Received-SPF: pass (google.com: domain of linux-kernel+bounces-63800-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240212-mainline-spi-precook-message-v1-0-a2373cd72d36@baylibre.com>
 <20240212-mainline-spi-precook-message-v1-5-a2373cd72d36@baylibre.com> <c06dfa1ecf88b07ef467ad7c08667d0cab400613.camel@gmail.com>
In-Reply-To: <c06dfa1ecf88b07ef467ad7c08667d0cab400613.camel@gmail.com>
From: David Lechner <dlechner@baylibre.com>
Date: Tue, 13 Feb 2024 09:27:58 -0600
Message-ID: <CAMknhBEU=iMzpE_P0KePL4cZZktBOGHRXaEox5a7XcVjXDT+Dg@mail.gmail.com>
Subject: Re: [PATCH 5/5] iio: adc: ad7380: use spi_optimize_message()
To: =?UTF-8?B?TnVubyBTw6E=?= <noname.nuno@gmail.com>
Cc: Mark Brown <broonie@kernel.org>, Martin Sperl <kernel@martin.sperl.org>, 
	David Jander <david@protonic.nl>, Jonathan Cameron <jic23@kernel.org>, 
	Michael Hennerich <michael.hennerich@analog.com>, =?UTF-8?B?TnVubyBTw6E=?= <nuno.sa@analog.com>, 
	Alain Volmat <alain.volmat@foss.st.com>, Maxime Coquelin <mcoquelin.stm32@gmail.com>, 
	Alexandre Torgue <alexandre.torgue@foss.st.com>, linux-spi@vger.kernel.org, 
	linux-kernel@vger.kernel.org, linux-stm32@st-md-mailman.stormreply.com, 
	linux-arm-kernel@lists.infradead.org, linux-iio@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Tue, Feb 13, 2024 at 3:47=E2=80=AFAM Nuno S=C3=A1 <noname.nuno@gmail.com=
> wrote:
>
> On Mon, 2024-02-12 at 17:26 -0600, David Lechner wrote:
> > This modifies the ad7380 ADC driver to use spi_optimize_message() to
> > optimize the SPI message for the buffered read operation. Since buffere=
d
> > reads reuse the same SPI message for each read, this can improve
> > performance by reducing the overhead of setting up some parts the SPI
> > message in each spi_sync() call.
> >
> > Signed-off-by: David Lechner <dlechner@baylibre.com>
> > ---
> >  drivers/iio/adc/ad7380.c | 52 ++++++++++++++++++++++++++++++++++++++++=
+------
> > -
> >  1 file changed, 45 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/iio/adc/ad7380.c b/drivers/iio/adc/ad7380.c
> > index abd746aef868..5c5d2642a474 100644
> > --- a/drivers/iio/adc/ad7380.c
> > +++ b/drivers/iio/adc/ad7380.c
> > @@ -133,6 +133,7 @@ struct ad7380_state {
> >       struct spi_device *spi;
> >       struct regulator *vref;
> >       struct regmap *regmap;
> > +     struct spi_message *msg;
> >       /*
> >        * DMA (thus cache coherency maintenance) requires the
> >        * transfer buffers to live in their own cache lines.
> > @@ -231,19 +232,55 @@ static int ad7380_debugfs_reg_access(struct iio_d=
ev
> > *indio_dev, u32 reg,
> >       return ret;
> >  }
> >
> > +static int ad7380_buffer_preenable(struct iio_dev *indio_dev)
> > +{
> > +     struct ad7380_state *st =3D iio_priv(indio_dev);
> > +     struct spi_transfer *xfer;
> > +     int ret;
> > +
> > +     st->msg =3D spi_message_alloc(1, GFP_KERNEL);
> > +     if (!st->msg)
> > +             return -ENOMEM;
> > +
> > +     xfer =3D list_first_entry(&st->msg->transfers, struct spi_transfe=
r,
> > +                             transfer_list);
> > +
> > +     xfer->bits_per_word =3D st->chip_info->channels[0].scan_type.real=
bits;
> > +     xfer->len =3D 4;
> > +     xfer->rx_buf =3D st->scan_data.raw;
> > +
> > +     ret =3D spi_optimize_message(st->spi, st->msg);
> > +     if (ret) {
> > +             spi_message_free(st->msg);
> > +             return ret;
> > +     }
> > +
> > +     return 0;
> > +}
> > +
> > +static int ad7380_buffer_postdisable(struct iio_dev *indio_dev)
> > +{
> > +     struct ad7380_state *st =3D iio_priv(indio_dev);
> > +
> > +     spi_unoptimize_message(st->msg);
> > +     spi_message_free(st->msg);
> > +
> > +     return 0;
> > +}
> > +
>
> Not such a big deal but unless I'm missing something we could have the
> spi_message (+ the transfer) statically allocated in struct ad7380_state =
and do
> the optimize only once at probe (naturally with proper devm action for
> unoptimize). Then we would not need to this for every buffer enable + dis=
able. I
> know in terms of performance it won't matter but it would be less code I =
guess.
>
> Am I missing something?

No, your understanding is correct for the current state of everything
in this series. So, we could do as you suggest, but I have a feeling
that future additions to this driver might require that it gets
changed back this way eventually.