RE: [PATCH 1/2] watchdog: Add watchdog driver for Sunplus SP7021

["xt.hu[胡先韬]" <xt.hu@cqplus1.com>]

Hi, Guenter Roeck :

Thanks for your review. Sorry I was so focused on fixing code and 
I forgot to respond to the email. I modify the code as you comment
and answer the question.

Thanks
Best Regards,
Xiantao Hu

> -----Original Message-----
> From: Guenter Roeck [mailto:groeck7@gmail.com] On Behalf Of Guenter Roeck
> Sent: Friday, November 12, 2021 10:46 PM
> To: xt.hu[胡先韬] <xt.hu@cqplus1.com>; wim@linux-watchdog.org; p.zabel@pengutronix.de;
> linux-kernel@vger.kernel.org; linux-watchdog@vger.kernel.org; robh+dt@kernel.org;
> devicetree@vger.kernel.org
> Cc: Wells Lu 呂芳騰 <wells.lu@sunplus.com>; qinjian[覃健] <qinjian@cqplus1.com>
> Subject: Re: [PATCH 1/2] watchdog: Add watchdog driver for Sunplus SP7021
> 
> On 11/12/21 2:59 AM, Xiantao Hu wrote:
> > Sunplus SP7021 requires watchdog timer support.
> > Add watchdog driver to enable this.
> >
> > Signed-off-by: Xiantao Hu <xt.hu@cqplus1.com>
...
> > +struct sp_wdt_priv {
> > +	struct watchdog_device wdev;
> > +	void __iomem *base;
> > +	void __iomem *miscellaneous;
> 
> Please find a better name for this variable. Also, unless I am
> missing something, it is only used in the init function. That means
> it can be passed to that function as parameter and doesn't need
> to be stored in sp_wdt_priv.
> 

I will remove the miscellaneous from sp_wdt_priv and pass
to that function as parameter.

> > +	struct clk *clk;
> > +	struct reset_control *rstc;
> > +};
...
> > +static int sp_wdt_set_timeout(struct watchdog_device *wdev,
> > +				   unsigned int timeout)
> > +{
> > +	wdev->timeout = timeout;
> > +
> 
> I would assume this also needs to set the new timeout in HW if
> the watchdog is running, or the watchdog could time out before
> userspace sends another ping.
> 

I will call the ping() after assigning timeout.

> > +	return 0;
> > +}
> > +
...
> > +
> > +static int sp_wdt_start(struct watchdog_device *wdev)
> > +{
> > +	int ret;
> > +
> > +	ret = sp_wdt_set_timeout(wdev, wdev->timeout);
> 
> That function never returns an error, so checking for it is
> pointless here.
> 

I will remove the variable ret.

> > +	if (ret < 0)
> > +		return ret;
> > +
...
> > +/*
> > + * 1.We need to reset watchdog flag(clear watchdog interrupt) here
> > + * because watchdog timer driver does not have an interrupt handler,
> > + * and before enalbe STC and RBUS watchdog timeout. Otherwise,
> 
> enable
> 

I will fix it.

> > + * the intr is always in the triggered state.
> > + * 2.enable STC and RBUS watchdog timeout trigger.
> > + * 3.watchdog conuter is running, need to be stopped.
> 
> counter
> 

I will fix it.

> > + */
...
> > +
> > +	wdt_res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
> > +	priv->miscellaneous =
> > +	    devm_ioremap(dev, wdt_res->start, resource_size(wdt_res));
> 
> Why not use devm_ioremap_resource() ? Or, for that matter,
> devm_platform_ioremap_resource() like above ?
> 

The function of this register is miscellaneous. The register 
accessed here shared by multiple drivers. Use the function 
of devm_platform_ioremap_resource() which internally
call request_mem_region() causes an error. So I handle it 
in this way. Any better ways?

> > +	if (IS_ERR(priv->miscellaneous))
> > +		return PTR_ERR(priv->miscellaneous);
> > +
> > +	priv->wdev.info = &sp_wdt_info;
> > +	priv->wdev.ops = &sp_wdt_ops;
> > +	priv->wdev.timeout = SP_WDT_MAX_TIMEOUT;
> > +	priv->wdev.max_timeout = SP_WDT_MAX_TIMEOUT;
> > +	priv->wdev.min_timeout = SP_WDT_MIN_TIMEOUT;
> 
> THis should really use max_hw_heartbeat_ms to let the core
> ping the watchdog if larger timeouts are desired.
> 

I will add the max_hw_heartbeat_ms and modify the 
ping ().

> > +	priv->wdev.parent = dev;
> > +
> > +	watchdog_set_drvdata(&priv->wdev, priv);
> > +	sp_wdt_hw_init(&priv->wdev);
> > +
> > +	watchdog_init_timeout(&priv->wdev, timeout, dev);
> > +	watchdog_set_nowayout(&priv->wdev, nowayout);
> > +	watchdog_stop_on_reboot(&priv->wdev);
> > +
> > +	err = devm_watchdog_register_device(dev, &priv->wdev);
> 
> Since you call watchdog_unregister_device() below you can not use
> the devm_ function here.
> 

I will drop the function remove() and not call watchdog_unregister_device() .
Use the devm_add_action_or_reset() add reset_control_assert() and
clk_disable_unprepare().

> > +	if (unlikely(err))
> 
> This is not time critical. Drop the unlikely.
> 

I will fix it.

> > +		return err;
> > +
> > +	dev_info(dev, "Watchdog enabled (timeout=%d sec%s.)\n",
> > +		 priv->wdev.timeout, nowayout ? ", nowayout" : "");
> > +
> > +	return 0;
> > +}
> > +
> > +static int sp_wdt_remove(struct platform_device *pdev)
> > +{
> > +	struct sp_wdt_priv *priv = platform_get_drvdata(pdev);
> > +
> > +	watchdog_unregister_device(&priv->wdev);
> > +
> > +	reset_control_assert(priv->rstc);
> > +	clk_disable_unprepare(priv->clk);
> > +
> > +	return 0;
> > +}
> > +
...
> > +
> > +static int __maybe_unused sp_wdt_resume(struct device *dev)
> > +{
> > +	struct sp_wdt_priv *priv = dev_get_drvdata(dev);
> > +
> > +	if (watchdog_active(&priv->wdev))
> > +		sp_wdt_start(&priv->wdev);
> > +
> Shouldn't the order be the opposite of the order in the suspend function ?
>
 
I will fix it.

> > +	reset_control_deassert(priv->rstc);
> > +	clk_prepare_enable(priv->clk);
> > +
> > +	return 0;
> > +}
> >
...
> >