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;
Message-ID: <fc0016b8c3d08fc139356c39668c6ab5d8c297fd.camel@gmail.com>
Subject: Re: [PATCH 1/1] MIPS: Add new Kconfig variable to avoid unaligned
 access instructions
From:   Yasha Cherikovsky <yasha.che3@gmail.com>
To:     Paul Burton <paul.burton@mips.com>
Cc:     Ralf Baechle <ralf@linux-mips.org>,
        James Hogan <jhogan@kernel.org>,
        "linux-mips@linux-mips.org" <linux-mips@linux-mips.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Date:   Tue, 25 Sep 2018 22:30:52 +0300
In-Reply-To: <20180925174510.rg3j4lfmwvlecnqt@pburton-laptop>
References: <20180920170306.9157-1-yasha.che3@gmail.com>
         <20180920170306.9157-2-yasha.che3@gmail.com>
         <20180925174510.rg3j4lfmwvlecnqt@pburton-laptop>
Content-Type: text/plain; charset="UTF-8"
User-Agent: Evolution 3.30.1 
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Hi Paul,

On Tue, 2018-09-25 at 17:45 +0000, Paul Burton wrote:
> Hi Yasha,
> 
> On Thu, Sep 20, 2018 at 08:03:06PM +0300, Yasha Cherikovsky wrote:
> > MIPSR6 doesn't support unaligned access instructions (lwl, lwr,
> > swl, swr).
> > The MIPS tree has some special cases to avoid these instructions,
> > and currently the code is testing for CONFIG_CPU_MIPSR6.
> > 
> > Declare a new Kconfig variable:
> > CONFIG_CPU_HAS_NO_UNALIGNED_LOAD_STORE,
> > and make CONFIG_CPU_MIPSR6 select it.
> > And switch all the special cases to test for the new variable.
> > 
> > Also, the new variable selects CONFIG_GENERIC_CSUM, to use
> > generic C checksum code (to avoid special assembly code that uses
> > the unsupported instructions).
> 
> Thanks for your patch :)
> 
> I think it would be cleaner to invert this logic & instead have the
> Kconfig entry indicate when kernel's build target *does* support the
> [ls]w[lr] instructions.
> 
> It would be good for the name to be clear that these instructions are
> what it's about too - "unaligned load store" is a little too vague
> for
> my liking. For example one could easily misconstrue it to mean
> something
> akin to the inverse of CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS,
> whereas
> in the MIPSr6 case many CPUs actually handle unaligned accesses in
> hardware when using the regular load/store instructions. They don't
> have
> the [ls]w[lr] instructions, but they don't need them because they
> handle
> unaligned accesses more naturally without needing us to be explicit
> about them.
> 
> How about we:
> 
>   - Add a Kconfig option CONFIG_CPU_SUPPORTS_LOAD_STORE_LR, and
> select
>     it for all existing pre-r6 targets (probably from CONFIG_CPU_*).
> 
>   - Change CONFIG_GENERIC_CSUM to default y if
>     !CONFIG_CPU_SUPPORTS_LOAD_STORE_LR, and drop the selects of it.
> 
> That would avoid the double-negative ("if we don't not support this")
> that the #ifndef's currently represent. It would also mean any future
> architecture/ISA targets beyond MIPSr6 automatically avoid the
> instructions.
> 
> Thanks,
>     Paul

Thanks for your feedback, I'll start preparing v2.

Looking in arch/mips/Kconfig, some CPU options start
with CPU_SUPPORTS_ and some with CPU_HAS_.
Which perfix should we use here?

Thanks,
Yasha