[4/5] i2c: designware: Add debug print for SDA hold time value

Message ID 20180529112754.13477-5-jarkko.nikula@linux.intel.com
State Superseded
Headers show
Series
  • i2c: designware: Improve debug prints
Related show

Commit Message

Jarkko Nikula May 29, 2018, 11:27 a.m.
SDA hold time is an important timing parameter and often reason for
arbitration lost errors if not set to a correct value. Add a debug print
for it in order to see what value gets programmed to a HW.

Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
---
 drivers/i2c/busses/i2c-designware-common.c | 4 ++++
 1 file changed, 4 insertions(+)

Comments

Andy Shevchenko May 29, 2018, 1:52 p.m. | #1
On Tue, 2018-05-29 at 14:27 +0300, Jarkko Nikula wrote:
> SDA hold time is an important timing parameter and often reason for
> arbitration lost errors if not set to a correct value. Add a debug
> print
> for it in order to see what value gets programmed to a HW.
> 
> Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
> ---
>  drivers/i2c/busses/i2c-designware-common.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/drivers/i2c/busses/i2c-designware-common.c
> b/drivers/i2c/busses/i2c-designware-common.c
> index 4f951486f74a..cb55d7e3af8e 100644
> --- a/drivers/i2c/busses/i2c-designware-common.c
> +++ b/drivers/i2c/busses/i2c-designware-common.c
> @@ -207,6 +207,10 @@ void i2c_dw_set_sda_hold(struct dw_i2c_dev *dev)
>  		 */
>  		if (!(dev->sda_hold_time & DW_IC_SDA_HOLD_RX_MASK))
>  			dev->sda_hold_time |= 1 <<
> DW_IC_SDA_HOLD_RX_SHIFT;
> +
> +		dev_dbg(dev->dev, "SDA Hold Time TX:RX = %d:%d\n",
> +			dev->sda_hold_time &
> ~(u32)DW_IC_SDA_HOLD_RX_MASK,

I'm not sure I understand why you need casting here.

> +			dev->sda_hold_time >>
> DW_IC_SDA_HOLD_RX_SHIFT);

>  	} else if (dev->sda_hold_time) {
>  		dev_warn(dev->dev,
>  			"Hardware too old to adjust SDA hold
> time.\n");
Jarkko Nikula May 30, 2018, 6:19 a.m. | #2
On 05/29/2018 04:52 PM, Andy Shevchenko wrote:
> On Tue, 2018-05-29 at 14:27 +0300, Jarkko Nikula wrote:
>> SDA hold time is an important timing parameter and often reason for
>> arbitration lost errors if not set to a correct value. Add a debug
>> print
>> for it in order to see what value gets programmed to a HW.
>>
>> Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
>> ---
>>   drivers/i2c/busses/i2c-designware-common.c | 4 ++++
>>   1 file changed, 4 insertions(+)
>>
>> diff --git a/drivers/i2c/busses/i2c-designware-common.c
>> b/drivers/i2c/busses/i2c-designware-common.c
>> index 4f951486f74a..cb55d7e3af8e 100644
>> --- a/drivers/i2c/busses/i2c-designware-common.c
>> +++ b/drivers/i2c/busses/i2c-designware-common.c
>> @@ -207,6 +207,10 @@ void i2c_dw_set_sda_hold(struct dw_i2c_dev *dev)
>>   		 */
>>   		if (!(dev->sda_hold_time & DW_IC_SDA_HOLD_RX_MASK))
>>   			dev->sda_hold_time |= 1 <<
>> DW_IC_SDA_HOLD_RX_SHIFT;
>> +
>> +		dev_dbg(dev->dev, "SDA Hold Time TX:RX = %d:%d\n",
>> +			dev->sda_hold_time &
>> ~(u32)DW_IC_SDA_HOLD_RX_MASK,
> 
> I'm not sure I understand why you need casting here.
> 
Me neither but gcc thinks otherwise in 64-bit build. It appears to 
convert that "(u32) & ~FOO" as u64 and complains "format ‘%d’ expects 
argument of type ‘int’, but argument 4 has type ‘long unsigned int’".

Sidenote, I need to change %d to %u.
Andy Shevchenko May 30, 2018, 1:49 p.m. | #3
On Wed, 2018-05-30 at 09:19 +0300, Jarkko Nikula wrote:
> On 05/29/2018 04:52 PM, Andy Shevchenko wrote:
> > On Tue, 2018-05-29 at 14:27 +0300, Jarkko Nikula wrote:

> > > +		dev_dbg(dev->dev, "SDA Hold Time TX:RX =
> > > %d:%d\n",
> > > +			dev->sda_hold_time &
> > > ~(u32)DW_IC_SDA_HOLD_RX_MASK,
> > 
> > I'm not sure I understand why you need casting here.

> Me neither but gcc thinks otherwise in 64-bit build. It appears to 
> convert that "(u32) & ~FOO" as u64 and complains "format ‘%d’ expects 
> argument of type ‘int’, but argument 4 has type ‘long unsigned int’".

Okay, we have, I think, the following options
 - (u32) explicit casting
 - lower_32_bits() macro call
 - defining mask with U type (not sure if ~ doesn't promote to UL)
 - defining negative mask explicitly as a number

All of them seems to me a compromise in one way or another (U suffix,
for example, will probably require to change all such definitions for
sake of consistency).

Patch

diff --git a/drivers/i2c/busses/i2c-designware-common.c b/drivers/i2c/busses/i2c-designware-common.c
index 4f951486f74a..cb55d7e3af8e 100644
--- a/drivers/i2c/busses/i2c-designware-common.c
+++ b/drivers/i2c/busses/i2c-designware-common.c
@@ -207,6 +207,10 @@  void i2c_dw_set_sda_hold(struct dw_i2c_dev *dev)
 		 */
 		if (!(dev->sda_hold_time & DW_IC_SDA_HOLD_RX_MASK))
 			dev->sda_hold_time |= 1 << DW_IC_SDA_HOLD_RX_SHIFT;
+
+		dev_dbg(dev->dev, "SDA Hold Time TX:RX = %d:%d\n",
+			dev->sda_hold_time & ~(u32)DW_IC_SDA_HOLD_RX_MASK,
+			dev->sda_hold_time >> DW_IC_SDA_HOLD_RX_SHIFT);
 	} else if (dev->sda_hold_time) {
 		dev_warn(dev->dev,
 			"Hardware too old to adjust SDA hold time.\n");