Re: [Xenomai] Sensoray 626 analogy driver

[Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org>]

On 03/24/2014 01:09 PM, Wojciech Domski wrote:
> W dniu 22.03.2014 17:36, Gilles Chanteperdrix pisze:
>> First, the coding style. You use mixed-case identifiers, whereas the
>> tradition is to use lower case with underscores, and you do not seem to
>> indent with tabs, which is also the rule for kernel code. Your code also
>> contains some code in commentaries, which makes it uselessly hard to read.
> Yes, this needs my attention. However, it is not critical.

Not critical, but mandatory.

>>
>> Second, I see a lot of busy wait loops, for instance for things such as
>> I2C transfers, but I2C transfers are usually slow, so busy waiting for
>> the end of an I2C transfer in a critical section (as the comments say),
>> will create large masking sections for the rest of the system, which is
>> unacceptable for Xenomai, as we try hard to guarantee low latencies.
> Yes, you have right. Could you elaborate on this a bit more?
> Inside the kernel module there are plenty of while loops. Could you tell 
> me if I
> could use  rtdm_task_busy_sleep() function inside those loops?
> 
> Would such solution be acceptable inside Xenomai kernel module?

The solution which would be acceptable is not to have busy waits, except
for very short durations. But for instance transferring a byte on I2C
takes around 20us at 400 kHz, a 20us masking section is unacceptable.
rtdm_task_busy_sleep, as the name suggests is a busy wait loop, so, no,
it is not acceptable either.

Use a thread or a timer to reschedule while you wait for the end of the
transfer, instead of busy waiting.


-- 
                                                                Gilles.