linux-kernel - Re: [PATCH v2 3/3] watchdog: sprd: change to use usleep

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] [day] [month] [year] [list]

Message-ID: <7b61713e-5808-d157-5b1a-08b1a275a631@roeck-us.net>
Date:   Sun, 8 Nov 2020 08:30:02 -0800
From:   Guenter Roeck <linux@...ck-us.net>
To:     Chunyan Zhang <zhang.lyra@...il.com>,
        Wim Van Sebroeck <wim@...ux-watchdog.org>
Cc:     linux-watchdog@...r.kernel.org, linux-kernel@...r.kernel.org,
        Orson Zhai <orsonzhai@...il.com>,
        Baolin Wang <baolin.wang7@...il.com>,
        Chunyan Zhang <chunyan.zhang@...soc.com>,
        Lingling Xu <ling_ling.xu@...soc.com>,
        Jingchao Ye <jingchao.ye@...soc.com>,
        Xiaoqing Wu <xiaoqing.wu@...soc.com>
Subject: Re: [PATCH v2 3/3] watchdog: sprd: change to use usleep_range()
 instead of busy loop

On 10/28/20 7:39 PM, Chunyan Zhang wrote:
> From: Chunyan Zhang <chunyan.zhang@...soc.com>
> 
> After changing to check busy bit for the previous loading operation instead
> of the current one, for most of cases, the busy bit is not set for the
> first time of read, so there's no need to check so frequently, so this
> patch use usleep_range() to replace cpu_relax() to avoid busy loop.
> 
> Also this patch change the max times to 11 which would be enough, since
> according to the specification, the busy bit would be set after a new
> loading operation and last 2 or 3 RTC clock cycles (about 60us~92us).
> 
> Fixes: 477603467009 ("watchdog: Add Spreadtrum watchdog driver")
> Original-by: Lingling Xu <ling_ling.xu@...soc.com>
> Signed-off-by: Chunyan Zhang <chunyan.zhang@...soc.com>

Reviewed-by: Guenter Roeck <linux@...ck-us.net>

> ---
>  drivers/watchdog/sprd_wdt.c | 10 ++++++----
>  1 file changed, 6 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/watchdog/sprd_wdt.c b/drivers/watchdog/sprd_wdt.c
> index b9b1daa9e2a4..e8097551dfcd 100644
> --- a/drivers/watchdog/sprd_wdt.c
> +++ b/drivers/watchdog/sprd_wdt.c
> @@ -53,7 +53,7 @@
>  
>  #define SPRD_WDT_CNT_HIGH_SHIFT		16
>  #define SPRD_WDT_LOW_VALUE_MASK		GENMASK(15, 0)
> -#define SPRD_WDT_LOAD_TIMEOUT		1000
> +#define SPRD_WDT_LOAD_TIMEOUT		11
>  
>  struct sprd_wdt {
>  	void __iomem *base;
> @@ -109,15 +109,17 @@ static int sprd_wdt_load_value(struct sprd_wdt *wdt, u32 timeout,
>  	u32 prtmr_step = pretimeout * SPRD_WDT_CNT_STEP;
>  
>  	/*
> -	 * Waiting the load value operation done,
> -	 * it needs two or three RTC clock cycles.
> +	 * Checking busy bit to make sure the previous loading operation is
> +	 * done. According to the specification, the busy bit would be set
> +	 * after a new loading operation and last 2 or 3 RTC clock
> +	 * cycles (about 60us~92us).
>  	 */
>  	do {
>  		val = readl_relaxed(wdt->base + SPRD_WDT_INT_RAW);
>  		if (!(val & SPRD_WDT_LD_BUSY_BIT))
>  			break;
>  
> -		cpu_relax();
> +		usleep_range(10, 100);
>  	} while (delay_cnt++ < SPRD_WDT_LOAD_TIMEOUT);
>  
>  	if (delay_cnt >= SPRD_WDT_LOAD_TIMEOUT)
>