1                        Soc-Camera Subsystem
   2                        ====================
   7The following terms are used in this document:
   8 - camera / camera device / camera sensor - a video-camera sensor chip, capable
   9   of connecting to a variety of systems and interfaces, typically uses i2c for
  10   control and configuration, and a parallel or a serial bus for data.
  11 - camera host - an interface, to which a camera is connected. Typically a
  12   specialised interface, present on many SoCs, e.g., PXA27x and PXA3xx, SuperH,
  13   AVR32, i.MX27, i.MX31.
  14 - camera host bus - a connection between a camera host and a camera. Can be
  15   parallel or serial, consists of data and control lines, e.g., clock, vertical
  16   and horizontal synchronization signals.
  18Purpose of the soc-camera subsystem
  21The soc-camera subsystem provides a unified API between camera host drivers and
  22camera sensor drivers. It implements a V4L2 interface to the user, currently
  23only the mmap method is supported.
  25This subsystem has been written to connect drivers for System-on-Chip (SoC)
  26video capture interfaces with drivers for CMOS camera sensor chips to enable
  27the reuse of sensor drivers with various hosts. The subsystem has been designed
  28to support multiple camera host interfaces and multiple cameras per interface,
  29although most applications have only one camera sensor.
  31Existing drivers
  34As of 2.6.27-rc4 there are two host drivers in the mainline: pxa_camera.c for
  35PXA27x SoCs and sh_mobile_ceu_camera.c for SuperH SoCs, and four sensor drivers:
  36mt9m001.c, mt9m111.c, mt9v022.c and a generic soc_camera_platform.c driver. This
  37list is not supposed to be updated, look for more examples in your tree.
  39Camera host API
  42A host camera driver is registered using the
  44soc_camera_host_register(struct soc_camera_host *);
  46function. The host object can be initialized as follows:
  48static struct soc_camera_host pxa_soc_camera_host = {
  49        .drv_name       = PXA_CAM_DRV_NAME,
  50        .ops            = &pxa_soc_camera_host_ops,
  53All camera host methods are passed in a struct soc_camera_host_ops:
  55static struct soc_camera_host_ops pxa_soc_camera_host_ops = {
  56        .owner          = THIS_MODULE,
  57        .add            = pxa_camera_add_device,
  58        .remove         = pxa_camera_remove_device,
  59        .suspend        = pxa_camera_suspend,
  60        .resume         = pxa_camera_resume,
  61        .set_fmt_cap    = pxa_camera_set_fmt_cap,
  62        .try_fmt_cap    = pxa_camera_try_fmt_cap,
  63        .init_videobuf  = pxa_camera_init_videobuf,
  64        .reqbufs        = pxa_camera_reqbufs,
  65        .poll           = pxa_camera_poll,
  66        .querycap       = pxa_camera_querycap,
  67        .try_bus_param  = pxa_camera_try_bus_param,
  68        .set_bus_param  = pxa_camera_set_bus_param,
  71.add and .remove methods are called when a sensor is attached to or detached
  72from the host, apart from performing host-internal tasks they shall also call
  73sensor driver's .init and .release methods respectively. .suspend and .resume
  74methods implement host's power-management functionality and its their
  75responsibility to call respective sensor's methods. .try_bus_param and
  76.set_bus_param are used to negotiate physical connection parameters between the
  77host and the sensor. .init_videobuf is called by soc-camera core when a
  78video-device is opened, further video-buffer management is implemented completely
  79by the specific camera host driver. The rest of the methods are called from
  80respective V4L2 operations.
  82Camera API
  85Sensor drivers can use struct soc_camera_link, typically provided by the
  86platform, and used to specify to which camera host bus the sensor is connected,
  87and arbitrarily provide platform .power and .reset methods for the camera.
  88soc_camera_device_register() and soc_camera_device_unregister() functions are
  89used to add a sensor driver to or remove one from the system. The registration
  90function takes a pointer to struct soc_camera_device as the only parameter.
  91This struct can be initialized as follows:
  93        /* link to driver operations */
  94        icd->ops        = &mt9m001_ops;
  95        /* link to the underlying physical (e.g., i2c) device */
  96        icd->control    = &client->dev;
  97        /* window geometry */
  98        icd->x_min      = 20;
  99        icd->y_min      = 12;
 100        icd->x_current  = 20;
 101        icd->y_current  = 12;
 102        icd->width_min  = 48;
 103        icd->width_max  = 1280;
 104        icd->height_min = 32;
 105        icd->height_max = 1024;
 106        icd->y_skip_top = 1;
 107        /* camera bus ID, typically obtained from platform data */
 108        icd->iface      = icl->bus_id;
 110struct soc_camera_ops provides .probe and .remove methods, which are called by
 111the soc-camera core, when a camera is matched against or removed from a camera
 112host bus, .init, .release, .suspend, and .resume are called from the camera host
 113driver as discussed above. Other members of this struct provide respective V4L2
 116struct soc_camera_device also links to an array of struct soc_camera_data_format,
 117listing pixel formats, supported by the camera.
 119VIDIOC_S_CROP and VIDIOC_S_FMT behaviour
 122Above user ioctls modify image geometry as follows:
 124VIDIOC_S_CROP: sets location and sizes of the sensor window. Unit is one sensor
 125pixel. Changing sensor window sizes preserves any scaling factors, therefore
 126user window sizes change as well.
 128VIDIOC_S_FMT: sets user window. Should preserve previously set sensor window as
 129much as possible by modifying scaling factors. If the sensor window cannot be
 130preserved precisely, it may be changed too.
 132In soc-camera there are two locations, where scaling and cropping can taks
 133place: in the camera driver and in the host driver. User ioctls are first passed
 134to the host driver, which then generally passes them down to the camera driver.
 135It is more efficient to perform scaling and cropping in the camera driver to
 136save camera bus bandwidth and maximise the framerate. However, if the camera
 137driver failed to set the required parameters with sufficient precision, the host
 138driver may decide to also use its own scaling and cropping to fulfill the user's
 141Camera drivers are interfaced to the soc-camera core and to host drivers over
 142the v4l2-subdev API, which is completely functional, it doesn't pass any data.
 143Therefore all camera drivers shall reply to .g_fmt() requests with their current
 144output geometry. This is necessary to correctly configure the camera bus.
 145.s_fmt() and .try_fmt() have to be implemented too. Sensor window and scaling
 146factors have to be maintained by camera drivers internally. According to the
 147V4L2 API all capture drivers must support the VIDIOC_CROPCAP ioctl, hence we
 148rely on camera drivers implementing .cropcap(). If the camera driver does not
 149support cropping, it may choose to not implement .s_crop(), but to enable
 150cropping support by the camera host driver at least the .g_crop method must be
 153User window geometry is kept in .user_width and .user_height fields in struct
 154soc_camera_device and used by the soc-camera core and host drivers. The core
 155updates these fields upon successful completion of a .s_fmt() call, but if these
 156fields change elsewhere, e.g., during .s_crop() processing, the host driver is
 157responsible for updating them.
 160Author: Guennadi Liakhovetski <>
 161 kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.