Hi Nemoto-san,
On Mon, Aug 1, 2016 at 4:23 PM, Atsushi Nemoto <anemo@xxxxxxxxxxxxx> wrote:
> On Thu, 14 Apr 2016 21:06:05 +0200, Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> wrote:
>> I've just updated my old rbtx4927 from v3.13-rc3 to v4.6-rc3.
>> Surprisingly, it boots to nfsroot without any kernel changes.
>
> Good news! I'm happy to hear that.
>
>> However, there's an annoying warning in the boot log:
>>
>> gpiod_direction_output_raw: invalid GPIO
>>
>> This is caused by the following code in arch/mips/txx9/rbtx4927/setup.c:
>>
>> static void __init rbtx4927_mem_setup(void)
>> {
>>
>> /* TX4927-SIO DTR on (PIO[15]) */
>> gpio_request(15, "sio-dtr");
>>
>> returns -EPROBE_DEFER
>>
>> gpio_direction_output(15, 1);
>>
>> VALIDATE_DESC triggers the warning.
Note that as of commit 54d77198fdfbc4f0 ("gpio: bail out silently on NULL
descriptors"), the warning is gone. But the underlying issue isn't.
>> Probably a silly GPIO conversion was missed during the last 2+ years?
>
> Maybe txx9_gpio_init() failed to add gpio_chip since it was called too
> early on startup.
>
> Nowadays gpiochip_add_data calls kzalloc so cannot be called from
> plat_mem_setup context.
That's correct. It failed (silently) with -12, which is -ENOMEM...
> Could you try moving these txx9_gpio_init(), gpio_request() and
> gpio_direction_output() calls to rbtx4927_arch_init() or
> rbtx4927_device_init()?
Thanks, that works! Will send patches soon...
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
