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;
Date:   Tue, 23 Aug 2022 10:42:44 +0800
From:   Baoquan He <bhe@redhat.com>
To:     Christoph Hellwig <hch@infradead.org>
Cc:     linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        akpm@linux-foundation.org, agordeev@linux.ibm.com,
        wangkefeng.wang@huawei.com, linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH v2 03/11] mm: ioremap: allow ARCH to have its own ioremap
 definition
Message-ID: <YwQ+pBNMGuHq2s+V@MiWiFi-R3L-srv>
References: <20220820003125.353570-1-bhe@redhat.com>
 <20220820003125.353570-4-bhe@redhat.com>
 <YwHXWdycBZr+bluG@infradead.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <YwHXWdycBZr+bluG@infradead.org>
Precedence: bulk

On 08/20/22 at 11:57pm, Christoph Hellwig wrote:
> On Sat, Aug 20, 2022 at 08:31:17AM +0800, Baoquan He wrote:
> > Architectures like xtensa, arc, can be converted to GENERIC_IOREMAP,
> > to take standard ioremap_prot() and ioremap_xxx() way. But they have
> > ARCH specific handling for ioremap() method, than standard ioremap()
> > method.
> 
> Do they?
> 
> For arc, the arc_uncached_addr_space case can be easily handled by
> arch_ioremap, and the xtensa case looks very similar to that.

I am worried it will impact ioremap_prot(). Arc has selected
HAVE_IOREMAP_PROT in Kconfig. Putting arc_uncached_addr_space() calling
into arch_ioremap() will change ioremap_prot(), right?

And I have the same about xtensa. You can see ioremap() and
ioremap_cache() will return different value since they take
XCHAL_KIO_BYPASS_VADDR and XCHAL_KIO_CACHED_VADDR differently. I haven't
figured out a way to handle them in arch_ioremap() differently.

static inline void __iomem *ioremap(unsigned long offset, unsigned long size)
{
        if (offset >= XCHAL_KIO_PADDR
            && offset - XCHAL_KIO_PADDR < XCHAL_KIO_SIZE)
                return (void*)(offset-XCHAL_KIO_PADDR+XCHAL_KIO_BYPASS_VADDR);
        else
                return xtensa_ioremap_nocache(offset, size);
}

static inline void __iomem *ioremap_cache(unsigned long offset,
                unsigned long size)
{
        if (offset >= XCHAL_KIO_PADDR
            && offset - XCHAL_KIO_PADDR < XCHAL_KIO_SIZE)
                return (void*)(offset-XCHAL_KIO_PADDR+XCHAL_KIO_CACHED_VADDR);
        else
                return xtensa_ioremap_cache(offset, size);
}

> 
> I'd really like to kill off arch definitions of ioremap going
> forward, as they should just be a special case of ioremap_prot
> by definition.