Allwinner A64: Issue on external rtc clock to wifi chip

Commit Message

Michael Nazzareno Trimarchi May 28, 2018, 7:08 a.m. UTC

Hi Maxime

On Mon, May 14, 2018 at 1:17 PM, Maxime Ripard
<[e-mail]> wrote:
> On Mon, May 14, 2018 at 03:12:49PM +0530, Jagan Teki wrote:
>> On Mon, May 14, 2018 at 2:36 PM, Maxime Ripard
>> <[e-mail]> wrote:
>> > On Mon, May 14, 2018 at 02:34:22PM +0530, Jagan Teki wrote:
>> >> On Mon, May 14, 2018 at 1:57 PM, Maxime Ripard
>> >> <[e-mail]> wrote:
>> >> > On Mon, May 14, 2018 at 01:34:56PM +0530, Jagan Teki wrote:
>> >> >> On Mon, May 14, 2018 at 1:27 PM, Maxime Ripard
>> >> >> <[e-mail]> wrote:
>> >> >> > Hi,
>> >> >> >
>> >> >> > On Mon, May 14, 2018 at 12:37:49PM +0530, Jagan Teki wrote:
>> >> >> >> Hi Maxime and All,
>> >> >> >>
>> >> >> >> We are trying to bring-up AP6330 Wifi chip for A64 board. We noticed
>> >> >> >> to have an external rtc clock has driven from wifi chip.
>> >> >> >>
>> >> >> >> So the devicetree is configured according to this as below.
>> >> >> >>
>> >> >> >> / {
>> >> >> >>         wifi_pwrseq: wifi-pwrseq {
>> >> >> >>                 compatible = "mmc-pwrseq-simple";
>> >> >> >>                 clocks = <&rtc 1>;
>> >> >> >>                 clock-names = "ext_clock";
>> >> >> >>                 reset-gpios = <&r_pio 0 2 GPIO_ACTIVE_LOW>; /* PL2 */
>> >> >> >>                 post-power-on-delay-ms = <400>;
>> >> >> >>         };
>> >> >> >> };
>> >> >> >>
>> >> >> >> &rtc {
>> >> >> >>         clock-output-names = "rtc-osc32k", "rtc-osc32k-out";
>> >> >> >>         clocks = <&osc32k>;
>> >> >> >>         #clock-cells = <1>;
>> >> >> >> };
>> >> >> >>
>> >> >> >> &mmc1 {
>> >> >> >>         pinctrl-names = "default";
>> >> >> >>         pinctrl-0 = <&mmc1_pins>;
>> >> >> >>         vmmc-supply = <&reg_dcdc1>;
>> >> >> >>         vqmmc-supply = <&reg_eldo1>;
>> >> >> >>         mmc-pwrseq = <&wifi_pwrseq>;
>> >> >> >>         bus-width = <4>;
>> >> >> >>         non-removable;
>> >> >> >>         status = "okay";
>> >> >> >>
>> >> >> >>         brcmf: wifi@1 {
>> >> >> >>                 reg = <1>;
>> >> >> >>                 compatible = "brcm,bcm4329-fmac";
>> >> >> >>                 interrupt-parent = <&r_pio>;
>> >> >> >>                 interrupts = <0 3 IRQ_TYPE_LEVEL_LOW>;  /* WL-WAKE-AP: PL3 */
>> >> >> >>                 interrupt-names = "host-wake";
>> >> >> >>         };
>> >> >> >> };
>> >> >> >>
>> >> >> >> And observed rtc-osc32k-out clock is never enabled[1] and the value of
>> >> >> >> LOSC_OUT_GATING is 0x0 which eventually not enabling
>> >> >> >> LOSC_OUT_GATING_EN
>> >> >> >>
>> >> >> >> Pls. let us know if we miss anything here?
>> >> >> >>
>> >> >> >> [1] https://paste.ubuntu.com/p/X2By4q8kD2/
>> >> >> >
>> >> >> > Could you paste your config and the logs from a boot to?
>> >> >>
>> >> >> .config
>> >> >> https://paste.ubuntu.com/p/w9w2KB7RFc/
>> >> >>
>> >> >> dmesg
>> >> >> https://paste.ubuntu.com/p/mrZGk5bWRR/
>> >> >
>> >> > This is kind of weird. Have you tested with a 4.17 kernel? We have
>> >> > runtime_pm changes lined up in next, so that might be a regression
>> >> > there, even though we tested it with Quentin at some point.
>> >>
>> >> This is 4.17-rc4 do you want to try it on 4.16 ?
>> >
>> > No, this is next-20180503. Please try with 4.17-rc4
>>
>> Couldn't find any different in behaviour [2]
>>
>> [2] https://paste.ubuntu.com/p/m3PGBwrv6W/
>
> It's hard to tell without the board, but have you looked at the return
> value of devm_clk_get in the pwrseq code?
>
> Enabling the clk ftrace events would also help.
>

The driver has one bug.

Michael


> Maxime
>
> --
> Maxime Ripard, Bootlin (formerly Free Electrons)
> Embedded Linux and Kernel engineering
> https://bootlin.com

Patch

diff --git a/drivers/rtc/rtc-sun6i.c b/drivers/rtc/rtc-sun6i.c
index 2e6fb27..a66f40c 100644
--- a/drivers/rtc/rtc-sun6i.c
+++ b/drivers/rtc/rtc-sun6i.c
@@ -74,7 +74,7 @@ 
 #define SUN6I_ALARM_CONFIG_WAKEUP              BIT(0)

 #define SUN6I_LOSC_OUT_GATING                  0x0060
-#define SUN6I_LOSC_OUT_GATING_EN               BIT(0)
+#define SUN6I_LOSC_OUT_GATING_EN               0

You need to pass bit_idx that is 0 and not BIT(0)