On Thu, Aug 17, 2017 at 01:27:38PM +0200, Pavel Machek wrote:
> Hi!
>
> > Hello Nate Case,
> >
> > The patch f46e9203d9a1: "leds: Add support for Philips PCA955x I2C
> > LED drivers" from Jul 16, 2008, leads to the following static checker
> > warning:
> >
> > drivers/leds/leds-pca955x.c:476 pca955x_probe()
> > warn: this array is probably non-NULL. 'pdata->leds + i->name'
> >
> > drivers/leds/leds-pca955x.c
> > 465 switch (pca955x_led->type) {
> > 466 case PCA955X_TYPE_NONE:
> > 467 break;
> > 468 case PCA955X_TYPE_GPIO:
> > 469 ngpios++;
> > 470 break;
> > 471 case PCA955X_TYPE_LED:
> > 472 /*
> > 473 * Platform data can specify LED names and
> > 474 * default triggers
> > 475 */
> > 476 if (pdata->leds[i].name)
> > ^^^^^^^^^^^^^^^^^^^
> > The comment implies that we should be testing pdata->leds[i].name[0] to
> > see if any string has been set?
>
> Someone was already submitting patch from this one, no?
>
> And please don't mark this one as a "bug report". There's no bug. Your
> code analysis tool found a way to make kernel code shorter... well, so
> what?
>
You didn't read my email. It was only one sentence long... :/
>From the comments, it looked like maybe a different test was intended.
Anyway, that's the point of these warnings is because many times the
bogus NULL test should be replaced with a correct test. I don't forward
the warning if it's obvious that the test should just be deleted.
I probably could review the warnings even more but I am pretty busy and
the code is obviously bogus and it's easier for the author.
regards,
dan carpenter
