2020-08-14 10:00:29

by Victor Ding

Subject: [PATCH v2] rtc: cmos: zero-init wkalrm when reading from CMOS

cmos_read_alarm() may leave certain fields of a struct rtc_wkalrm
untouched; therefore, these fields contain garbage if not properly
initialized, leading to inconsistent values when converting into
time64_t. This patch to zero initialize the struct before calling
cmos_read_alarm().

Note that this patch is not intended to produce a correct time64_t, it
is only to produce a consistent value. In the case of suspend/resume, a
correct time64_t is not necessary; a consistent value is sufficient to
correctly perform an equality test for t_current_expires and
t_saved_expires. Logic to deduce a correct time64_t is expensive and
hence should be avoided.

Signed-off-by: Victor Ding <[email protected]>
---

Changes in v2:
- Initialize the struct to 0 instead of -1;
- Initialize the whole struct rtc_wkalrm.

drivers/rtc/rtc-cmos.c | 2 ++
1 file changed, 2 insertions(+)

diff --git a/drivers/rtc/rtc-cmos.c b/drivers/rtc/rtc-cmos.c
index bcc96ab7793f..c633319cdb91 100644
--- a/drivers/rtc/rtc-cmos.c
+++ b/drivers/rtc/rtc-cmos.c
@@ -1006,6 +1006,7 @@ static int cmos_suspend(struct device *dev)
enable_irq_wake(cmos->irq);
}

+ memset(&cmos->saved_wkalrm, 0, sizeof(struct rtc_wkalrm));
cmos_read_alarm(dev, &cmos->saved_wkalrm);

dev_dbg(dev, "suspend%s, ctrl %02x\n",
@@ -1054,6 +1055,7 @@ static void cmos_check_wkalrm(struct device *dev)
return;
}

+ memset(&current_alarm, 0, sizeof(struct rtc_wkalrm));
cmos_read_alarm(dev, &current_alarm);
t_current_expires = rtc_tm_to_time64(&current_alarm.time);
t_saved_expires = rtc_tm_to_time64(&cmos->saved_wkalrm.time);
--
2.28.0.220.ged08abb693-goog

2020-08-20 23:59:14

by Alexandre Belloni

Subject: Re: [PATCH v2] rtc: cmos: zero-init wkalrm when reading from CMOS

On Fri, 14 Aug 2020 19:17:30 +1000, Victor Ding wrote:
> cmos_read_alarm() may leave certain fields of a struct rtc_wkalrm
> untouched; therefore, these fields contain garbage if not properly
> initialized, leading to inconsistent values when converting into
> time64_t. This patch to zero initialize the struct before calling
> cmos_read_alarm().
>
> Note that this patch is not intended to produce a correct time64_t, it
> is only to produce a consistent value. In the case of suspend/resume, a
> correct time64_t is not necessary; a consistent value is sufficient to
> correctly perform an equality test for t_current_expires and
> t_saved_expires. Logic to deduce a correct time64_t is expensive and
> hence should be avoided.

Applied, thanks!

[1/1] rtc: cmos: zero-init wkalrm when reading from CMOS
commit: c254bcd7231a3eeafc453f6ee3a483a2e7ff486e

Best regards,
--
Alexandre Belloni <[email protected]>