1      UHID - User-space I/O driver support for HID subsystem
   2     ========================================================
   4The HID subsystem needs two kinds of drivers. In this document we call them:
   6 1. The "HID I/O Driver" is the driver that performs raw data I/O to the
   7    low-level device. Internally, they register an hid_ll_driver structure with
   8    the HID core. They perform device setup, read raw data from the device and
   9    push it into the HID subsystem and they provide a callback so the HID
  10    subsystem can send data to the device.
  12 2. The "HID Device Driver" is the driver that parses HID reports and reacts on
  13    them. There are generic drivers like "generic-usb" and "generic-bluetooth"
  14    which adhere to the HID specification and provide the standardizes features.
  15    But there may be special drivers and quirks for each non-standard device out
  16    there. Internally, they use the hid_driver structure.
  18Historically, the USB stack was the first subsystem to provide an HID I/O
  19Driver. However, other standards like Bluetooth have adopted the HID specs and
  20may provide HID I/O Drivers, too. The UHID driver allows to implement HID I/O
  21Drivers in user-space and feed the data into the kernel HID-subsystem.
  23This allows user-space to operate on the same level as USB-HID, Bluetooth-HID
  24and similar. It does not provide a way to write HID Device Drivers, though. Use
  25hidraw for this purpose.
  27There is an example user-space application in ./samples/uhid/uhid-example.c
  29The UHID API
  32UHID is accessed through a character misc-device. The minor-number is allocated
  33dynamically so you need to rely on udev (or similar) to create the device node.
  34This is /dev/uhid by default.
  36If a new device is detected by your HID I/O Driver and you want to register this
  37device with the HID subsystem, then you need to open /dev/uhid once for each
  38device you want to register. All further communication is done by read()'ing or
  39write()'ing "struct uhid_event" objects. Non-blocking operations are supported
  40by setting O_NONBLOCK.
  42struct uhid_event {
  43        __u32 type;
  44        union {
  45                struct uhid_create_req create;
  46                struct uhid_data_req data;
  47                ...
  48        } u;
  51The "type" field contains the ID of the event. Depending on the ID different
  52payloads are sent. You must not split a single event across multiple read()'s or
  53multiple write()'s. A single event must always be sent as a whole. Furthermore,
  54only a single event can be sent per read() or write(). Pending data is ignored.
  55If you want to handle multiple events in a single syscall, then use vectored
  56I/O with readv()/writev().
  58The first thing you should do is sending an UHID_CREATE event. This will
  59register the device. UHID will respond with an UHID_START event. You can now
  60start sending data to and reading data from UHID. However, unless UHID sends the
  61UHID_OPEN event, the internally attached HID Device Driver has no user attached.
  62That is, you might put your device asleep unless you receive the UHID_OPEN
  63event. If you receive the UHID_OPEN event, you should start I/O. If the last
  64user closes the HID device, you will receive an UHID_CLOSE event. This may be
  65followed by an UHID_OPEN event again and so on. There is no need to perform
  66reference-counting in user-space. That is, you will never receive multiple
  67UHID_OPEN events without an UHID_CLOSE event. The HID subsystem performs
  68ref-counting for you.
  69You may decide to ignore UHID_OPEN/UHID_CLOSE, though. I/O is allowed even
  70though the device may have no users.
  72If you want to send data to the HID subsystem, you send an HID_INPUT event with
  73your raw data payload. If the kernel wants to send data to the device, you will
  74read an UHID_OUTPUT or UHID_OUTPUT_EV event.
  76If your device disconnects, you should send an UHID_DESTROY event. This will
  77unregister the device. You can now send UHID_CREATE again to register a new
  79If you close() the fd, the device is automatically unregistered and destroyed
  84write() allows you to modify the state of the device and feed input data into
  85the kernel. The following types are supported: UHID_CREATE, UHID_DESTROY and
  86UHID_INPUT. The kernel will parse the event immediately and if the event ID is
  87not supported, it will return -EOPNOTSUPP. If the payload is invalid, then
  88-EINVAL is returned, otherwise, the amount of data that was read is returned and
  89the request was handled successfully.
  92  This creates the internal HID device. No I/O is possible until you send this
  93  event to the kernel. The payload is of type struct uhid_create_req and
  94  contains information about your device. You can start I/O now.
  97  This destroys the internal HID device. No further I/O will be accepted. There
  98  may still be pending messages that you can receive with read() but no further
  99  UHID_INPUT events can be sent to the kernel.
 100  You can create a new device by sending UHID_CREATE again. There is no need to
 101  reopen the character device.
 104  You must send UHID_CREATE before sending input to the kernel! This event
 105  contains a data-payload. This is the raw data that you read from your device.
 106  The kernel will parse the HID reports and react on it.
 109  If you receive a UHID_FEATURE request you must answer with this request. You
 110  must copy the "id" field from the request into the answer. Set the "err" field
 111  to 0 if no error occured or to EIO if an I/O error occurred.
 112  If "err" is 0 then you should fill the buffer of the answer with the results
 113  of the feature request and set "size" correspondingly.
 117read() will return a queued ouput report. These output reports can be of type
 119reaction is required to any of them but you should handle them according to your
 120needs. Only UHID_OUTPUT and UHID_OUTPUT_EV have payloads.
 123  This is sent when the HID device is started. Consider this as an answer to
 124  UHID_CREATE. This is always the first event that is sent.
 126  UHID_STOP:
 127  This is sent when the HID device is stopped. Consider this as an answer to
 129  If the kernel HID device driver closes the device manually (that is, you
 130  didn't send UHID_DESTROY) then you should consider this device closed and send
 131  an UHID_DESTROY event. You may want to reregister your device, though. This is
 132  always the last message that is sent to you unless you reopen the device with
 135  UHID_OPEN:
 136  This is sent when the HID device is opened. That is, the data that the HID
 137  device provides is read by some other process. You may ignore this event but
 138  it is useful for power-management. As long as you haven't received this event
 139  there is actually no other process that reads your data so there is no need to
 140  send UHID_INPUT events to the kernel.
 143  This is sent when there are no more processes which read the HID data. It is
 144  the counterpart of UHID_OPEN and you may as well ignore this event.
 147  This is sent if the HID device driver wants to send raw data to the I/O
 148  device. You should read the payload and forward it to the device. The payload
 149  is of type "struct uhid_data_req".
 150  This may be received even though you haven't received UHID_OPEN, yet.
 153  Same as UHID_OUTPUT but this contains a "struct input_event" as payload. This
 154  is called for force-feedback, LED or similar events which are received through
 155  an input device by the HID subsystem. You should convert this into raw reports
 156  and send them to your device similar to events of type UHID_OUTPUT.
 159  This event is sent if the kernel driver wants to perform a feature request as
 160  described in the HID specs. The report-type and report-number are available in
 161  the payload.
 162  The kernel serializes feature requests so there will never be two in parallel.
 163  However, if you fail to respond with a UHID_FEATURE_ANSWER in a time-span of 5
 164  seconds, then the requests will be dropped and a new one might be sent.
 165  Therefore, the payload also contains an "id" field that identifies every
 166  request.
 168Document by:
 169  David Herrmann <>
 170 kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.