double advance in kernel MMIO read fast path

Re: [PATCH 1/2] LoongArch: KVM: Fix PC double advance in kernel MMIO read fast path

[Bibo Mao]

On 2026/7/23 下午7:15, Zeng Chi wrote:
> From: Zeng Chi <zengchi@kylinos.cn>
> 
> In the in-kernel MMIO read fast path of kvm_emu_mmio_read(),
> kvm_complete_mmio_read() already advances the guest PC via update_pc().
> The explicit update_pc() call right after it advances the PC a second
> time, so PC moves forward by 8 instead of 4 and the instruction
> following the MMIO read is silently skipped.
> 
> The user space MMIO read completion path in kvm_arch_vcpu_ioctl_run()
> calls kvm_complete_mmio_read() only once, and the MMIO write fast path
> advances the PC exactly once as well.
> 
> Remove the redundant update_pc() so the kernel MMIO read fast path
> advances the PC by a single instruction.
> 
> Fixes: 80edf90831a2 ("LoongArch: KVM: Add sign extension with kernel MMIO read emulation")
> Cc: stable@vger.kernel.org
> Signed-off-by: Zeng Chi <zengchi@kylinos.cn>
> ---
>   arch/loongarch/kvm/exit.c | 1 -
>   1 file changed, 1 deletion(-)
> 
> diff --git a/arch/loongarch/kvm/exit.c b/arch/loongarch/kvm/exit.c
> index 8572b63478bb..482ba17bcd3d 100644
> --- a/arch/loongarch/kvm/exit.c
> +++ b/arch/loongarch/kvm/exit.c
> @@ -481,7 +481,6 @@ int kvm_emu_mmio_read(struct kvm_vcpu *vcpu, larch_inst inst)
>   		srcu_read_unlock(&vcpu->kvm->srcu, idx);
>   		if (!ret) {
>   			kvm_complete_mmio_read(vcpu, run);
> -			update_pc(&vcpu->arch);
>   			vcpu->mmio_needed = 0;
>   			return EMULATE_DONE;
>   		}
> 
Hi Zeng,

Thanks for finding this problem, it is a big critical potential issue :(

Reviewed-by: Bibo Mao <maobibo@loongson.cn>

Re: [PATCH 1/2] LoongArch: KVM: Fix PC double advance in kernel MMIO read fast path

[Tao Cui]

From: Tao Cui <cuitao@kylinos.cn>

> In the in-kernel MMIO read fast path of kvm_emu_mmio_read(),
> kvm_complete_mmio_read() already advances the guest PC via update_pc().
> The explicit update_pc() call right after it advances the PC a second
> time, so PC moves forward by 8 instead of 4 and the instruction
> following the MMIO read is silently skipped.

Good catch!
Reviewed-by: Tao Cui <cuitao@kylinos.cn>