Re: [PATCH] rtc: ftrtc010: fix cast

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] rtc: ftrtc010: broken operator precedence in offset calculation
--

commit d205fb0cae60ed15e7685f77950862e58120dd5a
Author: Alexandre Belloni <alexandre.belloni@bootlin.com>

rtc: ftrtc010: fix cast

This commit removes unnecessary casts to timeu64_t from the hour, min, and sec
registers during the calculation of time and offset, as they will rollover
before exceeding a 32-bit integer.

Link: https://patch.msgid.link/20261002105946.8rdteIpO7uWGEcTlS65F8iSi4gsRoXi2kgq8FTSaDPA@z/

> diff --git a/drivers/rtc/rtc-ftrtc010.c b/drivers/rtc/rtc-ftrtc010.c
> index b29c96be40f4a..5332451bdaa05 100644
> --- a/drivers/rtc/rtc-ftrtc010.c
> +++ b/drivers/rtc/rtc-ftrtc010.c

[ ... ]

> @@ -92,8 +92,7 @@ static int ftrtc010_rtc_set_time(struct device *dev, struct rtc_time *tm)
>  	hour = readl(rtc->rtc_base + FTRTC010_RTC_HOUR);
>  	day = readl(rtc->rtc_base + FTRTC010_RTC_DAYS);
>  
> -	offset = time - ((timeu64_t)day * 86400 + (timeu64_t)hour * 3600 +
> -			 (timeu64_t)min * 60 + (timeu64_t)sec);
> +	offset = time - (timeu64_t)day * 86400 + hour * 3600 + min * 60 + sec;

[Severity: High]
Could this change introduce an operator precedence issue?

Because the grouping parentheses around the time components were removed,
the expression evaluates strictly from left to right. This subtracts only
the days from the system time, and erroneously adds the hour, min, and sec
components instead of subtracting them.

When user-space programs or system time synchronizations invoke
ftrtc010_rtc_set_time(), this will predictably corrupt the stored hardware
time offset written to the FTRTC010_RTC_RECORD register.

>  
>  	writel(offset, rtc->rtc_base + FTRTC010_RTC_RECORD);
>  	writel(0x01, rtc->rtc_base + FTRTC010_RTC_CR);

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261005110638.156559-1-alexandre.belloni@bootlin.com?part=1