linux-kernel - Re: [PATCH 5/7] watchdog: s3c2410: Introduce separate source clock

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <76ca9492-3829-6d3b-580c-983bc8409b1f@canonical.com>
Date:   Fri, 29 Oct 2021 10:18:08 +0200
From:   Krzysztof Kozlowski <krzysztof.kozlowski@...onical.com>
To:     Sam Protsenko <semen.protsenko@...aro.org>,
        Wim Van Sebroeck <wim@...ux-watchdog.org>,
        Guenter Roeck <linux@...ck-us.net>,
        Rob Herring <robh+dt@...nel.org>
Cc:     linux-watchdog@...r.kernel.org, devicetree@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
        linux-samsung-soc@...r.kernel.org
Subject: Re: [PATCH 5/7] watchdog: s3c2410: Introduce separate source clock

On 28/10/2021 20:35, Sam Protsenko wrote:
> Some Exynos chips (like Exynos850) have dedicated source clock. That
> clock is provided from device tree as "watchdog_src" clock. In such
> case, "watchdog" clock is just a peripheral clock used for register
> interface. If "watchdog_src" is present, use its rate instead of
> "watchdog" for all timer related calculations.

Please explain what is this source clock and remove the reference to
devicetree. Instead describe rather real HW. It's confusing now to have
one clock called watchdog and one watchdog source.

The source clock is the actual clock driving watchdog and it's counter,
right? Then let's document it and rename the variables to match reality
- one is pclk (or apb?) and second is counter or source?

> 
> Signed-off-by: Sam Protsenko <semen.protsenko@...aro.org>
> ---
>  drivers/watchdog/s3c2410_wdt.c | 33 +++++++++++++++++++++++++++------
>  1 file changed, 27 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/watchdog/s3c2410_wdt.c b/drivers/watchdog/s3c2410_wdt.c
> index a5ef7171a90e..bfc5872ca497 100644
> --- a/drivers/watchdog/s3c2410_wdt.c
> +++ b/drivers/watchdog/s3c2410_wdt.c
> @@ -126,6 +126,8 @@ struct s3c2410_wdt_variant {
>  struct s3c2410_wdt {
>  	struct device		*dev;
>  	struct clk		*clock;
> +	struct clk		*clock_src;
> +	unsigned long		freq_src;
>  	void __iomem		*reg_base;
>  	unsigned int		count;
>  	spinlock_t		lock;
> @@ -213,10 +215,8 @@ MODULE_DEVICE_TABLE(platform, s3c2410_wdt_ids);
>  
>  /* functions */
>  
> -static inline unsigned int s3c2410wdt_max_timeout(struct clk *clock)
> +static inline unsigned int s3c2410wdt_max_timeout(unsigned long freq)
>  {
> -	unsigned long freq = clk_get_rate(clock);
> -
>  	return S3C2410_WTCNT_MAXCNT / (freq / (S3C2410_WTCON_PRESCALE_MAX + 1)
>  				       / S3C2410_WTCON_MAXDIV);
>  }
> @@ -364,7 +364,7 @@ static int s3c2410wdt_set_heartbeat(struct watchdog_device *wdd,
>  				    unsigned int timeout)
>  {
>  	struct s3c2410_wdt *wdt = watchdog_get_drvdata(wdd);
> -	unsigned long freq = clk_get_rate(wdt->clock);
> +	unsigned long freq = wdt->freq_src;

This does not look good. You are using fixed frequency (from probe).

>  	unsigned int count;
>  	unsigned int divisor = 1;
>  	unsigned long wtcon;
> @@ -627,13 +627,27 @@ static int s3c2410wdt_probe(struct platform_device *pdev)
>  		return ret;
>  	}
>  
> +	/* "watchdog_src" clock is optional; if it's not present -- just skip */
> +	wdt->clock_src = devm_clk_get(dev, "watchdog_src");
> +	if (!IS_ERR(wdt->clock_src)) {
> +		ret = clk_prepare_enable(wdt->clock_src);
> +		if (ret < 0) {
> +			dev_err(dev, "failed to enable source clock\n");
> +			ret = PTR_ERR(wdt->clock_src);
> +			goto err_clk;
> +		}
> +		wdt->freq_src = clk_get_rate(wdt->clock_src);
> +	} else {
> +		wdt->freq_src = clk_get_rate(wdt->clock);
> +	}
> +
>  	wdt->wdt_device.min_timeout = 1;
> -	wdt->wdt_device.max_timeout = s3c2410wdt_max_timeout(wdt->clock);
> +	wdt->wdt_device.max_timeout = s3c2410wdt_max_timeout(wdt->freq_src);
>  
>  	ret = s3c2410wdt_cpufreq_register(wdt);
>  	if (ret < 0) {
>  		dev_err(dev, "failed to register cpufreq\n");
> -		goto err_clk;
> +		goto err_clk_src;
>  	}
>  
>  	watchdog_set_drvdata(&wdt->wdt_device, wdt);
> @@ -707,6 +721,10 @@ static int s3c2410wdt_probe(struct platform_device *pdev)
>   err_cpufreq:
>  	s3c2410wdt_cpufreq_deregister(wdt);
>  
> + err_clk_src:
> +	if (!IS_ERR(wdt->clock_src))
> +		clk_disable_unprepare(wdt->clock_src);

No. Errors in getting source clock should not be ignored, so you should
never store here ERR. You could store NULL. If() is anyway not needed in
both cases.

You can simplify all this and take pclk twice if src clock is missing.
Or assign src=pclk...

> +
>   err_clk:
>  	clk_disable_unprepare(wdt->clock);
>  
> @@ -727,6 +745,9 @@ static int s3c2410wdt_remove(struct platform_device *dev)
>  
>  	s3c2410wdt_cpufreq_deregister(wdt);
>  
> +	if (!IS_ERR(wdt->clock_src))
> +		clk_disable_unprepare(wdt->clock_src);
> +
>  	clk_disable_unprepare(wdt->clock);
>  
>  	return 0;
> 


Best regards,
Krzysztof