linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Hugo Villeneuve <hugo@hugovil.com>
To: Philipp Rosenberger <p.rosenberger@kunbus.com>
Cc: a.zummo@towertech.it, alexandre.belloni@bootlin.com,
	robh+dt@kernel.org, krzysztof.kozlowski+dt@linaro.org,
	linux-rtc@vger.kernel.org, devicetree@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	Hugo Villeneuve <hvilleneuve@dimonoff.com>
Subject: Re: [PATCH v3 12/14] rtc: pcf2127: support generic watchdog timing configuration
Date: Thu, 19 Jan 2023 12:48:53 -0500	[thread overview]
Message-ID: <20230119124853.9193fd073b8a95b45e20f41c@hugovil.com> (raw)
In-Reply-To: <72514fec-12e2-5b51-261d-9e171b46f5bb@kunbus.com>

On Wed, 18 Jan 2023 14:23:14 +0100
Philipp Rosenberger <p.rosenberger@kunbus.com> wrote:

> Hi Hugo,
> 
> shouldn't the timeout set with pcf2127_wdt_set_timeout() be in seconds? 
> With your changes this value is dependent on the configuration of the 
> timer source clock for watchdog timer. So with a default of 1/4Hz this 
> will be almost seconds * 4.
> 
> I think we need to do the same calculations as in the 
> pcf2127_watchdog_init() when calculating the timeout from the 
> PCF2127_WD_VAL_DEFAULT.
> 
> Best regards,
> Philipp

Hi Philipp,
you are right that the value store/computed inside the structure wdd (struct watchdog_device) should be in seconds, according to the header file documentation in watchdog.h.

However, in the PCF2127 datasheet, the value n that is stored in the PCF2127_REG_WD_VAL register does not represent a value in seconds, but a counter value. It is given by this equation:

    n = source_clock_frequency x timer_period_in_seconds

For the PCF2127, since the clock used is 1Hz, it works as the previous equation can be simplified as:

    n = timer_period_in_seconds

However, if the source clock is different than 1Hz, it would no longer work. Also, since the PCF2131 uses a default clock of 1/4 Hz, it also would not work.

That is the reason why I modified the watchdog timer value (n) computation to take the clock into account. I then use the desired timeout in seconds given by  PCF2127_WD_VAL_DEFAULT to compute the counter value (n).

So what I am proposing to do is to store the PCF2127_WD_VAL_DEFAULT value in wdd->timeout, as before, but convert it to a counter value in pcf2127_wdt_active_ping(). Or to only compute it once, I could define a new variable and compute/set it in pcf2127_watchdog_init().

Hugo.

 
> On 15.12.22 16:02, Hugo Villeneuve wrote:
> > From: Hugo Villeneuve <hvilleneuve@dimonoff.com>
> > 
> > Introduce in the configuration structure two new values to hold the
> > watchdog clock source and the min_hw_heartbeat_ms value.
> > 
> > The minimum and maximum timeout values are automatically computed from
> > the watchdog clock source value for each variant.
> > 
> > The PCF2131 has no 1Hz watchdog clock source, as is the case for
> > PCF2127/29.
> > 
> > The next best choice is using a 1/4Hz clock, giving a watchdog timeout
> > range between 4 and 1016s. By using the same register configuration as
> > for the PCF2127/29, the 1/4Hz clock source is selected.
> > 
> > Note: the PCF2127 datasheet gives a min/max range between 1 and 255s,
> > but it should be between 2 and 254s, because the watchdog is triggered
> > when the timer value reaches 1, not 0.
> > 
> > Signed-off-by: Hugo Villeneuve <hvilleneuve@dimonoff.com>
> > ---
> >   drivers/rtc/rtc-pcf2127.c | 56 +++++++++++++++++++++++++++++++++------
> >   1 file changed, 48 insertions(+), 8 deletions(-)
> > 
> > diff --git a/drivers/rtc/rtc-pcf2127.c b/drivers/rtc/rtc-pcf2127.c
> > index 11fbdab6bf01..3fd2fee4978b 100644
> > --- a/drivers/rtc/rtc-pcf2127.c
> > +++ b/drivers/rtc/rtc-pcf2127.c
> > @@ -157,9 +157,29 @@
> >   
> >   /* Watchdog timer value constants */
> >   #define PCF2127_WD_VAL_STOP		0
> > -#define PCF2127_WD_VAL_MIN		2
> > -#define PCF2127_WD_VAL_MAX		255
> > -#define PCF2127_WD_VAL_DEFAULT		60
> > +#define PCF2127_WD_VAL_DEFAULT		60 /* In seconds. */
> > +/* PCF2127/29 watchdog timer value constants */
> > +#define PCF2127_WD_CLOCK_HZ_X1000	1000 /* 1Hz */
> > +#define PCF2127_WD_MIN_HW_HEARTBEAT_MS	500
> > +/* PCF2131 watchdog timer value constants */
> > +#define PCF2131_WD_CLOCK_HZ_X1000	250  /* 1/4Hz */
> > +#define PCF2131_WD_MIN_HW_HEARTBEAT_MS	4000
> > +/*
> > + * Compute watchdog period, t, in seconds, from the WATCHDG_TIM_VAL register
> > + * value, n, and the clock frequency, f, in Hz.
> > + *
> > + * The PCF2127/29 datasheet gives t as:
> > + *   t = n / f
> > + * The PCF2131 datasheet gives t as:
> > + *   t = (n - 1) / f
> > + * For both variants, the watchdog is triggered when the WATCHDG_TIM_VAL reaches
> > + * the value 1, and not zero. Consequently, the equation from the PCF2131
> > + * datasheet seems to be the correct one for both variants.
> > + */
> > +#define WD_PERIOD_S(_n_, _f1000_) ((1000 * ((_n_) - 1)) / (_f1000_))
> > +
> > +/* Compute value of WATCHDG_TIM_VAL to obtain period t, in seconds. */
> > +#define WD_COUNTER(_t_, _f1000_) ((((_t_) * (_f1000_)) / 1000) + 1)
> >   
> >   /* Mask for currently enabled interrupts */
> >   #define PCF2127_CTRL1_IRQ_MASK (PCF2127_BIT_CTRL1_TSF1)
> > @@ -202,6 +222,11 @@ struct pcf21xx_config {
> >   	u8 reg_wd_val; /* Watchdog value register. */
> >   	u8 reg_clkout; /* Clkout register. */
> >   	u8 reg_reset;  /* Reset register if available. */
> > +
> > +	/* Watchdog configuration. */
> > +	int wdd_clock_hz_x1000; /* Value in Hz multiplicated by 1000 */
> > +	int wdd_min_hw_heartbeat_ms;
> > +
> >   	unsigned int ts_count;
> >   	struct pcf21xx_ts_config ts[4];
> >   	struct attribute_group attribute_group;
> > @@ -496,10 +521,19 @@ static int pcf2127_watchdog_init(struct device *dev, struct pcf2127 *pcf2127)
> >   	pcf2127->wdd.parent = dev;
> >   	pcf2127->wdd.info = &pcf2127_wdt_info;
> >   	pcf2127->wdd.ops = &pcf2127_watchdog_ops;
> > -	pcf2127->wdd.min_timeout = PCF2127_WD_VAL_MIN;
> > -	pcf2127->wdd.max_timeout = PCF2127_WD_VAL_MAX;
> > -	pcf2127->wdd.timeout = PCF2127_WD_VAL_DEFAULT;
> > -	pcf2127->wdd.min_hw_heartbeat_ms = 500;
> > +
> > +	pcf2127->wdd.min_timeout =
> > +		WD_PERIOD_S(2, pcf2127->cfg->wdd_clock_hz_x1000);
> > +	pcf2127->wdd.max_timeout =
> > +		WD_PERIOD_S(255, pcf2127->cfg->wdd_clock_hz_x1000);
> > +	pcf2127->wdd.timeout = WD_COUNTER(PCF2127_WD_VAL_DEFAULT,
> > +					  pcf2127->cfg->wdd_clock_hz_x1000);
> > +
> > +	dev_dbg(dev, "%s min = %ds\n", __func__, pcf2127->wdd.min_timeout);
> > +	dev_dbg(dev, "%s max = %ds\n", __func__, pcf2127->wdd.max_timeout);
> > +	dev_dbg(dev, "%s def = %d\n", __func__, pcf2127->wdd.timeout);
> > +
> > +	pcf2127->wdd.min_hw_heartbeat_ms = pcf2127->cfg->wdd_min_hw_heartbeat_ms;
> >   	pcf2127->wdd.status = WATCHDOG_NOWAYOUT_INIT_STATUS;
> >   
> >   	watchdog_set_drvdata(&pcf2127->wdd, pcf2127);
> > @@ -926,6 +960,8 @@ static struct pcf21xx_config pcf21xx_cfg[] = {
> >   		.reg_wd_ctl = PCF2127_REG_WD_CTL,
> >   		.reg_wd_val = PCF2127_REG_WD_VAL,
> >   		.reg_clkout = PCF2127_REG_CLKOUT,
> > +		.wdd_clock_hz_x1000 = PCF2127_WD_CLOCK_HZ_X1000,
> > +		.wdd_min_hw_heartbeat_ms = PCF2127_WD_MIN_HW_HEARTBEAT_MS,
> >   		.ts_count = 1,
> >   		.ts[0] = {
> >   			.regs_base = PCF2127_REG_TS1_BASE,
> > @@ -951,6 +987,8 @@ static struct pcf21xx_config pcf21xx_cfg[] = {
> >   		.reg_wd_ctl = PCF2127_REG_WD_CTL,
> >   		.reg_wd_val = PCF2127_REG_WD_VAL,
> >   		.reg_clkout = PCF2127_REG_CLKOUT,
> > +		.wdd_clock_hz_x1000 = PCF2127_WD_CLOCK_HZ_X1000,
> > +		.wdd_min_hw_heartbeat_ms = PCF2127_WD_MIN_HW_HEARTBEAT_MS,
> >   		.ts_count = 1,
> >   		.ts[0] = {
> >   			.regs_base = PCF2127_REG_TS1_BASE,
> > @@ -977,6 +1015,8 @@ static struct pcf21xx_config pcf21xx_cfg[] = {
> >   		.reg_wd_val = PCF2131_REG_WD_VAL,
> >   		.reg_clkout = PCF2131_REG_CLKOUT,
> >   		.reg_reset  = PCF2131_REG_SR_RESET,
> > +		.wdd_clock_hz_x1000 = PCF2131_WD_CLOCK_HZ_X1000,
> > +		.wdd_min_hw_heartbeat_ms = PCF2131_WD_MIN_HW_HEARTBEAT_MS,
> >   		.ts_count = 4,
> >   		.ts[0] = {
> >   			.regs_base = PCF2131_REG_TS1_BASE,
> > @@ -1215,7 +1255,7 @@ static int pcf2127_probe(struct device *dev, struct regmap *regmap,
> >   
> >   	/*
> >   	 * Watchdog timer enabled and reset pin /RST activated when timed out.
> > -	 * Select 1Hz clock source for watchdog timer.
> > +	 * Select 1Hz clock source for watchdog timer (1/4Hz for PCF2131).
> >   	 * Note: Countdown timer disabled and not available.
> >   	 * For pca2129, pcf2129 and pcf2131, only bit[7] is for Symbol WD_CD
> >   	 * of register watchdg_tim_ctl. The bit[6] is labeled
> 


-- 
Hugo Villeneuve <hugo@hugovil.com>

  reply	other threads:[~2023-01-19 17:49 UTC|newest]

Thread overview: 72+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-12-15 15:02 [PATCH v3 00/14] rtc: pcf2127: add PCF2131 driver Hugo Villeneuve
2022-12-15 15:02 ` [PATCH v3 01/14] rtc: pcf2127: add variant-specific configuration structure Hugo Villeneuve
2022-12-19  9:05   ` Bruno Thomsen
2022-12-19 15:15     ` Hugo Villeneuve
2022-12-19 17:17       ` Bruno Thomsen
2022-12-19 18:30         ` Hugo Villeneuve
2023-01-07 16:52   ` Bruno Thomsen
2022-12-15 15:02 ` [PATCH v3 02/14] rtc: pcf2127: adapt for time/date registers at any offset Hugo Villeneuve
2022-12-19  9:34   ` Bruno Thomsen
2022-12-19 16:27     ` Hugo Villeneuve
2023-01-07 16:49     ` Bruno Thomsen
2023-01-20 18:47   ` Alexandre Belloni
2023-01-23 15:54     ` Hugo Villeneuve
2023-06-21 18:18       ` Alexandre Belloni
2022-12-15 15:02 ` [PATCH v3 03/14] rtc: pcf2127: adapt for alarm " Hugo Villeneuve
2023-01-07 16:57   ` Bruno Thomsen
2023-01-20 17:10   ` Alexandre Belloni
2023-01-23 16:02     ` Hugo Villeneuve
2022-12-15 15:02 ` [PATCH v3 04/14] rtc: pcf2127: adapt for WD " Hugo Villeneuve
2023-01-07 16:59   ` Bruno Thomsen
2022-12-15 15:02 ` [PATCH v3 05/14] rtc: pcf2127: adapt for CLKOUT register " Hugo Villeneuve
2023-01-07 17:01   ` Bruno Thomsen
2022-12-15 15:02 ` [PATCH v3 06/14] rtc: pcf2127: add support for multiple TS functions Hugo Villeneuve
2023-01-07 17:58   ` Bruno Thomsen
2023-01-23 20:41     ` Hugo Villeneuve
2022-12-15 15:02 ` [PATCH v3 07/14] rtc: pcf2127: add support for PCF2131 RTC Hugo Villeneuve
2023-01-07 18:15   ` Bruno Thomsen
2023-01-23 19:06     ` Hugo Villeneuve
2023-01-20 18:57   ` Alexandre Belloni
2023-01-23 17:27     ` Hugo Villeneuve
2022-12-15 15:02 ` [PATCH v3 08/14] rtc: pcf2127: add support for PCF2131 interrupts on output INT_A Hugo Villeneuve
2023-01-07 18:17   ` Bruno Thomsen
2023-01-20 16:56   ` Alexandre Belloni
2023-01-23 20:52     ` Hugo Villeneuve
2023-05-11 17:19       ` Hugo Villeneuve
2023-06-21 19:24         ` Alexandre Belloni
2023-06-21 19:26           ` Alexandre Belloni
2023-06-22 14:21             ` Hugo Villeneuve
2022-12-15 15:02 ` [PATCH v3 09/14] rtc: pcf2127: set PWRMNG value for PCF2131 Hugo Villeneuve
2023-01-07 18:36   ` Bruno Thomsen
2023-01-20 16:39     ` Alexandre Belloni
2023-01-23 22:07       ` Hugo Villeneuve
2022-12-15 15:02 ` [PATCH v3 10/14] rtc: pcf2127: read and validate PCF2131 device signature Hugo Villeneuve
2023-01-20 17:01   ` Alexandre Belloni
2023-01-23 17:31     ` Hugo Villeneuve
2022-12-15 15:02 ` [PATCH v3 11/14] rtc: pcf2127: adapt time/date registers write sequence for PCF2131 Hugo Villeneuve
2023-01-07 18:44   ` Bruno Thomsen
2023-01-23 20:55     ` Hugo Villeneuve
2023-01-20 17:09   ` Alexandre Belloni
2023-01-23 21:57     ` Hugo Villeneuve
2023-06-21 19:36       ` Alexandre Belloni
2022-12-15 15:02 ` [PATCH v3 12/14] rtc: pcf2127: support generic watchdog timing configuration Hugo Villeneuve
2023-01-18 13:23   ` Philipp Rosenberger
2023-01-19 17:48     ` Hugo Villeneuve [this message]
2023-01-20  8:06       ` Philipp Rosenberger
2023-01-20 14:44         ` Hugo Villeneuve
2022-12-15 15:02 ` [PATCH v3 13/14] rtc: pcf2127: add flag for watchdog register value read support Hugo Villeneuve
2023-01-07 18:47   ` Bruno Thomsen
2022-12-15 15:02 ` [PATCH v3 14/14] dt-bindings: rtc: pcf2127: add PCF2131 Hugo Villeneuve
2022-12-16 13:24   ` Krzysztof Kozlowski
2022-12-19  9:14   ` Bruno Thomsen
2022-12-19 16:25     ` Hugo Villeneuve
2022-12-19 17:18       ` Bruno Thomsen
2022-12-19 18:31         ` Hugo Villeneuve
2023-01-20 19:05 ` [PATCH v3 00/14] rtc: pcf2127: add PCF2131 driver Alexandre Belloni
2023-01-23 15:51   ` Hugo Villeneuve
2023-06-21 14:14   ` Hugo Villeneuve
2023-06-21 16:59     ` Hugo Villeneuve
2023-06-21 18:14       ` Alexandre Belloni
2023-06-21 18:28         ` Hugo Villeneuve
2023-07-05 13:40           ` Hugo Villeneuve
2023-07-07 14:16             ` Alexandre Belloni

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230119124853.9193fd073b8a95b45e20f41c@hugovil.com \
    --to=hugo@hugovil.com \
    --cc=a.zummo@towertech.it \
    --cc=alexandre.belloni@bootlin.com \
    --cc=devicetree@vger.kernel.org \
    --cc=hvilleneuve@dimonoff.com \
    --cc=krzysztof.kozlowski+dt@linaro.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-rtc@vger.kernel.org \
    --cc=p.rosenberger@kunbus.com \
    --cc=robh+dt@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).