On Mar 6, 2012, at 10:28 PM, Rusty Russell <rusty at rustcorp.com.au> wrote:
> On Tue, 6 Mar 2012 10:53:16 +0000, Peter Maydell <peter.maydell at linaro.org> wrote:
>> On 5 March 2012 04:21, Rusty Russell <rusty at rustcorp.com.au> wrote:
>>> diff --git a/linux-headers/asm-arm/kvm.h b/linux-headers/asm-arm/kvm.h
>>> index ff88ca0..8caa83b 100644
>>> --- a/linux-headers/asm-arm/kvm.h
>>> +++ b/linux-headers/asm-arm/kvm.h
>>> @@ -55,6 +59,7 @@ struct kvm_regs {
>>> __u32 c1_sys;
>>> __u32 c2_base0;
>>> __u32 c2_base1;
>>> + __u32 c2_control;
>>> __u32 c3_dacr;
>>> } cp15;
>>
>> Which kernel tree is this in? I checked Christoffer's kvm-a15-patches
>> branch and Marc's kvm-v3.3-rc1 and kvm-v6-v3.3-rc3 branches and none
>> of those seem to have this extra field.
>
> This is in Christoffer's kvm-a15-v6 branch.
>
>> We may not have frozen the ABI here yet but I'm starting to think
>> we need to be a bit more careful about coordinating updates to it
>> between kernel trees: QEMU can't track more than one of them...
>
> Yes, I wouldn't take this yet, it doesn't appear anywhere else. I had
> to revert it to test Marc's tree, for example.
>
>> (Or for this particular case just move to MANY_REGS for accessing
>> the cp15 regs and then we can just ignore missing regs at runtime
>> rather than having compile failures or breakage.)
>
> Or we could complete the structure with all known cp15 registers, as I
> think we suggested we would?
>
> I'm happy to patch it like that, if there's agreement...
>
Sounds fine with me.
-Christoffer
