[PATCH] arm: mach-k3: am625_init: Add Erratum WA for RTC startup

[Nishanth Menon]

On 14:26-20230427, Bryan Brattlof wrote:
> From: Nishanth Menon <nm at ti.com>
> 
> Add erratum i2327 work around for initialization for RTC
> interrupt where interrupt is stuck for ever at startup. Unfortunately,
> this workaround needs to be applied under 1 second of boot.
> 
> Signed-off-by: Nishanth Menon <nm at ti.com>
> [bb at ti.com: rebased from 2021.01]
> Signed-off-by: Bryan Brattlof <bb at ti.com>
> ---

It will be good to state in the diffstat that am62a7 or other k3 SoCs
are not impacted: am62a7 has the hardware fix and other SoCs do not have
this rtc.

I dont think there is a need to respin the patch for that - just clarifying.


>  arch/arm/mach-k3/am625_init.c | 44 +++++++++++++++++++++++++++++++++++
>  1 file changed, 44 insertions(+)
> 
> diff --git a/arch/arm/mach-k3/am625_init.c b/arch/arm/mach-k3/am625_init.c
> index a91c15ca4e188..e90c18fdcc118 100644
> --- a/arch/arm/mach-k3/am625_init.c
> +++ b/arch/arm/mach-k3/am625_init.c
> @@ -15,6 +15,15 @@
>  #include <dm/uclass-internal.h>
>  #include <dm/pinctrl.h>
>  
> +#define RTC_BASE_ADDRESS		0x2b1f0000
> +#define REG_K3RTC_S_CNT_LSW		(RTC_BASE_ADDRESS + 0x18)
> +#define REG_K3RTC_KICK0			(RTC_BASE_ADDRESS + 0x70)
> +#define REG_K3RTC_KICK1			(RTC_BASE_ADDRESS + 0x74)
> +
> +/* Magic values for lock/unlock */
> +#define K3RTC_KICK0_UNLOCK_VALUE	0x83e70b13
> +#define K3RTC_KICK1_UNLOCK_VALUE	0x95a4f1e0
> +
>  /*
>   * This uninitialized global variable would normal end up in the .bss section,
>   * but the .bss is cleared between writing and reading this variable, so move
> @@ -71,6 +80,40 @@ static __maybe_unused void enable_mcu_esm_reset(void)
>  	writel(stat, CTRLMMR_MCU_RST_CTRL);
>  }
>  
> +#if defined(CONFIG_CPU_V7R)
> +
> +/*
> + * RTC Erratum i2327 Workaround
> + * Due to a bug in initial synchronization out of cold power on,
> + * IRQ status can get locked infinitely if we do not:
> + * a) unlock RTC
> + *
> + * This workaround *must* be applied within 1 second of power on,
> + * So, this is closest point to be able to guarantee the max
> + * timing.
> + */
> +void rtc_erratumi2327_init(void)
> +{
> +	u32 counter;
> +
> +	/*
> +	 * If counter has gone past 1, nothing we can do, leave
> +	 * system locked! This is the only way we know if RTC
> +	 * can be used for all practical purposes.
> +	 */
> +	counter = readl(REG_K3RTC_S_CNT_LSW);
> +	if (counter > 1)
> +		return;
> +	/*
> +	 * Need to set this up at the very start
> +	 * MUST BE DONE under 1 second of boot.
> +	 */
> +	writel(K3RTC_KICK0_UNLOCK_VALUE, REG_K3RTC_KICK0);
> +	writel(K3RTC_KICK1_UNLOCK_VALUE, REG_K3RTC_KICK1);
> +	return;
> +}
> +#endif
> +
>  void board_init_f(ulong dummy)
>  {
>  	struct udevice *dev;
> @@ -78,6 +121,7 @@ void board_init_f(ulong dummy)
>  
>  #if defined(CONFIG_CPU_V7R)
>  	setup_k3_mpu_regions();
> +	rtc_erratumi2327_init();
>  #endif
>  
>  	/*
> 
> base-commit: a25dcda452bf6a6de72764a8d990d72e5def643d
> -- 
> 2.40.0
> 

-- 
Regards,
Nishanth Menon
Key (0xDDB5849D1736249D) / Fingerprint: F8A2 8693 54EB 8232 17A3  1A34 DDB5 849D 1736 249D