On Sat, Jun 15, 2024 at 11:50:18AM +0100, Jonathan Cameron wrote:
> On Fri, 14 Jun 2024 11:36:58 -0300
> João Paulo Gonçalves <jpaulo.silvagoncalves@xxxxxxxxx> wrote:
>
> > From: João Paulo Gonçalves <joao.goncalves@xxxxxxxxxxx>
> >
> > The condition for checking if triggers belong to the same IIO device to
> > set attached_own_device is currently inverted, causing
> > iio_trigger_using_own() to return an incorrect value. Fix it by testing
> > for the correct return value of iio_validate_own_trigger().
> >
> > Cc: stable@xxxxxxxxxxxxxxx
> > Fixes: 517985ebc531 ("iio: trigger: Add simple trigger_validation helper")
> > Signed-off-by: João Paulo Gonçalves <joao.goncalves@xxxxxxxxxxx>
Reviewed-by: Francesco Dolcini <francesco.dolcini@xxxxxxxxxxx>
>
> Ouch. Can you give an example of resulting user visible result? That
> will help people decide whether to pick this up for their distro kernels
> etc. In some cases, looks like we'll get garbage timestamps and in others
> may get stale data (or garbage).
This was noticed while me and Joao were working on the ads1119 driver you
have been recently reviewing. We wanted to use iio_trigger_using_own()
and it was not behaving the right way. We looked into it and found the bug.
Given that I do not know the exact impact on the drivers that are using this
function.
> Odd no one has noticed this in the past whilst testing those dependent
> features in particular drivers and I worry a little that we may have bugs
> in the users as a result of iio_trigger_using_own() reporting the inverse
> of the intended. I've take a quick look at the users and 'think' they are
> ok, but would definitely like a few others to confirm.
All the users of iio_trigger_using_own() are older than the commit that
introduced the bug, it is safe to assume that they need the fix and
are expecting the function to behave the same way is documented and it was
before the bug was introduced.
The broken commit is not that old and less than 10 IIO drivers are using this
function. Given that I think that is not that odd that it took 1 year to find
the bug.
Francesco
