On 6/03/24 03:57, Andy Shevchenko wrote: > On Tue, Mar 05, 2024 at 09:23:07AM +0100, Geert Uytterhoeven wrote: >> On Tue, Mar 5, 2024 at 4:59 AM Chris Packham >> <chris.packham@xxxxxxxxxxxxxxxxxxx> wrote: > ... > >>> + priv->segment_gpios = devm_gpiod_get_array(dev, "segment", GPIOD_OUT_LOW); >>> + if (IS_ERR(priv->segment_gpios)) >>> + return PTR_ERR(priv->segment_gpios); >> This needs some validation of priv->segment_gpios->ndescs, else the >> call to gpiod_set_array_value_cansleep() in seg_led_update() may >> trigger an out-of-bounds access of the values bitmap. > Alternatively we can call gpiod_count() beforehand and check its result. Unless there are any objections I think I'll go with the ndescs check as it'll be easier to update to the subnode style in the future. It does mean there will be some extra allocations/frees (handled via the devm_ APIs) in the error case.
