linux-kernel - Re: [PATCH v7 12/12] iio: cros

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: <CAPUE2usGMaqieLW+L_Axou1GoVVOEnWDd6huAsqY21iKnMDPzQ@mail.gmail.com>
Date:   Sat, 28 Mar 2020 17:33:48 -0700
From:   Gwendal Grignou <gwendal@...omium.org>
To:     Jonathan Cameron <jic23@...23.retrosnub.co.uk>
Cc:     Benson Leung <bleung@...omium.org>,
        Enric Balletbo i Serra <enric.balletbo@...labora.com>,
        Jonathan Cameron <Jonathan.Cameron@...wei.com>,
        linux-iio <linux-iio@...r.kernel.org>,
        linux-kernel <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v7 12/12] iio: cros_ec: flush as hwfifo attribute

On Sat, Mar 28, 2020 at 10:22 AM Jonathan Cameron
<jic23@...23.retrosnub.co.uk> wrote:
>
> On Fri, 27 Mar 2020 15:34:43 -0700
> Gwendal Grignou <gwendal@...omium.org> wrote:
>
> > Add buffer/hwfifo_flush. It is not part of the ABI, but it follows ST
> > and HID lead: Tells the sensor hub to send to the host all pending
> > sensor events.
> >
> > Signed-off-by: Gwendal Grignou <gwendal@...omium.org>
>
> Unless I'm missing something there aren't any other drivers providing
> an explicit flush attribute.
The flush attribute comes from a  requirement from Android to ask the
sensorhub to flush the samples still in its FIFO queue. (see
https://source.android.com/devices/sensors/hal-interface#flush_sensor)
It has been implemented in the ST Android HAL, which expects a
hw_fifo_flush attribute.:
https://github.com/STMicroelectronics/STMems_Android_Sensor_HAL_IIO/blob/STMems_Android_Sensor_HAL_IIO/src/utils.cpp#L31

But I misread kernel ST code; as you said, the request to flush
appends only when the buffer is enabled/disabled or the sensor
suspended, it is not exposed to user space.

For Bosh sensor : there is a patch that was proposed a while back:
"http://lkml.iu.edu/hypermail/linux/kernel/1504.3/03270.html", but it
never reached mainline.

For HID, the attribute is defined in the HID specification (31C) :
https://www.usb.org/sites/default/files/hutrr59_-_usages_for_wearables_0.pdf
but I could not find a publicly available proposed change request that uses it.

Anyhow, it was a mistake to put this patch in the current patch set. I
need it on chromebook for supporting Android, but it should be
discussed more widely to have it part of the ABI, or define a better
solution.

> The nearest equivalent is the flush
> callback which reads out stuff that is in a fifo to be read, but which
> hasn't yet reached a watermark to trigger normal readback.
>
> Can we do something similar here?
>
> If not this needs ABI documentation in Documentation/ABI/testing/...
> I'm not keen on it in becoming general ABI unless I'm missing a
> strong argument in favour of it.
>
> Jonathan
Thank you for your support,
Gwendal.
>
>
> > ---
> > No changes in v7.
> > New in v6.
> >
> >  .../cros_ec_sensors/cros_ec_sensors_core.c    | 28 +++++++++++++++++++
> >  1 file changed, 28 insertions(+)
> >
> > diff --git a/drivers/iio/common/cros_ec_sensors/cros_ec_sensors_core.c b/drivers/iio/common/cros_ec_sensors/cros_ec_sensors_core.c
> > index c831915ca7e56..aaf124a82e0e4 100644
> > --- a/drivers/iio/common/cros_ec_sensors/cros_ec_sensors_core.c
> > +++ b/drivers/iio/common/cros_ec_sensors/cros_ec_sensors_core.c
> > @@ -113,6 +113,33 @@ static int cros_ec_sensor_set_ec_rate(struct cros_ec_sensors_core_state *st,
> >       return ret;
> >  }
> >
> > +static ssize_t cros_ec_sensors_flush(struct device *dev,
> > +                                  struct device_attribute *attr,
> > +                                  const char *buf, size_t len)
> > +{
> > +     struct iio_dev *indio_dev = dev_to_iio_dev(dev);
> > +     struct cros_ec_sensors_core_state *st = iio_priv(indio_dev);
> > +     int ret = 0;
> > +     bool flush;
> > +
> > +     ret = strtobool(buf, &flush);
> > +     if (ret < 0)
> > +             return ret;
> > +     if (!flush)
> > +             return -EINVAL;
> > +
> > +     mutex_lock(&st->cmd_lock);
> > +     st->param.cmd = MOTIONSENSE_CMD_FIFO_FLUSH;
> > +     ret = cros_ec_motion_send_host_cmd(st, 0);
> > +     if (ret != 0)
> > +             dev_warn(&indio_dev->dev, "Unable to flush sensor\n");
> > +     mutex_unlock(&st->cmd_lock);
> > +     return ret ? ret : len;
> > +}
> > +
> > +static IIO_DEVICE_ATTR(hwfifo_flush, 0644, NULL,
> > +                    cros_ec_sensors_flush, 0);
> > +
> >  static ssize_t cros_ec_sensor_set_report_latency(struct device *dev,
> >                                                struct device_attribute *attr,
> >                                                const char *buf, size_t len)
> > @@ -175,6 +202,7 @@ static ssize_t hwfifo_watermark_max_show(struct device *dev,
> >  static IIO_DEVICE_ATTR_RO(hwfifo_watermark_max, 0);
> >
> >  const struct attribute *cros_ec_sensor_fifo_attributes[] = {
> > +     &iio_dev_attr_hwfifo_flush.dev_attr.attr,
> >       &iio_dev_attr_hwfifo_timeout.dev_attr.attr,
> >       &iio_dev_attr_hwfifo_watermark_max.dev_attr.attr,
> >       NULL,
>