[RFC] UVC Video encoder (zephyr,videoenc) support

doc/releases/migration-guide-4.3.rst

@@ -131,6 +131,12 @@ Stepper
 
 * :dtcompatible:`zephyr,gpio-stepper` has been replaced by :dtcompatible:`zephyr,h-bridge-stepper`.
 
+USB
+===
+
+* The USB Video Class was configuring the framerate and format of the source video device.
+  This is now to be done by the application after the host selected the format (:github:`93192`).
+
 .. zephyr-keep-sorted-stop
 
 Bluetooth

doc/releases/release-notes-4.3.rst

@@ -279,6 +279,12 @@ New APIs and options
 
   * :c:macro:`__deprecated_version`
 
+* USB
+
+  * Video
+
+    * :c:func:`uvc_add_format`
+
 * Video
 
   * :c:member:`video_format.size` field

drivers/video/video_common.c

@@ -468,3 +468,24 @@ int video_estimate_fmt_size(struct video_format *fmt)
 
 	return 0;
 }
+
+int video_set_compose_format(const struct device *dev, struct video_format *fmt)
+{
+	struct video_selection sel = {
+		.type = fmt->type,
+		.target = VIDEO_SEL_TGT_COMPOSE,
+		.rect.left = 0,
+		.rect.top = 0,
+		.rect.width = fmt->width,
+		.rect.height = fmt->height,
+	};
+	int ret;
+
+	ret = video_set_selection(dev, &sel);
+	if (ret < 0 && ret != -ENOSYS) {
+		LOG_ERR("Unable to set selection compose");
+		return ret;
+	}
+
+	return video_set_format(dev, fmt);
+}

drivers/video/video_stm32_dcmipp.c

@@ -1359,22 +1359,93 @@ static int stm32_dcmipp_dequeue(const struct device *dev, struct video_buffer **
 }
 
 /*
- * TODO: caps aren't yet handled hence give back straight the caps given by the
- * source.  Normally this should be the intersection of what the source produces
- * vs what the DCMIPP can input (for pipe0) and, for pipe 1 and 2, for a given
- * input format, generate caps based on capabilities, color conversion, decimation
- * etc
+ * For MAIN / AUX pipe, it is necessary that the pitch is a multiple of 16 bytes.
+ * Give here the multiple in number of pixels, which depends on the format chosen
  */
+#define DCMIPP_CEIL_DIV_ROUND_UP_MUL(val, div, mul)						\
+		((((val) + (div) - 1) / (div) + (mul) - 1) / (mul) * (mul))
+#define DCMIPP_CEIL_DIV(val, div)								\
+		(((val) + (div) - 1) / (div))
+#define DCMIPP_VIDEO_FORMAT_CAP(format, pixmul)	{						\
+	.pixelformat = VIDEO_PIX_FMT_##format,							\
+	.width_min = DCMIPP_CEIL_DIV_ROUND_UP_MUL(CONFIG_VIDEO_STM32_DCMIPP_SENSOR_WIDTH,	\
+						  STM32_DCMIPP_MAX_PIPE_SCALE_FACTOR,		\
+						  pixmul),					\
+	.width_max = CONFIG_VIDEO_STM32_DCMIPP_SENSOR_WIDTH / (pixmul) * (pixmul),		\
+	.height_min = DCMIPP_CEIL_DIV(CONFIG_VIDEO_STM32_DCMIPP_SENSOR_HEIGHT,			\
+				      STM32_DCMIPP_MAX_PIPE_SCALE_FACTOR),			\
+	.height_max = CONFIG_VIDEO_STM32_DCMIPP_SENSOR_HEIGHT,					\
+	.width_step = pixmul, .height_step = 1,							\
+}
+
+static const struct video_format_cap stm32_dcmipp_dump_fmt[] = {
+	{
+		.pixelformat =
+			VIDEO_FOURCC_FROM_STR(CONFIG_VIDEO_STM32_DCMIPP_SENSOR_PIXEL_FORMAT),
+		.width_min = CONFIG_VIDEO_STM32_DCMIPP_SENSOR_WIDTH,
+		.width_max = CONFIG_VIDEO_STM32_DCMIPP_SENSOR_WIDTH,
+		.height_min = CONFIG_VIDEO_STM32_DCMIPP_SENSOR_HEIGHT,
+		.height_max = CONFIG_VIDEO_STM32_DCMIPP_SENSOR_HEIGHT,
+		.width_step = 1, .height_step = 1,
+	},
+	{0},
+};
+
+static const struct video_format_cap stm32_dcmipp_main_fmts[] = {
+	DCMIPP_VIDEO_FORMAT_CAP(RGB565, 8),
+	DCMIPP_VIDEO_FORMAT_CAP(YUYV, 8),
+	DCMIPP_VIDEO_FORMAT_CAP(YVYU, 8),
+	DCMIPP_VIDEO_FORMAT_CAP(GREY, 16),
+	DCMIPP_VIDEO_FORMAT_CAP(RGB24, 16),
+	DCMIPP_VIDEO_FORMAT_CAP(BGR24, 16),
+	DCMIPP_VIDEO_FORMAT_CAP(ARGB32, 4),
+	DCMIPP_VIDEO_FORMAT_CAP(ABGR32, 4),
+	DCMIPP_VIDEO_FORMAT_CAP(RGBA32, 4),
+	DCMIPP_VIDEO_FORMAT_CAP(BGRA32, 4),
+	DCMIPP_VIDEO_FORMAT_CAP(NV12, 16),
+	DCMIPP_VIDEO_FORMAT_CAP(NV21, 16),
+	DCMIPP_VIDEO_FORMAT_CAP(NV16, 16),
+	DCMIPP_VIDEO_FORMAT_CAP(NV61, 16),
+	DCMIPP_VIDEO_FORMAT_CAP(YUV420, 16),
+	DCMIPP_VIDEO_FORMAT_CAP(YVU420, 16),
+	{0},
+};
+
+static const struct video_format_cap stm32_dcmipp_aux_fmts[] = {
+	DCMIPP_VIDEO_FORMAT_CAP(RGB565, 8),
+	DCMIPP_VIDEO_FORMAT_CAP(YUYV, 8),
+	DCMIPP_VIDEO_FORMAT_CAP(YVYU, 8),
+	DCMIPP_VIDEO_FORMAT_CAP(GREY, 16),
+	DCMIPP_VIDEO_FORMAT_CAP(RGB24, 16),
+	DCMIPP_VIDEO_FORMAT_CAP(BGR24, 16),
+	DCMIPP_VIDEO_FORMAT_CAP(ARGB32, 4),
+	DCMIPP_VIDEO_FORMAT_CAP(ABGR32, 4),
+	DCMIPP_VIDEO_FORMAT_CAP(RGBA32, 4),
+	DCMIPP_VIDEO_FORMAT_CAP(BGRA32, 4),
+	{0},
+};
+
 static int stm32_dcmipp_get_caps(const struct device *dev, struct video_caps *caps)
 {
-	const struct stm32_dcmipp_config *config = dev->config;
-	int ret;
+	struct stm32_dcmipp_pipe_data *pipe = dev->data;
 
-	ret = video_get_caps(config->source_dev, caps);
+	switch (pipe->id) {
+	case DCMIPP_PIPE0:
+		caps->format_caps = stm32_dcmipp_dump_fmt;
+		break;
+	case DCMIPP_PIPE1:
+		caps->format_caps = stm32_dcmipp_main_fmts;
+		break;
+	case DCMIPP_PIPE2:
+		caps->format_caps = stm32_dcmipp_aux_fmts;
+		break;
+	default:
+		CODE_UNREACHABLE;
+	}
 
 	caps->min_vbuf_count = 1;
 
-	return ret;
+	return 0;
 }
 
 static int stm32_dcmipp_get_frmival(const struct device *dev, struct video_frmival *frmival)

include/zephyr/drivers/video.h

@@ -982,6 +982,22 @@ int64_t video_get_csi_link_freq(const struct device *dev, uint8_t bpp, uint8_t l
  */
 int video_estimate_fmt_size(struct video_format *fmt);
 
+/**
+ * @brief Set compose rectangle (if applicable) prior to setting format
+ *
+ * Some devices expose compose capabilities, allowing them to apply a transformation
+ * (downscale / upscale) to the frame. For those devices, it is necessary to set the
+ * compose rectangle before being able to apply the frame format (which must have the
+ * same width / height and the compose rectangle width / height.
+ * In order to allow non-compose aware application to be able to control such devices,
+ * introduce a helper which, if available, will apply the compose rectangle prior to
+ * setting the format.
+ *
+ * @param dev Video device to query.
+ * @param fmt Video format structure pointer
+ */
+int video_set_compose_format(const struct device *dev, struct video_format *fmt);
+
 /**
  * @defgroup video_pixel_formats Video pixel formats
  * The '|' characters separate the pixels or logical blocks, and spaces separate the bytes.

include/zephyr/usb/class/usbd_uvc.h

@@ -26,20 +26,33 @@
  */
 
 /**
- * @brief Set the video device that a UVC instance will use.
+ * @brief Set the video device that a UVC instance will use for control requests.
  *
- * It will query its supported controls, formats and frame rates, and use this information to
- * generate USB descriptors sent to the host.
- *
- * At runtime, it will forward all USB controls from the host to this device.
+ * It will query its supported video controls and frame intervals and use this information to
+ * generate the USB descriptors presented to the host. At runtime, it will forward all USB controls
+ * from the host to this device.
  *
  * @note This function must be called before @ref usbd_enable.
  *
  * @param uvc_dev The UVC device
- * @param video_dev The video device that this UVC instance controls
+ * @param video_dev The video device that this UVC instance send controls requests to
  */
 void uvc_set_video_dev(const struct device *uvc_dev, const struct device *video_dev);
 
+/**
+ * @brief Set the video format capabilities that a UVC instance will present to the host.
+ *
+ * This information will be used to generate USB descriptors.
+ * The particular format selected by the host can be queried with @ref video_get_format.
+ *
+ * @note This function must be called before @ref usbd_enable and before @ref uvc_set_video_dev.
+ *
+ * @param uvc_dev The UVC device to configure
+ * @param fmt The video format to add to this UVC instance
+ * @return 0 on success, negative value on error
+ */
+int uvc_add_format(const struct device *const uvc_dev, const struct video_format *const fmt);
+
 /**
  * @}
  */

samples/drivers/video/capture/src/main.c

@@ -98,10 +98,14 @@ int main(void)
 	struct video_frmival frmival;
 	struct video_frmival_enum fie;
 	enum video_buf_type type = VIDEO_BUF_TYPE_OUTPUT;
-#if (CONFIG_VIDEO_SOURCE_CROP_WIDTH && CONFIG_VIDEO_SOURCE_CROP_HEIGHT) ||	\
-	CONFIG_VIDEO_FRAME_HEIGHT || CONFIG_VIDEO_FRAME_WIDTH
-	struct video_selection sel = {
+#if (CONFIG_VIDEO_SOURCE_CROP_WIDTH && CONFIG_VIDEO_SOURCE_CROP_HEIGHT)
+	struct video_selection crop_sel = {
 		.type = VIDEO_BUF_TYPE_OUTPUT,
+		.target = VIDEO_SEL_TGT_CROP;
+		.rect.left = CONFIG_VIDEO_SOURCE_CROP_LEFT;
+		.rect.top = CONFIG_VIDEO_SOURCE_CROP_TOP;
+		.rect.width = CONFIG_VIDEO_SOURCE_CROP_WIDTH;
+		.rect.height = CONFIG_VIDEO_SOURCE_CROP_HEIGHT;
 	};
 #endif
 	unsigned int frame = 0;
@@ -149,20 +153,14 @@ int main(void)
 
 	/* Set the crop setting if necessary */
 #if CONFIG_VIDEO_SOURCE_CROP_WIDTH && CONFIG_VIDEO_SOURCE_CROP_HEIGHT
-	sel.target = VIDEO_SEL_TGT_CROP;
-	sel.rect.left = CONFIG_VIDEO_SOURCE_CROP_LEFT;
-	sel.rect.top = CONFIG_VIDEO_SOURCE_CROP_TOP;
-	sel.rect.width = CONFIG_VIDEO_SOURCE_CROP_WIDTH;
-	sel.rect.height = CONFIG_VIDEO_SOURCE_CROP_HEIGHT;
-	if (video_set_selection(video_dev, &sel)) {
+	if (video_set_selection(video_dev, &crop_sel)) {
 		LOG_ERR("Unable to set selection crop");
 		return 0;
 	}
 	LOG_INF("Selection crop set to (%u,%u)/%ux%u",
 		sel.rect.left, sel.rect.top, sel.rect.width, sel.rect.height);
 #endif
 
-#if CONFIG_VIDEO_FRAME_HEIGHT || CONFIG_VIDEO_FRAME_WIDTH
 #if CONFIG_VIDEO_FRAME_HEIGHT
 	fmt.height = CONFIG_VIDEO_FRAME_HEIGHT;
 #endif
@@ -171,39 +169,14 @@ int main(void)
 	fmt.width = CONFIG_VIDEO_FRAME_WIDTH;
 #endif
 
-	/*
-	 * Check (if possible) if targeted size is same as crop
-	 * and if compose is necessary
-	 */
-	sel.target = VIDEO_SEL_TGT_CROP;
-	err = video_get_selection(video_dev, &sel);
-	if (err < 0 && err != -ENOSYS) {
-		LOG_ERR("Unable to get selection crop");
-		return 0;
-	}
-
-	if (err == 0 && (sel.rect.width != fmt.width || sel.rect.height != fmt.height)) {
-		sel.target = VIDEO_SEL_TGT_COMPOSE;
-		sel.rect.left = 0;
-		sel.rect.top = 0;
-		sel.rect.width = fmt.width;
-		sel.rect.height = fmt.height;
-		err = video_set_selection(video_dev, &sel);
-		if (err < 0 && err != -ENOSYS) {
-			LOG_ERR("Unable to set selection compose");
-			return 0;
-		}
-	}
-#endif
-
 	if (strcmp(CONFIG_VIDEO_PIXEL_FORMAT, "")) {
 		fmt.pixelformat = VIDEO_FOURCC_FROM_STR(CONFIG_VIDEO_PIXEL_FORMAT);
 	}
 
 	LOG_INF("- Video format: %s %ux%u",
 		VIDEO_FOURCC_TO_STR(fmt.pixelformat), fmt.width, fmt.height);
 
-	if (video_set_format(video_dev, &fmt)) {
+	if (video_set_compose_format(video_dev, &fmt)) {
 		LOG_ERR("Unable to set format");
 		return 0;
 	}

samples/subsys/usb/uvc/Kconfig

@@ -6,4 +6,16 @@
 # tree, you cannot use them in your own application.
 source "samples/subsys/usb/common/Kconfig.sample_usbd"
 
+menu "UVC specific configuration"
+
+config VIDEO_MAX_RANGE_RESOLUTIONS
+	int "Maximum number of intermediate resolutions"
+	default 5
+	help
+	   Control the maximum number of resolution that will be advertised
+	   to the USB client in case of the video capture supports a range
+	   of resolutions.
+
+endmenu
+
 source "Kconfig.zephyr"

samples/subsys/usb/uvc/boards/stm32n6570_dk.conf

@@ -0,0 +1,3 @@
+# Enough for 2592x1944 YUYV frames
+CONFIG_VIDEO_BUFFER_POOL_SZ_MAX=10100000
+CONFIG_MAIN_STACK_SIZE=2048

samples/subsys/usb/uvc/boards/stm32n6570_dk_stm32n657xx_fsbl.conf

@@ -0,0 +1,3 @@
+# Enough for 2592x1944 YUYV frames
+CONFIG_VIDEO_BUFFER_POOL_SZ_MAX=10100000
+CONFIG_MAIN_STACK_SIZE=2048

samples/subsys/usb/uvc/boards/stm32n6570_dk_stm32n657xx_sb.conf

@@ -0,0 +1,3 @@
+# Enough for 2592x1944 YUYV frames
+CONFIG_VIDEO_BUFFER_POOL_SZ_MAX=10100000
+CONFIG_MAIN_STACK_SIZE=2048