Messages in this thread | | | Date | Wed, 5 Jun 2013 16:41:44 +0300 | Subject | Re: [PATCH v1.1] gpiolib: append SFI helpers for GPIO API | From | Andy Shevchenko <> |
| |
On Mon, Jun 3, 2013 at 3:26 PM, Joe Perches <joe@perches.com> wrote: > On Mon, 2013-06-03 at 11:16 +0300, Andy Shevchenko wrote: >> On Sun, 2013-06-02 at 19:06 -0700, Joe Perches wrote: >> > On Mon, 2013-06-03 at 11:59 +1000, Ryan Mallon wrote: >> > > On 31/05/13 19:27, Andy Shevchenko wrote: >> > > > + sfi_gpio_table = kmalloc(num * sizeof(*pentry), GFP_KERNEL); >> > > Use kcalloc when you have a size and a count. >> > kmalloc_array unless you want zero'ed memory >> One proposed to use kmemdup to clean up a bit. What do you think? > > Seems correct to use kmemdup but what validates num?
num comes from SFI firmware. We only could rely on what kmalloc/kmemdup returns.
> It seems like a possible error path exists there regardless > of whether kmemdup is used or not.
That's true. The advantage of kmemdup is an implicit memcpy. Thus, we could drop away that line.
-- With Best Regards, Andy Shevchenko
| |