linux/drivers/thermal/x86_pkg_temp_thermal.c

584 lines
15 KiB
C
Raw Normal View History

/*
* x86_pkg_temp_thermal driver
* Copyright (c) 2013, Intel Corporation.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License,
* version 2, as published by the Free Software Foundation.
*
* This program is distributed in the hope it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
* more details.
*
* You should have received a copy of the GNU General Public License along with
* this program; if not, write to the Free Software Foundation, Inc.
*
*/
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/module.h>
#include <linux/init.h>
#include <linux/err.h>
#include <linux/param.h>
#include <linux/device.h>
#include <linux/platform_device.h>
#include <linux/cpu.h>
#include <linux/smp.h>
#include <linux/slab.h>
#include <linux/pm.h>
#include <linux/thermal.h>
#include <linux/debugfs.h>
#include <asm/cpu_device_id.h>
#include <asm/mce.h>
/*
* Rate control delay: Idea is to introduce denounce effect
* This should be long enough to avoid reduce events, when
* threshold is set to a temperature, which is constantly
* violated, but at the short enough to take any action.
* The action can be remove threshold or change it to next
* interesting setting. Based on experiments, in around
* every 5 seconds under load will give us a significant
* temperature change.
*/
#define PKG_TEMP_THERMAL_NOTIFY_DELAY 5000
static int notify_delay_ms = PKG_TEMP_THERMAL_NOTIFY_DELAY;
module_param(notify_delay_ms, int, 0644);
MODULE_PARM_DESC(notify_delay_ms,
"User space notification delay in milli seconds.");
/* Number of trip points in thermal zone. Currently it can't
* be more than 2. MSR can allow setting and getting notifications
* for only 2 thresholds. This define enforces this, if there
* is some wrong values returned by cpuid for number of thresholds.
*/
#define MAX_NUMBER_OF_TRIPS 2
/* Limit number of package temp zones */
#define MAX_PKG_TEMP_ZONE_IDS 256
struct pkg_device {
struct list_head list;
u16 phys_proc_id;
u16 cpu;
u32 tj_max;
u32 msr_pkg_therm_low;
u32 msr_pkg_therm_high;
struct thermal_zone_device *tzone;
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
struct cpumask cpumask;
};
static struct thermal_zone_params pkg_temp_tz_params = {
.no_hwmon = true,
};
/* List maintaining number of package instances */
static LIST_HEAD(phy_dev_list);
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
/* Serializes interrupt notification, work and hotplug */
static DEFINE_SPINLOCK(pkg_temp_lock);
/* Protects zone operation in the work function against hotplug removal */
static DEFINE_MUTEX(thermal_zone_mutex);
/* Interrupt to work function schedule queue */
static DEFINE_PER_CPU(struct delayed_work, pkg_temp_thermal_threshold_work);
/* To track if the work is already scheduled on a package */
static u8 *pkg_work_scheduled;
static u16 max_phy_id;
/* Debug counters to show using debugfs */
static struct dentry *debugfs;
static unsigned int pkg_interrupt_cnt;
static unsigned int pkg_work_cnt;
static int pkg_temp_debugfs_init(void)
{
struct dentry *d;
debugfs = debugfs_create_dir("pkg_temp_thermal", NULL);
if (!debugfs)
return -ENOENT;
d = debugfs_create_u32("pkg_thres_interrupt", S_IRUGO, debugfs,
(u32 *)&pkg_interrupt_cnt);
if (!d)
goto err_out;
d = debugfs_create_u32("pkg_thres_work", S_IRUGO, debugfs,
(u32 *)&pkg_work_cnt);
if (!d)
goto err_out;
return 0;
err_out:
debugfs_remove_recursive(debugfs);
return -ENOENT;
}
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
/*
* Protection:
*
* - cpu hotplug: Read serialized by cpu hotplug lock
* Write must hold pkg_temp_lock
*
* - Other callsites: Must hold pkg_temp_lock
*/
static struct pkg_device *pkg_temp_thermal_get_dev(unsigned int cpu)
{
u16 phys_proc_id = topology_physical_package_id(cpu);
struct pkg_device *pkgdev;
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
list_for_each_entry(pkgdev, &phy_dev_list, list) {
if (pkgdev->phys_proc_id == phys_proc_id)
return pkgdev;
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
}
return NULL;
}
/*
* tj-max is is interesting because threshold is set relative to this
* temperature.
*/
static int get_tj_max(int cpu, u32 *tj_max)
{
u32 eax, edx, val;
int err;
err = rdmsr_safe_on_cpu(cpu, MSR_IA32_TEMPERATURE_TARGET, &eax, &edx);
if (err)
return err;
val = (eax >> 16) & 0xff;
*tj_max = val * 1000;
return val ? 0 : -EINVAL;
}
thermal: consistently use int for temperatures The thermal code uses int, long and unsigned long for temperatures in different places. Using an unsigned type limits the thermal framework to positive temperatures without need. Also several drivers currently will report temperatures near UINT_MAX for temperatures below 0°C. This will probably immediately shut the machine down due to overtemperature if started below 0°C. 'long' is 64bit on several architectures. This is not needed since INT_MAX °mC is above the melting point of all known materials. Consistently use a plain 'int' for temperatures throughout the thermal code and the drivers. This only changes the places in the drivers where the temperature is passed around as pointer, when drivers internally use another type this is not changed. Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de> Acked-by: Geert Uytterhoeven <geert+renesas@glider.be> Reviewed-by: Jean Delvare <jdelvare@suse.de> Reviewed-by: Lukasz Majewski <l.majewski@samsung.com> Reviewed-by: Darren Hart <dvhart@linux.intel.com> Reviewed-by: Heiko Stuebner <heiko@sntech.de> Reviewed-by: Peter Feuerer <peter@piie.net> Cc: Punit Agrawal <punit.agrawal@arm.com> Cc: Zhang Rui <rui.zhang@intel.com> Cc: Eduardo Valentin <edubezval@gmail.com> Cc: linux-pm@vger.kernel.org Cc: linux-kernel@vger.kernel.org Cc: Jean Delvare <jdelvare@suse.de> Cc: Peter Feuerer <peter@piie.net> Cc: Heiko Stuebner <heiko@sntech.de> Cc: Lukasz Majewski <l.majewski@samsung.com> Cc: Stephen Warren <swarren@wwwdotorg.org> Cc: Thierry Reding <thierry.reding@gmail.com> Cc: linux-acpi@vger.kernel.org Cc: platform-driver-x86@vger.kernel.org Cc: linux-arm-kernel@lists.infradead.org Cc: linux-omap@vger.kernel.org Cc: linux-samsung-soc@vger.kernel.org Cc: Guenter Roeck <linux@roeck-us.net> Cc: Rafael J. Wysocki <rjw@rjwysocki.net> Cc: Maxime Ripard <maxime.ripard@free-electrons.com> Cc: Darren Hart <dvhart@infradead.org> Cc: lm-sensors@lm-sensors.org Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2015-07-24 14:12:54 +08:00
static int sys_get_curr_temp(struct thermal_zone_device *tzd, int *temp)
{
struct pkg_device *pkgdev = tzd->devdata;
u32 eax, edx;
rdmsr_on_cpu(pkgdev->cpu, MSR_IA32_PACKAGE_THERM_STATUS, &eax, &edx);
if (eax & 0x80000000) {
*temp = pkgdev->tj_max - ((eax >> 16) & 0x7f) * 1000;
thermal: consistently use int for temperatures The thermal code uses int, long and unsigned long for temperatures in different places. Using an unsigned type limits the thermal framework to positive temperatures without need. Also several drivers currently will report temperatures near UINT_MAX for temperatures below 0°C. This will probably immediately shut the machine down due to overtemperature if started below 0°C. 'long' is 64bit on several architectures. This is not needed since INT_MAX °mC is above the melting point of all known materials. Consistently use a plain 'int' for temperatures throughout the thermal code and the drivers. This only changes the places in the drivers where the temperature is passed around as pointer, when drivers internally use another type this is not changed. Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de> Acked-by: Geert Uytterhoeven <geert+renesas@glider.be> Reviewed-by: Jean Delvare <jdelvare@suse.de> Reviewed-by: Lukasz Majewski <l.majewski@samsung.com> Reviewed-by: Darren Hart <dvhart@linux.intel.com> Reviewed-by: Heiko Stuebner <heiko@sntech.de> Reviewed-by: Peter Feuerer <peter@piie.net> Cc: Punit Agrawal <punit.agrawal@arm.com> Cc: Zhang Rui <rui.zhang@intel.com> Cc: Eduardo Valentin <edubezval@gmail.com> Cc: linux-pm@vger.kernel.org Cc: linux-kernel@vger.kernel.org Cc: Jean Delvare <jdelvare@suse.de> Cc: Peter Feuerer <peter@piie.net> Cc: Heiko Stuebner <heiko@sntech.de> Cc: Lukasz Majewski <l.majewski@samsung.com> Cc: Stephen Warren <swarren@wwwdotorg.org> Cc: Thierry Reding <thierry.reding@gmail.com> Cc: linux-acpi@vger.kernel.org Cc: platform-driver-x86@vger.kernel.org Cc: linux-arm-kernel@lists.infradead.org Cc: linux-omap@vger.kernel.org Cc: linux-samsung-soc@vger.kernel.org Cc: Guenter Roeck <linux@roeck-us.net> Cc: Rafael J. Wysocki <rjw@rjwysocki.net> Cc: Maxime Ripard <maxime.ripard@free-electrons.com> Cc: Darren Hart <dvhart@infradead.org> Cc: lm-sensors@lm-sensors.org Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2015-07-24 14:12:54 +08:00
pr_debug("sys_get_curr_temp %d\n", *temp);
return 0;
}
return -EINVAL;
}
static int sys_get_trip_temp(struct thermal_zone_device *tzd,
int trip, int *temp)
{
struct pkg_device *pkgdev = tzd->devdata;
unsigned long thres_reg_value;
u32 mask, shift, eax, edx;
int ret;
if (trip >= MAX_NUMBER_OF_TRIPS)
return -EINVAL;
if (trip) {
mask = THERM_MASK_THRESHOLD1;
shift = THERM_SHIFT_THRESHOLD1;
} else {
mask = THERM_MASK_THRESHOLD0;
shift = THERM_SHIFT_THRESHOLD0;
}
ret = rdmsr_on_cpu(pkgdev->cpu, MSR_IA32_PACKAGE_THERM_INTERRUPT,
&eax, &edx);
if (ret < 0)
return ret;
thres_reg_value = (eax & mask) >> shift;
if (thres_reg_value)
*temp = pkgdev->tj_max - thres_reg_value * 1000;
else
*temp = 0;
thermal: consistently use int for temperatures The thermal code uses int, long and unsigned long for temperatures in different places. Using an unsigned type limits the thermal framework to positive temperatures without need. Also several drivers currently will report temperatures near UINT_MAX for temperatures below 0°C. This will probably immediately shut the machine down due to overtemperature if started below 0°C. 'long' is 64bit on several architectures. This is not needed since INT_MAX °mC is above the melting point of all known materials. Consistently use a plain 'int' for temperatures throughout the thermal code and the drivers. This only changes the places in the drivers where the temperature is passed around as pointer, when drivers internally use another type this is not changed. Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de> Acked-by: Geert Uytterhoeven <geert+renesas@glider.be> Reviewed-by: Jean Delvare <jdelvare@suse.de> Reviewed-by: Lukasz Majewski <l.majewski@samsung.com> Reviewed-by: Darren Hart <dvhart@linux.intel.com> Reviewed-by: Heiko Stuebner <heiko@sntech.de> Reviewed-by: Peter Feuerer <peter@piie.net> Cc: Punit Agrawal <punit.agrawal@arm.com> Cc: Zhang Rui <rui.zhang@intel.com> Cc: Eduardo Valentin <edubezval@gmail.com> Cc: linux-pm@vger.kernel.org Cc: linux-kernel@vger.kernel.org Cc: Jean Delvare <jdelvare@suse.de> Cc: Peter Feuerer <peter@piie.net> Cc: Heiko Stuebner <heiko@sntech.de> Cc: Lukasz Majewski <l.majewski@samsung.com> Cc: Stephen Warren <swarren@wwwdotorg.org> Cc: Thierry Reding <thierry.reding@gmail.com> Cc: linux-acpi@vger.kernel.org Cc: platform-driver-x86@vger.kernel.org Cc: linux-arm-kernel@lists.infradead.org Cc: linux-omap@vger.kernel.org Cc: linux-samsung-soc@vger.kernel.org Cc: Guenter Roeck <linux@roeck-us.net> Cc: Rafael J. Wysocki <rjw@rjwysocki.net> Cc: Maxime Ripard <maxime.ripard@free-electrons.com> Cc: Darren Hart <dvhart@infradead.org> Cc: lm-sensors@lm-sensors.org Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2015-07-24 14:12:54 +08:00
pr_debug("sys_get_trip_temp %d\n", *temp);
return 0;
}
static int
sys_set_trip_temp(struct thermal_zone_device *tzd, int trip, int temp)
{
struct pkg_device *pkgdev = tzd->devdata;
u32 l, h, mask, shift, intr;
int ret;
if (trip >= MAX_NUMBER_OF_TRIPS || temp >= pkgdev->tj_max)
return -EINVAL;
ret = rdmsr_on_cpu(pkgdev->cpu, MSR_IA32_PACKAGE_THERM_INTERRUPT,
&l, &h);
if (ret < 0)
return ret;
if (trip) {
mask = THERM_MASK_THRESHOLD1;
shift = THERM_SHIFT_THRESHOLD1;
intr = THERM_INT_THRESHOLD1_ENABLE;
} else {
mask = THERM_MASK_THRESHOLD0;
shift = THERM_SHIFT_THRESHOLD0;
intr = THERM_INT_THRESHOLD0_ENABLE;
}
l &= ~mask;
/*
* When users space sets a trip temperature == 0, which is indication
* that, it is no longer interested in receiving notifications.
*/
if (!temp) {
l &= ~intr;
} else {
l |= (pkgdev->tj_max - temp)/1000 << shift;
l |= intr;
}
return wrmsr_on_cpu(pkgdev->cpu, MSR_IA32_PACKAGE_THERM_INTERRUPT, l, h);
}
static int sys_get_trip_type(struct thermal_zone_device *thermal, int trip,
enum thermal_trip_type *type)
{
*type = THERMAL_TRIP_PASSIVE;
return 0;
}
/* Thermal zone callback registry */
static struct thermal_zone_device_ops tzone_ops = {
.get_temp = sys_get_curr_temp,
.get_trip_temp = sys_get_trip_temp,
.get_trip_type = sys_get_trip_type,
.set_trip_temp = sys_set_trip_temp,
};
static bool pkg_thermal_rate_control(void)
{
return true;
}
/* Enable threshold interrupt on local package/cpu */
static inline void enable_pkg_thres_interrupt(void)
{
u8 thres_0, thres_1;
u32 l, h;
rdmsr(MSR_IA32_PACKAGE_THERM_INTERRUPT, l, h);
/* only enable/disable if it had valid threshold value */
thres_0 = (l & THERM_MASK_THRESHOLD0) >> THERM_SHIFT_THRESHOLD0;
thres_1 = (l & THERM_MASK_THRESHOLD1) >> THERM_SHIFT_THRESHOLD1;
if (thres_0)
l |= THERM_INT_THRESHOLD0_ENABLE;
if (thres_1)
l |= THERM_INT_THRESHOLD1_ENABLE;
wrmsr(MSR_IA32_PACKAGE_THERM_INTERRUPT, l, h);
}
/* Disable threshold interrupt on local package/cpu */
static inline void disable_pkg_thres_interrupt(void)
{
u32 l, h;
rdmsr(MSR_IA32_PACKAGE_THERM_INTERRUPT, l, h);
l &= ~(THERM_INT_THRESHOLD0_ENABLE | THERM_INT_THRESHOLD1_ENABLE);
wrmsr(MSR_IA32_PACKAGE_THERM_INTERRUPT, l, h);
}
static void pkg_temp_thermal_threshold_work_fn(struct work_struct *work)
{
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
struct thermal_zone_device *tzone = NULL;
int phy_id, cpu = smp_processor_id();
struct pkg_device *pkgdev;
u64 msr_val, wr_val;
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
mutex_lock(&thermal_zone_mutex);
spin_lock_irq(&pkg_temp_lock);
++pkg_work_cnt;
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
pkgdev = pkg_temp_thermal_get_dev(cpu);
if (!pkgdev) {
spin_unlock_irq(&pkg_temp_lock);
mutex_unlock(&thermal_zone_mutex);
return;
}
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
pkg_work_scheduled[phy_id] = 0;
rdmsrl(MSR_IA32_PACKAGE_THERM_STATUS, msr_val);
wr_val = msr_val & ~(THERM_LOG_THRESHOLD0 | THERM_LOG_THRESHOLD1);
if (wr_val != msr_val) {
wrmsrl(MSR_IA32_PACKAGE_THERM_STATUS, wr_val);
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
tzone = pkgdev->tzone;
}
enable_pkg_thres_interrupt();
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
spin_unlock_irq(&pkg_temp_lock);
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
/*
* If tzone is not NULL, then thermal_zone_mutex will prevent the
* concurrent removal in the cpu offline callback.
*/
if (tzone)
thermal_zone_device_update(tzone, THERMAL_EVENT_UNSPECIFIED);
mutex_unlock(&thermal_zone_mutex);
}
static int pkg_thermal_notify(u64 msr_val)
{
int cpu = smp_processor_id();
int phy_id = topology_physical_package_id(cpu);
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
struct pkg_device *pkgdev;
unsigned long flags;
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
spin_lock_irqsave(&pkg_temp_lock, flags);
++pkg_interrupt_cnt;
disable_pkg_thres_interrupt();
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
/* Work is per package, so scheduling it once is enough. */
pkgdev = pkg_temp_thermal_get_dev(cpu);
if (pkgdev && pkg_work_scheduled && !pkg_work_scheduled[phy_id]) {
pkg_work_scheduled[phy_id] = 1;
schedule_delayed_work_on(cpu,
&per_cpu(pkg_temp_thermal_threshold_work, cpu),
msecs_to_jiffies(notify_delay_ms));
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
}
spin_unlock_irqrestore(&pkg_temp_lock, flags);
return 0;
}
static int pkg_temp_thermal_device_add(unsigned int cpu)
{
u32 tj_max, eax, ebx, ecx, edx;
struct pkg_device *pkgdev;
int thres_count, err;
unsigned long flags;
u8 *temp;
cpuid(6, &eax, &ebx, &ecx, &edx);
thres_count = ebx & 0x07;
if (!thres_count)
return -ENODEV;
if (topology_physical_package_id(cpu) > MAX_PKG_TEMP_ZONE_IDS)
return -ENODEV;
thres_count = clamp_val(thres_count, 0, MAX_NUMBER_OF_TRIPS);
err = get_tj_max(cpu, &tj_max);
if (err)
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
return err;
pkgdev = kzalloc(sizeof(*pkgdev), GFP_KERNEL);
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
if (!pkgdev)
return -ENOMEM;
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
spin_lock_irqsave(&pkg_temp_lock, flags);
if (topology_physical_package_id(cpu) > max_phy_id)
max_phy_id = topology_physical_package_id(cpu);
temp = krealloc(pkg_work_scheduled,
(max_phy_id+1) * sizeof(u8), GFP_ATOMIC);
if (!temp) {
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
spin_unlock_irqrestore(&pkg_temp_lock, flags);
kfree(pkgdev);
return -ENOMEM;
}
pkg_work_scheduled = temp;
pkg_work_scheduled[topology_physical_package_id(cpu)] = 0;
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
spin_unlock_irqrestore(&pkg_temp_lock, flags);
pkgdev->phys_proc_id = topology_physical_package_id(cpu);
pkgdev->cpu = cpu;
pkgdev->tj_max = tj_max;
pkgdev->tzone = thermal_zone_device_register("x86_pkg_temp",
thres_count,
(thres_count == MAX_NUMBER_OF_TRIPS) ? 0x03 : 0x01,
pkgdev, &tzone_ops, &pkg_temp_tz_params, 0, 0);
if (IS_ERR(pkgdev->tzone)) {
err = PTR_ERR(pkgdev->tzone);
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
kfree(pkgdev);
return err;
}
/* Store MSR value for package thermal interrupt, to restore at exit */
rdmsr_on_cpu(cpu, MSR_IA32_PACKAGE_THERM_INTERRUPT,
&pkgdev->msr_pkg_therm_low,
&pkgdev->msr_pkg_therm_high);
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
cpumask_set_cpu(cpu, &pkgdev->cpumask);
spin_lock_irq(&pkg_temp_lock);
list_add_tail(&pkgdev->list, &phy_dev_list);
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
spin_unlock_irq(&pkg_temp_lock);
return 0;
}
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
static void put_core_offline(unsigned int cpu)
{
struct pkg_device *pkgdev = pkg_temp_thermal_get_dev(cpu);
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
bool lastcpu;
int target;
if (!pkgdev)
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
return;
target = cpumask_any_but(&pkgdev->cpumask, cpu);
cpumask_clear_cpu(cpu, &pkgdev->cpumask);
lastcpu = target >= nr_cpu_ids;
/*
* Remove the sysfs files, if this is the last cpu in the package
* before doing further cleanups.
*/
if (lastcpu) {
struct thermal_zone_device *tzone = pkgdev->tzone;
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
/*
* We must protect against a work function calling
* thermal_zone_update, after/while unregister. We null out
* the pointer under the zone mutex, so the worker function
* won't try to call.
*/
mutex_lock(&thermal_zone_mutex);
pkgdev->tzone = NULL;
mutex_unlock(&thermal_zone_mutex);
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
thermal_zone_device_unregister(tzone);
}
/*
* If this is the last CPU in the package, restore the interrupt
* MSR and remove the package reference from the array.
*/
if (lastcpu) {
/* Protect against work and interrupts */
spin_lock_irq(&pkg_temp_lock);
list_del(&pkgdev->list);
/*
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
* After this point nothing touches the MSR anymore. We
* must drop the lock to make the cross cpu call. This goes
* away once we move that code to the hotplug state machine.
*/
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
spin_unlock_irq(&pkg_temp_lock);
wrmsr_on_cpu(cpu, MSR_IA32_PACKAGE_THERM_INTERRUPT,
pkgdev->msr_pkg_therm_low,
pkgdev->msr_pkg_therm_high);
kfree(pkgdev);
}
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
/*
* Note, this is broken when work was really scheduled on the
* outgoing cpu because this will leave the work_scheduled flag set
* and the thermal interrupts disabled. Will be fixed in the next
* step as there is no way to fix it in a sane way with the per cpu
* work nonsense.
*/
cancel_delayed_work_sync(&per_cpu(pkg_temp_thermal_threshold_work, cpu));
}
static int get_core_online(unsigned int cpu)
{
struct pkg_device *pkgdev = pkg_temp_thermal_get_dev(cpu);
struct cpuinfo_x86 *c = &cpu_data(cpu);
/* Paranoia check */
if (!cpu_has(c, X86_FEATURE_DTHERM) || !cpu_has(c, X86_FEATURE_PTS))
return -ENODEV;
INIT_DELAYED_WORK(&per_cpu(pkg_temp_thermal_threshold_work, cpu),
pkg_temp_thermal_threshold_work_fn);
/* If the package exists, nothing to do */
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
if (pkgdev) {
cpumask_set_cpu(cpu, &pkgdev->cpumask);
return 0;
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
}
return pkg_temp_thermal_device_add(cpu);
}
static int pkg_temp_thermal_cpu_callback(struct notifier_block *nfb,
unsigned long action, void *hcpu)
{
unsigned int cpu = (unsigned long) hcpu;
switch (action & ~CPU_TASKS_FROZEN) {
case CPU_ONLINE:
case CPU_DOWN_FAILED:
get_core_online(cpu);
break;
case CPU_DOWN_PREPARE:
put_core_offline(cpu);
break;
}
return NOTIFY_OK;
}
static struct notifier_block pkg_temp_thermal_notifier __refdata = {
.notifier_call = pkg_temp_thermal_cpu_callback,
};
static const struct x86_cpu_id __initconst pkg_temp_thermal_ids[] = {
{ X86_VENDOR_INTEL, X86_FAMILY_ANY, X86_MODEL_ANY, X86_FEATURE_PTS },
{}
};
MODULE_DEVICE_TABLE(x86cpu, pkg_temp_thermal_ids);
static int __init pkg_temp_thermal_init(void)
{
int i;
if (!x86_match_cpu(pkg_temp_thermal_ids))
return -ENODEV;
cpu_notifier_register_begin();
for_each_online_cpu(i)
if (get_core_online(i))
goto err_ret;
__register_hotcpu_notifier(&pkg_temp_thermal_notifier);
cpu_notifier_register_done();
platform_thermal_package_notify = pkg_thermal_notify;
platform_thermal_package_rate_control = pkg_thermal_rate_control;
/* Don't care if it fails */
pkg_temp_debugfs_init();
return 0;
err_ret:
for_each_online_cpu(i)
put_core_offline(i);
cpu_notifier_register_done();
kfree(pkg_work_scheduled);
return -ENODEV;
}
module_init(pkg_temp_thermal_init)
static void __exit pkg_temp_thermal_exit(void)
{
int i;
platform_thermal_package_notify = NULL;
platform_thermal_package_rate_control = NULL;
cpu_notifier_register_begin();
__unregister_hotcpu_notifier(&pkg_temp_thermal_notifier);
for_each_online_cpu(i)
thermal/x86_pkg_temp: Sanitize locking The work cancellation code, the thermal zone unregistering, the work code and the interrupt notification function are racy against each other and against cpu hotplug and module exit. The random locking sprinkeled all over the place does not help anything and probably exists to make people feel good. The resulting issues (mainly use after free) are probably hard to trigger, but they clearly exist Protect the package list with a spinlock so it can be accessed from the interrupt notifier and also from the work function. The add/removal code in the hotplug callbacks take the lock for list manipulation. That makes sure that on removal neither the interrupt notifier nor the work function can access the about to be freed package structure anymore. The thermal zone unregistering is another trainwreck. It's not serialized against the work function. So unregistering the zone device can race with the work function and cause havoc. Protect the thermal zone with a mutex, which is held in the work function to make sure that the zone device is not being unregistered concurrently. To solve the module exit issues, we simply invoke the cpu offline callback and let it work its magic. For that it's required to keep track of the participating cpus in a package, because topology_core_mask is not affected by calling the offline callback for teardown of the driver, so it would never free the package as there is always a valid target in topology_core_mask. Use proper names for the locks so it's clear what they are for and add a pile of comments to explain the protection rules. It's amazing that fixing the locking and adding 30 lines of comments explaining it still removes more lines than it adds. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
2016-11-23 01:57:10 +08:00
put_core_offline(i);
cpu_notifier_register_done();
kfree(pkg_work_scheduled);
debugfs_remove_recursive(debugfs);
}
module_exit(pkg_temp_thermal_exit)
MODULE_DESCRIPTION("X86 PKG TEMP Thermal Driver");
MODULE_AUTHOR("Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>");
MODULE_LICENSE("GPL v2");