[PATCH v2 1/3] backlight: tdo24m: fix the spi cs between transfers

[PATCH v2 1/3] backlight: tdo24m: fix the spi cs between transfers

[Robert Jarzmik]

Currently the LCD display (TD035S) on the cm-x300 platform is broken and
remains blank.

The TD0245S specification requires that the chipselect is toggled
between commands sent to the panel. This was also the purpose of the
former patch of commit f64dcac0b124 ("backlight: tdo24m: ensure chip
select changes between transfers").

Unfortunately, the "cs_change" field of a SPI transfer is
misleading. Its true meaning is that for a SPI message holding multiple
transfers, the chip select is toggled between each transfer, but for the
last transfer it remains asserted.

In this driver, all the SPI messages contain exactly one transfer, which
means that each transfer is the last of its message, and as a
consequence the chip select is never toggled.

Actually, there was a second bug hidding the first one, hence the
problem was not seen until v4.6. This problem was fixed by commit
a52db659c79c ("spi: pxa2xx: Fix cs_change management") for PXA based
boards.

This fix makes the TD035S work again on a cm-x300 board. The same
applies to other PXA boards, ie. corgi and tosa.

Fixes: a52db659c79c ("spi: pxa2xx: Fix cs_change management")
Reported-by: Andrea Adami <andrea.adami@gmail.com>
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
---
Since v1: added 2 other panels
---
 drivers/video/backlight/corgi_lcd.c | 2 +-
 drivers/video/backlight/tdo24m.c    | 2 +-
 drivers/video/backlight/tosa_lcd.c  | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/video/backlight/corgi_lcd.c b/drivers/video/backlight/corgi_lcd.c
index d7c239ea3d09..f5574060f9c8 100644
--- a/drivers/video/backlight/corgi_lcd.c
+++ b/drivers/video/backlight/corgi_lcd.c
@@ -177,7 +177,7 @@ static int corgi_ssp_lcdtg_send(struct corgi_lcd *lcd, int adrs, uint8_t data)
 	struct spi_message msg;
 	struct spi_transfer xfer = {
 		.len		= 1,
-		.cs_change	= 1,
+		.cs_change	= 0,
 		.tx_buf		= lcd->buf,
 	};
 
diff --git a/drivers/video/backlight/tdo24m.c b/drivers/video/backlight/tdo24m.c
index eab1f842f9c0..e4bd63e9db6b 100644
--- a/drivers/video/backlight/tdo24m.c
+++ b/drivers/video/backlight/tdo24m.c
@@ -369,7 +369,7 @@ static int tdo24m_probe(struct spi_device *spi)
 
 	spi_message_init(m);
 
-	x->cs_change = 1;
+	x->cs_change = 0;
 	x->tx_buf = &lcd->buf[0];
 	spi_message_add_tail(x, m);
 
diff --git a/drivers/video/backlight/tosa_lcd.c b/drivers/video/backlight/tosa_lcd.c
index 6a41ea92737a..4dc5ee8debeb 100644
--- a/drivers/video/backlight/tosa_lcd.c
+++ b/drivers/video/backlight/tosa_lcd.c
@@ -49,7 +49,7 @@ static int tosa_tg_send(struct spi_device *spi, int adrs, uint8_t data)
 	struct spi_message msg;
 	struct spi_transfer xfer = {
 		.len		= 1,
-		.cs_change	= 1,
+		.cs_change	= 0,
 		.tx_buf		= buf,
 	};
 
-- 
2.11.0

[PATCH v2 2/3] backlight: tdo24m: add the reset line gpio

[Robert Jarzmik]

The Toppoly panels have a global reset line. Add an optional gpio
control for this line, for platforms which have the ability to drive it.

Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
---
 drivers/video/backlight/tdo24m.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/video/backlight/tdo24m.c b/drivers/video/backlight/tdo24m.c
index e4bd63e9db6b..e4b2dfabf192 100644
--- a/drivers/video/backlight/tdo24m.c
+++ b/drivers/video/backlight/tdo24m.c
@@ -10,6 +10,7 @@
  */
 
 #include <linux/module.h>
+#include <linux/gpio/consumer.h>
 #include <linux/kernel.h>
 #include <linux/init.h>
 #include <linux/device.h>
@@ -32,6 +33,7 @@ struct tdo24m {
 	struct spi_message	msg;
 	struct spi_transfer	xfer;
 	uint8_t			*buf;
+	struct gpio_desc	*xres;
 
 	int (*adj_mode)(struct tdo24m *lcd, int mode);
 	int color_invert;
@@ -364,6 +366,7 @@ static int tdo24m_probe(struct spi_device *spi)
 	if (lcd->buf = NULL)
 		return -ENOMEM;
 
+	lcd->xres = devm_gpiod_get(&spi->dev, "reset", GPIOD_OUT_LOW);
 	m = &lcd->msg;
 	x = &lcd->xfer;
 
-- 
2.11.0

Re: [PATCH v2 2/3] backlight: tdo24m: add the reset line gpio

[Daniel Thompson]

On Fri, Oct 13, 2017 at 09:42:48PM +0200, Robert Jarzmik wrote:
> The Toppoly panels have a global reset line. Add an optional gpio
> control for this line, for platforms which have the ability to drive it.
> 
> Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
> ---
>  drivers/video/backlight/tdo24m.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/video/backlight/tdo24m.c b/drivers/video/backlight/tdo24m.c
> index e4bd63e9db6b..e4b2dfabf192 100644
> --- a/drivers/video/backlight/tdo24m.c
> +++ b/drivers/video/backlight/tdo24m.c
> @@ -10,6 +10,7 @@
>   */
>  
>  #include <linux/module.h>
> +#include <linux/gpio/consumer.h>
>  #include <linux/kernel.h>
>  #include <linux/init.h>
>  #include <linux/device.h>
> @@ -32,6 +33,7 @@ struct tdo24m {
>  	struct spi_message	msg;
>  	struct spi_transfer	xfer;
>  	uint8_t			*buf;
> +	struct gpio_desc	*xres;

Why do we need a write-only variable?


>  	int (*adj_mode)(struct tdo24m *lcd, int mode);
>  	int color_invert;
> @@ -364,6 +366,7 @@ static int tdo24m_probe(struct spi_device *spi)
>  	if (lcd->buf = NULL)
>  		return -ENOMEM;
>  
> +	lcd->xres = devm_gpiod_get(&spi->dev, "reset", GPIOD_OUT_LOW);

devm_gpiod_get_optional() would be better, if only for its commenting
value.

Also, this adds a new optional property, doesn't the devicetree binding 
docs need to be update to match this?


Daniel.


>  	m = &lcd->msg;
>  	x = &lcd->xfer;
>  
> -- 
> 2.11.0
> 

Re: [PATCH v2 2/3] backlight: tdo24m: add the reset line gpio

[Robert Jarzmik]

Daniel Thompson <daniel.thompson@linaro.org> writes:

> On Fri, Oct 13, 2017 at 09:42:48PM +0200, Robert Jarzmik wrote:
>> The Toppoly panels have a global reset line. Add an optional gpio
>> control for this line, for platforms which have the ability to drive it.
>> 
>> Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
>> ---
>>  drivers/video/backlight/tdo24m.c | 3 +++
>>  1 file changed, 3 insertions(+)
>> 
>> diff --git a/drivers/video/backlight/tdo24m.c b/drivers/video/backlight/tdo24m.c
>> index e4bd63e9db6b..e4b2dfabf192 100644
>> --- a/drivers/video/backlight/tdo24m.c
>> +++ b/drivers/video/backlight/tdo24m.c
>> @@ -10,6 +10,7 @@
>>   */
>>  
>>  #include <linux/module.h>
>> +#include <linux/gpio/consumer.h>
>>  #include <linux/kernel.h>
>>  #include <linux/init.h>
>>  #include <linux/device.h>
>> @@ -32,6 +33,7 @@ struct tdo24m {
>>  	struct spi_message	msg;
>>  	struct spi_transfer	xfer;
>>  	uint8_t			*buf;
>> +	struct gpio_desc	*xres;
>
> Why do we need a write-only variable?
It's written to by devm_gpiod_get() below.

>> +	lcd->xres = devm_gpiod_get(&spi->dev, "reset", GPIOD_OUT_LOW);
>
> devm_gpiod_get_optional() would be better, if only for its commenting
> value.
Yes, you're perfectly right.

> Also, this adds a new optional property, doesn't the devicetree binding 
> docs need to be update to match this?
Ah yes, that too. I'll add it for v3.

Cheers.

-- 
Robert

Re: [PATCH v2 2/3] backlight: tdo24m: add the reset line gpio

[Daniel Thompson]

On Sat, Dec 16, 2017 at 12:39:50PM +0100, Robert Jarzmik wrote:
> Daniel Thompson <daniel.thompson@linaro.org> writes:
> > On Fri, Oct 13, 2017 at 09:42:48PM +0200, Robert Jarzmik wrote:
> >> The Toppoly panels have a global reset line. Add an optional gpio
> >> control for this line, for platforms which have the ability to drive it.
> >> 
> >> Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
> >> ---
> >>  drivers/video/backlight/tdo24m.c | 3 +++
> >>  1 file changed, 3 insertions(+)
> >> 
> >> diff --git a/drivers/video/backlight/tdo24m.c b/drivers/video/backlight/tdo24m.c
> >> index e4bd63e9db6b..e4b2dfabf192 100644
> >> --- a/drivers/video/backlight/tdo24m.c
> >> +++ b/drivers/video/backlight/tdo24m.c
> >> @@ -10,6 +10,7 @@
> >>   */
> >>  
> >>  #include <linux/module.h>
> >> +#include <linux/gpio/consumer.h>
> >>  #include <linux/kernel.h>
> >>  #include <linux/init.h>
> >>  #include <linux/device.h>
> >> @@ -32,6 +33,7 @@ struct tdo24m {
> >>  	struct spi_message	msg;
> >>  	struct spi_transfer	xfer;
> >>  	uint8_t			*buf;
> >> +	struct gpio_desc	*xres;
> >
> > Why do we need a write-only variable?
> It's written to by devm_gpiod_get() below.

Yes, but it not *read* anywhere because we rely on devm to release the GPIO.
Why do we need a write-only variable?


> >> +	lcd->xres = devm_gpiod_get(&spi->dev, "reset", GPIOD_OUT_LOW);
> >
> > devm_gpiod_get_optional() would be better, if only for its commenting
> > value.
> Yes, you're perfectly right.
> 
> > Also, this adds a new optional property, doesn't the devicetree binding 
> > docs need to be update to match this?
> Ah yes, that too. I'll add it for v3.

Thanks!


Daniel.

Re: [PATCH v2 2/3] backlight: tdo24m: add the reset line gpio

[Robert Jarzmik]

Robert Jarzmik <robert.jarzmik@free.fr> writes:

> Daniel Thompson <daniel.thompson@linaro.org> writes:
>
>> On Fri, Oct 13, 2017 at 09:42:48PM +0200, Robert Jarzmik wrote:
>> Also, this adds a new optional property, doesn't the devicetree binding 
>> docs need to be update to match this?
> Ah yes, that too. I'll add it for v3.
Actually I won't ... because there is no description for tdo24m I could find.

If that's a problem for you, I'll drop that patch as well.

Cheers.

-- 
Robert

Re: [PATCH v2 2/3] backlight: tdo24m: add the reset line gpio

[Daniel Thompson]

On Sun, Dec 24, 2017 at 12:55:55PM +0100, Robert Jarzmik wrote:
> Robert Jarzmik <robert.jarzmik@free.fr> writes:
> 
> > Daniel Thompson <daniel.thompson@linaro.org> writes:
> >
> >> On Fri, Oct 13, 2017 at 09:42:48PM +0200, Robert Jarzmik wrote:
> >> Also, this adds a new optional property, doesn't the devicetree binding 
> >> docs need to be update to match this?
> > Ah yes, that too. I'll add it for v3.
> Actually I won't ... because there is no description for tdo24m I could find.
> 
> If that's a problem for you, I'll drop that patch as well.

I think I'm ok to drop this, if only because based on current upstream 
state I think this code will never be given a reset line anyway (or is
there a partner patch for mach-pxa that hasn't been posted yet).


Daniel.

[PATCH v2 3/3] backlight: tdo24m: add model and status sysfs entries

[Robert Jarzmik]

Add entries to query the panel ASIC to acquire the model number and the
status field. This comes in handy to check that the SPI connection is
functional and that the panel's firmware is alive.

Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
---
 drivers/video/backlight/tdo24m.c | 73 +++++++++++++++++++++++++++++++++++++++-
 1 file changed, 72 insertions(+), 1 deletion(-)

diff --git a/drivers/video/backlight/tdo24m.c b/drivers/video/backlight/tdo24m.c
index e4b2dfabf192..113c28c2145a 100644
--- a/drivers/video/backlight/tdo24m.c
+++ b/drivers/video/backlight/tdo24m.c
@@ -22,7 +22,7 @@
 
 #define POWER_IS_ON(pwr)	((pwr) <= FB_BLANK_NORMAL)
 
-#define TDO24M_SPI_BUFF_SIZE	(4)
+#define TDO24M_SPI_BUFF_SIZE	(6)
 #define MODE_QVGA	0
 #define MODE_VGA	1
 
@@ -33,6 +33,7 @@ struct tdo24m {
 	struct spi_message	msg;
 	struct spi_transfer	xfer;
 	uint8_t			*buf;
+	uint8_t			*rx_buf;
 	struct gpio_desc	*xres;
 
 	int (*adj_mode)(struct tdo24m *lcd, int mode);
@@ -219,6 +220,30 @@ static int tdo24m_writes(struct tdo24m *lcd, const uint32_t *array)
 	return err;
 }
 
+static int tdo24m_read_command(struct tdo24m *lcd, unsigned char cmd,
+			       unsigned int *id)
+{
+	struct spi_transfer *x = &lcd->xfer;
+	int err = 0, i;
+
+	lcd->buf[0] = cmd >> 1;
+	lcd->buf[1] = (cmd & 0x01) << 7;
+	lcd->buf[2] = 0;
+	lcd->buf[3] = 0;
+	lcd->buf[4] = 0;
+
+	x->len = 6;
+	err = spi_sync(lcd->spi_dev, &lcd->msg);
+
+	for (i = 0; i < sizeof(lcd->rx_buf) - 2; i++)
+		lcd->rx_buf[i] = (lcd->rx_buf[i + 1] << 2) |
+			(lcd->rx_buf[i + 2] >> 6);
+	*id = (lcd->rx_buf[0] << 24) | (lcd->rx_buf[1] << 16) |
+		(lcd->rx_buf[2] << 8) | lcd->rx_buf[3];
+
+	return err;
+}
+
 static int tdo24m_adj_mode(struct tdo24m *lcd, int mode)
 {
 	switch (mode) {
@@ -327,6 +352,43 @@ static int tdo24m_set_mode(struct lcd_device *ld, struct fb_videomode *m)
 	return lcd->adj_mode(lcd, mode);
 }
 
+static ssize_t tdo24m_model_show(struct device *dev,
+				 struct device_attribute *attr, char *buf)
+{
+	struct tdo24m *lcd = dev_get_drvdata(dev);
+	uint32_t data;
+	int err;
+
+	err = tdo24m_read_command(lcd, 0x04, &data);
+	if (!err)
+		return snprintf(buf, PAGE_SIZE, "0x%06x\n", data >> 8);
+
+	return err;
+}
+static DEVICE_ATTR_RO(tdo24m_model);
+
+static ssize_t tdo24m_status_show(struct device *dev,
+				  struct device_attribute *attr, char *buf)
+{
+	struct tdo24m *lcd = dev_get_drvdata(dev);
+	uint32_t data;
+	int err;
+
+	err = tdo24m_read_command(lcd, 0x09, &data);
+	if (!err)
+		return snprintf(buf, PAGE_SIZE, "0x%08x\n", data);
+
+	return err;
+}
+static DEVICE_ATTR_RO(tdo24m_status);
+
+static struct attribute *tdo24m_attrs[] = {
+	&dev_attr_tdo24m_model.attr,
+	&dev_attr_tdo24m_status.attr,
+	NULL,
+};
+ATTRIBUTE_GROUPS(tdo24m);
+
 static struct lcd_ops tdo24m_ops = {
 	.get_power	= tdo24m_get_power,
 	.set_power	= tdo24m_set_power,
@@ -365,6 +427,9 @@ static int tdo24m_probe(struct spi_device *spi)
 	lcd->buf = devm_kzalloc(&spi->dev, TDO24M_SPI_BUFF_SIZE, GFP_KERNEL);
 	if (lcd->buf = NULL)
 		return -ENOMEM;
+	lcd->rx_buf = devm_kzalloc(&spi->dev, TDO24M_SPI_BUFF_SIZE, GFP_KERNEL);
+	if (lcd->rx_buf = NULL)
+		return -ENOMEM;
 
 	lcd->xres = devm_gpiod_get(&spi->dev, "reset", GPIOD_OUT_LOW);
 	m = &lcd->msg;
@@ -374,6 +439,7 @@ static int tdo24m_probe(struct spi_device *spi)
 
 	x->cs_change = 0;
 	x->tx_buf = &lcd->buf[0];
+	x->rx_buf = &lcd->rx_buf[0];
 	spi_message_add_tail(x, m);
 
 	switch (model) {
@@ -396,6 +462,11 @@ static int tdo24m_probe(struct spi_device *spi)
 		return PTR_ERR(lcd->lcd_dev);
 
 	spi_set_drvdata(spi, lcd);
+
+	err = devm_device_add_groups(&spi->dev, tdo24m_groups);
+	if (err)
+		return err;
+
 	err = tdo24m_power(lcd, FB_BLANK_UNBLANK);
 	if (err)
 		return err;
-- 
2.11.0

Re: [PATCH v2 3/3] backlight: tdo24m: add model and status sysfs entries

[Daniel Thompson]

On Fri, Oct 13, 2017 at 09:42:49PM +0200, Robert Jarzmik wrote:
> Add entries to query the panel ASIC to acquire the model number and the
> status field. This comes in handy to check that the SPI connection is
> functional and that the panel's firmware is alive.
> 
> Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>

This feels to me somewhat like debug information being pushed into sysfs.
There are very few backlight drivers that add additional sysfs attributes
meaning the extra info here is not commonly pushed into the ABI for a
backlight and knowing the info in these properties does not really 
help userspace to *use* the hardware.

Does this kind of thing really belong in sysfs?


Daniel.

> ---
>  drivers/video/backlight/tdo24m.c | 73 +++++++++++++++++++++++++++++++++++++++-
>  1 file changed, 72 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/video/backlight/tdo24m.c b/drivers/video/backlight/tdo24m.c
> index e4b2dfabf192..113c28c2145a 100644
> --- a/drivers/video/backlight/tdo24m.c
> +++ b/drivers/video/backlight/tdo24m.c
> @@ -22,7 +22,7 @@
>  
>  #define POWER_IS_ON(pwr)	((pwr) <= FB_BLANK_NORMAL)
>  
> -#define TDO24M_SPI_BUFF_SIZE	(4)
> +#define TDO24M_SPI_BUFF_SIZE	(6)
>  #define MODE_QVGA	0
>  #define MODE_VGA	1
>  
> @@ -33,6 +33,7 @@ struct tdo24m {
>  	struct spi_message	msg;
>  	struct spi_transfer	xfer;
>  	uint8_t			*buf;
> +	uint8_t			*rx_buf;
>  	struct gpio_desc	*xres;
>  
>  	int (*adj_mode)(struct tdo24m *lcd, int mode);
> @@ -219,6 +220,30 @@ static int tdo24m_writes(struct tdo24m *lcd, const uint32_t *array)
>  	return err;
>  }
>  
> +static int tdo24m_read_command(struct tdo24m *lcd, unsigned char cmd,
> +			       unsigned int *id)
> +{
> +	struct spi_transfer *x = &lcd->xfer;
> +	int err = 0, i;
> +
> +	lcd->buf[0] = cmd >> 1;
> +	lcd->buf[1] = (cmd & 0x01) << 7;
> +	lcd->buf[2] = 0;
> +	lcd->buf[3] = 0;
> +	lcd->buf[4] = 0;
> +
> +	x->len = 6;
> +	err = spi_sync(lcd->spi_dev, &lcd->msg);
> +
> +	for (i = 0; i < sizeof(lcd->rx_buf) - 2; i++)
> +		lcd->rx_buf[i] = (lcd->rx_buf[i + 1] << 2) |
> +			(lcd->rx_buf[i + 2] >> 6);
> +	*id = (lcd->rx_buf[0] << 24) | (lcd->rx_buf[1] << 16) |
> +		(lcd->rx_buf[2] << 8) | lcd->rx_buf[3];
> +
> +	return err;
> +}
> +
>  static int tdo24m_adj_mode(struct tdo24m *lcd, int mode)
>  {
>  	switch (mode) {
> @@ -327,6 +352,43 @@ static int tdo24m_set_mode(struct lcd_device *ld, struct fb_videomode *m)
>  	return lcd->adj_mode(lcd, mode);
>  }
>  
> +static ssize_t tdo24m_model_show(struct device *dev,
> +				 struct device_attribute *attr, char *buf)
> +{
> +	struct tdo24m *lcd = dev_get_drvdata(dev);
> +	uint32_t data;
> +	int err;
> +
> +	err = tdo24m_read_command(lcd, 0x04, &data);
> +	if (!err)
> +		return snprintf(buf, PAGE_SIZE, "0x%06x\n", data >> 8);
> +
> +	return err;
> +}
> +static DEVICE_ATTR_RO(tdo24m_model);
> +
> +static ssize_t tdo24m_status_show(struct device *dev,
> +				  struct device_attribute *attr, char *buf)
> +{
> +	struct tdo24m *lcd = dev_get_drvdata(dev);
> +	uint32_t data;
> +	int err;
> +
> +	err = tdo24m_read_command(lcd, 0x09, &data);
> +	if (!err)
> +		return snprintf(buf, PAGE_SIZE, "0x%08x\n", data);
> +
> +	return err;
> +}
> +static DEVICE_ATTR_RO(tdo24m_status);
> +
> +static struct attribute *tdo24m_attrs[] = {
> +	&dev_attr_tdo24m_model.attr,
> +	&dev_attr_tdo24m_status.attr,
> +	NULL,
> +};
> +ATTRIBUTE_GROUPS(tdo24m);
> +
>  static struct lcd_ops tdo24m_ops = {
>  	.get_power	= tdo24m_get_power,
>  	.set_power	= tdo24m_set_power,
> @@ -365,6 +427,9 @@ static int tdo24m_probe(struct spi_device *spi)
>  	lcd->buf = devm_kzalloc(&spi->dev, TDO24M_SPI_BUFF_SIZE, GFP_KERNEL);
>  	if (lcd->buf = NULL)
>  		return -ENOMEM;
> +	lcd->rx_buf = devm_kzalloc(&spi->dev, TDO24M_SPI_BUFF_SIZE, GFP_KERNEL);
> +	if (lcd->rx_buf = NULL)
> +		return -ENOMEM;
>  
>  	lcd->xres = devm_gpiod_get(&spi->dev, "reset", GPIOD_OUT_LOW);
>  	m = &lcd->msg;
> @@ -374,6 +439,7 @@ static int tdo24m_probe(struct spi_device *spi)
>  
>  	x->cs_change = 0;
>  	x->tx_buf = &lcd->buf[0];
> +	x->rx_buf = &lcd->rx_buf[0];
>  	spi_message_add_tail(x, m);
>  
>  	switch (model) {
> @@ -396,6 +462,11 @@ static int tdo24m_probe(struct spi_device *spi)
>  		return PTR_ERR(lcd->lcd_dev);
>  
>  	spi_set_drvdata(spi, lcd);
> +
> +	err = devm_device_add_groups(&spi->dev, tdo24m_groups);
> +	if (err)
> +		return err;
> +
>  	err = tdo24m_power(lcd, FB_BLANK_UNBLANK);
>  	if (err)
>  		return err;
> -- 
> 2.11.0
> 

Re: [PATCH v2 3/3] backlight: tdo24m: add model and status sysfs entries

[Robert Jarzmik]

Daniel Thompson <daniel.thompson@linaro.org> writes:

> On Fri, Oct 13, 2017 at 09:42:49PM +0200, Robert Jarzmik wrote:
>> Add entries to query the panel ASIC to acquire the model number and the
>> status field. This comes in handy to check that the SPI connection is
>> functional and that the panel's firmware is alive.
>> 
>> Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
>
> This feels to me somewhat like debug information being pushed into sysfs.
> There are very few backlight drivers that add additional sysfs attributes
> meaning the extra info here is not commonly pushed into the ABI for a
> backlight and knowing the info in these properties does not really 
> help userspace to *use* the hardware.
>
> Does this kind of thing really belong in sysfs?
Ok, as you wish. Let's drop this patch from the serie for now.

Cheers.

--
Robert

Re: [PATCH v2 1/3] backlight: tdo24m: fix the spi cs between transfers

[Robert Jarzmik]

Robert Jarzmik <robert.jarzmik@free.fr> writes:

> Currently the LCD display (TD035S) on the cm-x300 platform is broken and
> remains blank.
... blablabla ...

Could I know when this serie can be cherry-picked please ?

Cheers.

--
Robert

Re: [PATCH v2 1/3] backlight: tdo24m: fix the spi cs between transfers

[Daniel Thompson]

On Thu, Dec 14, 2017 at 09:03:00PM +0100, Robert Jarzmik wrote:
> Robert Jarzmik <robert.jarzmik@free.fr> writes:
> 
> > Currently the LCD display (TD035S) on the cm-x300 platform is broken and
> > remains blank.
> ... blablabla ...
> 
> Could I know when this serie can be cherry-picked please ?

Sorry. This got put on the look-at-it-later pile (because I wanted to
think more about the sysfs attributed) but it got backlogged and buried.


> 
> Cheers.
> 
> --
> Robert

Re: [PATCH v2 1/3] backlight: tdo24m: fix the spi cs between transfers

[Daniel Thompson]

On Fri, Oct 13, 2017 at 09:42:47PM +0200, Robert Jarzmik wrote:
> Currently the LCD display (TD035S) on the cm-x300 platform is broken and
> remains blank.
> 
> The TD0245S specification requires that the chipselect is toggled
> between commands sent to the panel. This was also the purpose of the
> former patch of commit f64dcac0b124 ("backlight: tdo24m: ensure chip
> select changes between transfers").
> 
> Unfortunately, the "cs_change" field of a SPI transfer is
> misleading. Its true meaning is that for a SPI message holding multiple
> transfers, the chip select is toggled between each transfer, but for the
> last transfer it remains asserted.
> 
> In this driver, all the SPI messages contain exactly one transfer, which
> means that each transfer is the last of its message, and as a
> consequence the chip select is never toggled.
> 
> Actually, there was a second bug hidding the first one, hence the
> problem was not seen until v4.6. This problem was fixed by commit
> a52db659c79c ("spi: pxa2xx: Fix cs_change management") for PXA based
> boards.
> 
> This fix makes the TD035S work again on a cm-x300 board. The same
> applies to other PXA boards, ie. corgi and tosa.
> 
> Fixes: a52db659c79c ("spi: pxa2xx: Fix cs_change management")
> Reported-by: Andrea Adami <andrea.adami@gmail.com>
> Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>

Didn't I already ack this one?


Daniel.

> ---
> Since v1: added 2 other panels
> ---
>  drivers/video/backlight/corgi_lcd.c | 2 +-
>  drivers/video/backlight/tdo24m.c    | 2 +-
>  drivers/video/backlight/tosa_lcd.c  | 2 +-
>  3 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/video/backlight/corgi_lcd.c b/drivers/video/backlight/corgi_lcd.c
> index d7c239ea3d09..f5574060f9c8 100644
> --- a/drivers/video/backlight/corgi_lcd.c
> +++ b/drivers/video/backlight/corgi_lcd.c
> @@ -177,7 +177,7 @@ static int corgi_ssp_lcdtg_send(struct corgi_lcd *lcd, int adrs, uint8_t data)
>  	struct spi_message msg;
>  	struct spi_transfer xfer = {
>  		.len		= 1,
> -		.cs_change	= 1,
> +		.cs_change	= 0,
>  		.tx_buf		= lcd->buf,
>  	};
>  
> diff --git a/drivers/video/backlight/tdo24m.c b/drivers/video/backlight/tdo24m.c
> index eab1f842f9c0..e4bd63e9db6b 100644
> --- a/drivers/video/backlight/tdo24m.c
> +++ b/drivers/video/backlight/tdo24m.c
> @@ -369,7 +369,7 @@ static int tdo24m_probe(struct spi_device *spi)
>  
>  	spi_message_init(m);
>  
> -	x->cs_change = 1;
> +	x->cs_change = 0;
>  	x->tx_buf = &lcd->buf[0];
>  	spi_message_add_tail(x, m);
>  
> diff --git a/drivers/video/backlight/tosa_lcd.c b/drivers/video/backlight/tosa_lcd.c
> index 6a41ea92737a..4dc5ee8debeb 100644
> --- a/drivers/video/backlight/tosa_lcd.c
> +++ b/drivers/video/backlight/tosa_lcd.c
> @@ -49,7 +49,7 @@ static int tosa_tg_send(struct spi_device *spi, int adrs, uint8_t data)
>  	struct spi_message msg;
>  	struct spi_transfer xfer = {
>  		.len		= 1,
> -		.cs_change	= 1,
> +		.cs_change	= 0,
>  		.tx_buf		= buf,
>  	};
>  
> -- 
> 2.11.0
> 

Re: [PATCH v2 1/3] backlight: tdo24m: fix the spi cs between transfers

[Robert Jarzmik]

Daniel Thompson <daniel.thompson@linaro.org> writes:

> On Fri, Oct 13, 2017 at 09:42:47PM +0200, Robert Jarzmik wrote:
>> Currently the LCD display (TD035S) on the cm-x300 platform is broken and
>> remains blank.
> Didn't I already ack this one?
Ah yes, I forgot to report the Ack, sorry, I'll add it in the v3.

Cheers.

--
Robert