Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Fri, 24 Jan 2020 15:07:51 +0100
From:   Thomas Bogendoerfer <tsbogend@alpha.franken.de>
To:     Paul Burton <paulburton@kernel.org>
Cc:     Jiaxun Yang <jiaxun.yang@flygoat.com>, linux-mips@vger.kernel.org,
        chenhc@lemote.com, paul.burton@mips.com,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH] MIPS: Introduce aligned IO memory operations
Message-ID: <20200124140751.GA17030@alpha.franken.de>
References: <20200114122343.163685-1-jiaxun.yang@flygoat.com>
 <20200122184506.7zbzetn5xturxamj@pburton-laptop>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20200122184506.7zbzetn5xturxamj@pburton-laptop>
User-Agent: Mutt/1.5.23 (2014-03-12)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Jan 22, 2020 at 10:45:06AM -0800, Paul Burton wrote:
> Hi Jiaxun,
> 
> On Tue, Jan 14, 2020 at 08:23:43PM +0800, Jiaxun Yang wrote:
> > Some platforms, such as Loongson64 or QEMU/KVM, don't support unaligned
> > instructions like lwl or lwr in IO memory access. However, our current
> > IO memcpy/memset is wired to the generic implementation, which leads
> > to a fatal result.
> 
> Hmm, I wonder if we should just do this unconditionally on all systems.
> I can't think of a reason it'd ever be a good idea to use lwl/lwr on an
> MMIO device. Any thoughts on that?

depends on the type of device. I can see benefits for framebuffers
and memory devices since memset/memcpy are more optimised than the
function in this patch.

Thomas.

-- 
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]