linux-kernel - Re: [PATCH V2] rtc: zynqmp: Update seconds time programming logic

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: <20160420140310.GS29844@piout.net>
Date:	Wed, 20 Apr 2016 16:03:10 +0200
From:	Alexandre Belloni <alexandre.belloni@...e-electrons.com>
To:	Anurag Kumar Vulisha <anurag.kumar.vulisha@...inx.com>
Cc:	Alessandro Zummo <a.zummo@...ertech.it>,
	soren.brinkmann@...inx.com, Michal Simek <michal.simek@...inx.com>,
	rtc-linux@...glegroups.com, linux-arm-kernel@...ts.infradead.org,
	linux-kernel@...r.kernel.org,
	Punnaiah Choudary Kalluri <punnaia@...inx.com>,
	Anirudha Sarangi <anirudh@...inx.com>,
	Srikanth Vemula <svemula@...inx.com>,
	Srinivas Goud <sgoud@...inx.com>,
	Anurag Kumar Vulisha <anuragku@...inx.com>
Subject: Re: [PATCH V2] rtc: zynqmp: Update seconds time programming logic

On 20/04/2016 at 19:21:19 +0530, Anurag Kumar Vulisha wrote :
> We programe RTC time using SET_TIME_WRITE register and read the RTC
> current time using CURRENT_TIME register. When we set the time by
> writing into SET_TIME_WRITE Register and immediately try to read the
> rtc time from CURRENT_TIME register, the previous old value is
> returned instead of the new loaded time. This is because RTC takes
> nearly 1 sec to update the  new loaded value into the CURRENT_TIME
> register. This behaviour is expected in our RTC IP.
> 
> This patch updates the driver to read the current time from SET_TIME_WRITE
> register instead of CURRENT_TIME when rtc time is requested within an 1sec
> period after setting the RTC time. Doing so will ensure the correct time
> is given to the user.
> 
> Since there is an delay of 1sec in updating the CURRENT_TIME we are loading
> set time +1sec while programming the SET_TIME_WRITE register, doing this
> will give correct time without any delay when read from CURRENT_TIME.
> 
> This patch updates the above said.
> 
> Signed-off-by: Anurag Kumar Vulisha <anuragku@...inx.com>
> ---
>  Changes in v2:
>     1. Updated the Time programming logic as suggested by Alexandre Belloni
>     2. Changed the commit message
> ---
>  drivers/rtc/rtc-zynqmp.c | 41 ++++++++++++++++++++++++++++++++++++-----
>  1 file changed, 36 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/rtc/rtc-zynqmp.c b/drivers/rtc/rtc-zynqmp.c
> index f87f971..ba4203a 100644
> --- a/drivers/rtc/rtc-zynqmp.c
> +++ b/drivers/rtc/rtc-zynqmp.c
> @@ -64,6 +64,12 @@ static int xlnx_rtc_set_time(struct device *dev, struct rtc_time *tm)
>  	struct xlnx_rtc_dev *xrtcdev = dev_get_drvdata(dev);
>  	unsigned long new_time;
>  
> +	/*
> +	 * The value written will be updated after 1 sec into the
> +	 * seconds read register, so we need to program time +1 sec
> +	 * to get the correct time on read.
> +	 */
> +	tm->tm_sec += 1;
>  	new_time = rtc_tm_to_time64(tm);
>  
>  	if (new_time > RTC_SEC_MAX_VAL)
> @@ -78,14 +84,41 @@ static int xlnx_rtc_set_time(struct device *dev, struct rtc_time *tm)
>  
>  	writel(new_time, xrtcdev->reg_base + RTC_SET_TM_WR);
>  
> +	/*
> +	 * Clear the rtc interrupt status register after setting the
> +	 * time. During a read_time function, the code should read the
> +	 * RTC_INT_STATUS register and if bit 0 is still 0, it means
> +	 * that one second has not elapsed yet since RTC was set and
> +	 * the current time should be read from SET_TIME_READ register;
> +	 * otherwise, CURRENT_TIME register is read to report the time
> +	 */
> +	writel(RTC_INT_SEC, xrtcdev->reg_base + RTC_INT_STS);
> +
>  	return 0;
>  }
>  
>  static int xlnx_rtc_read_time(struct device *dev, struct rtc_time *tm)
>  {
> +	int status;
>  	struct xlnx_rtc_dev *xrtcdev = dev_get_drvdata(dev);
>  
> -	rtc_time64_to_tm(readl(xrtcdev->reg_base + RTC_CUR_TM), tm);
> +	status = readl(xrtcdev->reg_base + RTC_INT_STS);
> +
> +	if (status & RTC_INT_SEC) {
> +		/*
> +		 * RTC has updated the CURRENT_TIME with the time written into
> +		 * SET_TIME_WRITE register.
> +		 */
> +		rtc_time64_to_tm(readl(xrtcdev->reg_base + RTC_CUR_TM), tm);
> +	} else {
> +		/*
> +		 * Time written in SET_TIME_WRITE has not yet updated into
> +		 * the seconds read register, so read the time from the
> +		 * SET_TIME_WRITE instead of CURRENT_TIME register.
> +		 */
> +		rtc_time64_to_tm(readl(xrtcdev->reg_base + RTC_SET_TM_RD), tm);
> +		tm->tm_sec -= 1;

Well, I didn't point it out directly earlier but this doesn't work while
my example is working:
rtc_time64_to_tm(readl(xrtcdev->reg_base + RTC_SET_TM_RD) - 1, tm);
Think about what is happening when tm->tm_sec == 0 ...

> +	}
>  
>  	return rtc_valid_tm(tm);
>  }
> @@ -166,11 +199,9 @@ static irqreturn_t xlnx_rtc_interrupt(int irq, void *id)
>  	if (!(status & (RTC_INT_SEC | RTC_INT_ALRM)))
>  		return IRQ_NONE;
>  
> -	/* Clear interrupt */
> -	writel(status, xrtcdev->reg_base + RTC_INT_STS);
> +	/* Clear RTC_INT_ALRM interrupt only */
> +	writel(RTC_INT_ALRM, xrtcdev->reg_base + RTC_INT_STS);
>  
> -	if (status & RTC_INT_SEC)
> -		rtc_update_irq(xrtcdev->rtc, 1, RTC_IRQF | RTC_UF);
>  	if (status & RTC_INT_ALRM)
>  		rtc_update_irq(xrtcdev->rtc, 1, RTC_IRQF | RTC_AF);
>  
> -- 
> 2.1.2
> 

-- 
Alexandre Belloni, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com