Len,
yes it's true that it's polling-based which may be malfunction in some
cases.
but this mechanism works only in a system overheating/critical stage.
if the polling-base passive trip point doesn't work, the system goes to
critical shutdown just like it does without this patch.
But once it works, it can help a lot.
As it won't bring any side effect, I think this is a good enhancement
of the current thermal management.
Any comments on this?
thanks,
rui
On Thu, 2008-12-04 at 02:00 +0800, Matthew Garrett wrote:
> Due to poor thermal design or Linux driving hardware outside its thermal
> envelope, some systems will reach critical temperature and shut down
> under high load. This patch adds support for forcing a polling-based
> passive trip point if the firmware doesn't provide one. The assumption
> is made that the processor is the most practical means to reduce the
> dynamic heat generation, so hitting the passive thermal limit will cause
> the CPU to be throttled until the temperature stabalises around the
> defined value.
>
> UI is provided via a "passive" sysfs entry in the thermal zone
> directory. It accepts a decimal value in millidegrees celsius, or "0" to
> disable the functionality. Default behaviour is for this functionality
> to be disabled.
>
> Signed-off-by: Matthew Garrett <mjg@xxxxxxxxxx>
> ---
>
> Depends on the code to move the trip handling into the generic layer.
> Tested this on a couple of machines and it seems to work well - a
> logical followup might be to allow the passive trip point to be
> overridden on machines that provide it.
>
> drivers/thermal/thermal_sys.c | 77 +++++++++++++++++++++++++++++++++++++++++
> include/linux/thermal.h | 1 +
> 2 files changed, 78 insertions(+), 0 deletions(-)
>
> diff --git a/drivers/thermal/thermal_sys.c b/drivers/thermal/thermal_sys.c
> index ac0f91f..c5351e5 100644
> --- a/drivers/thermal/thermal_sys.c
> +++ b/drivers/thermal/thermal_sys.c
> @@ -214,9 +214,69 @@ trip_point_temp_show(struct device *dev, struct device_attribute *attr,
> return sprintf(buf, "%ld\n", temperature);
> }
>
> +static ssize_t
> +passive_store(struct device *dev, struct device_attribute *attr,
> + const char *buf, size_t count)
> +{
> + struct thermal_zone_device *tz = to_thermal_zone(dev);
> + struct thermal_cooling_device *cdev = NULL;
> + int state;
> +
> + if (!sscanf(buf, "%d\n", &state))
> + return -EINVAL;
> +
> + if (state && !tz->forced_passive) {
> + mutex_lock(&thermal_list_lock);
> + list_for_each_entry(cdev, &thermal_cdev_list, node) {
> + if (!strncmp("Processor", cdev->type,
> + sizeof("Processor")))
> + thermal_zone_bind_cooling_device(tz,
> + THERMAL_TRIPS_NONE,
> + cdev);
> + }
> + mutex_unlock(&thermal_list_lock);
> + } else if (!state && tz->forced_passive) {
> + mutex_lock(&thermal_list_lock);
> + list_for_each_entry(cdev, &thermal_cdev_list, node) {
> + if (!strncmp("Processor", cdev->type,
> + sizeof("Processor")))
> + thermal_zone_unbind_cooling_device(tz,
> + THERMAL_TRIPS_NONE,
> + cdev);
> + }
> + mutex_unlock(&thermal_list_lock);
> + }
> +
> + tz->tc1 = 1;
> + tz->tc2 = 1;
> +
> + if (!tz->passive_delay)
> + tz->passive_delay = 1000;
> +
> + if (!tz->polling_delay)
> + tz->polling_delay = 10000;
> +
> + tz->forced_passive = state;
> +
> + thermal_zone_device_update(tz);
> +
> + return count;
> +}
> +
> +static ssize_t
> +passive_show(struct device *dev, struct device_attribute *attr,
> + char *buf)
> +{
> + struct thermal_zone_device *tz = to_thermal_zone(dev);
> +
> + return sprintf(buf, "%d\n", tz->forced_passive);
> +}
> +
> static DEVICE_ATTR(type, 0444, type_show, NULL);
> static DEVICE_ATTR(temp, 0444, temp_show, NULL);
> static DEVICE_ATTR(mode, 0644, mode_show, mode_store);
> +static DEVICE_ATTR(passive, S_IRUGO | S_IWUSR, passive_show, \
> + passive_store);
>
> static struct device_attribute trip_point_attrs[] = {
> __ATTR(trip_point_0_type, 0444, trip_point_type_show, NULL),
> @@ -939,6 +999,11 @@ void thermal_zone_device_update(struct thermal_zone_device *tz)
> break;
> }
> }
> +
> + if (tz->forced_passive)
> + thermal_zone_device_passive(tz, temp, tz->forced_passive,
> + THERMAL_TRIPS_NONE);
> +
> tz->last_temperature = temp;
> if (tz->passive)
> thermal_zone_device_set_polling(tz, tz->passive_delay);
> @@ -977,8 +1042,10 @@ struct thermal_zone_device *thermal_zone_device_register(char *type,
> {
> struct thermal_zone_device *tz;
> struct thermal_cooling_device *pos;
> + enum thermal_trip_type trip_type;
> int result;
> int count;
> + int passive=0;
>
> if (strlen(type) >= THERMAL_NAME_LENGTH)
> return ERR_PTR(-EINVAL);
> @@ -1041,8 +1108,18 @@ struct thermal_zone_device *thermal_zone_device_register(char *type,
> TRIP_POINT_ATTR_ADD(&tz->device, count, result);
> if (result)
> goto unregister;
> + tz->ops->get_trip_type(tz, count, &trip_type);
> + if (trip_type == THERMAL_TRIP_PASSIVE)
> + passive=1;
> }
>
> + if (!passive)
> + result = device_create_file(&tz->device,
> + &dev_attr_passive);
> +
> + if (result)
> + goto unregister;
> +
> result = thermal_add_hwmon_sysfs(tz);
> if (result)
> goto unregister;
> diff --git a/include/linux/thermal.h b/include/linux/thermal.h
> index a81c615..1de8b9e 100644
> --- a/include/linux/thermal.h
> +++ b/include/linux/thermal.h
> @@ -113,6 +113,7 @@ struct thermal_zone_device {
> int polling_delay;
> int last_temperature;
> bool passive;
> + unsigned int forced_passive;
> struct thermal_zone_device_ops *ops;
> struct list_head cooling_devices;
> struct idr idr;
>
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
