Bug in omap3evm.c?

Bug in omap3evm.c?

[Rick Ball]

I think I've found a small problem in the board-omap3evm.c file under arch/arm/mach-omap2 (support for the TI/Mistral OMAP35x EVM board).

What I noticed is that the declaration for the array gpio_leds is initialized with one element (at line 380):

static struct gpio_led gpio_leds[] = {
        {
                .name                   = "omap3evm::ledb",
                /* normally not visible (board underside) */
                .default_trigger        = "default-on",
                .gpio                   = -EINVAL,      /* gets replaced */
                .active_low             = true,
        },
};
 
But then down at line 430 element 2 is set:

        /* TWL4030_GPIO_MAX + 1 == ledB (out, active low LED) */
        gpio_leds[2].gpio = gpio + TWL4030_GPIO_MAX + 1;

How did the array end up with 3 elements so that 2 would be a valid index?  It looks to me like it wouldn't flag an error, but would corrupt memory.  I'd submit a patch, but I'm not sure what this code was attempting to do...

Rick Ball

Re: Bug in omap3evm.c?

[Ranjith Lohithakshan]

On Tue, 11-May-10 4:38 AM +0530, Rick Ball wrote:
> I think I've found a small problem in the board-omap3evm.c file under arch/arm/mach-omap2 (support for the TI/Mistral OMAP35x EVM board).
> 
> What I noticed is that the declaration for the array gpio_leds is initialized with one element (at line 380):
> 
> static struct gpio_led gpio_leds[] = {
>         {
>                 .name                   = "omap3evm::ledb",
>                 /* normally not visible (board underside) */
>                 .default_trigger        = "default-on",
>                 .gpio                   = -EINVAL,      /* gets replaced */
>                 .active_low             = true,
>         },
> };
>  
> But then down at line 430 element 2 is set:
> 
>         /* TWL4030_GPIO_MAX + 1 == ledB (out, active low LED) */
>         gpio_leds[2].gpio = gpio + TWL4030_GPIO_MAX + 1;
> 
> How did the array end up with 3 elements so that 2 would be a valid index?  It looks to me like it wouldn't flag an error, but would corrupt memory.  I'd submit a patch, but I'm not sure what this code was attempting to do...

This appears to be like a bug. Possibly there would have been some
entries earlier which got removed, but the indexing done in the code
never got updated.

 - Ranjith