diff mbox

Thermal: Fix bug on generic thermal framework.

Message ID 1348538339.10877.193.camel@rui.sh.intel.com (mailing list archive)
State Not Applicable, archived
Headers show

Commit Message

Zhang, Rui Sept. 25, 2012, 1:58 a.m. UTC
On ?, 2012-09-25 at 10:12 +0900, jonghwa3.lee@samsung.com wrote:
> On 2012? 09? 24? 17:57, Zhang Rui wrote:
> > On ?, 2012-09-24 at 02:08 -0600, R, Durgadoss wrote:
> >> Hi,
> >>
> >> Patch is fine, but I think you have to re-base on top of
> >> Rui's -next branch here:
> >> git://git.kernel.org/pub/scm/linux/kernel/git/rzhang/linux.git
> >>
> >> Also, adding Rui to this mail, not sure whether he is in LKML/pm.
> >>
> >> Thanks,
> >> Durga
> >>
> >>> -----Original Message-----
> >>> From: Jonghwa Lee [mailto:jonghwa3.lee@samsung.com]
> >>> Sent: Monday, September 24, 2012 7:36 AM
> >>> To: linux-pml@vger.kernel.org
> >>> Cc: linux-kernel@vger.kernel.org; Brown, Len; Rafael J. Wysocki; Andrew
> >>> Morton; Amit Kachhap; R, Durgadoss; Jonghwa Lee
> >>> Subject: [PATCH] Thermal: Fix bug on generic thermal framework.
> >>>
> >>> When system fails to bind cooling devices to thermal zone device during
> >>> registering thermal zone device, it leaves registering without canceling
> >>> delayed work. It probably makes panic if polling rate is not enough to release
> >>> that work from workqueue. So it is better to ignore initialization of polling
> >>> work to prevent that unexpected state.
> >>>
> > Hi, Jonghwa,
> >
> > I still do not understand what the problem is.
> > Say if a cooling device fails to bind, the thermal zone device would
> > still work properly, just like the failure cooling device is not
> > referenced in this thermal zone.
> >
> > thanks,
> > rui
> Hi rui,
> No, it doesn't work properly. If it fails to bind some cool dev to
> thermal zone device, it frees thermal zone
> device without canceling delayed work. After freeing thermal zone
> device, system may call work function
> pointed NULL as the timer expired. Thus it requires skipping the
> initialization of polling work or canceling before
> the unregister.


hah, I see what the problem is.
ideally, if we fail to bind one cooling device, we should just ignore it
and continue to bind other, what do you think?

does the patch below fix your problem?
If yes, I'll try to rebase it on top of my next tree.

Signed-off-by: Zhang Rui <rui.zhang@intel.com>
---
 drivers/thermal/thermal_sys.c |   13 +++++--------
 1 files changed, 5 insertions(+), 8 deletions(-)

Comments

durgadoss.r@intel.com Sept. 25, 2012, 3:04 a.m. UTC | #1
Hi Rui,

> > > Hi, Jonghwa,

> > >

> > > I still do not understand what the problem is.

> > > Say if a cooling device fails to bind, the thermal zone device would

> > > still work properly, just like the failure cooling device is not

> > > referenced in this thermal zone.

> > >

> > > thanks,

> > > rui

> > Hi rui,

> > No, it doesn't work properly. If it fails to bind some cool dev to

> > thermal zone device, it frees thermal zone

> > device without canceling delayed work. After freeing thermal zone

> > device, system may call work function

> > pointed NULL as the timer expired. Thus it requires skipping the

> > initialization of polling work or canceling before

> > the unregister.

> 

> 

> hah, I see what the problem is.

> ideally, if we fail to bind one cooling device, we should just ignore it

> and continue to bind other, what do you think?


Yes, this is what we should do.

> 

> does the patch below fix your problem?

> If yes, I'll try to rebase it on top of my next tree.


This is already fixed in your -next tree, since you applied the
fair share patches 10/15. The function bind_tz(tz) does the
exact same thing, and continues.

Thanks,
Durga
Jonghwa Lee Sept. 25, 2012, 4:21 a.m. UTC | #2
On 2012? 09? 25? 12:04, R, Durgadoss wrote:
> Hi Rui,
>
>>>> Hi, Jonghwa,
>>>>
>>>> I still do not understand what the problem is.
>>>> Say if a cooling device fails to bind, the thermal zone device would
>>>> still work properly, just like the failure cooling device is not
>>>> referenced in this thermal zone.
>>>>
>>>> thanks,
>>>> rui
>>> Hi rui,
>>> No, it doesn't work properly. If it fails to bind some cool dev to
>>> thermal zone device, it frees thermal zone
>>> device without canceling delayed work. After freeing thermal zone
>>> device, system may call work function
>>> pointed NULL as the timer expired. Thus it requires skipping the
>>> initialization of polling work or canceling before
>>> the unregister.
>>
>> hah, I see what the problem is.
>> ideally, if we fail to bind one cooling device, we should just ignore it
>> and continue to bind other, what do you think?
> Yes, this is what we should do.
>
>> does the patch below fix your problem?
>> If yes, I'll try to rebase it on top of my next tree.
> This is already fixed in your -next tree, since you applied the
> fair share patches 10/15. The function bind_tz(tz) does the
> exact same thing, and continues.
>
> Thanks,
> Durga
I checked that it had been applied on -next branch, I don't commit this
patch any more.

Thanks,
Jonghwa

> N?????r??y???b?X???v?^?)?{.n?+????{??h?????}????z?&j:+v???????zZ+??+zf???h???~????i???z??w????????&?)?fl===

--
To unsubscribe from this list: send the line "unsubscribe linux-pm" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/drivers/thermal/thermal_sys.c b/drivers/thermal/thermal_sys.c
index 2ab31e4..c5e2c28 100644
--- a/drivers/thermal/thermal_sys.c
+++ b/drivers/thermal/thermal_sys.c
@@ -1343,20 +1343,17 @@  struct thermal_zone_device *thermal_zone_device_register(const char *type,
 
 	mutex_lock(&thermal_list_lock);
 	list_add_tail(&tz->node, &thermal_tz_list);
-	if (ops->bind)
-		list_for_each_entry(pos, &thermal_cdev_list, node) {
-		result = ops->bind(tz, pos);
-		if (result)
-			break;
-		}
+	if (ops->bind) {
+		list_for_each_entry(pos, &thermal_cdev_list, node)
+			ops->bind(tz, pos);
+	}
 	mutex_unlock(&thermal_list_lock);
 
 	INIT_DELAYED_WORK(&(tz->poll_queue), thermal_zone_device_check);
 
 	thermal_zone_device_update(tz);
 
-	if (!result)
-		return tz;
+	return tz;
 
 unregister:
 	release_idr(&thermal_tz_idr, &thermal_idr_lock, tz->id);