[bug report] iio: adc: add a driver for Qualcomm PM8xxx HK/XOADC

[bug report] iio: adc: add a driver for Qualcomm PM8xxx HK/XOADC

[Dan Carpenter]

[ No idea why this warning is showing up after 6 years. -dan ]

Hello Linus Walleij,

This is a semi-automatic email about new static checker warnings.

The patch 63c3ecd946d4: "iio: adc: add a driver for Qualcomm PM8xxx
HK/XOADC" from Apr 4, 2017, leads to the following Smatch complaint:

    drivers/iio/adc/qcom-pm8xxx-xoadc.c:769 pm8xxx_xoadc_parse_channel()
    error: we previously assumed 'hwchan' could be null (see line 761)

drivers/iio/adc/qcom-pm8xxx-xoadc.c
   760		hwchan = &hw_channels[0];

Here the hw_channels is variant->channels so it potentially is NULL
(maybe?  I don't know the context very well).

   761		while (hwchan && hwchan->datasheet_name) {
                       ^^^^^^
Check for NULL.

   762			if (hwchan->pre_scale_mux == pre_scale_mux &&
   763			    hwchan->amux_channel == amux_channel)
   764				break;
   765			hwchan++;
   766			chid++;
   767		}
   768		/* The sentinel does not have a name assigned */
   769		if (!hwchan->datasheet_name) {
                     ^^^^^^^^^^^^^^^^^^^^^^
Unchecked dereference.

   770			dev_err(dev, "could not locate channel %02x/%02x\n",
   771				pre_scale_mux, amux_channel);

regards,
dan carpenter

Re: [bug report] iio: adc: add a driver for Qualcomm PM8xxx HK/XOADC

[Linus Walleij]

On Wed, Mar 22, 2023 at 10:09 AM Dan Carpenter <error27@gmail.com> wrote:

> [ No idea why this warning is showing up after 6 years. -dan ]

:D

> This is a semi-automatic email about new static checker warnings.
>
> The patch 63c3ecd946d4: "iio: adc: add a driver for Qualcomm PM8xxx
> HK/XOADC" from Apr 4, 2017, leads to the following Smatch complaint:
>
>     drivers/iio/adc/qcom-pm8xxx-xoadc.c:769 pm8xxx_xoadc_parse_channel()
>     error: we previously assumed 'hwchan' could be null (see line 761)
>
> drivers/iio/adc/qcom-pm8xxx-xoadc.c
>    760          hwchan = &hw_channels[0];
>
> Here the hw_channels is variant->channels so it potentially is NULL
> (maybe?  I don't know the context very well).
>
>    761          while (hwchan && hwchan->datasheet_name) {
>                        ^^^^^^
> Check for NULL.

Actually there was a patch for this last week:
https://lore.kernel.org/linux-iio/20230314193709.15208-1-xhxgldhlpfy@gmail.com/

It seems the other static checker also found it now...

Yours,
Linus Walleij