Re: [PATCH 04/12] i2c: mux: populate the new *_atomic callbacks

From: Peter Rosin <peda@axentia.se>
To: Wolfram Sang <wsa+renesas@sang-engineering.com>,
	"linux-i2c@vger.kernel.org" <linux-i2c@vger.kernel.org>
Cc: "linux-renesas-soc@vger.kernel.org"
	<linux-renesas-soc@vger.kernel.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-arm-kernel@lists.infradead.org"
	<linux-arm-kernel@lists.infradead.org>,
	Stefan Lengfeld <contact@stefanchrist.eu>,
	"linux-omap@vger.kernel.org" <linux-omap@vger.kernel.org>,
	"linux-tegra@vger.kernel.org" <linux-tegra@vger.kernel.org>,
	Linus Walleij <linus.walleij@linaro.org>,
	Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Subject: Re: [PATCH 04/12] i2c: mux: populate the new *_atomic callbacks
Date: Wed, 3 Apr 2019 15:17:12 +0000	[thread overview]
Message-ID: <59d32414-3859-d614-dc41-cc0a1bc63301@axentia.se> (raw)
In-Reply-To: <20190403124019.8947-5-wsa+renesas@sang-engineering.com>

On 2019-04-03 14:40, Wolfram Sang wrote:
> If the parent adapter has atomic_xfer callbacks, populate them for the
> mux adapter as well. We can use the same translation function as for the
> non-atomic xfer callback.
> 
> Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
> ---
>  drivers/i2c/i2c-mux.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/i2c/i2c-mux.c b/drivers/i2c/i2c-mux.c
> index f330690b4125..603252fa1284 100644
> --- a/drivers/i2c/i2c-mux.c
> +++ b/drivers/i2c/i2c-mux.c
> @@ -310,12 +310,18 @@ int i2c_mux_add_adapter(struct i2c_mux_core *muxc,
>  		else
>  			priv->algo.master_xfer = __i2c_mux_master_xfer;
>  	}
> +	if (parent->algo->master_xfer_atomic)
> +		priv->algo.master_xfer_atomic = priv->algo.master_xfer;
> +
>  	if (parent->algo->smbus_xfer) {
>  		if (muxc->mux_locked)
>  			priv->algo.smbus_xfer = i2c_mux_smbus_xfer;
>  		else
>  			priv->algo.smbus_xfer = __i2c_mux_smbus_xfer;
>  	}
> +	if (parent->algo->smbus_xfer_atomic)
> +		priv->algo.smbus_xfer_atomic = priv->algo.smbus_xfer;
> +
>  	priv->algo.functionality = i2c_mux_functionality;
>  
>  	/* Now fill out new adapter structure */
> 

Hmmm, what happens if a driver implements .master_xfer and relies on
emulation for SMBus, and then someone implements .smbus_xfer_atomic
to handle some corner case at power-down?

Then someone hides the power-down device behind a mux. That would end
with xfers destined for .smbus_xfer_atomic being emulated by the
non-atomic .master_xfer, no?

Maybe too weird to care about?

I guess the question is if it is allowed to have .master_xfer_atomic
but not .master_xfer (and similarly for .smbus_xfer{,_atomic})? Maybe
that decision should be made explicit? And perhaps enforced?

I don't care deeply about the above though, so feel free to do
something about it, or

Reviewed-by: Peter Rosin <peda@axentia.se>

Cheers,
Peter