RE: [PATCH 2/3] alarmtimer: Using the alarmtimer_get_rtcdev for all posix clock interface
> rtcdev = NULL
>
> CPU0 CPU 1
>
> alarmtimer_rtc_add_device(A)
> if (rtcdev) alarmtimer_rtc_add_device(B)
>return -EBUSY; if (rtcdev)
> mutex_lock(); return -EBUSY;
> rtcdev = A; mutex_lock();
> mutex_unlock;
>
> bla = alarmtimer_rtc_get_device()
>
> So bla = A
> mutex_lock() returns
> rtcdev = B;
> mutex_unlock();
>
> The next call to alarmtimer_rtc_get_device() will return B. Not what
> you want. Maybe you want that, but then your patch is missing an
> explanation why you want that and why this would be a desired
> behaviour.
>
Based on current implementation alarmtimer_rtc_add_device() will just
be called once thru class interface, also &parent->p->mutex is there to protect
the class_intf->add_dev() calling, so we can remove rtcdev_lock.
Is it making sense?
Subject: [PATCH] alarmtimer: Removing the usage of lock rtcdev_lock
alarmtimer_rtc_add_device() will be called once at the time
alarmtimer_init() is called.
alarmtimer_init -->
alarmtimer_rtc_interface_setup -->
class_interface_register -->
mutex_lock(&parent->p->mutex)
...
class_intf->add_dev()/alarmtimer_rtc_add_device
...
mutex_unlock(&parent->p->mutex);
But also the mutex parent->p->mutex has protected the add_dev()
callback.
So here removing the lock of rtcdev_lock.
Signed-off-by: liu chuansheng
---
kernel/time/alarmtimer.c | 13 -
1 files changed, 4 insertions(+), 9 deletions(-)
diff --git a/kernel/time/alarmtimer.c b/kernel/time/alarmtimer.c
index eed3b26..660091c 100644
--- a/kernel/time/alarmtimer.c
+++ b/kernel/time/alarmtimer.c
@@ -51,7 +51,6 @@ static struct wakeup_source *ws;
/* rtc timer and device for setting alarm wakeups at suspend */
static struct rtc_timerrtctimer;
static struct rtc_device *rtcdev;
-static DEFINE_SPINLOCK(rtcdev_lock);
/**
* alarmtimer_get_rtcdev - Return selected rtcdevice
@@ -69,7 +68,6 @@ struct rtc_device *alarmtimer_get_rtcdev(void)
static int alarmtimer_rtc_add_device(struct device *dev,
struct class_interface *class_intf)
{
- unsigned long flags;
struct rtc_device *rtc = to_rtc_device(dev);
if (rtcdev)
@@ -80,13 +78,10 @@ static int alarmtimer_rtc_add_device(struct device *dev,
if (!device_may_wakeup(rtc->dev.parent))
return -1;
- spin_lock_irqsave(&rtcdev_lock, flags);
- if (!rtcdev) {
- rtcdev = rtc;
- /* hold a reference so it doesn't go away */
- get_device(dev);
- }
- spin_unlock_irqrestore(&rtcdev_lock, flags);
+ rtcdev = rtc;
+ /* hold a reference so it doesn't go away */
+ get_device(dev);
+
return 0;
}
--
1.7.0.4
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
RE: [PATCH 2/3] alarmtimer: Using the alarmtimer_get_rtcdev for all posix clock interface
> -Original Message-
> From: Thomas Gleixner [mailto:t...@linutronix.de]
> Sent: Friday, November 02, 2012 6:56 AM
> To: Liu, Chuansheng
> Cc: john.stu...@linaro.org; gre...@linuxfoundation.org;
> linux-kernel@vger.kernel.org
> Subject: Re: [PATCH 2/3] alarmtimer: Using the alarmtimer_get_rtcdev for all
> posix clock interface
>
> On Thu, 1 Nov 2012, Chuansheng Liu wrote:
> >
> > Some posix clock interface directly use the variable rtcdev,
> > cleanup it here by alarmtimer_get_rtcdev().
> >
> > Signed-off-by: liu chuansheng
> > ---
> > kernel/time/alarmtimer.c | 13 ++---
> > 1 files changed, 6 insertions(+), 7 deletions(-)
> >
> > diff --git a/kernel/time/alarmtimer.c b/kernel/time/alarmtimer.c
> > index 4fc17cb..5490fa8 100644
> > --- a/kernel/time/alarmtimer.c
> > +++ b/kernel/time/alarmtimer.c
> > @@ -86,11 +86,10 @@ static int alarmtimer_rtc_add_device(struct device
> *dev,
> > return -1;
> >
> > mutex_lock(&rtcdev_mutex);
> > - if (!rtcdev) {
> > - rtcdev = rtc;
> > - /* hold a reference so it doesn't go away */
> > - get_device(dev);
> > - }
> > + rtcdev = rtc;
> > + /* hold a reference so it doesn't go away */
> > + get_device(dev);
> > +
>
> Brilliant.
>
> rtcdev = NULL
>
> CPU0 CPU 1
>
> alarmtimer_rtc_add_device(A)
> if (rtcdev) alarmtimer_rtc_add_device(B)
>return -EBUSY; if (rtcdev)
> mutex_lock(); return -EBUSY;
> rtcdev = A; mutex_lock();
> mutex_unlock;
>
> bla = alarmtimer_rtc_get_device()
>
> So bla = A
> mutex_lock() returns
> rtcdev = B;
> mutex_unlock();
>
> The next call to alarmtimer_rtc_get_device() will return B. Not what
> you want. Maybe you want that, but then your patch is missing an
> explanation why you want that and why this would be a desired
> behaviour.
Thanks your pointing out, I am wrong.
>
> Thanks,
>
> tglx
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH 2/3] alarmtimer: Using the alarmtimer_get_rtcdev for all posix clock interface
On Thu, 1 Nov 2012, Chuansheng Liu wrote:
>
> Some posix clock interface directly use the variable rtcdev,
> cleanup it here by alarmtimer_get_rtcdev().
>
> Signed-off-by: liu chuansheng
> ---
> kernel/time/alarmtimer.c | 13 ++---
> 1 files changed, 6 insertions(+), 7 deletions(-)
>
> diff --git a/kernel/time/alarmtimer.c b/kernel/time/alarmtimer.c
> index 4fc17cb..5490fa8 100644
> --- a/kernel/time/alarmtimer.c
> +++ b/kernel/time/alarmtimer.c
> @@ -86,11 +86,10 @@ static int alarmtimer_rtc_add_device(struct device *dev,
> return -1;
>
> mutex_lock(&rtcdev_mutex);
> - if (!rtcdev) {
> - rtcdev = rtc;
> - /* hold a reference so it doesn't go away */
> - get_device(dev);
> - }
> + rtcdev = rtc;
> + /* hold a reference so it doesn't go away */
> + get_device(dev);
> +
Brilliant.
rtcdev = NULL
CPU0CPU 1
alarmtimer_rtc_add_device(A)
if (rtcdev) alarmtimer_rtc_add_device(B)
return -EBUSY; if (rtcdev)
mutex_lock(); return -EBUSY;
rtcdev = A; mutex_lock();
mutex_unlock;
bla = alarmtimer_rtc_get_device()
So bla = A
mutex_lock() returns
rtcdev = B;
mutex_unlock();
The next call to alarmtimer_rtc_get_device() will return B. Not what
you want. Maybe you want that, but then your patch is missing an
explanation why you want that and why this would be a desired
behaviour.
Thanks,
tglx
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
