Hi Rob,
On 2014-03-24 16:04, Rob Herring wrote:
On Fri, Mar 21, 2014 at 10:20 PM, Christoffer Dall
<christoffer.dall@xxxxxxxxxx> wrote:
Does anyone have anything they would like to discuss on a KVM/ARM
Call
on Tuesday?
I don't know if this warrants a call, but Peter Maydell had raised an
issue about the smc vs. hvc calling convention for PSCI. The usecase
is doing migration between a KVM guest and QEMU TCG (system emulation
on x86). Seems a bit crazy to me, but people are trying to do this.
Crazy? Nahh... Just mad. The ambulance will be with you shortly. ;-)
The OS will pick the smc/hvc call method from DT based on whichever
environment it boots on, but then after migration will be using the
"wrong" call method.
Knowing little about KVM internals, the solution I see here is for
KVM
to support PSCI calls using smc. This will also mean we don't have a
Only one (major) nit: if the new host doesn't implement EL3 (like the
Applied SoC), an SMC instruction will undef at EL1. Not what you
expected.
different DT for virt machine model depending on KVM vs. TCG. Another
approach would be the TCG PSCI emulation (which I'm implementing)
could trap and handle hvc calls instead, but I don't recall if the
architecture can support that as easily.
To what exception level do you expect to trap? If your PSCI code
conceptually sits at EL2, that's fine. If you're at EL3, you're in
trouble.
M.
--
Fast, cheap, reliable. Pick two.
_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
