Re: [RFC PATCH v3 2/5] ppc64: Fix semihosting on ppc64le

From: "Alex Bennée" <alex.bennee@linaro.org>
To: Peter Maydell <peter.maydell@linaro.org>
Cc: Leandro Lupori <leandro.lupori@eldorado.org.br>,
	danielhb413@gmail.com,
	Richard Henderson <richard.henderson@linaro.org>,
	qemu-devel@nongnu.org, groug@kaod.org, qemu-ppc@nongnu.org,
	clg@kaod.org, pbonzini@redhat.com, david@gibson.dropbear.id.au
Subject: Re: [RFC PATCH v3 2/5] ppc64: Fix semihosting on ppc64le
Date: Thu, 21 Apr 2022 10:43:25 +0100	[thread overview]
Message-ID: <87r15qvkhi.fsf@linaro.org> (raw)
In-Reply-To: <CAFEAcA9HtPAydeC+C5fbCQCFxjhnV9_tbncRFsH51Mf2MeHQyw@mail.gmail.com>

Peter Maydell <peter.maydell@linaro.org> writes:

> On Wed, 20 Apr 2022 at 20:52, Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> On 4/20/22 12:42, Peter Maydell wrote:
>> > On Mon, 18 Apr 2022 at 20:19, Leandro Lupori
>> > <leandro.lupori@eldorado.org.br> wrote:
>> >>
>> >> PPC64 CPUs can change its endian dynamically, so semihosting code
>> >> must check its MSR at run time to determine if byte swapping is
>> >> needed.
>> >
>> > Arm CPUs also change endianness dynamically, so why is this
>> > change PPC-specific ?
>>
>> I'm reasonably certain that we simply don't test armbe or aarch64_be semihosting.  Leandro
>> found this because qemu-system-ppc64 defaults to BE and qemu-system-aarch64 defaults to LE.
>
> Right, so if there's an existing bug here on arm we should fix that,
> and that probably means that the abstraction split between
> "arch-specific thing" and "non-arch-specific code" is different
> from "PPC just overrides the entire swap function".

I think the helper function cpu_virtio_is_big_endian is really just a
proxy for the data endianess mode of the guest. Perhaps it could be
re-named and then used by the semihosting code?

-- 
Alex Bennée