Re: [PATCH 10/16] iio: adc: max1027: Prevent single channel accesses during buffer reads

[Miquel Raynal]

Hi Jonathan,

Jonathan Cameron <jic23@xxxxxxxxxx> wrote on Mon, 30 Aug 2021 11:20:24
+0100:

> On Fri, 20 Aug 2021 07:30:07 +0000
> "Sa, Nuno" <Nuno.Sa@xxxxxxxxxx> wrote:
> 
> > > -----Original Message-----
> > > From: Sa, Nuno <Nuno.Sa@xxxxxxxxxx>
> > > Sent: Friday, August 20, 2021 9:21 AM
> > > To: Miquel Raynal <miquel.raynal@xxxxxxxxxxx>; Jonathan Cameron
> > > <jic23@xxxxxxxxxx>; Lars-Peter Clausen <lars@xxxxxxxxxx>
> > > Cc: Thomas Petazzoni <thomas.petazzoni@xxxxxxxxxxx>; linux-
> > > iio@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx
> > > Subject: RE: [PATCH 10/16] iio: adc: max1027: Prevent single channel
> > > accesses during buffer reads
> > > 
> > > [External]
> > > 
> > > 
> > >     
> > > > -----Original Message-----
> > > > From: Miquel Raynal <miquel.raynal@xxxxxxxxxxx>
> > > > Sent: Wednesday, August 18, 2021 1:12 PM
> > > > To: Jonathan Cameron <jic23@xxxxxxxxxx>; Lars-Peter Clausen
> > > > <lars@xxxxxxxxxx>
> > > > Cc: Thomas Petazzoni <thomas.petazzoni@xxxxxxxxxxx>; linux-
> > > > iio@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Miquel Raynal
> > > > <miquel.raynal@xxxxxxxxxxx>
> > > > Subject: [PATCH 10/16] iio: adc: max1027: Prevent single channel
> > > > accesses during buffer reads
> > > >
> > > > [External]
> > > >
> > > > When hardware buffers are enabled (the cnvst pin being the    
> > > trigger),    
> > > > one
> > > > should not mess with the device state by requesting a single channel
> > > > read. Prevent it with a iio_buffer_enabled() check.
> > > >
> > > > Signed-off-by: Miquel Raynal <miquel.raynal@xxxxxxxxxxx>
> > > > ---
> > > >  drivers/iio/adc/max1027.c | 2 ++
> > > >  1 file changed, 2 insertions(+)
> > > >
> > > > diff --git a/drivers/iio/adc/max1027.c b/drivers/iio/adc/max1027.c
> > > > index 223c9e4abd86..83526f3d7d3a 100644
> > > > --- a/drivers/iio/adc/max1027.c
> > > > +++ b/drivers/iio/adc/max1027.c
> > > > @@ -335,6 +335,8 @@ static int max1027_read_raw(struct iio_dev
> > > > *indio_dev,
> > > >
> > > >  	switch (mask) {
> > > >  	case IIO_CHAN_INFO_RAW:
> > > > +		if (iio_buffer_enabled(indio_dev))
> > > > +			return -EBUSY;    
> > > 
> > > I guess 'iio_device_claim_direct_mode()' would be a better option
> > > here? There's nothing preventing this check to pass and then,
> > > concurrently
> > > someone enables the buffer...
> > >     
> > 
> > Taking a second look, it seems that this check is already done [1]? Am I missing
> > I missing something?

You're right, I missed that too.

> > Also, I think we are returning with the 'st->lock' held...
> > 
> > [1]: https://elixir.bootlin.com/linux/latest/source/drivers/iio/adc/max1027.c#L247  
> Absolutely agree this should be done with iio_device_claim_direct_mode() to close the
> possible races.

Didn't know this helper, nice.

> I wonder why this one has been missed in all the cleanups of that stuff? Looks like
> a simple case, but I guess it wasn't immediately visible in the read_raw() function
> so no one noticed.
> 
> Jonathan
> 

Thanks,
Miquèl