Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Feedback-ID: ifd894703:Fastmail
Content-Type: text/plain;
        charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 16.0 \(3731.500.231\))
Subject: Re: [PATCH 1/2] MIPS: Allow MIPS32R2 kernel to run on P5600 and M5150
From:   Jiaxun Yang <jiaxun.yang@flygoat.com>
In-Reply-To: <alpine.DEB.2.21.2305301152080.42601@angie.orcam.me.uk>
Date:   Tue, 30 May 2023 12:54:49 +0100
Cc:     Paul Cercueil <paul@crapouillou.net>,
        "linux-mips@vger.kernel.org" <linux-mips@vger.kernel.org>,
        linux-kernel@vger.kernel.org,
        Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Content-Transfer-Encoding: quoted-printable
Message-Id: <BCECE3BF-7A4D-43BF-8AA8-2232D974CD71@flygoat.com>
References: <20230529135245.4085-1-jiaxun.yang@flygoat.com>
 <alpine.DEB.2.21.2305300321520.25569@angie.orcam.me.uk>
 <684C1A48-C743-4045-AF12-B0846FCE8EBE@flygoat.com>
 <alpine.DEB.2.21.2305301152080.42601@angie.orcam.me.uk>
To:     "Maciej W. Rozycki" <macro@orcam.me.uk>
Precedence: bulk


> 2023=E5=B9=B45=E6=9C=8830=E6=97=A5 12:07=EF=BC=8CMaciej W. Rozycki =
<macro@orcam.me.uk> =E5=86=99=E9=81=93=EF=BC=9A
>=20
> On Tue, 30 May 2023, Jiaxun Yang wrote:
>=20
>>>> M5150 and P5600 are two MIPS32R5 kernels, however as MIPS32R5 is
>>>> backward compatible with MIPS32R2 there is no reason to forbid
>>>> M5150 and P5600 on MIPS32R2 kernel.
>>>=20
>>> What problem are you trying to solve?  The CONFIG_SYS_HAS_CPU_* =
settings=20
>>> denote overall platform's support for the given CPU and have nothing =
to do=20
>>> with what architecture level a given kernel has been configured for. =
 You=20
>>> do need to get the settings right for your platform, just as you do =
in=20
>>> 2/2, but this 1/2 part looks wrong to me.
>>=20
>> Well the universal target is to allow R2 generic kernel to run on R5 =
CPUs.
>> As R5 is backward compatible we can just have one universal kernel =
binary.
>=20
> Sure, but this change is not needed for it.  You just need to declare=20=

> which ISA revisions your platform supports and leave `__get_cpu_type'=20=

> alone.  It has worked like that for a decade now.

I=E2=80=99m afraid it won=E2=80=99t work as you expected.

Actually I ran into a problem that `case CPU_P5600` in c-r4k.c is =
optimised out
by compiler, because the codepath is marked as unreachable.

Thanks
- Jiaxun

>=20
> Back in the day I used to run R1 kernels on R2 hardware myself.  And=20=

> maybe MIPS IV on R1 even, as we had MIPS Malta CPU modules with both =
MIPS=20
> IV devices (QED RM5261/RM7061) and MIPS64r1 devices (MIPS =
5Kc/20Kc/25Kf)=20
> and switching the kernel when swapping modules was a nuisance.  The =
Malta=20
> config still supports these devices although some may not exist =
anymore.
>=20
>  Maciej