Re: [PATCH v2 5/8] MIPS: jz4740: dts: Add bindings for the jz4740-wdt driver

[Paul Cercueil <paul@crapouillou.net>]

Hi PrasannaKumar,

Le mar. 2 janv. 2018 à 17:37, PrasannaKumar Muralidharan 
<prasannatsmkumar@gmail.com> a écrit :
> Hi Paul,
> 
> On 30 December 2017 at 19:21, Paul Cercueil <paul@crapouillou.net> 
> wrote:
>>  Also remove the watchdog platform_device from platform.c, since it
>>  wasn't used anywhere anyway.
>> 
>>  Signed-off-by: Paul Cercueil <paul@crapouillou.net>
>>  ---
>>   arch/mips/boot/dts/ingenic/jz4740.dtsi |  8 ++++++++
>>   arch/mips/jz4740/platform.c            | 16 ----------------
>>   2 files changed, 8 insertions(+), 16 deletions(-)
>> 
>>   v2: No change
>> 
>>  diff --git a/arch/mips/boot/dts/ingenic/jz4740.dtsi 
>> b/arch/mips/boot/dts/ingenic/jz4740.dtsi
>>  index cd5185bb90ae..26c6b561d6f7 100644
>>  --- a/arch/mips/boot/dts/ingenic/jz4740.dtsi
>>  +++ b/arch/mips/boot/dts/ingenic/jz4740.dtsi
>>  @@ -45,6 +45,14 @@
>>                  #clock-cells = <1>;
>>          };
>> 
>>  +       watchdog: watchdog@10002000 {
>>  +               compatible = "ingenic,jz4740-watchdog";
>>  +               reg = <0x10002000 0x10>;
>>  +
>>  +               clocks = <&cgu JZ4740_CLK_RTC>;
>>  +               clock-names = "rtc";
>>  +       };
>>  +
> 
> The watchdog driver calls jz4740_timer_enable_watchdog and
> jz4740_timer_disable_watchdog which defined in
> arch/mips/jz4740/timer.c. It accesses registers iomapped by timer
> code. Declaring register size as 0x10 does not show the real picture.
> Better use register size as 0x100 and let timer, wdt, pwm drivers to
> share them.

As you said, it accesses registers iomapped by timer code. So the 
watchdog
driver doesn't need to iomap them.

> Code from one of your branches
> (https://github.com/OpenDingux/linux/blob/for-upstream-clocksource/arch/mips/boot/dts/ingenic/jz4740.dtsi)
> does it. Can you prepare a patch series and send it?
> I have a patch set that moves timer code out of arch/mips/jz4740/ and
> does a similar thing for watchdog and pwm. As your new timer driver is
> better than the existing one I have not sent my patches yet. I would
> like to see it getting mainlined as it paves way for removing most of
> code in arch/mips/jz4740.

The whole 'for-upstream-clocksource' branch is supposed to go upstream,
but I can't do it in one big patchset without having lots of breakages 
with
my other patchsets (jz4770 SoC support, and jz4740 watchdog updates)
currently under review. That also makes it simpler to upstream than 
having
one single patchset that touches 6 different frameworks (MIPS, irq, 
clocks,
clocksource, watchdog, PWM).

So I will submit it in two steps, first the irq/clocks/clocksource 
drivers
(this patchset) hopefully for 4.16, and then the platform/watchdog/PWM 
fixes
for 4.17.

Kind regards,
-Paul