Re: [PATCH v3 3/4] rtc: Add support for configuring the UIP timeout for RTC reads

[Mateusz Jończyk]

W dniu 27.11.2023 o 20:25, Mario Limonciello pisze:
> The UIP timeout is hardcoded to 10ms for all RTC reads, but in some
> contexts this might not be enough time. Add a timeout parameter to
> mc146818_get_time() and mc146818_get_time_callback().
>
> If UIP timeout is configured by caller to be >=100 ms and a call
> takes this long, log a warning.
>
> Make all callers use 10ms to ensure no functional changes.
>
> Cc: stable@xxxxxxxxxxxxxxx # 6.1.y
> Fixes: ec5895c0f2d8 ("rtc: mc146818-lib: extract mc146818_avoid_UIP")
> Signed-off-by: Mario Limonciello <mario.limonciello@xxxxxxx>
> ---
> v2->v3:
>  * Logic adjustments
>  * Clarify warning message
> v1->v2:
>  * Add a warning if 100ms or more
>  * Add stable and fixes tags
[snip]
> diff --git a/drivers/rtc/rtc-mc146818-lib.c b/drivers/rtc/rtc-mc146818-lib.c
> index 43a28e82674e..ab077dde397b 100644
> --- a/drivers/rtc/rtc-mc146818-lib.c
> +++ b/drivers/rtc/rtc-mc146818-lib.c
> @@ -8,26 +8,31 @@
>  #include <linux/acpi.h>
>  #endif
>  
> +#define UIP_RECHECK_DELAY		100	/* usec */
> +#define UIP_RECHECK_DELAY_MS		(USEC_PER_MSEC / UIP_RECHECK_DELAY)
> +#define UIP_RECHECK_TIMEOUT_MS(x)	(x / UIP_RECHECK_DELAY_MS)
> +
>  /*
>   * Execute a function while the UIP (Update-in-progress) bit of the RTC is
> - * unset.
> + * unset. The timeout is configurable by the caller in ms.
>   *
>   * Warning: callback may be executed more then once.
>   */
>  bool mc146818_avoid_UIP(void (*callback)(unsigned char seconds, void *param),
> +			int timeout,
>  			void *param)
>  {
>  	int i;
>  	unsigned long flags;
>  	unsigned char seconds;
>  
> -	for (i = 0; i < 100; i++) {
> +	for (i = 0; i < UIP_RECHECK_TIMEOUT_MS(timeout); i++) {

Sorry, this will not work. UIP_RECHECK_DELAY_MS is 10, so
UIP_RECHECK_TIMEOUT_MS(timeout) will be 1 for timeout=10. Should be

      for (i = 0; UIP_RECHECK_TIMEOUT_MS(i) < timeout; i++) {

With this, for i == 99, UIP_RECHECK_TIMEOUT_MS(i) = 9
for i == 100, UIP_RECHECK_TIMEOUT_MS(i) = 10 and the loop correctly terminates.

The macro should probably be renamed UIP_RECHECK_LOOPS_MS as it converts
loop count to ms.

>  		spin_lock_irqsave(&rtc_lock, flags);
>  
>  		/*
>  		 * Check whether there is an update in progress during which the
>  		 * readout is unspecified. The maximum update time is ~2ms. Poll
> -		 * every 100 usec for completion.
> +		 * for completion.
>  		 *
>  		 * Store the second value before checking UIP so a long lasting
>  		 * NMI which happens to hit after the UIP check cannot make
> @@ -37,7 +42,7 @@ bool mc146818_avoid_UIP(void (*callback)(unsigned char seconds, void *param),
>  
>  		if (CMOS_READ(RTC_FREQ_SELECT) & RTC_UIP) {
>  			spin_unlock_irqrestore(&rtc_lock, flags);
> -			udelay(100);
> +			udelay(UIP_RECHECK_DELAY);
>  			continue;
>  		}
>  
> @@ -56,7 +61,7 @@ bool mc146818_avoid_UIP(void (*callback)(unsigned char seconds, void *param),
>  		 */
>  		if (CMOS_READ(RTC_FREQ_SELECT) & RTC_UIP) {
>  			spin_unlock_irqrestore(&rtc_lock, flags);
> -			udelay(100);
> +			udelay(UIP_RECHECK_DELAY);
>  			continue;
>  		}
>  
> @@ -72,6 +77,10 @@ bool mc146818_avoid_UIP(void (*callback)(unsigned char seconds, void *param),
>  		}
>  		spin_unlock_irqrestore(&rtc_lock, flags);
>  
> +		if (i >= UIP_RECHECK_TIMEOUT_MS(100))

Same, should be:

          if (UIP_RECHECK_TIMEOUT_MS(i) >= 100)

> +			pr_warn("Reading current time from RTC took around %d ms\n",
> +				UIP_RECHECK_TIMEOUT_MS(i));
> +
>  		return true;
>  	}
>  	return false;

[snip]

Greetings,

Mateusz