[patch 08/12] thermal: add sanity check for the passive attribute

[patch 08/12] thermal: add sanity check for the passive attribute

[akpm]

From: Frans Pop <elendil@planet.nl>

Values below 1000 milli-celsius don't make sense and can cause the system
to go into a thermal heart attack: the actual temperature will always be
lower and thus the system will be throttled down to its lowest setting.

An additional problem is that values below 1000 will show as 0 in
/proc/acpi/thermal/TZx/trip_points:passive.

cat passive
0
echo -n 90 >passive
bash: echo: write error: Invalid argument
echo -n 90000 >passive
cat passive
90000

Addresses http://bugzilla.kernel.org/show_bug.cgi?id=13918

Signed-off-by: Frans Pop <elendil@planet.nl>
Cc: Matthew Garrett <mjg@redhat.com>
Cc: Zhang Rui <rui.zhang@intel.com>
Cc: Len Brown <lenb@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 Documentation/thermal/sysfs-api.txt |    1 +
 drivers/thermal/thermal_sys.c       |    6 ++++++
 2 files changed, 7 insertions(+)

diff -puN Documentation/thermal/sysfs-api.txt~thermal-add-sanity-check-for-the-passive-attribute Documentation/thermal/sysfs-api.txt
--- a/Documentation/thermal/sysfs-api.txt~thermal-add-sanity-check-for-the-passive-attribute
+++ a/Documentation/thermal/sysfs-api.txt
@@ -206,6 +206,7 @@ passive
 	passive trip point for the zone. Activation is done by polling with
 	an interval of 1 second.
 	Unit: millidegrees Celsius
+	Valid values: 0 (disabled) or greater than 1000
 	RW, Optional
 
 *****************************
diff -puN drivers/thermal/thermal_sys.c~thermal-add-sanity-check-for-the-passive-attribute drivers/thermal/thermal_sys.c
--- a/drivers/thermal/thermal_sys.c~thermal-add-sanity-check-for-the-passive-attribute
+++ a/drivers/thermal/thermal_sys.c
@@ -225,6 +225,12 @@ passive_store(struct device *dev, struct
 	if (!sscanf(buf, "%d\n", &state))
 		return -EINVAL;
 
+	/* sanity check: values below 1000 millicelcius don't make sense
+	 * and can cause the system to go into a thermal heart attack
+	 */
+	if (state && state < 1000)
+		return -EINVAL;
+
 	if (state && !tz->forced_passive) {
 		mutex_lock(&thermal_list_lock);
 		list_for_each_entry(cdev, &thermal_cdev_list, node) {
_

Re: [patch 08/12] thermal: add sanity check for the passive attribute

[Matthew Garrett]

As noted before, I'm not a fan of this approach - I can't think of many 
reasons why it'd be necessary to use temperatures below 1 degree C, but 
this constraint isn't present anywhere else in the thermal code and, 
really, there's plenty of things that people can break if they just echo 
incorrect values into /sys.

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: [patch 08/12] thermal: add sanity check for the passive attribute

[Frans Pop]

On Tuesday 17 November 2009, Matthew Garrett wrote:
> As noted before, I'm not a fan of this approach - I can't think of many
> reasons why it'd be necessary to use temperatures below 1 degree C, but
> this constraint isn't present anywhere else in the thermal code and,
> really, there's plenty of things that people can break if they just echo
> incorrect values into /sys.

Personally I'd be in favor of adding more sanity checks rather than 
removing them. The cost of the check is virtually zero.

Cheers,
FJP

Re: [patch 08/12] thermal: add sanity check for the passive attribute

[Matthew Garrett]

On Tue, Nov 17, 2009 at 11:42:55PM +0100, Frans Pop wrote:
> On Tuesday 17 November 2009, Matthew Garrett wrote:
> > As noted before, I'm not a fan of this approach - I can't think of many
> > reasons why it'd be necessary to use temperatures below 1 degree C, but
> > this constraint isn't present anywhere else in the thermal code and,
> > really, there's plenty of things that people can break if they just echo
> > incorrect values into /sys.
> 
> Personally I'd be in favor of adding more sanity checks rather than 
> removing them. The cost of the check is virtually zero.

It makes something that's potentially useful to someone impossible, for 
the benefit of people who can crash their system in half a dozen other 
ways by writing inappropriate values into other files.

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: [patch 08/12] thermal: add sanity check for the passive attribute

[Frans Pop]

On Tuesday 17 November 2009, Matthew Garrett wrote:
> On Tue, Nov 17, 2009 at 11:42:55PM +0100, Frans Pop wrote:
> > On Tuesday 17 November 2009, Matthew Garrett wrote:
> > > As noted before, I'm not a fan of this approach - I can't think of
> > > many reasons why it'd be necessary to use temperatures below 1
> > > degree C, but this constraint isn't present anywhere else in the
> > > thermal code and, really, there's plenty of things that people can
> > > break if they just echo incorrect values into /sys.
> >
> > Personally I'd be in favor of adding more sanity checks rather than
> > removing them. The cost of the check is virtually zero.
>
> It makes something that's potentially useful to someone impossible, for
> the benefit of people who can crash their system in half a dozen other
> ways by writing inappropriate values into other files.

If you can give me one valid use case for someone running a system with a 
thermal zone check below 1C I'll agree with you.

Cheers,
FJP

Re: [patch 08/12] thermal: add sanity check for the passive attribute

[Frans Pop]

On Wednesday 18 November 2009, Frans Pop wrote:
> On Tuesday 17 November 2009, Matthew Garrett wrote:
> > On Tue, Nov 17, 2009 at 11:42:55PM +0100, Frans Pop wrote:
> > > On Tuesday 17 November 2009, Matthew Garrett wrote:
> > > > As noted before, I'm not a fan of this approach - I can't think of
> > > > many reasons why it'd be necessary to use temperatures below 1
> > > > degree C, but this constraint isn't present anywhere else in the
> > > > thermal code and, really, there's plenty of things that people can
> > > > break if they just echo incorrect values into /sys.
> > >
> > > Personally I'd be in favor of adding more sanity checks rather than
> > > removing them. The cost of the check is virtually zero.
> >
> > It makes something that's potentially useful to someone impossible,
> > for the benefit of people who can crash their system in half a dozen
> > other ways by writing inappropriate values into other files.
>
> If you can give me one valid use case for someone running a system with
> a thermal zone check below 1C I'll agree with you.

And to make it really realistic, it should be a use case which needs a 
value between 0 and 1 C and where a value of 1 doesn't work.