[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <c91c5ec9-a917-6d20-1401-c11e27d16f90@xen.org>
Date: Mon, 18 Sep 2023 13:18:32 +0100
From: Paul Durrant <xadimgnik@...il.com>
To: David Woodhouse <dwmw2@...radead.org>, kvm@...r.kernel.org,
linux-kernel@...r.kernel.org
Cc: Paul Durrant <pdurrant@...zon.com>,
Sean Christopherson <seanjc@...gle.com>,
Paolo Bonzini <pbonzini@...hat.com>
Subject: Re: [PATCH v2 09/12] KVM: selftests / xen: set
KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID
On 18/09/2023 12:59, David Woodhouse wrote:
> On Mon, 2023-09-18 at 11:21 +0000, Paul Durrant wrote:
>> From: Paul Durrant <pdurrant@...zon.com>
>>
>> If the capability (KVM_XEN_HVM_CONFIG_EVTCHN_SEND) is present then set
>> the guest's vCPU id to match the chosen vcpu_info offset.
>
> I think from KVM's point of view, the vcpu_id is still zero. As is the
> vcpu_idx. What you're setting is the *Xen* vcpu_id.
Ok. I'll clarify the terminology; and I'll need to set it before the
shinfo as noted on another thread.
>
> I like that it's *different* to the vcpu_id; we should definitely be
> testing that case.
How so?
> I don't quite know why the code was using
> vcpu_info[1] in the shinfo before when we were explicitly setting the
> address from userspace; I suppose it didn't matter.
>
Yes, I think it was entirely arbitrary.
>> Also make some cosmetic fixes to the code for clarity.
>>
>> Signed-off-by: Paul Durrant <pdurrant@...zon.com>
>> ---
>> Cc: Sean Christopherson <seanjc@...gle.com>
>> Cc: Paolo Bonzini <pbonzini@...hat.com>
>> Cc: David Woodhouse <dwmw2@...radead.org>
>>
>> v2:
>> - New in this version.
>> ---
>> .../selftests/kvm/x86_64/xen_shinfo_test.c | 19 +++++++++++++++----
>> 1 file changed, 15 insertions(+), 4 deletions(-)
>>
>> diff --git a/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c b/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
>> index 05898ad9f4d9..49d0c91ee078 100644
>> --- a/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
>> +++ b/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c
>> @@ -38,6 +38,8 @@
>> #define VCPU_INFO_VADDR (SHINFO_REGION_GVA + 0x40)
>> #define RUNSTATE_VADDR (SHINFO_REGION_GVA + PAGE_SIZE + PAGE_SIZE - 15)
>>
>> +#define VCPU_ID 1 /* Must correspond to offset of VCPU_INFO_[V]ADDR */
>>
>
> As well as being a bit clearer in the commit comment as noted above,
> let's call this XEN_VCPU_ID ?
>
Ok.
> With that cleaned up,
>
> Reviewed-by: David Woodhouse <dwmw@...zon.co.uk>
>
Thanks,
Paul
Powered by blists - more mailing lists