[PATCH] arm64: KVM: Fix AArch64 guest userspace exception injection

[Andrew Jones]

On Tue, Jan 12, 2016 at 06:44:34PM +0000, Marc Zyngier wrote:
> On 12/01/16 18:23, Andrew Jones wrote:
> > Hi Marc,
> > 
> > Please shoot me if the following statement is false.
> 
> I wouldn't do that. Having had the privilege to waste 10 months of my
> life doing a military service, I quickly discovered I didn't like
> weapons nor those who carry them...
> 
> > Without this patch, if a guest that is running in, e.g. PSR_MODE_EL0t,
> > tries to do, e.g. 'smc #0', then KVM will inject an undef exception,
> > which should lead to the guest resuming at VBAR_EL1 + 0x400, but instead
> > it resumes at VBAR_EL1 + 0x200.
> 
> Not quite. SMC is undefined at EL0 (see C6.6.165), so it is not trapped
> to EL2, but to EL1. KVM is not in the loop at all in that case.
> 
> > Now, if you haven't started loading your gun to shoot me yet, then I'm
> > quite confused as to why the unit test[1] I wrote for this works just
> > fine without this patch.
> 
> If you want to exercise that path, you have to access something that
> wouldn't trap to EL1, but that EL2 traps. I don't think we have much
> stuff so far that can be used at EL0 and would be trapped to EL2,
> unfortunately (the PMU code is probably the first thing we'll merge).
>
> In the meantime, this test case is fairly pointless, I'm afraid...

Ah, thanks for the clarification, and the lack of desire to shoot me.
I'll stash this test case for a later day.

drew