Re: [PATCH] rtc: s3c: Rewrite clock handling

[Krzysztof Kozlowski <krzk@kernel.org>]

On Mon, 21 Jan 2019 at 09:39, Marek Szyprowski <m.szyprowski@samsung.com> wrote:
>
> Hi Krzysztof,
>
> On 2019-01-19 21:17, Krzysztof Kozlowski wrote:
> > On Fri, Jan 18, 2019 at 02:27:54PM +0100, Marek Szyprowski wrote:
> >> s3c_rtc_enable/disable_clk() functions were designed to be called multiple
> >> times without reference counting, because they were initially used in
> > s/used/used only/
> > (if I understand correctly the logic)
> >
> >> alarm setting/clearing functions, which can be called both when alarm is
> >> already set or not. Later however, calls to those functions have been added to
> >> other places in the driver - like time and /proc reading callbacks, what
> >> results in broken alarm if any of such events happens after the alarm has
> >> been set. Fix this by simplifying s3c_rtc_enable/disable_clk() functions
> >> to rely on proper reference counting in clock core and move alarm enable
> >> counter to s3c_rtc_setaie() function.
> >>
> >> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
> >> ---
> >>  drivers/rtc/rtc-s3c.c | 67 ++++++++++++++++++-------------------------
> >>  1 file changed, 28 insertions(+), 39 deletions(-)
> >>
> >> diff --git a/drivers/rtc/rtc-s3c.c b/drivers/rtc/rtc-s3c.c
> >> index 04c68178c42d..e682977b4f6e 100644
> >> --- a/drivers/rtc/rtc-s3c.c
> >> +++ b/drivers/rtc/rtc-s3c.c
> >> @@ -39,7 +39,7 @@ struct s3c_rtc {
> >>      void __iomem *base;
> >>      struct clk *rtc_clk;
> >>      struct clk *rtc_src_clk;
> >> -    bool clk_disabled;
> >> +    bool alarm_enabled;
> >>
> >>      const struct s3c_rtc_data *data;
> >>
> >> @@ -47,7 +47,7 @@ struct s3c_rtc {
> >>      int irq_tick;
> >>
> >>      spinlock_t pie_lock;
> >> -    spinlock_t alarm_clk_lock;
> >> +    spinlock_t alarm_lock;
> > Maybe add short comment that it protects only "alarm_enabled" property?
> >
> >>
> >>      int ticnt_save;
> >>      int ticnt_en_save;
> >> @@ -70,44 +70,28 @@ struct s3c_rtc_data {
> >>
> >>  static int s3c_rtc_enable_clk(struct s3c_rtc *info)
> >>  {
> >> -    unsigned long irq_flags;
> >>      int ret = 0;
> >>
> >> -    spin_lock_irqsave(&info->alarm_clk_lock, irq_flags);
> >> +    ret = clk_enable(info->rtc_clk);
> >> +    if (ret)
> >> +            goto out;
> > The out label is now empty so just "return ret". It is easier to read -
> > no need to jump anywhere to see the simple return.
> >
> >>
> >> -    if (info->clk_disabled) {
> >> -            ret = clk_enable(info->rtc_clk);
> >> -            if (ret)
> >> +    if (info->data->needs_src_clk) {
> >> +            ret = clk_enable(info->rtc_src_clk);
> >> +            if (ret) {
> >> +                    clk_disable(info->rtc_clk);
> >>                      goto out;
> >> -
> >> -            if (info->data->needs_src_clk) {
> >> -                    ret = clk_enable(info->rtc_src_clk);
> >> -                    if (ret) {
> >> -                            clk_disable(info->rtc_clk);
> >> -                            goto out;
> >> -                    }
> >>              }
> >> -            info->clk_disabled = false;
> >>      }
> >> -
> >>  out:
> >> -    spin_unlock_irqrestore(&info->alarm_clk_lock, irq_flags);
> >> -
> >>      return ret;
> >>  }
> >>
> >>  static void s3c_rtc_disable_clk(struct s3c_rtc *info)
> >>  {
> >> -    unsigned long irq_flags;
> >> -
> >> -    spin_lock_irqsave(&info->alarm_clk_lock, irq_flags);
> >> -    if (!info->clk_disabled) {
> >> -            if (info->data->needs_src_clk)
> >> -                    clk_disable(info->rtc_src_clk);
> >> -            clk_disable(info->rtc_clk);
> >> -            info->clk_disabled = true;
> >> -    }
> >> -    spin_unlock_irqrestore(&info->alarm_clk_lock, irq_flags);
> >> +    if (info->data->needs_src_clk)
> >> +            clk_disable(info->rtc_src_clk);
> >> +    clk_disable(info->rtc_clk);
> >>  }
> >>
> >>  /* IRQ Handlers */
> >> @@ -135,6 +119,7 @@ static irqreturn_t s3c_rtc_alarmirq(int irq, void *id)
> >>  static int s3c_rtc_setaie(struct device *dev, unsigned int enabled)
> >>  {
> >>      struct s3c_rtc *info = dev_get_drvdata(dev);
> >> +    unsigned long flags;
> >>      unsigned int tmp;
> >>      int ret;
> >>
> >> @@ -151,17 +136,19 @@ static int s3c_rtc_setaie(struct device *dev, unsigned int enabled)
> >>
> >>      writeb(tmp, info->base + S3C2410_RTCALM);
> >>
> >> -    s3c_rtc_disable_clk(info);
> >> +    spin_lock_irqsave(&info->alarm_lock, flags);
> >>
> >> -    if (enabled) {
> >> -            ret = s3c_rtc_enable_clk(info);
> >> -            if (ret)
> >> -                    return ret;
> >> -    } else {
> >> +    if (info->alarm_enabled && !enabled)
> >>              s3c_rtc_disable_clk(info);
> >> -    }
> >> +    else if (!info->alarm_enabled && enabled)
> >> +            ret = s3c_rtc_enable_clk(info);
> >>
> >> -    return 0;
> >> +    info->alarm_enabled = enabled;
> >> +    spin_unlock_irqrestore(&info->alarm_lock, flags);
> >> +
> >> +    s3c_rtc_disable_clk(info);
> >> +
> >> +    return ret;
> >>  }
> >>
> >>  /* Set RTC frequency */
> >> @@ -357,10 +344,10 @@ static int s3c_rtc_setalarm(struct device *dev, struct rtc_wkalrm *alrm)
> >>
> >>      writeb(alrm_en, info->base + S3C2410_RTCALM);
> >>
> >> -    s3c_rtc_disable_clk(info);
> >> -
> >>      s3c_rtc_setaie(dev, alrm->enabled);
> >>
> >> +    s3c_rtc_disable_clk(info);
> > I do not understand this change - why do you have to move the disable
> > clk? The s3c_rtc_setaie() takes care about clock enabling/disabling for
> > the time of accessing registers.
>
> This was micro optimization, s3c_rtc_setaie() increases clock enable
> count, so by changing the call order we can avoid one disable/enable
> sequence.

OK.

> >> +
> >>      return 0;
> >>  }
> >>
> >> @@ -491,7 +478,7 @@ static int s3c_rtc_probe(struct platform_device *pdev)
> >>              return -EINVAL;
> >>      }
> >>      spin_lock_init(&info->pie_lock);
> >> -    spin_lock_init(&info->alarm_clk_lock);
> >> +    spin_lock_init(&info->alarm_lock);
> >>
> >>      platform_set_drvdata(pdev, info);
> >>
> >> @@ -591,6 +578,8 @@ static int s3c_rtc_probe(struct platform_device *pdev)
> >>
> >>      s3c_rtc_setfreq(info, 1);
> >>
> >> +    s3c_rtc_disable_clk(info);
> > I cannot find the reason why this is related to this particular change.
> > I mean, it looks reasonable because previously the clock looked like it
> > was enabled all the time... but maybe this should be separate pach?
>
> It wasn't enabled all the time, because the call to s3c_rtc_setfreq()
> disabled it (remember there was no clock enable reference counting!).
> Now once we have enable/disable reference counting, we need to keep them
> balanced.

I get it, thanks!

Best regards,
Krzysztof