Re: BBB IIO ADC access not working

[Michael Welling <mwelling@ieee.org>]

On Wed, Aug 19, 2015 at 01:35:10PM -0700, Greg Wilson-Lindberg wrote:
>  
> 
> > -----Original Message-----
> > From: Michael Welling [mailto:mwelling79@gmail.com] On Behalf 
> > Of Michael Welling
> > Sent: Wednesday, August 19, 2015 10:53 AM
> > To: Greg Wilson-Lindberg
> > Cc: Daniel Baluta; linux-iio@vger.kernel.org
> > Subject: Re: BBB IIO ADC access not working
> > 
> > On Wed, Aug 19, 2015 at 09:27:02AM -0700, Greg Wilson-Lindberg wrote:
> > >  
> > > 
> > > > -----Original Message-----
> > > > From: Michael Welling [mailto:mwelling79@gmail.com] On Behalf Of 
> > > > Michael Welling
> > > > Sent: Tuesday, August 18, 2015 3:31 PM
> > > > To: Greg Wilson-Lindberg
> > > > Cc: Daniel Baluta; linux-iio@vger.kernel.org
> > > > Subject: Re: BBB IIO ADC access not working
> > > > 
> > > > On Tue, Aug 18, 2015 at 03:11:13PM -0700, Greg 
> > Wilson-Lindberg wrote:
> > > > >  
> > > > > 
> > > > > > I found that handling the IRQENB_EOS interrupt in the ADC
> > > > is causing
> > > > > > an undesired effect. After enabling the ADC buffer mode and 
> > > > > > disabling it the coordinate readings work but the touch
> > > > release is
> > > > > > no longer detected.
> > > > > 
> > > > > Can we put your code in the touch screen handler? The tsc
> > > > code is writing to the IRQSTATUS register, not the IRQCLR 
> > register.
> > > > > 
> > > > 
> > > > Actually using the IRQCLR disables the IRQ instead of clearing it 
> > > > and that was not my intent. This is why the pen releases 
> > where not 
> > > > working, the tsc driver relies on the EOS interrupt to write the 
> > > > step register because it is using a one shot hardware mode.
> > > > 
> > > > Since the IRQENB_EOS is not used by the ADC driver it can be 
> > > > disabled when in buffered mode which is triggering faster 
> > than the 
> > > > isr can handle it.
> > > > 
> > > > See patch below.
> > > > 
> > > > Though it allows for the repeated enabling and disabling of the 
> > > > buffer the touchscreen gets a bunch of unexpected 
> > releases and the 
> > > > readings are very jittery.
> > > > 
> > > > diff --git a/drivers/iio/adc/ti_am335x_adc.c 
> > > > b/drivers/iio/adc/ti_am335x_adc.c index a0e7161..9945e4a 100644
> > > > --- a/drivers/iio/adc/ti_am335x_adc.c
> > > > +++ b/drivers/iio/adc/ti_am335x_adc.c
> > > > @@ -177,6 +177,7 @@ static int 
> > tiadc_buffer_preenable(struct iio_dev 
> > > > *indio_dev)
> > > >  				IRQENB_FIFO1OVRRUN |
> > > >  				IRQENB_FIFO1UNDRFLW));
> > > >  
> > > > +	tiadc_writel(adc_dev, REG_IRQCLR, IRQENB_EOS);
> > > >  	/* Flush FIFO. Needed in corner cases in simultaneous 
> > tsc/adc use 
> > > > */
> > > >  	fifo1count = tiadc_readl(adc_dev, REG_FIFO1CNT);
> > > >  	for (i = 0; i < fifo1count; i++)
> > > > @@ -213,6 +214,7 @@ static int tiadc_buffer_predisable(struct 
> > > > iio_dev *indio_dev)
> > > >  
> > > >  	tiadc_writel(adc_dev, REG_IRQCLR, (IRQENB_FIFO1THRES |
> > > >  				IRQENB_FIFO1OVRRUN |
> > > > IRQENB_FIFO1UNDRFLW));
> > > > +	tiadc_writel(adc_dev, REG_IRQENABLE, IRQENB_EOS);
> > > >  	am335x_tsc_se_clr(adc_dev->mfd_tscadc,
> > > > adc_dev->buffer_en_ch_steps);
> > > >  	adc_dev->buffer_en_ch_steps = 0;
> > > >  
> > > > 
> > > 
> > > I'm not sure that it's better, but it is different.  The 
> > touch screen still locks up, but mouse clicks work.
> > 
> > This is very disappointing.
> > 
> > Have you tried reading the ADCs using one shot mode recently?
> > 
> > IE cat /sys/bus/iio/devices/iio:device0/in_voltageX_raw
> 
> I tried that after the system had locked up and what I get back is the Device
> or resource is busy.  I restarted the BBB and tried before starting my application
> and it worked just fine. I started up the app and the touch screen was 
> locked up immediately, mouse clicks delayed about 1/2 second. And I get Device or resource busy from the cat.
>

Is the using while using /sys/bus/iio/devices/iio:device0/in_voltageX_raw in your program?
 
> > 
> > > 
> > > > 
> > > > 
> > > > 
> >