Merge tag 'pm-4.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management updates from Rafael Wysocki:
 "The majority of changes go into the Operating Performance Points (OPP)
  framework and cpufreq this time, followed by devfreq and some
  scattered updates all over.

  The OPP changes are mostly related to switching over from RCU-based
  synchronization, that turned out to be overly complicated and
  problematic, to reference counting using krefs.

  In the cpufreq land there are core cleanups, documentation updates, a
  new driver for Broadcom BMIPS SoCs, a new cpufreq-dt sub-driver for TI
  SoCs that require special handling, ARM64 SoCs support for the qoriq
  driver, intel_pstate updates, powernv driver update and assorted
  fixes.

  The devfreq changes are mostly fixes related to the sysfs interface
  and some Exynos drivers updates.

  Apart from that, the cpuidle menu governor will support per-CPU PM QoS
  constraints for the wakeup latency now, some bugs in the wakeup IRQs
  framework are fixed, the generic power domains framework should handle
  asynchronous invocations of *noirq suspend/resume callbacks from now
  on, the analyze_suspend.py script is updated and there is a new tool
  for intel_pstate diagnostics.

  Specifics:

   - Operating Performance Points (OPP) framework fixes, cleanups and
     switch over from RCU-based synchronization to reference counting
     using krefs (Viresh Kumar, Wei Yongjun, Dave Gerlach)

   - cpufreq core cleanups and documentation updates (Viresh Kumar,
     Rafael Wysocki)

   - New cpufreq driver for Broadcom BMIPS SoCs (Markus Mayer)

   - New cpufreq-dt sub-driver for TI SoCs requiring special handling,
     like in the AM335x, AM437x, DRA7x, and AM57x families, along with
     new DT bindings for it (Dave Gerlach, Paul Gortmaker)

   - ARM64 SoCs support for the qoriq cpufreq driver (Tang Yuantian)

   - intel_pstate driver updates including a new sysfs knob to control
     the driver's operation mode and fixes related to the no_turbo sysfs
     knob and the hardware-managed P-states feature support (Rafael
     Wysocki, Srinivas Pandruvada)

   - New interface to export ultra-turbo frequencies for the powernv
     cpufreq driver (Shilpasri Bhat)

   - Assorted fixes for cpufreq drivers (Arnd Bergmann, Dan Carpenter,
     Wei Yongjun)

   - devfreq core fixes, mostly related to the sysfs interface exported
     by it (Chanwoo Choi, Chris Diamand)

   - Updates of the exynos-bus and exynos-ppmu devfreq drivers (Chanwoo
     Choi)

   - Device PM QoS extension to support CPUs and support for per-CPU
     wakeup (device resume) latency constraints in the cpuidle menu
     governor (Alex Shi)

   - Wakeup IRQs framework fixes (Grygorii Strashko)

   - Generic power domains framework update including a fix to make it
     handle asynchronous invocations of *noirq suspend/resume callbacks
     correctly (Ulf Hansson, Geert Uytterhoeven)

   - Assorted fixes and cleanups in the core suspend/hibernate code, PM
     QoS framework and x86 ACPI idle support code (Corentin Labbe, Geert
     Uytterhoeven, Geliang Tang, John Keeping, Nick Desaulniers)

   - Update of the analyze_suspend.py script is updated to version 4.5
     offering multiple improvements (Todd Brandt)

   - New tool for intel_pstate diagnostics using the pstate_sample
     tracepoint (Doug Smythies)"

* tag 'pm-4.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (85 commits)
  MAINTAINERS: cpufreq: add bmips-cpufreq.c
  PM / QoS: Fix memory leak on resume_latency.notifiers
  PM / Documentation: Spelling s/wrtie/write/
  PM / sleep: Fix test_suspend after sleep state rework
  cpufreq: CPPC: add ACPI_PROCESSOR dependency
  cpufreq: make ti-cpufreq explicitly non-modular
  cpufreq: Do not clear real_cpus mask on policy init
  tools/power/x86: Debug utility for intel_pstate driver
  AnalyzeSuspend: fix drag and zoom bug in javascript
  PM / wakeirq: report a wakeup_event on dedicated wekup irq
  PM / wakeirq: Fix spurious wake-up events for dedicated wakeirqs
  PM / wakeirq: Enable dedicated wakeirq for suspend
  cpufreq: dt: Don't use generic platdev driver for ti-cpufreq platforms
  cpufreq: ti: Add cpufreq driver to determine available OPPs at runtime
  Documentation: dt: add bindings for ti-cpufreq
  PM / OPP: Expose _of_get_opp_desc_node as dev_pm_opp API
  cpufreq: qoriq: Don't look at clock implementation details
  cpufreq: qoriq: add ARM64 SoCs support
  PM / Domains: Provide dummy governors if CONFIG_PM_GENERIC_DOMAINS=n
  cpufreq: brcmstb-avs-cpufreq: remove unnecessary platform_set_drvdata()
  ...

Documentation/ABI/testing/sysfs-class-devfreq-event

+What:		/sys/class/devfreq-event/event(x)/
+Date:		January 2017
+Contact:	Chanwoo Choi <cw00.choi@samsung.com>
+Description:
+		Provide a place in sysfs for the devfreq-event objects.
+		This allows accessing various devfreq-event specific variables.
+		The name of devfreq-event object denoted as 'event(x)' which
+		includes the unique number of 'x' for each devfreq-event object.
+
+What:		/sys/class/devfreq-event/event(x)/name
+Date:		January 2017
+Contact:	Chanwoo Choi <cw00.choi@samsung.com>
+Description:
+		The /sys/class/devfreq-event/event(x)/name attribute contains
+		the name of the devfreq-event object. This attribute is
+		read-only.
+
+What:		/sys/class/devfreq-event/event(x)/enable_count
+Date:		January 2017
+Contact:	Chanwoo Choi <cw00.choi@samsung.com>
+Description:
+		The /sys/class/devfreq-event/event(x)/enable_count attribute
+		contains the reference count to enable the devfreq-event
+		object. If the device is enabled, the value of attribute is
+		greater than zero.

Documentation/cpu-freq/core.txt

@@ -8,6 +8,8 @@
 
 		    Dominik Brodowski  <linux@brodo.de>
 		     David Kimdon <dwhedon@debian.org>
+		Rafael J. Wysocki <rafael.j.wysocki@intel.com>
+		   Viresh Kumar <viresh.kumar@linaro.org>
 
 
 
@@ -36,10 +38,11 @@ speed limits (like LCD drivers on ARM architecture). Additionally, the
 kernel "constant" loops_per_jiffy is updated on frequency changes
 here.
 
-Reference counting is done by cpufreq_get_cpu and cpufreq_put_cpu,
-which make sure that the cpufreq processor driver is correctly
-registered with the core, and will not be unloaded until
-cpufreq_put_cpu is called.
+Reference counting of the cpufreq policies is done by cpufreq_cpu_get
+and cpufreq_cpu_put, which make sure that the cpufreq driver is
+correctly registered with the core, and will not be unloaded until
+cpufreq_put_cpu is called. That also ensures that the respective cpufreq
+policy doesn't get freed while being used.
 
 2. CPUFreq notifiers
 ====================
@@ -69,18 +72,16 @@ CPUFreq policy notifier is called twice for a policy transition:
 The phase is specified in the second argument to the notifier.
 
 The third argument, a void *pointer, points to a struct cpufreq_policy
-consisting of five values: cpu, min, max, policy and max_cpu_freq. min 
-and max are the lower and upper frequencies (in kHz) of the new
-policy, policy the new policy, cpu the number of the affected CPU; and 
-max_cpu_freq the maximum supported CPU frequency. This value is given 
-for informational purposes only.
+consisting of several values, including min, max (the lower and upper
+frequencies (in kHz) of the new policy).
 
 
 2.2 CPUFreq transition notifiers
 --------------------------------
 
-These are notified twice when the CPUfreq driver switches the CPU core
-frequency and this change has any external implications.
+These are notified twice for each online CPU in the policy, when the
+CPUfreq driver switches the CPU core frequency and this change has no
+any external implications.
 
 The second argument specifies the phase - CPUFREQ_PRECHANGE or
 CPUFREQ_POSTCHANGE.
@@ -90,6 +91,7 @@ values:
 cpu	- number of the affected CPU
 old	- old frequency
 new	- new frequency
+flags	- flags of the cpufreq driver
 
 3. CPUFreq Table Generation with Operating Performance Point (OPP)
 ==================================================================

Documentation/cpu-freq/cpu-drivers.txt

@@ -9,6 +9,8 @@
 
 
 		    Dominik Brodowski  <linux@brodo.de>
+		Rafael J. Wysocki <rafael.j.wysocki@intel.com>
+		   Viresh Kumar <viresh.kumar@linaro.org>
 
 
 
@@ -49,49 +51,65 @@ using cpufreq_register_driver()
 
 What shall this struct cpufreq_driver contain? 
 
-cpufreq_driver.name -		The name of this driver.
+ .name - The name of this driver.
 
-cpufreq_driver.init -		A pointer to the per-CPU initialization 
-				function.
+ .init - A pointer to the per-policy initialization function.
 
-cpufreq_driver.verify -		A pointer to a "verification" function.
+ .verify - A pointer to a "verification" function.
 
-cpufreq_driver.setpolicy _or_ 
-cpufreq_driver.target/
-target_index		-	See below on the differences.
+ .setpolicy _or_ .fast_switch _or_ .target _or_ .target_index - See
+ below on the differences.
 
 And optionally
 
-cpufreq_driver.exit -		A pointer to a per-CPU cleanup
-				function called during CPU_POST_DEAD
-				phase of cpu hotplug process.
+ .flags - Hints for the cpufreq core.
 
-cpufreq_driver.stop_cpu -	A pointer to a per-CPU stop function
-				called during CPU_DOWN_PREPARE phase of
-				cpu hotplug process.
+ .driver_data - cpufreq driver specific data.
 
-cpufreq_driver.resume -		A pointer to a per-CPU resume function
-				which is called with interrupts disabled
-				and _before_ the pre-suspend frequency
-				and/or policy is restored by a call to
-				->target/target_index or ->setpolicy.
+ .resolve_freq - Returns the most appropriate frequency for a target
+ frequency. Doesn't change the frequency though.
 
-cpufreq_driver.attr -		A pointer to a NULL-terminated list of
-				"struct freq_attr" which allow to
-				export values to sysfs.
+ .get_intermediate and target_intermediate - Used to switch to stable
+ frequency while changing CPU frequency.
 
-cpufreq_driver.get_intermediate
-and target_intermediate		Used to switch to stable frequency while
-				changing CPU frequency.
+ .get - Returns current frequency of the CPU.
+
+ .bios_limit - Returns HW/BIOS max frequency limitations for the CPU.
+
+ .exit - A pointer to a per-policy cleanup function called during
+ CPU_POST_DEAD phase of cpu hotplug process.
+
+ .stop_cpu - A pointer to a per-policy stop function called during
+ CPU_DOWN_PREPARE phase of cpu hotplug process.
+
+ .suspend - A pointer to a per-policy suspend function which is called
+ with interrupts disabled and _after_ the governor is stopped for the
+ policy.
+
+ .resume - A pointer to a per-policy resume function which is called
+ with interrupts disabled and _before_ the governor is started again.
+
+ .ready - A pointer to a per-policy ready function which is called after
+ the policy is fully initialized.
+
+ .attr - A pointer to a NULL-terminated list of "struct freq_attr" which
+ allow to export values to sysfs.
+
+ .boost_enabled - If set, boost frequencies are enabled.
+
+ .set_boost - A pointer to a per-policy function to enable/disable boost
+ frequencies.
 
 
 1.2 Per-CPU Initialization
 --------------------------
 
 Whenever a new CPU is registered with the device model, or after the
-cpufreq driver registers itself, the per-CPU initialization function 
-cpufreq_driver.init is called. It takes a struct cpufreq_policy
-*policy as argument. What to do now?
+cpufreq driver registers itself, the per-policy initialization function
+cpufreq_driver.init is called if no cpufreq policy existed for the CPU.
+Note that the .init() and .exit() routines are called only once for the
+policy and not for each CPU managed by the policy. It takes a struct
+cpufreq_policy *policy as argument. What to do now?
 
 If necessary, activate the CPUfreq support on your CPU.
 
@@ -117,47 +135,45 @@ policy->governor		must contain the "default policy" for
 				cpufreq_driver.setpolicy or
 				cpufreq_driver.target/target_index is called
 				with these values.
+policy->cpus			Update this with the masks of the
+				(online + offline) CPUs that do DVFS
+				along with this CPU (i.e.  that share
+				clock/voltage rails with it).
 
 For setting some of these values (cpuinfo.min[max]_freq, policy->min[max]), the
 frequency table helpers might be helpful. See the section 2 for more information
 on them.
 
-SMP systems normally have same clock source for a group of cpus. For these the
-.init() would be called only once for the first online cpu. Here the .init()
-routine must initialize policy->cpus with mask of all possible cpus (Online +
-Offline) that share the clock. Then the core would copy this mask onto
-policy->related_cpus and will reset policy->cpus to carry only online cpus.
-
 
 1.3 verify
-------------
+----------
 
 When the user decides a new policy (consisting of
 "policy,governor,min,max") shall be set, this policy must be validated
 so that incompatible values can be corrected. For verifying these
-values, a frequency table helper and/or the
-cpufreq_verify_within_limits(struct cpufreq_policy *policy, unsigned
-int min_freq, unsigned int max_freq) function might be helpful. See
-section 2 for details on frequency table helpers.
+values cpufreq_verify_within_limits(struct cpufreq_policy *policy,
+unsigned int min_freq, unsigned int max_freq) function might be helpful.
+See section 2 for details on frequency table helpers.
 
 You need to make sure that at least one valid frequency (or operating
 range) is within policy->min and policy->max. If necessary, increase
 policy->max first, and only if this is no solution, decrease policy->min.
 
 
-1.4 target/target_index or setpolicy?
-----------------------------
+1.4 target or target_index or setpolicy or fast_switch?
+-------------------------------------------------------
 
 Most cpufreq drivers or even most cpu frequency scaling algorithms 
-only allow the CPU to be set to one frequency. For these, you use the
-->target/target_index call.
+only allow the CPU frequency to be set to predefined fixed values. For
+these, you use the ->target(), ->target_index() or ->fast_switch()
+callbacks.
 
-Some cpufreq-capable processors switch the frequency between certain
-limits on their own. These shall use the ->setpolicy call
+Some cpufreq capable processors switch the frequency between certain
+limits on their own. These shall use the ->setpolicy() callback.
 
 
 1.5. target/target_index
--------------
+------------------------
 
 The target_index call has two arguments: struct cpufreq_policy *policy,
 and unsigned int index (into the exposed frequency table).
@@ -186,9 +202,20 @@ actual frequency must be determined using the following rules:
 Here again the frequency table helper might assist you - see section 2
 for details.
 
+1.6. fast_switch
+----------------
 
-1.6 setpolicy
----------------
+This function is used for frequency switching from scheduler's context.
+Not all drivers are expected to implement it, as sleeping from within
+this callback isn't allowed. This callback must be highly optimized to
+do switching as fast as possible.
+
+This function has two arguments: struct cpufreq_policy *policy and
+unsigned int target_frequency.
+
+
+1.7 setpolicy
+-------------
 
 The setpolicy call only takes a struct cpufreq_policy *policy as
 argument. You need to set the lower limit of the in-processor or
@@ -198,7 +225,7 @@ setting when policy->policy is CPUFREQ_POLICY_PERFORMANCE, and a
 powersaving-oriented setting when CPUFREQ_POLICY_POWERSAVE. Also check
 the reference implementation in drivers/cpufreq/longrun.c
 
-1.7 get_intermediate and target_intermediate
+1.8 get_intermediate and target_intermediate
 --------------------------------------------
 
 Only for drivers with target_index() and CPUFREQ_ASYNC_NOTIFICATION unset.
@@ -222,42 +249,36 @@ failures as core would send notifications for that.
 
 As most cpufreq processors only allow for being set to a few specific
 frequencies, a "frequency table" with some functions might assist in
-some work of the processor driver. Such a "frequency table" consists
-of an array of struct cpufreq_frequency_table entries, with any value in
-"driver_data" you want to use, and the corresponding frequency in
-"frequency". At the end of the table, you need to add a
-cpufreq_frequency_table entry with frequency set to CPUFREQ_TABLE_END. And
-if you want to skip one entry in the table, set the frequency to 
-CPUFREQ_ENTRY_INVALID. The entries don't need to be in ascending
-order.
-
-By calling cpufreq_table_validate_and_show(struct cpufreq_policy *policy,
-					struct cpufreq_frequency_table *table);
-the cpuinfo.min_freq and cpuinfo.max_freq values are detected, and
-policy->min and policy->max are set to the same values. This is
-helpful for the per-CPU initialization stage.
-
-int cpufreq_frequency_table_verify(struct cpufreq_policy *policy,
-                                   struct cpufreq_frequency_table *table);
-assures that at least one valid frequency is within policy->min and
-policy->max, and all other criteria are met. This is helpful for the
-->verify call.
-
-int cpufreq_frequency_table_target(struct cpufreq_policy *policy,
-                                   unsigned int target_freq,
-                                   unsigned int relation);
-
-is the corresponding frequency table helper for the ->target
-stage. Just pass the values to this function, and this function
-returns the number of the frequency table entry which contains
-the frequency the CPU shall be set to.
+some work of the processor driver. Such a "frequency table" consists of
+an array of struct cpufreq_frequency_table entries, with driver specific
+values in "driver_data", the corresponding frequency in "frequency" and
+flags set. At the end of the table, you need to add a
+cpufreq_frequency_table entry with frequency set to CPUFREQ_TABLE_END.
+And if you want to skip one entry in the table, set the frequency to
+CPUFREQ_ENTRY_INVALID. The entries don't need to be in sorted in any
+particular order, but if they are cpufreq core will do DVFS a bit
+quickly for them as search for best match is faster.
+
+By calling cpufreq_table_validate_and_show(), the cpuinfo.min_freq and
+cpuinfo.max_freq values are detected, and policy->min and policy->max
+are set to the same values. This is helpful for the per-CPU
+initialization stage.
+
+cpufreq_frequency_table_verify() assures that at least one valid
+frequency is within policy->min and policy->max, and all other criteria
+are met. This is helpful for the ->verify call.
+
+cpufreq_frequency_table_target() is the corresponding frequency table
+helper for the ->target stage. Just pass the values to this function,
+and this function returns the of the frequency table entry which
+contains the frequency the CPU shall be set to.
 
 The following macros can be used as iterators over cpufreq_frequency_table:
 
 cpufreq_for_each_entry(pos, table) - iterates over all entries of frequency
 table.
 
-cpufreq-for_each_valid_entry(pos, table) - iterates over all entries,
+cpufreq_for_each_valid_entry(pos, table) - iterates over all entries,
 excluding CPUFREQ_ENTRY_INVALID frequencies.
 Use arguments "pos" - a cpufreq_frequency_table * as a loop cursor and
 "table" - the cpufreq_frequency_table * you want to iterate over.

Documentation/cpu-freq/cpufreq-stats.txt

@@ -35,9 +35,9 @@ cpufreq stats provides following statistics (explained in detail below).
 -  trans_table
 
 All the statistics will be from the time the stats driver has been inserted
-to the time when a read of a particular statistic is done. Obviously, stats 
-driver will not have any information about the frequency transitions before
-the stats driver insertion.
+(or the time the stats were reset) to the time when a read of a particular
+statistic is done. Obviously, stats driver will not have any information
+about the frequency transitions before the stats driver insertion.
 
 --------------------------------------------------------------------------------
 <mysystem>:/sys/devices/system/cpu/cpu0/cpufreq/stats # ls -l
@@ -110,25 +110,13 @@ Config Main Menu
 		CPU Frequency scaling  --->
 			[*] CPU Frequency scaling
 			[*]   CPU frequency translation statistics
-			[*]     CPU frequency translation statistics details
 
 
 "CPU Frequency scaling" (CONFIG_CPU_FREQ) should be enabled to configure
 cpufreq-stats.
 
 "CPU frequency translation statistics" (CONFIG_CPU_FREQ_STAT) provides the
-basic statistics which includes time_in_state and total_trans.
+statistics which includes time_in_state, total_trans and trans_table.
 
-"CPU frequency translation statistics details" (CONFIG_CPU_FREQ_STAT_DETAILS)
-provides fine grained cpufreq stats by trans_table. The reason for having a
-separate config option for trans_table is:
-- trans_table goes against the traditional /sysfs rule of one value per
-  interface. It provides a whole bunch of value in a 2 dimensional matrix
-  form.
-
-Once these two options are enabled and your CPU supports cpufrequency, you
+Once this option is enabled and your CPU supports cpufrequency, you
 will be able to see the CPU frequency statistics in /sysfs.
-
-
-
-

Documentation/cpu-freq/governors.txt

@@ -10,6 +10,8 @@
 
 		    Dominik Brodowski  <linux@brodo.de>
             some additions and corrections by Nico Golde <nico@ngolde.de>
+		Rafael J. Wysocki <rafael.j.wysocki@intel.com>
+		   Viresh Kumar <viresh.kumar@linaro.org>
 
 
 
@@ -28,32 +30,27 @@ Contents:
 2.3  Userspace
 2.4  Ondemand
 2.5  Conservative
+2.6  Schedutil
 
 3.   The Governor Interface in the CPUfreq Core
 
+4.   References
 
 
 1. What Is A CPUFreq Governor?
 ==============================
 
 Most cpufreq drivers (except the intel_pstate and longrun) or even most
-cpu frequency scaling algorithms only offer the CPU to be set to one
-frequency. In order to offer dynamic frequency scaling, the cpufreq
-core must be able to tell these drivers of a "target frequency". So
-these specific drivers will be transformed to offer a "->target/target_index"
-call instead of the existing "->setpolicy" call. For "longrun", all
-stays the same, though.
+cpu frequency scaling algorithms only allow the CPU frequency to be set
+to predefined fixed values.  In order to offer dynamic frequency
+scaling, the cpufreq core must be able to tell these drivers of a
+"target frequency". So these specific drivers will be transformed to
+offer a "->target/target_index/fast_switch()" call instead of the
+"->setpolicy()" call. For set_policy drivers, all stays the same,
+though.
 
 How to decide what frequency within the CPUfreq policy should be used?
-That's done using "cpufreq governors". Two are already in this patch
--- they're the already existing "powersave" and "performance" which
-set the frequency statically to the lowest or highest frequency,
-respectively. At least two more such governors will be ready for
-addition in the near future, but likely many more as there are various
-different theories and models about dynamic frequency scaling
-around. Using such a generic interface as cpufreq offers to scaling
-governors, these can be tested extensively, and the best one can be
-selected for each specific use.
+That's done using "cpufreq governors".
 
 Basically, it's the following flow graph:
 
@@ -71,7 +68,7 @@ CPU can be set to switch independently	 |	   CPU can only be set
 		    /			       the limits of policy->{min,max}
 		   /			            \
 		  /				     \
-	Using the ->setpolicy call,		 Using the ->target/target_index call,
+	Using the ->setpolicy call,		 Using the ->target/target_index/fast_switch call,
 	    the limits and the			  the frequency closest
 	     "policy" is set.			  to target_freq is set.
 						  It is assured that it
@@ -109,114 +106,159 @@ directory.
 2.4 Ondemand
 ------------
 
-The CPUfreq governor "ondemand" sets the CPU depending on the
-current usage. To do this the CPU must have the capability to
-switch the frequency very quickly.  There are a number of sysfs file
-accessible parameters:
-
-sampling_rate: measured in uS (10^-6 seconds), this is how often you
-want the kernel to look at the CPU usage and to make decisions on
-what to do about the frequency.  Typically this is set to values of
-around '10000' or more. It's default value is (cmp. with users-guide.txt):
-transition_latency * 1000
-Be aware that transition latency is in ns and sampling_rate is in us, so you
-get the same sysfs value by default.
-Sampling rate should always get adjusted considering the transition latency
-To set the sampling rate 750 times as high as the transition latency
-in the bash (as said, 1000 is default), do:
-echo `$(($(cat cpuinfo_transition_latency) * 750 / 1000)) \
-    >ondemand/sampling_rate
-
-sampling_rate_min:
-The sampling rate is limited by the HW transition latency:
-transition_latency * 100
-Or by kernel restrictions:
-If CONFIG_NO_HZ_COMMON is set, the limit is 10ms fixed.
-If CONFIG_NO_HZ_COMMON is not set or nohz=off boot parameter is used, the
-limits depend on the CONFIG_HZ option:
-HZ=1000: min=20000us  (20ms)
-HZ=250:  min=80000us  (80ms)
-HZ=100:  min=200000us (200ms)
-The highest value of kernel and HW latency restrictions is shown and
-used as the minimum sampling rate.
-
-up_threshold: defines what the average CPU usage between the samplings
-of 'sampling_rate' needs to be for the kernel to make a decision on
-whether it should increase the frequency.  For example when it is set
-to its default value of '95' it means that between the checking
-intervals the CPU needs to be on average more than 95% in use to then
-decide that the CPU frequency needs to be increased.  
-
-ignore_nice_load: this parameter takes a value of '0' or '1'. When
-set to '0' (its default), all processes are counted towards the
-'cpu utilisation' value.  When set to '1', the processes that are
-run with a 'nice' value will not count (and thus be ignored) in the
-overall usage calculation.  This is useful if you are running a CPU
-intensive calculation on your laptop that you do not care how long it
-takes to complete as you can 'nice' it and prevent it from taking part
-in the deciding process of whether to increase your CPU frequency.
-
-sampling_down_factor: this parameter controls the rate at which the
-kernel makes a decision on when to decrease the frequency while running
-at top speed. When set to 1 (the default) decisions to reevaluate load
-are made at the same interval regardless of current clock speed. But
-when set to greater than 1 (e.g. 100) it acts as a multiplier for the
-scheduling interval for reevaluating load when the CPU is at its top
-speed due to high load. This improves performance by reducing the overhead
-of load evaluation and helping the CPU stay at its top speed when truly
-busy, rather than shifting back and forth in speed. This tunable has no
-effect on behavior at lower speeds/lower CPU loads.
-
-powersave_bias: this parameter takes a value between 0 to 1000. It
-defines the percentage (times 10) value of the target frequency that
-will be shaved off of the target. For example, when set to 100 -- 10%,
-when ondemand governor would have targeted 1000 MHz, it will target
-1000 MHz - (10% of 1000 MHz) = 900 MHz instead. This is set to 0
-(disabled) by default.
-When AMD frequency sensitivity powersave bias driver --
-drivers/cpufreq/amd_freq_sensitivity.c is loaded, this parameter
-defines the workload frequency sensitivity threshold in which a lower
-frequency is chosen instead of ondemand governor's original target.
-The frequency sensitivity is a hardware reported (on AMD Family 16h
-Processors and above) value between 0 to 100% that tells software how
-the performance of the workload running on a CPU will change when
-frequency changes. A workload with sensitivity of 0% (memory/IO-bound)
-will not perform any better on higher core frequency, whereas a
-workload with sensitivity of 100% (CPU-bound) will perform better
-higher the frequency. When the driver is loaded, this is set to 400
-by default -- for CPUs running workloads with sensitivity value below
-40%, a lower frequency is chosen. Unloading the driver or writing 0
-will disable this feature.
+The CPUfreq governor "ondemand" sets the CPU frequency depending on the
+current system load. Load estimation is triggered by the scheduler
+through the update_util_data->func hook; when triggered, cpufreq checks
+the CPU-usage statistics over the last period and the governor sets the
+CPU accordingly.  The CPU must have the capability to switch the
+frequency very quickly.
+
+Sysfs files:
+
+* sampling_rate:
+
+  Measured in uS (10^-6 seconds), this is how often you want the kernel
+  to look at the CPU usage and to make decisions on what to do about the
+  frequency.  Typically this is set to values of around '10000' or more.
+  It's default value is (cmp. with users-guide.txt): transition_latency
+  * 1000.  Be aware that transition latency is in ns and sampling_rate
+  is in us, so you get the same sysfs value by default.  Sampling rate
+  should always get adjusted considering the transition latency to set
+  the sampling rate 750 times as high as the transition latency in the
+  bash (as said, 1000 is default), do:
+
+  $ echo `$(($(cat cpuinfo_transition_latency) * 750 / 1000)) > ondemand/sampling_rate
+
+* sampling_rate_min:
+
+  The sampling rate is limited by the HW transition latency:
+  transition_latency * 100
+
+  Or by kernel restrictions:
+  - If CONFIG_NO_HZ_COMMON is set, the limit is 10ms fixed.
+  - If CONFIG_NO_HZ_COMMON is not set or nohz=off boot parameter is
+    used, the limits depend on the CONFIG_HZ option:
+    HZ=1000: min=20000us  (20ms)
+    HZ=250:  min=80000us  (80ms)
+    HZ=100:  min=200000us (200ms)
+
+  The highest value of kernel and HW latency restrictions is shown and
+  used as the minimum sampling rate.
+
+* up_threshold:
+
+  This defines what the average CPU usage between the samplings of
+  'sampling_rate' needs to be for the kernel to make a decision on
+  whether it should increase the frequency.  For example when it is set
+  to its default value of '95' it means that between the checking
+  intervals the CPU needs to be on average more than 95% in use to then
+  decide that the CPU frequency needs to be increased.
+
+* ignore_nice_load:
+
+  This parameter takes a value of '0' or '1'. When set to '0' (its
+  default), all processes are counted towards the 'cpu utilisation'
+  value.  When set to '1', the processes that are run with a 'nice'
+  value will not count (and thus be ignored) in the overall usage
+  calculation.  This is useful if you are running a CPU intensive
+  calculation on your laptop that you do not care how long it takes to
+  complete as you can 'nice' it and prevent it from taking part in the
+  deciding process of whether to increase your CPU frequency.
+
+* sampling_down_factor:
+
+  This parameter controls the rate at which the kernel makes a decision
+  on when to decrease the frequency while running at top speed. When set
+  to 1 (the default) decisions to reevaluate load are made at the same
+  interval regardless of current clock speed. But when set to greater
+  than 1 (e.g. 100) it acts as a multiplier for the scheduling interval
+  for reevaluating load when the CPU is at its top speed due to high
+  load. This improves performance by reducing the overhead of load
+  evaluation and helping the CPU stay at its top speed when truly busy,
+  rather than shifting back and forth in speed. This tunable has no
+  effect on behavior at lower speeds/lower CPU loads.
+
+* powersave_bias:
+
+  This parameter takes a value between 0 to 1000. It defines the
+  percentage (times 10) value of the target frequency that will be
+  shaved off of the target. For example, when set to 100 -- 10%, when
+  ondemand governor would have targeted 1000 MHz, it will target
+  1000 MHz - (10% of 1000 MHz) = 900 MHz instead. This is set to 0
+  (disabled) by default.
+
+  When AMD frequency sensitivity powersave bias driver --
+  drivers/cpufreq/amd_freq_sensitivity.c is loaded, this parameter
+  defines the workload frequency sensitivity threshold in which a lower
+  frequency is chosen instead of ondemand governor's original target.
+  The frequency sensitivity is a hardware reported (on AMD Family 16h
+  Processors and above) value between 0 to 100% that tells software how
+  the performance of the workload running on a CPU will change when
+  frequency changes. A workload with sensitivity of 0% (memory/IO-bound)
+  will not perform any better on higher core frequency, whereas a
+  workload with sensitivity of 100% (CPU-bound) will perform better
+  higher the frequency. When the driver is loaded, this is set to 400 by
+  default -- for CPUs running workloads with sensitivity value below
+  40%, a lower frequency is chosen. Unloading the driver or writing 0
+  will disable this feature.
 
 
 2.5 Conservative
 ----------------
 
 The CPUfreq governor "conservative", much like the "ondemand"
-governor, sets the CPU depending on the current usage.  It differs in
-behaviour in that it gracefully increases and decreases the CPU speed
-rather than jumping to max speed the moment there is any load on the
-CPU.  This behaviour more suitable in a battery powered environment.
-The governor is tweaked in the same manner as the "ondemand" governor
-through sysfs with the addition of:
-
-freq_step: this describes what percentage steps the cpu freq should be
-increased and decreased smoothly by.  By default the cpu frequency will
-increase in 5% chunks of your maximum cpu frequency.  You can change this
-value to anywhere between 0 and 100 where '0' will effectively lock your
-CPU at a speed regardless of its load whilst '100' will, in theory, make
-it behave identically to the "ondemand" governor.
-
-down_threshold: same as the 'up_threshold' found for the "ondemand"
-governor but for the opposite direction.  For example when set to its
-default value of '20' it means that if the CPU usage needs to be below
-20% between samples to have the frequency decreased.
-
-sampling_down_factor: similar functionality as in "ondemand" governor.
-But in "conservative", it controls the rate at which the kernel makes
-a decision on when to decrease the frequency while running in any
-speed. Load for frequency increase is still evaluated every
-sampling rate.
+governor, sets the CPU frequency depending on the current usage.  It
+differs in behaviour in that it gracefully increases and decreases the
+CPU speed rather than jumping to max speed the moment there is any load
+on the CPU. This behaviour is more suitable in a battery powered
+environment.  The governor is tweaked in the same manner as the
+"ondemand" governor through sysfs with the addition of:
+
+* freq_step:
+
+  This describes what percentage steps the cpu freq should be increased
+  and decreased smoothly by.  By default the cpu frequency will increase
+  in 5% chunks of your maximum cpu frequency.  You can change this value
+  to anywhere between 0 and 100 where '0' will effectively lock your CPU
+  at a speed regardless of its load whilst '100' will, in theory, make
+  it behave identically to the "ondemand" governor.
+
+* down_threshold:
+
+  Same as the 'up_threshold' found for the "ondemand" governor but for
+  the opposite direction.  For example when set to its default value of
+  '20' it means that if the CPU usage needs to be below 20% between
+  samples to have the frequency decreased.
+
+* sampling_down_factor:
+
+  Similar functionality as in "ondemand" governor.  But in
+  "conservative", it controls the rate at which the kernel makes a
+  decision on when to decrease the frequency while running in any speed.
+  Load for frequency increase is still evaluated every sampling rate.
+
+
+2.6 Schedutil
+-------------
+
+The "schedutil" governor aims at better integration with the Linux
+kernel scheduler.  Load estimation is achieved through the scheduler's
+Per-Entity Load Tracking (PELT) mechanism, which also provides
+information about the recent load [1].  This governor currently does
+load based DVFS only for tasks managed by CFS. RT and DL scheduler tasks
+are always run at the highest frequency.  Unlike all the other
+governors, the code is located under the kernel/sched/ directory.
+
+Sysfs files:
+
+* rate_limit_us:
+
+  This contains a value in microseconds. The governor waits for
+  rate_limit_us time before reevaluating the load again, after it has
+  evaluated the load once.
+
+For an in-depth comparison with the other governors refer to [2].
+
 
 3. The Governor Interface in the CPUfreq Core
 =============================================
@@ -225,26 +267,10 @@ A new governor must register itself with the CPUfreq core using
 "cpufreq_register_governor". The struct cpufreq_governor, which has to
 be passed to that function, must contain the following values:
 
-governor->name -	    A unique name for this governor
-governor->governor -	    The governor callback function
-governor->owner	-	    .THIS_MODULE for the governor module (if 
-			    appropriate)
-
-The governor->governor callback is called with the current (or to-be-set)
-cpufreq_policy struct for that CPU, and an unsigned int event. The
-following events are currently defined:
-
-CPUFREQ_GOV_START:   This governor shall start its duty for the CPU
-		     policy->cpu
-CPUFREQ_GOV_STOP:    This governor shall end its duty for the CPU
-		     policy->cpu
-CPUFREQ_GOV_LIMITS:  The limits for CPU policy->cpu have changed to
-		     policy->min and policy->max.
-
-If you need other "events" externally of your driver, _only_ use the
-cpufreq_governor_l(unsigned int cpu, unsigned int event) call to the
-CPUfreq core to ensure proper locking.
+governor->name - A unique name for this governor.
+governor->owner - .THIS_MODULE for the governor module (if appropriate).
 
+plus a set of hooks to the functions implementing the governor's logic.
 
 The CPUfreq governor may call the CPU processor driver using one of
 these two functions:
@@ -258,12 +284,18 @@ int __cpufreq_driver_target(struct cpufreq_policy *policy,
                                    unsigned int relation);
 
 target_freq must be within policy->min and policy->max, of course.
-What's the difference between these two functions? When your governor
-still is in a direct code path of a call to governor->governor, the
-per-CPU cpufreq lock is still held in the cpufreq core, and there's
-no need to lock it again (in fact, this would cause a deadlock). So
-use __cpufreq_driver_target only in these cases. In all other cases 
-(for example, when there's a "daemonized" function that wakes up 
-every second), use cpufreq_driver_target to lock the cpufreq per-CPU
-lock before the command is passed to the cpufreq processor driver.
+What's the difference between these two functions? When your governor is
+in a direct code path of a call to governor callbacks, like
+governor->start(), the policy->rwsem is still held in the cpufreq core,
+and there's no need to lock it again (in fact, this would cause a
+deadlock). So use __cpufreq_driver_target only in these cases. In all
+other cases (for example, when there's a "daemonized" function that
+wakes up every second), use cpufreq_driver_target to take policy->rwsem
+before the command is passed to the cpufreq driver.
+
+4. References
+=============
+
+[1] Per-entity load tracking: https://lwn.net/Articles/531853/
+[2] Improvements in CPU frequency management: https://lwn.net/Articles/682391/
 

Documentation/cpu-freq/index.txt

@@ -18,16 +18,29 @@
 
 Documents in this directory:
 ----------------------------
+
+amd-powernow.txt -	AMD powernow driver specific file.
+
+boost.txt -		Frequency boosting support.
+
 core.txt	-	General description of the CPUFreq core and
-			of CPUFreq notifiers
+			of CPUFreq notifiers.
+
+cpu-drivers.txt -	How to implement a new cpufreq processor driver.
 
-cpu-drivers.txt -	How to implement a new cpufreq processor driver
+cpufreq-nforce2.txt -	nVidia nForce2 platform specific file.
+
+cpufreq-stats.txt -	General description of sysfs cpufreq stats.
 
 governors.txt	-	What are cpufreq governors and how to
 			implement them?
 
 index.txt	-	File index, Mailing list and Links (this document)
 
+intel-pstate.txt -	Intel pstate cpufreq driver specific file.
+
+pcc-cpufreq.txt -	PCC cpufreq driver specific file.
+
 user-guide.txt	-	User Guide to CPUFreq
 
 
@@ -35,9 +48,7 @@ Mailing List
 ------------
 There is a CPU frequency changing CVS commit and general list where
 you can report bugs, problems or submit patches. To post a message,
-send an email to linux-pm@vger.kernel.org, to subscribe go to
-http://vger.kernel.org/vger-lists.html#linux-pm and follow the
-instructions there.
+send an email to linux-pm@vger.kernel.org.
 
 Links
 -----
@@ -48,7 +59,7 @@ how to access the CVS repository:
 * http://cvs.arm.linux.org.uk/
 
 the CPUFreq Mailing list:
-* http://vger.kernel.org/vger-lists.html#cpufreq
+* http://vger.kernel.org/vger-lists.html#linux-pm
 
 Clock and voltage scaling for the SA-1100:
 * http://www.lartmaker.nl/projects/scaling

Documentation/cpu-freq/intel-pstate.txt

@@ -85,6 +85,21 @@ Sysfs will show :
 Refer to "Intel® 64 and IA-32 Architectures Software Developer’s Manual
 Volume 3: System Programming Guide" to understand ratios.
 
+There is one more sysfs attribute in /sys/devices/system/cpu/intel_pstate/
+that can be used for controlling the operation mode of the driver:
+
+      status: Three settings are possible:
+      "off"     - The driver is not in use at this time.
+      "active"  - The driver works as a P-state governor (default).
+      "passive" - The driver works as a regular cpufreq one and collaborates
+                  with the generic cpufreq governors (it sets P-states as
+                  requested by those governors).
+      The current setting is returned by reads from this attribute.  Writing one
+      of the above strings to it changes the operation mode as indicated by that
+      string, if possible.  If HW-managed P-states (HWP) are enabled, it is not
+      possible to change the driver's operation mode and attempts to write to
+      this attribute will fail.
+
 cpufreq sysfs for Intel P-State
 
 Since this driver registers with cpufreq, cpufreq sysfs is also presented.

Documentation/cpu-freq/user-guide.txt

@@ -18,7 +18,7 @@
 Contents:
 ---------
 1. Supported Architectures and Processors
-1.1 ARM
+1.1 ARM and ARM64
 1.2 x86
 1.3 sparc64
 1.4 ppc
@@ -37,16 +37,10 @@ Contents:
 1. Supported Architectures and Processors
 =========================================
 
-1.1 ARM
--------
-
-The following ARM processors are supported by cpufreq:
-
-ARM Integrator
-ARM-SA1100
-ARM-SA1110
-Intel PXA
+1.1 ARM and ARM64
+-----------------
 
+Almost all ARM and ARM64 platforms support CPU frequency scaling.
 
 1.2 x86
 -------
@@ -69,6 +63,7 @@ Transmeta Crusoe
 Transmeta Efficeon
 VIA Cyrix 3 / C3
 various processors on some ACPI 2.0-compatible systems [*]
+And many more
 
 [*] Only if "ACPI Processor Performance States" are available
 to the ACPI<->BIOS interface.
@@ -147,10 +142,19 @@ mounted it at /sys, the cpufreq interface is located in a subdirectory
 "cpufreq" within the cpu-device directory
 (e.g. /sys/devices/system/cpu/cpu0/cpufreq/ for the first CPU).
 
+affected_cpus :			List of Online CPUs that require software
+				coordination of frequency.
+
+cpuinfo_cur_freq :		Current frequency of the CPU as obtained from
+				the hardware, in KHz. This is the frequency
+				the CPU actually runs at.
+
 cpuinfo_min_freq :		this file shows the minimum operating
 				frequency the processor can run at(in kHz) 
+
 cpuinfo_max_freq :		this file shows the maximum operating
 				frequency the processor can run at(in kHz) 
+
 cpuinfo_transition_latency	The time it takes on this CPU to
 				switch between two frequencies in nano
 				seconds. If unknown or known to be
@@ -163,25 +167,30 @@ cpuinfo_transition_latency	The time it takes on this CPU to
 				userspace daemon. Make sure to not
 				switch the frequency too often
 				resulting in performance loss.
-scaling_driver :		this file shows what cpufreq driver is
-				used to set the frequency on this CPU
+
+related_cpus :			List of Online + Offline CPUs that need software
+				coordination of frequency.
+
+scaling_available_frequencies : List of available frequencies, in KHz.
 
 scaling_available_governors :	this file shows the CPUfreq governors
 				available in this kernel. You can see the
 				currently activated governor in
 
+scaling_cur_freq :		Current frequency of the CPU as determined by
+				the governor and cpufreq core, in KHz. This is
+				the frequency the kernel thinks the CPU runs
+				at.
+
+scaling_driver :		this file shows what cpufreq driver is
+				used to set the frequency on this CPU
+
 scaling_governor,		and by "echoing" the name of another
 				governor you can change it. Please note
 				that some governors won't load - they only
 				work on some specific architectures or
 				processors.
 
-cpuinfo_cur_freq :		Current frequency of the CPU as obtained from
-				the hardware, in KHz. This is the frequency
-				the CPU actually runs at.
-
-scaling_available_frequencies : List of available frequencies, in KHz.
-
 scaling_min_freq and
 scaling_max_freq		show the current "policy limits" (in
 				kHz). By echoing new values into these
@@ -190,16 +199,11 @@ scaling_max_freq		show the current "policy limits" (in
 				first set scaling_max_freq, then
 				scaling_min_freq.
 
-affected_cpus :			List of Online CPUs that require software
-				coordination of frequency.
-
-related_cpus :			List of Online + Offline CPUs that need software
-				coordination of frequency.
-
-scaling_cur_freq :		Current frequency of the CPU as determined by
-				the governor and cpufreq core, in KHz. This is
-				the frequency the kernel thinks the CPU runs
-				at.
+scaling_setspeed		This can be read to get the currently programmed
+				value by the governor. This can be written to
+				change the current frequency for a group of
+				CPUs, represented by a policy. This is supported
+				currently only by the userspace governor.
 
 bios_limit :			If the BIOS tells the OS to limit a CPU to
 				lower frequencies, the user can read out the

Documentation/devicetree/bindings/cpufreq/ti-cpufreq.txt

+TI CPUFreq and OPP bindings
+================================
+
+Certain TI SoCs, like those in the am335x, am437x, am57xx, and dra7xx
+families support different OPPs depending on the silicon variant in use.
+The ti-cpufreq driver can use revision and an efuse value from the SoC to
+provide the OPP framework with supported hardware information. This is
+used to determine which OPPs from the operating-points-v2 table get enabled
+when it is parsed by the OPP framework.
+
+Required properties:
+--------------------
+In 'cpus' nodes:
+- operating-points-v2: Phandle to the operating-points-v2 table to use.
+
+In 'operating-points-v2' table:
+- compatible: Should be
+	- 'operating-points-v2-ti-cpu' for am335x, am43xx, and dra7xx/am57xx SoCs
+- syscon: A phandle pointing to a syscon node representing the control module
+	  register space of the SoC.
+
+Optional properties:
+--------------------
+For each opp entry in 'operating-points-v2' table:
+- opp-supported-hw: Two bitfields indicating:
+	1. Which revision of the SoC the OPP is supported by
+	2. Which eFuse bits indicate this OPP is available
+
+	A bitwise AND is performed against these values and if any bit
+	matches, the OPP gets enabled.
+
+Example:
+--------
+
+/* From arch/arm/boot/dts/am33xx.dtsi */
+cpus {
+	#address-cells = <1>;
+	#size-cells = <0>;
+	cpu@0 {
+		compatible = "arm,cortex-a8";
+		device_type = "cpu";
+		reg = <0>;
+
+		operating-points-v2 = <&cpu0_opp_table>;
+
+		clocks = <&dpll_mpu_ck>;
+		clock-names = "cpu";
+
+		clock-latency = <300000>; /* From omap-cpufreq driver */
+	};
+};
+
+/*
+ * cpu0 has different OPPs depending on SoC revision and some on revisions
+ * 0x2 and 0x4 have eFuse bits that indicate if they are available or not
+ */
+cpu0_opp_table: opp-table {
+	compatible = "operating-points-v2-ti-cpu";
+	syscon = <&scm_conf>;
+
+	/*
+	 * The three following nodes are marked with opp-suspend
+	 * because they can not be enabled simultaneously on a
+	 * single SoC.
+	 */
+	opp50@300000000 {
+		opp-hz = /bits/ 64 <300000000>;
+		opp-microvolt = <950000 931000 969000>;
+		opp-supported-hw = <0x06 0x0010>;
+		opp-suspend;
+	};
+
+	opp100@275000000 {
+		opp-hz = /bits/ 64 <275000000>;
+		opp-microvolt = <1100000 1078000 1122000>;
+		opp-supported-hw = <0x01 0x00FF>;
+		opp-suspend;
+	};
+
+	opp100@300000000 {
+		opp-hz = /bits/ 64 <300000000>;
+		opp-microvolt = <1100000 1078000 1122000>;
+		opp-supported-hw = <0x06 0x0020>;
+		opp-suspend;
+	};
+
+	opp100@500000000 {
+		opp-hz = /bits/ 64 <500000000>;
+		opp-microvolt = <1100000 1078000 1122000>;
+		opp-supported-hw = <0x01 0xFFFF>;
+	};
+
+	opp100@600000000 {
+		opp-hz = /bits/ 64 <600000000>;
+		opp-microvolt = <1100000 1078000 1122000>;
+		opp-supported-hw = <0x06 0x0040>;
+	};
+
+	opp120@600000000 {
+		opp-hz = /bits/ 64 <600000000>;
+		opp-microvolt = <1200000 1176000 1224000>;
+		opp-supported-hw = <0x01 0xFFFF>;
+	};
+
+	opp120@720000000 {
+		opp-hz = /bits/ 64 <720000000>;
+		opp-microvolt = <1200000 1176000 1224000>;
+		opp-supported-hw = <0x06 0x0080>;
+	};
+
+	oppturbo@720000000 {
+		opp-hz = /bits/ 64 <720000000>;
+		opp-microvolt = <1260000 1234800 1285200>;
+		opp-supported-hw = <0x01 0xFFFF>;
+	};
+
+	oppturbo@800000000 {
+		opp-hz = /bits/ 64 <800000000>;
+		opp-microvolt = <1260000 1234800 1285200>;
+		opp-supported-hw = <0x06 0x0100>;
+	};
+
+	oppnitro@1000000000 {
+		opp-hz = /bits/ 64 <1000000000>;
+		opp-microvolt = <1325000 1298500 1351500>;
+		opp-supported-hw = <0x04 0x0200>;
+	};
+};

Documentation/devicetree/bindings/devfreq/exynos-bus.txt

@@ -123,6 +123,20 @@ Detailed correlation between sub-blocks and power line according to Exynos SoC:
 		|--- FSYS
 		|--- FSYS2
 
+- In case of Exynos5433, there is VDD_INT power line as following:
+	VDD_INT |--- G2D (parent device)
+		|--- MSCL
+		|--- GSCL
+		|--- JPEG
+		|--- MFC
+		|--- HEVC
+		|--- BUS0
+		|--- BUS1
+		|--- BUS2
+		|--- PERIS (Fixed clock rate)
+		|--- PERIC (Fixed clock rate)
+		|--- FSYS  (Fixed clock rate)
+
 Example1:
 	Show the AXI buses of Exynos3250 SoC. Exynos3250 divides the buses to
 	power line (regulator). The MIF (Memory Interface) AXI bus is used to

Documentation/power/opp.txt

@@ -79,22 +79,6 @@ dependent subsystems such as cpufreq are left to the discretion of the SoC
 specific framework which uses the OPP library. Similar care needs to be taken
 care to refresh the cpufreq table in cases of these operations.
 
-WARNING on OPP List locking mechanism:
--------------------------------------------------
-OPP library uses RCU for exclusivity. RCU allows the query functions to operate
-in multiple contexts and this synchronization mechanism is optimal for a read
-intensive operations on data structure as the OPP library caters to.
-
-To ensure that the data retrieved are sane, the users such as SoC framework
-should ensure that the section of code operating on OPP queries are locked
-using RCU read locks. The opp_find_freq_{exact,ceil,floor},
-opp_get_{voltage, freq, opp_count} fall into this category.
-
-opp_{add,enable,disable} are updaters which use mutex and implement it's own
-RCU locking mechanisms. These functions should *NOT* be called under RCU locks
-and other contexts that prevent blocking functions in RCU or mutex operations
-from working.
-
 2. Initial OPP List Registration
 ================================
 The SoC implementation calls dev_pm_opp_add function iteratively to add OPPs per
@@ -137,15 +121,18 @@ functions return the matching pointer representing the opp if a match is
 found, else returns error. These errors are expected to be handled by standard
 error checks such as IS_ERR() and appropriate actions taken by the caller.
 
+Callers of these functions shall call dev_pm_opp_put() after they have used the
+OPP. Otherwise the memory for the OPP will never get freed and result in
+memleak.
+
 dev_pm_opp_find_freq_exact - Search for an OPP based on an *exact* frequency and
 	availability. This function is especially useful to enable an OPP which
 	is not available by default.
 	Example: In a case when SoC framework detects a situation where a
 	higher frequency could be made available, it can use this function to
 	find the OPP prior to call the dev_pm_opp_enable to actually make it available.
-	 rcu_read_lock();
 	 opp = dev_pm_opp_find_freq_exact(dev, 1000000000, false);
-	 rcu_read_unlock();
+	 dev_pm_opp_put(opp);
 	 /* dont operate on the pointer.. just do a sanity check.. */
 	 if (IS_ERR(opp)) {
 		pr_err("frequency not disabled!\n");
@@ -163,9 +150,8 @@ dev_pm_opp_find_freq_floor - Search for an available OPP which is *at most* the
 	frequency.
 	Example: To find the highest opp for a device:
 	 freq = ULONG_MAX;
-	 rcu_read_lock();
-	 dev_pm_opp_find_freq_floor(dev, &freq);
-	 rcu_read_unlock();
+	 opp = dev_pm_opp_find_freq_floor(dev, &freq);
+	 dev_pm_opp_put(opp);
 
 dev_pm_opp_find_freq_ceil - Search for an available OPP which is *at least* the
 	provided frequency. This function is useful while searching for a
@@ -173,17 +159,15 @@ dev_pm_opp_find_freq_ceil - Search for an available OPP which is *at least* the
 	frequency.
 	Example 1: To find the lowest opp for a device:
 	 freq = 0;
-	 rcu_read_lock();
-	 dev_pm_opp_find_freq_ceil(dev, &freq);
-	 rcu_read_unlock();
+	 opp = dev_pm_opp_find_freq_ceil(dev, &freq);
+	 dev_pm_opp_put(opp);
 	Example 2: A simplified implementation of a SoC cpufreq_driver->target:
 	 soc_cpufreq_target(..)
 	 {
 		/* Do stuff like policy checks etc. */
 		/* Find the best frequency match for the req */
-		rcu_read_lock();
 		opp = dev_pm_opp_find_freq_ceil(dev, &freq);
-		rcu_read_unlock();
+		dev_pm_opp_put(opp);
 		if (!IS_ERR(opp))
 			soc_switch_to_freq_voltage(freq);
 		else
@@ -208,9 +192,8 @@ dev_pm_opp_enable - Make a OPP available for operation.
 	implementation might choose to do something as follows:
 	 if (cur_temp < temp_low_thresh) {
 		/* Enable 1GHz if it was disabled */
-		rcu_read_lock();
 		opp = dev_pm_opp_find_freq_exact(dev, 1000000000, false);
-		rcu_read_unlock();
+		dev_pm_opp_put(opp);
 		/* just error check */
 		if (!IS_ERR(opp))
 			ret = dev_pm_opp_enable(dev, 1000000000);
@@ -224,9 +207,8 @@ dev_pm_opp_disable - Make an OPP to be not available for operation
 	choose to do something as follows:
 	 if (cur_temp > temp_high_thresh) {
 		/* Disable 1GHz if it was enabled */
-		rcu_read_lock();
 		opp = dev_pm_opp_find_freq_exact(dev, 1000000000, true);
-		rcu_read_unlock();
+		dev_pm_opp_put(opp);
 		/* just error check */
 		if (!IS_ERR(opp))
 			ret = dev_pm_opp_disable(dev, 1000000000);
@@ -249,10 +231,9 @@ dev_pm_opp_get_voltage - Retrieve the voltage represented by the opp pointer.
 	 soc_switch_to_freq_voltage(freq)
 	 {
 		/* do things */
-		rcu_read_lock();
 		opp = dev_pm_opp_find_freq_ceil(dev, &freq);
 		v = dev_pm_opp_get_voltage(opp);
-		rcu_read_unlock();
+		dev_pm_opp_put(opp);
 		if (v)
 			regulator_set_voltage(.., v);
 		/* do other things */
@@ -266,12 +247,12 @@ dev_pm_opp_get_freq - Retrieve the freq represented by the opp pointer.
 	 {
 		/* do things.. */
 		 max_freq = ULONG_MAX;
-		 rcu_read_lock();
 		 max_opp = dev_pm_opp_find_freq_floor(dev,&max_freq);
 		 requested_opp = dev_pm_opp_find_freq_ceil(dev,&freq);
 		 if (!IS_ERR(max_opp) && !IS_ERR(requested_opp))
 			r = soc_test_validity(max_opp, requested_opp);
-		 rcu_read_unlock();
+		 dev_pm_opp_put(max_opp);
+		 dev_pm_opp_put(requested_opp);
 		/* do other things */
 	 }
 	 soc_test_validity(..)
@@ -289,7 +270,6 @@ dev_pm_opp_get_opp_count - Retrieve the number of available opps for a device
 	 soc_notify_coproc_available_frequencies()
 	 {
 		/* Do things */
-		rcu_read_lock();
 		num_available = dev_pm_opp_get_opp_count(dev);
 		speeds = kzalloc(sizeof(u32) * num_available, GFP_KERNEL);
 		/* populate the table in increasing order */
@@ -298,8 +278,8 @@ dev_pm_opp_get_opp_count - Retrieve the number of available opps for a device
 			speeds[i] = freq;
 			freq++;
 			i++;
+			dev_pm_opp_put(opp);
 		}
-		rcu_read_unlock();
 
 		soc_notify_coproc(AVAILABLE_FREQs, speeds, num_available);
 		/* Do other things */

Documentation/power/states.txt

@@ -25,7 +25,7 @@ to be used subsequently to change to the one represented by that string.
 Consequently, there are two ways to cause the system to go into the
 Suspend-To-Idle sleep state.  The first one is to write "freeze" directly to
 /sys/power/state.  The second one is to write "s2idle" to /sys/power/mem_sleep
-and then to wrtie "mem" to /sys/power/state.  Similarly, there are two ways
+and then to write "mem" to /sys/power/state.  Similarly, there are two ways
 to cause the system to go into the Power-On Suspend sleep state (the strings to
 write to the control files in that case are "standby" or "shallow" and "mem",
 respectively) if that state is supported by the platform.  In turn, there is

MAINTAINERS

@@ -2692,6 +2692,13 @@ F:	drivers/irqchip/irq-brcmstb*
 F:	include/linux/bcm963xx_nvram.h
 F:	include/linux/bcm963xx_tag.h
 
+BROADCOM BMIPS CPUFREQ DRIVER
+M:	Markus Mayer <mmayer@broadcom.com>
+M:	bcm-kernel-feedback-list@broadcom.com
+L:	linux-pm@vger.kernel.org
+S:	Maintained
+F:	drivers/cpufreq/bmips-cpufreq.c
+
 BROADCOM TG3 GIGABIT ETHERNET DRIVER
 M:	Siva Reddy Kallam <siva.kallam@broadcom.com>
 M:	Prashant Sreedharan <prashant@broadcom.com>

arch/arm/configs/exynos_defconfig

@@ -24,7 +24,7 @@ CONFIG_ARM_APPENDED_DTB=y
 CONFIG_ARM_ATAG_DTB_COMPAT=y
 CONFIG_CMDLINE="root=/dev/ram0 rw ramdisk=8192 initrd=0x41000000,8M console=ttySAC1,115200 init=/linuxrc mem=256M"
 CONFIG_CPU_FREQ=y
-CONFIG_CPU_FREQ_STAT_DETAILS=y
+CONFIG_CPU_FREQ_STAT=y
 CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND=y
 CONFIG_CPU_FREQ_GOV_POWERSAVE=m
 CONFIG_CPU_FREQ_GOV_USERSPACE=m

arch/arm/configs/multi_v5_defconfig

@@ -58,7 +58,7 @@ CONFIG_ZBOOT_ROM_BSS=0x0
 CONFIG_ARM_APPENDED_DTB=y
 CONFIG_ARM_ATAG_DTB_COMPAT=y
 CONFIG_CPU_FREQ=y
-CONFIG_CPU_FREQ_STAT_DETAILS=y
+CONFIG_CPU_FREQ_STAT=y
 CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND=y
 CONFIG_CPU_IDLE=y
 CONFIG_ARM_KIRKWOOD_CPUIDLE=y

arch/arm/configs/multi_v7_defconfig

@@ -132,7 +132,7 @@ CONFIG_ARM_ATAG_DTB_COMPAT=y
 CONFIG_KEXEC=y
 CONFIG_EFI=y
 CONFIG_CPU_FREQ=y
-CONFIG_CPU_FREQ_STAT_DETAILS=y
+CONFIG_CPU_FREQ_STAT=y
 CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND=y
 CONFIG_CPU_FREQ_GOV_POWERSAVE=m
 CONFIG_CPU_FREQ_GOV_USERSPACE=m

arch/arm/configs/mvebu_v5_defconfig

@@ -44,7 +44,7 @@ CONFIG_ZBOOT_ROM_BSS=0x0
 CONFIG_ARM_APPENDED_DTB=y
 CONFIG_ARM_ATAG_DTB_COMPAT=y
 CONFIG_CPU_FREQ=y
-CONFIG_CPU_FREQ_STAT_DETAILS=y
+CONFIG_CPU_FREQ_STAT=y
 CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND=y
 CONFIG_CPU_IDLE=y
 CONFIG_ARM_KIRKWOOD_CPUIDLE=y

arch/arm/configs/pxa_defconfig

@@ -97,7 +97,7 @@ CONFIG_ZBOOT_ROM_BSS=0x0
 CONFIG_CMDLINE="root=/dev/ram0 ro"
 CONFIG_KEXEC=y
 CONFIG_CPU_FREQ=y
-CONFIG_CPU_FREQ_STAT_DETAILS=y
+CONFIG_CPU_FREQ_STAT=y
 CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND=y
 CONFIG_CPU_FREQ_GOV_POWERSAVE=m
 CONFIG_CPU_FREQ_GOV_USERSPACE=m

arch/arm/configs/shmobile_defconfig

@@ -38,7 +38,7 @@ CONFIG_ZBOOT_ROM_BSS=0x0
 CONFIG_ARM_APPENDED_DTB=y
 CONFIG_KEXEC=y
 CONFIG_CPU_FREQ=y
-CONFIG_CPU_FREQ_STAT_DETAILS=y
+CONFIG_CPU_FREQ_STAT=y
 CONFIG_CPU_FREQ_GOV_POWERSAVE=y
 CONFIG_CPU_FREQ_GOV_USERSPACE=y
 CONFIG_CPU_FREQ_GOV_ONDEMAND=y

arch/arm/mach-omap2/pm.c

@@ -130,17 +130,16 @@ static int __init omap2_set_init_voltage(char *vdd_name, char *clk_name,
 	freq = clk_get_rate(clk);
 	clk_put(clk);
 
-	rcu_read_lock();
 	opp = dev_pm_opp_find_freq_ceil(dev, &freq);
 	if (IS_ERR(opp)) {
-		rcu_read_unlock();
 		pr_err("%s: unable to find boot up OPP for vdd_%s\n",
 			__func__, vdd_name);
 		goto exit;
 	}
 
 	bootup_volt = dev_pm_opp_get_voltage(opp);
-	rcu_read_unlock();
+	dev_pm_opp_put(opp);
+
 	if (!bootup_volt) {
 		pr_err("%s: unable to find voltage corresponding to the bootup OPP for vdd_%s\n",
 		       __func__, vdd_name);

arch/mips/Kconfig

@@ -1703,6 +1703,8 @@ config CPU_BMIPS
 	select WEAK_ORDERING
 	select CPU_SUPPORTS_HIGHMEM
 	select CPU_HAS_PREFETCH
+	select CPU_SUPPORTS_CPUFREQ
+	select MIPS_EXTERNAL_TIMER
 	help
 	  Support for BMIPS32/3300/4350/4380 and BMIPS5000 processors.
 

arch/mips/configs/bmips_stb_defconfig

@@ -9,13 +9,20 @@ CONFIG_MIPS_O32_FP64_SUPPORT=y
 # CONFIG_SWAP is not set
 CONFIG_NO_HZ=y
 CONFIG_BLK_DEV_INITRD=y
-CONFIG_RD_GZIP=y
 CONFIG_EXPERT=y
 # CONFIG_VM_EVENT_COUNTERS is not set
 # CONFIG_SLUB_DEBUG is not set
 # CONFIG_BLK_DEV_BSG is not set
 # CONFIG_IOSCHED_DEADLINE is not set
 # CONFIG_IOSCHED_CFQ is not set
+CONFIG_CPU_FREQ=y
+CONFIG_CPU_FREQ_STAT=y
+CONFIG_CPU_FREQ_GOV_POWERSAVE=y
+CONFIG_CPU_FREQ_GOV_USERSPACE=y
+CONFIG_CPU_FREQ_GOV_ONDEMAND=y
+CONFIG_CPU_FREQ_GOV_CONSERVATIVE=y
+CONFIG_CPU_FREQ_GOV_SCHEDUTIL=y
+CONFIG_BMIPS_CPUFREQ=y
 CONFIG_NET=y
 CONFIG_PACKET=y
 CONFIG_PACKET_DIAG=y
@@ -24,7 +31,6 @@ CONFIG_INET=y
 # CONFIG_INET_XFRM_MODE_TRANSPORT is not set
 # CONFIG_INET_XFRM_MODE_TUNNEL is not set
 # CONFIG_INET_XFRM_MODE_BEET is not set
-# CONFIG_INET_LRO is not set
 # CONFIG_INET_DIAG is not set
 CONFIG_CFG80211=y
 CONFIG_NL80211_TESTMODE=y
@@ -34,8 +40,6 @@ CONFIG_DEVTMPFS=y
 CONFIG_DEVTMPFS_MOUNT=y
 # CONFIG_STANDALONE is not set
 # CONFIG_PREVENT_FIRMWARE_BUILD is not set
-CONFIG_PRINTK_TIME=y
-CONFIG_BRCMSTB_GISB_ARB=y
 CONFIG_MTD=y
 CONFIG_MTD_CFI=y
 CONFIG_MTD_CFI_INTELEXT=y
@@ -51,16 +55,15 @@ CONFIG_USB_USBNET=y
 # CONFIG_INPUT is not set
 # CONFIG_SERIO is not set
 # CONFIG_VT is not set
-# CONFIG_DEVKMEM is not set
 CONFIG_SERIAL_8250=y
 # CONFIG_SERIAL_8250_DEPRECATED_OPTIONS is not set
 CONFIG_SERIAL_8250_CONSOLE=y
 CONFIG_SERIAL_OF_PLATFORM=y
 # CONFIG_HW_RANDOM is not set
-CONFIG_POWER_SUPPLY=y
 CONFIG_POWER_RESET=y
 CONFIG_POWER_RESET_BRCMSTB=y
 CONFIG_POWER_RESET_SYSCON=y
+CONFIG_POWER_SUPPLY=y
 # CONFIG_HWMON is not set
 CONFIG_USB=y
 CONFIG_USB_EHCI_HCD=y
@@ -82,6 +85,7 @@ CONFIG_CIFS=y
 CONFIG_NLS_CODEPAGE_437=y
 CONFIG_NLS_ASCII=y
 CONFIG_NLS_ISO8859_1=y
+CONFIG_PRINTK_TIME=y
 CONFIG_DEBUG_FS=y
 CONFIG_MAGIC_SYSRQ=y
 CONFIG_CMDLINE_BOOL=y

arch/mips/configs/lemote2f_defconfig

@@ -40,7 +40,6 @@ CONFIG_PM_STD_PARTITION="/dev/hda3"
 CONFIG_CPU_FREQ=y
 CONFIG_CPU_FREQ_DEBUG=y
 CONFIG_CPU_FREQ_STAT=m
-CONFIG_CPU_FREQ_STAT_DETAILS=y
 CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND=y
 CONFIG_CPU_FREQ_GOV_POWERSAVE=m
 CONFIG_CPU_FREQ_GOV_USERSPACE=m

arch/powerpc/configs/ppc6xx_defconfig

@@ -62,7 +62,6 @@ CONFIG_MPC8610_HPCD=y
 CONFIG_GEF_SBC610=y
 CONFIG_CPU_FREQ=y
 CONFIG_CPU_FREQ_STAT=m
-CONFIG_CPU_FREQ_STAT_DETAILS=y
 CONFIG_CPU_FREQ_DEFAULT_GOV_USERSPACE=y
 CONFIG_CPU_FREQ_GOV_PERFORMANCE=y
 CONFIG_CPU_FREQ_GOV_POWERSAVE=m

arch/sh/configs/sh7785lcr_32bit_defconfig

@@ -25,7 +25,7 @@ CONFIG_SH_SH7785LCR=y
 CONFIG_NO_HZ=y
 CONFIG_HIGH_RES_TIMERS=y
 CONFIG_CPU_FREQ=y
-CONFIG_CPU_FREQ_STAT_DETAILS=y
+CONFIG_CPU_FREQ_STAT=y
 CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND=y
 CONFIG_SH_CPU_FREQ=y
 CONFIG_HEARTBEAT=y

arch/x86/kernel/acpi/cstate.c

@@ -12,7 +12,6 @@
 #include <linux/sched.h>
 
 #include <acpi/processor.h>
-#include <asm/acpi.h>
 #include <asm/mwait.h>
 #include <asm/special_insns.h>
 
@@ -89,7 +88,8 @@ static long acpi_processor_ffh_cstate_probe_cpu(void *_cx)
 	retval = 0;
 	/* If the HW does not support any sub-states in this C-state */
 	if (num_cstate_subtype == 0) {
-		pr_warn(FW_BUG "ACPI MWAIT C-state 0x%x not supported by HW (0x%x)\n", cx->address, edx_part);
+		pr_warn(FW_BUG "ACPI MWAIT C-state 0x%x not supported by HW (0x%x)\n",
+				cx->address, edx_part);
 		retval = -1;
 		goto out;
 	}
@@ -104,8 +104,8 @@ static long acpi_processor_ffh_cstate_probe_cpu(void *_cx)
 	if (!mwait_supported[cstate_type]) {
 		mwait_supported[cstate_type] = 1;
 		printk(KERN_DEBUG
-			"Monitor-Mwait will be used to enter C-%d "
-			"state\n", cx->type);
+			"Monitor-Mwait will be used to enter C-%d state\n",
+			cx->type);
 	}
 	snprintf(cx->desc,
 			ACPI_CX_DESC_LEN, "ACPI FFH INTEL MWAIT 0x%x",
@@ -166,6 +166,7 @@ EXPORT_SYMBOL_GPL(acpi_processor_ffh_cstate_enter);
 static int __init ffh_cstate_init(void)
 {
 	struct cpuinfo_x86 *c = &boot_cpu_data;
+
 	if (c->x86_vendor != X86_VENDOR_INTEL)
 		return -1;
 

drivers/acpi/processor_perflib.c

@@ -75,10 +75,8 @@ static int acpi_processor_ppc_notifier(struct notifier_block *nb,
 	struct acpi_processor *pr;
 	unsigned int ppc = 0;
 
-	if (event == CPUFREQ_START && ignore_ppc <= 0) {
+	if (ignore_ppc < 0)
 		ignore_ppc = 0;
-		return 0;
-	}
 
 	if (ignore_ppc)
 		return 0;

drivers/base/cpu.c

@@ -17,6 +17,7 @@
 #include <linux/of.h>
 #include <linux/cpufeature.h>
 #include <linux/tick.h>
+#include <linux/pm_qos.h>
 
 #include "base.h"
 
@@ -376,6 +377,7 @@ int register_cpu(struct cpu *cpu, int num)
 
 	per_cpu(cpu_sys_devices, num) = &cpu->dev;
 	register_cpu_under_node(num, cpu_to_node(num));
+	dev_pm_qos_expose_latency_limit(&cpu->dev, 0);
 
 	return 0;
 }

drivers/base/power/domain.c

@@ -130,7 +130,7 @@ static inline bool irq_safe_dev_in_no_sleep_domain(struct device *dev,
 
 	ret = pm_runtime_is_irq_safe(dev) && !genpd_is_irq_safe(genpd);
 
-	/* Warn once for each IRQ safe dev in no sleep domain */
+	/* Warn once if IRQ safe dev in no sleep domain */
 	if (ret)
 		dev_warn_once(dev, "PM domain %s will not be powered off\n",
 				genpd->name);
@@ -201,7 +201,7 @@ static void genpd_sd_counter_inc(struct generic_pm_domain *genpd)
 	smp_mb__after_atomic();
 }
 
-static int genpd_power_on(struct generic_pm_domain *genpd, bool timed)
+static int _genpd_power_on(struct generic_pm_domain *genpd, bool timed)
 {
 	unsigned int state_idx = genpd->state_idx;
 	ktime_t time_start;
@@ -231,7 +231,7 @@ static int genpd_power_on(struct generic_pm_domain *genpd, bool timed)
 	return ret;
 }
 
-static int genpd_power_off(struct generic_pm_domain *genpd, bool timed)
+static int _genpd_power_off(struct generic_pm_domain *genpd, bool timed)
 {
 	unsigned int state_idx = genpd->state_idx;
 	ktime_t time_start;
@@ -262,10 +262,10 @@ static int genpd_power_off(struct generic_pm_domain *genpd, bool timed)
 }
 
 /**
- * genpd_queue_power_off_work - Queue up the execution of genpd_poweroff().
+ * genpd_queue_power_off_work - Queue up the execution of genpd_power_off().
  * @genpd: PM domain to power off.
  *
- * Queue up the execution of genpd_poweroff() unless it's already been done
+ * Queue up the execution of genpd_power_off() unless it's already been done
  * before.
  */
 static void genpd_queue_power_off_work(struct generic_pm_domain *genpd)
@@ -274,14 +274,14 @@ static void genpd_queue_power_off_work(struct generic_pm_domain *genpd)
 }
 
 /**
- * genpd_poweron - Restore power to a given PM domain and its masters.
+ * genpd_power_on - Restore power to a given PM domain and its masters.
  * @genpd: PM domain to power up.
  * @depth: nesting count for lockdep.
  *
  * Restore power to @genpd and all of its masters so that it is possible to
  * resume a device belonging to it.
  */
-static int genpd_poweron(struct generic_pm_domain *genpd, unsigned int depth)
+static int genpd_power_on(struct generic_pm_domain *genpd, unsigned int depth)
 {
 	struct gpd_link *link;
 	int ret = 0;
@@ -300,7 +300,7 @@ static int genpd_poweron(struct generic_pm_domain *genpd, unsigned int depth)
 		genpd_sd_counter_inc(master);
 
 		genpd_lock_nested(master, depth + 1);
-		ret = genpd_poweron(master, depth + 1);
+		ret = genpd_power_on(master, depth + 1);
 		genpd_unlock(master);
 
 		if (ret) {
@@ -309,7 +309,7 @@ static int genpd_poweron(struct generic_pm_domain *genpd, unsigned int depth)
 		}
 	}
 
-	ret = genpd_power_on(genpd, true);
+	ret = _genpd_power_on(genpd, true);
 	if (ret)
 		goto err;
 
@@ -368,14 +368,14 @@ static int genpd_dev_pm_qos_notifier(struct notifier_block *nb,
 }
 
 /**
- * genpd_poweroff - Remove power from a given PM domain.
+ * genpd_power_off - Remove power from a given PM domain.
  * @genpd: PM domain to power down.
  * @is_async: PM domain is powered down from a scheduled work
  *
  * If all of the @genpd's devices have been suspended and all of its subdomains
  * have been powered down, remove power from @genpd.
  */
-static int genpd_poweroff(struct generic_pm_domain *genpd, bool is_async)
+static int genpd_power_off(struct generic_pm_domain *genpd, bool is_async)
 {
 	struct pm_domain_data *pdd;
 	struct gpd_link *link;
@@ -427,13 +427,13 @@ static int genpd_poweroff(struct generic_pm_domain *genpd, bool is_async)
 
 		/*
 		 * If sd_count > 0 at this point, one of the subdomains hasn't
-		 * managed to call genpd_poweron() for the master yet after
-		 * incrementing it.  In that case genpd_poweron() will wait
+		 * managed to call genpd_power_on() for the master yet after
+		 * incrementing it.  In that case genpd_power_on() will wait
 		 * for us to drop the lock, so we can call .power_off() and let
-		 * the genpd_poweron() restore power for us (this shouldn't
+		 * the genpd_power_on() restore power for us (this shouldn't
 		 * happen very often).
 		 */
-		ret = genpd_power_off(genpd, true);
+		ret = _genpd_power_off(genpd, true);
 		if (ret)
 			return ret;
 	}
@@ -459,7 +459,7 @@ static void genpd_power_off_work_fn(struct work_struct *work)
 	genpd = container_of(work, struct generic_pm_domain, power_off_work);
 
 	genpd_lock(genpd);
-	genpd_poweroff(genpd, true);
+	genpd_power_off(genpd, true);
 	genpd_unlock(genpd);
 }
 
@@ -578,7 +578,7 @@ static int genpd_runtime_suspend(struct device *dev)
 		return 0;
 
 	genpd_lock(genpd);
-	genpd_poweroff(genpd, false);
+	genpd_power_off(genpd, false);
 	genpd_unlock(genpd);
 
 	return 0;
@@ -618,7 +618,7 @@ static int genpd_runtime_resume(struct device *dev)
 	}
 
 	genpd_lock(genpd);
-	ret = genpd_poweron(genpd, 0);
+	ret = genpd_power_on(genpd, 0);
 	genpd_unlock(genpd);
 
 	if (ret)
@@ -658,7 +658,7 @@ static int genpd_runtime_resume(struct device *dev)
 	if (!pm_runtime_is_irq_safe(dev) ||
 		(pm_runtime_is_irq_safe(dev) && genpd_is_irq_safe(genpd))) {
 		genpd_lock(genpd);
-		genpd_poweroff(genpd, 0);
+		genpd_power_off(genpd, 0);
 		genpd_unlock(genpd);
 	}
 
@@ -674,9 +674,9 @@ static int __init pd_ignore_unused_setup(char *__unused)
 __setup("pd_ignore_unused", pd_ignore_unused_setup);
 
 /**
- * genpd_poweroff_unused - Power off all PM domains with no devices in use.
+ * genpd_power_off_unused - Power off all PM domains with no devices in use.
  */
-static int __init genpd_poweroff_unused(void)
+static int __init genpd_power_off_unused(void)
 {
 	struct generic_pm_domain *genpd;
 
@@ -694,7 +694,7 @@ static int __init genpd_poweroff_unused(void)
 
 	return 0;
 }
-late_initcall(genpd_poweroff_unused);
+late_initcall(genpd_power_off_unused);
 
 #if defined(CONFIG_PM_SLEEP) || defined(CONFIG_PM_GENERIC_DOMAINS_OF)
 
@@ -727,18 +727,20 @@ static bool genpd_dev_active_wakeup(struct generic_pm_domain *genpd,
 }
 
 /**
- * genpd_sync_poweroff - Synchronously power off a PM domain and its masters.
+ * genpd_sync_power_off - Synchronously power off a PM domain and its masters.
  * @genpd: PM domain to power off, if possible.
+ * @use_lock: use the lock.
+ * @depth: nesting count for lockdep.
  *
  * Check if the given PM domain can be powered off (during system suspend or
  * hibernation) and do that if so.  Also, in that case propagate to its masters.
  *
  * This function is only called in "noirq" and "syscore" stages of system power
- * transitions, so it need not acquire locks (all of the "noirq" callbacks are
- * executed sequentially, so it is guaranteed that it will never run twice in
- * parallel).
+ * transitions. The "noirq" callbacks may be executed asynchronously, thus in
+ * these cases the lock must be held.
  */
-static void genpd_sync_poweroff(struct generic_pm_domain *genpd)
+static void genpd_sync_power_off(struct generic_pm_domain *genpd, bool use_lock,
+				 unsigned int depth)
 {
 	struct gpd_link *link;
 
@@ -751,26 +753,35 @@ static void genpd_sync_poweroff(struct generic_pm_domain *genpd)
 
 	/* Choose the deepest state when suspending */
 	genpd->state_idx = genpd->state_count - 1;
-	genpd_power_off(genpd, false);
+	_genpd_power_off(genpd, false);
 
 	genpd->status = GPD_STATE_POWER_OFF;
 
 	list_for_each_entry(link, &genpd->slave_links, slave_node) {
 		genpd_sd_counter_dec(link->master);
-		genpd_sync_poweroff(link->master);
+
+		if (use_lock)
+			genpd_lock_nested(link->master, depth + 1);
+
+		genpd_sync_power_off(link->master, use_lock, depth + 1);
+
+		if (use_lock)
+			genpd_unlock(link->master);
 	}
 }
 
 /**
- * genpd_sync_poweron - Synchronously power on a PM domain and its masters.
+ * genpd_sync_power_on - Synchronously power on a PM domain and its masters.
  * @genpd: PM domain to power on.
+ * @use_lock: use the lock.
+ * @depth: nesting count for lockdep.
  *
  * This function is only called in "noirq" and "syscore" stages of system power
- * transitions, so it need not acquire locks (all of the "noirq" callbacks are
- * executed sequentially, so it is guaranteed that it will never run twice in
- * parallel).
+ * transitions. The "noirq" callbacks may be executed asynchronously, thus in
+ * these cases the lock must be held.
  */
-static void genpd_sync_poweron(struct generic_pm_domain *genpd)
+static void genpd_sync_power_on(struct generic_pm_domain *genpd, bool use_lock,
+				unsigned int depth)
 {
 	struct gpd_link *link;
 
@@ -778,11 +789,18 @@ static void genpd_sync_poweron(struct generic_pm_domain *genpd)
 		return;
 
 	list_for_each_entry(link, &genpd->slave_links, slave_node) {
-		genpd_sync_poweron(link->master);
 		genpd_sd_counter_inc(link->master);
+
+		if (use_lock)
+			genpd_lock_nested(link->master, depth + 1);
+
+		genpd_sync_power_on(link->master, use_lock, depth + 1);
+
+		if (use_lock)
+			genpd_unlock(link->master);
 	}
 
-	genpd_power_on(genpd, false);
+	_genpd_power_on(genpd, false);
 
 	genpd->status = GPD_STATE_ACTIVE;
 }
@@ -888,13 +906,10 @@ static int pm_genpd_suspend_noirq(struct device *dev)
 			return ret;
 	}
 
-	/*
-	 * Since all of the "noirq" callbacks are executed sequentially, it is
-	 * guaranteed that this function will never run twice in parallel for
-	 * the same PM domain, so it is not necessary to use locking here.
-	 */
+	genpd_lock(genpd);
 	genpd->suspended_count++;
-	genpd_sync_poweroff(genpd);
+	genpd_sync_power_off(genpd, true, 0);
+	genpd_unlock(genpd);
 
 	return 0;
 }
@@ -919,13 +934,10 @@ static int pm_genpd_resume_noirq(struct device *dev)
 	if (dev->power.wakeup_path && genpd_dev_active_wakeup(genpd, dev))
 		return 0;
 
-	/*
-	 * Since all of the "noirq" callbacks are executed sequentially, it is
-	 * guaranteed that this function will never run twice in parallel for
-	 * the same PM domain, so it is not necessary to use locking here.
-	 */
-	genpd_sync_poweron(genpd);
+	genpd_lock(genpd);
+	genpd_sync_power_on(genpd, true, 0);
 	genpd->suspended_count--;
+	genpd_unlock(genpd);
 
 	if (genpd->dev_ops.stop && genpd->dev_ops.start)
 		ret = pm_runtime_force_resume(dev);
@@ -1002,22 +1014,20 @@ static int pm_genpd_restore_noirq(struct device *dev)
 		return -EINVAL;
 
 	/*
-	 * Since all of the "noirq" callbacks are executed sequentially, it is
-	 * guaranteed that this function will never run twice in parallel for
-	 * the same PM domain, so it is not necessary to use locking here.
-	 *
 	 * At this point suspended_count == 0 means we are being run for the
 	 * first time for the given domain in the present cycle.
 	 */
+	genpd_lock(genpd);
 	if (genpd->suspended_count++ == 0)
 		/*
 		 * The boot kernel might put the domain into arbitrary state,
-		 * so make it appear as powered off to genpd_sync_poweron(),
+		 * so make it appear as powered off to genpd_sync_power_on(),
 		 * so that it tries to power it on in case it was really off.
 		 */
 		genpd->status = GPD_STATE_POWER_OFF;
 
-	genpd_sync_poweron(genpd);
+	genpd_sync_power_on(genpd, true, 0);
+	genpd_unlock(genpd);
 
 	if (genpd->dev_ops.stop && genpd->dev_ops.start)
 		ret = pm_runtime_force_resume(dev);
@@ -1072,9 +1082,9 @@ static void genpd_syscore_switch(struct device *dev, bool suspend)
 
 	if (suspend) {
 		genpd->suspended_count++;
-		genpd_sync_poweroff(genpd);
+		genpd_sync_power_off(genpd, false, 0);
 	} else {
-		genpd_sync_poweron(genpd);
+		genpd_sync_power_on(genpd, false, 0);
 		genpd->suspended_count--;
 	}
 }
@@ -2043,7 +2053,7 @@ int genpd_dev_pm_attach(struct device *dev)
 	dev->pm_domain->sync = genpd_dev_pm_sync;
 
 	genpd_lock(pd);
-	ret = genpd_poweron(pd, 0);
+	ret = genpd_power_on(pd, 0);
 	genpd_unlock(pd);
 out:
 	return ret ? -EPROBE_DEFER : 0;

drivers/base/power/opp/core.c

@@ -32,13 +32,7 @@ LIST_HEAD(opp_tables);
 /* Lock to allow exclusive modification to the device and opp lists */
 DEFINE_MUTEX(opp_table_lock);
 
-#define opp_rcu_lockdep_assert()					\
-do {									\
-	RCU_LOCKDEP_WARN(!rcu_read_lock_held() &&			\
-			 !lockdep_is_held(&opp_table_lock),		\
-			 "Missing rcu_read_lock() or "			\
-			 "opp_table_lock protection");			\
-} while (0)
+static void dev_pm_opp_get(struct dev_pm_opp *opp);
 
 static struct opp_device *_find_opp_dev(const struct device *dev,
 					struct opp_table *opp_table)
@@ -52,38 +46,46 @@ static struct opp_device *_find_opp_dev(const struct device *dev,
 	return NULL;
 }
 
+static struct opp_table *_find_opp_table_unlocked(struct device *dev)
+{
+	struct opp_table *opp_table;
+
+	list_for_each_entry(opp_table, &opp_tables, node) {
+		if (_find_opp_dev(dev, opp_table)) {
+			_get_opp_table_kref(opp_table);
+
+			return opp_table;
+		}
+	}
+
+	return ERR_PTR(-ENODEV);
+}
+
 /**
  * _find_opp_table() - find opp_table struct using device pointer
  * @dev:	device pointer used to lookup OPP table
  *
- * Search OPP table for one containing matching device. Does a RCU reader
- * operation to grab the pointer needed.
+ * Search OPP table for one containing matching device.
  *
  * Return: pointer to 'struct opp_table' if found, otherwise -ENODEV or
  * -EINVAL based on type of error.
  *
- * Locking: For readers, this function must be called under rcu_read_lock().
- * opp_table is a RCU protected pointer, which means that opp_table is valid
- * as long as we are under RCU lock.
- *
- * For Writers, this function must be called with opp_table_lock held.
+ * The callers must call dev_pm_opp_put_opp_table() after the table is used.
  */
 struct opp_table *_find_opp_table(struct device *dev)
 {
 	struct opp_table *opp_table;
 
-	opp_rcu_lockdep_assert();
-
 	if (IS_ERR_OR_NULL(dev)) {
 		pr_err("%s: Invalid parameters\n", __func__);
 		return ERR_PTR(-EINVAL);
 	}
 
-	list_for_each_entry_rcu(opp_table, &opp_tables, node)
-		if (_find_opp_dev(dev, opp_table))
-			return opp_table;
+	mutex_lock(&opp_table_lock);
+	opp_table = _find_opp_table_unlocked(dev);
+	mutex_unlock(&opp_table_lock);
 
-	return ERR_PTR(-ENODEV);
+	return opp_table;
 }
 
 /**
@@ -94,29 +96,15 @@ struct opp_table *_find_opp_table(struct device *dev)
  * return 0
  *
  * This is useful only for devices with single power supply.
- *
- * Locking: This function must be called under rcu_read_lock(). opp is a rcu
- * protected pointer. This means that opp which could have been fetched by
- * opp_find_freq_{exact,ceil,floor} functions is valid as long as we are
- * under RCU lock. The pointer returned by the opp_find_freq family must be
- * used in the same section as the usage of this function with the pointer
- * prior to unlocking with rcu_read_unlock() to maintain the integrity of the
- * pointer.
  */
 unsigned long dev_pm_opp_get_voltage(struct dev_pm_opp *opp)
 {
-	struct dev_pm_opp *tmp_opp;
-	unsigned long v = 0;
-
-	opp_rcu_lockdep_assert();
-
-	tmp_opp = rcu_dereference(opp);
-	if (IS_ERR_OR_NULL(tmp_opp))
+	if (IS_ERR_OR_NULL(opp)) {
 		pr_err("%s: Invalid parameters\n", __func__);
-	else
-		v = tmp_opp->supplies[0].u_volt;
+		return 0;
+	}
 
-	return v;
+	return opp->supplies[0].u_volt;
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_get_voltage);
 
@@ -126,29 +114,15 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_get_voltage);
  *
  * Return: frequency in hertz corresponding to the opp, else
  * return 0
- *
- * Locking: This function must be called under rcu_read_lock(). opp is a rcu
- * protected pointer. This means that opp which could have been fetched by
- * opp_find_freq_{exact,ceil,floor} functions is valid as long as we are
- * under RCU lock. The pointer returned by the opp_find_freq family must be
- * used in the same section as the usage of this function with the pointer
- * prior to unlocking with rcu_read_unlock() to maintain the integrity of the
- * pointer.
  */
 unsigned long dev_pm_opp_get_freq(struct dev_pm_opp *opp)
 {
-	struct dev_pm_opp *tmp_opp;
-	unsigned long f = 0;
-
-	opp_rcu_lockdep_assert();
-
-	tmp_opp = rcu_dereference(opp);
-	if (IS_ERR_OR_NULL(tmp_opp) || !tmp_opp->available)
+	if (IS_ERR_OR_NULL(opp) || !opp->available) {
 		pr_err("%s: Invalid parameters\n", __func__);
-	else
-		f = tmp_opp->rate;
+		return 0;
+	}
 
-	return f;
+	return opp->rate;
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_get_freq);
 
@@ -161,28 +135,15 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_get_freq);
  * quickly. Running on them for longer times may overheat the chip.
  *
  * Return: true if opp is turbo opp, else false.
- *
- * Locking: This function must be called under rcu_read_lock(). opp is a rcu
- * protected pointer. This means that opp which could have been fetched by
- * opp_find_freq_{exact,ceil,floor} functions is valid as long as we are
- * under RCU lock. The pointer returned by the opp_find_freq family must be
- * used in the same section as the usage of this function with the pointer
- * prior to unlocking with rcu_read_unlock() to maintain the integrity of the
- * pointer.
  */
 bool dev_pm_opp_is_turbo(struct dev_pm_opp *opp)
 {
-	struct dev_pm_opp *tmp_opp;
-
-	opp_rcu_lockdep_assert();
-
-	tmp_opp = rcu_dereference(opp);
-	if (IS_ERR_OR_NULL(tmp_opp) || !tmp_opp->available) {
+	if (IS_ERR_OR_NULL(opp) || !opp->available) {
 		pr_err("%s: Invalid parameters\n", __func__);
 		return false;
 	}
 
-	return tmp_opp->turbo;
+	return opp->turbo;
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_is_turbo);
 
@@ -191,52 +152,29 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_is_turbo);
  * @dev:	device for which we do this operation
  *
  * Return: This function returns the max clock latency in nanoseconds.
- *
- * Locking: This function takes rcu_read_lock().
  */
 unsigned long dev_pm_opp_get_max_clock_latency(struct device *dev)
 {
 	struct opp_table *opp_table;
 	unsigned long clock_latency_ns;
 
-	rcu_read_lock();
-
 	opp_table = _find_opp_table(dev);
 	if (IS_ERR(opp_table))
-		clock_latency_ns = 0;
-	else
+		return 0;
+
 	clock_latency_ns = opp_table->clock_latency_ns_max;
 
-	rcu_read_unlock();
+	dev_pm_opp_put_opp_table(opp_table);
+
 	return clock_latency_ns;
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_get_max_clock_latency);
 
-static int _get_regulator_count(struct device *dev)
-{
-	struct opp_table *opp_table;
-	int count;
-
-	rcu_read_lock();
-
-	opp_table = _find_opp_table(dev);
-	if (!IS_ERR(opp_table))
-		count = opp_table->regulator_count;
-	else
-		count = 0;
-
-	rcu_read_unlock();
-
-	return count;
-}
-
 /**
  * dev_pm_opp_get_max_volt_latency() - Get max voltage latency in nanoseconds
  * @dev: device for which we do this operation
  *
  * Return: This function returns the max voltage latency in nanoseconds.
- *
- * Locking: This function takes rcu_read_lock().
  */
 unsigned long dev_pm_opp_get_max_volt_latency(struct device *dev)
 {
@@ -250,35 +188,33 @@ unsigned long dev_pm_opp_get_max_volt_latency(struct device *dev)
 		unsigned long max;
 	} *uV;
 
-	count = _get_regulator_count(dev);
+	opp_table = _find_opp_table(dev);
+	if (IS_ERR(opp_table))
+		return 0;
+
+	count = opp_table->regulator_count;
 
 	/* Regulator may not be required for the device */
 	if (!count)
-		return 0;
+		goto put_opp_table;
 
 	regulators = kmalloc_array(count, sizeof(*regulators), GFP_KERNEL);
 	if (!regulators)
-		return 0;
+		goto put_opp_table;
 
 	uV = kmalloc_array(count, sizeof(*uV), GFP_KERNEL);
 	if (!uV)
 		goto free_regulators;
 
-	rcu_read_lock();
-
-	opp_table = _find_opp_table(dev);
-	if (IS_ERR(opp_table)) {
-		rcu_read_unlock();
-		goto free_uV;
-	}
-
 	memcpy(regulators, opp_table->regulators, count * sizeof(*regulators));
 
+	mutex_lock(&opp_table->lock);
+
 	for (i = 0; i < count; i++) {
 		uV[i].min = ~0;
 		uV[i].max = 0;
 
-		list_for_each_entry_rcu(opp, &opp_table->opp_list, node) {
+		list_for_each_entry(opp, &opp_table->opp_list, node) {
 			if (!opp->available)
 				continue;
 
@@ -289,7 +225,7 @@ unsigned long dev_pm_opp_get_max_volt_latency(struct device *dev)
 		}
 	}
 
-	rcu_read_unlock();
+	mutex_unlock(&opp_table->lock);
 
 	/*
 	 * The caller needs to ensure that opp_table (and hence the regulator)
@@ -301,10 +237,11 @@ unsigned long dev_pm_opp_get_max_volt_latency(struct device *dev)
 			latency_ns += ret * 1000;
 	}
 
-free_uV:
 	kfree(uV);
 free_regulators:
 	kfree(regulators);
+put_opp_table:
+	dev_pm_opp_put_opp_table(opp_table);
 
 	return latency_ns;
 }
@@ -317,8 +254,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_get_max_volt_latency);
  *
  * Return: This function returns the max transition latency, in nanoseconds, to
  * switch from one OPP to other.
- *
- * Locking: This function takes rcu_read_lock().
  */
 unsigned long dev_pm_opp_get_max_transition_latency(struct device *dev)
 {
@@ -328,32 +263,29 @@ unsigned long dev_pm_opp_get_max_transition_latency(struct device *dev)
 EXPORT_SYMBOL_GPL(dev_pm_opp_get_max_transition_latency);
 
 /**
- * dev_pm_opp_get_suspend_opp() - Get suspend opp
+ * dev_pm_opp_get_suspend_opp_freq() - Get frequency of suspend opp in Hz
  * @dev:	device for which we do this operation
  *
- * Return: This function returns pointer to the suspend opp if it is
- * defined and available, otherwise it returns NULL.
- *
- * Locking: This function must be called under rcu_read_lock(). opp is a rcu
- * protected pointer. The reason for the same is that the opp pointer which is
- * returned will remain valid for use with opp_get_{voltage, freq} only while
- * under the locked area. The pointer returned must be used prior to unlocking
- * with rcu_read_unlock() to maintain the integrity of the pointer.
+ * Return: This function returns the frequency of the OPP marked as suspend_opp
+ * if one is available, else returns 0;
  */
-struct dev_pm_opp *dev_pm_opp_get_suspend_opp(struct device *dev)
+unsigned long dev_pm_opp_get_suspend_opp_freq(struct device *dev)
 {
 	struct opp_table *opp_table;
-
-	opp_rcu_lockdep_assert();
+	unsigned long freq = 0;
 
 	opp_table = _find_opp_table(dev);
-	if (IS_ERR(opp_table) || !opp_table->suspend_opp ||
-	    !opp_table->suspend_opp->available)
-		return NULL;
+	if (IS_ERR(opp_table))
+		return 0;
+
+	if (opp_table->suspend_opp && opp_table->suspend_opp->available)
+		freq = dev_pm_opp_get_freq(opp_table->suspend_opp);
+
+	dev_pm_opp_put_opp_table(opp_table);
 
-	return opp_table->suspend_opp;
+	return freq;
 }
-EXPORT_SYMBOL_GPL(dev_pm_opp_get_suspend_opp);
+EXPORT_SYMBOL_GPL(dev_pm_opp_get_suspend_opp_freq);
 
 /**
  * dev_pm_opp_get_opp_count() - Get number of opps available in the opp table
@@ -361,8 +293,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_get_suspend_opp);
  *
  * Return: This function returns the number of available opps if there are any,
  * else returns 0 if none or the corresponding error value.
- *
- * Locking: This function takes rcu_read_lock().
  */
 int dev_pm_opp_get_opp_count(struct device *dev)
 {
@@ -370,23 +300,24 @@ int dev_pm_opp_get_opp_count(struct device *dev)
 	struct dev_pm_opp *temp_opp;
 	int count = 0;
 
-	rcu_read_lock();
-
 	opp_table = _find_opp_table(dev);
 	if (IS_ERR(opp_table)) {
 		count = PTR_ERR(opp_table);
 		dev_err(dev, "%s: OPP table not found (%d)\n",
 			__func__, count);
-		goto out_unlock;
+		return count;
 	}
 
-	list_for_each_entry_rcu(temp_opp, &opp_table->opp_list, node) {
+	mutex_lock(&opp_table->lock);
+
+	list_for_each_entry(temp_opp, &opp_table->opp_list, node) {
 		if (temp_opp->available)
 			count++;
 	}
 
-out_unlock:
-	rcu_read_unlock();
+	mutex_unlock(&opp_table->lock);
+	dev_pm_opp_put_opp_table(opp_table);
+
 	return count;
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_get_opp_count);
@@ -411,11 +342,8 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_get_opp_count);
  * This provides a mechanism to enable an opp which is not available currently
  * or the opposite as well.
  *
- * Locking: This function must be called under rcu_read_lock(). opp is a rcu
- * protected pointer. The reason for the same is that the opp pointer which is
- * returned will remain valid for use with opp_get_{voltage, freq} only while
- * under the locked area. The pointer returned must be used prior to unlocking
- * with rcu_read_unlock() to maintain the integrity of the pointer.
+ * The callers are required to call dev_pm_opp_put() for the returned OPP after
+ * use.
  */
 struct dev_pm_opp *dev_pm_opp_find_freq_exact(struct device *dev,
 					      unsigned long freq,
@@ -424,8 +352,6 @@ struct dev_pm_opp *dev_pm_opp_find_freq_exact(struct device *dev,
 	struct opp_table *opp_table;
 	struct dev_pm_opp *temp_opp, *opp = ERR_PTR(-ERANGE);
 
-	opp_rcu_lockdep_assert();
-
 	opp_table = _find_opp_table(dev);
 	if (IS_ERR(opp_table)) {
 		int r = PTR_ERR(opp_table);
@@ -434,14 +360,22 @@ struct dev_pm_opp *dev_pm_opp_find_freq_exact(struct device *dev,
 		return ERR_PTR(r);
 	}
 
-	list_for_each_entry_rcu(temp_opp, &opp_table->opp_list, node) {
+	mutex_lock(&opp_table->lock);
+
+	list_for_each_entry(temp_opp, &opp_table->opp_list, node) {
 		if (temp_opp->available == available &&
 				temp_opp->rate == freq) {
 			opp = temp_opp;
+
+			/* Increment the reference count of OPP */
+			dev_pm_opp_get(opp);
 			break;
 		}
 	}
 
+	mutex_unlock(&opp_table->lock);
+	dev_pm_opp_put_opp_table(opp_table);
+
 	return opp;
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_find_freq_exact);
@@ -451,14 +385,21 @@ static noinline struct dev_pm_opp *_find_freq_ceil(struct opp_table *opp_table,
 {
 	struct dev_pm_opp *temp_opp, *opp = ERR_PTR(-ERANGE);
 
-	list_for_each_entry_rcu(temp_opp, &opp_table->opp_list, node) {
+	mutex_lock(&opp_table->lock);
+
+	list_for_each_entry(temp_opp, &opp_table->opp_list, node) {
 		if (temp_opp->available && temp_opp->rate >= *freq) {
 			opp = temp_opp;
 			*freq = opp->rate;
+
+			/* Increment the reference count of OPP */
+			dev_pm_opp_get(opp);
 			break;
 		}
 	}
 
+	mutex_unlock(&opp_table->lock);
+
 	return opp;
 }
 
@@ -477,18 +418,14 @@ static noinline struct dev_pm_opp *_find_freq_ceil(struct opp_table *opp_table,
  * ERANGE:	no match found for search
  * ENODEV:	if device not found in list of registered devices
  *
- * Locking: This function must be called under rcu_read_lock(). opp is a rcu
- * protected pointer. The reason for the same is that the opp pointer which is
- * returned will remain valid for use with opp_get_{voltage, freq} only while
- * under the locked area. The pointer returned must be used prior to unlocking
- * with rcu_read_unlock() to maintain the integrity of the pointer.
+ * The callers are required to call dev_pm_opp_put() for the returned OPP after
+ * use.
  */
 struct dev_pm_opp *dev_pm_opp_find_freq_ceil(struct device *dev,
 					     unsigned long *freq)
 {
 	struct opp_table *opp_table;
-
-	opp_rcu_lockdep_assert();
+	struct dev_pm_opp *opp;
 
 	if (!dev || !freq) {
 		dev_err(dev, "%s: Invalid argument freq=%p\n", __func__, freq);
@@ -499,7 +436,11 @@ struct dev_pm_opp *dev_pm_opp_find_freq_ceil(struct device *dev,
 	if (IS_ERR(opp_table))
 		return ERR_CAST(opp_table);
 
-	return _find_freq_ceil(opp_table, freq);
+	opp = _find_freq_ceil(opp_table, freq);
+
+	dev_pm_opp_put_opp_table(opp_table);
+
+	return opp;
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_find_freq_ceil);
 
@@ -518,11 +459,8 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_find_freq_ceil);
  * ERANGE:	no match found for search
  * ENODEV:	if device not found in list of registered devices
  *
- * Locking: This function must be called under rcu_read_lock(). opp is a rcu
- * protected pointer. The reason for the same is that the opp pointer which is
- * returned will remain valid for use with opp_get_{voltage, freq} only while
- * under the locked area. The pointer returned must be used prior to unlocking
- * with rcu_read_unlock() to maintain the integrity of the pointer.
+ * The callers are required to call dev_pm_opp_put() for the returned OPP after
+ * use.
  */
 struct dev_pm_opp *dev_pm_opp_find_freq_floor(struct device *dev,
 					      unsigned long *freq)
@@ -530,8 +468,6 @@ struct dev_pm_opp *dev_pm_opp_find_freq_floor(struct device *dev,
 	struct opp_table *opp_table;
 	struct dev_pm_opp *temp_opp, *opp = ERR_PTR(-ERANGE);
 
-	opp_rcu_lockdep_assert();
-
 	if (!dev || !freq) {
 		dev_err(dev, "%s: Invalid argument freq=%p\n", __func__, freq);
 		return ERR_PTR(-EINVAL);
@@ -541,7 +477,9 @@ struct dev_pm_opp *dev_pm_opp_find_freq_floor(struct device *dev,
 	if (IS_ERR(opp_table))
 		return ERR_CAST(opp_table);
 
-	list_for_each_entry_rcu(temp_opp, &opp_table->opp_list, node) {
+	mutex_lock(&opp_table->lock);
+
+	list_for_each_entry(temp_opp, &opp_table->opp_list, node) {
 		if (temp_opp->available) {
 			/* go to the next node, before choosing prev */
 			if (temp_opp->rate > *freq)
@@ -550,6 +488,13 @@ struct dev_pm_opp *dev_pm_opp_find_freq_floor(struct device *dev,
 				opp = temp_opp;
 		}
 	}
+
+	/* Increment the reference count of OPP */
+	if (!IS_ERR(opp))
+		dev_pm_opp_get(opp);
+	mutex_unlock(&opp_table->lock);
+	dev_pm_opp_put_opp_table(opp_table);
+
 	if (!IS_ERR(opp))
 		*freq = opp->rate;
 
@@ -557,34 +502,6 @@ struct dev_pm_opp *dev_pm_opp_find_freq_floor(struct device *dev,
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_find_freq_floor);
 
-/*
- * The caller needs to ensure that opp_table (and hence the clk) isn't freed,
- * while clk returned here is used.
- */
-static struct clk *_get_opp_clk(struct device *dev)
-{
-	struct opp_table *opp_table;
-	struct clk *clk;
-
-	rcu_read_lock();
-
-	opp_table = _find_opp_table(dev);
-	if (IS_ERR(opp_table)) {
-		dev_err(dev, "%s: device opp doesn't exist\n", __func__);
-		clk = ERR_CAST(opp_table);
-		goto unlock;
-	}
-
-	clk = opp_table->clk;
-	if (IS_ERR(clk))
-		dev_err(dev, "%s: No clock available for the device\n",
-			__func__);
-
-unlock:
-	rcu_read_unlock();
-	return clk;
-}
-
 static int _set_opp_voltage(struct device *dev, struct regulator *reg,
 			    struct dev_pm_opp_supply *supply)
 {
@@ -680,8 +597,6 @@ static int _generic_set_opp(struct dev_pm_set_opp_data *data)
  *
  * This configures the power-supplies and clock source to the levels specified
  * by the OPP corresponding to the target_freq.
- *
- * Locking: This function takes rcu_read_lock().
  */
 int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq)
 {
@@ -700,9 +615,19 @@ int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq)
 		return -EINVAL;
 	}
 
-	clk = _get_opp_clk(dev);
-	if (IS_ERR(clk))
-		return PTR_ERR(clk);
+	opp_table = _find_opp_table(dev);
+	if (IS_ERR(opp_table)) {
+		dev_err(dev, "%s: device opp doesn't exist\n", __func__);
+		return PTR_ERR(opp_table);
+	}
+
+	clk = opp_table->clk;
+	if (IS_ERR(clk)) {
+		dev_err(dev, "%s: No clock available for the device\n",
+			__func__);
+		ret = PTR_ERR(clk);
+		goto put_opp_table;
+	}
 
 	freq = clk_round_rate(clk, target_freq);
 	if ((long)freq <= 0)
@@ -714,16 +639,8 @@ int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq)
 	if (old_freq == freq) {
 		dev_dbg(dev, "%s: old/new frequencies (%lu Hz) are same, nothing to do\n",
 			__func__, freq);
-		return 0;
-	}
-
-	rcu_read_lock();
-
-	opp_table = _find_opp_table(dev);
-	if (IS_ERR(opp_table)) {
-		dev_err(dev, "%s: device opp doesn't exist\n", __func__);
-		rcu_read_unlock();
-		return PTR_ERR(opp_table);
+		ret = 0;
+		goto put_opp_table;
 	}
 
 	old_opp = _find_freq_ceil(opp_table, &old_freq);
@@ -737,8 +654,7 @@ int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq)
 		ret = PTR_ERR(opp);
 		dev_err(dev, "%s: failed to find OPP for freq %lu (%d)\n",
 			__func__, freq, ret);
-		rcu_read_unlock();
-		return ret;
+		goto put_old_opp;
 	}
 
 	dev_dbg(dev, "%s: switching OPP: %lu Hz --> %lu Hz\n", __func__,
@@ -748,8 +664,8 @@ int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq)
 
 	/* Only frequency scaling */
 	if (!regulators) {
-		rcu_read_unlock();
-		return _generic_set_opp_clk_only(dev, clk, old_freq, freq);
+		ret = _generic_set_opp_clk_only(dev, clk, old_freq, freq);
+		goto put_opps;
 	}
 
 	if (opp_table->set_opp)
@@ -773,28 +689,26 @@ int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq)
 	data->new_opp.rate = freq;
 	memcpy(data->new_opp.supplies, opp->supplies, size);
 
-	rcu_read_unlock();
+	ret = set_opp(data);
 
-	return set_opp(data);
+put_opps:
+	dev_pm_opp_put(opp);
+put_old_opp:
+	if (!IS_ERR(old_opp))
+		dev_pm_opp_put(old_opp);
+put_opp_table:
+	dev_pm_opp_put_opp_table(opp_table);
+	return ret;
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_set_rate);
 
 /* OPP-dev Helpers */
-static void _kfree_opp_dev_rcu(struct rcu_head *head)
-{
-	struct opp_device *opp_dev;
-
-	opp_dev = container_of(head, struct opp_device, rcu_head);
-	kfree_rcu(opp_dev, rcu_head);
-}
-
 static void _remove_opp_dev(struct opp_device *opp_dev,
 			    struct opp_table *opp_table)
 {
 	opp_debug_unregister(opp_dev, opp_table);
 	list_del(&opp_dev->node);
-	call_srcu(&opp_table->srcu_head.srcu, &opp_dev->rcu_head,
-		  _kfree_opp_dev_rcu);
+	kfree(opp_dev);
 }
 
 struct opp_device *_add_opp_dev(const struct device *dev,
@@ -809,7 +723,7 @@ struct opp_device *_add_opp_dev(const struct device *dev,
 
 	/* Initialize opp-dev */
 	opp_dev->dev = dev;
-	list_add_rcu(&opp_dev->node, &opp_table->dev_list);
+	list_add(&opp_dev->node, &opp_table->dev_list);
 
 	/* Create debugfs entries for the opp_table */
 	ret = opp_debug_register(opp_dev, opp_table);
@@ -820,26 +734,12 @@ struct opp_device *_add_opp_dev(const struct device *dev,
 	return opp_dev;
 }
 
-/**
- * _add_opp_table() - Find OPP table or allocate a new one
- * @dev:	device for which we do this operation
- *
- * It tries to find an existing table first, if it couldn't find one, it
- * allocates a new OPP table and returns that.
- *
- * Return: valid opp_table pointer if success, else NULL.
- */
-static struct opp_table *_add_opp_table(struct device *dev)
+static struct opp_table *_allocate_opp_table(struct device *dev)
 {
 	struct opp_table *opp_table;
 	struct opp_device *opp_dev;
 	int ret;
 
-	/* Check for existing table for 'dev' first */
-	opp_table = _find_opp_table(dev);
-	if (!IS_ERR(opp_table))
-		return opp_table;
-
 	/*
 	 * Allocate a new OPP table. In the infrequent case where a new
 	 * device is needed to be added, we pay this penalty.
@@ -867,50 +767,45 @@ static struct opp_table *_add_opp_table(struct device *dev)
 				ret);
 	}
 
-	srcu_init_notifier_head(&opp_table->srcu_head);
+	BLOCKING_INIT_NOTIFIER_HEAD(&opp_table->head);
 	INIT_LIST_HEAD(&opp_table->opp_list);
+	mutex_init(&opp_table->lock);
+	kref_init(&opp_table->kref);
 
 	/* Secure the device table modification */
-	list_add_rcu(&opp_table->node, &opp_tables);
+	list_add(&opp_table->node, &opp_tables);
 	return opp_table;
 }
 
-/**
- * _kfree_device_rcu() - Free opp_table RCU handler
- * @head:	RCU head
- */
-static void _kfree_device_rcu(struct rcu_head *head)
+void _get_opp_table_kref(struct opp_table *opp_table)
 {
-	struct opp_table *opp_table = container_of(head, struct opp_table,
-						   rcu_head);
-
-	kfree_rcu(opp_table, rcu_head);
+	kref_get(&opp_table->kref);
 }
 
-/**
- * _remove_opp_table() - Removes a OPP table
- * @opp_table: OPP table to be removed.
- *
- * Removes/frees OPP table if it doesn't contain any OPPs.
- */
-static void _remove_opp_table(struct opp_table *opp_table)
+struct opp_table *dev_pm_opp_get_opp_table(struct device *dev)
 {
-	struct opp_device *opp_dev;
+	struct opp_table *opp_table;
 
-	if (!list_empty(&opp_table->opp_list))
-		return;
+	/* Hold our table modification lock here */
+	mutex_lock(&opp_table_lock);
 
-	if (opp_table->supported_hw)
-		return;
+	opp_table = _find_opp_table_unlocked(dev);
+	if (!IS_ERR(opp_table))
+		goto unlock;
 
-	if (opp_table->prop_name)
-		return;
+	opp_table = _allocate_opp_table(dev);
 
-	if (opp_table->regulators)
-		return;
+unlock:
+	mutex_unlock(&opp_table_lock);
 
-	if (opp_table->set_opp)
-		return;
+	return opp_table;
+}
+EXPORT_SYMBOL_GPL(dev_pm_opp_get_opp_table);
+
+static void _opp_table_kref_release(struct kref *kref)
+{
+	struct opp_table *opp_table = container_of(kref, struct opp_table, kref);
+	struct opp_device *opp_dev;
 
 	/* Release clk */
 	if (!IS_ERR(opp_table->clk))
@@ -924,63 +819,60 @@ static void _remove_opp_table(struct opp_table *opp_table)
 	/* dev_list must be empty now */
 	WARN_ON(!list_empty(&opp_table->dev_list));
 
-	list_del_rcu(&opp_table->node);
-	call_srcu(&opp_table->srcu_head.srcu, &opp_table->rcu_head,
-		  _kfree_device_rcu);
+	mutex_destroy(&opp_table->lock);
+	list_del(&opp_table->node);
+	kfree(opp_table);
+
+	mutex_unlock(&opp_table_lock);
 }
 
-/**
- * _kfree_opp_rcu() - Free OPP RCU handler
- * @head:	RCU head
- */
-static void _kfree_opp_rcu(struct rcu_head *head)
+void dev_pm_opp_put_opp_table(struct opp_table *opp_table)
 {
-	struct dev_pm_opp *opp = container_of(head, struct dev_pm_opp, rcu_head);
+	kref_put_mutex(&opp_table->kref, _opp_table_kref_release,
+		       &opp_table_lock);
+}
+EXPORT_SYMBOL_GPL(dev_pm_opp_put_opp_table);
 
-	kfree_rcu(opp, rcu_head);
+void _opp_free(struct dev_pm_opp *opp)
+{
+	kfree(opp);
 }
 
-/**
- * _opp_remove()  - Remove an OPP from a table definition
- * @opp_table:	points back to the opp_table struct this opp belongs to
- * @opp:	pointer to the OPP to remove
- * @notify:	OPP_EVENT_REMOVE notification should be sent or not
- *
- * This function removes an opp definition from the opp table.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * It is assumed that the caller holds required mutex for an RCU updater
- * strategy.
- */
-void _opp_remove(struct opp_table *opp_table, struct dev_pm_opp *opp,
-		 bool notify)
+static void _opp_kref_release(struct kref *kref)
 {
+	struct dev_pm_opp *opp = container_of(kref, struct dev_pm_opp, kref);
+	struct opp_table *opp_table = opp->opp_table;
+
 	/*
 	 * Notify the changes in the availability of the operable
 	 * frequency/voltage list.
 	 */
-	if (notify)
-		srcu_notifier_call_chain(&opp_table->srcu_head,
-					 OPP_EVENT_REMOVE, opp);
+	blocking_notifier_call_chain(&opp_table->head, OPP_EVENT_REMOVE, opp);
 	opp_debug_remove_one(opp);
-	list_del_rcu(&opp->node);
-	call_srcu(&opp_table->srcu_head.srcu, &opp->rcu_head, _kfree_opp_rcu);
+	list_del(&opp->node);
+	kfree(opp);
 
-	_remove_opp_table(opp_table);
+	mutex_unlock(&opp_table->lock);
+	dev_pm_opp_put_opp_table(opp_table);
 }
 
+static void dev_pm_opp_get(struct dev_pm_opp *opp)
+{
+	kref_get(&opp->kref);
+}
+
+void dev_pm_opp_put(struct dev_pm_opp *opp)
+{
+	kref_put_mutex(&opp->kref, _opp_kref_release, &opp->opp_table->lock);
+}
+EXPORT_SYMBOL_GPL(dev_pm_opp_put);
+
 /**
  * dev_pm_opp_remove()  - Remove an OPP from OPP table
  * @dev:	device for which we do this operation
  * @freq:	OPP to remove with matching 'freq'
  *
  * This function removes an opp from the opp table.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
 void dev_pm_opp_remove(struct device *dev, unsigned long freq)
 {
@@ -988,12 +880,11 @@ void dev_pm_opp_remove(struct device *dev, unsigned long freq)
 	struct opp_table *opp_table;
 	bool found = false;
 
-	/* Hold our table modification lock here */
-	mutex_lock(&opp_table_lock);
-
 	opp_table = _find_opp_table(dev);
 	if (IS_ERR(opp_table))
-		goto unlock;
+		return;
+
+	mutex_lock(&opp_table->lock);
 
 	list_for_each_entry(opp, &opp_table->opp_list, node) {
 		if (opp->rate == freq) {
@@ -1002,28 +893,23 @@ void dev_pm_opp_remove(struct device *dev, unsigned long freq)
 		}
 	}
 
-	if (!found) {
+	mutex_unlock(&opp_table->lock);
+
+	if (found) {
+		dev_pm_opp_put(opp);
+	} else {
 		dev_warn(dev, "%s: Couldn't find OPP with freq: %lu\n",
 			 __func__, freq);
-		goto unlock;
 	}
 
-	_opp_remove(opp_table, opp, true);
-unlock:
-	mutex_unlock(&opp_table_lock);
+	dev_pm_opp_put_opp_table(opp_table);
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_remove);
 
-struct dev_pm_opp *_allocate_opp(struct device *dev,
-				 struct opp_table **opp_table)
+struct dev_pm_opp *_opp_allocate(struct opp_table *table)
 {
 	struct dev_pm_opp *opp;
 	int count, supply_size;
-	struct opp_table *table;
-
-	table = _add_opp_table(dev);
-	if (!table)
-		return NULL;
 
 	/* Allocate space for at least one supply */
 	count = table->regulator_count ? table->regulator_count : 1;
@@ -1031,17 +917,13 @@ struct dev_pm_opp *_allocate_opp(struct device *dev,
 
 	/* allocate new OPP node and supplies structures */
 	opp = kzalloc(sizeof(*opp) + supply_size, GFP_KERNEL);
-	if (!opp) {
-		kfree(table);
+	if (!opp)
 		return NULL;
-	}
 
 	/* Put the supplies at the end of the OPP structure as an empty array */
 	opp->supplies = (struct dev_pm_opp_supply *)(opp + 1);
 	INIT_LIST_HEAD(&opp->node);
 
-	*opp_table = table;
-
 	return opp;
 }
 
@@ -1067,11 +949,21 @@ static bool _opp_supported_by_regulators(struct dev_pm_opp *opp,
 	return true;
 }
 
+/*
+ * Returns:
+ * 0: On success. And appropriate error message for duplicate OPPs.
+ * -EBUSY: For OPP with same freq/volt and is available. The callers of
+ *  _opp_add() must return 0 if they receive -EBUSY from it. This is to make
+ *  sure we don't print error messages unnecessarily if different parts of
+ *  kernel try to initialize the OPP table.
+ * -EEXIST: For OPP with same freq but different volt or is unavailable. This
+ *  should be considered an error by the callers of _opp_add().
+ */
 int _opp_add(struct device *dev, struct dev_pm_opp *new_opp,
 	     struct opp_table *opp_table)
 {
 	struct dev_pm_opp *opp;
-	struct list_head *head = &opp_table->opp_list;
+	struct list_head *head;
 	int ret;
 
 	/*
@@ -1082,7 +974,10 @@ int _opp_add(struct device *dev, struct dev_pm_opp *new_opp,
 	 * loop, don't replace it with head otherwise it will become an infinite
 	 * loop.
 	 */
-	list_for_each_entry_rcu(opp, &opp_table->opp_list, node) {
+	mutex_lock(&opp_table->lock);
+	head = &opp_table->opp_list;
+
+	list_for_each_entry(opp, &opp_table->opp_list, node) {
 		if (new_opp->rate > opp->rate) {
 			head = &opp->node;
 			continue;
@@ -1098,12 +993,21 @@ int _opp_add(struct device *dev, struct dev_pm_opp *new_opp,
 			 new_opp->supplies[0].u_volt, new_opp->available);
 
 		/* Should we compare voltages for all regulators here ? */
-		return opp->available &&
-		       new_opp->supplies[0].u_volt == opp->supplies[0].u_volt ? 0 : -EEXIST;
+		ret = opp->available &&
+		      new_opp->supplies[0].u_volt == opp->supplies[0].u_volt ? -EBUSY : -EEXIST;
+
+		mutex_unlock(&opp_table->lock);
+		return ret;
 	}
 
+	list_add(&new_opp->node, head);
+	mutex_unlock(&opp_table->lock);
+
 	new_opp->opp_table = opp_table;
-	list_add_rcu(&new_opp->node, head);
+	kref_init(&new_opp->kref);
+
+	/* Get a reference to the OPP table */
+	_get_opp_table_kref(opp_table);
 
 	ret = opp_debug_create_one(new_opp, opp_table);
 	if (ret)
@@ -1121,6 +1025,7 @@ int _opp_add(struct device *dev, struct dev_pm_opp *new_opp,
 
 /**
  * _opp_add_v1() - Allocate a OPP based on v1 bindings.
+ * @opp_table:	OPP table
  * @dev:	device for which we do this operation
  * @freq:	Frequency in Hz for this OPP
  * @u_volt:	Voltage in uVolts for this OPP
@@ -1133,12 +1038,6 @@ int _opp_add(struct device *dev, struct dev_pm_opp *new_opp,
  * NOTE: "dynamic" parameter impacts OPPs added by the dev_pm_opp_of_add_table
  * and freed by dev_pm_opp_of_remove_table.
  *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
- *
  * Return:
  * 0		On success OR
  *		Duplicate OPPs (both freq and volt are same) and opp->available
@@ -1146,22 +1045,16 @@ int _opp_add(struct device *dev, struct dev_pm_opp *new_opp,
  *		Duplicate OPPs (both freq and volt are same) and !opp->available
  * -ENOMEM	Memory allocation failure
  */
-int _opp_add_v1(struct device *dev, unsigned long freq, long u_volt,
-		bool dynamic)
+int _opp_add_v1(struct opp_table *opp_table, struct device *dev,
+		unsigned long freq, long u_volt, bool dynamic)
 {
-	struct opp_table *opp_table;
 	struct dev_pm_opp *new_opp;
 	unsigned long tol;
 	int ret;
 
-	/* Hold our table modification lock here */
-	mutex_lock(&opp_table_lock);
-
-	new_opp = _allocate_opp(dev, &opp_table);
-	if (!new_opp) {
-		ret = -ENOMEM;
-		goto unlock;
-	}
+	new_opp = _opp_allocate(opp_table);
+	if (!new_opp)
+		return -ENOMEM;
 
 	/* populate the opp table */
 	new_opp->rate = freq;
@@ -1173,22 +1066,23 @@ int _opp_add_v1(struct device *dev, unsigned long freq, long u_volt,
 	new_opp->dynamic = dynamic;
 
 	ret = _opp_add(dev, new_opp, opp_table);
-	if (ret)
+	if (ret) {
+		/* Don't return error for duplicate OPPs */
+		if (ret == -EBUSY)
+			ret = 0;
 		goto free_opp;
-
-	mutex_unlock(&opp_table_lock);
+	}
 
 	/*
 	 * Notify the changes in the availability of the operable
 	 * frequency/voltage list.
 	 */
-	srcu_notifier_call_chain(&opp_table->srcu_head, OPP_EVENT_ADD, new_opp);
+	blocking_notifier_call_chain(&opp_table->head, OPP_EVENT_ADD, new_opp);
 	return 0;
 
 free_opp:
-	_opp_remove(opp_table, new_opp, false);
-unlock:
-	mutex_unlock(&opp_table_lock);
+	_opp_free(new_opp);
+
 	return ret;
 }
 
@@ -1202,27 +1096,16 @@ int _opp_add_v1(struct device *dev, unsigned long freq, long u_volt,
  * specify the hierarchy of versions it supports. OPP layer will then enable
  * OPPs, which are available for those versions, based on its 'opp-supported-hw'
  * property.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
-int dev_pm_opp_set_supported_hw(struct device *dev, const u32 *versions,
-				unsigned int count)
+struct opp_table *dev_pm_opp_set_supported_hw(struct device *dev,
+			const u32 *versions, unsigned int count)
 {
 	struct opp_table *opp_table;
-	int ret = 0;
-
-	/* Hold our table modification lock here */
-	mutex_lock(&opp_table_lock);
+	int ret;
 
-	opp_table = _add_opp_table(dev);
-	if (!opp_table) {
-		ret = -ENOMEM;
-		goto unlock;
-	}
+	opp_table = dev_pm_opp_get_opp_table(dev);
+	if (!opp_table)
+		return ERR_PTR(-ENOMEM);
 
 	/* Make sure there are no concurrent readers while updating opp_table */
 	WARN_ON(!list_empty(&opp_table->opp_list));
@@ -1243,65 +1126,40 @@ int dev_pm_opp_set_supported_hw(struct device *dev, const u32 *versions,
 	}
 
 	opp_table->supported_hw_count = count;
-	mutex_unlock(&opp_table_lock);
-	return 0;
+
+	return opp_table;
 
 err:
-	_remove_opp_table(opp_table);
-unlock:
-	mutex_unlock(&opp_table_lock);
+	dev_pm_opp_put_opp_table(opp_table);
 
-	return ret;
+	return ERR_PTR(ret);
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_set_supported_hw);
 
 /**
  * dev_pm_opp_put_supported_hw() - Releases resources blocked for supported hw
- * @dev: Device for which supported-hw has to be put.
+ * @opp_table: OPP table returned by dev_pm_opp_set_supported_hw().
  *
  * This is required only for the V2 bindings, and is called for a matching
  * dev_pm_opp_set_supported_hw(). Until this is called, the opp_table structure
  * will not be freed.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
-void dev_pm_opp_put_supported_hw(struct device *dev)
+void dev_pm_opp_put_supported_hw(struct opp_table *opp_table)
 {
-	struct opp_table *opp_table;
-
-	/* Hold our table modification lock here */
-	mutex_lock(&opp_table_lock);
-
-	/* Check for existing table for 'dev' first */
-	opp_table = _find_opp_table(dev);
-	if (IS_ERR(opp_table)) {
-		dev_err(dev, "Failed to find opp_table: %ld\n",
-			PTR_ERR(opp_table));
-		goto unlock;
-	}
-
 	/* Make sure there are no concurrent readers while updating opp_table */
 	WARN_ON(!list_empty(&opp_table->opp_list));
 
 	if (!opp_table->supported_hw) {
-		dev_err(dev, "%s: Doesn't have supported hardware list\n",
+		pr_err("%s: Doesn't have supported hardware list\n",
 		       __func__);
-		goto unlock;
+		return;
 	}
 
 	kfree(opp_table->supported_hw);
 	opp_table->supported_hw = NULL;
 	opp_table->supported_hw_count = 0;
 
-	/* Try freeing opp_table if this was the last blocking resource */
-	_remove_opp_table(opp_table);
-
-unlock:
-	mutex_unlock(&opp_table_lock);
+	dev_pm_opp_put_opp_table(opp_table);
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_put_supported_hw);
 
@@ -1314,26 +1172,15 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_put_supported_hw);
  * specify the extn to be used for certain property names. The properties to
  * which the extension will apply are opp-microvolt and opp-microamp. OPP core
  * should postfix the property name with -<name> while looking for them.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
-int dev_pm_opp_set_prop_name(struct device *dev, const char *name)
+struct opp_table *dev_pm_opp_set_prop_name(struct device *dev, const char *name)
 {
 	struct opp_table *opp_table;
-	int ret = 0;
-
-	/* Hold our table modification lock here */
-	mutex_lock(&opp_table_lock);
+	int ret;
 
-	opp_table = _add_opp_table(dev);
-	if (!opp_table) {
-		ret = -ENOMEM;
-		goto unlock;
-	}
+	opp_table = dev_pm_opp_get_opp_table(dev);
+	if (!opp_table)
+		return ERR_PTR(-ENOMEM);
 
 	/* Make sure there are no concurrent readers while updating opp_table */
 	WARN_ON(!list_empty(&opp_table->opp_list));
@@ -1352,63 +1199,37 @@ int dev_pm_opp_set_prop_name(struct device *dev, const char *name)
 		goto err;
 	}
 
-	mutex_unlock(&opp_table_lock);
-	return 0;
+	return opp_table;
 
 err:
-	_remove_opp_table(opp_table);
-unlock:
-	mutex_unlock(&opp_table_lock);
+	dev_pm_opp_put_opp_table(opp_table);
 
-	return ret;
+	return ERR_PTR(ret);
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_set_prop_name);
 
 /**
  * dev_pm_opp_put_prop_name() - Releases resources blocked for prop-name
- * @dev: Device for which the prop-name has to be put.
+ * @opp_table: OPP table returned by dev_pm_opp_set_prop_name().
  *
  * This is required only for the V2 bindings, and is called for a matching
  * dev_pm_opp_set_prop_name(). Until this is called, the opp_table structure
  * will not be freed.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
-void dev_pm_opp_put_prop_name(struct device *dev)
+void dev_pm_opp_put_prop_name(struct opp_table *opp_table)
 {
-	struct opp_table *opp_table;
-
-	/* Hold our table modification lock here */
-	mutex_lock(&opp_table_lock);
-
-	/* Check for existing table for 'dev' first */
-	opp_table = _find_opp_table(dev);
-	if (IS_ERR(opp_table)) {
-		dev_err(dev, "Failed to find opp_table: %ld\n",
-			PTR_ERR(opp_table));
-		goto unlock;
-	}
-
 	/* Make sure there are no concurrent readers while updating opp_table */
 	WARN_ON(!list_empty(&opp_table->opp_list));
 
 	if (!opp_table->prop_name) {
-		dev_err(dev, "%s: Doesn't have a prop-name\n", __func__);
-		goto unlock;
+		pr_err("%s: Doesn't have a prop-name\n", __func__);
+		return;
 	}
 
 	kfree(opp_table->prop_name);
 	opp_table->prop_name = NULL;
 
-	/* Try freeing opp_table if this was the last blocking resource */
-	_remove_opp_table(opp_table);
-
-unlock:
-	mutex_unlock(&opp_table_lock);
+	dev_pm_opp_put_opp_table(opp_table);
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_put_prop_name);
 
@@ -1455,12 +1276,6 @@ static void _free_set_opp_data(struct opp_table *opp_table)
  * well.
  *
  * This must be called before any OPPs are initialized for the device.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
 struct opp_table *dev_pm_opp_set_regulators(struct device *dev,
 					    const char * const names[],
@@ -1470,13 +1285,9 @@ struct opp_table *dev_pm_opp_set_regulators(struct device *dev,
 	struct regulator *reg;
 	int ret, i;
 
-	mutex_lock(&opp_table_lock);
-
-	opp_table = _add_opp_table(dev);
-	if (!opp_table) {
-		ret = -ENOMEM;
-		goto unlock;
-	}
+	opp_table = dev_pm_opp_get_opp_table(dev);
+	if (!opp_table)
+		return ERR_PTR(-ENOMEM);
 
 	/* This should be called before OPPs are initialized */
 	if (WARN_ON(!list_empty(&opp_table->opp_list))) {
@@ -1518,7 +1329,6 @@ struct opp_table *dev_pm_opp_set_regulators(struct device *dev,
 	if (ret)
 		goto free_regulators;
 
-	mutex_unlock(&opp_table_lock);
 	return opp_table;
 
 free_regulators:
@@ -1529,9 +1339,7 @@ struct opp_table *dev_pm_opp_set_regulators(struct device *dev,
 	opp_table->regulators = NULL;
 	opp_table->regulator_count = 0;
 err:
-	_remove_opp_table(opp_table);
-unlock:
-	mutex_unlock(&opp_table_lock);
+	dev_pm_opp_put_opp_table(opp_table);
 
 	return ERR_PTR(ret);
 }
@@ -1540,22 +1348,14 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_set_regulators);
 /**
  * dev_pm_opp_put_regulators() - Releases resources blocked for regulator
  * @opp_table: OPP table returned from dev_pm_opp_set_regulators().
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
 void dev_pm_opp_put_regulators(struct opp_table *opp_table)
 {
 	int i;
 
-	mutex_lock(&opp_table_lock);
-
 	if (!opp_table->regulators) {
 		pr_err("%s: Doesn't have regulators set\n", __func__);
-		goto unlock;
+		return;
 	}
 
 	/* Make sure there are no concurrent readers while updating opp_table */
@@ -1570,11 +1370,7 @@ void dev_pm_opp_put_regulators(struct opp_table *opp_table)
 	opp_table->regulators = NULL;
 	opp_table->regulator_count = 0;
 
-	/* Try freeing opp_table if this was the last blocking resource */
-	_remove_opp_table(opp_table);
-
-unlock:
-	mutex_unlock(&opp_table_lock);
+	dev_pm_opp_put_opp_table(opp_table);
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_put_regulators);
 
@@ -1587,29 +1383,19 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_put_regulators);
  * regulators per device), instead of the generic OPP set rate helper.
  *
  * This must be called before any OPPs are initialized for the device.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
-int dev_pm_opp_register_set_opp_helper(struct device *dev,
+struct opp_table *dev_pm_opp_register_set_opp_helper(struct device *dev,
 			int (*set_opp)(struct dev_pm_set_opp_data *data))
 {
 	struct opp_table *opp_table;
 	int ret;
 
 	if (!set_opp)
-		return -EINVAL;
-
-	mutex_lock(&opp_table_lock);
+		return ERR_PTR(-EINVAL);
 
-	opp_table = _add_opp_table(dev);
-	if (!opp_table) {
-		ret = -ENOMEM;
-		goto unlock;
-	}
+	opp_table = dev_pm_opp_get_opp_table(dev);
+	if (!opp_table)
+		return ERR_PTR(-ENOMEM);
 
 	/* This should be called before OPPs are initialized */
 	if (WARN_ON(!list_empty(&opp_table->opp_list))) {
@@ -1625,47 +1411,28 @@ int dev_pm_opp_register_set_opp_helper(struct device *dev,
 
 	opp_table->set_opp = set_opp;
 
-	mutex_unlock(&opp_table_lock);
-	return 0;
+	return opp_table;
 
 err:
-	_remove_opp_table(opp_table);
-unlock:
-	mutex_unlock(&opp_table_lock);
+	dev_pm_opp_put_opp_table(opp_table);
 
-	return ret;
+	return ERR_PTR(ret);
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_register_set_opp_helper);
 
 /**
  * dev_pm_opp_register_put_opp_helper() - Releases resources blocked for
  *					   set_opp helper
- * @dev: Device for which custom set_opp helper has to be cleared.
+ * @opp_table: OPP table returned from dev_pm_opp_register_set_opp_helper().
  *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
+ * Release resources blocked for platform specific set_opp helper.
  */
-void dev_pm_opp_register_put_opp_helper(struct device *dev)
+void dev_pm_opp_register_put_opp_helper(struct opp_table *opp_table)
 {
-	struct opp_table *opp_table;
-
-	mutex_lock(&opp_table_lock);
-
-	/* Check for existing table for 'dev' first */
-	opp_table = _find_opp_table(dev);
-	if (IS_ERR(opp_table)) {
-		dev_err(dev, "Failed to find opp_table: %ld\n",
-			PTR_ERR(opp_table));
-		goto unlock;
-	}
-
 	if (!opp_table->set_opp) {
-		dev_err(dev, "%s: Doesn't have custom set_opp helper set\n",
+		pr_err("%s: Doesn't have custom set_opp helper set\n",
 		       __func__);
-		goto unlock;
+		return;
 	}
 
 	/* Make sure there are no concurrent readers while updating opp_table */
@@ -1673,11 +1440,7 @@ void dev_pm_opp_register_put_opp_helper(struct device *dev)
 
 	opp_table->set_opp = NULL;
 
-	/* Try freeing opp_table if this was the last blocking resource */
-	_remove_opp_table(opp_table);
-
-unlock:
-	mutex_unlock(&opp_table_lock);
+	dev_pm_opp_put_opp_table(opp_table);
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_register_put_opp_helper);
 
@@ -1691,12 +1454,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_register_put_opp_helper);
  * The opp is made available by default and it can be controlled using
  * dev_pm_opp_enable/disable functions.
  *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
- *
  * Return:
  * 0		On success OR
  *		Duplicate OPPs (both freq and volt are same) and opp->available
@@ -1706,7 +1463,17 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_register_put_opp_helper);
  */
 int dev_pm_opp_add(struct device *dev, unsigned long freq, unsigned long u_volt)
 {
-	return _opp_add_v1(dev, freq, u_volt, true);
+	struct opp_table *opp_table;
+	int ret;
+
+	opp_table = dev_pm_opp_get_opp_table(dev);
+	if (!opp_table)
+		return -ENOMEM;
+
+	ret = _opp_add_v1(opp_table, dev, freq, u_volt, true);
+
+	dev_pm_opp_put_opp_table(opp_table);
+	return ret;
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_add);
 
@@ -1716,41 +1483,30 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_add);
  * @freq:		OPP frequency to modify availability
  * @availability_req:	availability status requested for this opp
  *
- * Set the availability of an OPP with an RCU operation, opp_{enable,disable}
- * share a common logic which is isolated here.
+ * Set the availability of an OPP, opp_{enable,disable} share a common logic
+ * which is isolated here.
  *
  * Return: -EINVAL for bad pointers, -ENOMEM if no memory available for the
  * copy operation, returns 0 if no modification was done OR modification was
  * successful.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks to
- * keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex locking or synchronize_rcu() blocking calls cannot be used.
  */
 static int _opp_set_availability(struct device *dev, unsigned long freq,
 				 bool availability_req)
 {
 	struct opp_table *opp_table;
-	struct dev_pm_opp *new_opp, *tmp_opp, *opp = ERR_PTR(-ENODEV);
+	struct dev_pm_opp *tmp_opp, *opp = ERR_PTR(-ENODEV);
 	int r = 0;
 
-	/* keep the node allocated */
-	new_opp = kmalloc(sizeof(*new_opp), GFP_KERNEL);
-	if (!new_opp)
-		return -ENOMEM;
-
-	mutex_lock(&opp_table_lock);
-
 	/* Find the opp_table */
 	opp_table = _find_opp_table(dev);
 	if (IS_ERR(opp_table)) {
 		r = PTR_ERR(opp_table);
 		dev_warn(dev, "%s: Device OPP not found (%d)\n", __func__, r);
-		goto unlock;
+		return r;
 	}
 
+	mutex_lock(&opp_table->lock);
+
 	/* Do we have the frequency? */
 	list_for_each_entry(tmp_opp, &opp_table->opp_list, node) {
 		if (tmp_opp->rate == freq) {
@@ -1758,6 +1514,7 @@ static int _opp_set_availability(struct device *dev, unsigned long freq,
 			break;
 		}
 	}
+
 	if (IS_ERR(opp)) {
 		r = PTR_ERR(opp);
 		goto unlock;
@@ -1766,29 +1523,20 @@ static int _opp_set_availability(struct device *dev, unsigned long freq,
 	/* Is update really needed? */
 	if (opp->available == availability_req)
 		goto unlock;
-	/* copy the old data over */
-	*new_opp = *opp;
 
-	/* plug in new node */
-	new_opp->available = availability_req;
-
-	list_replace_rcu(&opp->node, &new_opp->node);
-	mutex_unlock(&opp_table_lock);
-	call_srcu(&opp_table->srcu_head.srcu, &opp->rcu_head, _kfree_opp_rcu);
+	opp->available = availability_req;
 
 	/* Notify the change of the OPP availability */
 	if (availability_req)
-		srcu_notifier_call_chain(&opp_table->srcu_head,
-					 OPP_EVENT_ENABLE, new_opp);
+		blocking_notifier_call_chain(&opp_table->head, OPP_EVENT_ENABLE,
+					     opp);
 	else
-		srcu_notifier_call_chain(&opp_table->srcu_head,
-					 OPP_EVENT_DISABLE, new_opp);
-
-	return 0;
+		blocking_notifier_call_chain(&opp_table->head,
+					     OPP_EVENT_DISABLE, opp);
 
 unlock:
-	mutex_unlock(&opp_table_lock);
-	kfree(new_opp);
+	mutex_unlock(&opp_table->lock);
+	dev_pm_opp_put_opp_table(opp_table);
 	return r;
 }
 
@@ -1801,12 +1549,6 @@ static int _opp_set_availability(struct device *dev, unsigned long freq,
  * corresponding error value. It is meant to be used for users an OPP available
  * after being temporarily made unavailable with dev_pm_opp_disable.
  *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function indirectly uses RCU and mutex locks to keep the
- * integrity of the internal data structures. Callers should ensure that
- * this function is *NOT* called under RCU protection or in contexts where
- * mutex locking or synchronize_rcu() blocking calls cannot be used.
- *
  * Return: -EINVAL for bad pointers, -ENOMEM if no memory available for the
  * copy operation, returns 0 if no modification was done OR modification was
  * successful.
@@ -1827,12 +1569,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_enable);
  * control by users to make this OPP not available until the circumstances are
  * right to make it available again (with a call to dev_pm_opp_enable).
  *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function indirectly uses RCU and mutex locks to keep the
- * integrity of the internal data structures. Callers should ensure that
- * this function is *NOT* called under RCU protection or in contexts where
- * mutex locking or synchronize_rcu() blocking calls cannot be used.
- *
  * Return: -EINVAL for bad pointers, -ENOMEM if no memory available for the
  * copy operation, returns 0 if no modification was done OR modification was
  * successful.
@@ -1844,41 +1580,78 @@ int dev_pm_opp_disable(struct device *dev, unsigned long freq)
 EXPORT_SYMBOL_GPL(dev_pm_opp_disable);
 
 /**
- * dev_pm_opp_get_notifier() - find notifier_head of the device with opp
- * @dev:	device pointer used to lookup OPP table.
+ * dev_pm_opp_register_notifier() - Register OPP notifier for the device
+ * @dev:	Device for which notifier needs to be registered
+ * @nb:		Notifier block to be registered
  *
- * Return: pointer to  notifier head if found, otherwise -ENODEV or
- * -EINVAL based on type of error casted as pointer. value must be checked
- *  with IS_ERR to determine valid pointer or error result.
+ * Return: 0 on success or a negative error value.
+ */
+int dev_pm_opp_register_notifier(struct device *dev, struct notifier_block *nb)
+{
+	struct opp_table *opp_table;
+	int ret;
+
+	opp_table = _find_opp_table(dev);
+	if (IS_ERR(opp_table))
+		return PTR_ERR(opp_table);
+
+	ret = blocking_notifier_chain_register(&opp_table->head, nb);
+
+	dev_pm_opp_put_opp_table(opp_table);
+
+	return ret;
+}
+EXPORT_SYMBOL(dev_pm_opp_register_notifier);
+
+/**
+ * dev_pm_opp_unregister_notifier() - Unregister OPP notifier for the device
+ * @dev:	Device for which notifier needs to be unregistered
+ * @nb:		Notifier block to be unregistered
  *
- * Locking: This function must be called under rcu_read_lock(). opp_table is a
- * RCU protected pointer. The reason for the same is that the opp pointer which
- * is returned will remain valid for use with opp_get_{voltage, freq} only while
- * under the locked area. The pointer returned must be used prior to unlocking
- * with rcu_read_unlock() to maintain the integrity of the pointer.
+ * Return: 0 on success or a negative error value.
  */
-struct srcu_notifier_head *dev_pm_opp_get_notifier(struct device *dev)
+int dev_pm_opp_unregister_notifier(struct device *dev,
+				   struct notifier_block *nb)
 {
-	struct opp_table *opp_table = _find_opp_table(dev);
+	struct opp_table *opp_table;
+	int ret;
 
+	opp_table = _find_opp_table(dev);
 	if (IS_ERR(opp_table))
-		return ERR_CAST(opp_table); /* matching type */
+		return PTR_ERR(opp_table);
+
+	ret = blocking_notifier_chain_unregister(&opp_table->head, nb);
 
-	return &opp_table->srcu_head;
+	dev_pm_opp_put_opp_table(opp_table);
+
+	return ret;
 }
-EXPORT_SYMBOL_GPL(dev_pm_opp_get_notifier);
+EXPORT_SYMBOL(dev_pm_opp_unregister_notifier);
 
 /*
  * Free OPPs either created using static entries present in DT or even the
  * dynamically added entries based on remove_all param.
  */
-void _dev_pm_opp_remove_table(struct device *dev, bool remove_all)
+void _dev_pm_opp_remove_table(struct opp_table *opp_table, struct device *dev,
+			      bool remove_all)
 {
-	struct opp_table *opp_table;
 	struct dev_pm_opp *opp, *tmp;
 
-	/* Hold our table modification lock here */
-	mutex_lock(&opp_table_lock);
+	/* Find if opp_table manages a single device */
+	if (list_is_singular(&opp_table->dev_list)) {
+		/* Free static OPPs */
+		list_for_each_entry_safe(opp, tmp, &opp_table->opp_list, node) {
+			if (remove_all || !opp->dynamic)
+				dev_pm_opp_put(opp);
+		}
+	} else {
+		_remove_opp_dev(_find_opp_dev(dev, opp_table), opp_table);
+	}
+}
+
+void _dev_pm_opp_find_and_remove_table(struct device *dev, bool remove_all)
+{
+	struct opp_table *opp_table;
 
 	/* Check for existing table for 'dev' */
 	opp_table = _find_opp_table(dev);
@@ -1890,22 +1663,12 @@ void _dev_pm_opp_remove_table(struct device *dev, bool remove_all)
 			     IS_ERR_OR_NULL(dev) ?
 					"Invalid device" : dev_name(dev),
 			     error);
-		goto unlock;
+		return;
 	}
 
-	/* Find if opp_table manages a single device */
-	if (list_is_singular(&opp_table->dev_list)) {
-		/* Free static OPPs */
-		list_for_each_entry_safe(opp, tmp, &opp_table->opp_list, node) {
-			if (remove_all || !opp->dynamic)
-				_opp_remove(opp_table, opp, true);
-		}
-	} else {
-		_remove_opp_dev(_find_opp_dev(dev, opp_table), opp_table);
-	}
+	_dev_pm_opp_remove_table(opp_table, dev, remove_all);
 
-unlock:
-	mutex_unlock(&opp_table_lock);
+	dev_pm_opp_put_opp_table(opp_table);
 }
 
 /**
@@ -1914,15 +1677,9 @@ void _dev_pm_opp_remove_table(struct device *dev, bool remove_all)
  *
  * Free both OPPs created using static entries present in DT and the
  * dynamically added entries.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function indirectly uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
 void dev_pm_opp_remove_table(struct device *dev)
 {
-	_dev_pm_opp_remove_table(dev, true);
+	_dev_pm_opp_find_and_remove_table(dev, true);
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_remove_table);

drivers/base/power/opp/cpu.c

@@ -42,11 +42,6 @@
  *
  * WARNING: It is  important for the callers to ensure refreshing their copy of
  * the table if any of the mentioned functions have been invoked in the interim.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Since we just use the regular accessor functions to access the internal data
- * structures, we use RCU read lock inside this function. As a result, users of
- * this function DONOT need to use explicit locks for invoking.
  */
 int dev_pm_opp_init_cpufreq_table(struct device *dev,
 				  struct cpufreq_frequency_table **table)
@@ -56,19 +51,13 @@ int dev_pm_opp_init_cpufreq_table(struct device *dev,
 	int i, max_opps, ret = 0;
 	unsigned long rate;
 
-	rcu_read_lock();
-
 	max_opps = dev_pm_opp_get_opp_count(dev);
-	if (max_opps <= 0) {
-		ret = max_opps ? max_opps : -ENODATA;
-		goto out;
-	}
+	if (max_opps <= 0)
+		return max_opps ? max_opps : -ENODATA;
 
 	freq_table = kcalloc((max_opps + 1), sizeof(*freq_table), GFP_ATOMIC);
-	if (!freq_table) {
-		ret = -ENOMEM;
-		goto out;
-	}
+	if (!freq_table)
+		return -ENOMEM;
 
 	for (i = 0, rate = 0; i < max_opps; i++, rate++) {
 		/* find next rate */
@@ -83,6 +72,8 @@ int dev_pm_opp_init_cpufreq_table(struct device *dev,
 		/* Is Boost/turbo opp ? */
 		if (dev_pm_opp_is_turbo(opp))
 			freq_table[i].flags = CPUFREQ_BOOST_FREQ;
+
+		dev_pm_opp_put(opp);
 	}
 
 	freq_table[i].driver_data = i;
@@ -91,7 +82,6 @@ int dev_pm_opp_init_cpufreq_table(struct device *dev,
 	*table = &freq_table[0];
 
 out:
-	rcu_read_unlock();
 	if (ret)
 		kfree(freq_table);
 
@@ -147,12 +137,6 @@ void _dev_pm_opp_cpumask_remove_table(const struct cpumask *cpumask, bool of)
  * This removes the OPP tables for CPUs present in the @cpumask.
  * This should be used to remove all the OPPs entries associated with
  * the cpus in @cpumask.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
 void dev_pm_opp_cpumask_remove_table(const struct cpumask *cpumask)
 {
@@ -169,12 +153,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_cpumask_remove_table);
  * @cpumask.
  *
  * Returns -ENODEV if OPP table isn't already present.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
 int dev_pm_opp_set_sharing_cpus(struct device *cpu_dev,
 				const struct cpumask *cpumask)
@@ -184,13 +162,9 @@ int dev_pm_opp_set_sharing_cpus(struct device *cpu_dev,
 	struct device *dev;
 	int cpu, ret = 0;
 
-	mutex_lock(&opp_table_lock);
-
 	opp_table = _find_opp_table(cpu_dev);
-	if (IS_ERR(opp_table)) {
-		ret = PTR_ERR(opp_table);
-		goto unlock;
-	}
+	if (IS_ERR(opp_table))
+		return PTR_ERR(opp_table);
 
 	for_each_cpu(cpu, cpumask) {
 		if (cpu == cpu_dev->id)
@@ -213,8 +187,8 @@ int dev_pm_opp_set_sharing_cpus(struct device *cpu_dev,
 		/* Mark opp-table as multiple CPUs are sharing it now */
 		opp_table->shared_opp = OPP_TABLE_ACCESS_SHARED;
 	}
-unlock:
-	mutex_unlock(&opp_table_lock);
+
+	dev_pm_opp_put_opp_table(opp_table);
 
 	return ret;
 }
@@ -229,12 +203,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_set_sharing_cpus);
  *
  * Returns -ENODEV if OPP table isn't already present and -EINVAL if the OPP
  * table's status is access-unknown.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
 int dev_pm_opp_get_sharing_cpus(struct device *cpu_dev, struct cpumask *cpumask)
 {
@@ -242,17 +210,13 @@ int dev_pm_opp_get_sharing_cpus(struct device *cpu_dev, struct cpumask *cpumask)
 	struct opp_table *opp_table;
 	int ret = 0;
 
-	mutex_lock(&opp_table_lock);
-
 	opp_table = _find_opp_table(cpu_dev);
-	if (IS_ERR(opp_table)) {
-		ret = PTR_ERR(opp_table);
-		goto unlock;
-	}
+	if (IS_ERR(opp_table))
+		return PTR_ERR(opp_table);
 
 	if (opp_table->shared_opp == OPP_TABLE_ACCESS_UNKNOWN) {
 		ret = -EINVAL;
-		goto unlock;
+		goto put_opp_table;
 	}
 
 	cpumask_clear(cpumask);
@@ -264,8 +228,8 @@ int dev_pm_opp_get_sharing_cpus(struct device *cpu_dev, struct cpumask *cpumask)
 		cpumask_set_cpu(cpu_dev->id, cpumask);
 	}
 
-unlock:
-	mutex_unlock(&opp_table_lock);
+put_opp_table:
+	dev_pm_opp_put_opp_table(opp_table);
 
 	return ret;
 }

drivers/base/power/opp/of.c

@@ -24,9 +24,11 @@
 
 static struct opp_table *_managed_opp(const struct device_node *np)
 {
-	struct opp_table *opp_table;
+	struct opp_table *opp_table, *managed_table = NULL;
+
+	mutex_lock(&opp_table_lock);
 
-	list_for_each_entry_rcu(opp_table, &opp_tables, node) {
+	list_for_each_entry(opp_table, &opp_tables, node) {
 		if (opp_table->np == np) {
 			/*
 			 * Multiple devices can point to the same OPP table and
@@ -35,14 +37,18 @@ static struct opp_table *_managed_opp(const struct device_node *np)
 			 * But the OPPs will be considered as shared only if the
 			 * OPP table contains a "opp-shared" property.
 			 */
-			if (opp_table->shared_opp == OPP_TABLE_ACCESS_SHARED)
-				return opp_table;
+			if (opp_table->shared_opp == OPP_TABLE_ACCESS_SHARED) {
+				_get_opp_table_kref(opp_table);
+				managed_table = opp_table;
+			}
 
-			return NULL;
+			break;
 		}
 	}
 
-	return NULL;
+	mutex_unlock(&opp_table_lock);
+
+	return managed_table;
 }
 
 void _of_init_opp_table(struct opp_table *opp_table, struct device *dev)
@@ -229,34 +235,28 @@ static int opp_parse_supplies(struct dev_pm_opp *opp, struct device *dev,
  * @dev:	device pointer used to lookup OPP table.
  *
  * Free OPPs created using static entries present in DT.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function indirectly uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
 void dev_pm_opp_of_remove_table(struct device *dev)
 {
-	_dev_pm_opp_remove_table(dev, false);
+	_dev_pm_opp_find_and_remove_table(dev, false);
 }
 EXPORT_SYMBOL_GPL(dev_pm_opp_of_remove_table);
 
 /* Returns opp descriptor node for a device, caller must do of_node_put() */
-static struct device_node *_of_get_opp_desc_node(struct device *dev)
+struct device_node *dev_pm_opp_of_get_opp_desc_node(struct device *dev)
 {
 	/*
-	 * TODO: Support for multiple OPP tables.
-	 *
 	 * There should be only ONE phandle present in "operating-points-v2"
 	 * property.
 	 */
 
 	return of_parse_phandle(dev->of_node, "operating-points-v2", 0);
 }
+EXPORT_SYMBOL_GPL(dev_pm_opp_of_get_opp_desc_node);
 
 /**
  * _opp_add_static_v2() - Allocate static OPPs (As per 'v2' DT bindings)
+ * @opp_table:	OPP table
  * @dev:	device for which we do this operation
  * @np:		device node
  *
@@ -264,12 +264,6 @@ static struct device_node *_of_get_opp_desc_node(struct device *dev)
  * opp can be controlled using dev_pm_opp_enable/disable functions and may be
  * removed by dev_pm_opp_remove.
  *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
- *
  * Return:
  * 0		On success OR
  *		Duplicate OPPs (both freq and volt are same) and opp->available
@@ -278,22 +272,17 @@ static struct device_node *_of_get_opp_desc_node(struct device *dev)
  * -ENOMEM	Memory allocation failure
  * -EINVAL	Failed parsing the OPP node
  */
-static int _opp_add_static_v2(struct device *dev, struct device_node *np)
+static int _opp_add_static_v2(struct opp_table *opp_table, struct device *dev,
+			      struct device_node *np)
 {
-	struct opp_table *opp_table;
 	struct dev_pm_opp *new_opp;
 	u64 rate;
 	u32 val;
 	int ret;
 
-	/* Hold our table modification lock here */
-	mutex_lock(&opp_table_lock);
-
-	new_opp = _allocate_opp(dev, &opp_table);
-	if (!new_opp) {
-		ret = -ENOMEM;
-		goto unlock;
-	}
+	new_opp = _opp_allocate(opp_table);
+	if (!new_opp)
+		return -ENOMEM;
 
 	ret = of_property_read_u64(np, "opp-hz", &rate);
 	if (ret < 0) {
@@ -327,8 +316,12 @@ static int _opp_add_static_v2(struct device *dev, struct device_node *np)
 		goto free_opp;
 
 	ret = _opp_add(dev, new_opp, opp_table);
-	if (ret)
+	if (ret) {
+		/* Don't return error for duplicate OPPs */
+		if (ret == -EBUSY)
+			ret = 0;
 		goto free_opp;
+	}
 
 	/* OPP to select on device suspend */
 	if (of_property_read_bool(np, "opp-suspend")) {
@@ -345,8 +338,6 @@ static int _opp_add_static_v2(struct device *dev, struct device_node *np)
 	if (new_opp->clock_latency_ns > opp_table->clock_latency_ns_max)
 		opp_table->clock_latency_ns_max = new_opp->clock_latency_ns;
 
-	mutex_unlock(&opp_table_lock);
-
 	pr_debug("%s: turbo:%d rate:%lu uv:%lu uvmin:%lu uvmax:%lu latency:%lu\n",
 		 __func__, new_opp->turbo, new_opp->rate,
 		 new_opp->supplies[0].u_volt, new_opp->supplies[0].u_volt_min,
@@ -356,13 +347,12 @@ static int _opp_add_static_v2(struct device *dev, struct device_node *np)
 	 * Notify the changes in the availability of the operable
 	 * frequency/voltage list.
 	 */
-	srcu_notifier_call_chain(&opp_table->srcu_head, OPP_EVENT_ADD, new_opp);
+	blocking_notifier_call_chain(&opp_table->head, OPP_EVENT_ADD, new_opp);
 	return 0;
 
 free_opp:
-	_opp_remove(opp_table, new_opp, false);
-unlock:
-	mutex_unlock(&opp_table_lock);
+	_opp_free(new_opp);
+
 	return ret;
 }
 
@@ -373,41 +363,35 @@ static int _of_add_opp_table_v2(struct device *dev, struct device_node *opp_np)
 	struct opp_table *opp_table;
 	int ret = 0, count = 0;
 
-	mutex_lock(&opp_table_lock);
-
 	opp_table = _managed_opp(opp_np);
 	if (opp_table) {
 		/* OPPs are already managed */
 		if (!_add_opp_dev(dev, opp_table))
 			ret = -ENOMEM;
-		mutex_unlock(&opp_table_lock);
-		return ret;
+		goto put_opp_table;
 	}
-	mutex_unlock(&opp_table_lock);
+
+	opp_table = dev_pm_opp_get_opp_table(dev);
+	if (!opp_table)
+		return -ENOMEM;
 
 	/* We have opp-table node now, iterate over it and add OPPs */
 	for_each_available_child_of_node(opp_np, np) {
 		count++;
 
-		ret = _opp_add_static_v2(dev, np);
+		ret = _opp_add_static_v2(opp_table, dev, np);
 		if (ret) {
 			dev_err(dev, "%s: Failed to add OPP, %d\n", __func__,
 				ret);
-			goto free_table;
+			_dev_pm_opp_remove_table(opp_table, dev, false);
+			goto put_opp_table;
 		}
 	}
 
 	/* There should be one of more OPP defined */
-	if (WARN_ON(!count))
-		return -ENOENT;
-
-	mutex_lock(&opp_table_lock);
-
-	opp_table = _find_opp_table(dev);
-	if (WARN_ON(IS_ERR(opp_table))) {
-		ret = PTR_ERR(opp_table);
-		mutex_unlock(&opp_table_lock);
-		goto free_table;
+	if (WARN_ON(!count)) {
+		ret = -ENOENT;
+		goto put_opp_table;
 	}
 
 	opp_table->np = opp_np;
@@ -416,12 +400,8 @@ static int _of_add_opp_table_v2(struct device *dev, struct device_node *opp_np)
 	else
 		opp_table->shared_opp = OPP_TABLE_ACCESS_EXCLUSIVE;
 
-	mutex_unlock(&opp_table_lock);
-
-	return 0;
-
-free_table:
-	dev_pm_opp_of_remove_table(dev);
+put_opp_table:
+	dev_pm_opp_put_opp_table(opp_table);
 
 	return ret;
 }
@@ -429,9 +409,10 @@ static int _of_add_opp_table_v2(struct device *dev, struct device_node *opp_np)
 /* Initializes OPP tables based on old-deprecated bindings */
 static int _of_add_opp_table_v1(struct device *dev)
 {
+	struct opp_table *opp_table;
 	const struct property *prop;
 	const __be32 *val;
-	int nr;
+	int nr, ret = 0;
 
 	prop = of_find_property(dev->of_node, "operating-points", NULL);
 	if (!prop)
@@ -449,18 +430,27 @@ static int _of_add_opp_table_v1(struct device *dev)
 		return -EINVAL;
 	}
 
+	opp_table = dev_pm_opp_get_opp_table(dev);
+	if (!opp_table)
+		return -ENOMEM;
+
 	val = prop->value;
 	while (nr) {
 		unsigned long freq = be32_to_cpup(val++) * 1000;
 		unsigned long volt = be32_to_cpup(val++);
 
-		if (_opp_add_v1(dev, freq, volt, false))
-			dev_warn(dev, "%s: Failed to add OPP %ld\n",
-				 __func__, freq);
+		ret = _opp_add_v1(opp_table, dev, freq, volt, false);
+		if (ret) {
+			dev_err(dev, "%s: Failed to add OPP %ld (%d)\n",
+				__func__, freq, ret);
+			_dev_pm_opp_remove_table(opp_table, dev, false);
+			break;
+		}
 		nr -= 2;
 	}
 
-	return 0;
+	dev_pm_opp_put_opp_table(opp_table);
+	return ret;
 }
 
 /**
@@ -469,12 +459,6 @@ static int _of_add_opp_table_v1(struct device *dev)
  *
  * Register the initial OPP table with the OPP library for given device.
  *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function indirectly uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
- *
  * Return:
  * 0		On success OR
  *		Duplicate OPPs (both freq and volt are same) and opp->available
@@ -495,7 +479,7 @@ int dev_pm_opp_of_add_table(struct device *dev)
 	 * OPPs have two version of bindings now. The older one is deprecated,
 	 * try for the new binding first.
 	 */
-	opp_np = _of_get_opp_desc_node(dev);
+	opp_np = dev_pm_opp_of_get_opp_desc_node(dev);
 	if (!opp_np) {
 		/*
 		 * Try old-deprecated bindings for backward compatibility with
@@ -519,12 +503,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_of_add_table);
  *
  * This removes the OPP tables for CPUs present in the @cpumask.
  * This should be used only to remove static entries created from DT.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
 void dev_pm_opp_of_cpumask_remove_table(const struct cpumask *cpumask)
 {
@@ -537,12 +515,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_of_cpumask_remove_table);
  * @cpumask:	cpumask for which OPP table needs to be added.
  *
  * This adds the OPP tables for CPUs present in the @cpumask.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
 int dev_pm_opp_of_cpumask_add_table(const struct cpumask *cpumask)
 {
@@ -590,12 +562,6 @@ EXPORT_SYMBOL_GPL(dev_pm_opp_of_cpumask_add_table);
  * This updates the @cpumask with CPUs that are sharing OPPs with @cpu_dev.
  *
  * Returns -ENOENT if operating-points-v2 isn't present for @cpu_dev.
- *
- * Locking: The internal opp_table and opp structures are RCU protected.
- * Hence this function internally uses RCU updater strategy with mutex locks
- * to keep the integrity of the internal data structures. Callers should ensure
- * that this function is *NOT* called under RCU protection or in contexts where
- * mutex cannot be locked.
  */
 int dev_pm_opp_of_get_sharing_cpus(struct device *cpu_dev,
 				   struct cpumask *cpumask)
@@ -605,7 +571,7 @@ int dev_pm_opp_of_get_sharing_cpus(struct device *cpu_dev,
 	int cpu, ret = 0;
 
 	/* Get OPP descriptor node */
-	np = _of_get_opp_desc_node(cpu_dev);
+	np = dev_pm_opp_of_get_opp_desc_node(cpu_dev);
 	if (!np) {
 		dev_dbg(cpu_dev, "%s: Couldn't find opp node.\n", __func__);
 		return -ENOENT;
@@ -630,7 +596,7 @@ int dev_pm_opp_of_get_sharing_cpus(struct device *cpu_dev,
 		}
 
 		/* Get OPP descriptor node */
-		tmp_np = _of_get_opp_desc_node(tcpu_dev);
+		tmp_np = dev_pm_opp_of_get_opp_desc_node(tcpu_dev);
 		if (!tmp_np) {
 			dev_err(tcpu_dev, "%s: Couldn't find opp node.\n",
 				__func__);

drivers/base/power/opp/opp.h

@@ -16,11 +16,11 @@
 
 #include <linux/device.h>
 #include <linux/kernel.h>
+#include <linux/kref.h>
 #include <linux/list.h>
 #include <linux/limits.h>
 #include <linux/pm_opp.h>
-#include <linux/rculist.h>
-#include <linux/rcupdate.h>
+#include <linux/notifier.h>
 
 struct clk;
 struct regulator;
@@ -51,11 +51,9 @@ extern struct list_head opp_tables;
  * @node:	opp table node. The nodes are maintained throughout the lifetime
  *		of boot. It is expected only an optimal set of OPPs are
  *		added to the library by the SoC framework.
- *		RCU usage: opp table is traversed with RCU locks. node
- *		modification is possible realtime, hence the modifications
- *		are protected by the opp_table_lock for integrity.
  *		IMPORTANT: the opp nodes should be maintained in increasing
  *		order.
+ * @kref:	for reference count of the OPP.
  * @available:	true/false - marks if this OPP as available or not
  * @dynamic:	not-created from static DT entries.
  * @turbo:	true if turbo (boost) OPP
@@ -65,7 +63,6 @@ extern struct list_head opp_tables;
  * @clock_latency_ns: Latency (in nanoseconds) of switching to this OPP's
  *		frequency from any other OPP's frequency.
  * @opp_table:	points back to the opp_table struct this opp belongs to
- * @rcu_head:	RCU callback head used for deferred freeing
  * @np:		OPP's device node.
  * @dentry:	debugfs dentry pointer (per opp)
  *
@@ -73,6 +70,7 @@ extern struct list_head opp_tables;
  */
 struct dev_pm_opp {
 	struct list_head node;
+	struct kref kref;
 
 	bool available;
 	bool dynamic;
@@ -85,7 +83,6 @@ struct dev_pm_opp {
 	unsigned long clock_latency_ns;
 
 	struct opp_table *opp_table;
-	struct rcu_head rcu_head;
 
 	struct device_node *np;
 
@@ -98,7 +95,6 @@ struct dev_pm_opp {
  * struct opp_device - devices managed by 'struct opp_table'
  * @node:	list node
  * @dev:	device to which the struct object belongs
- * @rcu_head:	RCU callback head used for deferred freeing
  * @dentry:	debugfs dentry pointer (per device)
  *
  * This is an internal data structure maintaining the devices that are managed
@@ -107,7 +103,6 @@ struct dev_pm_opp {
 struct opp_device {
 	struct list_head node;
 	const struct device *dev;
-	struct rcu_head rcu_head;
 
 #ifdef CONFIG_DEBUG_FS
 	struct dentry *dentry;
@@ -125,12 +120,11 @@ enum opp_table_access {
  * @node:	table node - contains the devices with OPPs that
  *		have been registered. Nodes once added are not modified in this
  *		table.
- *		RCU usage: nodes are not modified in the table of opp_table,
- *		however addition is possible and is secured by opp_table_lock
- * @srcu_head:	notifier head to notify the OPP availability changes.
- * @rcu_head:	RCU callback head used for deferred freeing
+ * @head:	notifier head to notify the OPP availability changes.
  * @dev_list:	list of devices that share these OPPs
  * @opp_list:	table of opps
+ * @kref:	for reference count of the table.
+ * @lock:	mutex protecting the opp_list.
  * @np:		struct device_node pointer for opp's DT node.
  * @clock_latency_ns_max: Max clock latency in nanoseconds.
  * @shared_opp: OPP is shared between multiple devices.
@@ -151,18 +145,15 @@ enum opp_table_access {
  * This is an internal data structure maintaining the link to opps attached to
  * a device. This structure is not meant to be shared to users as it is
  * meant for book keeping and private to OPP library.
- *
- * Because the opp structures can be used from both rcu and srcu readers, we
- * need to wait for the grace period of both of them before freeing any
- * resources. And so we have used kfree_rcu() from within call_srcu() handlers.
  */
 struct opp_table {
 	struct list_head node;
 
-	struct srcu_notifier_head srcu_head;
-	struct rcu_head rcu_head;
+	struct blocking_notifier_head head;
 	struct list_head dev_list;
 	struct list_head opp_list;
+	struct kref kref;
+	struct mutex lock;
 
 	struct device_node *np;
 	unsigned long clock_latency_ns_max;
@@ -190,14 +181,17 @@ struct opp_table {
 };
 
 /* Routines internal to opp core */
+void _get_opp_table_kref(struct opp_table *opp_table);
 struct opp_table *_find_opp_table(struct device *dev);
 struct opp_device *_add_opp_dev(const struct device *dev, struct opp_table *opp_table);
-void _dev_pm_opp_remove_table(struct device *dev, bool remove_all);
-struct dev_pm_opp *_allocate_opp(struct device *dev, struct opp_table **opp_table);
+void _dev_pm_opp_remove_table(struct opp_table *opp_table, struct device *dev, bool remove_all);
+void _dev_pm_opp_find_and_remove_table(struct device *dev, bool remove_all);
+struct dev_pm_opp *_opp_allocate(struct opp_table *opp_table);
+void _opp_free(struct dev_pm_opp *opp);
 int _opp_add(struct device *dev, struct dev_pm_opp *new_opp, struct opp_table *opp_table);
-void _opp_remove(struct opp_table *opp_table, struct dev_pm_opp *opp, bool notify);
-int _opp_add_v1(struct device *dev, unsigned long freq, long u_volt, bool dynamic);
+int _opp_add_v1(struct opp_table *opp_table, struct device *dev, unsigned long freq, long u_volt, bool dynamic);
 void _dev_pm_opp_cpumask_remove_table(const struct cpumask *cpumask, bool of);
+struct opp_table *_add_opp_table(struct device *dev);
 
 #ifdef CONFIG_OF
 void _of_init_opp_table(struct opp_table *opp_table, struct device *dev);

drivers/base/power/qos.c

@@ -281,7 +281,7 @@ void dev_pm_qos_constraints_destroy(struct device *dev)
 	dev->power.qos = ERR_PTR(-ENODEV);
 	spin_unlock_irq(&dev->power.lock);
 
-	kfree(c->notifiers);
+	kfree(qos->resume_latency.notifiers);
 	kfree(qos);
 
  out:

drivers/base/power/wakeirq.c

@@ -141,6 +141,13 @@ static irqreturn_t handle_threaded_wake_irq(int irq, void *_wirq)
 	struct wake_irq *wirq = _wirq;
 	int res;
 
+	/* Maybe abort suspend? */
+	if (irqd_is_wakeup_set(irq_get_irq_data(irq))) {
+		pm_wakeup_event(wirq->dev, 0);
+
+		return IRQ_HANDLED;
+	}
+
 	/* We don't want RPM_ASYNC or RPM_NOWAIT here */
 	res = pm_runtime_resume(wirq->dev);
 	if (res < 0)
@@ -183,6 +190,9 @@ int dev_pm_set_dedicated_wake_irq(struct device *dev, int irq)
 	wirq->irq = irq;
 	irq_set_status_flags(irq, IRQ_NOAUTOEN);
 
+	/* Prevent deferred spurious wakeirqs with disable_irq_nosync() */
+	irq_set_status_flags(irq, IRQ_DISABLE_UNLAZY);
+
 	/*
 	 * Consumer device may need to power up and restore state
 	 * so we use a threaded irq.
@@ -312,8 +322,12 @@ void dev_pm_arm_wake_irq(struct wake_irq *wirq)
 	if (!wirq)
 		return;
 
-	if (device_may_wakeup(wirq->dev))
+	if (device_may_wakeup(wirq->dev)) {
+		if (wirq->status & WAKE_IRQ_DEDICATED_ALLOCATED)
+			enable_irq(wirq->irq);
+
 		enable_irq_wake(wirq->irq);
+	}
 }
 
 /**
@@ -328,6 +342,10 @@ void dev_pm_disarm_wake_irq(struct wake_irq *wirq)
 	if (!wirq)
 		return;
 
-	if (device_may_wakeup(wirq->dev))
+	if (device_may_wakeup(wirq->dev)) {
 		disable_irq_wake(wirq->irq);
+
+		if (wirq->status & WAKE_IRQ_DEDICATED_ALLOCATED)
+			disable_irq_nosync(wirq->irq);
+	}
 }

drivers/clk/tegra/clk-dfll.c

@@ -633,16 +633,12 @@ static int find_lut_index_for_rate(struct tegra_dfll *td, unsigned long rate)
 	struct dev_pm_opp *opp;
 	int i, uv;
 
-	rcu_read_lock();
-
 	opp = dev_pm_opp_find_freq_ceil(td->soc->dev, &rate);
-	if (IS_ERR(opp)) {
-		rcu_read_unlock();
+	if (IS_ERR(opp))
 		return PTR_ERR(opp);
-	}
-	uv = dev_pm_opp_get_voltage(opp);
 
-	rcu_read_unlock();
+	uv = dev_pm_opp_get_voltage(opp);
+	dev_pm_opp_put(opp);
 
 	for (i = 0; i < td->i2c_lut_size; i++) {
 		if (regulator_list_voltage(td->vdd_reg, td->i2c_lut[i]) == uv)
@@ -1440,8 +1436,6 @@ static int dfll_build_i2c_lut(struct tegra_dfll *td)
 	struct dev_pm_opp *opp;
 	int lut;
 
-	rcu_read_lock();
-
 	rate = ULONG_MAX;
 	opp = dev_pm_opp_find_freq_floor(td->soc->dev, &rate);
 	if (IS_ERR(opp)) {
@@ -1449,6 +1443,7 @@ static int dfll_build_i2c_lut(struct tegra_dfll *td)
 		goto out;
 	}
 	v_max = dev_pm_opp_get_voltage(opp);
+	dev_pm_opp_put(opp);
 
 	v = td->soc->cvb->min_millivolts * 1000;
 	lut = find_vdd_map_entry_exact(td, v);
@@ -1465,6 +1460,8 @@ static int dfll_build_i2c_lut(struct tegra_dfll *td)
 		if (v_opp <= td->soc->cvb->min_millivolts * 1000)
 			td->dvco_rate_min = dev_pm_opp_get_freq(opp);
 
+		dev_pm_opp_put(opp);
+
 		for (;;) {
 			v += max(1, (v_max - v) / (MAX_DFLL_VOLTAGES - j));
 			if (v >= v_opp)
@@ -1496,8 +1493,6 @@ static int dfll_build_i2c_lut(struct tegra_dfll *td)
 		ret = 0;
 
 out:
-	rcu_read_unlock();
-
 	return ret;
 }
 

drivers/cpufreq/Kconfig

@@ -37,14 +37,6 @@ config CPU_FREQ_STAT
 
 	  If in doubt, say N.
 
-config CPU_FREQ_STAT_DETAILS
-	bool "CPU frequency transition statistics details"
-	depends on CPU_FREQ_STAT
-	help
-	  Show detailed CPU frequency transition table in sysfs.
-
-	  If in doubt, say N.
-
 choice
 	prompt "Default CPUFreq governor"
 	default CPU_FREQ_DEFAULT_GOV_USERSPACE if ARM_SA1100_CPUFREQ || ARM_SA1110_CPUFREQ
@@ -271,6 +263,16 @@ config IA64_ACPI_CPUFREQ
 endif
 
 if MIPS
+config BMIPS_CPUFREQ
+	tristate "BMIPS CPUfreq Driver"
+	help
+	  This option adds a CPUfreq driver for BMIPS processors with
+	  support for configurable CPU frequency.
+
+	  For now, BMIPS5 chips are supported (such as the Broadcom 7425).
+
+	  If in doubt, say N.
+
 config LOONGSON2_CPUFREQ
 	tristate "Loongson2 CPUFreq Driver"
 	help
@@ -332,7 +334,7 @@ endif
 
 config QORIQ_CPUFREQ
 	tristate "CPU frequency scaling driver for Freescale QorIQ SoCs"
-	depends on OF && COMMON_CLK && (PPC_E500MC || ARM)
+	depends on OF && COMMON_CLK && (PPC_E500MC || ARM || ARM64)
 	depends on !CPU_THERMAL || THERMAL
 	select CLK_QORIQ
 	help

drivers/cpufreq/Kconfig.arm

@@ -247,6 +247,17 @@ config ARM_TEGRA124_CPUFREQ
 	help
 	  This adds the CPUFreq driver support for Tegra124 SOCs.
 
+config ARM_TI_CPUFREQ
+	bool "Texas Instruments CPUFreq support"
+	depends on ARCH_OMAP2PLUS
+	help
+	  This driver enables valid OPPs on the running platform based on
+	  values contained within the SoC in use. Enable this in order to
+	  use the cpufreq-dt driver on all Texas Instruments platforms that
+	  provide dt based operating-points-v2 tables with opp-supported-hw
+	  data provided. Required for cpufreq support on AM335x, AM437x,
+	  DRA7x, and AM57x platforms.
+
 config ARM_PXA2xx_CPUFREQ
 	tristate "Intel PXA2xx CPUfreq driver"
 	depends on PXA27x || PXA25x
@@ -257,7 +268,7 @@ config ARM_PXA2xx_CPUFREQ
 
 config ACPI_CPPC_CPUFREQ
 	tristate "CPUFreq driver based on the ACPI CPPC spec"
-	depends on ACPI
+	depends on ACPI_PROCESSOR
 	select ACPI_CPPC_LIB
 	default n
 	help

drivers/cpufreq/Makefile

@@ -77,6 +77,7 @@ obj-$(CONFIG_ARM_SPEAR_CPUFREQ)		+= spear-cpufreq.o
 obj-$(CONFIG_ARM_STI_CPUFREQ)		+= sti-cpufreq.o
 obj-$(CONFIG_ARM_TEGRA20_CPUFREQ)	+= tegra20-cpufreq.o
 obj-$(CONFIG_ARM_TEGRA124_CPUFREQ)	+= tegra124-cpufreq.o
+obj-$(CONFIG_ARM_TI_CPUFREQ)		+= ti-cpufreq.o
 obj-$(CONFIG_ARM_VEXPRESS_SPC_CPUFREQ)	+= vexpress-spc-cpufreq.o
 obj-$(CONFIG_ACPI_CPPC_CPUFREQ) += cppc_cpufreq.o
 obj-$(CONFIG_MACH_MVEBU_V7)		+= mvebu-cpufreq.o
@@ -98,6 +99,7 @@ obj-$(CONFIG_POWERNV_CPUFREQ)		+= powernv-cpufreq.o
 # Other platform drivers
 obj-$(CONFIG_AVR32_AT32AP_CPUFREQ)	+= at32ap-cpufreq.o
 obj-$(CONFIG_BFIN_CPU_FREQ)		+= blackfin-cpufreq.o
+obj-$(CONFIG_BMIPS_CPUFREQ)		+= bmips-cpufreq.o
 obj-$(CONFIG_CRIS_MACH_ARTPEC3)		+= cris-artpec3-cpufreq.o
 obj-$(CONFIG_ETRAXFS)			+= cris-etraxfs-cpufreq.o
 obj-$(CONFIG_IA64_ACPI_CPUFREQ)		+= ia64-acpi-cpufreq.o

drivers/cpufreq/bmips-cpufreq.c

+/*
+ * CPU frequency scaling for Broadcom BMIPS SoCs
+ *
+ * Copyright (c) 2017 Broadcom
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation version 2.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ */
+
+#include <linux/cpufreq.h>
+#include <linux/module.h>
+#include <linux/of_address.h>
+#include <linux/slab.h>
+
+/* for mips_hpt_frequency */
+#include <asm/time.h>
+
+#define BMIPS_CPUFREQ_PREFIX	"bmips"
+#define BMIPS_CPUFREQ_NAME	BMIPS_CPUFREQ_PREFIX "-cpufreq"
+
+#define TRANSITION_LATENCY	(25 * 1000)	/* 25 us */
+
+#define BMIPS5_CLK_DIV_SET_SHIFT	0x7
+#define BMIPS5_CLK_DIV_SHIFT		0x4
+#define BMIPS5_CLK_DIV_MASK		0xf
+
+enum bmips_type {
+	BMIPS5000,
+	BMIPS5200,
+};
+
+struct cpufreq_compat {
+	const char *compatible;
+	unsigned int bmips_type;
+	unsigned int clk_mult;
+	unsigned int max_freqs;
+};
+
+#define BMIPS(c, t, m, f) { \
+	.compatible = c, \
+	.bmips_type = (t), \
+	.clk_mult = (m), \
+	.max_freqs = (f), \
+}
+
+static struct cpufreq_compat bmips_cpufreq_compat[] = {
+	BMIPS("brcm,bmips5000", BMIPS5000, 8, 4),
+	BMIPS("brcm,bmips5200", BMIPS5200, 8, 4),
+	{ }
+};
+
+static struct cpufreq_compat *priv;
+
+static int htp_freq_to_cpu_freq(unsigned int clk_mult)
+{
+	return mips_hpt_frequency * clk_mult / 1000;
+}
+
+static struct cpufreq_frequency_table *
+bmips_cpufreq_get_freq_table(const struct cpufreq_policy *policy)
+{
+	struct cpufreq_frequency_table *table;
+	unsigned long cpu_freq;
+	int i;
+
+	cpu_freq = htp_freq_to_cpu_freq(priv->clk_mult);
+
+	table = kmalloc((priv->max_freqs + 1) * sizeof(*table), GFP_KERNEL);
+	if (!table)
+		return ERR_PTR(-ENOMEM);
+
+	for (i = 0; i < priv->max_freqs; i++) {
+		table[i].frequency = cpu_freq / (1 << i);
+		table[i].driver_data = i;
+	}
+	table[i].frequency = CPUFREQ_TABLE_END;
+
+	return table;
+}
+
+static unsigned int bmips_cpufreq_get(unsigned int cpu)
+{
+	unsigned int div;
+	uint32_t mode;
+
+	switch (priv->bmips_type) {
+	case BMIPS5200:
+	case BMIPS5000:
+		mode = read_c0_brcm_mode();
+		div = ((mode >> BMIPS5_CLK_DIV_SHIFT) & BMIPS5_CLK_DIV_MASK);
+		break;
+	default:
+		div = 0;
+	}
+
+	return htp_freq_to_cpu_freq(priv->clk_mult) / (1 << div);
+}
+
+static int bmips_cpufreq_target_index(struct cpufreq_policy *policy,
+				      unsigned int index)
+{
+	unsigned int div = policy->freq_table[index].driver_data;
+
+	switch (priv->bmips_type) {
+	case BMIPS5200:
+	case BMIPS5000:
+		change_c0_brcm_mode(BMIPS5_CLK_DIV_MASK << BMIPS5_CLK_DIV_SHIFT,
+				    (1 << BMIPS5_CLK_DIV_SET_SHIFT) |
+				    (div << BMIPS5_CLK_DIV_SHIFT));
+		break;
+	default:
+		return -ENOTSUPP;
+	}
+
+	return 0;
+}
+
+static int bmips_cpufreq_exit(struct cpufreq_policy *policy)
+{
+	kfree(policy->freq_table);
+
+	return 0;
+}
+
+static int bmips_cpufreq_init(struct cpufreq_policy *policy)
+{
+	struct cpufreq_frequency_table *freq_table;
+	int ret;
+
+	freq_table = bmips_cpufreq_get_freq_table(policy);
+	if (IS_ERR(freq_table)) {
+		ret = PTR_ERR(freq_table);
+		pr_err("%s: couldn't determine frequency table (%d).\n",
+			BMIPS_CPUFREQ_NAME, ret);
+		return ret;
+	}
+
+	ret = cpufreq_generic_init(policy, freq_table, TRANSITION_LATENCY);
+	if (ret)
+		bmips_cpufreq_exit(policy);
+	else
+		pr_info("%s: registered\n", BMIPS_CPUFREQ_NAME);
+
+	return ret;
+}
+
+static struct cpufreq_driver bmips_cpufreq_driver = {
+	.flags		= CPUFREQ_NEED_INITIAL_FREQ_CHECK,
+	.verify		= cpufreq_generic_frequency_table_verify,
+	.target_index	= bmips_cpufreq_target_index,
+	.get		= bmips_cpufreq_get,
+	.init		= bmips_cpufreq_init,
+	.exit		= bmips_cpufreq_exit,
+	.attr		= cpufreq_generic_attr,
+	.name		= BMIPS_CPUFREQ_PREFIX,
+};
+
+static int __init bmips_cpufreq_probe(void)
+{
+	struct cpufreq_compat *cc;
+	struct device_node *np;
+
+	for (cc = bmips_cpufreq_compat; cc->compatible; cc++) {
+		np = of_find_compatible_node(NULL, "cpu", cc->compatible);
+		if (np) {
+			of_node_put(np);
+			priv = cc;
+			break;
+		}
+	}
+
+	/* We hit the guard element of the array. No compatible CPU found. */
+	if (!cc->compatible)
+		return -ENODEV;
+
+	return cpufreq_register_driver(&bmips_cpufreq_driver);
+}
+device_initcall(bmips_cpufreq_probe);
+
+MODULE_AUTHOR("Markus Mayer <mmayer@broadcom.com>");
+MODULE_DESCRIPTION("CPUfreq driver for Broadcom BMIPS SoCs");
+MODULE_LICENSE("GPL");

drivers/cpufreq/brcmstb-avs-cpufreq.c

@@ -878,7 +878,6 @@ static int brcm_avs_prepare_init(struct platform_device *pdev)
 	iounmap(priv->avs_intr_base);
 unmap_base:
 	iounmap(priv->base);
-	platform_set_drvdata(pdev, NULL);
 
 	return ret;
 }
@@ -1042,7 +1041,6 @@ static int brcm_avs_cpufreq_remove(struct platform_device *pdev)
 	priv = platform_get_drvdata(pdev);
 	iounmap(priv->base);
 	iounmap(priv->avs_intr_base);
-	platform_set_drvdata(pdev, NULL);
 
 	return 0;
 }

drivers/cpufreq/cpufreq-dt-platdev.c

@@ -87,8 +87,6 @@ static const struct of_device_id machines[] __initconst = {
 	{ .compatible = "socionext,uniphier-ld11", },
 	{ .compatible = "socionext,uniphier-ld20", },
 
-	{ .compatible = "ti,am33xx", },
-	{ .compatible = "ti,dra7", },
 	{ .compatible = "ti,omap2", },
 	{ .compatible = "ti,omap3", },
 	{ .compatible = "ti,omap4", },

drivers/cpufreq/cpufreq-dt.c

@@ -148,7 +148,6 @@ static int cpufreq_init(struct cpufreq_policy *policy)
 	struct private_data *priv;
 	struct device *cpu_dev;
 	struct clk *cpu_clk;
-	struct dev_pm_opp *suspend_opp;
 	unsigned int transition_latency;
 	bool fallback = false;
 	const char *name;
@@ -252,11 +251,7 @@ static int cpufreq_init(struct cpufreq_policy *policy)
 	policy->driver_data = priv;
 	policy->clk = cpu_clk;
 
-	rcu_read_lock();
-	suspend_opp = dev_pm_opp_get_suspend_opp(cpu_dev);
-	if (suspend_opp)
-		policy->suspend_freq = dev_pm_opp_get_freq(suspend_opp) / 1000;
-	rcu_read_unlock();
+	policy->suspend_freq = dev_pm_opp_get_suspend_opp_freq(cpu_dev) / 1000;
 
 	ret = cpufreq_table_validate_and_show(policy, freq_table);
 	if (ret) {

drivers/cpufreq/cpufreq.c

@@ -1078,15 +1078,11 @@ static struct cpufreq_policy *cpufreq_policy_alloc(unsigned int cpu)
 	return NULL;
 }
 
-static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy, bool notify)
+static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy)
 {
 	struct kobject *kobj;
 	struct completion *cmp;
 
-	if (notify)
-		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
-					     CPUFREQ_REMOVE_POLICY, policy);
-
 	down_write(&policy->rwsem);
 	cpufreq_stats_free_table(policy);
 	kobj = &policy->kobj;
@@ -1104,7 +1100,7 @@ static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy, bool notify)
 	pr_debug("wait complete\n");
 }
 
-static void cpufreq_policy_free(struct cpufreq_policy *policy, bool notify)
+static void cpufreq_policy_free(struct cpufreq_policy *policy)
 {
 	unsigned long flags;
 	int cpu;
@@ -1117,7 +1113,7 @@ static void cpufreq_policy_free(struct cpufreq_policy *policy, bool notify)
 		per_cpu(cpufreq_cpu_data, cpu) = NULL;
 	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
 
-	cpufreq_policy_put_kobj(policy, notify);
+	cpufreq_policy_put_kobj(policy);
 	free_cpumask_var(policy->real_cpus);
 	free_cpumask_var(policy->related_cpus);
 	free_cpumask_var(policy->cpus);
@@ -1170,8 +1166,6 @@ static int cpufreq_online(unsigned int cpu)
 	if (new_policy) {
 		/* related_cpus should at least include policy->cpus. */
 		cpumask_copy(policy->related_cpus, policy->cpus);
-		/* Clear mask of registered CPUs */
-		cpumask_clear(policy->real_cpus);
 	}
 
 	/*
@@ -1244,17 +1238,12 @@ static int cpufreq_online(unsigned int cpu)
 			goto out_exit_policy;
 
 		cpufreq_stats_create_table(policy);
-		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
-				CPUFREQ_CREATE_POLICY, policy);
 
 		write_lock_irqsave(&cpufreq_driver_lock, flags);
 		list_add(&policy->policy_list, &cpufreq_policy_list);
 		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
 	}
 
-	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
-				     CPUFREQ_START, policy);
-
 	ret = cpufreq_init_policy(policy);
 	if (ret) {
 		pr_err("%s: Failed to initialize policy for cpu: %d (%d)\n",
@@ -1282,7 +1271,7 @@ static int cpufreq_online(unsigned int cpu)
 	if (cpufreq_driver->exit)
 		cpufreq_driver->exit(policy);
 out_free_policy:
-	cpufreq_policy_free(policy, !new_policy);
+	cpufreq_policy_free(policy);
 	return ret;
 }
 
@@ -1403,7 +1392,7 @@ static void cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif)
 	remove_cpu_dev_symlink(policy, dev);
 
 	if (cpumask_empty(policy->real_cpus))
-		cpufreq_policy_free(policy, true);
+		cpufreq_policy_free(policy);
 }
 
 /**

drivers/cpufreq/cpufreq_stats.c

@@ -24,9 +24,7 @@ struct cpufreq_stats {
 	unsigned int last_index;
 	u64 *time_in_state;
 	unsigned int *freq_table;
-#ifdef CONFIG_CPU_FREQ_STAT_DETAILS
 	unsigned int *trans_table;
-#endif
 };
 
 static int cpufreq_stats_update(struct cpufreq_stats *stats)
@@ -45,9 +43,7 @@ static void cpufreq_stats_clear_table(struct cpufreq_stats *stats)
 	unsigned int count = stats->max_state;
 
 	memset(stats->time_in_state, 0, count * sizeof(u64));
-#ifdef CONFIG_CPU_FREQ_STAT_DETAILS
 	memset(stats->trans_table, 0, count * count * sizeof(int));
-#endif
 	stats->last_time = get_jiffies_64();
 	stats->total_trans = 0;
 }
@@ -83,7 +79,6 @@ static ssize_t store_reset(struct cpufreq_policy *policy, const char *buf,
 	return count;
 }
 
-#ifdef CONFIG_CPU_FREQ_STAT_DETAILS
 static ssize_t show_trans_table(struct cpufreq_policy *policy, char *buf)
 {
 	struct cpufreq_stats *stats = policy->stats;
@@ -128,7 +123,6 @@ static ssize_t show_trans_table(struct cpufreq_policy *policy, char *buf)
 	return len;
 }
 cpufreq_freq_attr_ro(trans_table);
-#endif
 
 cpufreq_freq_attr_ro(total_trans);
 cpufreq_freq_attr_ro(time_in_state);
@@ -138,9 +132,7 @@ static struct attribute *default_attrs[] = {
 	&total_trans.attr,
 	&time_in_state.attr,
 	&reset.attr,
-#ifdef CONFIG_CPU_FREQ_STAT_DETAILS
 	&trans_table.attr,
-#endif
 	NULL
 };
 static struct attribute_group stats_attr_group = {
@@ -199,9 +191,7 @@ void cpufreq_stats_create_table(struct cpufreq_policy *policy)
 
 	alloc_size = count * sizeof(int) + count * sizeof(u64);
 
-#ifdef CONFIG_CPU_FREQ_STAT_DETAILS
 	alloc_size += count * count * sizeof(int);
-#endif
 
 	/* Allocate memory for time_in_state/freq_table/trans_table in one go */
 	stats->time_in_state = kzalloc(alloc_size, GFP_KERNEL);
@@ -210,9 +200,7 @@ void cpufreq_stats_create_table(struct cpufreq_policy *policy)
 
 	stats->freq_table = (unsigned int *)(stats->time_in_state + count);
 
-#ifdef CONFIG_CPU_FREQ_STAT_DETAILS
 	stats->trans_table = stats->freq_table + count;
-#endif
 
 	stats->max_state = count;
 
@@ -258,8 +246,6 @@ void cpufreq_stats_record_transition(struct cpufreq_policy *policy,
 	cpufreq_stats_update(stats);
 
 	stats->last_index = new_index;
-#ifdef CONFIG_CPU_FREQ_STAT_DETAILS
 	stats->trans_table[old_index * stats->max_state + new_index]++;
-#endif
 	stats->total_trans++;
 }

drivers/cpufreq/exynos5440-cpufreq.c

@@ -118,12 +118,10 @@ static int init_div_table(void)
 	unsigned int tmp, clk_div, ema_div, freq, volt_id;
 	struct dev_pm_opp *opp;
 
-	rcu_read_lock();
 	cpufreq_for_each_entry(pos, freq_tbl) {
 		opp = dev_pm_opp_find_freq_exact(dvfs_info->dev,
 					pos->frequency * 1000, true);
 		if (IS_ERR(opp)) {
-			rcu_read_unlock();
 			dev_err(dvfs_info->dev,
 				"failed to find valid OPP for %u KHZ\n",
 				pos->frequency);
@@ -140,6 +138,7 @@ static int init_div_table(void)
 
 		/* Calculate EMA */
 		volt_id = dev_pm_opp_get_voltage(opp);
+
 		volt_id = (MAX_VOLTAGE - volt_id) / VOLTAGE_STEP;
 		if (volt_id < PMIC_HIGH_VOLT) {
 			ema_div = (CPUEMA_HIGH << P0_7_CPUEMA_SHIFT) |
@@ -157,9 +156,9 @@ static int init_div_table(void)
 
 		__raw_writel(tmp, dvfs_info->base + XMU_PMU_P0_7 + 4 *
 						(pos - freq_tbl));
+		dev_pm_opp_put(opp);
 	}
 
-	rcu_read_unlock();
 	return 0;
 }
 

drivers/cpufreq/imx6q-cpufreq.c

@@ -53,16 +53,15 @@ static int imx6q_set_target(struct cpufreq_policy *policy, unsigned int index)
 	freq_hz = new_freq * 1000;
 	old_freq = clk_get_rate(arm_clk) / 1000;
 
-	rcu_read_lock();
 	opp = dev_pm_opp_find_freq_ceil(cpu_dev, &freq_hz);
 	if (IS_ERR(opp)) {
-		rcu_read_unlock();
 		dev_err(cpu_dev, "failed to find OPP for %ld\n", freq_hz);
 		return PTR_ERR(opp);
 	}
 
 	volt = dev_pm_opp_get_voltage(opp);
-	rcu_read_unlock();
+	dev_pm_opp_put(opp);
+
 	volt_old = regulator_get_voltage(arm_reg);
 
 	dev_dbg(cpu_dev, "%u MHz, %ld mV --> %u MHz, %ld mV\n",
@@ -321,14 +320,15 @@ static int imx6q_cpufreq_probe(struct platform_device *pdev)
 	 * freq_table initialised from OPP is therefore sorted in the
 	 * same order.
 	 */
-	rcu_read_lock();
 	opp = dev_pm_opp_find_freq_exact(cpu_dev,
 				  freq_table[0].frequency * 1000, true);
 	min_volt = dev_pm_opp_get_voltage(opp);
+	dev_pm_opp_put(opp);
 	opp = dev_pm_opp_find_freq_exact(cpu_dev,
 				  freq_table[--num].frequency * 1000, true);
 	max_volt = dev_pm_opp_get_voltage(opp);
-	rcu_read_unlock();
+	dev_pm_opp_put(opp);
+
 	ret = regulator_set_voltage_time(arm_reg, min_volt, max_volt);
 	if (ret > 0)
 		transition_latency += ret * 1000;

drivers/cpufreq/intel_pstate.c

@@ -358,6 +358,8 @@ static struct pstate_funcs pstate_funcs __read_mostly;
 static int hwp_active __read_mostly;
 static bool per_cpu_limits __read_mostly;
 
+static bool driver_registered __read_mostly;
+
 #ifdef CONFIG_ACPI
 static bool acpi_ppc;
 #endif
@@ -394,6 +396,7 @@ static struct perf_limits *limits = &performance_limits;
 static struct perf_limits *limits = &powersave_limits;
 #endif
 
+static DEFINE_MUTEX(intel_pstate_driver_lock);
 static DEFINE_MUTEX(intel_pstate_limits_lock);
 
 #ifdef CONFIG_ACPI
@@ -538,7 +541,6 @@ static void intel_pstate_exit_perf_limits(struct cpufreq_policy *policy)
 
 	acpi_processor_unregister_performance(policy->cpu);
 }
-
 #else
 static inline void intel_pstate_init_acpi_perf_limits(struct cpufreq_policy *policy)
 {
@@ -873,6 +875,9 @@ static void intel_pstate_hwp_set(struct cpufreq_policy *policy)
 
 		rdmsrl_on_cpu(cpu, MSR_HWP_CAPABILITIES, &cap);
 		hw_min = HWP_LOWEST_PERF(cap);
+		if (limits->no_turbo)
+			hw_max = HWP_GUARANTEED_PERF(cap);
+		else
 			hw_max = HWP_HIGHEST_PERF(cap);
 		range = hw_max - hw_min;
 
@@ -887,11 +892,6 @@ static void intel_pstate_hwp_set(struct cpufreq_policy *policy)
 
 		adj_range = max_perf_pct * range / 100;
 		max = hw_min + adj_range;
-		if (limits->no_turbo) {
-			hw_max = HWP_GUARANTEED_PERF(cap);
-			if (hw_max < max)
-				max = hw_max;
-		}
 
 		value &= ~HWP_MAX_PERF(~0L);
 		value |= HWP_MAX_PERF(max);
@@ -1007,37 +1007,59 @@ static int pid_param_get(void *data, u64 *val)
 }
 DEFINE_SIMPLE_ATTRIBUTE(fops_pid_param, pid_param_get, pid_param_set, "%llu\n");
 
+static struct dentry *debugfs_parent;
+
 struct pid_param {
 	char *name;
 	void *value;
+	struct dentry *dentry;
 };
 
 static struct pid_param pid_files[] = {
-	{"sample_rate_ms", &pid_params.sample_rate_ms},
-	{"d_gain_pct", &pid_params.d_gain_pct},
-	{"i_gain_pct", &pid_params.i_gain_pct},
-	{"deadband", &pid_params.deadband},
-	{"setpoint", &pid_params.setpoint},
-	{"p_gain_pct", &pid_params.p_gain_pct},
-	{NULL, NULL}
+	{"sample_rate_ms", &pid_params.sample_rate_ms, },
+	{"d_gain_pct", &pid_params.d_gain_pct, },
+	{"i_gain_pct", &pid_params.i_gain_pct, },
+	{"deadband", &pid_params.deadband, },
+	{"setpoint", &pid_params.setpoint, },
+	{"p_gain_pct", &pid_params.p_gain_pct, },
+	{NULL, NULL, }
 };
 
-static void __init intel_pstate_debug_expose_params(void)
+static void intel_pstate_debug_expose_params(void)
 {
-	struct dentry *debugfs_parent;
-	int i = 0;
+	int i;
 
 	debugfs_parent = debugfs_create_dir("pstate_snb", NULL);
 	if (IS_ERR_OR_NULL(debugfs_parent))
 		return;
-	while (pid_files[i].name) {
-		debugfs_create_file(pid_files[i].name, 0660,
+
+	for (i = 0; pid_files[i].name; i++) {
+		struct dentry *dentry;
+
+		dentry = debugfs_create_file(pid_files[i].name, 0660,
 					     debugfs_parent, pid_files[i].value,
 					     &fops_pid_param);
-		i++;
+		if (!IS_ERR(dentry))
+			pid_files[i].dentry = dentry;
 	}
 }
 
+static void intel_pstate_debug_hide_params(void)
+{
+	int i;
+
+	if (IS_ERR_OR_NULL(debugfs_parent))
+		return;
+
+	for (i = 0; pid_files[i].name; i++) {
+		debugfs_remove(pid_files[i].dentry);
+		pid_files[i].dentry = NULL;
+	}
+
+	debugfs_remove(debugfs_parent);
+	debugfs_parent = NULL;
+}
+
 /************************** debugfs end ************************/
 
 /************************** sysfs begin ************************/
@@ -1048,6 +1070,34 @@ static void __init intel_pstate_debug_expose_params(void)
 		return sprintf(buf, "%u\n", limits->object);		\
 	}
 
+static ssize_t intel_pstate_show_status(char *buf);
+static int intel_pstate_update_status(const char *buf, size_t size);
+
+static ssize_t show_status(struct kobject *kobj,
+			   struct attribute *attr, char *buf)
+{
+	ssize_t ret;
+
+	mutex_lock(&intel_pstate_driver_lock);
+	ret = intel_pstate_show_status(buf);
+	mutex_unlock(&intel_pstate_driver_lock);
+
+	return ret;
+}
+
+static ssize_t store_status(struct kobject *a, struct attribute *b,
+			    const char *buf, size_t count)
+{
+	char *p = memchr(buf, '\n', count);
+	int ret;
+
+	mutex_lock(&intel_pstate_driver_lock);
+	ret = intel_pstate_update_status(buf, p ? p - buf : count);
+	mutex_unlock(&intel_pstate_driver_lock);
+
+	return ret < 0 ? ret : count;
+}
+
 static ssize_t show_turbo_pct(struct kobject *kobj,
 				struct attribute *attr, char *buf)
 {
@@ -1055,12 +1105,22 @@ static ssize_t show_turbo_pct(struct kobject *kobj,
 	int total, no_turbo, turbo_pct;
 	uint32_t turbo_fp;
 
+	mutex_lock(&intel_pstate_driver_lock);
+
+	if (!driver_registered) {
+		mutex_unlock(&intel_pstate_driver_lock);
+		return -EAGAIN;
+	}
+
 	cpu = all_cpu_data[0];
 
 	total = cpu->pstate.turbo_pstate - cpu->pstate.min_pstate + 1;
 	no_turbo = cpu->pstate.max_pstate - cpu->pstate.min_pstate + 1;
 	turbo_fp = div_fp(no_turbo, total);
 	turbo_pct = 100 - fp_toint(mul_fp(turbo_fp, int_tofp(100)));
+
+	mutex_unlock(&intel_pstate_driver_lock);
+
 	return sprintf(buf, "%u\n", turbo_pct);
 }
 
@@ -1070,8 +1130,18 @@ static ssize_t show_num_pstates(struct kobject *kobj,
 	struct cpudata *cpu;
 	int total;
 
+	mutex_lock(&intel_pstate_driver_lock);
+
+	if (!driver_registered) {
+		mutex_unlock(&intel_pstate_driver_lock);
+		return -EAGAIN;
+	}
+
 	cpu = all_cpu_data[0];
 	total = cpu->pstate.turbo_pstate - cpu->pstate.min_pstate + 1;
+
+	mutex_unlock(&intel_pstate_driver_lock);
+
 	return sprintf(buf, "%u\n", total);
 }
 
@@ -1080,12 +1150,21 @@ static ssize_t show_no_turbo(struct kobject *kobj,
 {
 	ssize_t ret;
 
+	mutex_lock(&intel_pstate_driver_lock);
+
+	if (!driver_registered) {
+		mutex_unlock(&intel_pstate_driver_lock);
+		return -EAGAIN;
+	}
+
 	update_turbo_state();
 	if (limits->turbo_disabled)
 		ret = sprintf(buf, "%u\n", limits->turbo_disabled);
 	else
 		ret = sprintf(buf, "%u\n", limits->no_turbo);
 
+	mutex_unlock(&intel_pstate_driver_lock);
+
 	return ret;
 }
 
@@ -1099,12 +1178,20 @@ static ssize_t store_no_turbo(struct kobject *a, struct attribute *b,
 	if (ret != 1)
 		return -EINVAL;
 
+	mutex_lock(&intel_pstate_driver_lock);
+
+	if (!driver_registered) {
+		mutex_unlock(&intel_pstate_driver_lock);
+		return -EAGAIN;
+	}
+
 	mutex_lock(&intel_pstate_limits_lock);
 
 	update_turbo_state();
 	if (limits->turbo_disabled) {
 		pr_warn("Turbo disabled by BIOS or unavailable on processor\n");
 		mutex_unlock(&intel_pstate_limits_lock);
+		mutex_unlock(&intel_pstate_driver_lock);
 		return -EPERM;
 	}
 
@@ -1114,6 +1201,8 @@ static ssize_t store_no_turbo(struct kobject *a, struct attribute *b,
 
 	intel_pstate_update_policies();
 
+	mutex_unlock(&intel_pstate_driver_lock);
+
 	return count;
 }
 
@@ -1127,6 +1216,13 @@ static ssize_t store_max_perf_pct(struct kobject *a, struct attribute *b,
 	if (ret != 1)
 		return -EINVAL;
 
+	mutex_lock(&intel_pstate_driver_lock);
+
+	if (!driver_registered) {
+		mutex_unlock(&intel_pstate_driver_lock);
+		return -EAGAIN;
+	}
+
 	mutex_lock(&intel_pstate_limits_lock);
 
 	limits->max_sysfs_pct = clamp_t(int, input, 0 , 100);
@@ -1142,6 +1238,8 @@ static ssize_t store_max_perf_pct(struct kobject *a, struct attribute *b,
 
 	intel_pstate_update_policies();
 
+	mutex_unlock(&intel_pstate_driver_lock);
+
 	return count;
 }
 
@@ -1155,6 +1253,13 @@ static ssize_t store_min_perf_pct(struct kobject *a, struct attribute *b,
 	if (ret != 1)
 		return -EINVAL;
 
+	mutex_lock(&intel_pstate_driver_lock);
+
+	if (!driver_registered) {
+		mutex_unlock(&intel_pstate_driver_lock);
+		return -EAGAIN;
+	}
+
 	mutex_lock(&intel_pstate_limits_lock);
 
 	limits->min_sysfs_pct = clamp_t(int, input, 0 , 100);
@@ -1170,12 +1275,15 @@ static ssize_t store_min_perf_pct(struct kobject *a, struct attribute *b,
 
 	intel_pstate_update_policies();
 
+	mutex_unlock(&intel_pstate_driver_lock);
+
 	return count;
 }
 
 show_one(max_perf_pct, max_perf_pct);
 show_one(min_perf_pct, min_perf_pct);
 
+define_one_global_rw(status);
 define_one_global_rw(no_turbo);
 define_one_global_rw(max_perf_pct);
 define_one_global_rw(min_perf_pct);
@@ -1183,6 +1291,7 @@ define_one_global_ro(turbo_pct);
 define_one_global_ro(num_pstates);
 
 static struct attribute *intel_pstate_attributes[] = {
+	&status.attr,
 	&no_turbo.attr,
 	&turbo_pct.attr,
 	&num_pstates.attr,
@@ -1364,48 +1473,71 @@ static int core_get_max_pstate_physical(void)
 	return (value >> 8) & 0xFF;
 }
 
-static int core_get_max_pstate(void)
+static int core_get_tdp_ratio(u64 plat_info)
 {
-	u64 tar;
-	u64 plat_info;
-	int max_pstate;
-	int err;
-
-	rdmsrl(MSR_PLATFORM_INFO, plat_info);
-	max_pstate = (plat_info >> 8) & 0xFF;
-
-	err = rdmsrl_safe(MSR_TURBO_ACTIVATION_RATIO, &tar);
-	if (!err) {
-		/* Do some sanity checking for safety */
+	/* Check how many TDP levels present */
 	if (plat_info & 0x600000000) {
 		u64 tdp_ctrl;
 		u64 tdp_ratio;
 		int tdp_msr;
+		int err;
 
+		/* Get the TDP level (0, 1, 2) to get ratios */
 		err = rdmsrl_safe(MSR_CONFIG_TDP_CONTROL, &tdp_ctrl);
 		if (err)
-				goto skip_tar;
+			return err;
 
-			tdp_msr = MSR_CONFIG_TDP_NOMINAL + (tdp_ctrl & 0x3);
+		/* TDP MSR are continuous starting at 0x648 */
+		tdp_msr = MSR_CONFIG_TDP_NOMINAL + (tdp_ctrl & 0x03);
 		err = rdmsrl_safe(tdp_msr, &tdp_ratio);
 		if (err)
-				goto skip_tar;
+			return err;
 
 		/* For level 1 and 2, bits[23:16] contain the ratio */
-			if (tdp_ctrl)
+		if (tdp_ctrl & 0x03)
 			tdp_ratio >>= 16;
 
 		tdp_ratio &= 0xff; /* ratios are only 8 bits long */
-			if (tdp_ratio - 1 == tar) {
-				max_pstate = tar;
-				pr_debug("max_pstate=TAC %x\n", max_pstate);
-			} else {
-				goto skip_tar;
+		pr_debug("tdp_ratio %x\n", (int)tdp_ratio);
+
+		return (int)tdp_ratio;
 	}
+
+	return -ENXIO;
+}
+
+static int core_get_max_pstate(void)
+{
+	u64 tar;
+	u64 plat_info;
+	int max_pstate;
+	int tdp_ratio;
+	int err;
+
+	rdmsrl(MSR_PLATFORM_INFO, plat_info);
+	max_pstate = (plat_info >> 8) & 0xFF;
+
+	tdp_ratio = core_get_tdp_ratio(plat_info);
+	if (tdp_ratio <= 0)
+		return max_pstate;
+
+	if (hwp_active) {
+		/* Turbo activation ratio is not used on HWP platforms */
+		return tdp_ratio;
+	}
+
+	err = rdmsrl_safe(MSR_TURBO_ACTIVATION_RATIO, &tar);
+	if (!err) {
+		int tar_levels;
+
+		/* Do some sanity checking for safety */
+		tar_levels = tar & 0xff;
+		if (tdp_ratio - 1 == tar_levels) {
+			max_pstate = tar_levels;
+			pr_debug("max_pstate=TAC %x\n", max_pstate);
 		}
 	}
 
-skip_tar:
 	return max_pstate;
 }
 
@@ -2072,6 +2204,20 @@ static int intel_pstate_set_policy(struct cpufreq_policy *policy)
 
 static int intel_pstate_verify_policy(struct cpufreq_policy *policy)
 {
+	struct cpudata *cpu = all_cpu_data[policy->cpu];
+	struct perf_limits *perf_limits;
+
+	if (policy->policy == CPUFREQ_POLICY_PERFORMANCE)
+		perf_limits = &performance_limits;
+	else
+		perf_limits = &powersave_limits;
+
+	update_turbo_state();
+	policy->cpuinfo.max_freq = perf_limits->turbo_disabled ||
+					perf_limits->no_turbo ?
+					cpu->pstate.max_freq :
+					cpu->pstate.turbo_freq;
+
 	cpufreq_verify_within_cpu_limits(policy);
 
 	if (policy->policy != CPUFREQ_POLICY_POWERSAVE &&
@@ -2299,6 +2445,111 @@ static struct cpufreq_driver intel_cpufreq = {
 
 static struct cpufreq_driver *intel_pstate_driver = &intel_pstate;
 
+static void intel_pstate_driver_cleanup(void)
+{
+	unsigned int cpu;
+
+	get_online_cpus();
+	for_each_online_cpu(cpu) {
+		if (all_cpu_data[cpu]) {
+			if (intel_pstate_driver == &intel_pstate)
+				intel_pstate_clear_update_util_hook(cpu);
+
+			kfree(all_cpu_data[cpu]);
+			all_cpu_data[cpu] = NULL;
+		}
+	}
+	put_online_cpus();
+}
+
+static int intel_pstate_register_driver(void)
+{
+	int ret;
+
+	ret = cpufreq_register_driver(intel_pstate_driver);
+	if (ret) {
+		intel_pstate_driver_cleanup();
+		return ret;
+	}
+
+	mutex_lock(&intel_pstate_limits_lock);
+	driver_registered = true;
+	mutex_unlock(&intel_pstate_limits_lock);
+
+	if (intel_pstate_driver == &intel_pstate && !hwp_active &&
+	    pstate_funcs.get_target_pstate != get_target_pstate_use_cpu_load)
+		intel_pstate_debug_expose_params();
+
+	return 0;
+}
+
+static int intel_pstate_unregister_driver(void)
+{
+	if (hwp_active)
+		return -EBUSY;
+
+	if (intel_pstate_driver == &intel_pstate && !hwp_active &&
+	    pstate_funcs.get_target_pstate != get_target_pstate_use_cpu_load)
+		intel_pstate_debug_hide_params();
+
+	mutex_lock(&intel_pstate_limits_lock);
+	driver_registered = false;
+	mutex_unlock(&intel_pstate_limits_lock);
+
+	cpufreq_unregister_driver(intel_pstate_driver);
+	intel_pstate_driver_cleanup();
+
+	return 0;
+}
+
+static ssize_t intel_pstate_show_status(char *buf)
+{
+	if (!driver_registered)
+		return sprintf(buf, "off\n");
+
+	return sprintf(buf, "%s\n", intel_pstate_driver == &intel_pstate ?
+					"active" : "passive");
+}
+
+static int intel_pstate_update_status(const char *buf, size_t size)
+{
+	int ret;
+
+	if (size == 3 && !strncmp(buf, "off", size))
+		return driver_registered ?
+			intel_pstate_unregister_driver() : -EINVAL;
+
+	if (size == 6 && !strncmp(buf, "active", size)) {
+		if (driver_registered) {
+			if (intel_pstate_driver == &intel_pstate)
+				return 0;
+
+			ret = intel_pstate_unregister_driver();
+			if (ret)
+				return ret;
+		}
+
+		intel_pstate_driver = &intel_pstate;
+		return intel_pstate_register_driver();
+	}
+
+	if (size == 7 && !strncmp(buf, "passive", size)) {
+		if (driver_registered) {
+			if (intel_pstate_driver != &intel_pstate)
+				return 0;
+
+			ret = intel_pstate_unregister_driver();
+			if (ret)
+				return ret;
+		}
+
+		intel_pstate_driver = &intel_cpufreq;
+		return intel_pstate_register_driver();
+	}
+
+	return -EINVAL;
+}
+
 static int no_load __initdata;
 static int no_hwp __initdata;
 static int hwp_only __initdata;
@@ -2486,9 +2737,9 @@ static const struct x86_cpu_id hwp_support_ids[] __initconst = {
 
 static int __init intel_pstate_init(void)
 {
-	int cpu, rc = 0;
 	const struct x86_cpu_id *id;
 	struct cpu_defaults *cpu_def;
+	int rc = 0;
 
 	if (no_load)
 		return -ENODEV;
@@ -2520,45 +2771,29 @@ static int __init intel_pstate_init(void)
 	if (intel_pstate_platform_pwr_mgmt_exists())
 		return -ENODEV;
 
+	if (!hwp_active && hwp_only)
+		return -ENOTSUPP;
+
 	pr_info("Intel P-state driver initializing\n");
 
 	all_cpu_data = vzalloc(sizeof(void *) * num_possible_cpus());
 	if (!all_cpu_data)
 		return -ENOMEM;
 
-	if (!hwp_active && hwp_only)
-		goto out;
-
 	intel_pstate_request_control_from_smm();
 
-	rc = cpufreq_register_driver(intel_pstate_driver);
-	if (rc)
-		goto out;
-
-	if (intel_pstate_driver == &intel_pstate && !hwp_active &&
-	    pstate_funcs.get_target_pstate != get_target_pstate_use_cpu_load)
-		intel_pstate_debug_expose_params();
-
 	intel_pstate_sysfs_expose_params();
 
-	if (hwp_active)
-		pr_info("HWP enabled\n");
-
+	mutex_lock(&intel_pstate_driver_lock);
+	rc = intel_pstate_register_driver();
+	mutex_unlock(&intel_pstate_driver_lock);
+	if (rc)
 		return rc;
-out:
-	get_online_cpus();
-	for_each_online_cpu(cpu) {
-		if (all_cpu_data[cpu]) {
-			if (intel_pstate_driver == &intel_pstate)
-				intel_pstate_clear_update_util_hook(cpu);
 
-			kfree(all_cpu_data[cpu]);
-		}
-	}
+	if (hwp_active)
+		pr_info("HWP enabled\n");
 
-	put_online_cpus();
-	vfree(all_cpu_data);
-	return -ENODEV;
+	return 0;
 }
 device_initcall(intel_pstate_init);
 

drivers/cpufreq/mt8173-cpufreq.c

@@ -232,16 +232,14 @@ static int mtk_cpufreq_set_target(struct cpufreq_policy *policy,
 
 	freq_hz = freq_table[index].frequency * 1000;
 
-	rcu_read_lock();
 	opp = dev_pm_opp_find_freq_ceil(cpu_dev, &freq_hz);
 	if (IS_ERR(opp)) {
-		rcu_read_unlock();
 		pr_err("cpu%d: failed to find OPP for %ld\n",
 		       policy->cpu, freq_hz);
 		return PTR_ERR(opp);
 	}
 	vproc = dev_pm_opp_get_voltage(opp);
-	rcu_read_unlock();
+	dev_pm_opp_put(opp);
 
 	/*
 	 * If the new voltage or the intermediate voltage is higher than the
@@ -411,16 +409,14 @@ static int mtk_cpu_dvfs_info_init(struct mtk_cpu_dvfs_info *info, int cpu)
 
 	/* Search a safe voltage for intermediate frequency. */
 	rate = clk_get_rate(inter_clk);
-	rcu_read_lock();
 	opp = dev_pm_opp_find_freq_ceil(cpu_dev, &rate);
 	if (IS_ERR(opp)) {
-		rcu_read_unlock();
 		pr_err("failed to get intermediate opp for cpu%d\n", cpu);
 		ret = PTR_ERR(opp);
 		goto out_free_opp_table;
 	}
 	info->intermediate_voltage = dev_pm_opp_get_voltage(opp);
-	rcu_read_unlock();
+	dev_pm_opp_put(opp);
 
 	info->cpu_dev = cpu_dev;
 	info->proc_reg = proc_reg;

drivers/cpufreq/omap-cpufreq.c

@@ -63,16 +63,14 @@ static int omap_target(struct cpufreq_policy *policy, unsigned int index)
 	freq = ret;
 
 	if (mpu_reg) {
-		rcu_read_lock();
 		opp = dev_pm_opp_find_freq_ceil(mpu_dev, &freq);
 		if (IS_ERR(opp)) {
-			rcu_read_unlock();
 			dev_err(mpu_dev, "%s: unable to find MPU OPP for %d\n",
 				__func__, new_freq);
 			return -EINVAL;
 		}
 		volt = dev_pm_opp_get_voltage(opp);
-		rcu_read_unlock();
+		dev_pm_opp_put(opp);
 		tol = volt * OPP_TOLERANCE / 100;
 		volt_old = regulator_get_voltage(mpu_reg);
 	}

drivers/cpufreq/powernv-cpufreq.c

@@ -144,6 +144,7 @@ static struct powernv_pstate_info {
 	unsigned int max;
 	unsigned int nominal;
 	unsigned int nr_pstates;
+	bool wof_enabled;
 } powernv_pstate_info;
 
 /* Use following macros for conversions between pstate_id and index */
@@ -203,6 +204,7 @@ static int init_powernv_pstates(void)
 	const __be32 *pstate_ids, *pstate_freqs;
 	u32 len_ids, len_freqs;
 	u32 pstate_min, pstate_max, pstate_nominal;
+	u32 pstate_turbo, pstate_ultra_turbo;
 
 	power_mgt = of_find_node_by_path("/ibm,opal/power-mgt");
 	if (!power_mgt) {
@@ -225,8 +227,29 @@ static int init_powernv_pstates(void)
 		pr_warn("ibm,pstate-nominal not found\n");
 		return -ENODEV;
 	}
+
+	if (of_property_read_u32(power_mgt, "ibm,pstate-ultra-turbo",
+				 &pstate_ultra_turbo)) {
+		powernv_pstate_info.wof_enabled = false;
+		goto next;
+	}
+
+	if (of_property_read_u32(power_mgt, "ibm,pstate-turbo",
+				 &pstate_turbo)) {
+		powernv_pstate_info.wof_enabled = false;
+		goto next;
+	}
+
+	if (pstate_turbo == pstate_ultra_turbo)
+		powernv_pstate_info.wof_enabled = false;
+	else
+		powernv_pstate_info.wof_enabled = true;
+
+next:
 	pr_info("cpufreq pstate min %d nominal %d max %d\n", pstate_min,
 		pstate_nominal, pstate_max);
+	pr_info("Workload Optimized Frequency is %s in the platform\n",
+		(powernv_pstate_info.wof_enabled) ? "enabled" : "disabled");
 
 	pstate_ids = of_get_property(power_mgt, "ibm,pstate-ids", &len_ids);
 	if (!pstate_ids) {
@@ -268,6 +291,13 @@ static int init_powernv_pstates(void)
 			powernv_pstate_info.nominal = i;
 		else if (id == pstate_min)
 			powernv_pstate_info.min = i;
+
+		if (powernv_pstate_info.wof_enabled && id == pstate_turbo) {
+			int j;
+
+			for (j = i - 1; j >= (int)powernv_pstate_info.max; j--)
+				powernv_freqs[j].flags = CPUFREQ_BOOST_FREQ;
+		}
 	}
 
 	/* End of list marker entry */
@@ -305,9 +335,12 @@ static ssize_t cpuinfo_nominal_freq_show(struct cpufreq_policy *policy,
 struct freq_attr cpufreq_freq_attr_cpuinfo_nominal_freq =
 	__ATTR_RO(cpuinfo_nominal_freq);
 
+#define SCALING_BOOST_FREQS_ATTR_INDEX		2
+
 static struct freq_attr *powernv_cpu_freq_attr[] = {
 	&cpufreq_freq_attr_scaling_available_freqs,
 	&cpufreq_freq_attr_cpuinfo_nominal_freq,
+	&cpufreq_freq_attr_scaling_boost_freqs,
 	NULL,
 };
 
@@ -1013,11 +1046,22 @@ static int __init powernv_cpufreq_init(void)
 	register_reboot_notifier(&powernv_cpufreq_reboot_nb);
 	opal_message_notifier_register(OPAL_MSG_OCC, &powernv_cpufreq_opal_nb);
 
-	rc = cpufreq_register_driver(&powernv_cpufreq_driver);
-	if (!rc)
-		return 0;
+	if (powernv_pstate_info.wof_enabled)
+		powernv_cpufreq_driver.boost_enabled = true;
+	else
+		powernv_cpu_freq_attr[SCALING_BOOST_FREQS_ATTR_INDEX] = NULL;
 
+	rc = cpufreq_register_driver(&powernv_cpufreq_driver);
+	if (rc) {
 		pr_info("Failed to register the cpufreq driver (%d)\n", rc);
+		goto cleanup_notifiers;
+	}
+
+	if (powernv_pstate_info.wof_enabled)
+		cpufreq_enable_boost_support();
+
+	return 0;
+cleanup_notifiers:
 	unregister_all_notifiers();
 	clean_chip_info();
 out:

drivers/cpufreq/ppc_cbe_cpufreq_pmi.c

@@ -100,9 +100,6 @@ static int pmi_notifier(struct notifier_block *nb,
 	/* Should this really be called for CPUFREQ_ADJUST and CPUFREQ_NOTIFY
 	 * policy events?)
 	 */
-	if (event == CPUFREQ_START)
-		return 0;
-
 	node = cbe_cpu_to_node(policy->cpu);
 
 	pr_debug("got notified, event=%lu, node=%u\n", event, node);

drivers/cpufreq/qoriq-cpufreq.c

@@ -11,6 +11,7 @@
 #define pr_fmt(fmt)	KBUILD_MODNAME ": " fmt
 
 #include <linux/clk.h>
+#include <linux/clk-provider.h>
 #include <linux/cpufreq.h>
 #include <linux/cpu_cooling.h>
 #include <linux/errno.h>
@@ -37,53 +38,20 @@ struct cpu_data {
 	struct thermal_cooling_device *cdev;
 };
 
+/*
+ * Don't use cpufreq on this SoC -- used when the SoC would have otherwise
+ * matched a more generic compatible.
+ */
+#define SOC_BLACKLIST		1
+
 /**
  * struct soc_data - SoC specific data
- * @freq_mask: mask the disallowed frequencies
- * @flag: unique flags
+ * @flags: SOC_xxx
  */
 struct soc_data {
-	u32 freq_mask[4];
-	u32 flag;
-};
-
-#define FREQ_MASK	1
-/* see hardware specification for the allowed frqeuencies */
-static const struct soc_data sdata[] = {
-	{ /* used by p2041 and p3041 */
-		.freq_mask = {0x8, 0x8, 0x2, 0x2},
-		.flag = FREQ_MASK,
-	},
-	{ /* used by p5020 */
-		.freq_mask = {0x8, 0x2},
-		.flag = FREQ_MASK,
-	},
-	{ /* used by p4080, p5040 */
-		.freq_mask = {0},
-		.flag = 0,
-	},
+	u32 flags;
 };
 
-/*
- * the minimum allowed core frequency, in Hz
- * for chassis v1.0, >= platform frequency
- * for chassis v2.0, >= platform frequency / 2
- */
-static u32 min_cpufreq;
-static const u32 *fmask;
-
-#if defined(CONFIG_ARM)
-static int get_cpu_physical_id(int cpu)
-{
-	return topology_core_id(cpu);
-}
-#else
-static int get_cpu_physical_id(int cpu)
-{
-	return get_hard_smp_processor_id(cpu);
-}
-#endif
-
 static u32 get_bus_freq(void)
 {
 	struct device_node *soc;
@@ -101,9 +69,10 @@ static u32 get_bus_freq(void)
 	return sysfreq;
 }
 
-static struct device_node *cpu_to_clk_node(int cpu)
+static struct clk *cpu_to_clk(int cpu)
 {
-	struct device_node *np, *clk_np;
+	struct device_node *np;
+	struct clk *clk;
 
 	if (!cpu_present(cpu))
 		return NULL;
@@ -112,37 +81,28 @@ static struct device_node *cpu_to_clk_node(int cpu)
 	if (!np)
 		return NULL;
 
-	clk_np = of_parse_phandle(np, "clocks", 0);
-	if (!clk_np)
-		return NULL;
-
+	clk = of_clk_get(np, 0);
 	of_node_put(np);
-
-	return clk_np;
+	return clk;
 }
 
 /* traverse cpu nodes to get cpu mask of sharing clock wire */
 static void set_affected_cpus(struct cpufreq_policy *policy)
 {
-	struct device_node *np, *clk_np;
 	struct cpumask *dstp = policy->cpus;
+	struct clk *clk;
 	int i;
 
-	np = cpu_to_clk_node(policy->cpu);
-	if (!np)
-		return;
-
 	for_each_present_cpu(i) {
-		clk_np = cpu_to_clk_node(i);
-		if (!clk_np)
+		clk = cpu_to_clk(i);
+		if (IS_ERR(clk)) {
+			pr_err("%s: no clock for cpu %d\n", __func__, i);
 			continue;
+		}
 
-		if (clk_np == np)
+		if (clk_is_match(policy->clk, clk))
 			cpumask_set_cpu(i, dstp);
-
-		of_node_put(clk_np);
 	}
-	of_node_put(np);
 }
 
 /* reduce the duplicated frequencies in frequency table */
@@ -198,10 +158,11 @@ static void freq_table_sort(struct cpufreq_frequency_table *freq_table,
 
 static int qoriq_cpufreq_cpu_init(struct cpufreq_policy *policy)
 {
-	struct device_node *np, *pnode;
+	struct device_node *np;
 	int i, count, ret;
-	u32 freq, mask;
+	u32 freq;
 	struct clk *clk;
+	const struct clk_hw *hwclk;
 	struct cpufreq_frequency_table *table;
 	struct cpu_data *data;
 	unsigned int cpu = policy->cpu;
@@ -221,17 +182,13 @@ static int qoriq_cpufreq_cpu_init(struct cpufreq_policy *policy)
 		goto err_nomem2;
 	}
 
-	pnode = of_parse_phandle(np, "clocks", 0);
-	if (!pnode) {
-		pr_err("%s: could not get clock information\n", __func__);
-		goto err_nomem2;
-	}
+	hwclk = __clk_get_hw(policy->clk);
+	count = clk_hw_get_num_parents(hwclk);
 
-	count = of_property_count_strings(pnode, "clock-names");
 	data->pclk = kcalloc(count, sizeof(struct clk *), GFP_KERNEL);
 	if (!data->pclk) {
 		pr_err("%s: no memory\n", __func__);
-		goto err_node;
+		goto err_nomem2;
 	}
 
 	table = kcalloc(count + 1, sizeof(*table), GFP_KERNEL);
@@ -240,22 +197,10 @@ static int qoriq_cpufreq_cpu_init(struct cpufreq_policy *policy)
 		goto err_pclk;
 	}
 
-	if (fmask)
-		mask = fmask[get_cpu_physical_id(cpu)];
-	else
-		mask = 0x0;
-
 	for (i = 0; i < count; i++) {
-		clk = of_clk_get(pnode, i);
+		clk = clk_hw_get_parent_by_index(hwclk, i)->clk;
 		data->pclk[i] = clk;
 		freq = clk_get_rate(clk);
-		/*
-		 * the clock is valid if its frequency is not masked
-		 * and large than minimum allowed frequency.
-		 */
-		if (freq < min_cpufreq || (mask & (1 << i)))
-			table[i].frequency = CPUFREQ_ENTRY_INVALID;
-		else
 		table[i].frequency = freq / 1000;
 		table[i].driver_data = i;
 	}
@@ -282,7 +227,6 @@ static int qoriq_cpufreq_cpu_init(struct cpufreq_policy *policy)
 	policy->cpuinfo.transition_latency = u64temp + 1;
 
 	of_node_put(np);
-	of_node_put(pnode);
 
 	return 0;
 
@@ -290,10 +234,7 @@ static int qoriq_cpufreq_cpu_init(struct cpufreq_policy *policy)
 	kfree(table);
 err_pclk:
 	kfree(data->pclk);
-err_node:
-	of_node_put(pnode);
 err_nomem2:
-	policy->driver_data = NULL;
 	kfree(data);
 err_np:
 	of_node_put(np);
@@ -357,12 +298,25 @@ static struct cpufreq_driver qoriq_cpufreq_driver = {
 	.attr		= cpufreq_generic_attr,
 };
 
+static const struct soc_data blacklist = {
+	.flags = SOC_BLACKLIST,
+};
+
 static const struct of_device_id node_matches[] __initconst = {
-	{ .compatible = "fsl,p2041-clockgen", .data = &sdata[0], },
-	{ .compatible = "fsl,p3041-clockgen", .data = &sdata[0], },
-	{ .compatible = "fsl,p5020-clockgen", .data = &sdata[1], },
-	{ .compatible = "fsl,p4080-clockgen", .data = &sdata[2], },
-	{ .compatible = "fsl,p5040-clockgen", .data = &sdata[2], },
+	/* e6500 cannot use cpufreq due to erratum A-008083 */
+	{ .compatible = "fsl,b4420-clockgen", &blacklist },
+	{ .compatible = "fsl,b4860-clockgen", &blacklist },
+	{ .compatible = "fsl,t2080-clockgen", &blacklist },
+	{ .compatible = "fsl,t4240-clockgen", &blacklist },
+
+	{ .compatible = "fsl,ls1012a-clockgen", },
+	{ .compatible = "fsl,ls1021a-clockgen", },
+	{ .compatible = "fsl,ls1043a-clockgen", },
+	{ .compatible = "fsl,ls1046a-clockgen", },
+	{ .compatible = "fsl,ls1088a-clockgen", },
+	{ .compatible = "fsl,ls2080a-clockgen", },
+	{ .compatible = "fsl,p4080-clockgen", },
+	{ .compatible = "fsl,qoriq-clockgen-1.0", },
 	{ .compatible = "fsl,qoriq-clockgen-2.0", },
 	{}
 };
@@ -380,16 +334,12 @@ static int __init qoriq_cpufreq_init(void)
 
 	match = of_match_node(node_matches, np);
 	data = match->data;
-	if (data) {
-		if (data->flag)
-			fmask = data->freq_mask;
-		min_cpufreq = get_bus_freq();
-	} else {
-		min_cpufreq = get_bus_freq() / 2;
-	}
 
 	of_node_put(np);
 
+	if (data && data->flags & SOC_BLACKLIST)
+		return -ENODEV;
+
 	ret = cpufreq_register_driver(&qoriq_cpufreq_driver);
 	if (!ret)
 		pr_info("Freescale QorIQ CPU frequency scaling driver\n");

drivers/cpufreq/s3c2416-cpufreq.c

@@ -400,7 +400,6 @@ static int s3c2416_cpufreq_driver_init(struct cpufreq_policy *policy)
 	rate = clk_get_rate(s3c_freq->hclk);
 	if (rate < 133 * 1000 * 1000) {
 		pr_err("cpufreq: HCLK not at 133MHz\n");
-		clk_put(s3c_freq->hclk);
 		ret = -EINVAL;
 		goto err_armclk;
 	}

drivers/cpufreq/sti-cpufreq.c

@@ -160,6 +160,7 @@ static int sti_cpufreq_set_opp_info(void)
 	int pcode, substrate, major, minor;
 	int ret;
 	char name[MAX_PCODE_NAME_LEN];
+	struct opp_table *opp_table;
 
 	reg_fields = sti_cpufreq_match();
 	if (!reg_fields) {
@@ -211,20 +212,20 @@ static int sti_cpufreq_set_opp_info(void)
 
 	snprintf(name, MAX_PCODE_NAME_LEN, "pcode%d", pcode);
 
-	ret = dev_pm_opp_set_prop_name(dev, name);
-	if (ret) {
+	opp_table = dev_pm_opp_set_prop_name(dev, name);
+	if (IS_ERR(opp_table)) {
 		dev_err(dev, "Failed to set prop name\n");
-		return ret;
+		return PTR_ERR(opp_table);
 	}
 
 	version[0] = BIT(major);
 	version[1] = BIT(minor);
 	version[2] = BIT(substrate);
 
-	ret = dev_pm_opp_set_supported_hw(dev, version, VERSION_ELEMENTS);
-	if (ret) {
+	opp_table = dev_pm_opp_set_supported_hw(dev, version, VERSION_ELEMENTS);
+	if (IS_ERR(opp_table)) {
 		dev_err(dev, "Failed to set supported hardware\n");
-		return ret;
+		return PTR_ERR(opp_table);
 	}
 
 	dev_dbg(dev, "pcode: %d major: %d minor: %d substrate: %d\n",

drivers/cpufreq/ti-cpufreq.c

+/*
+ * TI CPUFreq/OPP hw-supported driver
+ *
+ * Copyright (C) 2016-2017 Texas Instruments, Inc.
+ *	 Dave Gerlach <d-gerlach@ti.com>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that 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.
+ */
+
+#include <linux/cpu.h>
+#include <linux/io.h>
+#include <linux/mfd/syscon.h>
+#include <linux/init.h>
+#include <linux/of.h>
+#include <linux/of_platform.h>
+#include <linux/pm_opp.h>
+#include <linux/regmap.h>
+#include <linux/slab.h>
+
+#define REVISION_MASK				0xF
+#define REVISION_SHIFT				28
+
+#define AM33XX_800M_ARM_MPU_MAX_FREQ		0x1E2F
+#define AM43XX_600M_ARM_MPU_MAX_FREQ		0xFFA
+
+#define DRA7_EFUSE_HAS_OD_MPU_OPP		11
+#define DRA7_EFUSE_HAS_HIGH_MPU_OPP		15
+#define DRA7_EFUSE_HAS_ALL_MPU_OPP		23
+
+#define DRA7_EFUSE_NOM_MPU_OPP			BIT(0)
+#define DRA7_EFUSE_OD_MPU_OPP			BIT(1)
+#define DRA7_EFUSE_HIGH_MPU_OPP			BIT(2)
+
+#define VERSION_COUNT				2
+
+struct ti_cpufreq_data;
+
+struct ti_cpufreq_soc_data {
+	unsigned long (*efuse_xlate)(struct ti_cpufreq_data *opp_data,
+				     unsigned long efuse);
+	unsigned long efuse_fallback;
+	unsigned long efuse_offset;
+	unsigned long efuse_mask;
+	unsigned long efuse_shift;
+	unsigned long rev_offset;
+};
+
+struct ti_cpufreq_data {
+	struct device *cpu_dev;
+	struct device_node *opp_node;
+	struct regmap *syscon;
+	const struct ti_cpufreq_soc_data *soc_data;
+};
+
+static unsigned long amx3_efuse_xlate(struct ti_cpufreq_data *opp_data,
+				      unsigned long efuse)
+{
+	if (!efuse)
+		efuse = opp_data->soc_data->efuse_fallback;
+	/* AM335x and AM437x use "OPP disable" bits, so invert */
+	return ~efuse;
+}
+
+static unsigned long dra7_efuse_xlate(struct ti_cpufreq_data *opp_data,
+				      unsigned long efuse)
+{
+	unsigned long calculated_efuse = DRA7_EFUSE_NOM_MPU_OPP;
+
+	/*
+	 * The efuse on dra7 and am57 parts contains a specific
+	 * value indicating the highest available OPP.
+	 */
+
+	switch (efuse) {
+	case DRA7_EFUSE_HAS_ALL_MPU_OPP:
+	case DRA7_EFUSE_HAS_HIGH_MPU_OPP:
+		calculated_efuse |= DRA7_EFUSE_HIGH_MPU_OPP;
+	case DRA7_EFUSE_HAS_OD_MPU_OPP:
+		calculated_efuse |= DRA7_EFUSE_OD_MPU_OPP;
+	}
+
+	return calculated_efuse;
+}
+
+static struct ti_cpufreq_soc_data am3x_soc_data = {
+	.efuse_xlate = amx3_efuse_xlate,
+	.efuse_fallback = AM33XX_800M_ARM_MPU_MAX_FREQ,
+	.efuse_offset = 0x07fc,
+	.efuse_mask = 0x1fff,
+	.rev_offset = 0x600,
+};
+
+static struct ti_cpufreq_soc_data am4x_soc_data = {
+	.efuse_xlate = amx3_efuse_xlate,
+	.efuse_fallback = AM43XX_600M_ARM_MPU_MAX_FREQ,
+	.efuse_offset = 0x0610,
+	.efuse_mask = 0x3f,
+	.rev_offset = 0x600,
+};
+
+static struct ti_cpufreq_soc_data dra7_soc_data = {
+	.efuse_xlate = dra7_efuse_xlate,
+	.efuse_offset = 0x020c,
+	.efuse_mask = 0xf80000,
+	.efuse_shift = 19,
+	.rev_offset = 0x204,
+};
+
+/**
+ * ti_cpufreq_get_efuse() - Parse and return efuse value present on SoC
+ * @opp_data: pointer to ti_cpufreq_data context
+ * @efuse_value: Set to the value parsed from efuse
+ *
+ * Returns error code if efuse not read properly.
+ */
+static int ti_cpufreq_get_efuse(struct ti_cpufreq_data *opp_data,
+				u32 *efuse_value)
+{
+	struct device *dev = opp_data->cpu_dev;
+	u32 efuse;
+	int ret;
+
+	ret = regmap_read(opp_data->syscon, opp_data->soc_data->efuse_offset,
+			  &efuse);
+	if (ret) {
+		dev_err(dev,
+			"Failed to read the efuse value from syscon: %d\n",
+			ret);
+		return ret;
+	}
+
+	efuse = (efuse & opp_data->soc_data->efuse_mask);
+	efuse >>= opp_data->soc_data->efuse_shift;
+
+	*efuse_value = opp_data->soc_data->efuse_xlate(opp_data, efuse);
+
+	return 0;
+}
+
+/**
+ * ti_cpufreq_get_rev() - Parse and return rev value present on SoC
+ * @opp_data: pointer to ti_cpufreq_data context
+ * @revision_value: Set to the value parsed from revision register
+ *
+ * Returns error code if revision not read properly.
+ */
+static int ti_cpufreq_get_rev(struct ti_cpufreq_data *opp_data,
+			      u32 *revision_value)
+{
+	struct device *dev = opp_data->cpu_dev;
+	u32 revision;
+	int ret;
+
+	ret = regmap_read(opp_data->syscon, opp_data->soc_data->rev_offset,
+			  &revision);
+	if (ret) {
+		dev_err(dev,
+			"Failed to read the revision number from syscon: %d\n",
+			ret);
+		return ret;
+	}
+
+	*revision_value = BIT((revision >> REVISION_SHIFT) & REVISION_MASK);
+
+	return 0;
+}
+
+static int ti_cpufreq_setup_syscon_register(struct ti_cpufreq_data *opp_data)
+{
+	struct device *dev = opp_data->cpu_dev;
+	struct device_node *np = opp_data->opp_node;
+
+	opp_data->syscon = syscon_regmap_lookup_by_phandle(np,
+							"syscon");
+	if (IS_ERR(opp_data->syscon)) {
+		dev_err(dev,
+			"\"syscon\" is missing, cannot use OPPv2 table.\n");
+		return PTR_ERR(opp_data->syscon);
+	}
+
+	return 0;
+}
+
+static const struct of_device_id ti_cpufreq_of_match[] = {
+	{ .compatible = "ti,am33xx", .data = &am3x_soc_data, },
+	{ .compatible = "ti,am4372", .data = &am4x_soc_data, },
+	{ .compatible = "ti,dra7", .data = &dra7_soc_data },
+	{},
+};
+
+static int ti_cpufreq_init(void)
+{
+	u32 version[VERSION_COUNT];
+	struct device_node *np;
+	const struct of_device_id *match;
+	struct ti_cpufreq_data *opp_data;
+	int ret;
+
+	np = of_find_node_by_path("/");
+	match = of_match_node(ti_cpufreq_of_match, np);
+	if (!match)
+		return -ENODEV;
+
+	opp_data = kzalloc(sizeof(*opp_data), GFP_KERNEL);
+	if (!opp_data)
+		return -ENOMEM;
+
+	opp_data->soc_data = match->data;
+
+	opp_data->cpu_dev = get_cpu_device(0);
+	if (!opp_data->cpu_dev) {
+		pr_err("%s: Failed to get device for CPU0\n", __func__);
+		return -ENODEV;
+	}
+
+	opp_data->opp_node = dev_pm_opp_of_get_opp_desc_node(opp_data->cpu_dev);
+	if (!opp_data->opp_node) {
+		dev_info(opp_data->cpu_dev,
+			 "OPP-v2 not supported, cpufreq-dt will attempt to use legacy tables.\n");
+		goto register_cpufreq_dt;
+	}
+
+	ret = ti_cpufreq_setup_syscon_register(opp_data);
+	if (ret)
+		goto fail_put_node;
+
+	/*
+	 * OPPs determine whether or not they are supported based on
+	 * two metrics:
+	 *	0 - SoC Revision
+	 *	1 - eFuse value
+	 */
+	ret = ti_cpufreq_get_rev(opp_data, &version[0]);
+	if (ret)
+		goto fail_put_node;
+
+	ret = ti_cpufreq_get_efuse(opp_data, &version[1]);
+	if (ret)
+		goto fail_put_node;
+
+	of_node_put(opp_data->opp_node);
+
+	ret = PTR_ERR_OR_ZERO(dev_pm_opp_set_supported_hw(opp_data->cpu_dev,
+							  version, VERSION_COUNT));
+	if (ret) {
+		dev_err(opp_data->cpu_dev,
+			"Failed to set supported hardware\n");
+		goto fail_put_node;
+	}
+
+register_cpufreq_dt:
+	platform_device_register_simple("cpufreq-dt", -1, NULL, 0);
+
+	return 0;
+
+fail_put_node:
+	of_node_put(opp_data->opp_node);
+
+	return ret;
+}
+device_initcall(ti_cpufreq_init);

drivers/cpuidle/governors/menu.c

@@ -19,6 +19,7 @@
 #include <linux/tick.h>
 #include <linux/sched.h>
 #include <linux/math64.h>
+#include <linux/cpu.h>
 
 /*
  * Please note when changing the tuning values:
@@ -280,17 +281,23 @@ static unsigned int get_typical_interval(struct menu_device *data)
 static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 {
 	struct menu_device *data = this_cpu_ptr(&menu_devices);
+	struct device *device = get_cpu_device(dev->cpu);
 	int latency_req = pm_qos_request(PM_QOS_CPU_DMA_LATENCY);
 	int i;
 	unsigned int interactivity_req;
 	unsigned int expected_interval;
 	unsigned long nr_iowaiters, cpu_load;
+	int resume_latency = dev_pm_qos_read_value(device);
 
 	if (data->needs_update) {
 		menu_update(drv, dev);
 		data->needs_update = 0;
 	}
 
+	/* resume_latency is 0 means no restriction */
+	if (resume_latency && resume_latency < latency_req)
+		latency_req = resume_latency;
+
 	/* Special case when user has set very strict latency requirement */
 	if (unlikely(latency_req == 0))
 		return 0;
@@ -357,9 +364,9 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 		if (s->disabled || su->disable)
 			continue;
 		if (s->target_residency > data->predicted_us)
-			continue;
+			break;
 		if (s->exit_latency > latency_req)
-			continue;
+			break;
 
 		data->last_state_idx = i;
 	}

drivers/devfreq/devfreq-event.c

@@ -306,7 +306,7 @@ struct devfreq_event_dev *devfreq_event_add_edev(struct device *dev,
 						struct devfreq_event_desc *desc)
 {
 	struct devfreq_event_dev *edev;
-	static atomic_t event_no = ATOMIC_INIT(0);
+	static atomic_t event_no = ATOMIC_INIT(-1);
 	int ret;
 
 	if (!dev || !desc)
@@ -329,7 +329,7 @@ struct devfreq_event_dev *devfreq_event_add_edev(struct device *dev,
 	edev->dev.class = devfreq_event_class;
 	edev->dev.release = devfreq_event_release_edev;
 
-	dev_set_name(&edev->dev, "event.%d", atomic_inc_return(&event_no) - 1);
+	dev_set_name(&edev->dev, "event%d", atomic_inc_return(&event_no));
 	ret = device_register(&edev->dev);
 	if (ret < 0) {
 		put_device(&edev->dev);

drivers/devfreq/devfreq.c

@@ -111,18 +111,16 @@ static void devfreq_set_freq_table(struct devfreq *devfreq)
 		return;
 	}
 
-	rcu_read_lock();
 	for (i = 0, freq = 0; i < profile->max_state; i++, freq++) {
 		opp = dev_pm_opp_find_freq_ceil(devfreq->dev.parent, &freq);
 		if (IS_ERR(opp)) {
 			devm_kfree(devfreq->dev.parent, profile->freq_table);
 			profile->max_state = 0;
-			rcu_read_unlock();
 			return;
 		}
+		dev_pm_opp_put(opp);
 		profile->freq_table[i] = freq;
 	}
-	rcu_read_unlock();
 }
 
 /**
@@ -130,7 +128,7 @@ static void devfreq_set_freq_table(struct devfreq *devfreq)
  * @devfreq:	the devfreq instance
  * @freq:	the update target frequency
  */
-static int devfreq_update_status(struct devfreq *devfreq, unsigned long freq)
+int devfreq_update_status(struct devfreq *devfreq, unsigned long freq)
 {
 	int lev, prev_lev, ret = 0;
 	unsigned long cur_time;
@@ -166,6 +164,7 @@ static int devfreq_update_status(struct devfreq *devfreq, unsigned long freq)
 	devfreq->last_stat_updated = cur_time;
 	return ret;
 }
+EXPORT_SYMBOL(devfreq_update_status);
 
 /**
  * find_devfreq_governor() - find devfreq governor from name
@@ -474,11 +473,15 @@ static int devfreq_notifier_call(struct notifier_block *nb, unsigned long type,
 }
 
 /**
- * _remove_devfreq() - Remove devfreq from the list and release its resources.
- * @devfreq:	the devfreq struct
+ * devfreq_dev_release() - Callback for struct device to release the device.
+ * @dev:	the devfreq device
+ *
+ * Remove devfreq from the list and release its resources.
  */
-static void _remove_devfreq(struct devfreq *devfreq)
+static void devfreq_dev_release(struct device *dev)
 {
+	struct devfreq *devfreq = to_devfreq(dev);
+
 	mutex_lock(&devfreq_list_lock);
 	if (IS_ERR(find_device_devfreq(devfreq->dev.parent))) {
 		mutex_unlock(&devfreq_list_lock);
@@ -499,19 +502,6 @@ static void _remove_devfreq(struct devfreq *devfreq)
 	kfree(devfreq);
 }
 
-/**
- * devfreq_dev_release() - Callback for struct device to release the device.
- * @dev:	the devfreq device
- *
- * This calls _remove_devfreq() if _remove_devfreq() is not called.
- */
-static void devfreq_dev_release(struct device *dev)
-{
-	struct devfreq *devfreq = to_devfreq(dev);
-
-	_remove_devfreq(devfreq);
-}
-
 /**
  * devfreq_add_device() - Add devfreq feature to the device
  * @dev:	the device to add devfreq feature.
@@ -527,6 +517,7 @@ struct devfreq *devfreq_add_device(struct device *dev,
 {
 	struct devfreq *devfreq;
 	struct devfreq_governor *governor;
+	static atomic_t devfreq_no = ATOMIC_INIT(-1);
 	int err = 0;
 
 	if (!dev || !profile || !governor_name) {
@@ -538,15 +529,14 @@ struct devfreq *devfreq_add_device(struct device *dev,
 	devfreq = find_device_devfreq(dev);
 	mutex_unlock(&devfreq_list_lock);
 	if (!IS_ERR(devfreq)) {
-		dev_err(dev, "%s: Unable to create devfreq for the device. It already has one.\n", __func__);
+		dev_err(dev, "%s: Unable to create devfreq for the device.\n",
+			__func__);
 		err = -EINVAL;
 		goto err_out;
 	}
 
 	devfreq = kzalloc(sizeof(struct devfreq), GFP_KERNEL);
 	if (!devfreq) {
-		dev_err(dev, "%s: Unable to create devfreq for the device\n",
-			__func__);
 		err = -ENOMEM;
 		goto err_out;
 	}
@@ -569,18 +559,21 @@ struct devfreq *devfreq_add_device(struct device *dev,
 		mutex_lock(&devfreq->lock);
 	}
 
-	dev_set_name(&devfreq->dev, "%s", dev_name(dev));
+	dev_set_name(&devfreq->dev, "devfreq%d",
+				atomic_inc_return(&devfreq_no));
 	err = device_register(&devfreq->dev);
 	if (err) {
 		mutex_unlock(&devfreq->lock);
 		goto err_out;
 	}
 
-	devfreq->trans_table =	devm_kzalloc(&devfreq->dev, sizeof(unsigned int) *
+	devfreq->trans_table =	devm_kzalloc(&devfreq->dev,
+						sizeof(unsigned int) *
 						devfreq->profile->max_state *
 						devfreq->profile->max_state,
 						GFP_KERNEL);
-	devfreq->time_in_state = devm_kzalloc(&devfreq->dev, sizeof(unsigned long) *
+	devfreq->time_in_state = devm_kzalloc(&devfreq->dev,
+						sizeof(unsigned long) *
 						devfreq->profile->max_state,
 						GFP_KERNEL);
 	devfreq->last_stat_updated = jiffies;
@@ -939,6 +932,9 @@ static ssize_t governor_store(struct device *dev, struct device_attribute *attr,
 	if (df->governor == governor) {
 		ret = 0;
 		goto out;
+	} else if (df->governor->immutable || governor->immutable) {
+		ret = -EINVAL;
+		goto out;
 	}
 
 	if (df->governor) {
@@ -968,13 +964,33 @@ static ssize_t available_governors_show(struct device *d,
 					struct device_attribute *attr,
 					char *buf)
 {
-	struct devfreq_governor *tmp_governor;
+	struct devfreq *df = to_devfreq(d);
 	ssize_t count = 0;
 
 	mutex_lock(&devfreq_list_lock);
-	list_for_each_entry(tmp_governor, &devfreq_governor_list, node)
+
+	/*
+	 * The devfreq with immutable governor (e.g., passive) shows
+	 * only own governor.
+	 */
+	if (df->governor->immutable) {
+		count = scnprintf(&buf[count], DEVFREQ_NAME_LEN,
+				   "%s ", df->governor_name);
+	/*
+	 * The devfreq device shows the registered governor except for
+	 * immutable governors such as passive governor .
+	 */
+	} else {
+		struct devfreq_governor *governor;
+
+		list_for_each_entry(governor, &devfreq_governor_list, node) {
+			if (governor->immutable)
+				continue;
 			count += scnprintf(&buf[count], (PAGE_SIZE - count - 2),
-				   "%s ", tmp_governor->name);
+					   "%s ", governor->name);
+		}
+	}
+
 	mutex_unlock(&devfreq_list_lock);
 
 	/* Truncate the trailing space */
@@ -1112,17 +1128,16 @@ static ssize_t available_frequencies_show(struct device *d,
 	ssize_t count = 0;
 	unsigned long freq = 0;
 
-	rcu_read_lock();
 	do {
 		opp = dev_pm_opp_find_freq_ceil(dev, &freq);
 		if (IS_ERR(opp))
 			break;
 
+		dev_pm_opp_put(opp);
 		count += scnprintf(&buf[count], (PAGE_SIZE - count - 2),
 				   "%lu ", freq);
 		freq++;
 	} while (1);
-	rcu_read_unlock();
 
 	/* Truncate the trailing space */
 	if (count)
@@ -1224,11 +1239,8 @@ subsys_initcall(devfreq_init);
  * @freq:	The frequency given to target function
  * @flags:	Flags handed from devfreq framework.
  *
- * Locking: This function must be called under rcu_read_lock(). opp is a rcu
- * protected pointer. The reason for the same is that the opp pointer which is
- * returned will remain valid for use with opp_get_{voltage, freq} only while
- * under the locked area. The pointer returned must be used prior to unlocking
- * with rcu_read_unlock() to maintain the integrity of the pointer.
+ * The callers are required to call dev_pm_opp_put() for the returned OPP after
+ * use.
  */
 struct dev_pm_opp *devfreq_recommended_opp(struct device *dev,
 					   unsigned long *freq,
@@ -1265,18 +1277,7 @@ EXPORT_SYMBOL(devfreq_recommended_opp);
  */
 int devfreq_register_opp_notifier(struct device *dev, struct devfreq *devfreq)
 {
-	struct srcu_notifier_head *nh;
-	int ret = 0;
-
-	rcu_read_lock();
-	nh = dev_pm_opp_get_notifier(dev);
-	if (IS_ERR(nh))
-		ret = PTR_ERR(nh);
-	rcu_read_unlock();
-	if (!ret)
-		ret = srcu_notifier_chain_register(nh, &devfreq->nb);
-
-	return ret;
+	return dev_pm_opp_register_notifier(dev, &devfreq->nb);
 }
 EXPORT_SYMBOL(devfreq_register_opp_notifier);
 
@@ -1292,18 +1293,7 @@ EXPORT_SYMBOL(devfreq_register_opp_notifier);
  */
 int devfreq_unregister_opp_notifier(struct device *dev, struct devfreq *devfreq)
 {
-	struct srcu_notifier_head *nh;
-	int ret = 0;
-
-	rcu_read_lock();
-	nh = dev_pm_opp_get_notifier(dev);
-	if (IS_ERR(nh))
-		ret = PTR_ERR(nh);
-	rcu_read_unlock();
-	if (!ret)
-		ret = srcu_notifier_chain_unregister(nh, &devfreq->nb);
-
-	return ret;
+	return dev_pm_opp_unregister_notifier(dev, &devfreq->nb);
 }
 EXPORT_SYMBOL(devfreq_unregister_opp_notifier);
 

drivers/devfreq/event/exynos-ppmu.c

@@ -17,13 +17,13 @@
 #include <linux/module.h>
 #include <linux/of_address.h>
 #include <linux/platform_device.h>
+#include <linux/regmap.h>
 #include <linux/suspend.h>
 #include <linux/devfreq-event.h>
 
 #include "exynos-ppmu.h"
 
 struct exynos_ppmu_data {
-	void __iomem *base;
 	struct clk *clk;
 };
 
@@ -33,6 +33,7 @@ struct exynos_ppmu {
 	unsigned int num_events;
 
 	struct device *dev;
+	struct regmap *regmap;
 
 	struct exynos_ppmu_data ppmu;
 };
@@ -107,20 +108,28 @@ static int exynos_ppmu_find_ppmu_id(struct devfreq_event_dev *edev)
 static int exynos_ppmu_disable(struct devfreq_event_dev *edev)
 {
 	struct exynos_ppmu *info = devfreq_event_get_drvdata(edev);
+	int ret;
 	u32 pmnc;
 
 	/* Disable all counters */
-	__raw_writel(PPMU_CCNT_MASK |
+	ret = regmap_write(info->regmap, PPMU_CNTENC,
+				PPMU_CCNT_MASK |
 				PPMU_PMCNT0_MASK |
 				PPMU_PMCNT1_MASK |
 				PPMU_PMCNT2_MASK |
-		     PPMU_PMCNT3_MASK,
-		     info->ppmu.base + PPMU_CNTENC);
+				PPMU_PMCNT3_MASK);
+	if (ret < 0)
+		return ret;
 
 	/* Disable PPMU */
-	pmnc = __raw_readl(info->ppmu.base + PPMU_PMNC);
+	ret = regmap_read(info->regmap, PPMU_PMNC, &pmnc);
+	if (ret < 0)
+		return ret;
+
 	pmnc &= ~PPMU_PMNC_ENABLE_MASK;
-	__raw_writel(pmnc, info->ppmu.base + PPMU_PMNC);
+	ret = regmap_write(info->regmap, PPMU_PMNC, pmnc);
+	if (ret < 0)
+		return ret;
 
 	return 0;
 }
@@ -129,29 +138,42 @@ static int exynos_ppmu_set_event(struct devfreq_event_dev *edev)
 {
 	struct exynos_ppmu *info = devfreq_event_get_drvdata(edev);
 	int id = exynos_ppmu_find_ppmu_id(edev);
+	int ret;
 	u32 pmnc, cntens;
 
 	if (id < 0)
 		return id;
 
 	/* Enable specific counter */
-	cntens = __raw_readl(info->ppmu.base + PPMU_CNTENS);
+	ret = regmap_read(info->regmap, PPMU_CNTENS, &cntens);
+	if (ret < 0)
+		return ret;
+
 	cntens |= (PPMU_CCNT_MASK | (PPMU_ENABLE << id));
-	__raw_writel(cntens, info->ppmu.base + PPMU_CNTENS);
+	ret = regmap_write(info->regmap, PPMU_CNTENS, cntens);
+	if (ret < 0)
+		return ret;
 
 	/* Set the event of Read/Write data count  */
-	__raw_writel(PPMU_RO_DATA_CNT | PPMU_WO_DATA_CNT,
-			info->ppmu.base + PPMU_BEVTxSEL(id));
+	ret = regmap_write(info->regmap, PPMU_BEVTxSEL(id),
+				PPMU_RO_DATA_CNT | PPMU_WO_DATA_CNT);
+	if (ret < 0)
+		return ret;
 
 	/* Reset cycle counter/performance counter and enable PPMU */
-	pmnc = __raw_readl(info->ppmu.base + PPMU_PMNC);
+	ret = regmap_read(info->regmap, PPMU_PMNC, &pmnc);
+	if (ret < 0)
+		return ret;
+
 	pmnc &= ~(PPMU_PMNC_ENABLE_MASK
 			| PPMU_PMNC_COUNTER_RESET_MASK
 			| PPMU_PMNC_CC_RESET_MASK);
 	pmnc |= (PPMU_ENABLE << PPMU_PMNC_ENABLE_SHIFT);
 	pmnc |= (PPMU_ENABLE << PPMU_PMNC_COUNTER_RESET_SHIFT);
 	pmnc |= (PPMU_ENABLE << PPMU_PMNC_CC_RESET_SHIFT);
-	__raw_writel(pmnc, info->ppmu.base + PPMU_PMNC);
+	ret = regmap_write(info->regmap, PPMU_PMNC, pmnc);
+	if (ret < 0)
+		return ret;
 
 	return 0;
 }
@@ -161,40 +183,64 @@ static int exynos_ppmu_get_event(struct devfreq_event_dev *edev,
 {
 	struct exynos_ppmu *info = devfreq_event_get_drvdata(edev);
 	int id = exynos_ppmu_find_ppmu_id(edev);
-	u32 pmnc, cntenc;
+	unsigned int total_count, load_count;
+	unsigned int pmcnt3_high, pmcnt3_low;
+	unsigned int pmnc, cntenc;
+	int ret;
 
 	if (id < 0)
 		return -EINVAL;
 
 	/* Disable PPMU */
-	pmnc = __raw_readl(info->ppmu.base + PPMU_PMNC);
+	ret = regmap_read(info->regmap, PPMU_PMNC, &pmnc);
+	if (ret < 0)
+		return ret;
+
 	pmnc &= ~PPMU_PMNC_ENABLE_MASK;
-	__raw_writel(pmnc, info->ppmu.base + PPMU_PMNC);
+	ret = regmap_write(info->regmap, PPMU_PMNC, pmnc);
+	if (ret < 0)
+		return ret;
 
 	/* Read cycle count */
-	edata->total_count = __raw_readl(info->ppmu.base + PPMU_CCNT);
+	ret = regmap_read(info->regmap, PPMU_CCNT, &total_count);
+	if (ret < 0)
+		return ret;
+	edata->total_count = total_count;
 
 	/* Read performance count */
 	switch (id) {
 	case PPMU_PMNCNT0:
 	case PPMU_PMNCNT1:
 	case PPMU_PMNCNT2:
-		edata->load_count
-			= __raw_readl(info->ppmu.base + PPMU_PMNCT(id));
+		ret = regmap_read(info->regmap, PPMU_PMNCT(id), &load_count);
+		if (ret < 0)
+			return ret;
+		edata->load_count = load_count;
 		break;
 	case PPMU_PMNCNT3:
-		edata->load_count =
-			((__raw_readl(info->ppmu.base + PPMU_PMCNT3_HIGH) << 8)
-			| __raw_readl(info->ppmu.base + PPMU_PMCNT3_LOW));
+		ret = regmap_read(info->regmap, PPMU_PMCNT3_HIGH, &pmcnt3_high);
+		if (ret < 0)
+			return ret;
+
+		ret = regmap_read(info->regmap, PPMU_PMCNT3_LOW, &pmcnt3_low);
+		if (ret < 0)
+			return ret;
+
+		edata->load_count = ((pmcnt3_high << 8) | pmcnt3_low);
 		break;
 	default:
 		return -EINVAL;
 	}
 
 	/* Disable specific counter */
-	cntenc = __raw_readl(info->ppmu.base + PPMU_CNTENC);
+	ret = regmap_read(info->regmap, PPMU_CNTENC, &cntenc);
+	if (ret < 0)
+		return ret;
+
 	cntenc |= (PPMU_CCNT_MASK | (PPMU_ENABLE << id));
-	__raw_writel(cntenc, info->ppmu.base + PPMU_CNTENC);
+	ret = regmap_write(info->regmap, PPMU_CNTENC, cntenc);
+	if (ret < 0)
+		return ret;
 
 	dev_dbg(&edev->dev, "%s (event: %ld/%ld)\n", edev->desc->name,
 					edata->load_count, edata->total_count);
@@ -214,36 +260,93 @@ static const struct devfreq_event_ops exynos_ppmu_ops = {
 static int exynos_ppmu_v2_disable(struct devfreq_event_dev *edev)
 {
 	struct exynos_ppmu *info = devfreq_event_get_drvdata(edev);
+	int ret;
 	u32 pmnc, clear;
 
 	/* Disable all counters */
 	clear = (PPMU_CCNT_MASK | PPMU_PMCNT0_MASK | PPMU_PMCNT1_MASK
 		| PPMU_PMCNT2_MASK | PPMU_PMCNT3_MASK);
+	ret = regmap_write(info->regmap, PPMU_V2_FLAG, clear);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_INTENC, clear);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_CNTENC, clear);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_CNT_RESET, clear);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_CIG_CFG0, 0x0);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_CIG_CFG1, 0x0);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_CIG_CFG2, 0x0);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_CIG_RESULT, 0x0);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_CNT_AUTO, 0x0);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_CH_EV0_TYPE, 0x0);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_CH_EV1_TYPE, 0x0);
+	if (ret < 0)
+		return ret;
 
-	__raw_writel(clear, info->ppmu.base + PPMU_V2_FLAG);
-	__raw_writel(clear, info->ppmu.base + PPMU_V2_INTENC);
-	__raw_writel(clear, info->ppmu.base + PPMU_V2_CNTENC);
-	__raw_writel(clear, info->ppmu.base + PPMU_V2_CNT_RESET);
-
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_CIG_CFG0);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_CIG_CFG1);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_CIG_CFG2);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_CIG_RESULT);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_CNT_AUTO);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_CH_EV0_TYPE);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_CH_EV1_TYPE);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_CH_EV2_TYPE);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_CH_EV3_TYPE);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_SM_ID_V);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_SM_ID_A);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_SM_OTHERS_V);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_SM_OTHERS_A);
-	__raw_writel(0x0, info->ppmu.base + PPMU_V2_INTERRUPT_RESET);
+	ret = regmap_write(info->regmap, PPMU_V2_CH_EV2_TYPE, 0x0);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_CH_EV3_TYPE, 0x0);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_SM_ID_V, 0x0);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_SM_ID_A, 0x0);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_SM_OTHERS_V, 0x0);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_SM_OTHERS_A, 0x0);
+	if (ret < 0)
+		return ret;
+
+	ret = regmap_write(info->regmap, PPMU_V2_INTERRUPT_RESET, 0x0);
+	if (ret < 0)
+		return ret;
 
 	/* Disable PPMU */
-	pmnc = __raw_readl(info->ppmu.base + PPMU_V2_PMNC);
+	ret = regmap_read(info->regmap, PPMU_V2_PMNC, &pmnc);
+	if (ret < 0)
+		return ret;
+
 	pmnc &= ~PPMU_PMNC_ENABLE_MASK;
-	__raw_writel(pmnc, info->ppmu.base + PPMU_V2_PMNC);
+	ret = regmap_write(info->regmap, PPMU_V2_PMNC, pmnc);
+	if (ret < 0)
+		return ret;
 
 	return 0;
 }
@@ -251,30 +354,43 @@ static int exynos_ppmu_v2_disable(struct devfreq_event_dev *edev)
 static int exynos_ppmu_v2_set_event(struct devfreq_event_dev *edev)
 {
 	struct exynos_ppmu *info = devfreq_event_get_drvdata(edev);
+	unsigned int pmnc, cntens;
 	int id = exynos_ppmu_find_ppmu_id(edev);
-	u32 pmnc, cntens;
+	int ret;
 
 	/* Enable all counters */
-	cntens = __raw_readl(info->ppmu.base + PPMU_V2_CNTENS);
+	ret = regmap_read(info->regmap, PPMU_V2_CNTENS, &cntens);
+	if (ret < 0)
+		return ret;
+
 	cntens |= (PPMU_CCNT_MASK | (PPMU_ENABLE << id));
-	__raw_writel(cntens, info->ppmu.base + PPMU_V2_CNTENS);
+	ret = regmap_write(info->regmap, PPMU_V2_CNTENS, cntens);
+	if (ret < 0)
+		return ret;
 
 	/* Set the event of Read/Write data count  */
 	switch (id) {
 	case PPMU_PMNCNT0:
 	case PPMU_PMNCNT1:
 	case PPMU_PMNCNT2:
-		__raw_writel(PPMU_V2_RO_DATA_CNT | PPMU_V2_WO_DATA_CNT,
-				info->ppmu.base + PPMU_V2_CH_EVx_TYPE(id));
+		ret = regmap_write(info->regmap, PPMU_V2_CH_EVx_TYPE(id),
+				PPMU_V2_RO_DATA_CNT | PPMU_V2_WO_DATA_CNT);
+		if (ret < 0)
+			return ret;
 		break;
 	case PPMU_PMNCNT3:
-		__raw_writel(PPMU_V2_EVT3_RW_DATA_CNT,
-				info->ppmu.base + PPMU_V2_CH_EVx_TYPE(id));
+		ret = regmap_write(info->regmap, PPMU_V2_CH_EVx_TYPE(id),
+				PPMU_V2_EVT3_RW_DATA_CNT);
+		if (ret < 0)
+			return ret;
 		break;
 	}
 
 	/* Reset cycle counter/performance counter and enable PPMU */
-	pmnc = __raw_readl(info->ppmu.base + PPMU_V2_PMNC);
+	ret = regmap_read(info->regmap, PPMU_V2_PMNC, &pmnc);
+	if (ret < 0)
+		return ret;
+
 	pmnc &= ~(PPMU_PMNC_ENABLE_MASK
 			| PPMU_PMNC_COUNTER_RESET_MASK
 			| PPMU_PMNC_CC_RESET_MASK
@@ -284,7 +400,10 @@ static int exynos_ppmu_v2_set_event(struct devfreq_event_dev *edev)
 	pmnc |= (PPMU_ENABLE << PPMU_PMNC_COUNTER_RESET_SHIFT);
 	pmnc |= (PPMU_ENABLE << PPMU_PMNC_CC_RESET_SHIFT);
 	pmnc |= (PPMU_V2_MODE_MANUAL << PPMU_V2_PMNC_START_MODE_SHIFT);
-	__raw_writel(pmnc, info->ppmu.base + PPMU_V2_PMNC);
+
+	ret = regmap_write(info->regmap, PPMU_V2_PMNC, pmnc);
+	if (ret < 0)
+		return ret;
 
 	return 0;
 }
@@ -294,37 +413,61 @@ static int exynos_ppmu_v2_get_event(struct devfreq_event_dev *edev,
 {
 	struct exynos_ppmu *info = devfreq_event_get_drvdata(edev);
 	int id = exynos_ppmu_find_ppmu_id(edev);
-	u32 pmnc, cntenc;
-	u32 pmcnt_high, pmcnt_low;
-	u64 load_count = 0;
+	int ret;
+	unsigned int pmnc, cntenc;
+	unsigned int pmcnt_high, pmcnt_low;
+	unsigned int total_count, count;
+	unsigned long load_count = 0;
 
 	/* Disable PPMU */
-	pmnc = __raw_readl(info->ppmu.base + PPMU_V2_PMNC);
+	ret = regmap_read(info->regmap, PPMU_V2_PMNC, &pmnc);
+	if (ret < 0)
+		return ret;
+
 	pmnc &= ~PPMU_PMNC_ENABLE_MASK;
-	__raw_writel(pmnc, info->ppmu.base + PPMU_V2_PMNC);
+	ret = regmap_write(info->regmap, PPMU_V2_PMNC, pmnc);
+	if (ret < 0)
+		return ret;
 
 	/* Read cycle count and performance count */
-	edata->total_count = __raw_readl(info->ppmu.base + PPMU_V2_CCNT);
+	ret = regmap_read(info->regmap, PPMU_V2_CCNT, &total_count);
+	if (ret < 0)
+		return ret;
+	edata->total_count = total_count;
 
 	switch (id) {
 	case PPMU_PMNCNT0:
 	case PPMU_PMNCNT1:
 	case PPMU_PMNCNT2:
-		load_count = __raw_readl(info->ppmu.base + PPMU_V2_PMNCT(id));
+		ret = regmap_read(info->regmap, PPMU_V2_PMNCT(id), &count);
+		if (ret < 0)
+			return ret;
+		load_count = count;
 		break;
 	case PPMU_PMNCNT3:
-		pmcnt_high = __raw_readl(info->ppmu.base + PPMU_V2_PMCNT3_HIGH);
-		pmcnt_low = __raw_readl(info->ppmu.base + PPMU_V2_PMCNT3_LOW);
-		load_count = ((u64)((pmcnt_high & 0xff)) << 32)
-			   + (u64)pmcnt_low;
+		ret = regmap_read(info->regmap, PPMU_V2_PMCNT3_HIGH,
+						&pmcnt_high);
+		if (ret < 0)
+			return ret;
+
+		ret = regmap_read(info->regmap, PPMU_V2_PMCNT3_LOW, &pmcnt_low);
+		if (ret < 0)
+			return ret;
+
+		load_count = ((u64)((pmcnt_high & 0xff)) << 32)+ (u64)pmcnt_low;
 		break;
 	}
 	edata->load_count = load_count;
 
 	/* Disable all counters */
-	cntenc = __raw_readl(info->ppmu.base + PPMU_V2_CNTENC);
+	ret = regmap_read(info->regmap, PPMU_V2_CNTENC, &cntenc);
+	if (ret < 0)
+		return 0;
+
 	cntenc |= (PPMU_CCNT_MASK | (PPMU_ENABLE << id));
-	__raw_writel(cntenc, info->ppmu.base + PPMU_V2_CNTENC);
+	ret = regmap_write(info->regmap, PPMU_V2_CNTENC, cntenc);
+	if (ret < 0)
+		return ret;
 
 	dev_dbg(&edev->dev, "%25s (load: %ld / %ld)\n", edev->desc->name,
 					edata->load_count, edata->total_count);
@@ -411,10 +554,19 @@ static int of_get_devfreq_events(struct device_node *np,
 	return 0;
 }
 
-static int exynos_ppmu_parse_dt(struct exynos_ppmu *info)
+static struct regmap_config exynos_ppmu_regmap_config = {
+	.reg_bits = 32,
+	.val_bits = 32,
+	.reg_stride = 4,
+};
+
+static int exynos_ppmu_parse_dt(struct platform_device *pdev,
+				struct exynos_ppmu *info)
 {
 	struct device *dev = info->dev;
 	struct device_node *np = dev->of_node;
+	struct resource *res;
+	void __iomem *base;
 	int ret = 0;
 
 	if (!np) {
@@ -423,10 +575,17 @@ static int exynos_ppmu_parse_dt(struct exynos_ppmu *info)
 	}
 
 	/* Maps the memory mapped IO to control PPMU register */
-	info->ppmu.base = of_iomap(np, 0);
-	if (IS_ERR_OR_NULL(info->ppmu.base)) {
-		dev_err(dev, "failed to map memory region\n");
-		return -ENOMEM;
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	base = devm_ioremap_resource(dev, res);
+	if (IS_ERR(base))
+		return PTR_ERR(base);
+
+	exynos_ppmu_regmap_config.max_register = resource_size(res) - 4;
+	info->regmap = devm_regmap_init_mmio(dev, base,
+					&exynos_ppmu_regmap_config);
+	if (IS_ERR(info->regmap)) {
+		dev_err(dev, "failed to initialize regmap\n");
+		return PTR_ERR(info->regmap);
 	}
 
 	info->ppmu.clk = devm_clk_get(dev, "ppmu");
@@ -438,15 +597,10 @@ static int exynos_ppmu_parse_dt(struct exynos_ppmu *info)
 	ret = of_get_devfreq_events(np, info);
 	if (ret < 0) {
 		dev_err(dev, "failed to parse exynos ppmu dt node\n");
-		goto err;
+		return ret;
 	}
 
 	return 0;
-
-err:
-	iounmap(info->ppmu.base);
-
-	return ret;
 }
 
 static int exynos_ppmu_probe(struct platform_device *pdev)
@@ -463,7 +617,7 @@ static int exynos_ppmu_probe(struct platform_device *pdev)
 	info->dev = &pdev->dev;
 
 	/* Parse dt data to get resource */
-	ret = exynos_ppmu_parse_dt(info);
+	ret = exynos_ppmu_parse_dt(pdev, info);
 	if (ret < 0) {
 		dev_err(&pdev->dev,
 			"failed to parse devicetree for resource\n");
@@ -476,8 +630,7 @@ static int exynos_ppmu_probe(struct platform_device *pdev)
 	if (!info->edev) {
 		dev_err(&pdev->dev,
 			"failed to allocate memory devfreq-event devices\n");
-		ret = -ENOMEM;
-		goto err;
+		return -ENOMEM;
 	}
 	edev = info->edev;
 	platform_set_drvdata(pdev, info);
@@ -488,17 +641,16 @@ static int exynos_ppmu_probe(struct platform_device *pdev)
 			ret = PTR_ERR(edev[i]);
 			dev_err(&pdev->dev,
 				"failed to add devfreq-event device\n");
-			goto err;
+			return PTR_ERR(edev[i]);
 		}
+
+		pr_info("exynos-ppmu: new PPMU device registered %s (%s)\n",
+			dev_name(&pdev->dev), desc[i].name);
 	}
 
 	clk_prepare_enable(info->ppmu.clk);
 
 	return 0;
-err:
-	iounmap(info->ppmu.base);
-
-	return ret;
 }
 
 static int exynos_ppmu_remove(struct platform_device *pdev)
@@ -506,7 +658,6 @@ static int exynos_ppmu_remove(struct platform_device *pdev)
 	struct exynos_ppmu *info = platform_get_drvdata(pdev);
 
 	clk_disable_unprepare(info->ppmu.clk);
-	iounmap(info->ppmu.base);
 
 	return 0;
 }

drivers/devfreq/exynos-bus.c

@@ -103,18 +103,17 @@ static int exynos_bus_target(struct device *dev, unsigned long *freq, u32 flags)
 	int ret = 0;
 
 	/* Get new opp-bus instance according to new bus clock */
-	rcu_read_lock();
 	new_opp = devfreq_recommended_opp(dev, freq, flags);
 	if (IS_ERR(new_opp)) {
 		dev_err(dev, "failed to get recommended opp instance\n");
-		rcu_read_unlock();
 		return PTR_ERR(new_opp);
 	}
 
 	new_freq = dev_pm_opp_get_freq(new_opp);
 	new_volt = dev_pm_opp_get_voltage(new_opp);
+	dev_pm_opp_put(new_opp);
+
 	old_freq = bus->curr_freq;
-	rcu_read_unlock();
 
 	if (old_freq == new_freq)
 		return 0;
@@ -147,8 +146,8 @@ static int exynos_bus_target(struct device *dev, unsigned long *freq, u32 flags)
 	}
 	bus->curr_freq = new_freq;
 
-	dev_dbg(dev, "Set the frequency of bus (%lukHz -> %lukHz)\n",
-			old_freq/1000, new_freq/1000);
+	dev_dbg(dev, "Set the frequency of bus (%luHz -> %luHz, %luHz)\n",
+			old_freq, new_freq, clk_get_rate(bus->clk));
 out:
 	mutex_unlock(&bus->lock);
 
@@ -214,17 +213,16 @@ static int exynos_bus_passive_target(struct device *dev, unsigned long *freq,
 	int ret = 0;
 
 	/* Get new opp-bus instance according to new bus clock */
-	rcu_read_lock();
 	new_opp = devfreq_recommended_opp(dev, freq, flags);
 	if (IS_ERR(new_opp)) {
 		dev_err(dev, "failed to get recommended opp instance\n");
-		rcu_read_unlock();
 		return PTR_ERR(new_opp);
 	}
 
 	new_freq = dev_pm_opp_get_freq(new_opp);
+	dev_pm_opp_put(new_opp);
+
 	old_freq = bus->curr_freq;
-	rcu_read_unlock();
 
 	if (old_freq == new_freq)
 		return 0;
@@ -241,8 +239,8 @@ static int exynos_bus_passive_target(struct device *dev, unsigned long *freq,
 	*freq = new_freq;
 	bus->curr_freq = new_freq;
 
-	dev_dbg(dev, "Set the frequency of bus (%lukHz -> %lukHz)\n",
-			old_freq/1000, new_freq/1000);
+	dev_dbg(dev, "Set the frequency of bus (%luHz -> %luHz, %luHz)\n",
+			old_freq, new_freq, clk_get_rate(bus->clk));
 out:
 	mutex_unlock(&bus->lock);
 
@@ -358,16 +356,14 @@ static int exynos_bus_parse_of(struct device_node *np,
 
 	rate = clk_get_rate(bus->clk);
 
-	rcu_read_lock();
 	opp = devfreq_recommended_opp(dev, &rate, 0);
 	if (IS_ERR(opp)) {
 		dev_err(dev, "failed to find dev_pm_opp\n");
-		rcu_read_unlock();
 		ret = PTR_ERR(opp);
 		goto err_opp;
 	}
 	bus->curr_freq = dev_pm_opp_get_freq(opp);
-	rcu_read_unlock();
+	dev_pm_opp_put(opp);
 
 	return 0;
 

drivers/devfreq/governor.h

@@ -38,4 +38,6 @@ extern void devfreq_interval_update(struct devfreq *devfreq,
 extern int devfreq_add_governor(struct devfreq_governor *governor);
 extern int devfreq_remove_governor(struct devfreq_governor *governor);
 
+extern int devfreq_update_status(struct devfreq *devfreq, unsigned long freq);
+
 #endif /* _GOVERNOR_H */

drivers/devfreq/governor_passive.c

@@ -59,14 +59,14 @@ static int devfreq_passive_get_target_freq(struct devfreq *devfreq,
 	 * list of parent device. Because in this case, *freq is temporary
 	 * value which is decided by ondemand governor.
 	 */
-	rcu_read_lock();
 	opp = devfreq_recommended_opp(parent_devfreq->dev.parent, freq, 0);
-	rcu_read_unlock();
 	if (IS_ERR(opp)) {
 		ret = PTR_ERR(opp);
 		goto out;
 	}
 
+	dev_pm_opp_put(opp);
+
 	/*
 	 * Get the OPP table's index of decided freqeuncy by governor
 	 * of parent device.
@@ -112,6 +112,11 @@ static int update_devfreq_passive(struct devfreq *devfreq, unsigned long freq)
 	if (ret < 0)
 		goto out;
 
+	if (devfreq->profile->freq_table
+		&& (devfreq_update_status(devfreq, freq)))
+		dev_err(&devfreq->dev,
+			"Couldn't update frequency transition information.\n");
+
 	devfreq->previous_freq = freq;
 
 out:
@@ -179,6 +184,7 @@ static int devfreq_passive_event_handler(struct devfreq *devfreq,
 
 static struct devfreq_governor devfreq_passive = {
 	.name = "passive",
+	.immutable = 1,
 	.get_target_freq = devfreq_passive_get_target_freq,
 	.event_handler = devfreq_passive_event_handler,
 };

drivers/devfreq/governor_userspace.c

 /*
- *  linux/drivers/devfreq/governor_simpleondemand.c
+ *  linux/drivers/devfreq/governor_userspace.c
  *
  *  Copyright (C) 2011 Samsung Electronics
  *	MyungJoo Ham <myungjoo.ham@samsung.com>
@@ -50,7 +50,6 @@ static ssize_t store_freq(struct device *dev, struct device_attribute *attr,
 	unsigned long wanted;
 	int err = 0;
 
-
 	mutex_lock(&devfreq->lock);
 	data = devfreq->data;
 
@@ -112,7 +111,13 @@ static int userspace_init(struct devfreq *devfreq)
 
 static void userspace_exit(struct devfreq *devfreq)
 {
+	/*
+	 * Remove the sysfs entry, unless this is being called after
+	 * device_del(), which should have done this already via kobject_del().
+	 */
+	if (devfreq->dev.kobj.sd)
 		sysfs_remove_group(&devfreq->dev.kobj, &dev_attr_group);
+
 	kfree(devfreq->data);
 	devfreq->data = NULL;
 }

drivers/devfreq/rk3399_dmc.c

@@ -91,17 +91,13 @@ static int rk3399_dmcfreq_target(struct device *dev, unsigned long *freq,
 	unsigned long target_volt, target_rate;
 	int err;
 
-	rcu_read_lock();
 	opp = devfreq_recommended_opp(dev, freq, flags);
-	if (IS_ERR(opp)) {
-		rcu_read_unlock();
+	if (IS_ERR(opp))
 		return PTR_ERR(opp);
-	}
 
 	target_rate = dev_pm_opp_get_freq(opp);
 	target_volt = dev_pm_opp_get_voltage(opp);
-
-	rcu_read_unlock();
+	dev_pm_opp_put(opp);
 
 	if (dmcfreq->rate == target_rate)
 		return 0;
@@ -422,15 +418,13 @@ static int rk3399_dmcfreq_probe(struct platform_device *pdev)
 
 	data->rate = clk_get_rate(data->dmc_clk);
 
-	rcu_read_lock();
 	opp = devfreq_recommended_opp(dev, &data->rate, 0);
-	if (IS_ERR(opp)) {
-		rcu_read_unlock();
+	if (IS_ERR(opp))
 		return PTR_ERR(opp);
-	}
+
 	data->rate = dev_pm_opp_get_freq(opp);
 	data->volt = dev_pm_opp_get_voltage(opp);
-	rcu_read_unlock();
+	dev_pm_opp_put(opp);
 
 	rk3399_devfreq_dmc_profile.initial_freq = data->rate;
 

drivers/devfreq/tegra-devfreq.c

@@ -487,15 +487,13 @@ static int tegra_devfreq_target(struct device *dev, unsigned long *freq,
 	struct dev_pm_opp *opp;
 	unsigned long rate = *freq * KHZ;
 
-	rcu_read_lock();
 	opp = devfreq_recommended_opp(dev, &rate, flags);
 	if (IS_ERR(opp)) {
-		rcu_read_unlock();
 		dev_err(dev, "Failed to find opp for %lu KHz\n", *freq);
 		return PTR_ERR(opp);
 	}
 	rate = dev_pm_opp_get_freq(opp);
-	rcu_read_unlock();
+	dev_pm_opp_put(opp);
 
 	clk_set_min_rate(tegra->emc_clock, rate);
 	clk_set_rate(tegra->emc_clock, 0);

drivers/thermal/cpu_cooling.c

@@ -297,8 +297,6 @@ static int build_dyn_power_table(struct cpufreq_cooling_device *cpufreq_device,
 	if (!power_table)
 		return -ENOMEM;
 
-	rcu_read_lock();
-
 	for (freq = 0, i = 0;
 	     opp = dev_pm_opp_find_freq_ceil(dev, &freq), !IS_ERR(opp);
 	     freq++, i++) {
@@ -306,13 +304,13 @@ static int build_dyn_power_table(struct cpufreq_cooling_device *cpufreq_device,
 		u64 power;
 
 		if (i >= num_opps) {
-			rcu_read_unlock();
 			ret = -EAGAIN;
 			goto free_power_table;
 		}
 
 		freq_mhz = freq / 1000000;
 		voltage_mv = dev_pm_opp_get_voltage(opp) / 1000;
+		dev_pm_opp_put(opp);
 
 		/*
 		 * Do the multiplication with MHz and millivolt so as
@@ -328,8 +326,6 @@ static int build_dyn_power_table(struct cpufreq_cooling_device *cpufreq_device,
 		power_table[i].power = power;
 	}
 
-	rcu_read_unlock();
-
 	if (i != num_opps) {
 		ret = PTR_ERR(opp);
 		goto free_power_table;
@@ -433,13 +429,10 @@ static int get_static_power(struct cpufreq_cooling_device *cpufreq_device,
 		return 0;
 	}
 
-	rcu_read_lock();
-
 	opp = dev_pm_opp_find_freq_exact(cpufreq_device->cpu_dev, freq_hz,
 					 true);
 	voltage = dev_pm_opp_get_voltage(opp);
-
-	rcu_read_unlock();
+	dev_pm_opp_put(opp);
 
 	if (voltage == 0) {
 		dev_warn_ratelimited(cpufreq_device->cpu_dev,

drivers/thermal/devfreq_cooling.c

@@ -113,15 +113,15 @@ static int partition_enable_opps(struct devfreq_cooling_device *dfc,
 		unsigned int freq = dfc->freq_table[i];
 		bool want_enable = i >= cdev_state ? true : false;
 
-		rcu_read_lock();
 		opp = dev_pm_opp_find_freq_exact(dev, freq, !want_enable);
-		rcu_read_unlock();
 
 		if (PTR_ERR(opp) == -ERANGE)
 			continue;
 		else if (IS_ERR(opp))
 			return PTR_ERR(opp);
 
+		dev_pm_opp_put(opp);
+
 		if (want_enable)
 			ret = dev_pm_opp_enable(dev, freq);
 		else
@@ -221,15 +221,12 @@ get_static_power(struct devfreq_cooling_device *dfc, unsigned long freq)
 	if (!dfc->power_ops->get_static_power)
 		return 0;
 
-	rcu_read_lock();
-
 	opp = dev_pm_opp_find_freq_exact(dev, freq, true);
 	if (IS_ERR(opp) && (PTR_ERR(opp) == -ERANGE))
 		opp = dev_pm_opp_find_freq_exact(dev, freq, false);
 
 	voltage = dev_pm_opp_get_voltage(opp) / 1000; /* mV */
-
-	rcu_read_unlock();
+	dev_pm_opp_put(opp);
 
 	if (voltage == 0) {
 		dev_warn_ratelimited(dev,
@@ -412,18 +409,14 @@ static int devfreq_cooling_gen_tables(struct devfreq_cooling_device *dfc)
 		unsigned long power_dyn, voltage;
 		struct dev_pm_opp *opp;
 
-		rcu_read_lock();
-
 		opp = dev_pm_opp_find_freq_floor(dev, &freq);
 		if (IS_ERR(opp)) {
-			rcu_read_unlock();
 			ret = PTR_ERR(opp);
 			goto free_tables;
 		}
 
 		voltage = dev_pm_opp_get_voltage(opp) / 1000; /* mV */
-
-		rcu_read_unlock();
+		dev_pm_opp_put(opp);
 
 		if (dfc->power_ops) {
 			power_dyn = get_dynamic_power(dfc, freq, voltage);

include/linux/cpufreq.h

@@ -31,7 +31,7 @@
 
 #define CPUFREQ_ETERNAL			(-1)
 #define CPUFREQ_NAME_LEN		16
-/* Print length for names. Extra 1 space for accomodating '\n' in prints */
+/* Print length for names. Extra 1 space for accommodating '\n' in prints */
 #define CPUFREQ_NAME_PLEN		(CPUFREQ_NAME_LEN + 1)
 
 struct cpufreq_governor;
@@ -115,7 +115,7 @@ struct cpufreq_policy {
 	 *   guarantee that frequency can be changed on any CPU sharing the
 	 *   policy and that the change will affect all of the policy CPUs then.
 	 * - fast_switch_enabled is to be set by governors that support fast
-	 *   freqnency switching with the help of cpufreq_enable_fast_switch().
+	 *   frequency switching with the help of cpufreq_enable_fast_switch().
 	 */
 	bool			fast_switch_possible;
 	bool			fast_switch_enabled;
@@ -415,9 +415,6 @@ static inline void cpufreq_resume(void) {}
 /* Policy Notifiers  */
 #define CPUFREQ_ADJUST			(0)
 #define CPUFREQ_NOTIFY			(1)
-#define CPUFREQ_START			(2)
-#define CPUFREQ_CREATE_POLICY		(3)
-#define CPUFREQ_REMOVE_POLICY		(4)
 
 #ifdef CONFIG_CPU_FREQ
 int cpufreq_register_notifier(struct notifier_block *nb, unsigned int list);

include/linux/devfreq.h

@@ -104,6 +104,8 @@ struct devfreq_dev_profile {
  * struct devfreq_governor - Devfreq policy governor
  * @node:		list node - contains registered devfreq governors
  * @name:		Governor's name
+ * @immutable:		Immutable flag for governor. If the value is 1,
+ *			this govenror is never changeable to other governor.
  * @get_target_freq:	Returns desired operating frequency for the device.
  *			Basically, get_target_freq will run
  *			devfreq_dev_profile.get_dev_status() to get the
@@ -121,6 +123,7 @@ struct devfreq_governor {
 	struct list_head node;
 
 	const char name[DEVFREQ_NAME_LEN];
+	const unsigned int immutable;
 	int (*get_target_freq)(struct devfreq *this, unsigned long *freq);
 	int (*event_handler)(struct devfreq *devfreq,
 				unsigned int event, void *data);

include/linux/pm_domain.h

@@ -182,6 +182,9 @@ static inline int pm_genpd_remove(struct generic_pm_domain *genpd)
 {
 	return -ENOTSUPP;
 }
+
+#define simple_qos_governor		(*(struct dev_power_governor *)(NULL))
+#define pm_domain_always_on_gov		(*(struct dev_power_governor *)(NULL))
 #endif
 
 static inline int pm_genpd_add_device(struct generic_pm_domain *genpd,

include/linux/pm_opp.h

@@ -78,6 +78,9 @@ struct dev_pm_set_opp_data {
 
 #if defined(CONFIG_PM_OPP)
 
+struct opp_table *dev_pm_opp_get_opp_table(struct device *dev);
+void dev_pm_opp_put_opp_table(struct opp_table *opp_table);
+
 unsigned long dev_pm_opp_get_voltage(struct dev_pm_opp *opp);
 
 unsigned long dev_pm_opp_get_freq(struct dev_pm_opp *opp);
@@ -88,7 +91,7 @@ int dev_pm_opp_get_opp_count(struct device *dev);
 unsigned long dev_pm_opp_get_max_clock_latency(struct device *dev);
 unsigned long dev_pm_opp_get_max_volt_latency(struct device *dev);
 unsigned long dev_pm_opp_get_max_transition_latency(struct device *dev);
-struct dev_pm_opp *dev_pm_opp_get_suspend_opp(struct device *dev);
+unsigned long dev_pm_opp_get_suspend_opp_freq(struct device *dev);
 
 struct dev_pm_opp *dev_pm_opp_find_freq_exact(struct device *dev,
 					      unsigned long freq,
@@ -99,6 +102,7 @@ struct dev_pm_opp *dev_pm_opp_find_freq_floor(struct device *dev,
 
 struct dev_pm_opp *dev_pm_opp_find_freq_ceil(struct device *dev,
 					     unsigned long *freq);
+void dev_pm_opp_put(struct dev_pm_opp *opp);
 
 int dev_pm_opp_add(struct device *dev, unsigned long freq,
 		   unsigned long u_volt);
@@ -108,22 +112,30 @@ int dev_pm_opp_enable(struct device *dev, unsigned long freq);
 
 int dev_pm_opp_disable(struct device *dev, unsigned long freq);
 
-struct srcu_notifier_head *dev_pm_opp_get_notifier(struct device *dev);
-int dev_pm_opp_set_supported_hw(struct device *dev, const u32 *versions,
-				unsigned int count);
-void dev_pm_opp_put_supported_hw(struct device *dev);
-int dev_pm_opp_set_prop_name(struct device *dev, const char *name);
-void dev_pm_opp_put_prop_name(struct device *dev);
+int dev_pm_opp_register_notifier(struct device *dev, struct notifier_block *nb);
+int dev_pm_opp_unregister_notifier(struct device *dev, struct notifier_block *nb);
+
+struct opp_table *dev_pm_opp_set_supported_hw(struct device *dev, const u32 *versions, unsigned int count);
+void dev_pm_opp_put_supported_hw(struct opp_table *opp_table);
+struct opp_table *dev_pm_opp_set_prop_name(struct device *dev, const char *name);
+void dev_pm_opp_put_prop_name(struct opp_table *opp_table);
 struct opp_table *dev_pm_opp_set_regulators(struct device *dev, const char * const names[], unsigned int count);
 void dev_pm_opp_put_regulators(struct opp_table *opp_table);
-int dev_pm_opp_register_set_opp_helper(struct device *dev, int (*set_opp)(struct dev_pm_set_opp_data *data));
-void dev_pm_opp_register_put_opp_helper(struct device *dev);
+struct opp_table *dev_pm_opp_register_set_opp_helper(struct device *dev, int (*set_opp)(struct dev_pm_set_opp_data *data));
+void dev_pm_opp_register_put_opp_helper(struct opp_table *opp_table);
 int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq);
 int dev_pm_opp_set_sharing_cpus(struct device *cpu_dev, const struct cpumask *cpumask);
 int dev_pm_opp_get_sharing_cpus(struct device *cpu_dev, struct cpumask *cpumask);
 void dev_pm_opp_remove_table(struct device *dev);
 void dev_pm_opp_cpumask_remove_table(const struct cpumask *cpumask);
 #else
+static inline struct opp_table *dev_pm_opp_get_opp_table(struct device *dev)
+{
+	return ERR_PTR(-ENOTSUPP);
+}
+
+static inline void dev_pm_opp_put_opp_table(struct opp_table *opp_table) {}
+
 static inline unsigned long dev_pm_opp_get_voltage(struct dev_pm_opp *opp)
 {
 	return 0;
@@ -159,9 +171,9 @@ static inline unsigned long dev_pm_opp_get_max_transition_latency(struct device
 	return 0;
 }
 
-static inline struct dev_pm_opp *dev_pm_opp_get_suspend_opp(struct device *dev)
+static inline unsigned long dev_pm_opp_get_suspend_opp_freq(struct device *dev)
 {
-	return NULL;
+	return 0;
 }
 
 static inline struct dev_pm_opp *dev_pm_opp_find_freq_exact(struct device *dev,
@@ -182,6 +194,8 @@ static inline struct dev_pm_opp *dev_pm_opp_find_freq_ceil(struct device *dev,
 	return ERR_PTR(-ENOTSUPP);
 }
 
+static inline void dev_pm_opp_put(struct dev_pm_opp *opp) {}
+
 static inline int dev_pm_opp_add(struct device *dev, unsigned long freq,
 					unsigned long u_volt)
 {
@@ -202,35 +216,39 @@ static inline int dev_pm_opp_disable(struct device *dev, unsigned long freq)
 	return 0;
 }
 
-static inline struct srcu_notifier_head *dev_pm_opp_get_notifier(
-							struct device *dev)
+static inline int dev_pm_opp_register_notifier(struct device *dev, struct notifier_block *nb)
 {
-	return ERR_PTR(-ENOTSUPP);
+	return -ENOTSUPP;
+}
+
+static inline int dev_pm_opp_unregister_notifier(struct device *dev, struct notifier_block *nb)
+{
+	return -ENOTSUPP;
 }
 
-static inline int dev_pm_opp_set_supported_hw(struct device *dev,
+static inline struct opp_table *dev_pm_opp_set_supported_hw(struct device *dev,
 							    const u32 *versions,
 							    unsigned int count)
 {
-	return -ENOTSUPP;
+	return ERR_PTR(-ENOTSUPP);
 }
 
-static inline void dev_pm_opp_put_supported_hw(struct device *dev) {}
+static inline void dev_pm_opp_put_supported_hw(struct opp_table *opp_table) {}
 
-static inline int dev_pm_opp_register_set_opp_helper(struct device *dev,
+static inline struct opp_table *dev_pm_opp_register_set_opp_helper(struct device *dev,
 			int (*set_opp)(struct dev_pm_set_opp_data *data))
 {
-	return -ENOTSUPP;
+	return ERR_PTR(-ENOTSUPP);
 }
 
-static inline void dev_pm_opp_register_put_opp_helper(struct device *dev) {}
+static inline void dev_pm_opp_register_put_opp_helper(struct opp_table *opp_table) {}
 
-static inline int dev_pm_opp_set_prop_name(struct device *dev, const char *name)
+static inline struct opp_table *dev_pm_opp_set_prop_name(struct device *dev, const char *name)
 {
-	return -ENOTSUPP;
+	return ERR_PTR(-ENOTSUPP);
 }
 
-static inline void dev_pm_opp_put_prop_name(struct device *dev) {}
+static inline void dev_pm_opp_put_prop_name(struct opp_table *opp_table) {}
 
 static inline struct opp_table *dev_pm_opp_set_regulators(struct device *dev, const char * const names[], unsigned int count)
 {
@@ -270,6 +288,7 @@ void dev_pm_opp_of_remove_table(struct device *dev);
 int dev_pm_opp_of_cpumask_add_table(const struct cpumask *cpumask);
 void dev_pm_opp_of_cpumask_remove_table(const struct cpumask *cpumask);
 int dev_pm_opp_of_get_sharing_cpus(struct device *cpu_dev, struct cpumask *cpumask);
+struct device_node *dev_pm_opp_of_get_opp_desc_node(struct device *dev);
 #else
 static inline int dev_pm_opp_of_add_table(struct device *dev)
 {
@@ -293,6 +312,11 @@ static inline int dev_pm_opp_of_get_sharing_cpus(struct device *cpu_dev, struct
 {
 	return -ENOTSUPP;
 }
+
+static inline struct device_node *dev_pm_opp_of_get_opp_desc_node(struct device *dev)
+{
+	return NULL;
+}
 #endif
 
 #endif		/* __LINUX_OPP_H__ */

include/linux/pm_qos.h

@@ -6,7 +6,6 @@
  */
 #include <linux/plist.h>
 #include <linux/notifier.h>
-#include <linux/miscdevice.h>
 #include <linux/device.h>
 #include <linux/workqueue.h>
 

kernel/power/suspend_test.c

@@ -166,7 +166,7 @@ static int __init setup_test_suspend(char *value)
 			return 0;
 	}
 
-	for (i = 0; pm_labels[i]; i++)
+	for (i = PM_SUSPEND_MIN; i < PM_SUSPEND_MAX; i++)
 		if (!strcmp(pm_labels[i], suspend_type)) {
 			test_state_label = pm_labels[i];
 			return 0;

kernel/power/swap.c

@@ -201,7 +201,7 @@ void free_all_swap_pages(int swap)
 		struct swsusp_extent *ext;
 		unsigned long offset;
 
-		ext = container_of(node, struct swsusp_extent, node);
+		ext = rb_entry(node, struct swsusp_extent, node);
 		rb_erase(node, &swsusp_extents);
 		for (offset = ext->start; offset <= ext->end; offset++)
 			swap_free(swp_entry(swap, offset));

scripts/analyze_suspend.py

@@ -24,11 +24,6 @@
 #	   https://01.org/suspendresume
 #	 Source repo
 #	   https://github.com/01org/suspendresume
-#	 Documentation
-#	   Getting Started
-#	     https://01.org/suspendresume/documentation/getting-started
-#	   Command List:
-#	     https://01.org/suspendresume/documentation/command-list
 #
 # Description:
 #	 This tool is designed to assist kernel and OS developers in optimizing
@@ -66,6 +61,8 @@ import platform
 from datetime import datetime
 import struct
 import ConfigParser
+from threading import Thread
+from subprocess import call, Popen, PIPE
 
 # ----------------- CLASSES --------------------
 
@@ -75,11 +72,15 @@ import ConfigParser
 #	 store system values and test parameters
 class SystemValues:
 	ansi = False
-	version = '4.2'
+	version = '4.5'
 	verbose = False
 	addlogs = False
-	mindevlen = 0.001
-	mincglen = 1.0
+	mindevlen = 0.0
+	mincglen = 0.0
+	cgphase = ''
+	cgtest = -1
+	callloopmaxgap = 0.0001
+	callloopmaxlen = 0.005
 	srgap = 0
 	cgexp = False
 	outdir = ''
@@ -92,6 +93,7 @@ class SystemValues:
 		'device_pm_callback_end',
 		'device_pm_callback_start'
 	]
+	logmsg = ''
 	testcommand = ''
 	mempath = '/dev/mem'
 	powerfile = '/sys/power/state'
@@ -117,19 +119,19 @@ class SystemValues:
 	usetracemarkers = True
 	usekprobes = True
 	usedevsrc = False
+	useprocmon = False
 	notestrun = False
+	mixedphaseheight = True
 	devprops = dict()
-	postresumetime = 0
+	predelay = 0
+	postdelay = 0
+	procexecfmt = 'ps - (?P<ps>.*)$'
 	devpropfmt = '# Device Properties: .*'
 	tracertypefmt = '# tracer: (?P<t>.*)'
 	firmwarefmt = '# fwsuspend (?P<s>[0-9]*) fwresume (?P<r>[0-9]*)$'
-	postresumefmt = '# post resume time (?P<t>[0-9]*)$'
 	stampfmt = '# suspend-(?P<m>[0-9]{2})(?P<d>[0-9]{2})(?P<y>[0-9]{2})-'+\
 				'(?P<H>[0-9]{2})(?P<M>[0-9]{2})(?P<S>[0-9]{2})'+\
 				' (?P<host>.*) (?P<mode>.*) (?P<kernel>.*)$'
-	kprobecolor = 'rgba(204,204,204,0.5)'
-	synccolor = 'rgba(204,204,204,0.5)'
-	debugfuncs = []
 	tracefuncs = {
 		'sys_sync': dict(),
 		'pm_prepare_console': dict(),
@@ -152,44 +154,66 @@ class SystemValues:
 		'CPU_OFF': {
 			'func':'_cpu_down',
 			'args_x86_64': {'cpu':'%di:s32'},
-			'format': 'CPU_OFF[{cpu}]',
-			'mask': 'CPU_.*_DOWN'
+			'format': 'CPU_OFF[{cpu}]'
 		},
 		'CPU_ON': {
 			'func':'_cpu_up',
 			'args_x86_64': {'cpu':'%di:s32'},
-			'format': 'CPU_ON[{cpu}]',
-			'mask': 'CPU_.*_UP'
+			'format': 'CPU_ON[{cpu}]'
 		},
 	}
 	dev_tracefuncs = {
 		# general wait/delay/sleep
-		'msleep': { 'args_x86_64': {'time':'%di:s32'} },
-		'udelay': { 'func':'__const_udelay', 'args_x86_64': {'loops':'%di:s32'} },
-		'acpi_os_stall': dict(),
+		'msleep': { 'args_x86_64': {'time':'%di:s32'}, 'ub': 1 },
+		'schedule_timeout_uninterruptible': { 'args_x86_64': {'timeout':'%di:s32'}, 'ub': 1 },
+		'schedule_timeout': { 'args_x86_64': {'timeout':'%di:s32'}, 'ub': 1 },
+		'udelay': { 'func':'__const_udelay', 'args_x86_64': {'loops':'%di:s32'}, 'ub': 1 },
+		'usleep_range': { 'args_x86_64': {'min':'%di:s32', 'max':'%si:s32'}, 'ub': 1 },
+		'mutex_lock_slowpath': { 'func':'__mutex_lock_slowpath', 'ub': 1 },
+		'acpi_os_stall': {'ub': 1},
 		# ACPI
 		'acpi_resume_power_resources': dict(),
 		'acpi_ps_parse_aml': dict(),
 		# filesystem
 		'ext4_sync_fs': dict(),
+		# 80211
+		'iwlagn_mac_start': dict(),
+		'iwlagn_alloc_bcast_station': dict(),
+		'iwl_trans_pcie_start_hw': dict(),
+		'iwl_trans_pcie_start_fw': dict(),
+		'iwl_run_init_ucode': dict(),
+		'iwl_load_ucode_wait_alive': dict(),
+		'iwl_alive_start': dict(),
+		'iwlagn_mac_stop': dict(),
+		'iwlagn_mac_suspend': dict(),
+		'iwlagn_mac_resume': dict(),
+		'iwlagn_mac_add_interface': dict(),
+		'iwlagn_mac_remove_interface': dict(),
+		'iwlagn_mac_change_interface': dict(),
+		'iwlagn_mac_config': dict(),
+		'iwlagn_configure_filter': dict(),
+		'iwlagn_mac_hw_scan': dict(),
+		'iwlagn_bss_info_changed': dict(),
+		'iwlagn_mac_channel_switch': dict(),
+		'iwlagn_mac_flush': dict(),
 		# ATA
 		'ata_eh_recover': { 'args_x86_64': {'port':'+36(%di):s32'} },
 		# i915
-		'i915_gem_restore_gtt_mappings': dict(),
+		'i915_gem_resume': dict(),
+		'i915_restore_state': dict(),
 		'intel_opregion_setup': dict(),
+		'g4x_pre_enable_dp': dict(),
+		'vlv_pre_enable_dp': dict(),
+		'chv_pre_enable_dp': dict(),
+		'g4x_enable_dp': dict(),
+		'vlv_enable_dp': dict(),
+		'intel_hpd_init': dict(),
+		'intel_opregion_register': dict(),
 		'intel_dp_detect': dict(),
 		'intel_hdmi_detect': dict(),
 		'intel_opregion_init': dict(),
+		'intel_fbdev_set_suspend': dict(),
 	}
-	kprobes_postresume = [
-		{
-			'name': 'ataportrst',
-			'func': 'ata_eh_recover',
-			'args': {'port':'+36(%di):s32'},
-			'format': 'ata{port}_port_reset',
-			'mask': 'ata.*_port_reset'
-		}
-	]
 	kprobes = dict()
 	timeformat = '%.3f'
 	def __init__(self):
@@ -198,6 +222,7 @@ class SystemValues:
 			self.embedded = True
 			self.addlogs = True
 			self.htmlfile = os.environ['LOG_FILE']
+		self.archargs = 'args_'+platform.machine()
 		self.hostname = platform.node()
 		if(self.hostname == ''):
 			self.hostname = 'localhost'
@@ -214,6 +239,13 @@ class SystemValues:
 		if num < 0 or num > 6:
 			return
 		self.timeformat = '%.{0}f'.format(num)
+	def setOutputFolder(self, value):
+		args = dict()
+		n = datetime.now()
+		args['date'] = n.strftime('%y%m%d')
+		args['time'] = n.strftime('%H%M%S')
+		args['hostname'] = self.hostname
+		self.outdir = value.format(**args)
 	def setOutputFile(self):
 		if((self.htmlfile == '') and (self.dmesgfile != '')):
 			m = re.match('(?P<name>.*)_dmesg\.txt$', self.dmesgfile)
@@ -253,10 +285,14 @@ class SystemValues:
 			self.testdir+'/'+self.prefix+'_'+self.suspendmode+'.html'
 		if not os.path.isdir(self.testdir):
 			os.mkdir(self.testdir)
-	def setDeviceFilter(self, devnames):
-		self.devicefilter = string.split(devnames)
+	def setDeviceFilter(self, value):
+		self.devicefilter = []
+		if value:
+			value = value.split(',')
+		for i in value:
+			self.devicefilter.append(i.strip())
 	def rtcWakeAlarmOn(self):
-		os.system('echo 0 > '+self.rtcpath+'/wakealarm')
+		call('echo 0 > '+self.rtcpath+'/wakealarm', shell=True)
 		outD = open(self.rtcpath+'/date', 'r').read().strip()
 		outT = open(self.rtcpath+'/time', 'r').read().strip()
 		mD = re.match('^(?P<y>[0-9]*)-(?P<m>[0-9]*)-(?P<d>[0-9]*)', outD)
@@ -272,12 +308,12 @@ class SystemValues:
 			# if hardware time fails, use the software time
 			nowtime = int(datetime.now().strftime('%s'))
 		alarm = nowtime + self.rtcwaketime
-		os.system('echo %d > %s/wakealarm' % (alarm, self.rtcpath))
+		call('echo %d > %s/wakealarm' % (alarm, self.rtcpath), shell=True)
 	def rtcWakeAlarmOff(self):
-		os.system('echo 0 > %s/wakealarm' % self.rtcpath)
+		call('echo 0 > %s/wakealarm' % self.rtcpath, shell=True)
 	def initdmesg(self):
 		# get the latest time stamp from the dmesg log
-		fp = os.popen('dmesg')
+		fp = Popen('dmesg', stdout=PIPE).stdout
 		ktime = '0'
 		for line in fp:
 			line = line.replace('\r\n', '')
@@ -291,7 +327,7 @@ class SystemValues:
 		self.dmesgstart = float(ktime)
 	def getdmesg(self):
 		# store all new dmesg lines since initdmesg was called
-		fp = os.popen('dmesg')
+		fp = Popen('dmesg', stdout=PIPE).stdout
 		op = open(self.dmesgfile, 'a')
 		for line in fp:
 			line = line.replace('\r\n', '')
@@ -317,18 +353,11 @@ class SystemValues:
 	def getFtraceFilterFunctions(self, current):
 		rootCheck(True)
 		if not current:
-			os.system('cat '+self.tpath+'available_filter_functions')
+			call('cat '+self.tpath+'available_filter_functions', shell=True)
 			return
 		fp = open(self.tpath+'available_filter_functions')
 		master = fp.read().split('\n')
 		fp.close()
-		if len(self.debugfuncs) > 0:
-			for i in self.debugfuncs:
-				if i in master:
-					print i
-				else:
-					print self.colorText(i)
-		else:
 		for i in self.tracefuncs:
 			if 'func' in self.tracefuncs[i]:
 				i = self.tracefuncs[i]['func']
@@ -351,22 +380,15 @@ class SystemValues:
 		fp = open(self.tpath+'set_graph_function', 'w')
 		fp.write(flist)
 		fp.close()
-	def kprobeMatch(self, name, target):
-		if name not in self.kprobes:
-			return False
-		if re.match(self.kprobes[name]['mask'], target):
-			return True
-		return False
 	def basicKprobe(self, name):
-		self.kprobes[name] = {'name': name,'func': name,'args': dict(),'format': name,'mask': name}
+		self.kprobes[name] = {'name': name,'func': name,'args': dict(),'format': name}
 	def defaultKprobe(self, name, kdata):
 		k = kdata
-		for field in ['name', 'format', 'mask', 'func']:
+		for field in ['name', 'format', 'func']:
 			if field not in k:
 				k[field] = name
-		archargs = 'args_'+platform.machine()
-		if archargs in k:
-			k['args'] = k[archargs]
+		if self.archargs in k:
+			k['args'] = k[self.archargs]
 		else:
 			k['args'] = dict()
 			k['format'] = name
@@ -403,49 +425,80 @@ class SystemValues:
 		out = fmt.format(**arglist)
 		out = out.replace(' ', '_').replace('"', '')
 		return out
-	def kprobeText(self, kprobe):
-		name, fmt, func, args = kprobe['name'], kprobe['format'], kprobe['func'], kprobe['args']
+	def kprobeText(self, kname, kprobe):
+		name = fmt = func = kname
+		args = dict()
+		if 'name' in kprobe:
+			name = kprobe['name']
+		if 'format' in kprobe:
+			fmt = kprobe['format']
+		if 'func' in kprobe:
+			func = kprobe['func']
+		if self.archargs in kprobe:
+			args = kprobe[self.archargs]
+		if 'args' in kprobe:
+			args = kprobe['args']
 		if re.findall('{(?P<n>[a-z,A-Z,0-9]*)}', func):
-			doError('Kprobe "%s" has format info in the function name "%s"' % (name, func), False)
+			doError('Kprobe "%s" has format info in the function name "%s"' % (name, func))
 		for arg in re.findall('{(?P<n>[a-z,A-Z,0-9]*)}', fmt):
 			if arg not in args:
-				doError('Kprobe "%s" is missing argument "%s"' % (name, arg), False)
+				doError('Kprobe "%s" is missing argument "%s"' % (name, arg))
 		val = 'p:%s_cal %s' % (name, func)
 		for i in sorted(args):
 			val += ' %s=%s' % (i, args[i])
 		val += '\nr:%s_ret %s $retval\n' % (name, func)
 		return val
-	def addKprobes(self):
+	def addKprobes(self, output=False):
+		if len(sysvals.kprobes) < 1:
+			return
+		if output:
+			print('    kprobe functions in this kernel:')
 		# first test each kprobe
-		print('INITIALIZING KPROBES...')
 		rejects = []
+		# sort kprobes: trace, ub-dev, custom, dev
+		kpl = [[], [], [], []]
 		for name in sorted(self.kprobes):
-			if not self.testKprobe(self.kprobes[name]):
+			res = self.colorText('YES', 32)
+			if not self.testKprobe(name, self.kprobes[name]):
+				res = self.colorText('NO')
 				rejects.append(name)
+			else:
+				if name in self.tracefuncs:
+					kpl[0].append(name)
+				elif name in self.dev_tracefuncs:
+					if 'ub' in self.dev_tracefuncs[name]:
+						kpl[1].append(name)
+					else:
+						kpl[3].append(name)
+				else:
+					kpl[2].append(name)
+			if output:
+				print('         %s: %s' % (name, res))
+		kplist = kpl[0] + kpl[1] + kpl[2] + kpl[3]
 		# remove all failed ones from the list
 		for name in rejects:
-			vprint('Skipping KPROBE: %s' % name)
 			self.kprobes.pop(name)
+		# set the kprobes all at once
 		self.fsetVal('', 'kprobe_events')
 		kprobeevents = ''
-		# set the kprobes all at once
-		for kp in self.kprobes:
-			val = self.kprobeText(self.kprobes[kp])
-			vprint('Adding KPROBE: %s\n%s' % (kp, val.strip()))
-			kprobeevents += self.kprobeText(self.kprobes[kp])
+		for kp in kplist:
+			kprobeevents += self.kprobeText(kp, self.kprobes[kp])
 		self.fsetVal(kprobeevents, 'kprobe_events')
 		# verify that the kprobes were set as ordered
 		check = self.fgetVal('kprobe_events')
-		linesout = len(kprobeevents.split('\n'))
-		linesack = len(check.split('\n'))
+		linesout = len(kprobeevents.split('\n')) - 1
+		linesack = len(check.split('\n')) - 1
+		if output:
+			res = '%d/%d' % (linesack, linesout)
 			if linesack < linesout:
-			# if not, try appending the kprobes 1 by 1
-			for kp in self.kprobes:
-				kprobeevents = self.kprobeText(self.kprobes[kp])
-				self.fsetVal(kprobeevents, 'kprobe_events', 'a')
+				res = self.colorText(res, 31)
+			else:
+				res = self.colorText(res, 32)
+			print('    working kprobe functions enabled: %s' % res)
 		self.fsetVal('1', 'events/kprobes/enable')
-	def testKprobe(self, kprobe):
-		kprobeevents = self.kprobeText(kprobe)
+	def testKprobe(self, kname, kprobe):
+		self.fsetVal('0', 'events/kprobes/enable')
+		kprobeevents = self.kprobeText(kname, kprobe)
 		if not kprobeevents:
 			return False
 		try:
@@ -463,8 +516,9 @@ class SystemValues:
 		if not os.path.exists(file):
 			return False
 		try:
-			fp = open(file, mode)
+			fp = open(file, mode, 0)
 			fp.write(val)
+			fp.flush()
 			fp.close()
 		except:
 			pass
@@ -491,21 +545,17 @@ class SystemValues:
 		for name in self.dev_tracefuncs:
 			self.defaultKprobe(name, self.dev_tracefuncs[name])
 	def isCallgraphFunc(self, name):
-		if len(self.debugfuncs) < 1 and self.suspendmode == 'command':
-			return True
-		if name in self.debugfuncs:
+		if len(self.tracefuncs) < 1 and self.suspendmode == 'command':
 			return True
-		funclist = []
 		for i in self.tracefuncs:
 			if 'func' in self.tracefuncs[i]:
-				funclist.append(self.tracefuncs[i]['func'])
+				f = self.tracefuncs[i]['func']
 			else:
-				funclist.append(i)
-		if name in funclist:
+				f = i
+			if name == f:
 				return True
 		return False
 	def initFtrace(self, testing=False):
-		tp = self.tpath
 		print('INITIALIZING FTRACE...')
 		# turn trace off
 		self.fsetVal('0', 'tracing_on')
@@ -518,18 +568,7 @@ class SystemValues:
 		# go no further if this is just a status check
 		if testing:
 			return
-		if self.usekprobes:
-			# add tracefunc kprobes so long as were not using full callgraph
-			if(not self.usecallgraph or len(self.debugfuncs) > 0):
-				for name in self.tracefuncs:
-					self.defaultKprobe(name, self.tracefuncs[name])
-				if self.usedevsrc:
-					for name in self.dev_tracefuncs:
-						self.defaultKprobe(name, self.dev_tracefuncs[name])
-			else:
-				self.usedevsrc = False
-			self.addKprobes()
-		# initialize the callgraph trace, unless this is an x2 run
+		# initialize the callgraph trace
 		if(self.usecallgraph):
 			# set trace type
 			self.fsetVal('function_graph', 'current_tracer')
@@ -545,11 +584,6 @@ class SystemValues:
 			self.fsetVal('context-info', 'trace_options')
 			self.fsetVal('graph-time', 'trace_options')
 			self.fsetVal('0', 'max_graph_depth')
-			if len(self.debugfuncs) > 0:
-				self.setFtraceFilterFunctions(self.debugfuncs)
-			elif self.suspendmode == 'command':
-				self.fsetVal('', 'set_graph_function')
-			else:
 			cf = ['dpm_run_callback']
 			if(self.usetraceeventsonly):
 				cf += ['dpm_prepare', 'dpm_complete']
@@ -559,6 +593,15 @@ class SystemValues:
 				else:
 					cf.append(fn)
 			self.setFtraceFilterFunctions(cf)
+		# initialize the kprobe trace
+		elif self.usekprobes:
+			for name in self.tracefuncs:
+				self.defaultKprobe(name, self.tracefuncs[name])
+			if self.usedevsrc:
+				for name in self.dev_tracefuncs:
+					self.defaultKprobe(name, self.dev_tracefuncs[name])
+			print('INITIALIZING KPROBES...')
+			self.addKprobes(self.verbose)
 		if(self.usetraceevents):
 			# turn trace events on
 			events = iter(self.traceevents)
@@ -590,10 +633,10 @@ class SystemValues:
 			if(os.path.exists(tp+f) == False):
 				return False
 		return True
-	def colorText(self, str):
+	def colorText(self, str, color=31):
 		if not self.ansi:
 			return str
-		return '\x1B[31;40m'+str+'\x1B[m'
+		return '\x1B[%d;40m%s\x1B[m' % (color, str)
 
 sysvals = SystemValues()
 
@@ -625,8 +668,8 @@ class DevProps:
 		if self.xtraclass:
 			return ' '+self.xtraclass
 		if self.async:
-			return ' async'
-		return ' sync'
+			return ' async_device'
+		return ' sync_device'
 
 # Class: DeviceNode
 # Description:
@@ -646,8 +689,6 @@ class DeviceNode:
 #	 The primary container for suspend/resume test data. There is one for
 #	 each test run. The data is organized into a cronological hierarchy:
 #	 Data.dmesg {
-#		root structure, started as dmesg & ftrace, but now only ftrace
-#		contents: times for suspend start/end, resume start/end, fwdata
 #		phases {
 #			10 sequential, non-overlapping phases of S/R
 #			contents: times for phase start/end, order/color data for html
@@ -658,7 +699,7 @@ class DeviceNode:
 #					contents: start/stop times, pid/cpu/driver info
 #						parents/children, html id for timeline/callgraph
 #						optionally includes an ftrace callgraph
-#						optionally includes intradev trace events
+#						optionally includes dev/ps data
 #				}
 #			}
 #		}
@@ -671,19 +712,24 @@ class Data:
 	end = 0.0   # test end
 	tSuspended = 0.0 # low-level suspend start
 	tResumed = 0.0   # low-level resume start
+	tKernSus = 0.0   # kernel level suspend start
+	tKernRes = 0.0   # kernel level resume end
 	tLow = 0.0       # time spent in low-level suspend (standby/freeze)
 	fwValid = False  # is firmware data available
 	fwSuspend = 0    # time spent in firmware suspend
 	fwResume = 0     # time spent in firmware resume
 	dmesgtext = []   # dmesg text file in memory
+	pstl = 0         # process timeline
 	testnumber = 0
 	idstr = ''
 	html_device_id = 0
 	stamp = 0
 	outfile = ''
-	dev_ubiquitous = ['msleep', 'udelay']
+	devpids = []
+	kerror = False
 	def __init__(self, num):
-		idchar = 'abcdefghijklmnopqrstuvwxyz'
+		idchar = 'abcdefghij'
+		self.pstl = dict()
 		self.testnumber = num
 		self.idstr = idchar[num]
 		self.dmesgtext = []
@@ -714,16 +760,39 @@ class Data:
 		self.devicegroups = []
 		for phase in self.phases:
 			self.devicegroups.append([phase])
-	def getStart(self):
-		return self.dmesg[self.phases[0]]['start']
+		self.errorinfo = {'suspend':[],'resume':[]}
+	def extractErrorInfo(self, dmesg):
+		error = ''
+		tm = 0.0
+		for i in range(len(dmesg)):
+			if 'Call Trace:' in dmesg[i]:
+				m = re.match('[ \t]*(\[ *)(?P<ktime>[0-9\.]*)(\]) .*', dmesg[i])
+				if not m:
+					continue
+				tm = float(m.group('ktime'))
+				if tm < self.start or tm > self.end:
+					continue
+				for j in range(i-10, i+1):
+					error += dmesg[j]
+				continue
+			if error:
+				m = re.match('[ \t]*\[ *[0-9\.]*\]  \[\<[0-9a-fA-F]*\>\] .*', dmesg[i])
+				if m:
+					error += dmesg[i]
+				else:
+					if tm < self.tSuspended:
+						dir = 'suspend'
+					else:
+						dir = 'resume'
+					error = error.replace('<', '&lt').replace('>', '&gt')
+					vprint('kernel error found in %s at %f' % (dir, tm))
+					self.errorinfo[dir].append((tm, error))
+					self.kerror = True
+					error = ''
 	def setStart(self, time):
 		self.start = time
-		self.dmesg[self.phases[0]]['start'] = time
-	def getEnd(self):
-		return self.dmesg[self.phases[-1]]['end']
 	def setEnd(self, time):
 		self.end = time
-		self.dmesg[self.phases[-1]]['end'] = time
 	def isTraceEventOutsideDeviceCalls(self, pid, time):
 		for phase in self.phases:
 			list = self.dmesg[phase]['list']
@@ -733,39 +802,67 @@ class Data:
 					time < d['end']):
 					return False
 		return True
-	def targetDevice(self, phaselist, start, end, pid=-1):
+	def sourcePhase(self, start):
+		for phase in self.phases:
+			pend = self.dmesg[phase]['end']
+			if start <= pend:
+				return phase
+		return 'resume_complete'
+	def sourceDevice(self, phaselist, start, end, pid, type):
 		tgtdev = ''
 		for phase in phaselist:
 			list = self.dmesg[phase]['list']
 			for devname in list:
 				dev = list[devname]
-				if(pid >= 0 and dev['pid'] != pid):
+				# pid must match
+				if dev['pid'] != pid:
 					continue
 				devS = dev['start']
 				devE = dev['end']
+				if type == 'device':
+					# device target event is entirely inside the source boundary
 					if(start < devS or start >= devE or end <= devS or end > devE):
 						continue
+				elif type == 'thread':
+					# thread target event will expand the source boundary
+					if start < devS:
+						dev['start'] = start
+					if end > devE:
+						dev['end'] = end
 				tgtdev = dev
 				break
 		return tgtdev
 	def addDeviceFunctionCall(self, displayname, kprobename, proc, pid, start, end, cdata, rdata):
-		machstart = self.dmesg['suspend_machine']['start']
-		machend = self.dmesg['resume_machine']['end']
-		tgtdev = self.targetDevice(self.phases, start, end, pid)
-		if not tgtdev and start >= machstart and end < machend:
-			# device calls in machine phases should be serial
-			tgtdev = self.targetDevice(['suspend_machine', 'resume_machine'], start, end)
+		# try to place the call in a device
+		tgtdev = self.sourceDevice(self.phases, start, end, pid, 'device')
+		# calls with device pids that occur outside device bounds are dropped
+		# TODO: include these somehow
+		if not tgtdev and pid in self.devpids:
+			return False
+		# try to place the call in a thread
+		if not tgtdev:
+			tgtdev = self.sourceDevice(self.phases, start, end, pid, 'thread')
+		# create new thread blocks, expand as new calls are found
 		if not tgtdev:
-			if 'scsi_eh' in proc:
-				self.newActionGlobal(proc, start, end, pid)
-				self.addDeviceFunctionCall(displayname, kprobename, proc, pid, start, end, cdata, rdata)
+			if proc == '<...>':
+				threadname = 'kthread-%d' % (pid)
 			else:
-				vprint('IGNORE: %s[%s](%d) [%f - %f] | %s | %s | %s' % (displayname, kprobename,
-					pid, start, end, cdata, rdata, proc))
+				threadname = '%s-%d' % (proc, pid)
+			tgtphase = self.sourcePhase(start)
+			self.newAction(tgtphase, threadname, pid, '', start, end, '', ' kth', '')
+			return self.addDeviceFunctionCall(displayname, kprobename, proc, pid, start, end, cdata, rdata)
+		# this should not happen
+		if not tgtdev:
+			vprint('[%f - %f] %s-%d %s %s %s' % \
+				(start, end, proc, pid, kprobename, cdata, rdata))
 			return False
-		# detail block fits within tgtdev
+		# place the call data inside the src element of the tgtdev
 		if('src' not in tgtdev):
 			tgtdev['src'] = []
+		dtf = sysvals.dev_tracefuncs
+		ubiquitous = False
+		if kprobename in dtf and 'ub' in dtf[kprobename]:
+			ubiquitous = True
 		title = cdata+' '+rdata
 		mstr = '\(.*\) *(?P<args>.*) *\((?P<caller>.*)\+.* arg1=(?P<ret>.*)'
 		m = re.match(mstr, title)
@@ -777,14 +874,81 @@ class Data:
 				r = ''
 			else:
 				r = 'ret=%s ' % r
-			l = '%0.3fms' % ((end - start) * 1000)
-			if kprobename in self.dev_ubiquitous:
-				title = '%s(%s) <- %s, %s(%s)' % (displayname, a, c, r, l)
-			else:
-				title = '%s(%s) %s(%s)' % (displayname, a, r, l)
-		e = TraceEvent(title, kprobename, start, end - start)
+			if ubiquitous and c in dtf and 'ub' in dtf[c]:
+				return False
+		color = sysvals.kprobeColor(kprobename)
+		e = DevFunction(displayname, a, c, r, start, end, ubiquitous, proc, pid, color)
 		tgtdev['src'].append(e)
 		return True
+	def overflowDevices(self):
+		# get a list of devices that extend beyond the end of this test run
+		devlist = []
+		for phase in self.phases:
+			list = self.dmesg[phase]['list']
+			for devname in list:
+				dev = list[devname]
+				if dev['end'] > self.end:
+					devlist.append(dev)
+		return devlist
+	def mergeOverlapDevices(self, devlist):
+		# merge any devices that overlap devlist
+		for dev in devlist:
+			devname = dev['name']
+			for phase in self.phases:
+				list = self.dmesg[phase]['list']
+				if devname not in list:
+					continue
+				tdev = list[devname]
+				o = min(dev['end'], tdev['end']) - max(dev['start'], tdev['start'])
+				if o <= 0:
+					continue
+				dev['end'] = tdev['end']
+				if 'src' not in dev or 'src' not in tdev:
+					continue
+				dev['src'] += tdev['src']
+				del list[devname]
+	def usurpTouchingThread(self, name, dev):
+		# the caller test has priority of this thread, give it to him
+		for phase in self.phases:
+			list = self.dmesg[phase]['list']
+			if name in list:
+				tdev = list[name]
+				if tdev['start'] - dev['end'] < 0.1:
+					dev['end'] = tdev['end']
+					if 'src' not in dev:
+						dev['src'] = []
+					if 'src' in tdev:
+						dev['src'] += tdev['src']
+					del list[name]
+				break
+	def stitchTouchingThreads(self, testlist):
+		# merge any threads between tests that touch
+		for phase in self.phases:
+			list = self.dmesg[phase]['list']
+			for devname in list:
+				dev = list[devname]
+				if 'htmlclass' not in dev or 'kth' not in dev['htmlclass']:
+					continue
+				for data in testlist:
+					data.usurpTouchingThread(devname, dev)
+	def optimizeDevSrc(self):
+		# merge any src call loops to reduce timeline size
+		for phase in self.phases:
+			list = self.dmesg[phase]['list']
+			for dev in list:
+				if 'src' not in list[dev]:
+					continue
+				src = list[dev]['src']
+				p = 0
+				for e in sorted(src, key=lambda event: event.time):
+					if not p or not e.repeat(p):
+						p = e
+						continue
+					# e is another iteration of p, move it into p
+					p.end = e.end
+					p.length = p.end - p.time
+					p.count += 1
+					src.remove(e)
 	def trimTimeVal(self, t, t0, dT, left):
 		if left:
 			if(t > t0):
@@ -804,6 +968,8 @@ class Data:
 		self.tSuspended = self.trimTimeVal(self.tSuspended, t0, dT, left)
 		self.tResumed = self.trimTimeVal(self.tResumed, t0, dT, left)
 		self.start = self.trimTimeVal(self.start, t0, dT, left)
+		self.tKernSus = self.trimTimeVal(self.tKernSus, t0, dT, left)
+		self.tKernRes = self.trimTimeVal(self.tKernRes, t0, dT, left)
 		self.end = self.trimTimeVal(self.end, t0, dT, left)
 		for phase in self.phases:
 			p = self.dmesg[phase]
@@ -832,36 +998,6 @@ class Data:
 			else:
 				self.trimTime(self.tSuspended, \
 					self.tResumed-self.tSuspended, False)
-	def newPhaseWithSingleAction(self, phasename, devname, start, end, color):
-		for phase in self.phases:
-			self.dmesg[phase]['order'] += 1
-		self.html_device_id += 1
-		devid = '%s%d' % (self.idstr, self.html_device_id)
-		list = dict()
-		list[devname] = \
-			{'start': start, 'end': end, 'pid': 0, 'par': '',
-			'length': (end-start), 'row': 0, 'id': devid, 'drv': '' };
-		self.dmesg[phasename] = \
-			{'list': list, 'start': start, 'end': end,
-			'row': 0, 'color': color, 'order': 0}
-		self.phases = self.sortedPhases()
-	def newPhase(self, phasename, start, end, color, order):
-		if(order < 0):
-			order = len(self.phases)
-		for phase in self.phases[order:]:
-			self.dmesg[phase]['order'] += 1
-		if(order > 0):
-			p = self.phases[order-1]
-			self.dmesg[p]['end'] = start
-		if(order < len(self.phases)):
-			p = self.phases[order]
-			self.dmesg[p]['start'] = end
-		list = dict()
-		self.dmesg[phasename] = \
-			{'list': list, 'start': start, 'end': end,
-			'row': 0, 'color': color, 'order': order}
-		self.phases = self.sortedPhases()
-		self.devicegroups.append([phasename])
 	def setPhase(self, phase, ktime, isbegin):
 		if(isbegin):
 			self.dmesg[phase]['start'] = ktime
@@ -881,7 +1017,7 @@ class Data:
 		for t in sorted(tmp):
 			slist.append(tmp[t])
 		return slist
-	def fixupInitcalls(self, phase, end):
+	def fixupInitcalls(self, phase):
 		# if any calls never returned, clip them at system resume end
 		phaselist = self.dmesg[phase]['list']
 		for devname in phaselist:
@@ -893,37 +1029,23 @@ class Data:
 						break
 				vprint('%s (%s): callback didnt return' % (devname, phase))
 	def deviceFilter(self, devicefilter):
-		# remove all by the relatives of the filter devnames
-		filter = []
-		for phase in self.phases:
-			list = self.dmesg[phase]['list']
-			for name in devicefilter:
-				dev = name
-				while(dev in list):
-					if(dev not in filter):
-						filter.append(dev)
-					dev = list[dev]['par']
-				children = self.deviceDescendants(name, phase)
-				for dev in children:
-					if(dev not in filter):
-						filter.append(dev)
 		for phase in self.phases:
 			list = self.dmesg[phase]['list']
 			rmlist = []
 			for name in list:
-				pid = list[name]['pid']
-				if(name not in filter and pid >= 0):
+				keep = False
+				for filter in devicefilter:
+					if filter in name or \
+						('drv' in list[name] and filter in list[name]['drv']):
+						keep = True
+				if not keep:
 					rmlist.append(name)
 			for name in rmlist:
 				del list[name]
 	def fixupInitcallsThatDidntReturn(self):
 		# if any calls never returned, clip them at system resume end
 		for phase in self.phases:
-			self.fixupInitcalls(phase, self.getEnd())
-	def isInsideTimeline(self, start, end):
-		if(self.start <= start and self.end > start):
-			return True
-		return False
+			self.fixupInitcalls(phase)
 	def phaseOverlap(self, phases):
 		rmgroups = []
 		newgroup = []
@@ -940,30 +1062,35 @@ class Data:
 			self.devicegroups.remove(group)
 		self.devicegroups.append(newgroup)
 	def newActionGlobal(self, name, start, end, pid=-1, color=''):
-		# if event starts before timeline start, expand timeline
-		if(start < self.start):
-			self.setStart(start)
-		# if event ends after timeline end, expand the timeline
-		if(end > self.end):
-			self.setEnd(end)
-		# which phase is this device callback or action "in"
-		targetphase = "none"
+		# which phase is this device callback or action in
+		targetphase = 'none'
 		htmlclass = ''
 		overlap = 0.0
 		phases = []
 		for phase in self.phases:
 			pstart = self.dmesg[phase]['start']
 			pend = self.dmesg[phase]['end']
+			# see if the action overlaps this phase
 			o = max(0, min(end, pend) - max(start, pstart))
 			if o > 0:
 				phases.append(phase)
+			# set the target phase to the one that overlaps most
 			if o > overlap:
 				if overlap > 0 and phase == 'post_resume':
 					continue
 				targetphase = phase
 				overlap = o
+		# if no target phase was found, pin it to the edge
+		if targetphase == 'none':
+			p0start = self.dmesg[self.phases[0]]['start']
+			if start <= p0start:
+				targetphase = self.phases[0]
+			else:
+				targetphase = self.phases[-1]
 		if pid == -2:
 			htmlclass = ' bg'
+		elif pid == -3:
+			htmlclass = ' ps'
 		if len(phases) > 1:
 			htmlclass = ' bg'
 			self.phaseOverlap(phases)
@@ -985,29 +1112,13 @@ class Data:
 			while(name in list):
 				name = '%s[%d]' % (origname, i)
 				i += 1
-		list[name] = {'start': start, 'end': end, 'pid': pid, 'par': parent,
-					  'length': length, 'row': 0, 'id': devid, 'drv': drv }
+		list[name] = {'name': name, 'start': start, 'end': end, 'pid': pid,
+			'par': parent, 'length': length, 'row': 0, 'id': devid, 'drv': drv }
 		if htmlclass:
 			list[name]['htmlclass'] = htmlclass
 		if color:
 			list[name]['color'] = color
 		return name
-	def deviceIDs(self, devlist, phase):
-		idlist = []
-		list = self.dmesg[phase]['list']
-		for devname in list:
-			if devname in devlist:
-				idlist.append(list[devname]['id'])
-		return idlist
-	def deviceParentID(self, devname, phase):
-		pdev = ''
-		pdevid = ''
-		list = self.dmesg[phase]['list']
-		if devname in list:
-			pdev = list[devname]['par']
-		if pdev in list:
-			return list[pdev]['id']
-		return pdev
 	def deviceChildren(self, devname, phase):
 		devlist = []
 		list = self.dmesg[phase]['list']
@@ -1015,21 +1126,15 @@ class Data:
 			if(list[child]['par'] == devname):
 				devlist.append(child)
 		return devlist
-	def deviceDescendants(self, devname, phase):
-		children = self.deviceChildren(devname, phase)
-		family = children
-		for child in children:
-			family += self.deviceDescendants(child, phase)
-		return family
-	def deviceChildrenIDs(self, devname, phase):
-		devlist = self.deviceChildren(devname, phase)
-		return self.deviceIDs(devlist, phase)
 	def printDetails(self):
+		vprint('Timeline Details:')
 		vprint('          test start: %f' % self.start)
+		vprint('kernel suspend start: %f' % self.tKernSus)
 		for phase in self.phases:
 			dc = len(self.dmesg[phase]['list'])
 			vprint('    %16s: %f - %f (%d devices)' % (phase, \
 				self.dmesg[phase]['start'], self.dmesg[phase]['end'], dc))
+		vprint('   kernel resume end: %f' % self.tKernRes)
 		vprint('            test end: %f' % self.end)
 	def deviceChildrenAllPhases(self, devname):
 		devlist = []
@@ -1108,21 +1213,134 @@ class Data:
 				if width != '0.000000' and length >= mindevlen:
 					devlist.append(dev)
 			self.tdevlist[phase] = devlist
-
-# Class: TraceEvent
+	def addHorizontalDivider(self, devname, devend):
+		phase = 'suspend_prepare'
+		self.newAction(phase, devname, -2, '', \
+			self.start, devend, '', ' sec', '')
+		if phase not in self.tdevlist:
+			self.tdevlist[phase] = []
+		self.tdevlist[phase].append(devname)
+		d = DevItem(0, phase, self.dmesg[phase]['list'][devname])
+		return d
+	def addProcessUsageEvent(self, name, times):
+		# get the start and end times for this process
+		maxC = 0
+		tlast = 0
+		start = -1
+		end = -1
+		for t in sorted(times):
+			if tlast == 0:
+				tlast = t
+				continue
+			if name in self.pstl[t]:
+				if start == -1 or tlast < start:
+					start = tlast
+				if end == -1 or t > end:
+					end = t
+			tlast = t
+		if start == -1 or end == -1:
+			return 0
+		# add a new action for this process and get the object
+		out = self.newActionGlobal(name, start, end, -3)
+		if not out:
+			return 0
+		phase, devname = out
+		dev = self.dmesg[phase]['list'][devname]
+		# get the cpu exec data
+		tlast = 0
+		clast = 0
+		cpuexec = dict()
+		for t in sorted(times):
+			if tlast == 0 or t <= start or t > end:
+				tlast = t
+				continue
+			list = self.pstl[t]
+			c = 0
+			if name in list:
+				c = list[name]
+			if c > maxC:
+				maxC = c
+			if c != clast:
+				key = (tlast, t)
+				cpuexec[key] = c
+				tlast = t
+				clast = c
+		dev['cpuexec'] = cpuexec
+		return maxC
+	def createProcessUsageEvents(self):
+		# get an array of process names
+		proclist = []
+		for t in self.pstl:
+			pslist = self.pstl[t]
+			for ps in pslist:
+				if ps not in proclist:
+					proclist.append(ps)
+		# get a list of data points for suspend and resume
+		tsus = []
+		tres = []
+		for t in sorted(self.pstl):
+			if t < self.tSuspended:
+				tsus.append(t)
+			else:
+				tres.append(t)
+		# process the events for suspend and resume
+		if len(proclist) > 0:
+			vprint('Process Execution:')
+		for ps in proclist:
+			c = self.addProcessUsageEvent(ps, tsus)
+			if c > 0:
+				vprint('%25s (sus): %d' % (ps, c))
+			c = self.addProcessUsageEvent(ps, tres)
+			if c > 0:
+				vprint('%25s (res): %d' % (ps, c))
+
+# Class: DevFunction
 # Description:
-#	 A container for trace event data found in the ftrace file
-class TraceEvent:
-	text = ''
-	time = 0.0
-	length = 0.0
-	title = ''
+#	 A container for kprobe function data we want in the dev timeline
+class DevFunction:
 	row = 0
-	def __init__(self, a, n, t, l):
-		self.title = a
-		self.text = n
-		self.time = t
-		self.length = l
+	count = 1
+	def __init__(self, name, args, caller, ret, start, end, u, proc, pid, color):
+		self.name = name
+		self.args = args
+		self.caller = caller
+		self.ret = ret
+		self.time = start
+		self.length = end - start
+		self.end = end
+		self.ubiquitous = u
+		self.proc = proc
+		self.pid = pid
+		self.color = color
+	def title(self):
+		cnt = ''
+		if self.count > 1:
+			cnt = '(x%d)' % self.count
+		l = '%0.3fms' % (self.length * 1000)
+		if self.ubiquitous:
+			title = '%s(%s)%s <- %s, %s(%s)' % \
+				(self.name, self.args, cnt, self.caller, self.ret, l)
+		else:
+			title = '%s(%s) %s%s(%s)' % (self.name, self.args, self.ret, cnt, l)
+		return title.replace('"', '')
+	def text(self):
+		if self.count > 1:
+			text = '%s(x%d)' % (self.name, self.count)
+		else:
+			text = self.name
+		return text
+	def repeat(self, tgt):
+		# is the tgt call just a repeat of this call (e.g. are we in a loop)
+		dt = self.time - tgt.end
+		# only combine calls if -all- attributes are identical
+		if tgt.caller == self.caller and \
+			tgt.name == self.name and tgt.args == self.args and \
+			tgt.proc == self.proc and tgt.pid == self.pid and \
+			tgt.ret == self.ret and dt >= 0 and \
+			dt <= sysvals.callloopmaxgap and \
+			self.length < sysvals.callloopmaxlen:
+			return True
+		return False
 
 # Class: FTraceLine
 # Description:
@@ -1226,7 +1444,6 @@ class FTraceLine:
 			print('%s -- %f (%02d): %s() { (%.3f us)' % (dev, self.time, \
 				self.depth, self.name, self.length*1000000))
 	def startMarker(self):
-		global sysvals
 		# Is this the starting line of a suspend?
 		if not self.fevent:
 			return False
@@ -1506,6 +1723,16 @@ class FTraceCallGraph:
 					l.depth, l.name, l.length*1000000))
 		print(' ')
 
+class DevItem:
+	def __init__(self, test, phase, dev):
+		self.test = test
+		self.phase = phase
+		self.dev = dev
+	def isa(self, cls):
+		if 'htmlclass' in self.dev and cls in self.dev['htmlclass']:
+			return True
+		return False
+
 # Class: Timeline
 # Description:
 #	 A container for a device timeline which calculates
@@ -1517,12 +1744,11 @@ class Timeline:
 	rowH = 30	# device row height
 	bodyH = 0	# body height
 	rows = 0	# total timeline rows
-	phases = []
-	rowmaxlines = dict()
-	rowcount = dict()
+	rowlines = dict()
 	rowheight = dict()
-	def __init__(self, rowheight):
+	def __init__(self, rowheight, scaleheight):
 		self.rowH = rowheight
+		self.scaleH = scaleheight
 		self.html = {
 			'header': '',
 			'timeline': '',
@@ -1537,21 +1763,19 @@ class Timeline:
 	#	 The total number of rows needed to display this phase of the timeline
 	def getDeviceRows(self, rawlist):
 		# clear all rows and set them to undefined
-		lendict = dict()
+		sortdict = dict()
 		for item in rawlist:
 			item.row = -1
-			lendict[item] = item.length
-		list = []
-		for i in sorted(lendict, key=lendict.get, reverse=True):
-			list.append(i)
-		remaining = len(list)
+			sortdict[item] = item.length
+		sortlist = sorted(sortdict, key=sortdict.get, reverse=True)
+		remaining = len(sortlist)
 		rowdata = dict()
 		row = 1
 		# try to pack each row with as many ranges as possible
 		while(remaining > 0):
 			if(row not in rowdata):
 				rowdata[row] = []
-			for i in list:
+			for i in sortlist:
 				if(i.row >= 0):
 					continue
 				s = i.time
@@ -1575,81 +1799,86 @@ class Timeline:
 	#	 Organize the timeline entries into the smallest
 	#	 number of rows possible, with no entry overlapping
 	# Arguments:
-	#	 list: the list of devices/actions for a single phase
-	#	 devlist: string list of device names to use
+	#	 devlist: the list of devices/actions in a group of contiguous phases
 	# Output:
 	#	 The total number of rows needed to display this phase of the timeline
-	def getPhaseRows(self, dmesg, devlist):
+	def getPhaseRows(self, devlist, row=0):
 		# clear all rows and set them to undefined
 		remaining = len(devlist)
 		rowdata = dict()
-		row = 0
-		lendict = dict()
+		sortdict = dict()
 		myphases = []
+		# initialize all device rows to -1 and calculate devrows
 		for item in devlist:
-			if item[0] not in self.phases:
-				self.phases.append(item[0])
-			if item[0] not in myphases:
-				myphases.append(item[0])
-				self.rowmaxlines[item[0]] = dict()
-				self.rowheight[item[0]] = dict()
-			dev = dmesg[item[0]]['list'][item[1]]
+			dev = item.dev
+			tp = (item.test, item.phase)
+			if tp not in myphases:
+				myphases.append(tp)
 			dev['row'] = -1
-			lendict[item] = float(dev['end']) - float(dev['start'])
+			# sort by length 1st, then name 2nd
+			sortdict[item] = (float(dev['end']) - float(dev['start']), item.dev['name'])
 			if 'src' in dev:
 				dev['devrows'] = self.getDeviceRows(dev['src'])
-		lenlist = []
-		for i in sorted(lendict, key=lendict.get, reverse=True):
-			lenlist.append(i)
+		# sort the devlist by length so that large items graph on top
+		sortlist = sorted(sortdict, key=sortdict.get, reverse=True)
 		orderedlist = []
-		for item in lenlist:
-			dev = dmesg[item[0]]['list'][item[1]]
-			if dev['pid'] == -2:
+		for item in sortlist:
+			if item.dev['pid'] == -2:
 				orderedlist.append(item)
-		for item in lenlist:
+		for item in sortlist:
 			if item not in orderedlist:
 				orderedlist.append(item)
-		# try to pack each row with as many ranges as possible
+		# try to pack each row with as many devices as possible
 		while(remaining > 0):
 			rowheight = 1
 			if(row not in rowdata):
 				rowdata[row] = []
 			for item in orderedlist:
-				dev = dmesg[item[0]]['list'][item[1]]
+				dev = item.dev
 				if(dev['row'] < 0):
 					s = dev['start']
 					e = dev['end']
 					valid = True
 					for ritem in rowdata[row]:
-						rs = ritem['start']
-						re = ritem['end']
+						rs = ritem.dev['start']
+						re = ritem.dev['end']
 						if(not (((s <= rs) and (e <= rs)) or
 							((s >= re) and (e >= re)))):
 							valid = False
 							break
 					if(valid):
-						rowdata[row].append(dev)
+						rowdata[row].append(item)
 						dev['row'] = row
 						remaining -= 1
 						if 'devrows' in dev and dev['devrows'] > rowheight:
 							rowheight = dev['devrows']
-			for phase in myphases:
-				self.rowmaxlines[phase][row] = rowheight
-				self.rowheight[phase][row] = rowheight * self.rowH
+			for t, p in myphases:
+				if t not in self.rowlines or t not in self.rowheight:
+					self.rowlines[t] = dict()
+					self.rowheight[t] = dict()
+				if p not in self.rowlines[t] or p not in self.rowheight[t]:
+					self.rowlines[t][p] = dict()
+					self.rowheight[t][p] = dict()
+				rh = self.rowH
+				# section headers should use a different row height
+				if len(rowdata[row]) == 1 and \
+					'htmlclass' in rowdata[row][0].dev and \
+					'sec' in rowdata[row][0].dev['htmlclass']:
+					rh = 15
+				self.rowlines[t][p][row] = rowheight
+				self.rowheight[t][p][row] = rowheight * rh
 			row += 1
 		if(row > self.rows):
 			self.rows = int(row)
-		for phase in myphases:
-			self.rowcount[phase] = row
 		return row
-	def phaseRowHeight(self, phase, row):
-		return self.rowheight[phase][row]
-	def phaseRowTop(self, phase, row):
+	def phaseRowHeight(self, test, phase, row):
+		return self.rowheight[test][phase][row]
+	def phaseRowTop(self, test, phase, row):
 		top = 0
-		for i in sorted(self.rowheight[phase]):
+		for i in sorted(self.rowheight[test][phase]):
 			if i >= row:
 				break
-			top += self.rowheight[phase][i]
+			top += self.rowheight[test][phase][i]
 		return top
 	# Function: calcTotalRows
 	# Description:
@@ -1657,19 +1886,21 @@ class Timeline:
 	def calcTotalRows(self):
 		maxrows = 0
 		standardphases = []
-		for phase in self.phases:
+		for t in self.rowlines:
+			for p in self.rowlines[t]:
 				total = 0
-			for i in sorted(self.rowmaxlines[phase]):
-				total += self.rowmaxlines[phase][i]
+				for i in sorted(self.rowlines[t][p]):
+					total += self.rowlines[t][p][i]
 				if total > maxrows:
 					maxrows = total
-			if total == self.rowcount[phase]:
-				standardphases.append(phase)
+				if total == len(self.rowlines[t][p]):
+					standardphases.append((t, p))
 		self.height = self.scaleH + (maxrows*self.rowH)
 		self.bodyH = self.height - self.scaleH
-		for phase in standardphases:
-			for i in sorted(self.rowheight[phase]):
-				self.rowheight[phase][i] = self.bodyH/self.rowcount[phase]
+		# if there is 1 line per row, draw them the standard way
+		for t, p in standardphases:
+			for i in sorted(self.rowheight[t][p]):
+				self.rowheight[t][p][i] = self.bodyH/len(self.rowlines[t][p])
 	# Function: createTimeScale
 	# Description:
 	#	 Create the timescale for a timeline block
@@ -1716,7 +1947,6 @@ class Timeline:
 #	 A list of values describing the properties of these test runs
 class TestProps:
 	stamp = ''
-	tracertype = ''
 	S0i3 = False
 	fwdata = []
 	ftrace_line_fmt_fg = \
@@ -1734,14 +1964,13 @@ class TestProps:
 	def __init__(self):
 		self.ktemp = dict()
 	def setTracerType(self, tracer):
-		self.tracertype = tracer
 		if(tracer == 'function_graph'):
 			self.cgformat = True
 			self.ftrace_line_fmt = self.ftrace_line_fmt_fg
 		elif(tracer == 'nop'):
 			self.ftrace_line_fmt = self.ftrace_line_fmt_nop
 		else:
-			doError('Invalid tracer format: [%s]' % tracer, False)
+			doError('Invalid tracer format: [%s]' % tracer)
 
 # Class: TestRun
 # Description:
@@ -1756,6 +1985,51 @@ class TestRun:
 		self.ftemp = dict()
 		self.ttemp = dict()
 
+class ProcessMonitor:
+	proclist = dict()
+	running = False
+	def procstat(self):
+		c = ['cat /proc/[1-9]*/stat 2>/dev/null']
+		process = Popen(c, shell=True, stdout=PIPE)
+		running = dict()
+		for line in process.stdout:
+			data = line.split()
+			pid = data[0]
+			name = re.sub('[()]', '', data[1])
+			user = int(data[13])
+			kern = int(data[14])
+			kjiff = ujiff = 0
+			if pid not in self.proclist:
+				self.proclist[pid] = {'name' : name, 'user' : user, 'kern' : kern}
+			else:
+				val = self.proclist[pid]
+				ujiff = user - val['user']
+				kjiff = kern - val['kern']
+				val['user'] = user
+				val['kern'] = kern
+			if ujiff > 0 or kjiff > 0:
+				running[pid] = ujiff + kjiff
+		result = process.wait()
+		out = ''
+		for pid in running:
+			jiffies = running[pid]
+			val = self.proclist[pid]
+			if out:
+				out += ','
+			out += '%s-%s %d' % (val['name'], pid, jiffies)
+		return 'ps - '+out
+	def processMonitor(self, tid):
+		while self.running:
+			out = self.procstat()
+			if out:
+				sysvals.fsetVal(out, 'trace_marker')
+	def start(self):
+		self.thread = Thread(target=self.processMonitor, args=(0,))
+		self.running = True
+		self.thread.start()
+	def stop(self):
+		self.running = False
+
 # ----------------- FUNCTIONS --------------------
 
 # Function: vprint
@@ -1764,7 +2038,7 @@ class TestRun:
 # Arguments:
 #	 msg: the debug/log message to print
 def vprint(msg):
-	global sysvals
+	sysvals.logmsg += msg+'\n'
 	if(sysvals.verbose):
 		print(msg)
 
@@ -1775,8 +2049,6 @@ def vprint(msg):
 # Arguments:
 #	 m: the valid re.match output for the stamp line
 def parseStamp(line, data):
-	global sysvals
-
 	m = re.match(sysvals.stampfmt, line)
 	data.stamp = {'time': '', 'host': '', 'mode': ''}
 	dt = datetime(int(m.group('y'))+2000, int(m.group('m')),
@@ -1788,6 +2060,14 @@ def parseStamp(line, data):
 	data.stamp['kernel'] = m.group('kernel')
 	sysvals.hostname = data.stamp['host']
 	sysvals.suspendmode = data.stamp['mode']
+	if sysvals.suspendmode == 'command' and sysvals.ftracefile != '':
+		modes = ['on', 'freeze', 'standby', 'mem']
+		out = Popen(['grep', 'suspend_enter', sysvals.ftracefile],
+			stderr=PIPE, stdout=PIPE).stdout.read()
+		m = re.match('.* suspend_enter\[(?P<mode>.*)\]', out)
+		if m and m.group('mode') in ['1', '2', '3']:
+			sysvals.suspendmode = modes[int(m.group('mode'))]
+			data.stamp['mode'] = sysvals.suspendmode
 	if not sysvals.stamp:
 		sysvals.stamp = data.stamp
 
@@ -1817,18 +2097,17 @@ def diffStamp(stamp1, stamp2):
 #	 required for primary parsing. Set the usetraceevents and/or
 #	 usetraceeventsonly flags in the global sysvals object
 def doesTraceLogHaveTraceEvents():
-	global sysvals
-
 	# check for kprobes
 	sysvals.usekprobes = False
-	out = os.system('grep -q "_cal: (" '+sysvals.ftracefile)
+	out = call('grep -q "_cal: (" '+sysvals.ftracefile, shell=True)
 	if(out == 0):
 		sysvals.usekprobes = True
 	# check for callgraph data on trace event blocks
-	out = os.system('grep -q "_cpu_down()" '+sysvals.ftracefile)
+	out = call('grep -q "_cpu_down()" '+sysvals.ftracefile, shell=True)
 	if(out == 0):
 		sysvals.usekprobes = True
-	out = os.popen('head -1 '+sysvals.ftracefile).read().replace('\n', '')
+	out = Popen(['head', '-1', sysvals.ftracefile],
+		stderr=PIPE, stdout=PIPE).stdout.read().replace('\n', '')
 	m = re.match(sysvals.stampfmt, out)
 	if m and m.group('mode') == 'command':
 		sysvals.usetraceeventsonly = True
@@ -1838,14 +2117,14 @@ def doesTraceLogHaveTraceEvents():
 	sysvals.usetraceeventsonly = True
 	sysvals.usetraceevents = False
 	for e in sysvals.traceevents:
-		out = os.system('grep -q "'+e+': " '+sysvals.ftracefile)
+		out = call('grep -q "'+e+': " '+sysvals.ftracefile, shell=True)
 		if(out != 0):
 			sysvals.usetraceeventsonly = False
 		if(e == 'suspend_resume' and out == 0):
 			sysvals.usetraceevents = True
 	# determine is this log is properly formatted
 	for e in ['SUSPEND START', 'RESUME COMPLETE']:
-		out = os.system('grep -q "'+e+'" '+sysvals.ftracefile)
+		out = call('grep -q "'+e+'" '+sysvals.ftracefile, shell=True)
 		if(out != 0):
 			sysvals.usetracemarkers = False
 
@@ -1860,8 +2139,6 @@ def doesTraceLogHaveTraceEvents():
 # Arguments:
 #	 testruns: the array of Data objects obtained from parseKernelLog
 def appendIncompleteTraceLog(testruns):
-	global sysvals
-
 	# create TestRun vessels for ftrace parsing
 	testcnt = len(testruns)
 	testidx = 0
@@ -2052,7 +2329,6 @@ def appendIncompleteTraceLog(testruns):
 								dev['ftrace'] = cg
 						break
 
-		if(sysvals.verbose):
 		test.data.printDetails()
 
 # Function: parseTraceLog
@@ -2064,14 +2340,12 @@ def appendIncompleteTraceLog(testruns):
 # Output:
 #	 An array of Data objects
 def parseTraceLog():
-	global sysvals
-
 	vprint('Analyzing the ftrace data...')
 	if(os.path.exists(sysvals.ftracefile) == False):
-		doError('%s does not exist' % sysvals.ftracefile, False)
+		doError('%s does not exist' % sysvals.ftracefile)
 
 	sysvals.setupAllKprobes()
-	tracewatch = ['suspend_enter']
+	tracewatch = []
 	if sysvals.usekprobes:
 		tracewatch += ['sync_filesystems', 'freeze_processes', 'syscore_suspend',
 			'syscore_resume', 'resume_console', 'thaw_processes', 'CPU_ON', 'CPU_OFF']
@@ -2102,17 +2376,13 @@ def parseTraceLog():
 		if(m):
 			tp.setTracerType(m.group('t'))
 			continue
-		# post resume time line: did this test run include post-resume data
-		m = re.match(sysvals.postresumefmt, line)
-		if(m):
-			t = int(m.group('t'))
-			if(t > 0):
-				sysvals.postresumetime = t
-			continue
 		# device properties line
 		if(re.match(sysvals.devpropfmt, line)):
 			devProps(line)
 			continue
+		# ignore all other commented lines
+		if line[0] == '#':
+			continue
 		# ftrace line: parse only valid lines
 		m = re.match(tp.ftrace_line_fmt, line)
 		if(not m):
@@ -2142,20 +2412,36 @@ def parseTraceLog():
 			testrun = TestRun(data)
 			testruns.append(testrun)
 			parseStamp(tp.stamp, data)
-			if len(tp.fwdata) > data.testnumber:
-				data.fwSuspend, data.fwResume = tp.fwdata[data.testnumber]
-				if(data.fwSuspend > 0 or data.fwResume > 0):
-					data.fwValid = True
 			data.setStart(t.time)
+			data.tKernSus = t.time
 			continue
 		if(not data):
 			continue
+		# process cpu exec line
+		if t.type == 'tracing_mark_write':
+			m = re.match(sysvals.procexecfmt, t.name)
+			if(m):
+				proclist = dict()
+				for ps in m.group('ps').split(','):
+					val = ps.split()
+					if not val:
+						continue
+					name = val[0].replace('--', '-')
+					proclist[name] = int(val[1])
+				data.pstl[t.time] = proclist
+				continue
 		# find the end of resume
 		if(t.endMarker()):
-			if(sysvals.usetracemarkers and sysvals.postresumetime > 0):
-				phase = 'post_resume'
-				data.newPhase(phase, t.time, t.time, '#F0F0F0', -1)
 			data.setEnd(t.time)
+			if data.tKernRes == 0.0:
+				data.tKernRes = t.time
+			if data.dmesg['resume_complete']['end'] < 0:
+				data.dmesg['resume_complete']['end'] = t.time
+			if sysvals.suspendmode == 'mem' and len(tp.fwdata) > data.testnumber:
+				data.fwSuspend, data.fwResume = tp.fwdata[data.testnumber]
+				if(data.tSuspended != 0 and data.tResumed != 0 and \
+					(data.fwSuspend > 0 or data.fwResume > 0)):
+					data.fwValid = True
 			if(not sysvals.usetracemarkers):
 				# no trace markers? then quit and be sure to finish recording
 				# the event we used to trigger resume end
@@ -2190,8 +2476,14 @@ def parseTraceLog():
 				if(name.split('[')[0] in tracewatch):
 					continue
 				# -- phase changes --
+				# start of kernel suspend
+				if(re.match('suspend_enter\[.*', t.name)):
+					if(isbegin):
+						data.dmesg[phase]['start'] = t.time
+						data.tKernSus = t.time
+					continue
 				# suspend_prepare start
-				if(re.match('dpm_prepare\[.*', t.name)):
+				elif(re.match('dpm_prepare\[.*', t.name)):
 					phase = 'suspend_prepare'
 					if(not isbegin):
 						data.dmesg[phase]['end'] = t.time
@@ -2291,6 +2583,8 @@ def parseTraceLog():
 				p = m.group('p')
 				if(n and p):
 					data.newAction(phase, n, pid, p, t.time, -1, drv)
+					if pid not in data.devpids:
+						data.devpids.append(pid)
 			# device callback finish
 			elif(t.type == 'device_pm_callback_end'):
 				m = re.match('(?P<drv>.*) (?P<d>.*), err.*', t.name);
@@ -2332,6 +2626,12 @@ def parseTraceLog():
 				else:
 					e['end'] = t.time
 					e['rdata'] = kprobedata
+				# end of kernel resume
+				if(kprobename == 'pm_notifier_call_chain' or \
+					kprobename == 'pm_restore_console'):
+					data.dmesg[phase]['end'] = t.time
+					data.tKernRes = t.time
+
 		# callgraph processing
 		elif sysvals.usecallgraph:
 			# create a callgraph object for the data
@@ -2348,24 +2648,37 @@ def parseTraceLog():
 	if sysvals.suspendmode == 'command':
 		for test in testruns:
 			for p in test.data.phases:
-				if p == 'resume_complete':
+				if p == 'suspend_prepare':
 					test.data.dmesg[p]['start'] = test.data.start
 					test.data.dmesg[p]['end'] = test.data.end
 				else:
-					test.data.dmesg[p]['start'] = test.data.start
-					test.data.dmesg[p]['end'] = test.data.start
-			test.data.tSuspended = test.data.start
-			test.data.tResumed = test.data.start
+					test.data.dmesg[p]['start'] = test.data.end
+					test.data.dmesg[p]['end'] = test.data.end
+			test.data.tSuspended = test.data.end
+			test.data.tResumed = test.data.end
 			test.data.tLow = 0
 			test.data.fwValid = False
 
-	for test in testruns:
+	# dev source and procmon events can be unreadable with mixed phase height
+	if sysvals.usedevsrc or sysvals.useprocmon:
+		sysvals.mixedphaseheight = False
+
+	for i in range(len(testruns)):
+		test = testruns[i]
+		data = test.data
+		# find the total time range for this test (begin, end)
+		tlb, tle = data.start, data.end
+		if i < len(testruns) - 1:
+			tle = testruns[i+1].data.start
+		# add the process usage data to the timeline
+		if sysvals.useprocmon:
+			data.createProcessUsageEvents()
 		# add the traceevent data to the device hierarchy
 		if(sysvals.usetraceevents):
 			# add actual trace funcs
 			for name in test.ttemp:
 				for event in test.ttemp[name]:
-					test.data.newActionGlobal(name, event['begin'], event['end'], event['pid'])
+					data.newActionGlobal(name, event['begin'], event['end'], event['pid'])
 			# add the kprobe based virtual tracefuncs as actual devices
 			for key in tp.ktemp:
 				name, pid = key
@@ -2373,24 +2686,20 @@ def parseTraceLog():
 					continue
 				for e in tp.ktemp[key]:
 					kb, ke = e['begin'], e['end']
-					if kb == ke or not test.data.isInsideTimeline(kb, ke):
+					if kb == ke or tlb > kb or tle <= kb:
 						continue
-					test.data.newActionGlobal(e['name'], kb, ke, pid)
+					color = sysvals.kprobeColor(name)
+					data.newActionGlobal(e['name'], kb, ke, pid, color)
 			# add config base kprobes and dev kprobes
+			if sysvals.usedevsrc:
 				for key in tp.ktemp:
 					name, pid = key
-				if name in sysvals.tracefuncs:
+					if name in sysvals.tracefuncs or name not in sysvals.dev_tracefuncs:
 						continue
 					for e in tp.ktemp[key]:
 						kb, ke = e['begin'], e['end']
-					if kb == ke or not test.data.isInsideTimeline(kb, ke):
+						if kb == ke or tlb > kb or tle <= kb:
 							continue
-					color = sysvals.kprobeColor(e['name'])
-					if name not in sysvals.dev_tracefuncs:
-						# config base kprobe
-						test.data.newActionGlobal(e['name'], kb, ke, -2, color)
-					elif sysvals.usedevsrc:
-						# dev kprobe
 						data.addDeviceFunctionCall(e['name'], name, e['proc'], pid, kb,
 							ke, e['cdata'], e['rdata'])
 		if sysvals.usecallgraph:
@@ -2407,7 +2716,7 @@ def parseTraceLog():
 							id+', ignoring this callback')
 						continue
 					# match cg data to devices
-					if sysvals.suspendmode == 'command' or not cg.deviceMatch(pid, test.data):
+					if sysvals.suspendmode == 'command' or not cg.deviceMatch(pid, data):
 						sortkey = '%f%f%d' % (cg.start, cg.end, pid)
 						sortlist[sortkey] = cg
 			# create blocks for orphan cg data
@@ -2416,10 +2725,9 @@ def parseTraceLog():
 				name = cg.list[0].name
 				if sysvals.isCallgraphFunc(name):
 					vprint('Callgraph found for task %d: %.3fms, %s' % (cg.pid, (cg.end - cg.start)*1000, name))
-					cg.newActionFromFunction(test.data)
+					cg.newActionFromFunction(data)
 
 	if sysvals.suspendmode == 'command':
-		if(sysvals.verbose):
 		for data in testdata:
 			data.printDetails()
 		return testdata
@@ -2429,7 +2737,7 @@ def parseTraceLog():
 		lp = data.phases[0]
 		for p in data.phases:
 			if(data.dmesg[p]['start'] < 0 and data.dmesg[p]['end'] < 0):
-				print('WARNING: phase "%s" is missing!' % p)
+				vprint('WARNING: phase "%s" is missing!' % p)
 			if(data.dmesg[p]['start'] < 0):
 				data.dmesg[p]['start'] = data.dmesg[lp]['end']
 				if(p == 'resume_machine'):
@@ -2438,60 +2746,27 @@ def parseTraceLog():
 					data.tLow = 0
 			if(data.dmesg[p]['end'] < 0):
 				data.dmesg[p]['end'] = data.dmesg[p]['start']
+			if(p != lp and not ('machine' in p and 'machine' in lp)):
+				data.dmesg[lp]['end'] = data.dmesg[p]['start']
 			lp = p
 
 		if(len(sysvals.devicefilter) > 0):
 			data.deviceFilter(sysvals.devicefilter)
 		data.fixupInitcallsThatDidntReturn()
-		if(sysvals.verbose):
+		if sysvals.usedevsrc:
+			data.optimizeDevSrc()
 		data.printDetails()
 
+	# x2: merge any overlapping devices between test runs
+	if sysvals.usedevsrc and len(testdata) > 1:
+		tc = len(testdata)
+		for i in range(tc - 1):
+			devlist = testdata[i].overflowDevices()
+			for j in range(i + 1, tc):
+				testdata[j].mergeOverlapDevices(devlist)
+		testdata[0].stitchTouchingThreads(testdata[1:])
 	return testdata
 
-# Function: loadRawKernelLog
-# Description:
-#	 Load a raw kernel log that wasn't created by this tool, it might be
-#	 possible to extract a valid suspend/resume log
-def loadRawKernelLog(dmesgfile):
-	global sysvals
-
-	stamp = {'time': '', 'host': '', 'mode': 'mem', 'kernel': ''}
-	stamp['time'] = datetime.now().strftime('%B %d %Y, %I:%M:%S %p')
-	stamp['host'] = sysvals.hostname
-
-	testruns = []
-	data = 0
-	lf = open(dmesgfile, 'r')
-	for line in lf:
-		line = line.replace('\r\n', '')
-		idx = line.find('[')
-		if idx > 1:
-			line = line[idx:]
-		m = re.match('[ \t]*(\[ *)(?P<ktime>[0-9\.]*)(\]) (?P<msg>.*)', line)
-		if(not m):
-			continue
-		msg = m.group("msg")
-		m = re.match('PM: Syncing filesystems.*', msg)
-		if(m):
-			if(data):
-				testruns.append(data)
-			data = Data(len(testruns))
-			data.stamp = stamp
-		if(data):
-			m = re.match('.* *(?P<k>[0-9]\.[0-9]{2}\.[0-9]-.*) .*', msg)
-			if(m):
-				stamp['kernel'] = m.group('k')
-			m = re.match('PM: Preparing system for (?P<m>.*) sleep', msg)
-			if(m):
-				stamp['mode'] = m.group('m')
-			data.dmesgtext.append(line)
-	if(data):
-		testruns.append(data)
-		sysvals.stamp = stamp
-		sysvals.suspendmode = stamp['mode']
-	lf.close()
-	return testruns
-
 # Function: loadKernelLog
 # Description:
 #	 [deprecated for kernel 3.15.0 or newer]
@@ -2499,15 +2774,16 @@ def loadRawKernelLog(dmesgfile):
 #	 The dmesg filename is taken from sysvals
 # Output:
 #	 An array of empty Data objects with only their dmesgtext attributes set
-def loadKernelLog():
-	global sysvals
-
+def loadKernelLog(justtext=False):
 	vprint('Analyzing the dmesg data...')
 	if(os.path.exists(sysvals.dmesgfile) == False):
-		doError('%s does not exist' % sysvals.dmesgfile, False)
+		doError('%s does not exist' % sysvals.dmesgfile)
 
+	if justtext:
+		dmesgtext = []
 	# there can be multiple test runs in a single file
 	tp = TestProps()
+	tp.stamp = datetime.now().strftime('# suspend-%m%d%y-%H%M%S localhost mem unknown')
 	testruns = []
 	data = 0
 	lf = open(sysvals.dmesgfile, 'r')
@@ -2528,6 +2804,9 @@ def loadKernelLog():
 		if(not m):
 			continue
 		msg = m.group("msg")
+		if justtext:
+			dmesgtext.append(line)
+			continue
 		if(re.match('PM: Syncing filesystems.*', msg)):
 			if(data):
 				testruns.append(data)
@@ -2537,24 +2816,24 @@ def loadKernelLog():
 				data.fwSuspend, data.fwResume = tp.fwdata[data.testnumber]
 				if(data.fwSuspend > 0 or data.fwResume > 0):
 					data.fwValid = True
-		if(re.match('ACPI: resume from mwait', msg)):
-			print('NOTE: This suspend appears to be freeze rather than'+\
-				' %s, it will be treated as such' % sysvals.suspendmode)
-			sysvals.suspendmode = 'freeze'
 		if(not data):
 			continue
+		m = re.match('.* *(?P<k>[0-9]\.[0-9]{2}\.[0-9]-.*) .*', msg)
+		if(m):
+			sysvals.stamp['kernel'] = m.group('k')
+		m = re.match('PM: Preparing system for (?P<m>.*) sleep', msg)
+		if(m):
+			sysvals.stamp['mode'] = sysvals.suspendmode = m.group('m')
 		data.dmesgtext.append(line)
-	if(data):
-		testruns.append(data)
 	lf.close()
 
-	if(len(testruns) < 1):
-		# bad log, but see if you can extract something meaningful anyway
-		testruns = loadRawKernelLog(sysvals.dmesgfile)
-
-	if(len(testruns) < 1):
-		doError(' dmesg log is completely unreadable: %s' \
-			% sysvals.dmesgfile, False)
+	if justtext:
+		return dmesgtext
+	if data:
+		testruns.append(data)
+	if len(testruns) < 1:
+		doError(' dmesg log has no suspend/resume data: %s' \
+			% sysvals.dmesgfile)
 
 	# fix lines with same timestamp/function with the call and return swapped
 	for data in testruns:
@@ -2586,8 +2865,6 @@ def loadKernelLog():
 # Output:
 #	 The filled Data object
 def parseKernelLog(data):
-	global sysvals
-
 	phase = 'suspend_runtime'
 
 	if(data.fwValid):
@@ -2645,7 +2922,6 @@ def parseKernelLog(data):
 	prevktime = -1.0
 	actions = dict()
 	for line in data.dmesgtext:
-		# -- preprocessing --
 		# parse each dmesg line into the time and message
 		m = re.match('[ \t]*(\[ *)(?P<ktime>[0-9\.]*)(\]) (?P<msg>.*)', line)
 		if(m):
@@ -2653,8 +2929,6 @@ def parseKernelLog(data):
 			try:
 				ktime = float(val)
 			except:
-				doWarning('INVALID DMESG LINE: '+\
-					line.replace('\n', ''), 'dmesg')
 				continue
 			msg = m.group('msg')
 			# initialize data start to first line time
@@ -2672,12 +2946,12 @@ def parseKernelLog(data):
 			phase = 'resume_noirq'
 			data.dmesg[phase]['start'] = ktime
 
-		# -- phase changes --
 		# suspend start
 		if(re.match(dm['suspend_prepare'], msg)):
 			phase = 'suspend_prepare'
 			data.dmesg[phase]['start'] = ktime
 			data.setStart(ktime)
+			data.tKernSus = ktime
 		# suspend start
 		elif(re.match(dm['suspend'], msg)):
 			data.dmesg['suspend_prepare']['end'] = ktime
@@ -2734,7 +3008,7 @@ def parseKernelLog(data):
 		elif(re.match(dm['post_resume'], msg)):
 			data.dmesg['resume_complete']['end'] = ktime
 			data.setEnd(ktime)
-			phase = 'post_resume'
+			data.tKernRes = ktime
 			break
 
 		# -- device callbacks --
@@ -2761,7 +3035,6 @@ def parseKernelLog(data):
 					dev['length'] = int(t)
 					dev['end'] = ktime
 
-		# -- non-devicecallback actions --
 		# if trace events are not available, these are better than nothing
 		if(not sysvals.usetraceevents):
 			# look for known actions
@@ -2821,7 +3094,6 @@ def parseKernelLog(data):
 		for event in actions[name]:
 			data.newActionGlobal(name, event['begin'], event['end'])
 
-	if(sysvals.verbose):
 	data.printDetails()
 	if(len(sysvals.devicefilter) > 0):
 		data.deviceFilter(sysvals.devicefilter)
@@ -2834,8 +3106,6 @@ def parseKernelLog(data):
 # Arguments:
 #	 testruns: array of Data objects from parseTraceLog
 def createHTMLSummarySimple(testruns, htmlfile):
-	global sysvals
-
 	# print out the basic summary of all the tests
 	hf = open(htmlfile, 'w')
 
@@ -2960,7 +3230,6 @@ def createHTMLSummarySimple(testruns, htmlfile):
 	hf.close()
 
 def htmlTitle():
-	global sysvals
 	modename = {
 		'freeze': 'Freeze (S0)',
 		'standby': 'Standby (S1)',
@@ -2993,13 +3262,14 @@ def ordinal(value):
 # Output:
 #	 True if the html file was created, false if it failed
 def createHTML(testruns):
-	global sysvals
-
 	if len(testruns) < 1:
 		print('ERROR: Not enough test data to build a timeline')
 		return
 
+	kerror = False
 	for data in testruns:
+		if data.kerror:
+			kerror = True
 		data.normalizeTime(testruns[-1].tSuspended)
 
 	x2changes = ['', 'absolute']
@@ -3009,53 +3279,59 @@ def createHTML(testruns):
 	headline_version = '<div class="version"><a href="https://01.org/suspendresume">AnalyzeSuspend v%s</a></div>' % sysvals.version
 	headline_stamp = '<div class="stamp">{0} {1} {2} {3}</div>\n'
 	html_devlist1 = '<button id="devlist1" class="devlist" style="float:left;">Device Detail%s</button>' % x2changes[0]
-	html_zoombox = '<center><button id="zoomin">ZOOM IN</button><button id="zoomout">ZOOM OUT</button><button id="zoomdef">ZOOM 1:1</button></center>\n'
+	html_zoombox = '<center><button id="zoomin">ZOOM IN +</button><button id="zoomout">ZOOM OUT -</button><button id="zoomdef">ZOOM 1:1</button></center>\n'
 	html_devlist2 = '<button id="devlist2" class="devlist" style="float:right;">Device Detail2</button>\n'
 	html_timeline = '<div id="dmesgzoombox" class="zoombox">\n<div id="{0}" class="timeline" style="height:{1}px">\n'
-	html_tblock = '<div id="block{0}" class="tblock" style="left:{1}%;width:{2}%;">\n'
+	html_tblock = '<div id="block{0}" class="tblock" style="left:{1}%;width:{2}%;"><div class="tback" style="height:{3}px"></div>\n'
 	html_device = '<div id="{0}" title="{1}" class="thread{7}" style="left:{2}%;top:{3}px;height:{4}px;width:{5}%;{8}">{6}</div>\n'
-	html_traceevent = '<div title="{0}" class="traceevent" style="left:{1}%;top:{2}px;height:{3}px;width:{4}%;line-height:{3}px;">{5}</div>\n'
+	html_error = '<div id="{1}" title="kernel error/warning" class="err" style="right:{0}%">ERROR&rarr;</div>\n'
+	html_traceevent = '<div title="{0}" class="traceevent{6}" style="left:{1}%;top:{2}px;height:{3}px;width:{4}%;line-height:{3}px;{7}">{5}</div>\n'
+	html_cpuexec = '<div class="jiffie" style="left:{0}%;top:{1}px;height:{2}px;width:{3}%;background:{4};"></div>\n'
 	html_phase = '<div class="phase" style="left:{0}%;width:{1}%;top:{2}px;height:{3}px;background-color:{4}">{5}</div>\n'
-	html_phaselet = '<div id="{0}" class="phaselet" style="left:{1}%;width:{2}%;background-color:{3}"></div>\n'
+	html_phaselet = '<div id="{0}" class="phaselet" style="left:{1}%;width:{2}%;background:{3}"></div>\n'
 	html_legend = '<div id="p{3}" class="square" style="left:{0}%;background-color:{1}">&nbsp;{2}</div>\n'
 	html_timetotal = '<table class="time1">\n<tr>'\
-		'<td class="green">{2} Suspend Time: <b>{0} ms</b></td>'\
-		'<td class="yellow">{2} Resume Time: <b>{1} ms</b></td>'\
+		'<td class="green" title="{3}">{2} Suspend Time: <b>{0} ms</b></td>'\
+		'<td class="yellow" title="{4}">{2} Resume Time: <b>{1} ms</b></td>'\
 		'</tr>\n</table>\n'
 	html_timetotal2 = '<table class="time1">\n<tr>'\
-		'<td class="green">{3} Suspend Time: <b>{0} ms</b></td>'\
-		'<td class="gray">'+sysvals.suspendmode+' time: <b>{1} ms</b></td>'\
-		'<td class="yellow">{3} Resume Time: <b>{2} ms</b></td>'\
+		'<td class="green" title="{4}">{3} Suspend Time: <b>{0} ms</b></td>'\
+		'<td class="gray" title="time spent in low-power mode with clock running">'+sysvals.suspendmode+' time: <b>{1} ms</b></td>'\
+		'<td class="yellow" title="{5}">{3} Resume Time: <b>{2} ms</b></td>'\
 		'</tr>\n</table>\n'
 	html_timetotal3 = '<table class="time1">\n<tr>'\
 		'<td class="green">Execution Time: <b>{0} ms</b></td>'\
 		'<td class="yellow">Command: <b>{1}</b></td>'\
 		'</tr>\n</table>\n'
 	html_timegroups = '<table class="time2">\n<tr>'\
-		'<td class="green">{4}Kernel Suspend: {0} ms</td>'\
+		'<td class="green" title="time from kernel enter_state({5}) to firmware mode [kernel time only]">{4}Kernel Suspend: {0} ms</td>'\
 		'<td class="purple">{4}Firmware Suspend: {1} ms</td>'\
 		'<td class="purple">{4}Firmware Resume: {2} ms</td>'\
-		'<td class="yellow">{4}Kernel Resume: {3} ms</td>'\
+		'<td class="yellow" title="time from firmware mode to return from kernel enter_state({5}) [kernel time only]">{4}Kernel Resume: {3} ms</td>'\
 		'</tr>\n</table>\n'
 
 	# html format variables
-	rowheight = 30
-	devtextS = '14px'
-	devtextH = '30px'
-	hoverZ = 'z-index:10;'
-
+	hoverZ = 'z-index:8;'
 	if sysvals.usedevsrc:
 		hoverZ = ''
+	scaleH = 20
+	scaleTH = 20
+	if kerror:
+		scaleH = 40
+		scaleTH = 60
 
 	# device timeline
 	vprint('Creating Device Timeline...')
 
-	devtl = Timeline(rowheight)
+	devtl = Timeline(30, scaleH)
 
 	# Generate the header for this timeline
 	for data in testruns:
 		tTotal = data.end - data.start
-		tEnd = data.dmesg['resume_complete']['end']
+		sktime = (data.dmesg['suspend_machine']['end'] - \
+			data.tKernSus) * 1000
+		rktime = (data.dmesg['resume_complete']['end'] - \
+			data.dmesg['resume_machine']['start']) * 1000
 		if(tTotal == 0):
 			print('ERROR: No timeline data')
 			sys.exit()
@@ -3072,59 +3348,85 @@ def createHTML(testruns):
 			thtml = html_timetotal3.format(run_time, testdesc)
 			devtl.html['header'] += thtml
 		elif data.fwValid:
-			suspend_time = '%.0f'%((data.tSuspended-data.start)*1000 + \
-				(data.fwSuspend/1000000.0))
-			resume_time = '%.0f'%((tEnd-data.tSuspended)*1000 + \
-				(data.fwResume/1000000.0))
+			suspend_time = '%.0f'%(sktime + (data.fwSuspend/1000000.0))
+			resume_time = '%.0f'%(rktime + (data.fwResume/1000000.0))
 			testdesc1 = 'Total'
 			testdesc2 = ''
+			stitle = 'time from kernel enter_state(%s) to low-power mode [kernel & firmware time]' % sysvals.suspendmode
+			rtitle = 'time from low-power mode to return from kernel enter_state(%s) [firmware & kernel time]' % sysvals.suspendmode
 			if(len(testruns) > 1):
 				testdesc1 = testdesc2 = ordinal(data.testnumber+1)
 				testdesc2 += ' '
 			if(data.tLow == 0):
 				thtml = html_timetotal.format(suspend_time, \
-					resume_time, testdesc1)
+					resume_time, testdesc1, stitle, rtitle)
 			else:
 				thtml = html_timetotal2.format(suspend_time, low_time, \
-					resume_time, testdesc1)
+					resume_time, testdesc1, stitle, rtitle)
 			devtl.html['header'] += thtml
-			sktime = '%.3f'%((data.dmesg['suspend_machine']['end'] - \
-				data.getStart())*1000)
 			sftime = '%.3f'%(data.fwSuspend / 1000000.0)
 			rftime = '%.3f'%(data.fwResume / 1000000.0)
-			rktime = '%.3f'%((data.dmesg['resume_complete']['end'] - \
-				data.dmesg['resume_machine']['start'])*1000)
-			devtl.html['header'] += html_timegroups.format(sktime, \
-				sftime, rftime, rktime, testdesc2)
+			devtl.html['header'] += html_timegroups.format('%.3f'%sktime, \
+				sftime, rftime, '%.3f'%rktime, testdesc2, sysvals.suspendmode)
 		else:
-			suspend_time = '%.0f'%((data.tSuspended-data.start)*1000)
-			resume_time = '%.0f'%((tEnd-data.tSuspended)*1000)
+			suspend_time = '%.3f' % sktime
+			resume_time = '%.3f' % rktime
 			testdesc = 'Kernel'
+			stitle = 'time from kernel enter_state(%s) to firmware mode [kernel time only]' % sysvals.suspendmode
+			rtitle = 'time from firmware mode to return from kernel enter_state(%s) [kernel time only]' % sysvals.suspendmode
 			if(len(testruns) > 1):
 				testdesc = ordinal(data.testnumber+1)+' '+testdesc
 			if(data.tLow == 0):
 				thtml = html_timetotal.format(suspend_time, \
-					resume_time, testdesc)
+					resume_time, testdesc, stitle, rtitle)
 			else:
 				thtml = html_timetotal2.format(suspend_time, low_time, \
-					resume_time, testdesc)
+					resume_time, testdesc, stitle, rtitle)
 			devtl.html['header'] += thtml
 
 	# time scale for potentially multiple datasets
 	t0 = testruns[0].start
 	tMax = testruns[-1].end
-	tSuspended = testruns[-1].tSuspended
 	tTotal = tMax - t0
 
 	# determine the maximum number of rows we need to draw
+	fulllist = []
+	threadlist = []
+	pscnt = 0
+	devcnt = 0
 	for data in testruns:
 		data.selectTimelineDevices('%f', tTotal, sysvals.mindevlen)
 		for group in data.devicegroups:
 			devlist = []
 			for phase in group:
 				for devname in data.tdevlist[phase]:
-					devlist.append((phase,devname))
-			devtl.getPhaseRows(data.dmesg, devlist)
+					d = DevItem(data.testnumber, phase, data.dmesg[phase]['list'][devname])
+					devlist.append(d)
+					if d.isa('kth'):
+						threadlist.append(d)
+					else:
+						if d.isa('ps'):
+							pscnt += 1
+						else:
+							devcnt += 1
+						fulllist.append(d)
+			if sysvals.mixedphaseheight:
+				devtl.getPhaseRows(devlist)
+	if not sysvals.mixedphaseheight:
+		if len(threadlist) > 0 and len(fulllist) > 0:
+			if pscnt > 0 and devcnt > 0:
+				msg = 'user processes & device pm callbacks'
+			elif pscnt > 0:
+				msg = 'user processes'
+			else:
+				msg = 'device pm callbacks'
+			d = testruns[0].addHorizontalDivider(msg, testruns[-1].end)
+			fulllist.insert(0, d)
+		devtl.getPhaseRows(fulllist)
+		if len(threadlist) > 0:
+			d = testruns[0].addHorizontalDivider('asynchronous kernel threads', testruns[-1].end)
+			threadlist.insert(0, d)
+			devtl.getPhaseRows(threadlist, devtl.rows)
 	devtl.calcTotalRows()
 
 	# create bounding box, add buttons
@@ -3145,18 +3447,6 @@ def createHTML(testruns):
 
 	# draw each test run chronologically
 	for data in testruns:
-		# if nore than one test, draw a block to represent user mode
-		if(data.testnumber > 0):
-			m0 = testruns[data.testnumber-1].end
-			mMax = testruns[data.testnumber].start
-			mTotal = mMax - m0
-			name = 'usermode%d' % data.testnumber
-			top = '%d' % devtl.scaleH
-			left = '%f' % (((m0-t0)*100.0)/tTotal)
-			width = '%f' % ((mTotal*100.0)/tTotal)
-			title = 'user mode (%0.3f ms) ' % (mTotal*1000)
-			devtl.html['timeline'] += html_device.format(name, \
-				title, left, top, '%d'%devtl.bodyH, width, '', '', '')
 		# now draw the actual timeline blocks
 		for dir in phases:
 			# draw suspend and resume blocks separately
@@ -3169,13 +3459,16 @@ def createHTML(testruns):
 			else:
 				m0 = testruns[data.testnumber].tSuspended
 				mMax = testruns[data.testnumber].end
+				# in an x2 run, remove any gap between blocks
+				if len(testruns) > 1 and data.testnumber == 0:
+					mMax = testruns[1].start
 				mTotal = mMax - m0
 				left = '%f' % ((((m0-t0)*100.0)+sysvals.srgap/2)/tTotal)
 			# if a timeline block is 0 length, skip altogether
 			if mTotal == 0:
 				continue
 			width = '%f' % (((mTotal*100.0)-sysvals.srgap/2)/tTotal)
-			devtl.html['timeline'] += html_tblock.format(bname, left, width)
+			devtl.html['timeline'] += html_tblock.format(bname, left, width, devtl.scaleH)
 			for b in sorted(phases[dir]):
 				# draw the phase color background
 				phase = data.dmesg[b]
@@ -3185,6 +3478,12 @@ def createHTML(testruns):
 				devtl.html['timeline'] += html_phase.format(left, width, \
 					'%.3f'%devtl.scaleH, '%.3f'%devtl.bodyH, \
 					data.dmesg[b]['color'], '')
+			for e in data.errorinfo[dir]:
+				# draw red lines for any kernel errors found
+				t, err = e
+				right = '%f' % (((mMax-t)*100.0)/mTotal)
+				devtl.html['timeline'] += html_error.format(right, err)
+			for b in sorted(phases[dir]):
 				# draw the devices for this phase
 				phaselist = data.dmesg[b]['list']
 				for d in data.tdevlist[b]:
@@ -3196,46 +3495,62 @@ def createHTML(testruns):
 					xtrastyle = ''
 					if 'htmlclass' in dev:
 						xtraclass = dev['htmlclass']
-						xtrainfo = dev['htmlclass']
 					if 'color' in dev:
 						xtrastyle = 'background-color:%s;' % dev['color']
 					if(d in sysvals.devprops):
 						name = sysvals.devprops[d].altName(d)
 						xtraclass = sysvals.devprops[d].xtraClass()
 						xtrainfo = sysvals.devprops[d].xtraInfo()
+					elif xtraclass == ' kth':
+						xtrainfo = ' kernel_thread'
 					if('drv' in dev and dev['drv']):
 						drv = ' {%s}' % dev['drv']
-					rowheight = devtl.phaseRowHeight(b, dev['row'])
-					rowtop = devtl.phaseRowTop(b, dev['row'])
+					rowheight = devtl.phaseRowHeight(data.testnumber, b, dev['row'])
+					rowtop = devtl.phaseRowTop(data.testnumber, b, dev['row'])
 					top = '%.3f' % (rowtop + devtl.scaleH)
 					left = '%f' % (((dev['start']-m0)*100)/mTotal)
 					width = '%f' % (((dev['end']-dev['start'])*100)/mTotal)
 					length = ' (%0.3f ms) ' % ((dev['end']-dev['start'])*1000)
+					title = name+drv+xtrainfo+length
 					if sysvals.suspendmode == 'command':
-						title = name+drv+xtrainfo+length+'cmdexec'
+						title += sysvals.testcommand
+					elif xtraclass == ' ps':
+						if 'suspend' in b:
+							title += 'pre_suspend_process'
+						else:
+							title += 'post_resume_process'
 					else:
-						title = name+drv+xtrainfo+length+b
+						title += b
 					devtl.html['timeline'] += html_device.format(dev['id'], \
 						title, left, top, '%.3f'%rowheight, width, \
 						d+drv, xtraclass, xtrastyle)
+					if('cpuexec' in dev):
+						for t in sorted(dev['cpuexec']):
+							start, end = t
+							j = float(dev['cpuexec'][t]) / 5
+							if j > 1.0:
+								j = 1.0
+							height = '%.3f' % (rowheight/3)
+							top = '%.3f' % (rowtop + devtl.scaleH + 2*rowheight/3)
+							left = '%f' % (((start-m0)*100)/mTotal)
+							width = '%f' % ((end-start)*100/mTotal)
+							color = 'rgba(255, 0, 0, %f)' % j
+							devtl.html['timeline'] += \
+								html_cpuexec.format(left, top, height, width, color)
 					if('src' not in dev):
 						continue
 					# draw any trace events for this device
-					vprint('Debug trace events found for device %s' % d)
-					vprint('%20s %20s %10s %8s' % ('title', \
-						'name', 'time(ms)', 'length(ms)'))
 					for e in dev['src']:
-						vprint('%20s %20s %10.3f %8.3f' % (e.title, \
-							e.text, e.time*1000, e.length*1000))
-						height = devtl.rowH
+						height = '%.3f' % devtl.rowH
 						top = '%.3f' % (rowtop + devtl.scaleH + (e.row*devtl.rowH))
 						left = '%f' % (((e.time-m0)*100)/mTotal)
 						width = '%f' % (e.length*100/mTotal)
-						color = 'rgba(204,204,204,0.5)'
+						xtrastyle = ''
+						if e.color:
+							xtrastyle = 'background:%s;' % e.color
 						devtl.html['timeline'] += \
-							html_traceevent.format(e.title, \
-								left, top, '%.3f'%height, \
-								width, e.text)
+							html_traceevent.format(e.title(), \
+								left, top, height, width, e.text(), '', xtrastyle)
 			# draw the time scale, try to make the number of labels readable
 			devtl.html['timeline'] += devtl.createTimeScale(m0, mMax, tTotal, dir)
 			devtl.html['timeline'] += '</div>\n'
@@ -3284,8 +3599,7 @@ def createHTML(testruns):
 		t2 {color:black;font:25px Times;}\n\
 		t3 {color:black;font:20px Times;white-space:nowrap;}\n\
 		t4 {color:black;font:bold 30px Times;line-height:60px;white-space:nowrap;}\n\
-		cS {color:blue;font:bold 11px Times;}\n\
-		cR {color:red;font:bold 11px Times;}\n\
+		cS {font:bold 13px Times;}\n\
 		table {width:100%;}\n\
 		.gray {background-color:rgba(80,80,80,0.1);}\n\
 		.green {background-color:rgba(204,255,204,0.4);}\n\
@@ -3302,20 +3616,22 @@ def createHTML(testruns):
 		.pf:'+cgchk+' + label {background:url(\'data:image/svg+xml;utf,<?xml version="1.0" standalone="no"?><svg xmlns="http://www.w3.org/2000/svg" height="18" width="18" version="1.1"><circle cx="9" cy="9" r="8" stroke="black" stroke-width="1" fill="white"/><rect x="4" y="8" width="10" height="2" style="fill:black;stroke-width:0"/><rect x="8" y="4" width="2" height="10" style="fill:black;stroke-width:0"/></svg>\') no-repeat left center;}\n\
 		.pf:'+cgnchk+' ~ label {background:url(\'data:image/svg+xml;utf,<?xml version="1.0" standalone="no"?><svg xmlns="http://www.w3.org/2000/svg" height="18" width="18" version="1.1"><circle cx="9" cy="9" r="8" stroke="black" stroke-width="1" fill="white"/><rect x="4" y="8" width="10" height="2" style="fill:black;stroke-width:0"/></svg>\') no-repeat left center;}\n\
 		.pf:'+cgchk+' ~ *:not(:nth-child(2)) {display:none;}\n\
-		.zoombox {position:relative;width:100%;overflow-x:scroll;}\n\
+		.zoombox {position:relative;width:100%;overflow-x:scroll;-webkit-user-select:none;-moz-user-select:none;user-select:none;}\n\
 		.timeline {position:relative;font-size:14px;cursor:pointer;width:100%; overflow:hidden;background:linear-gradient(#cccccc, white);}\n\
-		.thread {position:absolute;height:0%;overflow:hidden;line-height:'+devtextH+';font-size:'+devtextS+';border:1px solid;text-align:center;white-space:nowrap;background-color:rgba(204,204,204,0.5);}\n\
-		.thread.sync {background-color:'+sysvals.synccolor+';}\n\
-		.thread.bg {background-color:'+sysvals.kprobecolor+';}\n\
+		.thread {position:absolute;height:0%;overflow:hidden;z-index:7;line-height:30px;font-size:14px;border:1px solid;text-align:center;white-space:nowrap;}\n\
+		.thread.ps {border-radius:3px;background:linear-gradient(to top, #ccc, #eee);}\n\
 		.thread:hover {background-color:white;border:1px solid red;'+hoverZ+'}\n\
+		.thread.sec,.thread.sec:hover {background-color:black;border:0;color:white;line-height:15px;font-size:10px;}\n\
 		.hover {background-color:white;border:1px solid red;'+hoverZ+'}\n\
 		.hover.sync {background-color:white;}\n\
-		.hover.bg {background-color:white;}\n\
-		.traceevent {position:absolute;font-size:10px;overflow:hidden;color:black;text-align:center;white-space:nowrap;border-radius:5px;border:1px solid black;background:linear-gradient(to bottom right,rgba(204,204,204,1),rgba(150,150,150,1));}\n\
-		.traceevent:hover {background:white;}\n\
+		.hover.bg,.hover.kth,.hover.sync,.hover.ps {background-color:white;}\n\
+		.jiffie {position:absolute;pointer-events: none;z-index:8;}\n\
+		.traceevent {position:absolute;font-size:10px;z-index:7;overflow:hidden;color:black;text-align:center;white-space:nowrap;border-radius:5px;border:1px solid black;background:linear-gradient(to bottom right,#CCC,#969696);}\n\
+		.traceevent:hover {color:white;font-weight:bold;border:1px solid white;}\n\
 		.phase {position:absolute;overflow:hidden;border:0px;text-align:center;}\n\
 		.phaselet {position:absolute;overflow:hidden;border:0px;text-align:center;height:100px;font-size:24px;}\n\
-		.t {z-index:2;position:absolute;pointer-events:none;top:0%;height:100%;border-right:1px solid black;}\n\
+		.t {position:absolute;line-height:'+('%d'%scaleTH)+'px;pointer-events:none;top:0;height:100%;border-right:1px solid black;z-index:6;}\n\
+		.err {position:absolute;top:0%;height:100%;border-right:3px solid red;color:red;font:bold 14px Times;line-height:18px;}\n\
 		.legend {position:relative; width:100%; height:40px; text-align:center;margin-bottom:20px}\n\
 		.legend .square {position:absolute;cursor:pointer;top:10px; width:0px;height:20px;border:1px solid;padding-left:20px;}\n\
 		button {height:40px;width:200px;margin-bottom:20px;margin-top:20px;font-size:24px;}\n\
@@ -3327,7 +3643,8 @@ def createHTML(testruns):
 		a:active {color:white;}\n\
 		.version {position:relative;float:left;color:white;font-size:10px;line-height:30px;margin-left:10px;}\n\
 		#devicedetail {height:100px;box-shadow:5px 5px 20px black;}\n\
-		.tblock {position:absolute;height:100%;}\n\
+		.tblock {position:absolute;height:100%;background-color:#ddd;}\n\
+		.tback {position:absolute;width:100%;background:linear-gradient(#ccc, #ddd);}\n\
 		.bg {z-index:1;}\n\
 	</style>\n</head>\n<body>\n'
 
@@ -3342,6 +3659,8 @@ def createHTML(testruns):
 	# write the test title and general info header
 	if(sysvals.stamp['time'] != ""):
 		hf.write(headline_version)
+		if sysvals.logmsg:
+			hf.write('<button id="showtest" class="logbtn">log</button>')
 		if sysvals.addlogs and sysvals.dmesgfile:
 			hf.write('<button id="showdmesg" class="logbtn">dmesg</button>')
 		if sysvals.addlogs and sysvals.ftracefile:
@@ -3359,6 +3678,9 @@ def createHTML(testruns):
 	# draw the colored boxes for the device detail section
 	for data in testruns:
 		hf.write('<div id="devicedetail%d">\n' % data.testnumber)
+		pscolor = 'linear-gradient(to top left, #ccc, #eee)'
+		hf.write(html_phaselet.format('pre_suspend_process', \
+			'0', '0', pscolor))
 		for b in data.phases:
 			phase = data.dmesg[b]
 			length = phase['end']-phase['start']
@@ -3366,13 +3688,17 @@ def createHTML(testruns):
 			width = '%.3f' % ((length*100.0)/tTotal)
 			hf.write(html_phaselet.format(b, left, width, \
 				data.dmesg[b]['color']))
+		hf.write(html_phaselet.format('post_resume_process', \
+			'0', '0', pscolor))
 		if sysvals.suspendmode == 'command':
-			hf.write(html_phaselet.format('cmdexec', '0', '0', \
-				data.dmesg['resume_complete']['color']))
+			hf.write(html_phaselet.format('cmdexec', '0', '0', pscolor))
 		hf.write('</div>\n')
 	hf.write('</div>\n')
 
 	# write the ftrace data (callgraph)
+	if sysvals.cgtest >= 0 and len(testruns) > sysvals.cgtest:
+		data = testruns[sysvals.cgtest]
+	else:
 		data = testruns[-1]
 	if(sysvals.usecallgraph and not sysvals.embedded):
 		hf.write('<section id="callgraphs" class="callgraph">\n')
@@ -3383,6 +3709,8 @@ def createHTML(testruns):
 		html_func_leaf = '<article>{0} {1}</article>\n'
 		num = 0
 		for p in data.phases:
+			if sysvals.cgphase and p != sysvals.cgphase:
+				continue
 			list = data.dmesg[p]['list']
 			for devname in data.sortedDevices(p):
 				if('ftrace' not in list[devname]):
@@ -3420,11 +3748,15 @@ def createHTML(testruns):
 				hf.write(html_func_end)
 		hf.write('\n\n    </section>\n')
 
+	# add the test log as a hidden div
+	if sysvals.logmsg:
+		hf.write('<div id="testlog" style="display:none;">\n'+sysvals.logmsg+'</div>\n')
 	# add the dmesg log as a hidden div
 	if sysvals.addlogs and sysvals.dmesgfile:
 		hf.write('<div id="dmesglog" style="display:none;">\n')
 		lf = open(sysvals.dmesgfile, 'r')
 		for line in lf:
+			line = line.replace('<', '&lt').replace('>', '&gt')
 			hf.write(line)
 		lf.close()
 		hf.write('</div>\n')
@@ -3475,8 +3807,9 @@ def addScriptCode(hf, testruns):
 	script_code = \
 	'<script type="text/javascript">\n'+detail+\
 	'	var resolution = -1;\n'\
+	'	var dragval = [0, 0];\n'\
 	'	function redrawTimescale(t0, tMax, tS) {\n'\
-	'		var rline = \'<div class="t" style="left:0;border-left:1px solid black;border-right:0;"><cR><-R</cR></div>\';\n'\
+	'		var rline = \'<div class="t" style="left:0;border-left:1px solid black;border-right:0;"><cS>&larr;R</cS></div>\';\n'\
 	'		var tTotal = tMax - t0;\n'\
 	'		var list = document.getElementsByClassName("tblock");\n'\
 	'		for (var i = 0; i < list.length; i++) {\n'\
@@ -3501,7 +3834,7 @@ def addScriptCode(hf, testruns):
 	'					pos = 100 - (((j)*tS*100)/mTotal) - divEdge;\n'\
 	'					val = (j-divTotal+1)*tS;\n'\
 	'					if(j == divTotal - 1)\n'\
-	'						htmlline = \'<div class="t" style="right:\'+pos+\'%"><cS>S-></cS></div>\';\n'\
+	'						htmlline = \'<div class="t" style="right:\'+pos+\'%"><cS>S&rarr;</cS></div>\';\n'\
 	'					else\n'\
 	'						htmlline = \'<div class="t" style="right:\'+pos+\'%">\'+val+\'ms</div>\';\n'\
 	'				}\n'\
@@ -3513,6 +3846,7 @@ def addScriptCode(hf, testruns):
 	'	function zoomTimeline() {\n'\
 	'		var dmesg = document.getElementById("dmesg");\n'\
 	'		var zoombox = document.getElementById("dmesgzoombox");\n'\
+	'		var left = zoombox.scrollLeft;\n'\
 	'		var val = parseFloat(dmesg.style.width);\n'\
 	'		var newval = 100;\n'\
 	'		var sh = window.outerWidth / 2;\n'\
@@ -3520,12 +3854,12 @@ def addScriptCode(hf, testruns):
 	'			newval = val * 1.2;\n'\
 	'			if(newval > 910034) newval = 910034;\n'\
 	'			dmesg.style.width = newval+"%";\n'\
-	'			zoombox.scrollLeft = ((zoombox.scrollLeft + sh) * newval / val) - sh;\n'\
+	'			zoombox.scrollLeft = ((left + sh) * newval / val) - sh;\n'\
 	'		} else if (this.id == "zoomout") {\n'\
 	'			newval = val / 1.2;\n'\
 	'			if(newval < 100) newval = 100;\n'\
 	'			dmesg.style.width = newval+"%";\n'\
-	'			zoombox.scrollLeft = ((zoombox.scrollLeft + sh) * newval / val) - sh;\n'\
+	'			zoombox.scrollLeft = ((left + sh) * newval / val) - sh;\n'\
 	'		} else {\n'\
 	'			zoombox.scrollLeft = 0;\n'\
 	'			dmesg.style.width = "100%";\n'\
@@ -3542,8 +3876,12 @@ def addScriptCode(hf, testruns):
 	'		resolution = tS[i];\n'\
 	'		redrawTimescale(t0, tMax, tS[i]);\n'\
 	'	}\n'\
+	'	function deviceName(title) {\n'\
+	'		var name = title.slice(0, title.indexOf(" ("));\n'\
+	'		return name;\n'\
+	'	}\n'\
 	'	function deviceHover() {\n'\
-	'		var name = this.title.slice(0, this.title.indexOf(" ("));\n'\
+	'		var name = deviceName(this.title);\n'\
 	'		var dmesg = document.getElementById("dmesg");\n'\
 	'		var dev = dmesg.getElementsByClassName("thread");\n'\
 	'		var cpu = -1;\n'\
@@ -3552,7 +3890,7 @@ def addScriptCode(hf, testruns):
 	'		else if(name.match("CPU_OFF\[[0-9]*\]"))\n'\
 	'			cpu = parseInt(name.slice(8));\n'\
 	'		for (var i = 0; i < dev.length; i++) {\n'\
-	'			dname = dev[i].title.slice(0, dev[i].title.indexOf(" ("));\n'\
+	'			dname = deviceName(dev[i].title);\n'\
 	'			var cname = dev[i].className.slice(dev[i].className.indexOf("thread"));\n'\
 	'			if((cpu >= 0 && dname.match("CPU_O[NF]*\\\[*"+cpu+"\\\]")) ||\n'\
 	'				(name == dname))\n'\
@@ -3578,7 +3916,7 @@ def addScriptCode(hf, testruns):
 	'			total[2] = (total[2]+total[4])/2;\n'\
 	'		}\n'\
 	'		var devtitle = document.getElementById("devicedetailtitle");\n'\
-	'		var name = title.slice(0, title.indexOf(" ("));\n'\
+	'		var name = deviceName(title);\n'\
 	'		if(cpu >= 0) name = "CPU"+cpu;\n'\
 	'		var driver = "";\n'\
 	'		var tS = "<t2>(</t2>";\n'\
@@ -3600,7 +3938,7 @@ def addScriptCode(hf, testruns):
 	'	function deviceDetail() {\n'\
 	'		var devinfo = document.getElementById("devicedetail");\n'\
 	'		devinfo.style.display = "block";\n'\
-	'		var name = this.title.slice(0, this.title.indexOf(" ("));\n'\
+	'		var name = deviceName(this.title);\n'\
 	'		var cpu = -1;\n'\
 	'		if(name.match("CPU_ON\[[0-9]*\]"))\n'\
 	'			cpu = parseInt(name.slice(7));\n'\
@@ -3615,7 +3953,7 @@ def addScriptCode(hf, testruns):
 	'		var pd = pdata[0];\n'\
 	'		var total = [0.0, 0.0, 0.0];\n'\
 	'		for (var i = 0; i < dev.length; i++) {\n'\
-	'			dname = dev[i].title.slice(0, dev[i].title.indexOf(" ("));\n'\
+	'			dname = deviceName(dev[i].title);\n'\
 	'			if((cpu >= 0 && dname.match("CPU_O[NF]*\\\[*"+cpu+"\\\]")) ||\n'\
 	'				(name == dname))\n'\
 	'			{\n'\
@@ -3656,7 +3994,7 @@ def addScriptCode(hf, testruns):
 	'					phases[i].title = phases[i].id+" "+pd[phases[i].id]+" ms";\n'\
 	'					left += w;\n'\
 	'					var time = "<t4 style=\\"font-size:"+fs+"px\\">"+pd[phases[i].id]+" ms<br></t4>";\n'\
-	'					var pname = "<t3 style=\\"font-size:"+fs2+"px\\">"+phases[i].id.replace("_", " ")+"</t3>";\n'\
+	'					var pname = "<t3 style=\\"font-size:"+fs2+"px\\">"+phases[i].id.replace(new RegExp("_", "g"), " ")+"</t3>";\n'\
 	'					phases[i].innerHTML = time+pname;\n'\
 	'				} else {\n'\
 	'					phases[i].style.width = "0%";\n'\
@@ -3677,12 +4015,7 @@ def addScriptCode(hf, testruns):
 	'		}\n'\
 	'	}\n'\
 	'	function devListWindow(e) {\n'\
-	'		var sx = e.clientX;\n'\
-	'		if(sx > window.innerWidth - 440)\n'\
-	'			sx = window.innerWidth - 440;\n'\
-	'		var cfg="top="+e.screenY+", left="+sx+", width=440, height=720, scrollbars=yes";\n'\
-	'		var win = window.open("", "_blank", cfg);\n'\
-	'		if(window.chrome) win.moveBy(sx, 0);\n'\
+	'		var win = window.open();\n'\
 	'		var html = "<title>"+e.target.innerHTML+"</title>"+\n'\
 	'			"<style type=\\"text/css\\">"+\n'\
 	'			"   ul {list-style-type:circle;padding-left:10px;margin-left:10px;}"+\n'\
@@ -3692,6 +4025,12 @@ def addScriptCode(hf, testruns):
 	'			dt = devtable[1];\n'\
 	'		win.document.write(html+dt);\n'\
 	'	}\n'\
+	'	function errWindow() {\n'\
+	'		var text = this.id;\n'\
+	'		var win = window.open();\n'\
+	'		win.document.write("<pre>"+text+"</pre>");\n'\
+	'		win.document.close();\n'\
+	'	}\n'\
 	'	function logWindow(e) {\n'\
 	'		var name = e.target.id.slice(4);\n'\
 	'		var win = window.open();\n'\
@@ -3702,16 +4041,46 @@ def addScriptCode(hf, testruns):
 	'	}\n'\
 	'	function onClickPhase(e) {\n'\
 	'	}\n'\
+	'	function onMouseDown(e) {\n'\
+	'		dragval[0] = e.clientX;\n'\
+	'		dragval[1] = document.getElementById("dmesgzoombox").scrollLeft;\n'\
+	'		document.onmousemove = onMouseMove;\n'\
+	'	}\n'\
+	'	function onMouseMove(e) {\n'\
+	'		var zoombox = document.getElementById("dmesgzoombox");\n'\
+	'		zoombox.scrollLeft = dragval[1] + dragval[0] - e.clientX;\n'\
+	'	}\n'\
+	'	function onMouseUp(e) {\n'\
+	'		document.onmousemove = null;\n'\
+	'	}\n'\
+	'	function onKeyPress(e) {\n'\
+	'		var c = e.charCode;\n'\
+	'		if(c != 42 && c != 43 && c != 45) return;\n'\
+	'		var click = document.createEvent("Events");\n'\
+	'		click.initEvent("click", true, false);\n'\
+	'		if(c == 43)  \n'\
+	'			document.getElementById("zoomin").dispatchEvent(click);\n'\
+	'		else if(c == 45)\n'\
+	'			document.getElementById("zoomout").dispatchEvent(click);\n'\
+	'		else if(c == 42)\n'\
+	'			document.getElementById("zoomdef").dispatchEvent(click);\n'\
+	'	}\n'\
 	'	window.addEventListener("resize", function () {zoomTimeline();});\n'\
 	'	window.addEventListener("load", function () {\n'\
 	'		var dmesg = document.getElementById("dmesg");\n'\
 	'		dmesg.style.width = "100%"\n'\
+	'		dmesg.onmousedown = onMouseDown;\n'\
+	'		document.onmouseup = onMouseUp;\n'\
+	'		document.onkeypress = onKeyPress;\n'\
 	'		document.getElementById("zoomin").onclick = zoomTimeline;\n'\
 	'		document.getElementById("zoomout").onclick = zoomTimeline;\n'\
 	'		document.getElementById("zoomdef").onclick = zoomTimeline;\n'\
 	'		var list = document.getElementsByClassName("square");\n'\
 	'		for (var i = 0; i < list.length; i++)\n'\
 	'			list[i].onclick = onClickPhase;\n'\
+	'		var list = document.getElementsByClassName("err");\n'\
+	'		for (var i = 0; i < list.length; i++)\n'\
+	'			list[i].onclick = errWindow;\n'\
 	'		var list = document.getElementsByClassName("logbtn");\n'\
 	'		for (var i = 0; i < list.length; i++)\n'\
 	'			list[i].onclick = logWindow;\n'\
@@ -3734,9 +4103,7 @@ def addScriptCode(hf, testruns):
 #	 Execute system suspend through the sysfs interface, then copy the output
 #	 dmesg and ftrace files to the test output directory.
 def executeSuspend():
-	global sysvals
-
-	t0 = time.time()*1000
+	pm = ProcessMonitor()
 	tp = sysvals.tpath
 	fwdata = []
 	# mark the start point in the kernel ring buffer just as we start
@@ -3745,30 +4112,39 @@ def executeSuspend():
 	if(sysvals.usecallgraph or sysvals.usetraceevents):
 		print('START TRACING')
 		sysvals.fsetVal('1', 'tracing_on')
+		if sysvals.useprocmon:
+			pm.start()
 	# execute however many s/r runs requested
 	for count in range(1,sysvals.execcount+1):
-		# if this is test2 and there's a delay, start here
+		# x2delay in between test runs
 		if(count > 1 and sysvals.x2delay > 0):
-			tN = time.time()*1000
-			while (tN - t0) < sysvals.x2delay:
-				tN = time.time()*1000
-				time.sleep(0.001)
-		# initiate suspend
-		if(sysvals.usecallgraph or sysvals.usetraceevents):
-			sysvals.fsetVal('SUSPEND START', 'trace_marker')
-		if sysvals.suspendmode == 'command':
+			sysvals.fsetVal('WAIT %d' % sysvals.x2delay, 'trace_marker')
+			time.sleep(sysvals.x2delay/1000.0)
+			sysvals.fsetVal('WAIT END', 'trace_marker')
+		# start message
+		if sysvals.testcommand != '':
 			print('COMMAND START')
-			if(sysvals.rtcwake):
-				print('will issue an rtcwake in %d seconds' % sysvals.rtcwaketime)
-				sysvals.rtcWakeAlarmOn()
-			os.system(sysvals.testcommand)
 		else:
 			if(sysvals.rtcwake):
 				print('SUSPEND START')
-				print('will autoresume in %d seconds' % sysvals.rtcwaketime)
-				sysvals.rtcWakeAlarmOn()
 			else:
 				print('SUSPEND START (press a key to resume)')
+		# set rtcwake
+		if(sysvals.rtcwake):
+			print('will issue an rtcwake in %d seconds' % sysvals.rtcwaketime)
+			sysvals.rtcWakeAlarmOn()
+		# start of suspend trace marker
+		if(sysvals.usecallgraph or sysvals.usetraceevents):
+			sysvals.fsetVal('SUSPEND START', 'trace_marker')
+		# predelay delay
+		if(count == 1 and sysvals.predelay > 0):
+			sysvals.fsetVal('WAIT %d' % sysvals.predelay, 'trace_marker')
+			time.sleep(sysvals.predelay/1000.0)
+			sysvals.fsetVal('WAIT END', 'trace_marker')
+		# initiate suspend or command
+		if sysvals.testcommand != '':
+			call(sysvals.testcommand+' 2>&1', shell=True);
+		else:
 			pf = open(sysvals.powerfile, 'w')
 			pf.write(sysvals.suspendmode)
 			# execution will pause here
@@ -3776,26 +4152,27 @@ def executeSuspend():
 				pf.close()
 			except:
 				pass
-		t0 = time.time()*1000
 		if(sysvals.rtcwake):
 			sysvals.rtcWakeAlarmOff()
+		# postdelay delay
+		if(count == sysvals.execcount and sysvals.postdelay > 0):
+			sysvals.fsetVal('WAIT %d' % sysvals.postdelay, 'trace_marker')
+			time.sleep(sysvals.postdelay/1000.0)
+			sysvals.fsetVal('WAIT END', 'trace_marker')
 		# return from suspend
 		print('RESUME COMPLETE')
 		if(sysvals.usecallgraph or sysvals.usetraceevents):
 			sysvals.fsetVal('RESUME COMPLETE', 'trace_marker')
-		if(sysvals.suspendmode == 'mem'):
+		if(sysvals.suspendmode == 'mem' or sysvals.suspendmode == 'command'):
 			fwdata.append(getFPDT(False))
-	# look for post resume events after the last test run
-	t = sysvals.postresumetime
-	if(t > 0):
-		print('Waiting %d seconds for POST-RESUME trace events...' % t)
-		time.sleep(t)
 	# stop ftrace
 	if(sysvals.usecallgraph or sysvals.usetraceevents):
+		if sysvals.useprocmon:
+			pm.stop()
 		sysvals.fsetVal('0', 'tracing_on')
 		print('CAPTURING TRACE')
 		writeDatafileHeader(sysvals.ftracefile, fwdata)
-		os.system('cat '+tp+'trace >> '+sysvals.ftracefile)
+		call('cat '+tp+'trace >> '+sysvals.ftracefile, shell=True)
 		sysvals.fsetVal('', 'trace')
 		devProps()
 	# grab a copy of the dmesg output
@@ -3804,17 +4181,12 @@ def executeSuspend():
 	sysvals.getdmesg()
 
 def writeDatafileHeader(filename, fwdata):
-	global sysvals
-
-	prt = sysvals.postresumetime
 	fp = open(filename, 'a')
 	fp.write(sysvals.teststamp+'\n')
-	if(sysvals.suspendmode == 'mem'):
+	if(sysvals.suspendmode == 'mem' or sysvals.suspendmode == 'command'):
 		for fw in fwdata:
 			if(fw):
 				fp.write('# fwsuspend %u fwresume %u\n' % (fw[0], fw[1]))
-	if(prt > 0):
-		fp.write('# post resume time %u\n' % prt)
 	fp.close()
 
 # Function: setUSBDevicesAuto
@@ -3824,18 +4196,16 @@ def writeDatafileHeader(filename, fwdata):
 #	 to always-on since the kernel cant determine if the device can
 #	 properly autosuspend
 def setUSBDevicesAuto():
-	global sysvals
-
 	rootCheck(True)
 	for dirname, dirnames, filenames in os.walk('/sys/devices'):
 		if(re.match('.*/usb[0-9]*.*', dirname) and
 			'idVendor' in filenames and 'idProduct' in filenames):
-			os.system('echo auto > %s/power/control' % dirname)
+			call('echo auto > %s/power/control' % dirname, shell=True)
 			name = dirname.split('/')[-1]
-			desc = os.popen('cat %s/product 2>/dev/null' % \
-				dirname).read().replace('\n', '')
-			ctrl = os.popen('cat %s/power/control 2>/dev/null' % \
-				dirname).read().replace('\n', '')
+			desc = Popen(['cat', '%s/product' % dirname],
+				stderr=PIPE, stdout=PIPE).stdout.read().replace('\n', '')
+			ctrl = Popen(['cat', '%s/power/control' % dirname],
+				stderr=PIPE, stdout=PIPE).stdout.read().replace('\n', '')
 			print('control is %s for %6s: %s' % (ctrl, name, desc))
 
 # Function: yesno
@@ -3872,8 +4242,6 @@ def ms2nice(val):
 #	 Detect all the USB hosts and devices currently connected and add
 #	 a list of USB device names to sysvals for better timeline readability
 def detectUSB():
-	global sysvals
-
 	field = {'idVendor':'', 'idProduct':'', 'product':'', 'speed':''}
 	power = {'async':'', 'autosuspend':'', 'autosuspend_delay_ms':'',
 			 'control':'', 'persist':'', 'runtime_enabled':'',
@@ -3899,12 +4267,12 @@ def detectUSB():
 		if(re.match('.*/usb[0-9]*.*', dirname) and
 			'idVendor' in filenames and 'idProduct' in filenames):
 			for i in field:
-				field[i] = os.popen('cat %s/%s 2>/dev/null' % \
-					(dirname, i)).read().replace('\n', '')
+				field[i] = Popen(['cat', '%s/%s' % (dirname, i)],
+					stderr=PIPE, stdout=PIPE).stdout.read().replace('\n', '')
 			name = dirname.split('/')[-1]
 			for i in power:
-				power[i] = os.popen('cat %s/power/%s 2>/dev/null' % \
-					(dirname, i)).read().replace('\n', '')
+				power[i] = Popen(['cat', '%s/power/%s' % (dirname, i)],
+					stderr=PIPE, stdout=PIPE).stdout.read().replace('\n', '')
 			if(re.match('usb[0-9]*', name)):
 				first = '%-8s' % name
 			else:
@@ -3928,7 +4296,6 @@ def detectUSB():
 # Description:
 #	 Retrieve a list of properties for all devices in the trace log
 def devProps(data=0):
-	global sysvals
 	props = dict()
 
 	if data:
@@ -3953,7 +4320,7 @@ def devProps(data=0):
 		return
 
 	if(os.path.exists(sysvals.ftracefile) == False):
-		doError('%s does not exist' % sysvals.ftracefile, False)
+		doError('%s does not exist' % sysvals.ftracefile)
 
 	# first get the list of devices we need properties for
 	msghead = 'Additional data added by AnalyzeSuspend'
@@ -3976,7 +4343,7 @@ def devProps(data=0):
 		m = re.match('.*: (?P<drv>.*) (?P<d>.*), parent: *(?P<p>.*), .*', m.group('msg'));
 		if(not m):
 			continue
-		drv, dev, par = m.group('drv'), m.group('d'), m.group('p')
+		dev = m.group('d')
 		if dev not in props:
 			props[dev] = DevProps()
 	tf.close()
@@ -4052,7 +4419,6 @@ def devProps(data=0):
 # Output:
 #	 A string list of the available modes
 def getModes():
-	global sysvals
 	modes = ''
 	if(os.path.exists(sysvals.powerfile)):
 		fp = open(sysvals.powerfile, 'r')
@@ -4066,8 +4432,6 @@ def getModes():
 # Arguments:
 #	 output: True to output the info to stdout, False otherwise
 def getFPDT(output):
-	global sysvals
-
 	rectype = {}
 	rectype[0] = 'Firmware Basic Boot Performance Record'
 	rectype[1] = 'S3 Performance Table Record'
@@ -4078,19 +4442,19 @@ def getFPDT(output):
 	rootCheck(True)
 	if(not os.path.exists(sysvals.fpdtpath)):
 		if(output):
-			doError('file does not exist: %s' % sysvals.fpdtpath, False)
+			doError('file does not exist: %s' % sysvals.fpdtpath)
 		return False
 	if(not os.access(sysvals.fpdtpath, os.R_OK)):
 		if(output):
-			doError('file is not readable: %s' % sysvals.fpdtpath, False)
+			doError('file is not readable: %s' % sysvals.fpdtpath)
 		return False
 	if(not os.path.exists(sysvals.mempath)):
 		if(output):
-			doError('file does not exist: %s' % sysvals.mempath, False)
+			doError('file does not exist: %s' % sysvals.mempath)
 		return False
 	if(not os.access(sysvals.mempath, os.R_OK)):
 		if(output):
-			doError('file is not readable: %s' % sysvals.mempath, False)
+			doError('file is not readable: %s' % sysvals.mempath)
 		return False
 
 	fp = open(sysvals.fpdtpath, 'rb')
@@ -4100,7 +4464,7 @@ def getFPDT(output):
 	if(len(buf) < 36):
 		if(output):
 			doError('Invalid FPDT table data, should '+\
-				'be at least 36 bytes', False)
+				'be at least 36 bytes')
 		return False
 
 	table = struct.unpack('4sIBB6s8sI4sI', buf[0:36])
@@ -4199,7 +4563,6 @@ def getFPDT(output):
 # Output:
 #	 True if the test will work, False if not
 def statusCheck(probecheck=False):
-	global sysvals
 	status = True
 
 	print('Checking this system (%s)...' % platform.node())
@@ -4282,37 +4645,14 @@ def statusCheck(probecheck=False):
 	if not probecheck:
 		return status
 
-	if (sysvals.usecallgraph and len(sysvals.debugfuncs) > 0) or len(sysvals.kprobes) > 0:
-		sysvals.initFtrace(True)
-
-	# verify callgraph debugfuncs
-	if sysvals.usecallgraph and len(sysvals.debugfuncs) > 0:
-		print('    verifying these ftrace callgraph functions work:')
-		sysvals.setFtraceFilterFunctions(sysvals.debugfuncs)
-		fp = open(sysvals.tpath+'set_graph_function', 'r')
-		flist = fp.read().split('\n')
-		fp.close()
-		for func in sysvals.debugfuncs:
-			res = sysvals.colorText('NO')
-			if func in flist:
-				res = 'YES'
-			else:
-				for i in flist:
-					if ' [' in i and func == i.split(' ')[0]:
-						res = 'YES'
-						break
-			print('         %s: %s' % (func, res))
-
 	# verify kprobes
-	if len(sysvals.kprobes) > 0:
-		print('    verifying these kprobes work:')
-		for name in sorted(sysvals.kprobes):
-			if name in sysvals.tracefuncs:
-				continue
-			res = sysvals.colorText('NO')
-			if sysvals.testKprobe(sysvals.kprobes[name]):
-				res = 'YES'
-			print('         %s: %s' % (name, res))
+	if sysvals.usekprobes:
+		for name in sysvals.tracefuncs:
+			sysvals.defaultKprobe(name, sysvals.tracefuncs[name])
+		if sysvals.usedevsrc:
+			for name in sysvals.dev_tracefuncs:
+				sysvals.defaultKprobe(name, sysvals.dev_tracefuncs[name])
+		sysvals.addKprobes(True)
 
 	return status
 
@@ -4322,33 +4662,20 @@ def statusCheck(probecheck=False):
 # Arguments:
 #	 msg: the error message to print
 #	 help: True if printHelp should be called after, False otherwise
-def doError(msg, help):
+def doError(msg, help=False):
 	if(help == True):
 		printHelp()
 	print('ERROR: %s\n') % msg
 	sys.exit()
 
-# Function: doWarning
-# Description:
-#	 generic warning function for non-catastrophic anomalies
-# Arguments:
-#	 msg: the warning message to print
-#	 file: If not empty, a filename to request be sent to the owner for debug
-def doWarning(msg, file=''):
-	print('/* %s */') % msg
-	if(file):
-		print('/* For a fix, please send this'+\
-			' %s file to <todd.e.brandt@intel.com> */' % file)
-
 # Function: rootCheck
 # Description:
 #	 quick check to see if we have root access
 def rootCheck(fatal):
-	global sysvals
 	if(os.access(sysvals.powerfile, os.W_OK)):
 		return True
 	if fatal:
-		doError('This command must be run as root', False)
+		doError('This command must be run as root')
 	return False
 
 # Function: getArgInt
@@ -4389,71 +4716,61 @@ def getArgFloat(name, args, min, max, main=True):
 		doError(name+': value should be between %f and %f' % (min, max), True)
 	return val
 
-# Function: rerunTest
-# Description:
-#	 generate an output from an existing set of ftrace/dmesg logs
-def rerunTest():
-	global sysvals
-
-	if(sysvals.ftracefile != ''):
-		doesTraceLogHaveTraceEvents()
-	if(sysvals.dmesgfile == '' and not sysvals.usetraceeventsonly):
-		doError('recreating this html output '+\
-			'requires a dmesg file', False)
-	sysvals.setOutputFile()
-	vprint('Output file: %s' % sysvals.htmlfile)
+def processData():
 	print('PROCESSING DATA')
 	if(sysvals.usetraceeventsonly):
 		testruns = parseTraceLog()
+		if sysvals.dmesgfile:
+			dmesgtext = loadKernelLog(True)
+			for data in testruns:
+				data.extractErrorInfo(dmesgtext)
 	else:
 		testruns = loadKernelLog()
 		for data in testruns:
 			parseKernelLog(data)
-		if(sysvals.ftracefile != ''):
+		if(sysvals.ftracefile and (sysvals.usecallgraph or sysvals.usetraceevents)):
 			appendIncompleteTraceLog(testruns)
 	createHTML(testruns)
 
+# Function: rerunTest
+# Description:
+#	 generate an output from an existing set of ftrace/dmesg logs
+def rerunTest():
+	if sysvals.ftracefile:
+		doesTraceLogHaveTraceEvents()
+	if not sysvals.dmesgfile and not sysvals.usetraceeventsonly:
+		doError('recreating this html output requires a dmesg file')
+	sysvals.setOutputFile()
+	vprint('Output file: %s' % sysvals.htmlfile)
+	if(os.path.exists(sysvals.htmlfile) and not os.access(sysvals.htmlfile, os.W_OK)):
+		doError('missing permission to write to %s' % sysvals.htmlfile)
+	processData()
+
 # Function: runTest
 # Description:
 #	 execute a suspend/resume, gather the logs, and generate the output
 def runTest(subdir, testpath=''):
-	global sysvals
-
 	# prepare for the test
 	sysvals.initFtrace()
 	sysvals.initTestOutput(subdir, testpath)
-
-	vprint('Output files:\n    %s' % sysvals.dmesgfile)
-	if(sysvals.usecallgraph or
-		sysvals.usetraceevents or
-		sysvals.usetraceeventsonly):
-		vprint('    %s' % sysvals.ftracefile)
-	vprint('    %s' % sysvals.htmlfile)
+	vprint('Output files:\n\t%s\n\t%s\n\t%s' % \
+		(sysvals.dmesgfile, sysvals.ftracefile, sysvals.htmlfile))
 
 	# execute the test
 	executeSuspend()
 	sysvals.cleanupFtrace()
+	processData()
 
-	# analyze the data and create the html output
-	print('PROCESSING DATA')
-	if(sysvals.usetraceeventsonly):
-		# data for kernels 3.15 or newer is entirely in ftrace
-		testruns = parseTraceLog()
-	else:
-		# data for kernels older than 3.15 is primarily in dmesg
-		testruns = loadKernelLog()
-		for data in testruns:
-			parseKernelLog(data)
-		if(sysvals.usecallgraph or sysvals.usetraceevents):
-			appendIncompleteTraceLog(testruns)
-	createHTML(testruns)
+	# if running as root, change output dir owner to sudo_user
+	if os.path.isdir(sysvals.testdir) and os.getuid() == 0 and \
+		'SUDO_USER' in os.environ:
+		cmd = 'chown -R {0}:{0} {1} > /dev/null 2>&1'
+		call(cmd.format(os.environ['SUDO_USER'], sysvals.testdir), shell=True)
 
 # Function: runSummary
 # Description:
 #	 create a summary of tests in a sub-directory
 def runSummary(subdir, output):
-	global sysvals
-
 	# get a list of ftrace output files
 	files = []
 	for dirname, dirnames, filenames in os.walk(subdir):
@@ -4509,12 +4826,12 @@ def checkArgBool(value):
 # Description:
 #	 Configure the script via the info in a config file
 def configFromFile(file):
-	global sysvals
 	Config = ConfigParser.ConfigParser()
 
-	ignorekprobes = False
 	Config.read(file)
 	sections = Config.sections()
+	overridekprobes = False
+	overridedevkprobes = False
 	if 'Settings' in sections:
 		for opt in Config.options('Settings'):
 			value = Config.get('Settings', opt).lower()
@@ -4524,19 +4841,19 @@ def configFromFile(file):
 				sysvals.addlogs = checkArgBool(value)
 			elif(opt.lower() == 'dev'):
 				sysvals.usedevsrc = checkArgBool(value)
-			elif(opt.lower() == 'ignorekprobes'):
-				ignorekprobes = checkArgBool(value)
+			elif(opt.lower() == 'proc'):
+				sysvals.useprocmon = checkArgBool(value)
 			elif(opt.lower() == 'x2'):
 				if checkArgBool(value):
 					sysvals.execcount = 2
 			elif(opt.lower() == 'callgraph'):
 				sysvals.usecallgraph = checkArgBool(value)
-			elif(opt.lower() == 'callgraphfunc'):
-				sysvals.debugfuncs = []
-				if value:
-					value = value.split(',')
-				for i in value:
-					sysvals.debugfuncs.append(i.strip())
+			elif(opt.lower() == 'override-timeline-functions'):
+				overridekprobes = checkArgBool(value)
+			elif(opt.lower() == 'override-dev-timeline-functions'):
+				overridedevkprobes = checkArgBool(value)
+			elif(opt.lower() == 'devicefilter'):
+				sysvals.setDeviceFilter(value)
 			elif(opt.lower() == 'expandcg'):
 				sysvals.cgexp = checkArgBool(value)
 			elif(opt.lower() == 'srgap'):
@@ -4548,8 +4865,10 @@ def configFromFile(file):
 				sysvals.testcommand = value
 			elif(opt.lower() == 'x2delay'):
 				sysvals.x2delay = getArgInt('-x2delay', value, 0, 60000, False)
-			elif(opt.lower() == 'postres'):
-				sysvals.postresumetime = getArgInt('-postres', value, 0, 3600, False)
+			elif(opt.lower() == 'predelay'):
+				sysvals.predelay = getArgInt('-predelay', value, 0, 60000, False)
+			elif(opt.lower() == 'postdelay'):
+				sysvals.postdelay = getArgInt('-postdelay', value, 0, 60000, False)
 			elif(opt.lower() == 'rtcwake'):
 				sysvals.rtcwake = True
 				sysvals.rtcwaketime = getArgInt('-rtcwake', value, 0, 3600, False)
@@ -4557,53 +4876,50 @@ def configFromFile(file):
 				sysvals.setPrecision(getArgInt('-timeprec', value, 0, 6, False))
 			elif(opt.lower() == 'mindev'):
 				sysvals.mindevlen = getArgFloat('-mindev', value, 0.0, 10000.0, False)
+			elif(opt.lower() == 'callloop-maxgap'):
+				sysvals.callloopmaxgap = getArgFloat('-callloop-maxgap', value, 0.0, 1.0, False)
+			elif(opt.lower() == 'callloop-maxlen'):
+				sysvals.callloopmaxgap = getArgFloat('-callloop-maxlen', value, 0.0, 1.0, False)
 			elif(opt.lower() == 'mincg'):
 				sysvals.mincglen = getArgFloat('-mincg', value, 0.0, 10000.0, False)
-			elif(opt.lower() == 'kprobecolor'):
-				try:
-					val = int(value, 16)
-					sysvals.kprobecolor = '#'+value
-				except:
-					sysvals.kprobecolor = value
-			elif(opt.lower() == 'synccolor'):
-				try:
-					val = int(value, 16)
-					sysvals.synccolor = '#'+value
-				except:
-					sysvals.synccolor = value
 			elif(opt.lower() == 'output-dir'):
-				args = dict()
-				n = datetime.now()
-				args['date'] = n.strftime('%y%m%d')
-				args['time'] = n.strftime('%H%M%S')
-				args['hostname'] = sysvals.hostname
-				sysvals.outdir = value.format(**args)
+				sysvals.setOutputFolder(value)
 
 	if sysvals.suspendmode == 'command' and not sysvals.testcommand:
-		doError('No command supplied for mode "command"', False)
+		doError('No command supplied for mode "command"')
+
+	# compatibility errors
 	if sysvals.usedevsrc and sysvals.usecallgraph:
-		doError('dev and callgraph cannot both be true', False)
-	if sysvals.usecallgraph and sysvals.execcount > 1:
-		doError('-x2 is not compatible with -f', False)
+		doError('-dev is not compatible with -f')
+	if sysvals.usecallgraph and sysvals.useprocmon:
+		doError('-proc is not compatible with -f')
 
-	if ignorekprobes:
-		return
+	if overridekprobes:
+		sysvals.tracefuncs = dict()
+	if overridedevkprobes:
+		sysvals.dev_tracefuncs = dict()
 
 	kprobes = dict()
-	archkprobe = 'Kprobe_'+platform.machine()
-	if archkprobe in sections:
-		for name in Config.options(archkprobe):
-			kprobes[name] = Config.get(archkprobe, name)
-	if 'Kprobe' in sections:
-		for name in Config.options('Kprobe'):
-			kprobes[name] = Config.get('Kprobe', name)
+	kprobesec = 'dev_timeline_functions_'+platform.machine()
+	if kprobesec in sections:
+		for name in Config.options(kprobesec):
+			text = Config.get(kprobesec, name)
+			kprobes[name] = (text, True)
+	kprobesec = 'timeline_functions_'+platform.machine()
+	if kprobesec in sections:
+		for name in Config.options(kprobesec):
+			if name in kprobes:
+				doError('Duplicate timeline function found "%s"' % (name))
+			text = Config.get(kprobesec, name)
+			kprobes[name] = (text, False)
 
 	for name in kprobes:
 		function = name
 		format = name
 		color = ''
 		args = dict()
-		data = kprobes[name].split()
+		text, dev = kprobes[name]
+		data = text.split()
 		i = 0
 		for val in data:
 			# bracketted strings are special formatting, read them separately
@@ -4626,28 +4942,30 @@ def configFromFile(file):
 				args[d[0]] = d[1]
 			i += 1
 		if not function or not format:
-			doError('Invalid kprobe: %s' % name, False)
+			doError('Invalid kprobe: %s' % name)
 		for arg in re.findall('{(?P<n>[a-z,A-Z,0-9]*)}', format):
 			if arg not in args:
-				doError('Kprobe "%s" is missing argument "%s"' % (name, arg), False)
-		if name in sysvals.kprobes:
-			doError('Duplicate kprobe found "%s"' % (name), False)
-		vprint('Adding KPROBE: %s %s %s %s' % (name, function, format, args))
-		sysvals.kprobes[name] = {
+				doError('Kprobe "%s" is missing argument "%s"' % (name, arg))
+		if (dev and name in sysvals.dev_tracefuncs) or (not dev and name in sysvals.tracefuncs):
+			doError('Duplicate timeline function found "%s"' % (name))
+
+		kp = {
 			'name': name,
 			'func': function,
 			'format': format,
-			'args': args,
-			'mask': re.sub('{(?P<n>[a-z,A-Z,0-9]*)}', '.*', format)
+			sysvals.archargs: args
 		}
 		if color:
-			sysvals.kprobes[name]['color'] = color
+			kp['color'] = color
+		if dev:
+			sysvals.dev_tracefuncs[name] = kp
+		else:
+			sysvals.tracefuncs[name] = kp
 
 # Function: printHelp
 # Description:
 #	 print out the help text
 def printHelp():
-	global sysvals
 	modes = getModes()
 
 	print('')
@@ -4672,34 +4990,37 @@ def printHelp():
 	print('  [general]')
 	print('   -h           Print this help text')
 	print('   -v           Print the current tool version')
-	print('    -config file Pull arguments and config options from a file')
+	print('   -config fn   Pull arguments and config options from file fn')
 	print('   -verbose     Print extra information during execution and analysis')
 	print('   -status      Test to see if the system is enabled to run this tool')
 	print('   -modes       List available suspend modes')
 	print('   -m mode      Mode to initiate for suspend %s (default: %s)') % (modes, sysvals.suspendmode)
 	print('   -o subdir    Override the output subdirectory')
-	print('  [advanced]')
 	print('   -rtcwake t   Use rtcwake to autoresume after <t> seconds (default: disabled)')
 	print('   -addlogs     Add the dmesg and ftrace logs to the html output')
-	print('    -multi n d  Execute <n> consecutive tests at <d> seconds intervals. The outputs will')
-	print('                be created in a new subdirectory with a summary page.')
 	print('   -srgap       Add a visible gap in the timeline between sus/res (default: disabled)')
-	print('    -cmd {s}    Instead of suspend/resume, run a command, e.g. "sync -d"')
+	print('  [advanced]')
+	print('   -cmd {s}     Run the timeline over a custom command, e.g. "sync -d"')
+	print('   -proc        Add usermode process info into the timeline (default: disabled)')
+	print('   -dev         Add kernel function calls and threads to the timeline (default: disabled)')
+	print('   -x2          Run two suspend/resumes back to back (default: disabled)')
+	print('   -x2delay t   Include t ms delay between multiple test runs (default: 0 ms)')
+	print('   -predelay t  Include t ms delay before 1st suspend (default: 0 ms)')
+	print('   -postdelay t Include t ms delay after last resume (default: 0 ms)')
 	print('   -mindev ms   Discard all device blocks shorter than ms milliseconds (e.g. 0.001 for us)')
-	print('    -mincg  ms  Discard all callgraphs shorter than ms milliseconds (e.g. 0.001 for us)')
-	print('    -timeprec N Number of significant digits in timestamps (0:S, [3:ms], 6:us)')
+	print('   -multi n d   Execute <n> consecutive tests at <d> seconds intervals. The outputs will')
+	print('                be created in a new subdirectory with a summary page.')
 	print('  [debug]')
 	print('   -f           Use ftrace to create device callgraphs (default: disabled)')
 	print('   -expandcg    pre-expand the callgraph data in the html output (default: disabled)')
 	print('   -flist       Print the list of functions currently being captured in ftrace')
 	print('   -flistall    Print all functions capable of being captured in ftrace')
 	print('   -fadd file   Add functions to be graphed in the timeline from a list in a text file')
-	print('    -filter "d1 d2 ..." Filter out all but this list of device names')
-	print('    -dev        Display common low level functions in the timeline')
-	print('  [post-resume task analysis]')
-	print('    -x2         Run two suspend/resumes back to back (default: disabled)')
-	print('    -x2delay t  Minimum millisecond delay <t> between the two test runs (default: 0 ms)')
-	print('    -postres t  Time after resume completion to wait for post-resume events (default: 0 S)')
+	print('   -filter "d1,d2,..." Filter out all but this comma-delimited list of device names')
+	print('   -mincg  ms   Discard all callgraphs shorter than ms milliseconds (e.g. 0.001 for us)')
+	print('   -cgphase P   Only show callgraph data for phase P (e.g. suspend_late)')
+	print('   -cgtest N    Only show callgraph data for test N (e.g. 0 or 1 in an x2 run)')
+	print('   -timeprec N  Number of significant digits in timestamps (0:S, [3:ms], 6:us)')
 	print('  [utilities]')
 	print('   -fpdt        Print out the contents of the ACPI Firmware Performance Data Table')
 	print('   -usbtopo     Print out the current USB topology with power info')
@@ -4739,26 +5060,22 @@ if __name__ == '__main__':
 			sys.exit()
 		elif(arg == '-x2'):
 			sysvals.execcount = 2
-			if(sysvals.usecallgraph):
-				doError('-x2 is not compatible with -f', False)
 		elif(arg == '-x2delay'):
 			sysvals.x2delay = getArgInt('-x2delay', args, 0, 60000)
-		elif(arg == '-postres'):
-			sysvals.postresumetime = getArgInt('-postres', args, 0, 3600)
+		elif(arg == '-predelay'):
+			sysvals.predelay = getArgInt('-predelay', args, 0, 60000)
+		elif(arg == '-postdelay'):
+			sysvals.postdelay = getArgInt('-postdelay', args, 0, 60000)
 		elif(arg == '-f'):
 			sysvals.usecallgraph = True
-			if(sysvals.execcount > 1):
-				doError('-x2 is not compatible with -f', False)
-			if(sysvals.usedevsrc):
-				doError('-dev is not compatible with -f', False)
 		elif(arg == '-addlogs'):
 			sysvals.addlogs = True
 		elif(arg == '-verbose'):
 			sysvals.verbose = True
+		elif(arg == '-proc'):
+			sysvals.useprocmon = True
 		elif(arg == '-dev'):
 			sysvals.usedevsrc = True
-			if(sysvals.usecallgraph):
-				doError('-dev is not compatible with -f', False)
 		elif(arg == '-rtcwake'):
 			sysvals.rtcwake = True
 			sysvals.rtcwaketime = getArgInt('-rtcwake', args, 0, 3600)
@@ -4768,6 +5085,21 @@ if __name__ == '__main__':
 			sysvals.mindevlen = getArgFloat('-mindev', args, 0.0, 10000.0)
 		elif(arg == '-mincg'):
 			sysvals.mincglen = getArgFloat('-mincg', args, 0.0, 10000.0)
+		elif(arg == '-cgtest'):
+			sysvals.cgtest = getArgInt('-cgtest', args, 0, 1)
+		elif(arg == '-cgphase'):
+			try:
+				val = args.next()
+			except:
+				doError('No phase name supplied', True)
+			d = Data(0)
+			if val not in d.phases:
+				doError('Invalid phase, valid phaess are %s' % d.phases, True)
+			sysvals.cgphase = val
+		elif(arg == '-callloop-maxgap'):
+			sysvals.callloopmaxgap = getArgFloat('-callloop-maxgap', args, 0.0, 1.0)
+		elif(arg == '-callloop-maxlen'):
+			sysvals.callloopmaxlen = getArgFloat('-callloop-maxlen', args, 0.0, 1.0)
 		elif(arg == '-cmd'):
 			try:
 				val = args.next()
@@ -4788,14 +5120,14 @@ if __name__ == '__main__':
 				val = args.next()
 			except:
 				doError('No subdirectory name supplied', True)
-			sysvals.outdir = val
+			sysvals.setOutputFolder(val)
 		elif(arg == '-config'):
 			try:
 				val = args.next()
 			except:
 				doError('No text file supplied', True)
 			if(os.path.exists(val) == False):
-				doError('%s does not exist' % val, False)
+				doError('%s does not exist' % val)
 			configFromFile(val)
 		elif(arg == '-fadd'):
 			try:
@@ -4803,7 +5135,7 @@ if __name__ == '__main__':
 			except:
 				doError('No text file supplied', True)
 			if(os.path.exists(val) == False):
-				doError('%s does not exist' % val, False)
+				doError('%s does not exist' % val)
 			sysvals.addFtraceFilterFunctions(val)
 		elif(arg == '-dmesg'):
 			try:
@@ -4813,7 +5145,7 @@ if __name__ == '__main__':
 			sysvals.notestrun = True
 			sysvals.dmesgfile = val
 			if(os.path.exists(sysvals.dmesgfile) == False):
-				doError('%s does not exist' % sysvals.dmesgfile, False)
+				doError('%s does not exist' % sysvals.dmesgfile)
 		elif(arg == '-ftrace'):
 			try:
 				val = args.next()
@@ -4822,7 +5154,7 @@ if __name__ == '__main__':
 			sysvals.notestrun = True
 			sysvals.ftracefile = val
 			if(os.path.exists(sysvals.ftracefile) == False):
-				doError('%s does not exist' % sysvals.ftracefile, False)
+				doError('%s does not exist' % sysvals.ftracefile)
 		elif(arg == '-summary'):
 			try:
 				val = args.next()
@@ -4832,7 +5164,7 @@ if __name__ == '__main__':
 			cmdarg = val
 			sysvals.notestrun = True
 			if(os.path.isdir(val) == False):
-				doError('%s is not accesible' % val, False)
+				doError('%s is not accesible' % val)
 		elif(arg == '-filter'):
 			try:
 				val = args.next()
@@ -4842,6 +5174,12 @@ if __name__ == '__main__':
 		else:
 			doError('Invalid argument: '+arg, True)
 
+	# compatibility errors
+	if(sysvals.usecallgraph and sysvals.usedevsrc):
+		doError('-dev is not compatible with -f')
+	if(sysvals.usecallgraph and sysvals.useprocmon):
+		doError('-proc is not compatible with -f')
+
 	# callgraph size cannot exceed device size
 	if sysvals.mincglen < sysvals.mindevlen:
 		sysvals.mincglen = sysvals.mindevlen
@@ -4855,8 +5193,7 @@ if __name__ == '__main__':
 		elif(cmd == 'usbtopo'):
 			detectUSB()
 		elif(cmd == 'modes'):
-			modes = getModes()
-			print modes
+			print getModes()
 		elif(cmd == 'flist'):
 			sysvals.getFtraceFilterFunctions(True)
 		elif(cmd == 'flistall'):

tools/power/x86/intel_pstate_tracer/intel_pstate_tracer.py

+#!/usr/bin/python
+# -*- coding: utf-8 -*-
+#
+""" This utility can be used to debug and tune the performance of the
+intel_pstate driver. This utility can be used in two ways:
+- If there is Linux trace file with pstate_sample events enabled, then
+this utility can parse the trace file and generate performance plots.
+- If user has not specified a trace file as input via command line parameters,
+then this utility enables and collects trace data for a user specified interval
+and generates performance plots.
+
+Prerequisites:
+    Python version 2.7.x
+    gnuplot 5.0 or higher
+    gnuplot-py 1.8
+    (Most of the distributions have these required packages. They may be called
+     gnuplot-py, phython-gnuplot. )
+
+    HWP (Hardware P-States are disabled)
+    Kernel config for Linux trace is enabled
+
+    see print_help(): for Usage and Output details
+
+"""
+from __future__ import print_function
+from datetime import datetime
+import subprocess
+import os
+import time
+import re
+import sys
+import getopt
+import Gnuplot
+from numpy import *
+from decimal import *
+
+__author__ = "Srinivas Pandruvada"
+__copyright__ = " Copyright (c) 2017, Intel Corporation. "
+__license__ = "GPL version 2"
+
+
+MAX_CPUS = 256
+
+# Define the csv file columns
+C_COMM = 18
+C_GHZ = 17
+C_ELAPSED = 16
+C_SAMPLE = 15
+C_DURATION = 14
+C_LOAD = 13
+C_BOOST = 12
+C_FREQ = 11
+C_TSC = 10
+C_APERF = 9
+C_MPERF = 8
+C_TO = 7
+C_FROM = 6
+C_SCALED = 5
+C_CORE = 4
+C_USEC = 3
+C_SEC = 2
+C_CPU = 1
+
+global sample_num, last_sec_cpu, last_usec_cpu, start_time, testname
+
+# 11 digits covers uptime to 115 days
+getcontext().prec = 11
+
+sample_num =0
+last_sec_cpu = [0] * MAX_CPUS
+last_usec_cpu = [0] * MAX_CPUS
+
+def print_help():
+    print('intel_pstate_tracer.py:')
+    print('  Usage:')
+    print('    If the trace file is available, then to simply parse and plot, use (sudo not required):')
+    print('      ./intel_pstate_tracer.py [-c cpus] -t <trace_file> -n <test_name>')
+    print('    Or')
+    print('      ./intel_pstate_tracer.py [--cpu cpus] ---trace_file <trace_file> --name <test_name>')
+    print('    To generate trace file, parse and plot, use (sudo required):')
+    print('      sudo ./intel_pstate_tracer.py [-c cpus] -i <interval> -n <test_name>')
+    print('    Or')
+    print('      sudo ./intel_pstate_tracer.py [--cpu cpus] --interval <interval> --name <test_name>')
+    print('    Optional argument:')
+    print('      cpus:  comma separated list of CPUs')
+    print('  Output:')
+    print('    If not already present, creates a "results/test_name" folder in the current working directory with:')
+    print('      cpu.csv - comma seperated values file with trace contents and some additional calculations.')
+    print('      cpu???.csv - comma seperated values file for CPU number ???.')
+    print('      *.png - a variety of PNG format plot files created from the trace contents and the additional calculations.')
+    print('  Notes:')
+    print('    Avoid the use of _ (underscore) in test names, because in gnuplot it is a subscript directive.')
+    print('    Maximum number of CPUs is {0:d}. If there are more the script will abort with an error.'.format(MAX_CPUS))
+    print('    Off-line CPUs cause the script to list some warnings, and create some empty files. Use the CPU mask feature for a clean run.')
+    print('    Empty y range warnings for autoscaled plots can occur and can be ignored.')
+
+def plot_perf_busy_with_sample(cpu_index):
+    """ Plot method to per cpu information """
+
+    file_name = 'cpu{:0>3}.csv'.format(cpu_index)
+    if os.path.exists(file_name):
+        output_png = "cpu%03d_perf_busy_vs_samples.png" % cpu_index
+        g_plot = common_all_gnuplot_settings(output_png)
+        g_plot('set yrange [0:40]')
+        g_plot('set y2range [0:200]')
+        g_plot('set y2tics 0, 10')
+        g_plot('set title "{} : cpu perf busy vs. sample : CPU {:0>3} : {:%F %H:%M}"'.format(testname, cpu_index, datetime.now()))
+#       Override common
+        g_plot('set xlabel "Samples"')
+        g_plot('set ylabel "P-State"')
+        g_plot('set y2label "Scaled Busy/performance/io-busy(%)"')
+        set_4_plot_linestyles(g_plot)
+        g_plot('plot "' + file_name + '" using {:d}:{:d} with linespoints linestyle 1 axis x1y2 title "performance",\\'.format(C_SAMPLE, C_CORE))
+        g_plot('"' + file_name + '" using {:d}:{:d} with linespoints linestyle 2 axis x1y2 title "scaled-busy",\\'.format(C_SAMPLE, C_SCALED))
+        g_plot('"' + file_name + '" using {:d}:{:d} with linespoints linestyle 3 axis x1y2 title "io-boost",\\'.format(C_SAMPLE, C_BOOST))
+        g_plot('"' + file_name + '" using {:d}:{:d} with linespoints linestyle 4 axis x1y1 title "P-State"'.format(C_SAMPLE, C_TO))
+
+def plot_perf_busy(cpu_index):
+    """ Plot some per cpu information """
+
+    file_name = 'cpu{:0>3}.csv'.format(cpu_index)
+    if os.path.exists(file_name):
+        output_png = "cpu%03d_perf_busy.png" % cpu_index
+        g_plot = common_all_gnuplot_settings(output_png)
+        g_plot('set yrange [0:40]')
+        g_plot('set y2range [0:200]')
+        g_plot('set y2tics 0, 10')
+        g_plot('set title "{} : perf busy : CPU {:0>3} : {:%F %H:%M}"'.format(testname, cpu_index, datetime.now()))
+        g_plot('set ylabel "P-State"')
+        g_plot('set y2label "Scaled Busy/performance/io-busy(%)"')
+        set_4_plot_linestyles(g_plot)
+        g_plot('plot "' + file_name + '" using {:d}:{:d} with linespoints linestyle 1 axis x1y2 title "performance",\\'.format(C_ELAPSED, C_CORE))
+        g_plot('"' + file_name + '" using {:d}:{:d} with linespoints linestyle 2 axis x1y2 title "scaled-busy",\\'.format(C_ELAPSED, C_SCALED))
+        g_plot('"' + file_name + '" using {:d}:{:d} with linespoints linestyle 3 axis x1y2 title "io-boost",\\'.format(C_ELAPSED, C_BOOST))
+        g_plot('"' + file_name + '" using {:d}:{:d} with linespoints linestyle 4 axis x1y1 title "P-State"'.format(C_ELAPSED, C_TO))
+
+def plot_durations(cpu_index):
+    """ Plot per cpu durations """
+
+    file_name = 'cpu{:0>3}.csv'.format(cpu_index)
+    if os.path.exists(file_name):
+        output_png = "cpu%03d_durations.png" % cpu_index
+        g_plot = common_all_gnuplot_settings(output_png)
+#       Should autoscale be used here? Should seconds be used here?
+        g_plot('set yrange [0:5000]')
+        g_plot('set ytics 0, 500')
+        g_plot('set title "{} : durations : CPU {:0>3} : {:%F %H:%M}"'.format(testname, cpu_index, datetime.now()))
+        g_plot('set ylabel "Timer Duration (MilliSeconds)"')
+#       override common
+        g_plot('set key off')
+        set_4_plot_linestyles(g_plot)
+        g_plot('plot "' + file_name + '" using {:d}:{:d} with linespoints linestyle 1 axis x1y1'.format(C_ELAPSED, C_DURATION))
+
+def plot_loads(cpu_index):
+    """ Plot per cpu loads """
+
+    file_name = 'cpu{:0>3}.csv'.format(cpu_index)
+    if os.path.exists(file_name):
+        output_png = "cpu%03d_loads.png" % cpu_index
+        g_plot = common_all_gnuplot_settings(output_png)
+        g_plot('set yrange [0:100]')
+        g_plot('set ytics 0, 10')
+        g_plot('set title "{} : loads : CPU {:0>3} : {:%F %H:%M}"'.format(testname, cpu_index, datetime.now()))
+        g_plot('set ylabel "CPU load (percent)"')
+#       override common
+        g_plot('set key off')
+        set_4_plot_linestyles(g_plot)
+        g_plot('plot "' + file_name + '" using {:d}:{:d} with linespoints linestyle 1 axis x1y1'.format(C_ELAPSED, C_LOAD))
+
+def plot_pstate_cpu_with_sample():
+    """ Plot all cpu information """
+
+    if os.path.exists('cpu.csv'):
+        output_png = 'all_cpu_pstates_vs_samples.png'
+        g_plot = common_all_gnuplot_settings(output_png)
+        g_plot('set yrange [0:40]')
+#       override common
+        g_plot('set xlabel "Samples"')
+        g_plot('set ylabel "P-State"')
+        g_plot('set title "{} : cpu pstate vs. sample : {:%F %H:%M}"'.format(testname, datetime.now()))
+        title_list = subprocess.check_output('ls cpu???.csv | sed -e \'s/.csv//\'',shell=True).replace('\n', ' ')
+        plot_str = "plot for [i in title_list] i.'.csv' using {:d}:{:d} pt 7 ps 1 title i".format(C_SAMPLE, C_TO)
+        g_plot('title_list = "{}"'.format(title_list))
+        g_plot(plot_str)
+
+def plot_pstate_cpu():
+    """ Plot all cpu information from csv files """
+
+    output_png = 'all_cpu_pstates.png'
+    g_plot = common_all_gnuplot_settings(output_png)
+    g_plot('set yrange [0:40]')
+    g_plot('set ylabel "P-State"')
+    g_plot('set title "{} : cpu pstates : {:%F %H:%M}"'.format(testname, datetime.now()))
+
+#    the following command is really cool, but doesn't work with the CPU masking option because it aborts on the first missing file.
+#    plot_str = 'plot for [i=0:*] file=sprintf("cpu%03d.csv",i) title_s=sprintf("cpu%03d",i) file using 16:7 pt 7 ps 1 title title_s'
+#
+    title_list = subprocess.check_output('ls cpu???.csv | sed -e \'s/.csv//\'',shell=True).replace('\n', ' ')
+    plot_str = "plot for [i in title_list] i.'.csv' using {:d}:{:d} pt 7 ps 1 title i".format(C_ELAPSED, C_TO)
+    g_plot('title_list = "{}"'.format(title_list))
+    g_plot(plot_str)
+
+def plot_load_cpu():
+    """ Plot all cpu loads """
+
+    output_png = 'all_cpu_loads.png'
+    g_plot = common_all_gnuplot_settings(output_png)
+    g_plot('set yrange [0:100]')
+    g_plot('set ylabel "CPU load (percent)"')
+    g_plot('set title "{} : cpu loads : {:%F %H:%M}"'.format(testname, datetime.now()))
+
+    title_list = subprocess.check_output('ls cpu???.csv | sed -e \'s/.csv//\'',shell=True).replace('\n', ' ')
+    plot_str = "plot for [i in title_list] i.'.csv' using {:d}:{:d} pt 7 ps 1 title i".format(C_ELAPSED, C_LOAD)
+    g_plot('title_list = "{}"'.format(title_list))
+    g_plot(plot_str)
+
+def plot_frequency_cpu():
+    """ Plot all cpu frequencies """
+
+    output_png = 'all_cpu_frequencies.png'
+    g_plot = common_all_gnuplot_settings(output_png)
+    g_plot('set yrange [0:4]')
+    g_plot('set ylabel "CPU Frequency (GHz)"')
+    g_plot('set title "{} : cpu frequencies : {:%F %H:%M}"'.format(testname, datetime.now()))
+
+    title_list = subprocess.check_output('ls cpu???.csv | sed -e \'s/.csv//\'',shell=True).replace('\n', ' ')
+    plot_str = "plot for [i in title_list] i.'.csv' using {:d}:{:d} pt 7 ps 1 title i".format(C_ELAPSED, C_FREQ)
+    g_plot('title_list = "{}"'.format(title_list))
+    g_plot(plot_str)
+
+def plot_duration_cpu():
+    """ Plot all cpu durations """
+
+    output_png = 'all_cpu_durations.png'
+    g_plot = common_all_gnuplot_settings(output_png)
+    g_plot('set yrange [0:5000]')
+    g_plot('set ytics 0, 500')
+    g_plot('set ylabel "Timer Duration (MilliSeconds)"')
+    g_plot('set title "{} : cpu durations : {:%F %H:%M}"'.format(testname, datetime.now()))
+
+    title_list = subprocess.check_output('ls cpu???.csv | sed -e \'s/.csv//\'',shell=True).replace('\n', ' ')
+    plot_str = "plot for [i in title_list] i.'.csv' using {:d}:{:d} pt 7 ps 1 title i".format(C_ELAPSED, C_DURATION)
+    g_plot('title_list = "{}"'.format(title_list))
+    g_plot(plot_str)
+
+def plot_scaled_cpu():
+    """ Plot all cpu scaled busy """
+
+    output_png = 'all_cpu_scaled.png'
+    g_plot = common_all_gnuplot_settings(output_png)
+#   autoscale this one, no set y range
+    g_plot('set ylabel "Scaled Busy (Unitless)"')
+    g_plot('set title "{} : cpu scaled busy : {:%F %H:%M}"'.format(testname, datetime.now()))
+
+    title_list = subprocess.check_output('ls cpu???.csv | sed -e \'s/.csv//\'',shell=True).replace('\n', ' ')
+    plot_str = "plot for [i in title_list] i.'.csv' using {:d}:{:d} pt 7 ps 1 title i".format(C_ELAPSED, C_SCALED)
+    g_plot('title_list = "{}"'.format(title_list))
+    g_plot(plot_str)
+
+def plot_boost_cpu():
+    """ Plot all cpu IO Boosts """
+
+    output_png = 'all_cpu_boost.png'
+    g_plot = common_all_gnuplot_settings(output_png)
+    g_plot('set yrange [0:100]')
+    g_plot('set ylabel "CPU IO Boost (percent)"')
+    g_plot('set title "{} : cpu io boost : {:%F %H:%M}"'.format(testname, datetime.now()))
+
+    title_list = subprocess.check_output('ls cpu???.csv | sed -e \'s/.csv//\'',shell=True).replace('\n', ' ')
+    plot_str = "plot for [i in title_list] i.'.csv' using {:d}:{:d} pt 7 ps 1 title i".format(C_ELAPSED, C_BOOST)
+    g_plot('title_list = "{}"'.format(title_list))
+    g_plot(plot_str)
+
+def plot_ghz_cpu():
+    """ Plot all cpu tsc ghz """
+
+    output_png = 'all_cpu_ghz.png'
+    g_plot = common_all_gnuplot_settings(output_png)
+#   autoscale this one, no set y range
+    g_plot('set ylabel "TSC Frequency (GHz)"')
+    g_plot('set title "{} : cpu TSC Frequencies (Sanity check calculation) : {:%F %H:%M}"'.format(testname, datetime.now()))
+
+    title_list = subprocess.check_output('ls cpu???.csv | sed -e \'s/.csv//\'',shell=True).replace('\n', ' ')
+    plot_str = "plot for [i in title_list] i.'.csv' using {:d}:{:d} pt 7 ps 1 title i".format(C_ELAPSED, C_GHZ)
+    g_plot('title_list = "{}"'.format(title_list))
+    g_plot(plot_str)
+
+def common_all_gnuplot_settings(output_png):
+    """ common gnuplot settings for multiple CPUs one one graph. """
+
+    g_plot = common_gnuplot_settings()
+    g_plot('set output "' + output_png + '"')
+    return(g_plot)
+
+def common_gnuplot_settings():
+    """ common gnuplot settings. """
+
+    g_plot = Gnuplot.Gnuplot(persist=1)
+#   The following line is for rigor only. It seems to be assumed for .csv files
+    g_plot('set datafile separator \",\"')
+    g_plot('set ytics nomirror')
+    g_plot('set xtics nomirror')
+    g_plot('set xtics font ", 10"')
+    g_plot('set ytics font ", 10"')
+    g_plot('set tics out scale 1.0')
+    g_plot('set grid')
+    g_plot('set key out horiz')
+    g_plot('set key bot center')
+    g_plot('set key samplen 2 spacing .8 font ", 9"')
+    g_plot('set term png size 1200, 600')
+    g_plot('set title font ", 11"')
+    g_plot('set ylabel font ", 10"')
+    g_plot('set xlabel font ", 10"')
+    g_plot('set xlabel offset 0, 0.5')
+    g_plot('set xlabel "Elapsed Time (Seconds)"')
+    return(g_plot)
+
+def set_4_plot_linestyles(g_plot):
+    """ set the linestyles used for 4 plots in 1 graphs. """
+
+    g_plot('set style line 1 linetype 1 linecolor rgb "green" pointtype -1')
+    g_plot('set style line 2 linetype 1 linecolor rgb "red" pointtype -1')
+    g_plot('set style line 3 linetype 1 linecolor rgb "purple" pointtype -1')
+    g_plot('set style line 4 linetype 1 linecolor rgb "blue" pointtype -1')
+
+def store_csv(cpu_int, time_pre_dec, time_post_dec, core_busy, scaled, _from, _to, mperf, aperf, tsc, freq_ghz, io_boost, common_comm, load, duration_ms, sample_num, elapsed_time, tsc_ghz):
+    """ Store master csv file information """
+
+    global graph_data_present
+
+    if cpu_mask[cpu_int] == 0:
+        return
+
+    try:
+        f_handle = open('cpu.csv', 'a')
+        string_buffer = "CPU_%03u, %05u, %06u, %u, %u, %u, %u, %u, %u, %u, %.4f, %u, %.2f, %.3f, %u, %.3f, %.3f, %s\n" % (cpu_int, int(time_pre_dec), int(time_post_dec), int(core_busy), int(scaled), int(_from), int(_to), int(mperf), int(aperf), int(tsc), freq_ghz, int(io_boost), load, duration_ms, sample_num, elapsed_time, tsc_ghz, common_comm)
+        f_handle.write(string_buffer);
+        f_handle.close()
+    except:
+        print('IO error cpu.csv')
+        return
+
+    graph_data_present = True;
+
+def split_csv():
+    """ seperate the all csv file into per CPU csv files. """
+
+    global current_max_cpu
+
+    if os.path.exists('cpu.csv'):
+        for index in range(0, current_max_cpu + 1):
+            if cpu_mask[int(index)] != 0:
+                os.system('grep -m 1 common_cpu cpu.csv > cpu{:0>3}.csv'.format(index))
+                os.system('grep CPU_{:0>3} cpu.csv >> cpu{:0>3}.csv'.format(index, index))
+
+def cleanup_data_files():
+    """ clean up existing data files """
+
+    if os.path.exists('cpu.csv'):
+        os.remove('cpu.csv')
+    f_handle = open('cpu.csv', 'a')
+    f_handle.write('common_cpu, common_secs, common_usecs, core_busy, scaled_busy, from, to, mperf, aperf, tsc, freq, boost, load, duration_ms, sample_num, elapsed_time, tsc_ghz, common_comm')
+    f_handle.write('\n')
+    f_handle.close()
+
+def clear_trace_file():
+    """ Clear trace file """
+
+    try:
+        f_handle = open('/sys/kernel/debug/tracing/trace', 'w')
+        f_handle.close()
+    except:
+        print('IO error clearing trace file ')
+        quit()
+
+def enable_trace():
+    """ Enable trace """
+
+    try:
+       open('/sys/kernel/debug/tracing/events/power/pstate_sample/enable'
+                 , 'w').write("1")
+    except:
+        print('IO error enabling trace ')
+        quit()
+
+def disable_trace():
+    """ Disable trace """
+
+    try:
+       open('/sys/kernel/debug/tracing/events/power/pstate_sample/enable'
+                 , 'w').write("0")
+    except:
+        print('IO error disabling trace ')
+        quit()
+
+def set_trace_buffer_size():
+    """ Set trace buffer size """
+
+    try:
+       open('/sys/kernel/debug/tracing/buffer_size_kb'
+                 , 'w').write("10240")
+    except:
+        print('IO error setting trace buffer size ')
+        quit()
+
+def read_trace_data(filename):
+    """ Read and parse trace data """
+
+    global current_max_cpu
+    global sample_num, last_sec_cpu, last_usec_cpu, start_time
+
+    try:
+        data = open(filename, 'r').read()
+    except:
+        print('Error opening ', filename)
+        quit()
+
+    for line in data.splitlines():
+        search_obj = \
+            re.search(r'(^(.*?)\[)((\d+)[^\]])(.*?)(\d+)([.])(\d+)(.*?core_busy=)(\d+)(.*?scaled=)(\d+)(.*?from=)(\d+)(.*?to=)(\d+)(.*?mperf=)(\d+)(.*?aperf=)(\d+)(.*?tsc=)(\d+)(.*?freq=)(\d+)'
+                      , line)
+
+        if search_obj:
+            cpu = search_obj.group(3)
+            cpu_int = int(cpu)
+            cpu = str(cpu_int)
+
+            time_pre_dec = search_obj.group(6)
+            time_post_dec = search_obj.group(8)
+            core_busy = search_obj.group(10)
+            scaled = search_obj.group(12)
+            _from = search_obj.group(14)
+            _to = search_obj.group(16)
+            mperf = search_obj.group(18)
+            aperf = search_obj.group(20)
+            tsc = search_obj.group(22)
+            freq = search_obj.group(24)
+            common_comm = search_obj.group(2).replace(' ', '')
+
+            # Not all kernel versions have io_boost field
+            io_boost = '0'
+            search_obj = re.search(r'.*?io_boost=(\d+)', line)
+            if search_obj:
+                io_boost = search_obj.group(1)
+
+            if sample_num == 0 :
+                start_time = Decimal(time_pre_dec) + Decimal(time_post_dec) / Decimal(1000000)
+            sample_num += 1
+
+            if last_sec_cpu[cpu_int] == 0 :
+                last_sec_cpu[cpu_int] = time_pre_dec
+                last_usec_cpu[cpu_int] = time_post_dec
+            else :
+                duration_us = (int(time_pre_dec) - int(last_sec_cpu[cpu_int])) * 1000000 + (int(time_post_dec) - int(last_usec_cpu[cpu_int]))
+                duration_ms = Decimal(duration_us) / Decimal(1000)
+                last_sec_cpu[cpu_int] = time_pre_dec
+                last_usec_cpu[cpu_int] = time_post_dec
+                elapsed_time = Decimal(time_pre_dec) + Decimal(time_post_dec) / Decimal(1000000) - start_time
+                load = Decimal(int(mperf)*100)/ Decimal(tsc)
+                freq_ghz = Decimal(freq)/Decimal(1000000)
+#               Sanity check calculation, typically anomalies indicate missed samples
+#               However, check for 0 (should never occur)
+                tsc_ghz = Decimal(0)
+                if duration_ms != Decimal(0) :
+                    tsc_ghz = Decimal(tsc)/duration_ms/Decimal(1000000)
+                store_csv(cpu_int, time_pre_dec, time_post_dec, core_busy, scaled, _from, _to, mperf, aperf, tsc, freq_ghz, io_boost, common_comm, load, duration_ms, sample_num, elapsed_time, tsc_ghz)
+
+            if cpu_int > current_max_cpu:
+                current_max_cpu = cpu_int
+# End of for each trace line loop
+# Now seperate the main overall csv file into per CPU csv files.
+    split_csv()
+
+interval = ""
+filename = ""
+cpu_list = ""
+testname = ""
+graph_data_present = False;
+
+valid1 = False
+valid2 = False
+
+cpu_mask = zeros((MAX_CPUS,), dtype=int)
+
+try:
+    opts, args = getopt.getopt(sys.argv[1:],"ht:i:c:n:",["help","trace_file=","interval=","cpu=","name="])
+except getopt.GetoptError:
+    print_help()
+    sys.exit(2)
+for opt, arg in opts:
+    if opt == '-h':
+        print()
+        sys.exit()
+    elif opt in ("-t", "--trace_file"):
+        valid1 = True
+        location = os.path.realpath(os.path.join(os.getcwd(), os.path.dirname(__file__)))
+        filename = os.path.join(location, arg)
+    elif opt in ("-i", "--interval"):
+        valid1 = True
+        interval = arg
+    elif opt in ("-c", "--cpu"):
+        cpu_list = arg
+    elif opt in ("-n", "--name"):
+        valid2 = True
+        testname = arg
+
+if not (valid1 and valid2):
+    print_help()
+    sys.exit()
+
+if cpu_list:
+    for p in re.split("[,]", cpu_list):
+        if int(p) < MAX_CPUS :
+            cpu_mask[int(p)] = 1
+else:
+    for i in range (0, MAX_CPUS):
+        cpu_mask[i] = 1
+
+if not os.path.exists('results'):
+    os.mkdir('results')
+
+os.chdir('results')
+if os.path.exists(testname):
+    print('The test name directory already exists. Please provide a unique test name. Test re-run not supported, yet.')
+    sys.exit()
+os.mkdir(testname)
+os.chdir(testname)
+
+# Temporary (or perhaps not)
+cur_version = sys.version_info
+print('python version (should be >= 2.7):')
+print(cur_version)
+
+# Left as "cleanup" for potential future re-run ability.
+cleanup_data_files()
+
+if interval:
+    filename = "/sys/kernel/debug/tracing/trace"
+    clear_trace_file()
+    set_trace_buffer_size()
+    enable_trace()
+    print('Sleeping for ', interval, 'seconds')
+    time.sleep(int(interval))
+    disable_trace()
+
+current_max_cpu = 0
+
+read_trace_data(filename)
+
+if graph_data_present == False:
+    print('No valid data to plot')
+    sys.exit(2)
+
+for cpu_no in range(0, current_max_cpu + 1):
+    plot_perf_busy_with_sample(cpu_no)
+    plot_perf_busy(cpu_no)
+    plot_durations(cpu_no)
+    plot_loads(cpu_no)
+
+plot_pstate_cpu_with_sample()
+plot_pstate_cpu()
+plot_load_cpu()
+plot_frequency_cpu()
+plot_duration_cpu()
+plot_scaled_cpu()
+plot_boost_cpu()
+plot_ghz_cpu()
+
+os.chdir('../../')