linux-kernel - Re: [RFC PATCH v4 1/1] iio/scmi: Adding support for IIO SCMI Based Sensors

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20210131204220.GB8355@e120937-lin>
Date:   Sun, 31 Jan 2021 20:42:20 +0000
From:   Cristian Marussi <cristian.marussi@....com>
To:     Jonathan Cameron <jic23@...nel.org>
Cc:     Jyoti Bhayana <jbhayana@...gle.com>,
        Hartmut Knaack <knaack.h@....de>,
        Lars-Peter Clausen <lars@...afoo.de>,
        Peter Meerwald-Stadler <pmeerw@...erw.net>,
        Mauro Carvalho Chehab <mchehab+huawei@...nel.org>,
        "David S. Miller" <davem@...emloft.net>,
        Rob Herring <robh@...nel.org>,
        Lukas Bulwahn <lukas.bulwahn@...il.com>,
        linux-kernel@...r.kernel.org, linux-iio@...r.kernel.org,
        sudeep.holla@....com, egranata@...gle.com,
        mikhail.golubev@...nsynergy.com, Igor.Skalkin@...nsynergy.com,
        Peter.hilber@...nsynergy.com, ankitarora@...gle.com
Subject: Re: [RFC PATCH v4 1/1] iio/scmi: Adding support for IIO SCMI Based
 Sensors

Hi

a clarification down below regarding something I pointed out in the
other thread (just to be sure I have not pointed out something
plain wrong :D)

Thanks

Cristian

On Sun, Jan 31, 2021 at 01:11:41PM +0000, Jonathan Cameron wrote:
> On Fri, 29 Jan 2021 22:18:18 +0000
> Jyoti Bhayana <jbhayana@...gle.com> wrote:
> 
> > This change provides ARM SCMI Protocol based IIO device.
> > This driver provides support for Accelerometer and Gyroscope using
> > SCMI Sensor Protocol extensions added in the SCMIv3.0 ARM specification
> > 
> > Signed-off-by: Jyoti Bhayana <jbhayana@...gle.com>
> 
> A few minor things noticed on a fresh read through, but mostly I think
> we are down to figuring out how to deal with the range (as discussed
> in the thread continuing on v3).
> 
> On another note, probably time to drop the RFC or give a bit more detail
> on why you think this isn't ready to be applied.
> 
> Thanks,
> 
> Jonathan
> 
[snip]

> > +
> > +static int scmi_iio_dev_probe(struct scmi_device *sdev)
> > +{
> > +	const struct scmi_sensor_info *sensor_info;
> > +	struct scmi_handle *handle = sdev->handle;
> > +	struct device *dev = &sdev->dev;
> > +	struct iio_dev *scmi_iio_dev;
> > +	u16 nr_sensors;
> > +	int err, i;
> > +
> > +	if (!handle || !handle->sensor_ops) {
> > +		dev_err(dev, "SCMI device has no sensor interface\n");
> I'm going to guess we can't actually get here because the registration
> would't have happened if either of those are true?
> If so perhaps drop the error message.
> 
> > +		return -EINVAL;
> > +	}
> > +
> > +	nr_sensors = handle->sensor_ops->count_get(handle);
> > +	if (!nr_sensors) {
> > +		dev_dbg(dev, "0 sensors found via SCMI bus\n");
> -ENODEV maybe?
> > +		return -EINVAL;
> > +	}
> > +
> > +	dev_dbg(dev, "%d sensors found via SCMI bus\n", nr_sensors);
> 
> Clear out any debug prints out that don't provide info that can't be obtained
> farily easily from elsewhere.  In this case they will either be registered
> or not and we'll get error messages.
> These sort of prints bitrot over time so we want to limit them to the truely
> useful.
> 
> > +
> > +	for (i = 0; i < nr_sensors; i++) {
> > +		sensor_info = handle->sensor_ops->info_get(handle, i);
> > +		if (!sensor_info) {
> > +			dev_err(dev, "SCMI sensor %d has missing info\n", i);
> > +			return -EINVAL;
> > +		}
> > +
> > +		/* Skipping scalar sensor,as this driver only supports accel and gyro */
> > +		if (sensor_info->num_axis == 0)
> > +			continue;
> 
> So there is a situation where this driver never creates anything?  In that path I'd
> like to see an -ENODEV error return.
> 
You mean -ENODEV only if this driver does not find at least one
good/supported GYRO/ACCEL sensor right ?

I would expect a system to possibly expose a bunch of other SCMI sensors
maybe unsupported by this IIO driver but currently handled by other
drivers, as an example on JUNO a number of temps/volts/currents sensors
are exposed and handled by the SCMI hwmon driver.


> > +
> > +		err = scmi_alloc_iiodev(dev, handle, sensor_info,
> > +					&scmi_iio_dev);
> > +		if (err < 0) {
> > +			dev_err(dev,
> > +				"failed to allocate IIO device for sensor %s: %d\n",
> > +				sensor_info->name, err);
> > +			return err;
> > +		}
> > +
> > +		err = scmi_iio_buffers_setup(scmi_iio_dev);
> > +		if (err < 0) {
> > +			dev_err(dev,
> > +				"IIO buffer setup error at sensor %s: %d\n",
> > +				sensor_info->name, err);
> > +			return err;
> > +		}
> > +
> > +		err = devm_iio_device_register(dev, scmi_iio_dev);
> > +		if (err) {
> > +			dev_err(dev,
> > +				"IIO device registration failed at sensor %s: %d\n",
> > +				sensor_info->name, err);
> > +			return err;
> > +		}
> > +	}
> > +	return err;
> > +}
> > +
> > +static const struct scmi_device_id scmi_id_table[] = {
> > +	{ SCMI_PROTOCOL_SENSOR, "iiodev" },
> 
> I'm curious on this.  What actually causes a match on that
> iiodev?  From digging around the scmi core am I right in thinking
> that this iiodev id needs to be explicitly listed?
> 
> It would be good to include any changes needed there in this
> series.
> 
> > +	{},
> > +};
> > +
> > +MODULE_DEVICE_TABLE(scmi, scmi_id_table);
> > +
> > +static struct scmi_driver scmi_iiodev_driver = {
> > +	.name = "scmi-sensor-iiodev",
> > +	.probe = scmi_iio_dev_probe,
> > +	.id_table = scmi_id_table,
> > +};
> > +
> > +module_scmi_driver(scmi_iiodev_driver);
> > +
> > +MODULE_AUTHOR("Jyoti Bhayana <jbhayana@...gle.com>");
> > +MODULE_DESCRIPTION("SCMI IIO Driver");
> > +MODULE_LICENSE("GPL v2");
>