On 13/12/12 21:52, Christoffer Dall wrote:
>
>
> On Thursday, December 6, 2012, Marc Zyngier wrote:
>
> Move the "unlikely" statement to indicate that it is the
> "not initialized" state of the vgic that is unlikely.
>
> Not that it does anything on ARM...
>
> Signed-off-by: Marc Zyngier <marc.zyngier@xxxxxxx <javascript:;>>
> ---
> arch/arm/kvm/arm.c | 9 ++++++---
> 1 file changed, 6 insertions(+), 3 deletions(-)
>
> diff --git a/arch/arm/kvm/arm.c b/arch/arm/kvm/arm.c
> index fea8fbb..b307d27 100644
> --- a/arch/arm/kvm/arm.c
> +++ b/arch/arm/kvm/arm.c
> @@ -661,9 +661,12 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu
> *vcpu, struct kvm_run *run)
> if (unlikely(vcpu->arch.target < 0))
> return -ENOEXEC;
>
> - /* Initalize the VGIC before running a vcpu the first time
> on this VM */
> - if (unlikely(irqchip_in_kernel(vcpu->kvm) &&
> - !vgic_initialized(vcpu->kvm))) {
> + /*
> + * Initialize the VGIC before running a vcpu the first time on
> + * this VM
>
>
> super nit: this is a sentence that could be followed by a dot '.' :)
>
>
> + */
> + if (irqchip_in_kernel(vcpu->kvm) &&
> + unlikely(!vgic_initialized(vcpu->kvm))) {
> ret = kvm_vgic_init(vcpu->kvm);
> if (ret)
> return ret;
>
>
> hmm, could we move this check to kvm_arch_vcpu_load to reduce the number
> of branches we have at each vmexit?
This is not in the main loop, but just before. And there is a
kvm_vcpu_load() call for each of the kvm_arch_vcpu_ioctl_run() calls. So
I'm afraid it wouldn't change a thing.
>
> --
> 1.8.0.1
>
>
>
> _______________________________________________
> kvmarm mailing list
> kvmarm@xxxxxxxxxxxxxxxxxxxxx <javascript:;>
> https://lists.cs.columbia.edu/cucslists/listinfo/kvmarm
>
--
Jazz is not dead. It just smells funny...
_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/cucslists/listinfo/kvmarm
