On 2020-11-26 15:06, Alexandru Elisei wrote:
Hi Marc,
This patch looks correct to me, I checked in the Arm ARM DDI 0487F.b
and indeed
all accesses to the PMU registers are UNDEFINED if the PMU is not
present.
I checked all the accessors and now all the PMU registers that KVM
emulates will
inject an undefined exception if the VCPU feature isn't set. There's
one register
that we don't emulate, PMMIR_EL1, I suppose that's because it's part of
PMU
ARMv8.4 and KVM advertises ARMv8.1; if the guest tries to access it, it
will get
an undefined exception and KVM will print a warning in
emulate_sys_reg().
Funny that. I wrote a patch for that a long while ago, and obviously
never did anything with it [1]... Actually, the whole series was
silently
dropped. I guess I had other things to think about at the time!
Let me pick that up again.
Reviewed-by: Alexandru Elisei <alexandru.elisei@xxxxxxx>
Thanks!
M.
[1]
https://lore.kernel.org/kvmarm/20200216185324.32596-6-maz@xxxxxxxxxx/
--
Jazz is not dead. It just smells funny...
_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
