Re: [PATCH 01/18] Make TSC offset writes non-preemptible

[Avi Kivity <avi@redhat.com>]

On 07/13/2010 05:25 AM, Zachary Amsden wrote:
> Ensure that the storing of the offset and the reading of the TSC
> are never preempted by taking a spinlock.  While the lock is overkill
> now, it is useful later in this patch series.
>
>
>    

This patch has both code movement and semantic changes.  Please split 
into two so it's easier to understand which are which.

>   /*
> - * writes 'guest_tsc' into guest's timestamp counter "register"
> - * guest_tsc = host_tsc + tsc_offset ==>  tsc_offset = guest_tsc - host_tsc
> + * writes 'offset' into guest's timestamp counter offset register
>    */
> -static void guest_write_tsc(u64 guest_tsc, u64 host_tsc)
> +static void vmx_write_tsc_offset(struct kvm_vcpu *vcpu, u64 offset)
>    

Shouldn't offset be an s64?  Doesn't matter in practice, of course, but 
offset would usually be negative, not some huge number that happens to 
roll the guest tsc around.

>   {
> -	vmcs_write64(TSC_OFFSET, guest_tsc - host_tsc);
> +	vmcs_write64(TSC_OFFSET, offset);
>   }
>
>
>    

>
> +void guest_write_tsc(struct kvm_vcpu *vcpu, u64 data)
> +{
> +	struct kvm *kvm = vcpu->kvm;
> +	u64 offset;
> +
> +	spin_lock(&kvm->arch.tsc_write_lock);
> +	offset = data - native_read_tsc();
> +	kvm_x86_ops->write_tsc_offset(vcpu, offset);
> +	spin_unlock(&kvm->arch.tsc_write_lock);
> +
> +	/* Reset of TSC must disable overshoot protection below */
> +	vcpu->arch.hv_clock.tsc_timestamp = 0;
> +}
> +EXPORT_SYMBOL_GPL(guest_write_tsc);
> +
>    

Exported symbols should start with kvm_.


-- 
error compiling committee.c: too many arguments to function