Hi James,
     Thanks a lot for your comments.

2017-05-04 23:49 GMT+08:00 James Morse [off-list ref]:

Hi Dongjiu Geng,

On 02/05/17 16:29, James Morse wrote:

quoted

I think we need a new API for injecting SError for SEI from Qemu/kvmtool, but it
shouldn't be related to the RAS extensions. All v8.0 CPUs have HCR_EL2.VSE, so
we need to know KVM supports this API.

Thinking about this some more, it is slightly more nuanced, KVM can always
provide an API to inject SError, but it can only set the VSESR if the CPU has
the RAS Extensions.

James, do you mean we need to add a new API instead of adding the
VSESR in the old API kvm_inject_vabt?

 /**
  * kvm_inject_vabt - inject an async abort / SError into the guest
  * @vcpu: The VCPU to receive the exception
  *
  * It is assumed that this code is called from the VCPU thread and that the
  * VCPU therefore is not currently executing guest code.
  */
 void kvm_inject_vabt(struct kvm_vcpu *vcpu)
 {
     vcpu_set_hcr(vcpu, vcpu_get_hcr(vcpu) | HCR_VSE);
 }

Only offering the inject-SError API call if we can also set the VSESR looks a
bit funny, but no-one has needed the no-ESR version so far.

I still don't think we should let user-space make the 'RAS Extensions means
VSESR' logical step.

 Ok. got it.

So my comments on this patch become:
Don't read id registers directly, use cpus_have_cap() which handles features
that differ across CPUs or were turned off at compile time.

Please don't call this 'KVM_CAP_ARM_RAS_EXTENSION', if we are building an API to
inject SError, call it that instead.

Ok, thanks for your suggestion.

Thanks,

James
_______________________________________________
kvmarm mailing list
kvmarm at lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm