>perhaps have it dump_stack() as if a conflict did exist ? sound good
>to you Michael ?
Yeah have it make noise.
But there is another problem at one spot.
static int __init bfin_gpio_init(void)
{
str_ident = kcalloc(MAX_RESOURCES,
sizeof(struct str_ident), GFP_KERNEL);
if (str_ident == NULL)
return -ENOMEM;
memset(str_ident, 0, MAX_RESOURCES * sizeof(struct str_ident));
printk(KERN_INFO "Blackfin GPIO Controller\n");
return 0;
}
arch_initcall(bfin_gpio_init);
early serial init also utilizes the peripheral request api - however at
that point bfin_gpio_init didn't allocate memory for the labels.
So we always have two zombies (allocated pin functions without labels)
Zero pointers are handled - so we don't see issues.
The only workaround I could think of is to statically allocate memory
holding the labels.
-Michael
>-----Original Message-----
>From: Mike Frysinger [mailto:[EMAIL PROTECTED]
>Sent: Donnerstag, 6. Dezember 2007 16:22
>To: Michael Hennerich
>Cc: [EMAIL PROTECTED];
[email protected];
>Wu, Bryan
>Subject: Re: [Linux-kernel-commits]
[3995]trunk/sound/soc/blackfin/bf5xx-
>ac97.c: bug[#3725]for BF54X,GPIO 19 label should be NULL
>
>On Dec 6, 2007 4:20 AM, Wu, Bryan <[EMAIL PROTECTED]> wrote:
>> 1. gpio_request and peripheral_request should check (label != NULL)
>
>perhaps have it dump_stack() as if a conflict did exist ? sound good
>to you Michael ?
>-mike
_______________________________________________
Linux-kernel-commits mailing list
[email protected]
http://blackfin.uclinux.org/mailman/listinfo/linux-kernel-commits
