video: gc2145: addition of CSI output support #89571

avolmat-st · 2025-05-06T20:30:31Z

The PR adds GC2145 driver support to allow controlling a CSI based GC2145 sensor.
This is the setup used on the STM32MP13-DK board in which case the GC2145 is connected to the ST-MIPID02 CSI bridge (PR to be pushed as well) into the DCMIPP (ongoing PR 89407 with added parallel support to be pushed).

The current driver currently support 320x240 / 640x480 and 1600x1200 resolution, however for the time being I haven't been able to make the 320x240 resolution work in CSI (probably a PIXELRATE issue with the CSI receiver timing to be adjusted ...) so I only allow 640x480 and 1600x1200 resolution in CSI for the moment.

I've also added standard CSI2 DT macros into video.h so that not all sensor or receiver will have to define them.

I do not have a setup with a DVP GC2145 so could someone run a sanity test with those patches in a DVP mode.

josuah

Using the patch soup snippets to test with UVC:

git remote add tvai https://github.com/tinyvision-ai-inc/zephyr
git fetch tvai
git cherry-pick tvai/pr-usb-uvc~9..tvai/pr-usb-uvc
west build -b arduino_nicla_vision/stm32h747xx/m7 samples/subsys/usb/uvc/
west flash

This still works!

dts/bindings/video/galaxycore,gc2145.yaml

josuah · 2025-05-08T23:35:11Z

include/zephyr/drivers/video.h

+#define VIDEO_MIPI_CSI2_DT_YUV422_8	0x1e
+#define VIDEO_MIPI_CSI2_DT_YUV422_10	0x1f
+#define VIDEO_MIPI_CSI2_DT_RGB444	0x20
+#define VIDEO_MIPI_CSI2_DT_RGB555	0x21
+#define VIDEO_MIPI_CSI2_DT_RGB565	0x22
+#define VIDEO_MIPI_CSI2_DT_RGB666	0x23
+#define VIDEO_MIPI_CSI2_DT_RGB888	0x24
+#define VIDEO_MIPI_CSI2_DT_RAW6		0x28
+#define VIDEO_MIPI_CSI2_DT_RAW7		0x29
+#define VIDEO_MIPI_CSI2_DT_RAW8		0x2a
+#define VIDEO_MIPI_CSI2_DT_RAW10	0x2b
+#define VIDEO_MIPI_CSI2_DT_RAW12	0x2c
+#define VIDEO_MIPI_CSI2_DT_RAW14	0x2d


This seems to mostly match the naming of include/media/mipi-csi2.h.

josuah

Suggesting a change for the small syntax bugfix remote-endpoint -> remote-endpoint-label if that is ok to ask.
Otherwise I can open a separate PR as it is not a bug introduced by this and ~~CI does not fail due to this~~.

[EDIT: It does but https://github.com//pull/89627 is where the fix is]

avolmat-st · 2025-05-10T14:21:07Z

Corrected the PIXEL_RATE part that I got wrong within the sensor side when doing the ST_MIPID02 driver at the same time.
With this capture is fine on STM32MP13xx (DCMIPP) + ST_MIPID02 + GC2145.
I'm also going to push the ST_MIPID02 PR now.

avolmat-st · 2025-05-10T14:52:57Z

Small update to avoid if (err) style error checking and instead use MISRA validated if (err < 0).

josuah

If interested in improving this now, using DT_INST_ENDPOINT_BY_ID(inst, pid, eid) can be nicer, but I do not see this as blocking at least IMHO.

Thank you for extending this sensorr!

josuah · 2025-05-10T16:12:08Z

drivers/video/gc2145.c

+/* MIPI-CSI registers - on page 3 */
+#define GC2145_REG_DPHY_MODE1		0x01
+#define GC2145_DPHY_MODE1_CLK_EN	BIT(0)
+#define GC2145_DPHY_MODE1_LANE0_EN	BIT(1)
+#define GC2145_DPHY_MODE1_LANE1_EN	BIT(2)
+#define GC2145_DPHY_MODE1_CLK_LANE_P2S_SEL	BIT(7)
+
+#define GC2145_REG_DPHY_MODE2	0x02
+#define GC2145_DPHY_MODE2_CLK_DIFF(a)		((a) & 0x07)
+#define GC2145_DPHY_MODE2_LANE0_DIFF(a)	(((a) & 0x07) << 4)
+
+#define GC2145_REG_DPHY_MODE3	0x03
+#define GC2145_DPHY_MODE3_LANE1_DIFF(a)	((a) & 0x07)
+#define GC2145_DPHY_MODE3_CLK_DELAY		BIT(4)
+#define GC2145_DPHY_MODE3_LANE0_DELAY		BIT(5)
+#define GC2145_DPHY_MODE3_LANE1_DELAY		BIT(6)
+
+#define GC2145_REG_FIFO_FULL_LVL_LOW	0x04
+#define GC2145_REG_FIFO_FULL_LVL_HIGH	0x05
+#define GC2145_REG_FIFO_MODE		0x06
+#define GC2145_FIFO_MODE_READ_GATE	BIT(3)
+#define GC2145_FIFO_MODE_MIPI_CLK_MODULE	BIT(7)
+
+#define GC2145_REG_BUF_CSI2_MODE	0x10
+#define GC2145_CSI2_MODE_DOUBLE		BIT(0)
+#define GC2145_CSI2_MODE_RAW8		BIT(2)
+#define GC2145_CSI2_MODE_MIPI_EN	BIT(4)
+#define GC2145_CSI2_MODE_EN		BIT(7)
+
+#define GC2145_REG_MIPI_DT		0x11
+#define GC2145_REG_LWC_LOW		0x12
+#define GC2145_REG_LWC_HIGH		0x13
+#define GC2145_REG_DPHY_MODE		0x15
+#define GC2145_DPHY_MODE_TRIGGER_PROG	BIT(4)
+
+#define GC2145_REG_FIFO_GATE_MODE	0x17
+#define GC2145_REG_T_LPX		0x21
+#define GC2145_REG_T_CLK_HS_PREPARE	0x22
+#define GC2145_REG_T_CLK_ZERO		0x23
+#define GC2145_REG_T_CLK_PRE		0x24
+#define GC2145_REG_T_CLK_POST		0x25
+#define GC2145_REG_T_CLK_TRAIL		0x26
+#define GC2145_REG_T_HS_EXIT		0x27
+#define GC2145_REG_T_WAKEUP		0x28
+#define GC2145_REG_T_HS_PREPARE		0x29
+#define GC2145_REG_T_HS_ZERO		0x2a
+#define GC2145_REG_T_HS_TRAIL		0x2b
+


I will try to port it to the CCI library, I got started for DVP and will do it after the merge

dts/bindings/video/galaxycore,gc2145.yaml

josuah · 2025-05-10T16:20:25Z

drivers/video/gc2145.c

+	.bus_type = DT_PROP_OR(DT_CHILD(DT_INST_CHILD(0, port), endpoint), bus_type,
+			       VIDEO_BUS_TYPE_PARALLEL),


It might be possible to use the macro DT_INST_ENDPOINT_BY_ID(inst, pid, eid) for walking down the port and endpoint:

zephyr/include/zephyr/devicetree/port-endpoint.h

Lines 105 to 163 in c35bb0d

/**

* @brief Get an endpoint node from its id and its parent port id

*

* Given a device instance number, a port ID and an endpoint ID, return the endpoint node.

* It handles various ways of how a port and an endpoint could be defined as described in

* @ref DT_INST_PORT_BY_ID and below.

*

* Example usage with DT_INST_ENDPOINT_BY_ID() to get the @c endpoint or @c endpoint@0 node:

*

* @code{.c}

* DT_INST_ENDPOINT_BY_ID(inst, 0, 0)

* @endcode

*

* Example devicetree fragment:

*

* @code{.dts}

* &device {

* port {

* endpoint {

* };

* };

* };

* @endcode

*

* @code{.dts}

* &device {

* port {

* #address-cells = <1>;

* #size-cells = <0>;

* endpoint@0 {

* reg = <0x0>;

* };

* };

* };

* @endcode

*

* @code{.dts}

* &device {

* ports {

* #address-cells = <1>;

* #size-cells = <0>;

* port@0 {

* reg = <0x0>;

* #address-cells = <1>;

* #size-cells = <0>;

* endpoint@0 {

* reg = <0x0>;

* };

* };

* };

* };

* @endcode

*

* @param inst instance number

* @param pid port ID

* @param eid endpoint ID

* @return endpoint node associated with @p eid and @p pid

*/

#define DT_INST_ENDPOINT_BY_ID(inst, pid, eid) \

avolmat-st · 2025-05-11T08:19:08Z

I fixed the macro to get the bus-type. Should have done that already in the first place. Thanks for noticing.

avolmat-st · 2025-05-11T10:22:58Z

Apparently some tests are not successful but I have to say that I do not understand what is failing.

josuah · 2025-05-11T11:45:01Z

Apparently some tests are not successful but I have to say that I do not understand what is failing.

Maybe the CI issues will get resolved once https://github.com/zephyrproject-rtos/zephyr/pull/89571/files#r2080630046 is resolved (fix: merge #89627 ?)

devicetree error: 'remote-endpoint-label' is marked as required in 'properties:' in .../galaxycore,gc2145.yaml, but does not appear in <Node /soc/i2c@40005c00/gc2145@3c/port/endpoint>
https://github.com/zephyrproject-rtos/zephyr/actions/runs/14953906764/job/42007171754?pr=89571#step:12:927

avolmat-st · 2025-05-11T13:11:24Z

Apparently some tests are not successful but I have to say that I do not understand what is failing.

Maybe the CI issues will get resolved once https://github.com/zephyrproject-rtos/zephyr/pull/89571/files#r2080630046 is resolved (fix: merge #89627 ?)

oh yeah indeed. That #89627 should now be ready I think so we should be able to merge it before this one. Or maybe adding in the description that this PR depends on the other one might help the CI build ?

josuah · 2025-05-11T14:23:41Z

Or maybe adding in the description that this PR depends on the other one might help the CI build ?

You might need to rebase this PR on top of #89627's commits. Alas Github is not able to grok the order in which it should apply patches like maintainers would in a mailing list.

avolmat-st · 2025-05-12T15:21:35Z

Will be rebased on top of #89764 since this is the PR which now expose the VIDEO_CID_LINK_FREQUENCY control and the MIPI_CSI2 data type macros.
Upon rebase, I will update the driver to expose the VIDEO_CID_LINK_FREQUENCY to replace the VIDEO_CID_PIXEL_RATE control.

loicpoulain · 2025-05-17T19:23:29Z

include/zephyr/drivers/video.h

+#define VIDEO_MIPI_CSI2_DT_RAW10	0x2b
+#define VIDEO_MIPI_CSI2_DT_RAW12	0x2c
+#define VIDEO_MIPI_CSI2_DT_RAW14	0x2d
+


Would be good to define all the CSI2 data types now?

Alright yes !! Just converting to draft for the moment in order to avoid polluting the PR listing, while updating this.

avolmat-st · 2025-05-17T21:51:29Z

The MIPI data type macros are removed since I moved them into ST_MIPID02 PR which is a dependency for this PR since it also introduce the LINK_FREQUENCY ctrl. Last push allows you to see the latest modification (introduction of LINK_FREQUENCY). Then I will rebase this PR once ST_MIPID02 serie get agreed and CCI PR get merged.
(this PR will have to be rebased on top of main branch to also take into account the removal of the endpoint parameters).

josuah

We might need to review it again when dependencies get merged, but LGTM in the meantime.
One minor issue observed... that is probably me chasing ghosts rather than finding bugs!

josuah · 2025-05-19T17:15:02Z

drivers/video/gc2145.c

@@ -1082,6 +1236,14 @@ static int gc2145_set_fmt(const struct device *dev, enum video_endpoint_id ep,
 		return ret;
 	}

+	if (cfg->bus_type == VIDEO_BUS_TYPE_CSI2_DPHY) {
+		ret = gc2145_config_csi(dev, fmt->pixelformat, fmt->width, fmt->height);


If the format is not supported for CSI but it is for DVP (like 320x240), then drv_data->fmt = *fmt would still be applied, but the format would not be valid. Probably not a big issue though.

Moved the drv_data->fmt setting after CSI check.

avolmat-st · 2025-05-25T20:18:32Z

Simply rebased on HEAD.

Add standard data-type macros within the video.h in order to avoid having to redefine them all in each and every CSI based driver. Signed-off-by: Alain Volmat <alain.volmat@foss.st.com>

avolmat-st · 2025-05-31T14:47:04Z

Fixed @josuah comment and added 3 commits from PR ST_MIPID02 in order to be able to properly build the driver.

josuah

Still works!

I could not spot any issue left. LGTM thank you!

josuah · 2025-05-31T19:34:31Z

drivers/video/video_ctrls.h

+	/** Menu integer type, standard or driver-defined menu */
+	VIDEO_CTRL_TYPE_INTEGER_MENU = 6,


[typo]
The commit message still has MENU_INTEGER in it.

Add a new INTEGER_MENU type allowing to store signed 64bits integer into a menu. Signed-off-by: Alain Volmat <alain.volmat@foss.st.com>

Add a ctrl VIDEO_CID_LINK_FREQ to indicate to a sink the frequency at which a device streams data over CSI2. Since not all source device currently provide the LINK_FREQ control, add a helper function to retrieve it, with a fall-back by doing an approximate via the PIXEL_RATE control. Signed-off-by: Alain Volmat <alain.volmat@foss.st.com>

Do not enable by default the DVP streaming to avoid conflict with the introduction of the MIPI mode of the GC2145. Signed-off-by: Alain Volmat <alain.volmat@foss.st.com>

Add possibility to use the gc2145 sensor in CSI mode by adding the bus-type property in the device-tree. Signed-off-by: Alain Volmat <alain.volmat@foss.st.com>

Since QVGA resolution is currently not working in CSI, use VGA as default resolution to avoid failing during the init of the driver. Signed-off-by: Alain Volmat <alain.volmat@foss.st.com>

avolmat-st · 2025-06-01T19:14:57Z

Following a comment in PR 89764 which contains the
video: common: addition of INTEGER_MENU control type
video: add LINK_FREQ ctrl and a helper to retrieve it

I update this PR to also have the same valid commits.

Now that LINK_FREQ ctrl is no more READ_ONLY by default, add setting the READ_ONLY flag within the driver here since with this driver the LINK_FREQ ctrl is readl-only.

sonarqubecloud · 2025-06-01T19:41:28Z

Quality Gate passed

Issues
2 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

github-actions bot added the area: Video Video subsystem label May 6, 2025

github-actions bot requested review from josuah, loicpoulain and ngphibang May 6, 2025 20:31

avolmat-st force-pushed the video_gc2145_csi_support branch from 4c5ccf6 to c561479 Compare May 6, 2025 20:37

josuah added area: Drivers area: camera labels May 6, 2025

josuah mentioned this pull request May 8, 2025

meta: Track progress on moving video drivers and devicetree to use the video-interfaces binding and the endpoint DT macros #80514

Open

josuah reviewed May 8, 2025

View reviewed changes

josuah self-requested a review May 8, 2025 23:42

josuah requested changes May 8, 2025

View reviewed changes

avolmat-st force-pushed the video_gc2145_csi_support branch from c561479 to b016299 Compare May 10, 2025 14:19

avolmat-st force-pushed the video_gc2145_csi_support branch from b016299 to 20cb045 Compare May 10, 2025 14:52

avolmat-st mentioned this pull request May 10, 2025

video: st_mipid02: addition of ST MIPID02 CSI bridge #89764

Merged

josuah reviewed May 10, 2025

View reviewed changes

josuah previously approved these changes May 10, 2025

View reviewed changes

avolmat-st dismissed josuah’s stale review via 123e2dc May 11, 2025 08:17

avolmat-st force-pushed the video_gc2145_csi_support branch from 20cb045 to 123e2dc Compare May 11, 2025 08:17

avolmat-st assigned josuah and unassigned josuah May 11, 2025

loicpoulain reviewed May 17, 2025

View reviewed changes

avolmat-st marked this pull request as draft May 17, 2025 20:08

avolmat-st force-pushed the video_gc2145_csi_support branch from 123e2dc to 8425649 Compare May 17, 2025 21:47

josuah previously approved these changes May 19, 2025

View reviewed changes

avolmat-st dismissed josuah’s stale review via 511449f May 25, 2025 19:51

avolmat-st force-pushed the video_gc2145_csi_support branch 2 times, most recently from 511449f to 1946a51 Compare May 25, 2025 20:17

avolmat-st marked this pull request as ready for review May 25, 2025 20:18

video: addition of CSI2 standard data-type defines

da8d5b5

Add standard data-type macros within the video.h in order to avoid having to redefine them all in each and every CSI based driver. Signed-off-by: Alain Volmat <alain.volmat@foss.st.com>

avolmat-st force-pushed the video_gc2145_csi_support branch from 1946a51 to cc52cc4 Compare May 31, 2025 14:45

github-actions bot assigned josuah and ngphibang May 31, 2025

josuah previously approved these changes May 31, 2025

View reviewed changes

Alain Volmat added 5 commits June 1, 2025 21:06

video: common: addition of INTEGER_MENU control type

c976456

Add a new INTEGER_MENU type allowing to store signed 64bits integer into a menu. Signed-off-by: Alain Volmat <alain.volmat@foss.st.com>

video: gc2145: avoid default DVP enabling

7e35a3f

Do not enable by default the DVP streaming to avoid conflict with the introduction of the MIPI mode of the GC2145. Signed-off-by: Alain Volmat <alain.volmat@foss.st.com>

video: gc2145: add CSI mode of the GC2145 sensor

e0ccd1e

Add possibility to use the gc2145 sensor in CSI mode by adding the bus-type property in the device-tree. Signed-off-by: Alain Volmat <alain.volmat@foss.st.com>

video: gc2145: use VGA resolution as default

c509b92

Since QVGA resolution is currently not working in CSI, use VGA as default resolution to avoid failing during the init of the driver. Signed-off-by: Alain Volmat <alain.volmat@foss.st.com>

avolmat-st dismissed josuah’s stale review via c509b92 June 1, 2025 19:12

avolmat-st force-pushed the video_gc2145_csi_support branch from cc52cc4 to c509b92 Compare June 1, 2025 19:12

ngphibang approved these changes Jun 1, 2025

View reviewed changes

josuah approved these changes Jun 1, 2025

View reviewed changes

kartben merged commit 5fa26d9 into zephyrproject-rtos:main Jun 2, 2025
26 checks passed

		.bus_type = DT_PROP_OR(DT_CHILD(DT_INST_CHILD(0, port), endpoint), bus_type,
		VIDEO_BUS_TYPE_PARALLEL),

	/**
	* @brief Get an endpoint node from its id and its parent port id
	*
	* Given a device instance number, a port ID and an endpoint ID, return the endpoint node.
	* It handles various ways of how a port and an endpoint could be defined as described in
	* @ref DT_INST_PORT_BY_ID and below.
	*
	* Example usage with DT_INST_ENDPOINT_BY_ID() to get the @c endpoint or @c endpoint@0 node:
	*
	* @code{.c}
	* DT_INST_ENDPOINT_BY_ID(inst, 0, 0)
	* @endcode
	*
	* Example devicetree fragment:
	*
	* @code{.dts}
	* &device {
	* port {
	* endpoint {
	* };
	* };
	* };
	* @endcode
	*
	* @code{.dts}
	* &device {
	* port {
	* #address-cells = <1>;
	* #size-cells = <0>;
	* endpoint@0 {
	* reg = <0x0>;
	* };
	* };
	* };
	* @endcode
	*
	* @code{.dts}
	* &device {
	* ports {
	* #address-cells = <1>;
	* #size-cells = <0>;
	* port@0 {
	* reg = <0x0>;
	* #address-cells = <1>;
	* #size-cells = <0>;
	* endpoint@0 {
	* reg = <0x0>;
	* };
	* };
	* };
	* };
	* @endcode
	*
	* @param inst instance number
	* @param pid port ID
	* @param eid endpoint ID
	* @return endpoint node associated with @p eid and @p pid
	*/
	#define DT_INST_ENDPOINT_BY_ID(inst, pid, eid) \

		/** Menu integer type, standard or driver-defined menu */
		VIDEO_CTRL_TYPE_INTEGER_MENU = 6,

video: gc2145: addition of CSI output support #89571

video: gc2145: addition of CSI output support #89571

Uh oh!

Conversation

avolmat-st commented May 6, 2025

Uh oh!

josuah left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

josuah left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

avolmat-st commented May 10, 2025

Uh oh!

avolmat-st commented May 10, 2025

Uh oh!

josuah left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

avolmat-st commented May 11, 2025

Uh oh!

avolmat-st commented May 11, 2025

Uh oh!

josuah commented May 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

avolmat-st commented May 11, 2025

Uh oh!

josuah commented May 11, 2025

Uh oh!

avolmat-st commented May 12, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

avolmat-st commented May 17, 2025

Uh oh!

josuah left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

avolmat-st commented May 25, 2025

Uh oh!

avolmat-st commented May 31, 2025

Uh oh!

josuah left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

avolmat-st commented Jun 1, 2025

Uh oh!

sonarqubecloud bot commented Jun 1, 2025

Quality Gate passed

Uh oh!

Uh oh!

Uh oh!

josuah left a comment •

edited

Loading

josuah commented May 11, 2025 •

edited

Loading