[PATCHv3 2/3] ARM: mxs: cfa10049: Switch bus i2c1 to bitbanging

[Alexandre Belloni]

On 02/07/2013 18:33, Marek Vasut wrote:
> Dear Alexandre Belloni,
> 
>> On 02/07/2013 13:50, Alexandre Belloni wrote:
>>> On 02/07/2013 13:45, Fabio Estevam wrote:
>>>> Shouldn't this be
>>>>
>>>>          i2c at 1 {
>>>>          
>>>>                  reg = <1>; ?
>>>
>>> No, we have 4 devices on that mux and 2 pins to select the muxing.
>>
>> OK, got it working.
>>
>> So, the results:
>>
>> bitbanging:
>>
>> # time cat /sys/bus/iio/devices/iio\:device1/in_voltage0_raw
>> 2637
>> real	0m 0.09s
>> user	0m 0.01s
>> sys	0m 0.01s
>>
>>
>> i2c-mxs PIO mode:
>>
>> # time cat /sys/bus/iio/devices/iio\:device1/in_voltage0_raw
>> [   35.007650] [sched_delayed] sched: RT throttling activated
>> 2627
>> real	0m 7.14s
>> user	0m 0.02s
>> sys	0m 0.01s
>>
>>
>> i2c-mxs PIO mode without LRADC:
>>
>> # time cat /sys/bus/iio/devices/iio\:device1/in_voltage0_raw
>> [   18.007432] [sched_delayed] sched: RT throttling activated
>> 2629
>> real	0m 7.09s
>> user	0m 0.00s
>> sys	0m 0.03s
>>
>>
>> i2c-mxs DMA mode:
>>
>> # time cat /sys/bus/iio/devices/iio\:device1/in_voltage0_raw
>> 2631
>> real	0m 0.12s
>> user	0m 0.01s
>> sys	0m 0.01s
>>
>>
>> It seems fine for me.
> 
> I think I'm getting a little lost in these gazilions of i2c and lradc threads. 
> Can we not create one thread and keep the related stuff in there instead of 
> discussing it all around !?
> 
> Only one question comes to mind with this email -- what do LRADC and I2C have to 
> do with each other here ?
> 

Yeah, sorry, I meant the lradc touchscreen support. This seemed to
trigger the issue for Fabio but as my testing shows, this is not the
case for me, I get the issue with PIO, whether the lradc touchscreen
support is activated or not.

I think Torsten is the one that investigated it the most :

http://www.spinics.net/lists/linux-i2c/msg12619.html


> It'd be nice if someone could summarize on what I should focus and possibly 
> prepare a testcase.
> 

On my setup, it happens on every i2c read that are done in PIO mode.
But, my setup may be a bit unconventional as we are using a i2c gpio muxer.

Regards,

-- 
Alexandre Belloni, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com