Received-SPF: pass (google.com: domain of linux-kernel+bounces-71481-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Precedence: bulk
Mime-Version: 1.0
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain; charset=UTF-8
Date: Mon, 19 Feb 2024 15:31:45 +0100
Message-Id: <CZ94MU7BQIOE.2KYB4TWVOJISN@bootlin.com>
Subject: Re: [PATCH 07/13] i2c: nomadik: support short xfer timeouts using
 waitqueue & hrtimer
Cc: "Andi Shyti" <andi.shyti@kernel.org>, "Rob Herring"
 <robh+dt@kernel.org>, "Krzysztof Kozlowski"
 <krzysztof.kozlowski+dt@linaro.org>, "Conor Dooley" <conor+dt@kernel.org>,
 "Thomas Bogendoerfer" <tsbogend@alpha.franken.de>,
 <linux-arm-kernel@lists.infradead.org>, <linux-i2c@vger.kernel.org>,
 <devicetree@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
 <linux-mips@vger.kernel.org>, "Gregory Clement"
 <gregory.clement@bootlin.com>, "Vladimir Kondratiev"
 <vladimir.kondratiev@mobileye.com>, "Thomas Petazzoni"
 <thomas.petazzoni@bootlin.com>, "Tawfik Bayouk"
 <tawfik.bayouk@mobileye.com>
To: "Linus Walleij" <linus.walleij@linaro.org>
From: =?utf-8?q?Th=C3=A9o_Lebrun?= <theo.lebrun@bootlin.com>
References: <20240215-mbly-i2c-v1-0-19a336e91dca@bootlin.com>
 <20240215-mbly-i2c-v1-7-19a336e91dca@bootlin.com>
 <CACRpkdYC=vVBA-s6GmsaED=NdXfsr0JDzzF+x8q8C3tqQ0F8YQ@mail.gmail.com>
In-Reply-To: <CACRpkdYC=vVBA-s6GmsaED=NdXfsr0JDzzF+x8q8C3tqQ0F8YQ@mail.gmail.com>

Hello,

On Mon Feb 19, 2024 at 3:19 PM CET, Linus Walleij wrote:
> On Thu, Feb 15, 2024 at 5:52=E2=80=AFPM Th=C3=A9o Lebrun <theo.lebrun@boo=
tlin.com> wrote:
>
> > Replace the completion by a waitqueue for synchronization from IRQ
> > handler to task. For short timeouts, use hrtimers, else use timers.
> > Usecase: avoid blocking the I2C bus for too long when an issue occurs.
> >
> > The threshold picked is one jiffy: if timeout is below that, use
> > hrtimers. This threshold is NOT configurable.
> >
> > Implement behavior but do NOT change fetching of timeout. This means th=
e
> > timeout is unchanged (200ms) and the hrtimer case will never trigger.
> >
> > A waitqueue is used because it supports both desired timeout approaches=
.
> > See wait_event_timeout() and wait_event_hrtimeout(). An atomic boolean
> > serves as synchronization condition.
> >
> > Signed-off-by: Th=C3=A9o Lebrun <theo.lebrun@bootlin.com>
>
> Part of me want to go and fix completions to handle hrtimer timeouts
> for submicrosecond timeouts, BUT I realized that this is a bit thick
> request for a simple driver, so just a suggestion for something we could
> do one day. This is fine with me.
> Reviewed-by: Linus Walleij <linus.walleij@linaro.org>

Indeed having to switch to another abstraction because we desire another
timeout method is nonsensical. Completion supporting hrtimeouts would
make sense. As you said though, this is too much for a simple driver.

Thanks,

--
Th=C3=A9o Lebrun, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

------------------------------------------------------------------------