Re: [PATCH v12 00/10] hvf: Implement Apple Silicon Support

[Peter Maydell <peter.maydell@linaro.org>]

On Sat, 25 Sept 2021 at 18:22, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
> So this part of Xilinx Versal:
>
>      * When loading an OS, we turn on QEMU's PSCI implementation with SMC
>      * as the PSCI conduit. When there's no -kernel, we assume the user
>      * provides EL3 firmware to handle PSCI.
>      */
>     if (machine->kernel_filename) {
>         psci_conduit = QEMU_PSCI_CONDUIT_SMC;
>     }
>
> abuses the fact that -kernel command line expect a *Linux* kernel able
> to read the provided dtb which describe SMC. >This won't work if we
> manually provide a crafted dtb with another conduit

...which is not a supported thing to do (it won't work with the
'virt' board either)...

>, or if we pass any
> other binary (not Linux, not particularly ELF, since -kernel can load
> many formats).

-kernel means "assume I am a Linux kernel and boot me accordingly".
Sometimes it works for other things, but there are no guarantees.
Among other things it always means "start me at EL2 or EL1, not EL3",
so PSCI handling via SMC doesn't get in anybody's way even if they're
not using it. (People who want a pure "load this ELF file" should
use the generic-loader.)

The basic distinction the versal-virt board is making here is
"we have EL3 firmware to run in the guest" vs "we don't";
for the same reason that the virt board does, in the former
case we disable QEMU's internal PSCI implementation, and in
the latter we enable it. In theory the orangepi board code
should do the same, because if we're really running guest
code at EL3 it probably is going to assume it has SMC, and
QEMU taking control instead is going to confuse it.

-- PMM