iio: documentation: Document proximity sensor label use

Add an entry to Documentation/ABI/testing/sysfs-bus-iio for
the new device label sysfs-attribute support.

And document the standardized labels which may be used with proximity
sensors to hint userspace about the intended use of the sensor.

Using labels to differentiate between the multiple proximity sensors
which a modern laptop/tablet may have was discussed in this thread:
https://lore.kernel.org/linux-iio/9f9b0ff6-3bf1-63c4-eb36-901cecd7c4d9@redhat.com/

As mentioned there the "proximity-wifi*" labels are already being used
in this manner on some chromebooks, see e.g.:
arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
arch/arm64/boot/dts/qcom/sc7180-trogdor-lte-sku.dtsi

And the "proximity-palmrest" and "proximity-lap" labels are intended
to be used with the lap and palmrest sensors found in recent Lenovo
ThinkPad models.

Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Cc: Mark Pearson <mpearson@lenovo.com>
Cc: Bastien Nocera <hadess@hadess.net>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20210215191003.698888-1-hdegoede@redhat.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: Add relative hysteresis in ABI documentation

Add relative hysteresis in ABI documentation for als sensor.

Signed-off-by: Ye Xiang <xiang.ye@intel.com>
Link: https://lore.kernel.org/r/20210207070048.23935-4-xiang.ye@intel.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: hid-sensor-als: Add relative hysteresis support

Hid sensor als use relative hysteresis, this patch adds the support.

Signed-off-by: Ye Xiang <xiang.ye@intel.com>
Link: https://lore.kernel.org/r/20210207070048.23935-3-xiang.ye@intel.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: Add relative sensitivity support

Some hid sensors may use relative sensitivity such as als sensor.
This patch adds relative sensitivity checking for all hid sensors.

Signed-off-by: Ye Xiang <xiang.ye@intel.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Link: https://lore.kernel.org/r/20210207070048.23935-2-xiang.ye@intel.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

hid-sensors: Add more data fields for sensitivity checking

Before, when reading/writing the hysteresis of als, incli-3d, press, and
rotation sensor, we will get invalid argument error.

This patch add more sensitivity data fields for these sensors, so that
these sensors can get sensitivity index and return correct hysteresis
value.

Signed-off-by: Ye Xiang <xiang.ye@intel.com>
Link: https://lore.kernel.org/r/20210201054921.18214-3-xiang.ye@intel.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: hid-sensors: Move get sensitivity attribute to hid-sensor-common

No functional change has been made with this patch. The main intent here
is to reduce code repetition of getting sensitivity attribute.

In the current implementation, sensor_hub_input_get_attribute_info() is
called from multiple drivers to get attribute info for sensitivity
field. Moving this to common place will avoid code repetition.

Signed-off-by: Ye Xiang <xiang.ye@intel.com>
Acked-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Link: https://lore.kernel.org/r/20210201054921.18214-2-xiang.ye@intel.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

docs: iio: mark "repeat" sysfs attribute as optional

Show that this field is optional, just like the shift value.

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Link: https://lore.kernel.org/r/20210215143511.25471-6-luca@lucaceresoli.net
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

docs: iio: fix bullet list formatting

This 2nd-level bullet list is not properly ReST-formatted and thus it gets
rendered as a unique paragraph quite unreadable.  Fix by adding spaces as
needed.

While there also swap "shift" and "repeat" so they are in the correct
order.

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Link: https://lore.kernel.org/r/20210215143511.25471-5-luca@lucaceresoli.net
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

docs: iio: document the 'index' attribute too

Two out of three attributes are documented, document the third one too.

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Link: https://lore.kernel.org/r/20210215143511.25471-4-luca@lucaceresoli.net
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

docs: iio: fix directory naming

This directory is a, well, directory, not a file.

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Link: https://lore.kernel.org/r/20210215143511.25471-3-luca@lucaceresoli.net
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

docs: iio: fix example formatting

This line is part of the code snippet, so it has to be nested in order
to be rendered correctly.

Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Link: https://lore.kernel.org/r/20210215143511.25471-2-luca@lucaceresoli.net
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: use getter/setter functions

Use getter and setter functions, for a variety of data types.

Signed-off-by: Julia Lawall <Julia.Lawall@inria.fr>
Link: https://lore.kernel.org/r/20210209211315.1261791-1-Julia.Lawall@inria.fr
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

tools: iio: convert iio_generic_buffer to use new IIO buffer API

This change makes use of the new IIO buffer API to read data from an IIO
buffer.
It doesn't read the /sys/bus/iio/devices/iio:deviceX/scan_elements dir
anymore, it reads /sys/bus/iio/devices/iio:deviceX/bufferY, where all the
scan_elements have been merged together with the old/classical buffer
attributes.

And it makes use of the new IIO_BUFFER_GET_FD_IOCTL ioctl to get an FD for
the IIO buffer for which to read data from.
It also does a quick sanity check to see that -EBUSY is returned if reading
the chardev after the ioctl() has succeeded.

This was tested with the following cases:
 1. Tested buffer0 works with ioctl()
 2. Tested that buffer0 can't be opened via /dev/iio:deviceX after ioctl()
    This check should be omitted under normal operation; it's being done
    here to check that the driver change is sane
 3. Moved valid buffer0 to be buffer1, and tested that data comes from it

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-25-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

tools: iio: privatize globals and functions in iio_generic_buffer.c file

Mostly a tidy-up.
But also helps to understand the limits of scope of these functions and
globals.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-24-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

tools: iio: make iioutils_get_type() private in iio_utils

This is a bit of a tidy-up, but also helps with extending the
iioutils_get_type() function a bit, as we don't need to use it outside of
the iio_utils.c file. So, we'll need to update it only in one place.

With this change, the 'unsigned' types are updated to 'unsigned int' in the
iioutils_get_type() function definition.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-23-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: core: rename 'dev' -> 'indio_dev' in iio_device_alloc()

The 'dev' variable name usually refers to 'struct device' types. However in
iio_device_alloc() this was used for the 'struct iio_dev' type, which was
sometimes causing minor confusions.

This change renames the variable to 'indio_dev', which is the usual name
used around IIO for 'struct iio_dev' type objects.
It makes grepping a bit easier as well.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-22-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: buffer: add ioctl() to support opening extra buffers for IIO device

With this change, an ioctl() call is added to open a character device for a
buffer. The ioctl() number is 'i' 0x91, which follows the
IIO_GET_EVENT_FD_IOCTL ioctl.

The ioctl() will return an FD for the requested buffer index. The indexes
are the same from the /sys/iio/devices/iio:deviceX/bufferY (i.e. the Y
variable).

Since there doesn't seem to be a sane way to return the FD for buffer0 to
be the same FD for the /dev/iio:deviceX, this ioctl() will return another
FD for buffer0 (or the first buffer). This duplicate FD will be able to
access the same buffer object (for buffer0) as accessing directly the
/dev/iio:deviceX chardev.

Also, there is no IIO_BUFFER_GET_BUFFER_COUNT ioctl() implemented, as the
index for each buffer (and the count) can be deduced from the
'/sys/bus/iio/devices/iio:deviceX/bufferY' folders (i.e the number of
bufferY folders).

Used following C code to test this:
-------------------------------------------------------------------

 #include <stdio.h>
 #include <stdlib.h>
 #include <unistd.h>
 #include <sys/ioctl.h>
 #include <fcntl.h"
 #include <errno.h>

 #define IIO_BUFFER_GET_FD_IOCTL      _IOWR('i', 0x91, int)

int main(int argc, char *argv[])
{
        int fd;
        int fd1;
        int ret;

        if ((fd = open("/dev/iio:device0", O_RDWR))<0) {
                fprintf(stderr, "Error open() %d errno %d\n",fd, errno);
                return -1;
        }

        fprintf(stderr, "Using FD %d\n", fd);

        fd1 = atoi(argv[1]);

        ret = ioctl(fd, IIO_BUFFER_GET_FD_IOCTL, &fd1);
        if (ret < 0) {
                fprintf(stderr, "Error for buffer %d ioctl() %d errno %d\n", fd1, ret, errno);
                close(fd);
                return -1;
        }

        fprintf(stderr, "Got FD %d\n", fd1);

        close(fd1);
        close(fd);

        return 0;
}
-------------------------------------------------------------------

Results are:
-------------------------------------------------------------------
 # ./test 0
 Using FD 3
 Got FD 4

 # ./test 1
 Using FD 3
 Got FD 4

 # ./test 2
 Using FD 3
 Got FD 4

 # ./test 3
 Using FD 3
 Got FD 4

 # ls /sys/bus/iio/devices/iio\:device0
 buffer  buffer0  buffer1  buffer2  buffer3  dev
 in_voltage_sampling_frequency  in_voltage_scale
 in_voltage_scale_available
 name  of_node  power  scan_elements  subsystem  uevent
-------------------------------------------------------------------

iio:device0 has some fake kfifo buffers attached to an IIO device.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-21-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: buffer: introduce support for attaching more IIO buffers

With this change, calling iio_device_attach_buffer() will actually attach
more buffers.
Right now this doesn't do any validation of whether a buffer is attached
twice; maybe that can be added later (if needed). Attaching a buffer more
than once should yield noticeably bad results.

The first buffer is the legacy buffer, so a reference is kept to it.

At this point, accessing the data for the extra buffers (that are added
after the first one) isn't possible yet.

The iio_device_attach_buffer() is also changed to return an error code,
which for now is -ENOMEM if the array could not be realloc-ed for more
buffers.
To adapt to this new change iio_device_attach_buffer() is called last in
all place where it's called. The realloc failure is a bit difficult to
handle during un-managed calls when unwinding, so it's better to have this
as the last error in the setup_buffer calls.

At this point, no driver should call iio_device_attach_buffer() directly,
it should call one of the {devm_}iio_triggered_buffer_setup() or
devm_iio_kfifo_buffer_setup() or devm_iio_dmaengine_buffer_setup()
functions. This makes iio_device_attach_buffer() a bit easier to handle.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-20-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: dummy: iio_simple_dummy_buffer: use triggered buffer core calls

The iio_simple_dummy_configure_buffer() function is essentially a
re-implementation of the iio_triggered_buffer_setup() function.

This change makes use of the iio_triggered_buffer_setup() function. The
reason is so that we don't have to modify the iio_device_attach_buffer()
function in this driver as well.

One minor drawback is that the pollfunc name may not be 100% identical
with the one in the original code, but since it's an example, it should be
a big problem.

This change does a minor re-arranging of the included iio headers, as a
minor tidy-up.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-19-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: buffer: move __iio_buffer_free_sysfs_and_mask() before alloc

The __iio_buffer_free_sysfs_and_mask() function will be used in
iio_buffer_alloc_sysfs_and_mask() when multiple buffers will be attached to
the IIO device.
This will need to be used to cleanup resources on each buffer, when the
buffers cleanup unwind will occur on the error path.

The move is done in this patch to make the patch that adds multiple buffers
per IIO device a bit cleaner.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-18-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: core: wrap iio device & buffer into struct for character devices

In order to keep backwards compatibility with the current chardev
mechanism, and in order to add support for multiple buffers per IIO device,
we need to pass both the IIO device & IIO buffer to the chardev.

This is particularly needed for the iio_buffer_read_outer() function, where
we need to pass another buffer object than 'indio_dev->buffer'.

Since we'll also open some chardevs via anon inodes, we can pass extra
buffers in that function by assigning another object to the
iio_dev_buffer_pair object.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-17-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: buffer: dmaengine: obtain buffer object from attribute

The reference to the IIO buffer object is stored on the attribute object.
So we need to unwind it to obtain it.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-16-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: buffer: wrap all buffer attributes into iio_dev_attr

This change wraps all buffer attributes into iio_dev_attr objects, and
assigns a reference to the IIO buffer they belong to.

With the addition of multiple IIO buffers per one IIO device, we need a way
to know which IIO buffer is being enabled/disabled/controlled.

We know that all buffer attributes are device_attributes. So we can wrap
them with a iio_dev_attr types. In the iio_dev_attr type, we can also hold
a reference to an IIO buffer.
So, we end up being able to allocate wrapped attributes for all buffer
attributes (even the one from other drivers).

The neat part with this mechanism, is that we don't need to add any extra
cleanup, because these attributes are being added to a dynamic list that
will get cleaned up via iio_free_chan_devattr_list().

With this change, the 'buffer->scan_el_dev_attr_list' list is being renamed
to 'buffer->buffer_attr_list', effectively merging (or finalizing the
merge) of the buffer/ & scan_elements/ attributes internally.

Accessing these new buffer attributes can now be done via
'to_iio_dev_attr(attr)->buffer' inside the show/store handlers.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-15-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: add reference to iio buffer on iio_dev_attr

This change adds a reference to a 'struct iio_buffer' object on the
iio_dev_attr object. This way, we can use the created iio_dev_attr objects
on per-buffer basis (since they're allocated anyway).

A minor downside of this change is that the number of parameters on
__iio_add_chan_devattr() grows by 1. This looks like it could do with a bit
of a re-think.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-14-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: core: merge buffer/ & scan_elements/ attributes

With this change, we create a new directory for the IIO device called
buffer0, under which both the old buffer/ and scan_elements/ are stored.

This is done to simplify the addition of multiple IIO buffers per IIO
device. Otherwise we would need to add a bufferX/ and scan_elementsX/
directory for each IIO buffer.
With the current way of storing attribute groups, we can't have directories
stored under each other (i.e. scan_elements/ under buffer/), so the best
approach moving forward is to merge their attributes.

The old/legacy buffer/ & scan_elements/ groups are not stored on the opaque
IIO device object. This way the IIO buffer can have just a single
attribute_group object, saving a bit of memory when adding multiple IIO
buffers.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-13-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: buffer: group attr count and attr alloc

If we want to merge the attributes of the buffer/ and scan_elements/
directories, we'll need to count all attributes first, then (depending on
the attribute group) either allocate 2 attribute groups, or a single one.
Historically an IIO buffer was described by 2 subdirectories under
/sys/bus/iio/iio:devicesX (i.e. buffer/ and scan_elements/); these subdirs
were actually 2 separate attribute groups on the iio_buffer object.

Moving forward, if we want to allow more than one buffer per IIO device,
keeping 2 subdirectories for each IIO buffer is a bit cumbersome
(especially for userpace ABI). So, we will merge the attributes of these 2
subdirs under a /sys/bus/iio/iio:devicesX/bufferY subdirectory. To do this,
we need to count all attributes first, and then distribute them based on
which buffer this is. For the first buffer, we'll need to also allocate the
legacy 2 attribute groups (for buffer/ and scan_elements/), and also a
/sys/bus/iio/iio:devicesX/buffer0 attribute group.

For buffer1 and above, just a single attribute group will be allocated (the
merged one).

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-12-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: core: rework iio device group creation

Up until now, the device groups that an IIO device had were limited to 6.
Two of these groups would account for buffer attributes (the buffer/ and
scan_elements/ directories).

Since we want to add multiple buffers per IIO device, this number may not
be enough, when adding a second buffer. So, this change reallocates the
groups array whenever an IIO device group is added, via a
iio_device_register_sysfs_group() helper.

This also means that the groups array should be assigned to
'indio_dev.dev.groups' really late, right before {cdev_}device_add() is
called to do the entire setup.
And we also must take care to free this array when the sysfs resources are
being cleaned up.

With this change we can also move the 'groups' & 'groupcounter' fields to
the iio_dev_opaque object. Up until now, this didn't make a whole lot of
sense (especially since we weren't sure how multibuffer support would look
like in the end).
But doing it now kills one birds with one stone.

An alternative, would be to add a configurable Kconfig symbol
CONFIG_IIO_MAX_BUFFERS_PER_DEVICE (or something like that) and compute a
static maximum of the groups we can support per IIO device. But that would
probably annoy a few people since that would make the system less
configurable.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-11-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: core-trigger: make iio_device_register_trigger_consumer() an int return

Oddly enough the noop function is an int-return. This one seems to be void.
This change converts it to int, because we want to change how groups are
registered. With that change this function could error out with -ENOMEM.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-10-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: core: register chardev only if needed

We only need a chardev if we need to support buffers and/or events.

With this change, a chardev will be created only if an IIO buffer is
attached OR an event_interface is configured.

Otherwise, no chardev will be created, and the IIO device will get
registered with the 'device_add()' call.

Quite a lot of IIO devices don't really need a chardev, so this is a minor
improvement to the IIO core, as the IIO device will take up (slightly)
fewer resources.

In order to not create a chardev, we mostly just need to not initialize the
indio_dev->dev.devt field. If that is un-initialized, cdev_device_add()
behaves like device_add().

This change has a small chance of breaking some userspace ABI, because it
removes un-needed chardevs. While these chardevs (that are being removed)
have always been unusable, it is likely that some scripts may check their
existence (for whatever logic).
And we also hope that before opening these chardevs, userspace would have
already checked for some pre-conditions to make sure that opening these
chardevs makes sense.
For the most part, there is also the hope that it would be easier to change
userspace code than revert this. But in the case that reverting this is
required, it should be easy enough to do it.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-9-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

docs: ioctl-number.rst: reserve IIO subsystem ioctl() space

Currently, only the 'i' 0x90 ioctl() actually exists and is defined in
'include/uapi/linux/iio/events.h'.

It's the IIO_GET_EVENT_FD_IOCTL, which is used to retrieve and FD for
reading events from an IIO device.
We will want to add more ioct() numbers, so with this change the 'i'
0x90-0x9F space is reserved for IIO ioctl() calls.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-8-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: buffer-dma,adi-axi-adc: introduce devm_iio_dmaengine_buffer_setup()

This change does a conversion of the devm_iio_dmaengine_buffer_alloc() to
devm_iio_dmaengine_buffer_setup(). This will allocate an IIO DMA buffer and
attach it to the IIO device, similar to devm_iio_triggered_buffer_setup()
(though the underlying code is different, the final logic is the same).

Since the only user of the devm_iio_dmaengine_buffer_alloc() was the
adi-axi-adc driver, this change does the replacement in a single go in the
driver.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-7-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: kfifo: un-export devm_iio_kfifo_allocate() function

At this point all drivers should use devm_iio_kfifo_buffer_setup() instead
of manually allocating via devm_iio_kfifo_allocate() and assigning ops and
modes.

With this change, the devm_iio_kfifo_allocate() will be made private to the
IIO core, since all drivers should call either
devm_iio_kfifo_buffer_setup() or devm_iio_triggered_buffer_setup() to
create a kfifo buffer.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-6-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: accel: sca3000: use devm_iio_kfifo_buffer_setup() helper

This change makes use of the devm_iio_kfifo_buffer_setup() helper, however
the unwind order is changed.
The life-time of the kfifo object is attached to the parent device object.
This is to make the driver a bit more consistent with the other IIO
drivers, even though (as it is now before this change) it shouldn't be a
problem.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-5-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: make use of devm_iio_kfifo_buffer_setup() helper

All drivers that already call devm_iio_kfifo_allocate() &
iio_device_attach_buffer() are simple to convert to
iio_device_attach_kfifo_buffer() in a single go.

This change does that; the unwind order is preserved.
What is important, is that the devm_iio_kfifo_buffer_setup() be called
after the indio_dev->modes is assigned, to make sure that
INDIO_BUFFER_SOFTWARE flag is set and not overridden by the assignment to
indio_dev->modes.

Also, the INDIO_BUFFER_SOFTWARE has been removed from the assignments of
'indio_dev->modes' because it is set by devm_iio_kfifo_buffer_setup().

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Reviewed-by: Gwendal Grignou <gwendal@chromium.org>
Reviewed-by: Matt Ranostay <matt.ranostay@konsulko.com>x
Link: https://lore.kernel.org/r/20210215104043.91251-4-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: kfifo: add devm_iio_kfifo_buffer_setup() helper

This change adds the devm_iio_kfifo_buffer_setup() helper/short-hand,
which groups the simple routine of allocating a kfifo buffers via
devm_iio_kfifo_allocate() and calling iio_device_attach_buffer().

The mode_flags parameter is required, as the IIO kfifo supports 2 modes:
INDIO_BUFFER_SOFTWARE & INDIO_BUFFER_TRIGGERED.
The setup_ops parameter is optional.

This function will be a bit more useful when needing to define multiple
buffers per IIO device.

The naming for this function has been inspired from
iio_triggered_buffer_setup() since that one does a kfifo alloc + a pollfunc
alloc. So, this should have a more familiar ring to what it is.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210215104043.91251-3-alexandru.ardelean@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: Add basic unit test for iio_format_value()

The IIO core provides a function to do formatting of fixedpoint numbers.

In the past there have been some issues with the implementation of the
function where for example negative numbers were not handled correctly.

Introduce a basic unit test based on kunit that tests the function and
ensures that the generated output matches the expected output.

This gives us some confidence that future modifications to the function
implementation will not break ABI compatibility.

To run the unit tests follow the kunit documentation and add

  CONFIG_IIO=y
  CONFIG_IIO_TEST_FORMAT=y

to the .kunitconfig and run

  > ./tools/testing/kunit/kunit.py run
  Configuring KUnit Kernel ...
  Building KUnit Kernel ...
  Starting KUnit Kernel ...
  ============================================================
  ======== [PASSED] iio-format ========
  [PASSED] iio_test_iio_format_value_integer
  [PASSED] iio_test_iio_format_value_fixedpoint
  [PASSED] iio_test_iio_format_value_fractional
  [PASSED] iio_test_iio_format_value_fractional_log2
  [PASSED] iio_test_iio_format_value_multiple
  ============================================================
  Testing complete. 21 tests run. 0 failed. 0 crashed.
  Elapsed time: 8.242s total, 0.001s configuring, 3.865s building, 0.000s running

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Link: https://lore.kernel.org/r/20201215191743.2725-3-lars@metafoo.de
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: iio_format_value(): Fix IIO_VAL_FRACTIONAL_LOG2 values between -1.0 and 0.0

When formatting a value using IIO_VAL_FRACTIONAL_LOG2 and the values is
between -1 and 0 the sign is omitted.

We need the same trick as for IIO_VAL_FRACTIONAL to make sure this gets
formatted correctly.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Link: https://lore.kernel.org/r/20201215191743.2725-2-lars@metafoo.de
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: iio_format_value(): Use signed temporary for IIO_VAL_FRACTIONAL_LOG2

IIO_VAL_FRACTIONAL_LOG2 works with signed values, yet the temporary we use
is unsigned. This works at the moment because the variable is implicitly
cast to signed everywhere where it is used.

But it will certainly be cleaner to use a signed variable in the first
place.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Link: https://lore.kernel.org/r/20201215191743.2725-1-lars@metafoo.de
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: proximity: sx9500: Fix a spelling postive to positive

s/postive/positive/

Signed-off-by: Bhaskar Chowdhury <unixbhaskar@gmail.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20210210085704.1228068-1-unixbhaskar@gmail.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: accel: kxcjk-1013: Set label based on accel-location on 2-accel yoga-style 2-in-1s

Some 2-in-1 laptops / convertibles with 360° (yoga-style) hinges,
use 2 KXCJ91008 accelerometers:
1 in their display using an ACPI HID of "KIOX010A"; and
1 in their base    using an ACPI HID of "KIOX020A"

Since in this case we know the location of each accelerometer,
set the label for the accelerometers to the standardized
"accel-display" resp. "accel-base" labels. This way userspace
can use the labels to get the location.

This was tested on a Medion Akoya E2228T MD60250.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20210207160901.110643-4-hdegoede@redhat.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: accel: bmc150: Set label based on accel-location on 2-accel yoga-style 2-in-1s

Some 2-in-1 laptops / convertibles with 360° (yoga-style) hinges,
use 2 bmc150 accelerometers, defined by a single BOSC0200 ACPI
device node (1 in their base and 1 in their display).

Since in this case we know the location of each accelerometer,
set the label for the accelerometers to the standardized
"accel-display" resp. "accel-base" labels. This way userspace
can use the labels to get the location.

This was tested on a Lenovo ThinkPad Yoga 11e 4th gen (N3450 CPU).

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20210207160901.110643-3-hdegoede@redhat.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: core: Allow drivers to specify a label without it coming from of

Only set indio_dev->label from of/dt if there actually is a label
specified in of.

This allows drivers to set a label without this being overwritten with
NULL when there is no label specified in of. This is esp. useful on
devices where of is not used at all, such as your typical x86/ACPI device.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20210207160901.110643-2-hdegoede@redhat.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

bindings: iio: adc: Add documentation for ADS131E0x ADC driver

Add a device tree binding documentation for Texas Instruments
ADS131E0x ADC family driver.

Signed-off-by: Tomislav Denis <tomislav.denis@avl.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20210202084107.3260-3-tomislav.denis@avl.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: adc: Add driver for Texas Instruments ADS131E0x ADC family

The ADS131E0x are a family of multichannel, simultaneous sampling,
24-bit, delta-sigma, analog-to-digital converters (ADCs) with a
built-in programmable gain amplifier (PGA), internal reference
and an onboard oscillator.

Datasheet: https://www.ti.com/lit/ds/symlink/ads131e08.pdf
Signed-off-by: Tomislav Denis <tomislav.denis@avl.com>
Link: https://lore.kernel.org/r/20210202084107.3260-2-tomislav.denis@avl.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: accel: Add support for the Bosch-Sensortec BMI088

The BMI088 is a combined module with both accelerometer and gyroscope.
This adds the accelerometer driver support for the SPI interface.
The gyroscope part is already supported by the BMG160 driver.

Signed-off-by: Mike Looijmans <mike.looijmans@topic.nl>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20210125150732.23873-2-mike.looijmans@topic.nl
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

dt-bindings: iio: accel: Add bmi088 accelerometer bindings

This adds the device-tree bindings for the Bosch Sensortec BMI088 IMU,
the accelerometer part.

Signed-off-by: Mike Looijmans <mike.looijmans@topic.nl>
Reviewed-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20210125150732.23873-1-mike.looijmans@topic.nl
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio:adc:dac:ad5791 typo fix of regster

change 'regster' to 'register'

Signed-off-by: zuoqilin <zuoqilin@yulong.com>
Link: https://lore.kernel.org/r/20210128021905.963-1-zuoqilin1@163.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

MAINTAINERS: iio: move Peter Meerwald-Stadler to CREDITS

Haven't had much time lately and moved on to different things.
Thanks Jonathan for the gentle introduction to Linux land.

Signed-off-by: Peter Meerwald-Stadler <pmeerw@pmeerw.net>
Reviewed-by: Matt Ranostay <matt.ranostay@konsulko.com>
Link: https://lore.kernel.org/r/20210125195654.580465-1-pmeerw@pmeerw.net
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: adc: stm32-adc: enable timestamping for non-DMA usage

For non-DMA usage, we have an easy way to associate a timestamp with a
sample: iio_pollfunc_store_time stores a timestamp in the primary
trigger IRQ handler and stm32_adc_trigger_handler runs in the IRQ thread
to push out the buffer along with the timestamp.

For this to work, the driver needs to register an IIO_TIMESTAMP channel.
Do this.

For DMA, it's not as easy, because we don't push the buffers out of
stm32_adc_trigger, but out of stm32_adc_dma_buffer_done, which runs in
a tasklet scheduled after a DMA completion.

Preferably, the DMA controller would copy us the timestamp into that buffer
as well. Until this is implemented, restrict timestamping support to
only PIO. For low-frequency sampling, PIO is probably good enough.

Cc: Holger Assmann <has@pengutronix.de>
Acked-by: Fabrice Gasnier <fabrice.gasnier@foss.st.com>
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.kernel.org/r/20210125194824.30549-1-a.fatoum@pengutronix.de
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: hid-sensor-rotation: Fix quaternion data not correct

Because the data of HID_USAGE_SENSOR_ORIENT_QUATERNION defined by ISH FW
is s16, but quaternion data type is in_rot_quaternion_type(le:s16/32X4>>0),
need to transform data type from s16 to s32

May require manual backporting.

Fixes: fc18dddc0625 ("iio: hid-sensors: Added device rotation support")
Signed-off-by: Ye Xiang <xiang.ye@intel.com>
Link: https://lore.kernel.org/r/20210130102546.31397-1-xiang.ye@intel.com
Cc: <Stable@vger.kernel.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

counter: 104-quad-8: Remove IIO counter ABI

The IIO counter driver has been superseded by the Counter subsystem as
discussed in [1]. This patch removes the IIO counter ABI from the
104-QUAD-8 driver.

[1] https://lore.kernel.org/lkml/20210119104105.000010df@Huawei.com/

Cc: Syed Nayyar Waris <syednwaris@gmail.com>
Signed-off-by: William Breathitt Gray <vilhelm.gray@gmail.com>
Link: https://lore.kernel.org/r/98a39983d5df761c058a469d1346fd8ffdef8516.1611973018.git.vilhelm.gray@gmail.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

counter: stm32-lptimer-cnt: remove iio counter abi

Currently, the STM32 LP Timer counter driver registers into both IIO and
counter subsystems, which is redundant.

Remove the IIO counter ABI and IIO registration from the STM32 LP Timer
counter driver since it's been superseded by the Counter subsystem
as discussed in [1].

Keep only the counter subsystem related part.
Move a part of the ABI documentation into a driver comment.

This also removes a duplicate ABI warning
$ scripts/get_abi.pl validate
...
/sys/bus/iio/devices/iio:deviceX/in_count0_preset is defined 2 times:
  ./Documentation/ABI/testing/sysfs-bus-iio-timer-stm32:100
  ./Documentation/ABI/testing/sysfs-bus-iio-lptimer-stm32:0

[1] https://lkml.org/lkml/2021/1/19/347

Acked-by: William Breathitt Gray <vilhelm.gray@gmail.com>
Signed-off-by: Fabrice Gasnier <fabrice.gasnier@foss.st.com>
Link: https://lore.kernel.org/r/1611926542-2490-1-git-send-email-fabrice.gasnier@foss.st.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio:adc:stm32-adc: Add HAS_IOMEM dependency

Seems that there are config combinations in which this driver gets enabled
and hence selects the MFD, but with out HAS_IOMEM getting pulled in
via some other route.  MFD is entirely contained in an
if HAS_IOMEM block, leading to the build issue in this bugzilla.

https://bugzilla.kernel.org/show_bug.cgi?id=209889

Cc: <Stable@vger.kernel.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

iio: st_sensors: fix typo in comment

s/timetamping/timestamping/

Cc: trivial@kernel.org
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Link: https://lore.kernel.org/r/20210121153945.5499-1-a.fatoum@pengutronix.de
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

staging: unisys: visornic: Fix repeated words in comments

This patch removes some repeated words in code comments.

Signed-off-by: Klemen Košir <klemen.kosir@kream.io>
Link: https://lore.kernel.org/r/20210303090134.16007-1-klemen.kosir@kream.io
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: unisys: visornic: enhance visornic to use channel_interrupt

visornic uses timer to check the response queue and drain
it if needed periodically. On the other hand, visorbus
provides periodic work to serve such request, therefore,
timer should be replaced by channel_interrupt.

Signed-off-by: Song Chen <chensong_2000@189.cn>
Link: https://lore.kernel.org/r/1610439940-5590-1-git-send-email-chensong_2000@189.cn
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging/greybus: eliminate use of NAME_SIZE for strings

Instead, depend on the size of the destination buffer for easier
refactoring.

Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Link: https://lore.kernel.org/r/20210221154258.119503-1-memxor@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8188eu: Replace one-element array with flexible-array in struct ndis_802_11_var_ie

There is a regular need in the kernel to provide a way to declare having
a dynamically sized set of trailing elements in a structure. Kernel code
should always use “flexible array members”[1] for these cases. The older
style of one-element or zero-length arrays should no longer be used[2].

Use flexible-array member in struct ndis_802_11_var_ie, instead of
one-element array.

Also, this helps with the ongoing efforts to enable -Warray-bounds by
fixing the following warning:

drivers/staging/rtl8188eu/core/rtw_wlan_util.c: In function ‘HT_caps_handler’:
drivers/staging/rtl8188eu/core/rtw_wlan_util.c:665:65: warning: array subscript 2 is above array bounds of ‘u8[1]’ {aka ‘unsigned char[1]’} [-Warray-bounds]
  665 |    if ((pmlmeinfo->HT_caps.ampdu_params_info & 0x3) > (pIE->data[i] & 0x3))
      |                                                        ~~~~~~~~~^~~

[1] https://en.wikipedia.org/wiki/Flexible_array_member
[2] https://www.kernel.org/doc/html/v5.9/process/deprecated.html#zero-length-and-one-element-arrays

Link: https://github.com/KSPP/linux/issues/79
Link: https://github.com/KSPP/linux/issues/109
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Link: https://lore.kernel.org/r/20210304020228.GA45139@embeddedor
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: HalBtc8723b1Ant: remove useless typedef's

This commit fixes the following checkpatch.pl warnings:

    WARNING: do not add new typedefs
    #24: FILE: hal/HalBtc8723b1Ant.h:24:
    +typedef enum _BT_INFO_SRC_8723B_1ANT {

    WARNING: do not add new typedefs
    #31: FILE: hal/HalBtc8723b1Ant.h:31:
    +typedef enum _BT_8723B_1ANT_BT_STATUS {

    WARNING: do not add new typedefs
    #41: FILE: hal/HalBtc8723b1Ant.h:41:
    +typedef enum _BT_8723B_1ANT_WIFI_STATUS {

    WARNING: do not add new typedefs
    #51: FILE: hal/HalBtc8723b1Ant.h:51:
    +typedef enum _BT_8723B_1ANT_COEX_ALGO {

    WARNING: do not add new typedefs
    #66: FILE: hal/HalBtc8723b1Ant.h:66:
    +typedef struct _COEX_DM_8723B_1ANT {

    WARNING: do not add new typedefs
    #121: FILE: hal/HalBtc8723b1Ant.h:121:
    +typedef struct _COEX_STA_8723B_1ANT {

Signed-off-by: Marco Cesati <marco.cesati@gmail.com>
Link: https://lore.kernel.org/r/20210305100146.30687-1-marco.cesati@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: HalBtc8723b2Ant: remove useless typedef's

This commit fixes the following checkpatch.pl warnings:

    WARNING: do not add new typedefs
    #19: FILE: hal/HalBtc8723b2Ant.h:19:
    +typedef enum _BT_INFO_SRC_8723B_2ANT {

    WARNING: do not add new typedefs
    #26: FILE: hal/HalBtc8723b2Ant.h:26:
    +typedef enum _BT_8723B_2ANT_BT_STATUS {

    WARNING: do not add new typedefs
    #36: FILE: hal/HalBtc8723b2Ant.h:36:
    +typedef enum _BT_8723B_2ANT_COEX_ALGO {

    WARNING: do not add new typedefs
    #51: FILE: hal/HalBtc8723b2Ant.h:51:
    +typedef struct _COEX_DM_8723B_2ANT {

    WARNING: do not add new typedefs
    #104: FILE: hal/HalBtc8723b2Ant.h:104:
    +typedef struct _COEX_STA_8723B_2ANT {

Signed-off-by: Marco Cesati <marco.cesati@gmail.com>
Link: https://lore.kernel.org/r/20210305101151.13137-1-marco.cesati@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8723bs: remove duplicate pstat->hwaddr check

IS_MCAST(pstat->hwaddr) checked twice in a row in
odm_RefreshRateAdaptiveMaskCE(). Remove the second check.

Signed-off-by: Denis Efremov <efremov@linux.com>
Link: https://lore.kernel.org/r/20210305155001.61951-1-efremov@linux.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8723bs: remove useless typedef's in HalPhyRf.h

This commit fixes the following checkpatch.pl warnings:

    WARNING: do not add new typedefs
    #11: FILE: hal/HalPhyRf.h:11:
    +typedef enum _SPUR_CAL_METHOD {

    WARNING: do not add new typedefs
    #16: FILE: hal/HalPhyRf.h:16:
    +typedef enum _PWRTRACK_CONTROL_METHOD {

    WARNING: do not add new typedefs
    #27: FILE: hal/HalPhyRf.h:27:
    +typedef struct _TXPWRTRACK_CFG {

Signed-off-by: Marco Cesati <marco.cesati@gmail.com>
Link: https://lore.kernel.org/r/20210305144906.18850-1-marco.cesati@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8723bs: removed unused code block

Removed conditional code block checked by unused CONFIG_TX_BCAST2UNI

This patch does the cleaning required in TODO file:

find and remove code blocks guarded by never set CONFIG_FOO defines

Signed-off-by: Fabio Aiuto <fabioaiuto83@gmail.com>
Link: https://lore.kernel.org/r/20210305191713.GA2466@agape.jhs
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8188eu: rename parameters of odm_phy_status_query()

Rename parameters of odm_phy_status_query() to avoid camel case.

pDM_Odm -> dm_odm
pPhyInfo -> phy_info
pPhyStatus -> phy_status
pPktinfo -> pkt_info

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Link: https://lore.kernel.org/r/20210306081133.1272-2-straube.linux@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8188eu: rename ODM_PhyStatusQuery()

Rename ODM_PhyStatusQuery() to odm_phy_status_query() to avoid camel
case.

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Link: https://lore.kernel.org/r/20210306081133.1272-1-straube.linux@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vc04_services: vchiq_arm: fix error return code of vchiq_release_internal() and vchiq_use_internal()

When arm_state is NULL, no error return code of vchiq_release_internal()
and vchiq_use_internal() is assigned.
To fix this bug, ret is assigned with VCHIQ_ERROR.

Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Link: https://lore.kernel.org/r/20210306132245.16811-1-baijiaju1990@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ralink-gdma: Check return code of device_reset

The device_reset() function is marked as "__must_check", thus the static
analysis tool "sparse" complains that in ralink-gdma its return value is
ignored. Log a warning in case it returns an error.

Signed-off-by: Giovanni Gherdovich <bobdc9664@seznam.cz>
Link: https://lore.kernel.org/r/20210306141322.7516-1-bobdc9664@seznam.cz
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wimax: i2400m: fix some incorrect type warnings

Fix some "incorrect type in assignment" warnings reported by sparse in fw.c

sparse warnings:
wimax/i2400m/fw.c:266:27: warning: cast to restricted __le32
wimax/i2400m/fw.c:266:25: warning: incorrect type in assignment (different base types)
wimax/i2400m/fw.c:267:27: warning: cast to restricted __le32
wimax/i2400m/fw.c:267:25: warning: incorrect type in assignment (different base types)
wimax/i2400m/fw.c:268:27: warning: cast to restricted __le32
wimax/i2400m/fw.c:268:25: warning: incorrect type in assignment (different base types)
wimax/i2400m/fw.c:269:27: warning: cast to restricted __le32
wimax/i2400m/fw.c:269:25: warning: incorrect type in assignment (different base types)

Signed-off-by: Darryl T. Agostinelli <dagostinelli@gmail.com>
Link: https://lore.kernel.org/r/20210308144839.2364329-1-dagostinelli@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wimax/i2400m: convert __le32 type to host byte-order

fix sparse type warning by converting __le32 types
to host byte-order types before comparison

Signed-off-by: karthik alapati <mail@karthek.com>
Link: https://lore.kernel.org/r/87f93e091f736cb422f1d557fa5a2ac752f057a8.1613921277.git.mail@karthek.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wimax/i2400m: fix byte-order issue

fix sparse byte-order warnings by converting host byte-order
type to __le16 byte-order types before assigning to hdr.length

Signed-off-by: karthik alapati <mail@karthek.com>
Link: https://lore.kernel.org/r/0ae5c5c4c646506d8be871e7be5705542671a1d5.1613921277.git.mail@karthek.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8192e: remove redundant variable shadowing

In function rtl92e_start_adapter() automatic variable 'i' referenced only
within certain loops, used as iteration counter. Control flow can't get
into such loop w/o 'i = 0' assignment.

It's redundant to shadow this variable by creating scope around loop.

This patch fixes the following sparse warning:

warning: symbol 'i' shadows an earlier one

Signed-off-by: Nikolay Kyx <knv418@gmail.com>
Link: https://lore.kernel.org/r/20210302133217.145994-1-knv418@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8192e: reformat rx_ts_record struct

This change uses a space instead of tabs between the type and name of
each member of the struct.

Signed-off-by: William Durand <will+git@drnd.me>
Link: https://lore.kernel.org/r/20210301215335.767-10-will+git@drnd.me
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8192e: rename RxLastFragNum to rx_last_frag_num in rx_ts_record struct

Rename RxLastFragNum to rx_last_frag_num to silence a checkpatch warning
about CamelCase.

Signed-off-by: William Durand <will+git@drnd.me>
Link: https://lore.kernel.org/r/20210301215335.767-9-will+git@drnd.me
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8192e: rename RxLastSeqNum to rx_last_seq_num in rx_ts_record struct

Rename RxLastSeqNum to rx_last_seq_num to silence a checkpatch warning
about CamelCase.

Signed-off-by: William Durand <will+git@drnd.me>
Link: https://lore.kernel.org/r/20210301215335.767-8-will+git@drnd.me
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8192e: rename RxAdmittedBARecord to rx_admitted_ba_record in rx_ts_record struct

Rename RxAdmittedBARecord to rx_admitted_ba_record to silence a
checkpatch warning about CamelCase.

Signed-off-by: William Durand <will+git@drnd.me>
Link: https://lore.kernel.org/r/20210301215335.767-7-will+git@drnd.me
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8192e: rename RxPktPendingTimer to rx_pkt_pending_timer in rx_ts_record struct

Rename RxPktPendingTimer to rx_pkt_pending_timer to silence a checkpatch
warning about CamelCase.

Signed-off-by: William Durand <will+git@drnd.me>
Link: https://lore.kernel.org/r/20210301215335.767-6-will+git@drnd.me
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8192e: rename RxPendingPktList to rx_pending_pkt_list in rx_ts_record struct

Rename RxPendingPktList to rx_pending_pkt_list to silence a checkpatch
warning about CamelCase.

Signed-off-by: William Durand <will+git@drnd.me>
Link: https://lore.kernel.org/r/20210301215335.767-5-will+git@drnd.me
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8192e: rename RxTimeoutIndicateSeq to rx_timeout_indicate_seq in rx_ts_record struct

Rename RxTimeoutIndicateSeq to rx_timeout_indicate_seq to silence a
checkpatch warning about CamelCase.

Signed-off-by: William Durand <will+git@drnd.me>
Link: https://lore.kernel.org/r/20210301215335.767-4-will+git@drnd.me
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8192e: rename RxIndicateSeq to rx_indicate_seq in rx_ts_record struct

Rename RxIndicateSeq to rx_indicate_seq to silence a checkpatch warning
about CamelCase.

Signed-off-by: William Durand <will+git@drnd.me>
Link: https://lore.kernel.org/r/20210301215335.767-3-will+git@drnd.me
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8192e: rename TsCommonInfo to ts_common_info in rx_ts_record struct

Rename TsCommonInfo to ts_common_info to silence a checkpatch warning
about CamelCase.

Signed-off-by: William Durand <will+git@drnd.me>
Link: https://lore.kernel.org/r/20210301215335.767-2-will+git@drnd.me
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8723bs: os_dep: Replace sprintf with scnprintf

The use of sprintf with format string here means that there is a risk
that the writes will go out of bounds, replace it with scnprintf.

In one block of the translate_scan function sprintf is only called once
(it's not being used to concatenate strings) so there is no need to keep
the pointer "p", remove it.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Candy Febriyanto <cfebriyanto@gmail.com>
Link: https://lore.kernel.org/r/d76c5f1db8dbf02ac0ab954b0971ce24e5a8b9bd.1614610197.git.cfebriyanto@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8723bs: hal: Replace sprintf with scnprintf

The use of sprintf with format string here means that there is a risk
that the writes will go out of bounds, replace it with scnprintf.

Also avoid unnecessarily passing "%s" on some of the function calls.

Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Candy Febriyanto <cfebriyanto@gmail.com>
Link: https://lore.kernel.org/r/ed564fb9e325f757b2b937df37689d40b96d8831.1614610197.git.cfebriyanto@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8723bs: core: Replace sprintf with scnprintf

The use of sprintf with format string here means that there is a risk
that the writes will go out of bounds, replace it with scnprintf.

In on_action_public_default the variable "cnt" isn't being used for
anything meaningful so remove it.

Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Candy Febriyanto <cfebriyanto@gmail.com>
Link: https://lore.kernel.org/r/6c260641164d6856fc4447555fe739124cb27c19.1614610197.git.cfebriyanto@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging:rtl8712: replace cap_* definitions with native kernel WLAN_CAPABILITY_*

cap_* definitions duplicate WLAN_CAPABILITY_*. Remove cap_* definitions,
improve code consistency.

Reviewed-by: Mike Ximing Chen <mike.ximing.chen@intel.com>
Signed-off-by: Ivan Safonov <insafonov@gmail.com>
Link: https://lore.kernel.org/r/20210227222236.581490-5-insafonov@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging:rtl8712: use IEEE80211_FCTL_* kernel definitions

_TO_DS_, _FROM_DS_, _MORE_FRAG_, _RETRY_, _PWRMGT_, _MORE_DATA_,
_PRIVACY_, _ORDER_ definitions are duplicate IEEE80211_FCTL_*
kernel definitions.

Signed-off-by: Ivan Safonov <insafonov@gmail.com>
Link: https://lore.kernel.org/r/20210227222236.581490-4-insafonov@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging:rtl8712: remove unused definitions from wifi.h

These definitions are not used and will not be useful in the future.

Signed-off-by: Ivan Safonov <insafonov@gmail.com>
Link: https://lore.kernel.org/r/20210227222236.581490-3-insafonov@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging:rtl8712: replace get_(d|s)a with ieee80211_get_(D|S)A

get_da()/get_sa() duplicate native ieee80211_get_(D|S)A functions.
Remove get_(d|s)a, use ieee80211_get_(D|S)A instead.

Signed-off-by: Ivan Safonov <insafonov@gmail.com>
Link: https://lore.kernel.org/r/20210227222236.581490-2-insafonov@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: fwserial: minor coding style fix

Fixes this checkpatch warning
WARNING: Integer promotion: Using 'h' in '%04hx' is unnecessary

Signed-off-by: Lee Gibson <leegib@gmail.com>
Link: https://lore.kernel.org/r/20210226114800.316897-1-leegib@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: comedi: dt2814: Clear stale AI data on detach

When the Comedi "detach" handler is called, it is possible that an extra
A/D conversion (triggered during termination of a Comedi asynchronous
command) is still in progress.  In that case, the FINISH bit in the
Status register will eventually get set and there will be stale data
waiting to be read from the A/D Data register.  The interrupt handler
will also be called if still connected at the time.  That should all be
mostly harmless, but it would be better to wait for any such conversion
to complete and clear any stale data during the "detach".  Add a custom
"detach" handler `dt2814_detach()` to do that if an interrupt handler
has been set up.  (There is no need to do it if no interrupt handler was
set up because Comedi asynchronous command support is disabled in that
case.)

Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Link: https://lore.kernel.org/r/20210301165757.243065-7-abbotti@mev.co.uk
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: comedi: dt2814: Remove struct dt2814_private

The `ntrig` member of the `struct dt2814_private` pointed to by
`dev->private` is no longer used as a counter to determine the end of
acquisition for a Comedi asynchronous command.  The other member
`curadscan` is also unused.  Remove the allocation of the private data
during initialization and remove the definition of `struct
dt2814_private` since they are no longer needed.

Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Link: https://lore.kernel.org/r/20210301165757.243065-6-abbotti@mev.co.uk
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: comedi: dt2814: Fix asynchronous command interrupt handling

The support for asynchronous commands in this driver is currently
broken.  If interrupts are enabled, the interrupt handler is called at
the end of every A/D conversion.  A/D conversions could be due to
software-triggered conversions resulting from Comedi `INSN_READ`
instruction handling, or due to timer-trigger conversions enabled when
a Comedi asynchronous command is set up.  We only want the interrupt
handler to read a sample from the A/D Data register for timer-triggered
conversions, but currently it always reads the A/D Data register.  Since
the A/D Data register is read twice (to read a 12-bit value from an
8-bit register), that probably interferes with the reading for
software-triggered conversions.

The interrupt handler does not currently do anything with the data, it
just ignores it.  It should be written to the Comedi buffer if handling
an asynchronous command.

Other problems are that the driver has no Comedi `cancel` handler to
call when the asynchronous command is being stopped manually, and it
does not handle "infinite" acquisitions (when the command's `stop_src ==
TRIG_NONE`) properly.

Change the interrupt handler to check the timer enable (ENB) bit to
check the asynchronous command is active and return early if not
enabled.  Also check the error (ERR) and "conversion complete" (FINISH)
bits, and return early if neither is set.  Then the sample can be read
from the A/D Data register to clear the ERR and FINISH bits.  If the ERR
bit was set, terminate the acquisition with an error, otherwise write
the data to the Comedi buffer and check for end of acquisition.  Replace
the current check for end of acquisition, using the count of completed
scans in `scans_done` (updated by calls to `comedi_buf_write_samples()`)
when `stop_src == TRIG_COUNT`) and allowing "infinite" acquisitions when
`stop_src == TRIG_NONE`.

Add a `cancel` handler function `dt2814_ai_cancel()` that will be called
when the end of acquisition event is processed and when the acquisition
is stopped manually.  It turns off the timer enable (ENB) bit in the
Control register, leaving the current channel selected.

Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Link: https://lore.kernel.org/r/20210301165757.243065-5-abbotti@mev.co.uk
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: comedi: dt2814: Call dt2814_ai_clear() during initialization

The Comedi "attach" handler `dt2814_attach()` writes to the Control
register to turn off the timer enable 'ENB' bit, which triggers a
conversion.  It then sleeps awhile and checks the Status register,
returning an error if the ERR bit is set.  However, the ERR bit could
have been set due to the conversion being triggered while the A/D
converter was busy (unlikely) or due to the conversion being triggered
before some previous sample had been read from the A/D Data register.

Replace the existing code with a call to `dt2814_ai_clear()` which waits
for any conversion to finish and then clears any unread data or error
condition.  A non-zero return value from `dt2814_ai_clear()` indicates a
time-out while waiting for the A/D converter to become non-busy.  Return
an error in that case.

Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Link: https://lore.kernel.org/r/20210301165757.243065-4-abbotti@mev.co.uk
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: comedi: dt2814: Don't wait for conversion in interrupt handler

When the interrupt handler decides the final sample has been acquired,
it turns off the timer enable (ENB) bit in the Command register.  That
triggers another A/D conversion.  The interrupt handler currently waits
for that to finish and then reads the resulting, unwanted sample.  Since
the functions for handling Comedi read instructions and for setting up
asynchronous commands now call `dt2814_ai_clear()` to wait for and
discard any spurious A/D conversion, let's remove that code from the
interrupt handler.

Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Link: https://lore.kernel.org/r/20210301165757.243065-3-abbotti@mev.co.uk
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: comedi: dt2814: Clear stale AI data before operation

When performing a Comedi read instruction or setting up an asynchronous
command on the AI subdevice, clear any stale data on the A/D registers
by waiting for the Status register's BUSY bit to clear (if set) and then
if the FINISH or ERR bit is set, reading the A/D Data register twice to
clear the stale data.

Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Link: https://lore.kernel.org/r/20210301165757.243065-2-abbotti@mev.co.uk
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: clocking-wizard: Remove the hardcoding of the clock outputs

The number of output clocks are configurable in the hardware.
Currently the driver registers the maximum number of outputs.
Fix the same by registering only the outputs that are there.

Signed-off-by: Shubhrajyoti Datta <shubhrajyoti.datta@xilinx.com>
Link: https://lore.kernel.org/r/1614172241-17326-8-git-send-email-shubhrajyoti.datta@xilinx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: clocking-wizard: Add support for fractional support

Currently the set rate granularity is to integral divisors.
Add support for the fractional divisors.
Only the first output0 is fractional in the hardware.

Signed-off-by: Shubhrajyoti Datta <shubhrajyoti.datta@xilinx.com>
Link: https://lore.kernel.org/r/1614172241-17326-7-git-send-email-shubhrajyoti.datta@xilinx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: clocking-wizard: Add support for dynamic reconfiguration

The patch adds support for dynamic reconfiguration of clock output rate.
Output clocks are registered as dividers and set rate callback function
is used for dynamic reconfiguration.

Signed-off-by: Shubhrajyoti Datta <shubhrajyoti.datta@xilinx.com>
Co-developed-by: Chirag Parekh <chirag.parekh@xilinx.com>
Link: https://lore.kernel.org/r/1614172241-17326-6-git-send-email-shubhrajyoti.datta@xilinx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: clocking-wizard: Allow changing of parent rate for single output

If there is only one output then allow changing of the parent rate.

Signed-off-by: Shubhrajyoti Datta <shubhrajyoti.datta@xilinx.com>
Link: https://lore.kernel.org/r/1614172241-17326-5-git-send-email-shubhrajyoti.datta@xilinx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: clocking-wizard: Update the fixed factor divisors

Update the fixed factor clock registration to register the divisors.

Signed-off-by: Shubhrajyoti Datta <shubhrajyoti.datta@xilinx.com>
Link: https://lore.kernel.org/r/1614172241-17326-4-git-send-email-shubhrajyoti.datta@xilinx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: clocking-wizard: Rename speed-grade to xlnx,speed-grade

Rename speed-grade to xlnx,speed-grade

Signed-off-by: Shubhrajyoti Datta <shubhrajyoti.datta@xilinx.com>
Link: https://lore.kernel.org/r/1614172241-17326-3-git-send-email-shubhrajyoti.datta@xilinx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>