Skip to contentSkip to conversation

media: ov08x40: Add Signal Integration Adjustments for specific project

PATCH Superseded Chen, Jason Z 3 participants 4 messages Oct 20, 2023 – Oct 30, 2023

Subsystems (3)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 media: ov08x40: Add Signal Integration Adjustments for specific project

From: Jason Chen <[email protected]>

Due to certain MIPI hardware designs using overly long cables, there is a loss of signal strength, resulting in failed signal integration. The patch includes changes to adjust the driving strength in the register settings for a specific project.

---
 drivers/media/i2c/ov08x40.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)
drivers/media/i2c/ov08x40.c +17 -0
index 0b3b906ebef..4f13e23b325 100644
@@ -160,6 +160,18 @@ static const struct ov08x40_reg mipi_data_rate_800mbps[] = {
 	{0x6002, 0x2e},
 };
 
+static const struct ov08x40_reg mipi_si_changed_regs[] = {
+	{0x481b, 0x2c}, /* HS Exit: Data Tx TEOT - reducing 8ns */
+	{0x4826, 0x42}, /* HS Entry: Data Tx TREOT - raising 8ns */
+	{0x4829, 0x54}, /* HS Exit: Data Tx TREOT - reducing 8ns */
+	{0x4885, 0x1f}, /* driving strength change */
+};
+
+struct ov08x40_reg_list si_regs = {
+	.regs = mipi_si_changed_regs,
+	.num_of_regs = ARRAY_SIZE(mipi_si_changed_regs),
+};
+
 static const struct ov08x40_reg mode_3856x2416_regs[] = {
 	{0x5000, 0x5d},
 	{0x5001, 0x20},
@@ -2917,6 +2929,11 @@ static int ov08x40_start_streaming(struct ov08x40 *ov08x)
 		return ret;
 	}
 
+	/* Apply SI change to current project */
+	reg_list = &si_regs;
+
+	ov08x40_write_reg_list(ov08x, reg_list);
+
 	/* Apply default values of current mode */
 	reg_list = &ov08x->cur_mode->reg_list;
 	ret = ov08x40_write_reg_list(ov08x, reg_list);
Signature · 1 line
2.34.1
#1 lore raw

Quoting Chen, Jason Z (2023-10-20 05:09:08)

Quoted 6 lines

From: Jason Chen <[email protected]>

Due to certain MIPI hardware designs using overly long cables, there is a loss of signal strength, resulting in failed signal integration. The patch includes changes to adjust the driving strength in the register settings for a specific project.

For a specific project? Will this now affect 'all' projects using this sensor? Will that be adverse in anyway?

Do 'short' cables still function with this patch?

Quoted 11 lines
---
 drivers/media/i2c/ov08x40.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)
drivers/media/i2c/ov08x40.c +0 -0
index 0b3b906ebef..4f13e23b325 100644

drivers/media/i2c/ov08x40.c static const struct ov08x40_reg mipi_data_rate_800mbps[] = { · line 167

Quoted 8 lines
drivers/media/i2c/ov08x40.c +5 -0
@@ -160,6 +160,18 @@ static const struct ov08x40_reg mipi_data_rate_800mbps[] = {
        {0x6002, 0x2e},
 };
 
+static const struct ov08x40_reg mipi_si_changed_regs[] = {
+       {0x481b, 0x2c}, /* HS Exit: Data Tx TEOT - reducing 8ns */
+       {0x4826, 0x42}, /* HS Entry: Data Tx TREOT - raising 8ns */
+       {0x4829, 0x54}, /* HS Exit: Data Tx TREOT - reducing 8ns */
+       {0x4885, 0x1f}, /* driving strength change */

Are these changes 'relative' to already written values?

Quoted 11 lines
drivers/media/i2c/ov08x40.c +7 -0
+};
+
+struct ov08x40_reg_list si_regs = {
+       .regs = mipi_si_changed_regs,
+       .num_of_regs = ARRAY_SIZE(mipi_si_changed_regs),
+};
+
 static const struct ov08x40_reg mode_3856x2416_regs[] = {
        {0x5000, 0x5d},
        {0x5001, 0x20},

drivers/media/i2c/ov08x40.c static int ov08x40_start_streaming(struct ov08x40 *ov08x) · line 2936

Quoted 8 lines
drivers/media/i2c/ov08x40.c +5 -0
@@ -2917,6 +2929,11 @@ static int ov08x40_start_streaming(struct ov08x40 *ov08x)
                return ret;
        }
 
+       /* Apply SI change to current project */
+       reg_list = &si_regs;
+
+       ov08x40_write_reg_list(ov08x, reg_list);
+

This looks odd. Why isn't this just:

	ov08x40_write_reg_list(0v08x, &si_regs);
Quoted 6 lines
drivers/media/i2c/ov08x40.c +0 -0
        /* Apply default values of current mode */
        reg_list = &ov08x->cur_mode->reg_list;
        ret = ov08x40_write_reg_list(ov08x, reg_list);
Signature · 1 line
2.34.1
#2 lore raw
Quoted 26 lines · Kieran Bingham

Quoting Chen, Jason Z (2023-10-20 05:09:08)

Quoted 6 lines

From: Jason Chen <[email protected]>

Due to certain MIPI hardware designs using overly long cables, there is a loss of signal strength, resulting in failed signal integration. The patch includes changes to adjust the driving strength in the register settings for a specific project.

For a specific project? Will this now affect 'all' projects using this sensor? Will that be adverse in anyway?

Do 'short' cables still function with this patch?

Quoted 13 lines
---
 drivers/media/i2c/ov08x40.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)
drivers/media/i2c/ov08x40.c +0 -0
index 0b3b906ebef..4f13e23b325 100644
@@ -160,6 +160,18 @@ static const struct ov08x40_reg mipi_data_rate_800mbps[] = {
        {0x6002, 0x2e},
 };
 
Quoted 7 lines

drivers/media/i2c/ov08x40.c static const struct ov08x40_reg mipi_data_rate_800mbps[] = { · line 167

Quoted 5 lines
drivers/media/i2c/ov08x40.c +5 -0
+static const struct ov08x40_reg mipi_si_changed_regs[] = {
+       {0x481b, 0x2c}, /* HS Exit: Data Tx TEOT - reducing 8ns */
+       {0x4826, 0x42}, /* HS Entry: Data Tx TREOT - raising 8ns */
+       {0x4829, 0x54}, /* HS Exit: Data Tx TREOT - reducing 8ns */
+       {0x4885, 0x1f}, /* driving strength change */

Are these changes 'relative' to already written values?

Looks like it's time to address MIPI PHY timings properly, by standardizing DT/ACPI device properties to convey timing information, and implementing helpers to parse those properties and calculate timing parameters for drivers.

Quoted 27 lines
Quoted 11 lines
drivers/media/i2c/ov08x40.c +7 -0
+};
+
+struct ov08x40_reg_list si_regs = {
+       .regs = mipi_si_changed_regs,
+       .num_of_regs = ARRAY_SIZE(mipi_si_changed_regs),
+};
+
 static const struct ov08x40_reg mode_3856x2416_regs[] = {
        {0x5000, 0x5d},
        {0x5001, 0x20},

drivers/media/i2c/ov08x40.c static int ov08x40_start_streaming(struct ov08x40 *ov08x) · line 2936

Quoted 8 lines
drivers/media/i2c/ov08x40.c +5 -0
@@ -2917,6 +2929,11 @@ static int ov08x40_start_streaming(struct ov08x40 *ov08x)
                return ret;
        }
 
+       /* Apply SI change to current project */
+       reg_list = &si_regs;
+
+       ov08x40_write_reg_list(ov08x, reg_list);
+

This looks odd. Why isn't this just:

	ov08x40_write_reg_list(0v08x, &si_regs);
Quoted 3 lines
drivers/media/i2c/ov08x40.c +0 -0
        /* Apply default values of current mode */
        reg_list = &ov08x->cur_mode->reg_list;
        ret = ov08x40_write_reg_list(ov08x, reg_list);
Signature · 3 lines
Regards,

Laurent Pinchart
#3 lore raw

Thanks for the review, Laurent and Kieran

Quoted 2 lines

For a specific project? Will this now affect 'all' projects using this sensor? Will that be adverse in anyway?

This was my initial concern with the patch very first, so I have added the quirk infrastructure to distinguish it using dmi_check_system. However, the sensor vendor replies that if another platform uses the same clock frequency, it shouldn't pose a significant issue.

Quoted 1 line

Do 'short' cables still function with this patch?

Yes, the change only slightly modifies the HS Tx TEOT/TREOT part. The value remains within the DPHY tolerance.

Quoted 1 line

Are these changes 'relative' to already written values?

It's mostly related to DPHY timing changes, as I mentioned in the comments. Therefore, it's independent of other sensor's capabilitiles like exposure time.

Quoted 1 line

Looks like it's time to address MIPI PHY timings properly, by standardizing DT/ACPI device properties to convey timing information, and implementing helpers to parse those properties and calculate timing parameters for drivers.

This is a positive direction. However, I'd like to provide some context.
This project encountered issues like insufficient driving strength and inadequate HS preparing time. The sensor vendor only provided relative registers to assist in debugging such issues. The detail are dependent on sensor vendor willingness to share this confidential information.
So I am not sure if other sensor vendors, like Sony, will provide such information for creating an universal API to access DPHY timing.

Quoted 1 line

This looks odd. Why isn't this just:

Thank you for this. I'll make the update.

-----Original Message-----
From: Laurent Pinchart <[email protected]>
Sent: Tuesday, October 24, 2023 8:49 AM
To: Kieran Bingham <[email protected]>
Cc: Chen, Jason Z <[email protected]>; Yeh, Andy <[email protected]>; [email protected]; [email protected]; [email protected]; Zhang, Qingwu <[email protected]>
Subject: Re: [PATCH v1] media: ov08x40: Add Signal Integration Adjustments for specific project

Quoted 26 lines · Kieran Bingham

Quoting Chen, Jason Z (2023-10-20 05:09:08)

Quoted 6 lines

From: Jason Chen <[email protected]>

Due to certain MIPI hardware designs using overly long cables, there is a loss of signal strength, resulting in failed signal integration. The patch includes changes to adjust the driving strength in the register settings for a specific project.

For a specific project? Will this now affect 'all' projects using this sensor? Will that be adverse in anyway?

Do 'short' cables still function with this patch?

Quoted 13 lines
---
 drivers/media/i2c/ov08x40.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)
(quoted diff) +0 -0

b/drivers/media/i2c/ov08x40.c index 0b3b906ebef..4f13e23b325 100644

drivers/media/i2c/ov08x40.c +0 -0
@@ -160,6 +160,18 @@ static const struct ov08x40_reg mipi_data_rate_800mbps[] = {
        {0x6002, 0x2e},
 };
 
Quoted 7 lines

drivers/media/i2c/ov08x40.c static const struct ov08x40_reg mipi_data_rate_800mbps[] = { · line 167

Quoted 5 lines
drivers/media/i2c/ov08x40.c +5 -0
+static const struct ov08x40_reg mipi_si_changed_regs[] = {
+       {0x481b, 0x2c}, /* HS Exit: Data Tx TEOT - reducing 8ns */
+       {0x4826, 0x42}, /* HS Entry: Data Tx TREOT - raising 8ns */
+       {0x4829, 0x54}, /* HS Exit: Data Tx TREOT - reducing 8ns */
+       {0x4885, 0x1f}, /* driving strength change */

Are these changes 'relative' to already written values?

Looks like it's time to address MIPI PHY timings properly, by standardizing DT/ACPI device properties to convey timing information, and implementing helpers to parse those properties and calculate timing parameters for drivers.

Quoted 21 lines
Quoted 10 lines
drivers/media/i2c/ov08x40.c +6 -0
+};
+
+struct ov08x40_reg_list si_regs = {
+       .regs = mipi_si_changed_regs,
+       .num_of_regs = ARRAY_SIZE(mipi_si_changed_regs), };
+
 static const struct ov08x40_reg mode_3856x2416_regs[] = {
        {0x5000, 0x5d},
        {0x5001, 0x20},

drivers/media/i2c/ov08x40.c static int ov08x40_start_streaming(struct ov08x40 *ov08x) · line 2936

Quoted 8 lines
drivers/media/i2c/ov08x40.c +5 -0
@@ -2917,6 +2929,11 @@ static int ov08x40_start_streaming(struct ov08x40 *ov08x)
                return ret;
        }
 
+       /* Apply SI change to current project */
+       reg_list = &si_regs;
+
+       ov08x40_write_reg_list(ov08x, reg_list);
+

This looks odd. Why isn't this just:

Quoted 5 lines
	ov08x40_write_reg_list(0v08x, &si_regs);
Quoted 3 lines
drivers/media/i2c/ov08x40.c +0 -0
        /* Apply default values of current mode */
        reg_list = &ov08x->cur_mode->reg_list;
        ret = ov08x40_write_reg_list(ov08x, reg_list);

--
Regards,

Laurent Pinchart