On 19/10/2022 22:40, Martin Blumenstingl wrote:
Hi Anand,
On Wed, Oct 19, 2022 at 7:17 PM Anand Moon <linux.amoon@xxxxxxxxx> wrote:
[...]
+&pwm_AO_ab {
&pwm_AO_cd not _ab
No it has a conflict with CPU_B (vddcpu_b) PWM
Uh, you're right. That's probably why the Hardkernel team uses a
software based PWM implementation: [0]
In hindsight they should have used a different pad either for VDDCPU_B
or the fan.
I think the most pragmatic approach (since the "GPIO PWM" driver is
not upstream and I don't know if something like that would be accepted
upstream) is to use a GPIO based fan as you did in your initial patch.
Not sure what others think though.
Indeed I recall them using a SW pwm.
Yes it's perfectly ok to use a gpio-fan.
Anand, please fixup the ddr-active name and send a v2.
Thanks,
Neil
Best regards,
Martin
[0] https://github.com/hardkernel/linux/blob/c109dec94e7e819554830acfac4b6ed96e230179/arch/arm64/boot/dts/amlogic/meson64_odroidn2.dtsi#L356-L359
