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 v4] mips: add <asm-generic/io.h> including
From:   Jiaxun Yang <jiaxun.yang@flygoat.com>
In-Reply-To: <dae342ed-8999-4fa5-b719-322182580025@app.fastmail.com>
Date:   Fri, 19 May 2023 22:41:31 +0100
Cc:     "linux-mips@vger.kernel.org" <linux-mips@vger.kernel.org>,
        linux-kernel@vger.kernel.org,
        Thomas Bogendoerfer <tsbogend@alpha.franken.de>,
        Linux-Arch <linux-arch@vger.kernel.org>,
        Baoquan He <bhe@redhat.com>,
        Huacai Chen <chenhuacai@kernel.org>
Content-Transfer-Encoding: quoted-printable
Message-Id: <07FC59A6-862A-41E3-A8A9-545A6D0A42FE@flygoat.com>
References: <20230519195135.79600-1-jiaxun.yang@flygoat.com>
 <dae342ed-8999-4fa5-b719-322182580025@app.fastmail.com>
To:     Arnd Bergmann <arnd@arndb.de>
Precedence: bulk


> 2023=E5=B9=B45=E6=9C=8819=E6=97=A5 22:05=EF=BC=8CArnd Bergmann =
<arnd@arndb.de> =E5=86=99=E9=81=93=EF=BC=9A
>=20
> On Fri, May 19, 2023, at 21:51, Jiaxun Yang wrote:
>> With the adding, some default ioremap_xx methods defined in
>> asm-generic/io.h can be used. E.g the default ioremap_uc() returning
>> NULL.
>>=20
>> We also massaged various headers to avoid nested includes.
>>=20
>> Signed-off-by: Baoquan He <bhe@redhat.com>
>> Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
>> [jiaxun.yang@flygoat.com: Massage more headers, fix ioport defines]
>> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
>> Cc: Huacai Chen <chenhuacai@kernel.org>
>> Cc: Jiaxun Yang <jiaxun.yang@flygoat.com>
>> Cc: linux-mips@vger.kernel.org
>> ---
>> Tested against qemu malta 34Kf, boston I6500, Loongson64, hopefully
>> everything is fine now.
>=20
> Thanks a lot for figuring this out!
>>=20
>> @@ -44,6 +42,11 @@
>> # define __raw_ioswabq(a, x) (x)
>> # define ____raw_ioswabq(a, x) (x)
>>=20
>> +# define _ioswabb ioswabb
>> +# define _ioswabw ioswabw
>> +# define _ioswabl ioswabl
>> +# define _ioswabq ioswabq
>> +
>=20
> I'm missing something here, what are these macros used for in addition
> to the non-underscore versions?

Since now {in,out}{bwlq} have a `_` in prefix, it needs a corresponding
ioswab variant to work.

>=20
>> +#define memset_io memset_io
>> static inline void memset_io(volatile void __iomem *addr, unsigned=20
>> char val, int count)
>> {
>> memset((void __force *) addr, val, count);
>> }
>> +#define memcpy_fromio memcpy_fromio
>> static inline void memcpy_fromio(void *dst, const volatile void=20
>> __iomem *src, int count)
>> {
>> memcpy(dst, (void __force *) src, count);
>> }
>> +#define memcpy_toio memcpy_toio
>> static inline void memcpy_toio(volatile void __iomem *dst, const void=20=

>> *src, int count)
>> {
>> memcpy((void __force *) dst, src, count);
>=20
> These three could probably go away now, as they are identical
> to the asm-generic version. Not important though.

Will update in a follow-up patch, I=E2=80=99m planing to do some cleanup =
for those
headers later.

>=20
>> @@ -549,6 +555,47 @@ extern void (*_dma_cache_inv)(unsigned long =
start,=20
>> unsigned long size);
>> #define csr_out32(v, a) (*(volatile u32 *)((unsigned long)(a) +=20
>> __CSR_32_ADJUST) =3D (v))
>> #define csr_in32(a)    (*(volatile u32 *)((unsigned long)(a) +=20
>> __CSR_32_ADJUST))
>>=20
>> +
>> +#define __raw_readb __raw_readb
>> +#define __raw_readw __raw_readw
>> +#define __raw_readl __raw_readl
>> +#define __raw_readq __raw_readq
>> +#define __raw_writeb __raw_writeb
>> +#define __raw_writew __raw_writew
>> +#define __raw_writel __raw_writel
>> +#define __raw_writeq __raw_writeq
>> +
>> +#define readb readb
>> +#define readw readw
>> +#define readl readl
>> +#define writeb writeb
>> +#define writew writew
>> +#define writel writel
>> +
>> +#define readsb readsb
>> +#define readsw readsw
>> +#define readsl readsl
>> +#define readsq readsq
>> +#define writesb writesb
>> +#define writesw writesw
>> +#define writesl writesl
>> +#define writesq writesq
>=20
> As far as I can tell, the readsq()/writesq() helpers are currently
> only defined on 64-bit, it's probably best to leave it that way.

Will opt-out them on 32 bit build.

>=20
> On most other architectures, we also don't define __raw_readq()
> and __raw_writeq() on 32-bit because they lose the atomicity that
> might be required for FIFO accesses, but the existing MIPS version
> has them, so changing those should be a separate patch after it
> can be shown to not break anything.

Current implementation of __raw_readq __raw_readq  will check for =
CPU=E2=80=99s
64bit support and do a 64bit access if possible, it=E2=80=99s helpful =
for running
32 bit kernel on 64 bit CPUs.

Thanks
- Jiaxun

>=20
>     Arnd