On Wed, Jan 12, 2022, Li RongQing wrote:
> After support paravirtualized TLB shootdowns, steal_time.preempted
> includes not only KVM_VCPU_PREEMPTED, but also KVM_VCPU_FLUSH_TLB
>
> and kvm_vcpu_is_preempted should test only with KVM_VCPU_PREEMPTED
>
> Fixes: 858a43aae2367 ("KVM: X86: use paravirtualized TLB Shootdown")
> Signed-off-by: Li RongQing <lirongqing@xxxxxxxxx>
> ---
> arch/x86/kernel/kvm.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
> index 59abbda..a9202d9 100644
> --- a/arch/x86/kernel/kvm.c
> +++ b/arch/x86/kernel/kvm.c
> @@ -1025,8 +1025,8 @@ asm(
> ".type __raw_callee_save___kvm_vcpu_is_preempted, @function;"
> "__raw_callee_save___kvm_vcpu_is_preempted:"
> "movq __per_cpu_offset(,%rdi,8), %rax;"
> -"cmpb $0, " __stringify(KVM_STEAL_TIME_preempted) "+steal_time(%rax);"
> -"setne %al;"
> +"movb " __stringify(KVM_STEAL_TIME_preempted) "+steal_time(%rax), %al;"
> +"andb $" __stringify(KVM_VCPU_PREEMPTED) ", %al;"
Eww, the existing code is sketchy. It relies on the compiler to store _Bool/bool
in a single byte since %rax may be non-zero from the __per_cpu_offset(), and
modifying %al doesn't zero %rax[63:8]. I doubt gcc or clang use anything but a
single byte on x86-64, but "andl" is just as cheap so I don't see any harm in
being paranoid.
> "ret;"
> ".size __raw_callee_save___kvm_vcpu_is_preempted, .-__raw_callee_save___kvm_vcpu_is_preempted;"
> ".popsection");
> --
> 2.9.4
>
