Re: [PATCH v2] watchdog: iTCO_wdt: Account for rebooting on second timeout

[Guenter Roeck]

On Sun, May 30, 2021 at 01:24:23PM +0200, Jan Kiszka wrote:
> From: Jan Kiszka <jan.kiszka@xxxxxxxxxxx>
> 
> This was already attempted to fix via 1fccb73011ea: If the BIOS did not
> enable TCO SMIs, the timer definitely needs to trigger twice in order to
> cause a reboot. If TCO SMIs are on, as well as SMIs in general, we can
> continue to assume that the BIOS will perform a reboot on the first
> timeout.
> 
> QEMU with its ICH9 and related BIOS falls into the former category,
> currently taking twice the configured timeout in order to reboot the
> machine. For iTCO version that fall under turn_SMI_watchdog_clear_off,
> this is also true and was currently only addressed for v1, irrespective
> of the turn_SMI_watchdog_clear_off value.
> 
> Signed-off-by: Jan Kiszka <jan.kiszka@xxxxxxxxxxx>
> ---
> 
> Changes in v2:
>  - consider GBL_SMI_EN as well
> 
>  drivers/watchdog/iTCO_wdt.c | 12 +++++++++---
>  1 file changed, 9 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/watchdog/iTCO_wdt.c b/drivers/watchdog/iTCO_wdt.c
> index bf31d7b67a69..3f1324871cfd 100644
> --- a/drivers/watchdog/iTCO_wdt.c
> +++ b/drivers/watchdog/iTCO_wdt.c
> @@ -71,6 +71,8 @@
>  #define TCOBASE(p)	((p)->tco_res->start)
>  /* SMI Control and Enable Register */
>  #define SMI_EN(p)	((p)->smi_res->start)
> +#define TCO_EN		(1 << 13)
> +#define GBL_SMI_EN	(1 << 0)
>  
>  #define TCO_RLD(p)	(TCOBASE(p) + 0x00) /* TCO Timer Reload/Curr. Value */
>  #define TCOv1_TMR(p)	(TCOBASE(p) + 0x01) /* TCOv1 Timer Initial Value*/
> @@ -355,8 +357,12 @@ static int iTCO_wdt_set_timeout(struct watchdog_device *wd_dev, unsigned int t)
>  
>  	tmrval = seconds_to_ticks(p, t);
>  
> -	/* For TCO v1 the timer counts down twice before rebooting */
> -	if (p->iTCO_version == 1)
> +	/*
> +	 * If TCO SMIs are off, the timer counts down twice before rebooting.
> +	 * Otherwise, the BIOS generally reboots when the SMI triggers.
> +	 */
> +	if (p->smi_res &&
> +	    (SMI_EN(p) & (TCO_EN | GBL_SMI_EN)) != (TCO_EN | GBL_SMI_EN))
>  		tmrval /= 2;

This expands the scope of this adjustment to all versions, while at the same
time making it conditional for v1. Is this correct ? What for systems with v1
TCO where the above conditions are not met ?

Guenter

>  
>  	/* from the specs: */
> @@ -521,7 +527,7 @@ static int iTCO_wdt_probe(struct platform_device *pdev)
>  		 * Disables TCO logic generating an SMI#
>  		 */
>  		val32 = inl(SMI_EN(p));
> -		val32 &= 0xffffdfff;	/* Turn off SMI clearing watchdog */
> +		val32 &= ~TCO_EN;	/* Turn off SMI clearing watchdog */
>  		outl(val32, SMI_EN(p));
>  	}
>  
> -- 
> 2.26.2