Subject: Re: [PATCH v5 01/17] asm-generic/iomap.h: remove ARCH_HAS_IOREMAP_xx
 macros
To:     Baoquan He <bhe@redhat.com>, linux-kernel@vger.kernel.org
Cc:     linux-arch@vger.kernel.org, linux-mm@kvack.org, arnd@arndb.de,
        christophe.leroy@csgroup.eu, hch@infradead.org,
        agordeev@linux.ibm.com, wangkefeng.wang@huawei.com,
        schnelle@linux.ibm.com, David.Laight@ACULAB.COM, shorne@gmail.com,
        willy@infradead.org, loongarch@lists.linux.dev,
        linux-m68k@lists.linux-m68k.org, linux-mips@vger.kernel.org,
        linuxppc-dev@lists.ozlabs.org, x86@kernel.org,
        netdev@vger.kernel.org, Martin Habets <habetsm.xilinx@gmail.com>
References: <20230301034247.136007-1-bhe@redhat.com>
 <20230301034247.136007-2-bhe@redhat.com>
From:   Edward Cree <ecree.xilinx@gmail.com>
Message-ID: <7bd6db48-ffb1-7eb1-decf-afa8be032970@gmail.com>
Date:   Wed, 1 Mar 2023 04:38:10 +0000
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.14.0
MIME-Version: 1.0
In-Reply-To: <20230301034247.136007-2-bhe@redhat.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-GB
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 01/03/2023 03:42, Baoquan He wrote:
> diff --git a/drivers/net/ethernet/sfc/io.h b/drivers/net/ethernet/sfc/io.h
> index 30439cc83a89..07f99ad14bf3 100644
> --- a/drivers/net/ethernet/sfc/io.h
> +++ b/drivers/net/ethernet/sfc/io.h
> @@ -70,7 +70,7 @@
>   */
>  #ifdef CONFIG_X86_64
>  /* PIO is a win only if write-combining is possible */
> -#ifdef ARCH_HAS_IOREMAP_WC
> +#ifdef ioremap_wc
>  #define EFX_USE_PIO 1
>  #endif
>  #endif

So I don't know how valid what we're doing here is...

> diff --git a/include/asm-generic/iomap.h b/include/asm-generic/iomap.h
> index 08237ae8b840..196087a8126e 100644
> --- a/include/asm-generic/iomap.h
> +++ b/include/asm-generic/iomap.h
> @@ -93,15 +93,15 @@ extern void __iomem *ioport_map(unsigned long port, unsigned int nr);
>  extern void ioport_unmap(void __iomem *);
>  #endif
>  
> -#ifndef ARCH_HAS_IOREMAP_WC
> +#ifndef ioremap_wc
>  #define ioremap_wc ioremap
>  #endif

... but it looks like this will break it, since in sfc/io.h
 `#ifdef ioremap_wc` will always be true (if I'm correctly
 understanding what we get via #include <linux/io.h>, which I'm
 probably not because asm includes always confuse me).
I.e. we're not just interested in "can code that calls ioremap_wc
 compile?", we care about whether we actually get WC, because
 we're making an optimisation decision based on it.