[Please refrain from top-posting, it makes it harder to reply in
context...]
On Tue, 6 Nov 2012 01:32:51 +0000, "Sundaram, Senthilkumar"
<ssundara@xxxxxxxxxxxxxxxx> wrote:
> Still didn't get it. Are you saying that the load / store emulation
only
> applies to hardware that doesn't support Stage 2 translation /
> Virtualization Extensions?
What I'm trying to say it that load/store emulation is all about MMIO on
"hardware" that is emulated. This "hardware" have no entries in the state-2
translation page tables, thus any access traps into the hypervisor. This
basically covers *any* device but the GIC CPU interface and the generic
timers.
M.
> Thanks & Regards
> Senthil
>
>
> -----Original Message-----
> From: Marc Zyngier [mailto:marc.zyngier@xxxxxxx]
> Sent: Monday, November 05, 2012 11:18 PM
> To: Sundaram, Senthilkumar
> Cc: Christoffer Dall; kvmarm@xxxxxxxxxxxxxxxxxxxxx
> Subject: Re: [PATCH 4/5] KVM: ARM: Change instruction decoding
to
> use struct kvm_decode
>
> On 05/11/12 17:39, Sundaram, Senthilkumar wrote:
>> In the ARM V7 ARM, I don't see any provision for trapping Load Store
>> instructions in to Hyp mode( See section B.14). What am I missing?
>
> You're missing Stage-2 translation, or lack thereof.
>
> M.
--
Fast, cheap, reliable. Pick two.
_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/cucslists/listinfo/kvmarm
