| 1 | .. SPDX-License-Identifier: GPL-2.0 |
| 2 | |
| 3 | Writing camera sensor drivers |
| 4 | ============================= |
| 5 | |
| 6 | CSI-2 and parallel (BT.601 and BT.656) busses |
| 7 | --------------------------------------------- |
| 8 | |
| 9 | Please see :ref:`transmitter-receiver`. |
| 10 | |
| 11 | Handling clocks |
| 12 | --------------- |
| 13 | |
| 14 | Camera sensors have an internal clock tree including a PLL and a number of |
| 15 | divisors. The clock tree is generally configured by the driver based on a few |
| 16 | input parameters that are specific to the hardware:: the external clock frequency |
| 17 | and the link frequency. The two parameters generally are obtained from system |
| 18 | firmware. **No other frequencies should be used in any circumstances.** |
| 19 | |
| 20 | The reason why the clock frequencies are so important is that the clock signals |
| 21 | come out of the SoC, and in many cases a specific frequency is designed to be |
| 22 | used in the system. Using another frequency may cause harmful effects |
| 23 | elsewhere. Therefore only the pre-determined frequencies are configurable by the |
| 24 | user. |
| 25 | |
| 26 | ACPI |
| 27 | ~~~~ |
| 28 | |
| 29 | Read the ``clock-frequency`` _DSD property to denote the frequency. The driver |
| 30 | can rely on this frequency being used. |
| 31 | |
| 32 | Devicetree |
| 33 | ~~~~~~~~~~ |
| 34 | |
| 35 | The currently preferred way to achieve this is using ``assigned-clocks``, |
| 36 | ``assigned-clock-parents`` and ``assigned-clock-rates`` properties. See |
| 37 | ``Documentation/devicetree/bindings/clock/clock-bindings.txt`` for more |
| 38 | information. The driver then gets the frequency using ``clk_get_rate()``. |
| 39 | |
| 40 | This approach has the drawback that there's no guarantee that the frequency |
| 41 | hasn't been modified directly or indirectly by another driver, or supported by |
| 42 | the board's clock tree to begin with. Changes to the Common Clock Framework API |
| 43 | are required to ensure reliability. |
| 44 | |
| 45 | Frame size |
| 46 | ---------- |
| 47 | |
| 48 | There are two distinct ways to configure the frame size produced by camera |
| 49 | sensors. |
| 50 | |
| 51 | Freely configurable camera sensor drivers |
| 52 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 53 | |
| 54 | Freely configurable camera sensor drivers expose the device's internal |
| 55 | processing pipeline as one or more sub-devices with different cropping and |
| 56 | scaling configurations. The output size of the device is the result of a series |
| 57 | of cropping and scaling operations from the device's pixel array's size. |
| 58 | |
| 59 | An example of such a driver is the CCS driver (see ``drivers/media/i2c/ccs``). |
| 60 | |
| 61 | Register list based drivers |
| 62 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 63 | |
| 64 | Register list based drivers generally, instead of able to configure the device |
| 65 | they control based on user requests, are limited to a number of preset |
| 66 | configurations that combine a number of different parameters that on hardware |
| 67 | level are independent. How a driver picks such configuration is based on the |
| 68 | format set on a source pad at the end of the device's internal pipeline. |
| 69 | |
| 70 | Most sensor drivers are implemented this way, see e.g. |
| 71 | ``drivers/media/i2c/imx319.c`` for an example. |
| 72 | |
| 73 | Frame interval configuration |
| 74 | ---------------------------- |
| 75 | |
| 76 | There are two different methods for obtaining possibilities for different frame |
| 77 | intervals as well as configuring the frame interval. Which one to implement |
| 78 | depends on the type of the device. |
| 79 | |
| 80 | Raw camera sensors |
| 81 | ~~~~~~~~~~~~~~~~~~ |
| 82 | |
| 83 | Instead of a high level parameter such as frame interval, the frame interval is |
| 84 | a result of the configuration of a number of camera sensor implementation |
| 85 | specific parameters. Luckily, these parameters tend to be the same for more or |
| 86 | less all modern raw camera sensors. |
| 87 | |
| 88 | The frame interval is calculated using the following equation:: |
| 89 | |
| 90 | frame interval = (analogue crop width + horizontal blanking) * |
| 91 | (analogue crop height + vertical blanking) / pixel rate |
| 92 | |
| 93 | The formula is bus independent and is applicable for raw timing parameters on |
| 94 | large variety of devices beyond camera sensors. Devices that have no analogue |
| 95 | crop, use the full source image size, i.e. pixel array size. |
| 96 | |
| 97 | Horizontal and vertical blanking are specified by ``V4L2_CID_HBLANK`` and |
| 98 | ``V4L2_CID_VBLANK``, respectively. The unit of the ``V4L2_CID_HBLANK`` control |
| 99 | is pixels and the unit of the ``V4L2_CID_VBLANK`` is lines. The pixel rate in |
| 100 | the sensor's **pixel array** is specified by ``V4L2_CID_PIXEL_RATE`` in the same |
| 101 | sub-device. The unit of that control is pixels per second. |
| 102 | |
| 103 | Register list based drivers need to implement read-only sub-device nodes for the |
| 104 | purpose. Devices that are not register list based need these to configure the |
| 105 | device's internal processing pipeline. |
| 106 | |
| 107 | The first entity in the linear pipeline is the pixel array. The pixel array may |
| 108 | be followed by other entities that are there to allow configuring binning, |
| 109 | skipping, scaling or digital crop :ref:`v4l2-subdev-selections`. |
| 110 | |
| 111 | USB cameras etc. devices |
| 112 | ~~~~~~~~~~~~~~~~~~~~~~~~ |
| 113 | |
| 114 | USB video class hardware, as well as many cameras offering a similar higher |
| 115 | level interface natively, generally use the concept of frame interval (or frame |
| 116 | rate) on device level in firmware or hardware. This means lower level controls |
| 117 | implemented by raw cameras may not be used on uAPI (or even kAPI) to control the |
| 118 | frame interval on these devices. |
| 119 | |
| 120 | Power management |
| 121 | ---------------- |
| 122 | |
| 123 | Always use runtime PM to manage the power states of your device. Camera sensor |
| 124 | drivers are in no way special in this respect: they are responsible for |
| 125 | controlling the power state of the device they otherwise control as well. In |
| 126 | general, the device must be powered on at least when its registers are being |
| 127 | accessed and when it is streaming. |
| 128 | |
| 129 | Existing camera sensor drivers may rely on the old |
| 130 | struct v4l2_subdev_core_ops->s_power() callback for bridge or ISP drivers to |
| 131 | manage their power state. This is however **deprecated**. If you feel you need |
| 132 | to begin calling an s_power from an ISP or a bridge driver, instead please add |
| 133 | runtime PM support to the sensor driver you are using. Likewise, new drivers |
| 134 | should not use s_power. |
| 135 | |
| 136 | Please see examples in e.g. ``drivers/media/i2c/ov8856.c`` and |
| 137 | ``drivers/media/i2c/ccs/ccs-core.c``. The two drivers work in both ACPI |
| 138 | and DT based systems. |
| 139 | |
| 140 | Control framework |
| 141 | ~~~~~~~~~~~~~~~~~ |
| 142 | |
| 143 | ``v4l2_ctrl_handler_setup()`` function may not be used in the device's runtime |
| 144 | PM ``runtime_resume`` callback, as it has no way to figure out the power state |
| 145 | of the device. This is because the power state of the device is only changed |
| 146 | after the power state transition has taken place. The ``s_ctrl`` callback can be |
| 147 | used to obtain device's power state after the power state transition: |
| 148 | |
| 149 | .. c:function:: int pm_runtime_get_if_in_use(struct device *dev); |
| 150 | |
| 151 | The function returns a non-zero value if it succeeded getting the power count or |
| 152 | runtime PM was disabled, in either of which cases the driver may proceed to |
| 153 | access the device. |