Re: Re: [PATCH] rtc: rs5c372: Fix read the time from RTC is illegal When reading time from an uninitialized RTC chip, The value may be illegal

From: Alexandre Belloni
Date: Wed May 26 2021 - 08:13:05 EST


On 26/05/2021 19:35:31+0800, qxj511mail@xxxxxxxxx wrote:
> > If the time on the RTC is invalid, then rtc_valid_tm has to fail, there
> > is nothing to fix here.
>
> I mean, if you try to get the time or execute hwclock command, Invalid arguments are returned when the RTC chip is not initialized
> therefore,we need to get a valid time
>

Well, the time is invalid so hwclock needs to know it is invalid.
Returning a bogus time and date doesn't make it right. hwclock needs to
fail.

>
>
> qxj511mail@xxxxxxxxx
>
> From: Alexandre Belloni
> Date: 2021-05-26 05:18
> To: qxj511mail@xxxxxxxxx
> CC: Nobuhiro Iwamatsu; a.zummo; linux-rtc; linux-kernel; 邱晓金
> Subject: Re: Re: [PATCH] rtc: rs5c372: Fix read the time from RTC is illegal When reading time from an uninitialized RTC chip, The value may be illegal
> Hi,
>
> On 25/05/2021 19:24:14+0800, qxj511mail@xxxxxxxxx wrote:
> > Please briefly describe the patch contained in the email to the subject.
> > And, please write a description of the patch in the text
> > ----> The legal days are 1 to 31 and the legal month is less than or equal to 12.
> > But for an uninitialized RTC chip, the time is random.
> > Days and month may be zero, leading to RTC_valid_TM failed
> >
>
> If the time on the RTC is invalid, then rtc_valid_tm has to fail, there
> is nothing to fix here.
>
> > add description && fix build error, so I modefied my code :
> >
> > Signed-off-by: qiuxiaojin <qiuxiaojin@xxxxxxxx>
> > ---
> > drivers/rtc/rtc-rs5c372.c | 20 ++++++++++++++++++++
> > 1 file changed, 20 insertions(+)
> >
> > diff --git a/drivers/rtc/rtc-rs5c372.c b/drivers/rtc/rtc-rs5c372.c
> > index 3bd6eaa0dcf6..cb3f90983778 100644
> > --- a/drivers/rtc/rtc-rs5c372.c
> > +++ b/drivers/rtc/rtc-rs5c372.c
> > @@ -128,6 +128,9 @@ struct rs5c372 {
> > char *regs;
> > };
> >
> > +
> > +static int rs5c372_rtc_set_time(struct device *dev, struct rtc_time *tm);
> > +
> > static int rs5c_get_regs(struct rs5c372 *rs5c)
> > {
> > struct i2c_client *client = rs5c->client;
> > @@ -212,6 +215,7 @@ static int rs5c372_rtc_read_time(struct device *dev, struct rtc_time *tm)
> > struct rs5c372 *rs5c = i2c_get_clientdata(client);
> > int status = rs5c_get_regs(rs5c);
> > unsigned char ctrl2 = rs5c->regs[RS5C_REG_CTRL2];
> > + int flags_utime = 0;
> >
> > if (status < 0)
> > return status;
> > @@ -239,12 +243,28 @@ static int rs5c372_rtc_read_time(struct device *dev, struct rtc_time *tm)
> > tm->tm_wday = bcd2bin(rs5c->regs[RS5C372_REG_WDAY] & 0x07);
> > tm->tm_mday = bcd2bin(rs5c->regs[RS5C372_REG_DAY] & 0x3f);
> >
> > + /* The value read from the register may be zero, which is an illegal value */
> > + if (tm->tm_mday < 1) {
> > + flags_utime++;
> > + tm->tm_mday = 1;
> > + }
> > +
> > /* tm->tm_mon is zero-based */
> > tm->tm_mon = bcd2bin(rs5c->regs[RS5C372_REG_MONTH] & 0x1f) - 1;
> >
> > + if (tm->tm_mon < 0) {
> > + /* avoid illegal month */
> > + flags_utime++;
> > + tm->tm_mon = 0;
> > + }
> > +
> > /* year is 1900 + tm->tm_year */
> > tm->tm_year = bcd2bin(rs5c->regs[RS5C372_REG_YEAR]) + 100;
> >
> > + /* update legal time */
> > + if (flags_utime > 0)
> > + rs5c372_rtc_set_time(dev, tm);
> > +
> > dev_dbg(&client->dev, "%s: tm is secs=%d, mins=%d, hours=%d, "
> > "mday=%d, mon=%d, year=%d, wday=%d\n",
> > __func__,
> > --
> > 2.29.0
> >
> >
> >
> > qxj511mail@xxxxxxxxx
> >
> > From: Nobuhiro Iwamatsu
> > Date: 2021-05-24 12:40
> > To: qxj511mail
> > CC: Alessandro Zummo; Alexandre Belloni; linux-rtc; Linux Kernel Mailing List; qiuxiaojin
> > Subject: Re: [PATCH] rtc: rs5c372: Fix read the time from RTC is illegal When reading time from an uninitialized RTC chip, The value may be illegal
> > Hi,
> >
> > 2021年5月20日(木) 12:32 <qxj511mail@xxxxxxxxx>:
> > >
> > > From: qiuxiaojin <qiuxiaojin@xxxxxxxx>
> >
> > Please briefly describe the patch contained in the email to the subject.
> > And, please write a description of the patch in the text
> >
> > >
> > > Signed-off-by: qiuxiaojin <qiuxiaojin@xxxxxxxx>
> > > ---
> > > drivers/rtc/rtc-rs5c372.c | 16 ++++++++++++++++
> > > 1 file changed, 16 insertions(+)
> > >
> > > diff --git a/drivers/rtc/rtc-rs5c372.c b/drivers/rtc/rtc-rs5c372.c
> > > index 3bd6eaa0dcf6..ce61e15d5f3a 100644
> > > --- a/drivers/rtc/rtc-rs5c372.c
> > > +++ b/drivers/rtc/rtc-rs5c372.c
> > > @@ -212,6 +212,7 @@ static int rs5c372_rtc_read_time(struct device *dev, struct rtc_time *tm)
> > > struct rs5c372 *rs5c = i2c_get_clientdata(client);
> > > int status = rs5c_get_regs(rs5c);
> > > unsigned char ctrl2 = rs5c->regs[RS5C_REG_CTRL2];
> > > + int flags_utime = 0;
> > >
> > > if (status < 0)
> > > return status;
> > > @@ -239,12 +240,27 @@ static int rs5c372_rtc_read_time(struct device *dev, struct rtc_time *tm)
> > > tm->tm_wday = bcd2bin(rs5c->regs[RS5C372_REG_WDAY] & 0x07);
> > > tm->tm_mday = bcd2bin(rs5c->regs[RS5C372_REG_DAY] & 0x3f);
> > >
> > > + if (tm->tm_mday < 1) {
> > > + // The value read from the register may be zero, which is an illegal value
> >
> > Please use C89 style commet (/* */).
> >
> > > + flags_utime = flags_utime + 1;
> >
> > I like using ++ (flags_utime++).
> >
> > > + tm->tm_mday = 1;
> > > + }
> > > +
> > > /* tm->tm_mon is zero-based */
> > > tm->tm_mon = bcd2bin(rs5c->regs[RS5C372_REG_MONTH] & 0x1f) - 1;
> > >
> > > + if (tm->tm_mon < 0) {
> > > + flags_utime = flags_utime + 1;
> > > + tm->tm_mday = 0;
> > > + }
> > > +
> > > /* year is 1900 + tm->tm_year */
> > > tm->tm_year = bcd2bin(rs5c->regs[RS5C372_REG_YEAR]) + 100;
> > >
> > > + if (flags_utime > 0) {
> >
> > {} is unnecessary.
> >
> > > + rs5c372_rtc_set_time(dev, tm);
> >
> > A build error will occur because there is no declaration of
> > rs5c372_rtc_set_time.
> > Please make sure this can be compiled.
> >
> > > + }
> > > +
> > > dev_dbg(&client->dev, "%s: tm is secs=%d, mins=%d, hours=%d, "
> > > "mday=%d, mon=%d, year=%d, wday=%d\n",
> > > __func__,
> > > --
> > > 2.29.0
> > >
> >
> >
> > --
> > Nobuhiro Iwamatsu
> > iwamatsu at {nigauri.org / debian.org}
> > GPG ID: 40AD1FA6
>
>
>
> --
> Alexandre Belloni, co-owner and COO, Bootlin
> Embedded Linux and Kernel engineering
> https://bootlin.com

--
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com