Re: getting started w/ iio driver

Re: getting started w/ iio driver

[Enrico Weigelt, metux IT consult]

On 04.11.2017 21:38, Jonathan Cameron wrote:

>> How should I implement the sampling rate setting w/ separate devices ?
>> Maybe some master-slave concept where settings on the master
>> automatically affect the slaves ?
> 
> Yes it will be a bit ugly. Probably do it as a set of peers where writing any will change them all.

Ok. By the way: how to set the sampling-rate via iio ?
Is there a callback or device field for that ?

where should i tell the adc to start capture w/ ?
in pre- or post-enable buffer ops ?


--mtx


-- 
Enrico Weigelt, metux IT consult
Free software and Linux embedded engineering
info@metux.net -- +49-151-27565287

Re: getting started w/ iio driver

[Jonathan Cameron]

On Sat, 4 Nov 2017 21:51:34 +0100
"Enrico Weigelt, metux IT consult" <info@metux.net> wrote:

> On 04.11.2017 21:38, Jonathan Cameron wrote:
> 
> >> How should I implement the sampling rate setting w/ separate devices ?
> >> Maybe some master-slave concept where settings on the master
> >> automatically affect the slaves ?  
> > 
> > Yes it will be a bit ugly. Probably do it as a set of peers where writing any will change them all.  

Sorry, been a busy week - now have beer and am on right side of airport security ;)
> 
> Ok. By the way: how to set the sampling-rate via iio ?
> Is there a callback or device field for that ?
Yes - the raw_write callback will do it but you will also need to
add the IIO_CHAN_SPEC_SAMP_FREQ bit to the appropriate 
infomask_shared_by*
> 

> where should i tell the adc to start capture w/ ?
> in pre- or post-enable buffer ops ?

Conceptually in the post-enable.  The pre / post divide is
rather vague now as it the original plan to do it nice and cleanly
was flawed by some race conditions.  Now they both occur in a very
similar state...
> 
> 
> --mtx
> 
> 

Re: getting started w/ iio driver

[Phil Reid]

On 3/11/2017 11:12, Enrico Weigelt, metux IT consult wrote:
> Hi folks,
> 
> i'm currently developing an iio driver for a custom adc and looking for
> the optimal approach.
> 
> The device is made up of 3x 24bit-SD-ADCs and a fpga (as buffer+bridge)
> connected via EIM. All ADCs operate at the same (configurable, up to
> 500khz) sampling rate and each have their own hw buffer. For each buffer
> there's a separate interrupt line.
> 
> So, i'd need an irq handler, which fetches the data into a suitable
> buffer and push it into iio. is that what iio_triggered_buffer_setup()
> etc is for ? How does that relate to triggers ?
> 
I have something very similar. Multiple SD ADC's connected to an FPGA.

Multiple ADC channels clocked with common sample rate.

My approach was to have a master IIO device that controls common parameters
(sample rate etc) and an IIO device per channel.
I wanted the channel data to be separated into a separate stream and the
ability to start / stop channels independently.
The master device creates the child devices nd manages an interdependencies..

FGPA writes directly to memory buffers (FPGA is the DMA master)
The IIO system can provide a queue of memory buffers to fill.
And then in the hard IRQ it updates the target address.

I based the DMA stuff on Lars-Peter's dummy dma buffer example.
http://comments.gmane.org/gmane.linux.kernel.iio/20283

The dma buffer code is in mainline by I don't think the example driver is.




-- 
Regards
Phil Reid