From 3282e6b8f89eaeaf4915ee6cc57bcf06d1d6cead Mon Sep 17 00:00:00 2001 From: Sudeep Holla Date: Wed, 4 May 2016 17:50:59 +0100 Subject: [PATCH 01/42] x86/topology: Remove redundant ENABLE_TOPO_DEFINES Commit c8e56d20f2d1 ("x86: Kill CONFIG_X86_HT") removed CONFIG_X86_HT and defined ENABLE_TOPO_DEFINES always if CONFIG_SMP, which makes ENABLE_TOPO_DEFINES redundant. This patch removes the redundant ENABLE_TOPO_DEFINES and instead uses CONFIG_SMP directly Signed-off-by: Sudeep Holla Acked-by: Borislav Petkov Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1462380659-5968-1-git-send-email-sudeep.holla@arm.com Signed-off-by: Ingo Molnar --- arch/x86/include/asm/topology.h | 12 +----------- 1 file changed, 1 insertion(+), 11 deletions(-) diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h index 7f991bd5031b..c9a4ed73aef4 100644 --- a/arch/x86/include/asm/topology.h +++ b/arch/x86/include/asm/topology.h @@ -25,16 +25,6 @@ #ifndef _ASM_X86_TOPOLOGY_H #define _ASM_X86_TOPOLOGY_H -#ifdef CONFIG_X86_32 -# ifdef CONFIG_SMP -# define ENABLE_TOPO_DEFINES -# endif -#else -# ifdef CONFIG_SMP -# define ENABLE_TOPO_DEFINES -# endif -#endif - /* * to preserve the visibility of NUMA_NO_NODE definition, * moved to there from here. May be used independent of @@ -123,7 +113,7 @@ extern const struct cpumask *cpu_coregroup_mask(int cpu); #define topology_physical_package_id(cpu) (cpu_data(cpu).phys_proc_id) #define topology_core_id(cpu) (cpu_data(cpu).cpu_core_id) -#ifdef ENABLE_TOPO_DEFINES +#ifdef CONFIG_SMP #define topology_core_cpumask(cpu) (per_cpu(cpu_core_map, cpu)) #define topology_sibling_cpumask(cpu) (per_cpu(cpu_sibling_map, cpu)) From f0133acc7d4835cfbb86393b7d2a4fba7519585b Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Sun, 8 May 2016 20:58:40 +0200 Subject: [PATCH 02/42] x86/cpu: Correct comments and messages in P4 erratum 037 handling code Remove the linebreak in the conditional and s/errata/erratum/ as the singular is "erratum". No functionality change. Signed-off-by: Borislav Petkov Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1462733920-7224-1-git-send-email-bp@alien8.de Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/intel.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c index f71a34944b56..5354080f76c3 100644 --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -263,15 +263,14 @@ static void intel_workarounds(struct cpuinfo_x86 *c) } /* - * P4 Xeon errata 037 workaround. + * P4 Xeon erratum 037 workaround. * Hardware prefetcher may cause stale data to be loaded into the cache. */ if ((c->x86 == 15) && (c->x86_model == 1) && (c->x86_mask == 1)) { if (msr_set_bit(MSR_IA32_MISC_ENABLE, - MSR_IA32_MISC_ENABLE_PREFETCH_DISABLE_BIT) - > 0) { + MSR_IA32_MISC_ENABLE_PREFETCH_DISABLE_BIT) > 0) { pr_info("CPU: C0 stepping P4 Xeon detected.\n"); - pr_info("CPU: Disabling hardware prefetching (Errata 037)\n"); + pr_info("CPU: Disabling hardware prefetching (Erratum 037)\n"); } } From d6ff44d647a9c63911983da9979961fde306b3aa Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Sat, 14 May 2016 00:59:27 +0200 Subject: [PATCH 03/42] cpufreq: governor: CPUFREQ_GOV_LIMITS never fails None of the cpufreq governors currently in the tree will ever fail an invocation of the ->governor() callback with the event argument equal to CPUFREQ_GOV_LIMITS (unless invoked with incorrect arguments which doesn't matter anyway) and had it ever failed, the result of it wouldn't have been very clean. For this reason, rearrange the code in the core to ignore the return value of cpufreq_governor() when called with event equal to CPUFREQ_GOV_LIMITS. Signed-off-by: Rafael J. Wysocki Acked-by: Viresh Kumar --- drivers/cpufreq/cpufreq.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index 36bc11a106aa..e3e666109e21 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -2054,7 +2054,11 @@ static int cpufreq_start_governor(struct cpufreq_policy *policy) cpufreq_update_current_freq(policy); ret = cpufreq_governor(policy, CPUFREQ_GOV_START); - return ret ? ret : cpufreq_governor(policy, CPUFREQ_GOV_LIMITS); + if (ret) + return ret; + + cpufreq_governor(policy, CPUFREQ_GOV_LIMITS); + return 0; } int cpufreq_register_governor(struct cpufreq_governor *governor) @@ -2195,7 +2199,8 @@ static int cpufreq_set_policy(struct cpufreq_policy *policy, if (new_policy->governor == policy->governor) { pr_debug("cpufreq: governor limits update\n"); - return cpufreq_governor(policy, CPUFREQ_GOV_LIMITS); + cpufreq_governor(policy, CPUFREQ_GOV_LIMITS); + return 0; } pr_debug("governor switch\n"); From 9e8c0a899b2f89b8367d33668f6819892f0a00f2 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Sat, 14 May 2016 01:00:28 +0200 Subject: [PATCH 04/42] cpufreq: governor: Check transition latecy at init time only It is not necessary to check the governor's max_transition_latency attribute every time cpufreq_governor() runs, so check it only if the event argument is CPUFREQ_GOV_POLICY_INIT. Signed-off-by: Rafael J. Wysocki Acked-by: Viresh Kumar --- drivers/cpufreq/cpufreq.c | 25 +++++++++++++------------ 1 file changed, 13 insertions(+), 12 deletions(-) diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index e3e666109e21..a5f1d016d473 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -2011,23 +2011,24 @@ static int cpufreq_governor(struct cpufreq_policy *policy, unsigned int event) if (!policy->governor) return -EINVAL; - if (policy->governor->max_transition_latency && - policy->cpuinfo.transition_latency > - policy->governor->max_transition_latency) { - struct cpufreq_governor *gov = cpufreq_fallback_governor(); + if (event == CPUFREQ_GOV_POLICY_INIT) { + if (policy->governor->max_transition_latency && + policy->cpuinfo.transition_latency > + policy->governor->max_transition_latency) { + struct cpufreq_governor *gov = cpufreq_fallback_governor(); - if (gov) { - pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n", - policy->governor->name, gov->name); - policy->governor = gov; - } else { - return -EINVAL; + if (gov) { + pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n", + policy->governor->name, gov->name); + policy->governor = gov; + } else { + return -EINVAL; + } } - } - if (event == CPUFREQ_GOV_POLICY_INIT) if (!try_module_get(policy->governor->owner)) return -EINVAL; + } pr_debug("%s: for CPU %u, event %u\n", __func__, policy->cpu, event); From 16de72b9f4d07e393bea7b1221898b3be718a296 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Sat, 14 May 2016 01:01:08 +0200 Subject: [PATCH 05/42] cpufreq: governor: Simplify performance and powersave governors The performance and powersave cpufreq governors handle the CPUFREQ_GOV_START event in the same way as CPUFREQ_GOV_LIMITS. However, the cpufreq core always invokes cpufreq_governor() with the event argument equal to CPUFREQ_GOV_LIMITS right after invoking it with event equal to CPUFREQ_GOV_START. As a result, for both the governors in question, __cpufreq_driver_target() is executed twice in a row with the same arguments which is not useful. For this reason, simplify the performance and powersave governors to handle the CPUFREQ_GOV_LIMITS event only as that's going to be sufficient for the governor start too. Signed-off-by: Rafael J. Wysocki Acked-by: Viresh Kumar --- drivers/cpufreq/cpufreq_performance.c | 4 +--- drivers/cpufreq/cpufreq_powersave.c | 4 +--- 2 files changed, 2 insertions(+), 6 deletions(-) diff --git a/drivers/cpufreq/cpufreq_performance.c b/drivers/cpufreq/cpufreq_performance.c index af9f4b96f5a8..dd4dca3ab844 100644 --- a/drivers/cpufreq/cpufreq_performance.c +++ b/drivers/cpufreq/cpufreq_performance.c @@ -20,10 +20,8 @@ static int cpufreq_governor_performance(struct cpufreq_policy *policy, unsigned int event) { switch (event) { - case CPUFREQ_GOV_START: case CPUFREQ_GOV_LIMITS: - pr_debug("setting to %u kHz because of event %u\n", - policy->max, event); + pr_debug("setting to %u kHz\n", policy->max); __cpufreq_driver_target(policy, policy->max, CPUFREQ_RELATION_H); break; diff --git a/drivers/cpufreq/cpufreq_powersave.c b/drivers/cpufreq/cpufreq_powersave.c index b8b400232a74..691e573e46ee 100644 --- a/drivers/cpufreq/cpufreq_powersave.c +++ b/drivers/cpufreq/cpufreq_powersave.c @@ -20,10 +20,8 @@ static int cpufreq_governor_powersave(struct cpufreq_policy *policy, unsigned int event) { switch (event) { - case CPUFREQ_GOV_START: case CPUFREQ_GOV_LIMITS: - pr_debug("setting to %u kHz because of event %u\n", - policy->min, event); + pr_debug("setting to %u kHz\n", policy->min); __cpufreq_driver_target(policy, policy->min, CPUFREQ_RELATION_L); break; From a92604b419f47e1c5098632742d8e031f6e8fab1 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Sat, 14 May 2016 01:01:46 +0200 Subject: [PATCH 06/42] cpufreq: Split cpufreq_governor() into simpler functions The cpufreq_governor() routine is used by the cpufreq core to invoke the current governor's ->governor() callback with appropriate arguments and do some housekeeping related to that. Unfortunately, the way it mixes different governor events in one code path makes it rather hard to follow the code. For this reason, split cpufreq_governor() into five simpler functions that each will handle just one specific governor event and put all of the code related to the given event into its own function. This change is a prerequisite for a redesign of the cpufreq governor API that will be done subsequently. Signed-off-by: Rafael J. Wysocki Acked-by: Viresh Kumar --- drivers/cpufreq/cpufreq.c | 113 +++++++++++++++++++++++--------------- 1 file changed, 70 insertions(+), 43 deletions(-) diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index a5f1d016d473..268566e40684 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -74,19 +74,12 @@ static inline bool has_target(void) } /* internal prototypes */ -static int cpufreq_governor(struct cpufreq_policy *policy, unsigned int event); static unsigned int __cpufreq_get(struct cpufreq_policy *policy); +static int cpufreq_init_governor(struct cpufreq_policy *policy); +static void cpufreq_exit_governor(struct cpufreq_policy *policy); static int cpufreq_start_governor(struct cpufreq_policy *policy); - -static inline void cpufreq_exit_governor(struct cpufreq_policy *policy) -{ - (void)cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT); -} - -static inline void cpufreq_stop_governor(struct cpufreq_policy *policy) -{ - (void)cpufreq_governor(policy, CPUFREQ_GOV_STOP); -} +static void cpufreq_stop_governor(struct cpufreq_policy *policy); +static void cpufreq_governor_limits(struct cpufreq_policy *policy); /** * Two notifier lists: the "policy" list is involved in the @@ -1997,7 +1990,7 @@ __weak struct cpufreq_governor *cpufreq_fallback_governor(void) return NULL; } -static int cpufreq_governor(struct cpufreq_policy *policy, unsigned int event) +static int cpufreq_init_governor(struct cpufreq_policy *policy) { int ret; @@ -2011,57 +2004,91 @@ static int cpufreq_governor(struct cpufreq_policy *policy, unsigned int event) if (!policy->governor) return -EINVAL; - if (event == CPUFREQ_GOV_POLICY_INIT) { - if (policy->governor->max_transition_latency && - policy->cpuinfo.transition_latency > - policy->governor->max_transition_latency) { - struct cpufreq_governor *gov = cpufreq_fallback_governor(); + if (policy->governor->max_transition_latency && + policy->cpuinfo.transition_latency > + policy->governor->max_transition_latency) { + struct cpufreq_governor *gov = cpufreq_fallback_governor(); - if (gov) { - pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n", - policy->governor->name, gov->name); - policy->governor = gov; - } else { - return -EINVAL; - } - } - - if (!try_module_get(policy->governor->owner)) + if (gov) { + pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n", + policy->governor->name, gov->name); + policy->governor = gov; + } else { return -EINVAL; + } } - pr_debug("%s: for CPU %u, event %u\n", __func__, policy->cpu, event); + if (!try_module_get(policy->governor->owner)) + return -EINVAL; - ret = policy->governor->governor(policy, event); + pr_debug("%s: for CPU %u\n", __func__, policy->cpu); - if (event == CPUFREQ_GOV_POLICY_INIT) { - if (ret) - module_put(policy->governor->owner); - else - policy->governor->initialized++; - } else if (event == CPUFREQ_GOV_POLICY_EXIT) { - policy->governor->initialized--; + ret = policy->governor->governor(policy, CPUFREQ_GOV_POLICY_INIT); + if (ret) { module_put(policy->governor->owner); + return ret; } - return ret; + policy->governor->initialized++; + return 0; +} + +static void cpufreq_exit_governor(struct cpufreq_policy *policy) +{ + if (cpufreq_suspended || !policy->governor) + return; + + pr_debug("%s: for CPU %u\n", __func__, policy->cpu); + + policy->governor->governor(policy, CPUFREQ_GOV_POLICY_EXIT); + + policy->governor->initialized--; + module_put(policy->governor->owner); } static int cpufreq_start_governor(struct cpufreq_policy *policy) { int ret; + if (cpufreq_suspended) + return 0; + + if (!policy->governor) + return -EINVAL; + + pr_debug("%s: for CPU %u\n", __func__, policy->cpu); + if (cpufreq_driver->get && !cpufreq_driver->setpolicy) cpufreq_update_current_freq(policy); - ret = cpufreq_governor(policy, CPUFREQ_GOV_START); + ret = policy->governor->governor(policy, CPUFREQ_GOV_START); if (ret) return ret; - cpufreq_governor(policy, CPUFREQ_GOV_LIMITS); + policy->governor->governor(policy, CPUFREQ_GOV_LIMITS); return 0; } +static void cpufreq_stop_governor(struct cpufreq_policy *policy) +{ + if (cpufreq_suspended || !policy->governor) + return; + + pr_debug("%s: for CPU %u\n", __func__, policy->cpu); + + policy->governor->governor(policy, CPUFREQ_GOV_STOP); +} + +static void cpufreq_governor_limits(struct cpufreq_policy *policy) +{ + if (cpufreq_suspended || !policy->governor) + return; + + pr_debug("%s: for CPU %u\n", __func__, policy->cpu); + + policy->governor->governor(policy, CPUFREQ_GOV_LIMITS); +} + int cpufreq_register_governor(struct cpufreq_governor *governor) { int err; @@ -2200,7 +2227,7 @@ static int cpufreq_set_policy(struct cpufreq_policy *policy, if (new_policy->governor == policy->governor) { pr_debug("cpufreq: governor limits update\n"); - cpufreq_governor(policy, CPUFREQ_GOV_LIMITS); + cpufreq_governor_limits(policy); return 0; } @@ -2216,7 +2243,7 @@ static int cpufreq_set_policy(struct cpufreq_policy *policy, /* start new governor */ policy->governor = new_policy->governor; - ret = cpufreq_governor(policy, CPUFREQ_GOV_POLICY_INIT); + ret = cpufreq_init_governor(policy); if (!ret) { ret = cpufreq_start_governor(policy); if (!ret) { @@ -2230,7 +2257,7 @@ static int cpufreq_set_policy(struct cpufreq_policy *policy, pr_debug("starting governor %s failed\n", policy->governor->name); if (old_gov) { policy->governor = old_gov; - if (cpufreq_governor(policy, CPUFREQ_GOV_POLICY_INIT)) + if (cpufreq_init_governor(policy)) policy->governor = NULL; else cpufreq_start_governor(policy); @@ -2328,7 +2355,7 @@ static int cpufreq_boost_set_sw(int state) down_write(&policy->rwsem); policy->user_policy.max = policy->max; - cpufreq_governor(policy, CPUFREQ_GOV_LIMITS); + cpufreq_governor_limits(policy); up_write(&policy->rwsem); } } From e788892ba3cc71d385b75895f7a375fbc659ce86 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Thu, 2 Jun 2016 23:24:15 +0200 Subject: [PATCH 07/42] cpufreq: governor: Get rid of governor events The design of the cpufreq governor API is not very straightforward, as struct cpufreq_governor provides only one callback to be invoked from different code paths for different purposes. The purpose it is invoked for is determined by its second "event" argument, causing it to act as a "callback multiplexer" of sorts. Unfortunately, that leads to extra complexity in governors, some of which implement the ->governor() callback as a switch statement that simply checks the event argument and invokes a separate function to handle that specific event. That extra complexity can be eliminated by replacing the all-purpose ->governor() callback with a family of callbacks to carry out specific governor operations: initialization and exit, start and stop and policy limits updates. That also turns out to reduce the code size too, so do it. Signed-off-by: Rafael J. Wysocki Acked-by: Viresh Kumar --- .../platforms/cell/cpufreq_spudemand.c | 86 ++++++++--------- drivers/cpufreq/cpufreq.c | 31 +++--- drivers/cpufreq/cpufreq_conservative.c | 7 +- drivers/cpufreq/cpufreq_governor.c | 39 ++------ drivers/cpufreq/cpufreq_governor.h | 20 +++- drivers/cpufreq/cpufreq_ondemand.c | 7 +- drivers/cpufreq/cpufreq_performance.c | 17 +--- drivers/cpufreq/cpufreq_powersave.c | 17 +--- drivers/cpufreq/cpufreq_userspace.c | 96 +++++++++---------- include/linux/cpufreq.h | 14 +-- kernel/sched/cpufreq_schedutil.c | 34 ++----- 11 files changed, 161 insertions(+), 207 deletions(-) diff --git a/arch/powerpc/platforms/cell/cpufreq_spudemand.c b/arch/powerpc/platforms/cell/cpufreq_spudemand.c index 82607d621aca..88301e53f085 100644 --- a/arch/powerpc/platforms/cell/cpufreq_spudemand.c +++ b/arch/powerpc/platforms/cell/cpufreq_spudemand.c @@ -85,61 +85,57 @@ static void spu_gov_cancel_work(struct spu_gov_info_struct *info) cancel_delayed_work_sync(&info->work); } -static int spu_gov_govern(struct cpufreq_policy *policy, unsigned int event) +static int spu_gov_start(struct cpufreq_policy *policy) { unsigned int cpu = policy->cpu; - struct spu_gov_info_struct *info, *affected_info; + struct spu_gov_info_struct *info = &per_cpu(spu_gov_info, cpu); + struct spu_gov_info_struct *affected_info; int i; - int ret = 0; - info = &per_cpu(spu_gov_info, cpu); - - switch (event) { - case CPUFREQ_GOV_START: - if (!cpu_online(cpu)) { - printk(KERN_ERR "cpu %d is not online\n", cpu); - ret = -EINVAL; - break; - } - - if (!policy->cur) { - printk(KERN_ERR "no cpu specified in policy\n"); - ret = -EINVAL; - break; - } - - /* initialize spu_gov_info for all affected cpus */ - for_each_cpu(i, policy->cpus) { - affected_info = &per_cpu(spu_gov_info, i); - affected_info->policy = policy; - } - - info->poll_int = POLL_TIME; - - /* setup timer */ - spu_gov_init_work(info); - - break; - - case CPUFREQ_GOV_STOP: - /* cancel timer */ - spu_gov_cancel_work(info); - - /* clean spu_gov_info for all affected cpus */ - for_each_cpu (i, policy->cpus) { - info = &per_cpu(spu_gov_info, i); - info->policy = NULL; - } - - break; + if (!cpu_online(cpu)) { + printk(KERN_ERR "cpu %d is not online\n", cpu); + return -EINVAL; } - return ret; + if (!policy->cur) { + printk(KERN_ERR "no cpu specified in policy\n"); + return -EINVAL; + } + + /* initialize spu_gov_info for all affected cpus */ + for_each_cpu(i, policy->cpus) { + affected_info = &per_cpu(spu_gov_info, i); + affected_info->policy = policy; + } + + info->poll_int = POLL_TIME; + + /* setup timer */ + spu_gov_init_work(info); + + return 0; +} + +static void spu_gov_stop(struct cpufreq_policy *policy) +{ + unsigned int cpu = policy->cpu; + struct spu_gov_info_struct *info = &per_cpu(spu_gov_info, cpu); + int i; + + /* cancel timer */ + spu_gov_cancel_work(info); + + /* clean spu_gov_info for all affected cpus */ + for_each_cpu (i, policy->cpus) { + info = &per_cpu(spu_gov_info, i); + info->policy = NULL; + } } static struct cpufreq_governor spu_governor = { .name = "spudemand", - .governor = spu_gov_govern, + .start = spu_gov_start, + .stop = spu_gov_stop, .owner = THIS_MODULE, }; diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index 268566e40684..d98ff688b2f1 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -2023,10 +2023,12 @@ static int cpufreq_init_governor(struct cpufreq_policy *policy) pr_debug("%s: for CPU %u\n", __func__, policy->cpu); - ret = policy->governor->governor(policy, CPUFREQ_GOV_POLICY_INIT); - if (ret) { - module_put(policy->governor->owner); - return ret; + if (policy->governor->init) { + ret = policy->governor->init(policy); + if (ret) { + module_put(policy->governor->owner); + return ret; + } } policy->governor->initialized++; @@ -2040,7 +2042,8 @@ static void cpufreq_exit_governor(struct cpufreq_policy *policy) pr_debug("%s: for CPU %u\n", __func__, policy->cpu); - policy->governor->governor(policy, CPUFREQ_GOV_POLICY_EXIT); + if (policy->governor->exit) + policy->governor->exit(policy); policy->governor->initialized--; module_put(policy->governor->owner); @@ -2061,11 +2064,15 @@ static int cpufreq_start_governor(struct cpufreq_policy *policy) if (cpufreq_driver->get && !cpufreq_driver->setpolicy) cpufreq_update_current_freq(policy); - ret = policy->governor->governor(policy, CPUFREQ_GOV_START); - if (ret) - return ret; + if (policy->governor->start) { + ret = policy->governor->start(policy); + if (ret) + return ret; + } + + if (policy->governor->limits) + policy->governor->limits(policy); - policy->governor->governor(policy, CPUFREQ_GOV_LIMITS); return 0; } @@ -2076,7 +2083,8 @@ static void cpufreq_stop_governor(struct cpufreq_policy *policy) pr_debug("%s: for CPU %u\n", __func__, policy->cpu); - policy->governor->governor(policy, CPUFREQ_GOV_STOP); + if (policy->governor->stop) + policy->governor->stop(policy); } static void cpufreq_governor_limits(struct cpufreq_policy *policy) @@ -2086,7 +2094,8 @@ static void cpufreq_governor_limits(struct cpufreq_policy *policy) pr_debug("%s: for CPU %u\n", __func__, policy->cpu); - policy->governor->governor(policy, CPUFREQ_GOV_LIMITS); + if (policy->governor->limits) + policy->governor->limits(policy); } int cpufreq_register_governor(struct cpufreq_governor *governor) diff --git a/drivers/cpufreq/cpufreq_conservative.c b/drivers/cpufreq/cpufreq_conservative.c index 316df247e00d..2568508fca38 100644 --- a/drivers/cpufreq/cpufreq_conservative.c +++ b/drivers/cpufreq/cpufreq_conservative.c @@ -313,12 +313,7 @@ static void cs_start(struct cpufreq_policy *policy) } static struct dbs_governor cs_dbs_gov = { - .gov = { - .name = "conservative", - .governor = cpufreq_governor_dbs, - .max_transition_latency = TRANSITION_LATENCY_LIMIT, - .owner = THIS_MODULE, - }, + .gov = CPUFREQ_DBS_GOVERNOR_INITIALIZER("conservative"), .kobj_type = { .default_attrs = cs_attributes }, .gov_dbs_timer = cs_dbs_timer, .alloc = cs_alloc, diff --git a/drivers/cpufreq/cpufreq_governor.c b/drivers/cpufreq/cpufreq_governor.c index be498d56dd69..ca9927c15a71 100644 --- a/drivers/cpufreq/cpufreq_governor.c +++ b/drivers/cpufreq/cpufreq_governor.c @@ -389,7 +389,7 @@ static void free_policy_dbs_info(struct policy_dbs_info *policy_dbs, gov->free(policy_dbs); } -static int cpufreq_governor_init(struct cpufreq_policy *policy) +int cpufreq_dbs_governor_init(struct cpufreq_policy *policy) { struct dbs_governor *gov = dbs_governor_of(policy); struct dbs_data *dbs_data; @@ -474,8 +474,9 @@ static int cpufreq_governor_init(struct cpufreq_policy *policy) mutex_unlock(&gov_dbs_data_mutex); return ret; } +EXPORT_SYMBOL_GPL(cpufreq_dbs_governor_init); -static int cpufreq_governor_exit(struct cpufreq_policy *policy) +void cpufreq_dbs_governor_exit(struct cpufreq_policy *policy) { struct dbs_governor *gov = dbs_governor_of(policy); struct policy_dbs_info *policy_dbs = policy->governor_data; @@ -500,10 +501,10 @@ static int cpufreq_governor_exit(struct cpufreq_policy *policy) free_policy_dbs_info(policy_dbs, gov); mutex_unlock(&gov_dbs_data_mutex); - return 0; } +EXPORT_SYMBOL_GPL(cpufreq_dbs_governor_exit); -static int cpufreq_governor_start(struct cpufreq_policy *policy) +int cpufreq_dbs_governor_start(struct cpufreq_policy *policy) { struct dbs_governor *gov = dbs_governor_of(policy); struct policy_dbs_info *policy_dbs = policy->governor_data; @@ -539,14 +540,15 @@ static int cpufreq_governor_start(struct cpufreq_policy *policy) gov_set_update_util(policy_dbs, sampling_rate); return 0; } +EXPORT_SYMBOL_GPL(cpufreq_dbs_governor_start); -static int cpufreq_governor_stop(struct cpufreq_policy *policy) +void cpufreq_dbs_governor_stop(struct cpufreq_policy *policy) { gov_cancel_work(policy); - return 0; } +EXPORT_SYMBOL_GPL(cpufreq_dbs_governor_stop); -static int cpufreq_governor_limits(struct cpufreq_policy *policy) +void cpufreq_dbs_governor_limits(struct cpufreq_policy *policy) { struct policy_dbs_info *policy_dbs = policy->governor_data; @@ -560,26 +562,5 @@ static int cpufreq_governor_limits(struct cpufreq_policy *policy) gov_update_sample_delay(policy_dbs, 0); mutex_unlock(&policy_dbs->timer_mutex); - - return 0; } - -int cpufreq_governor_dbs(struct cpufreq_policy *policy, unsigned int event) -{ - if (event == CPUFREQ_GOV_POLICY_INIT) { - return cpufreq_governor_init(policy); - } else if (policy->governor_data) { - switch (event) { - case CPUFREQ_GOV_POLICY_EXIT: - return cpufreq_governor_exit(policy); - case CPUFREQ_GOV_START: - return cpufreq_governor_start(policy); - case CPUFREQ_GOV_STOP: - return cpufreq_governor_stop(policy); - case CPUFREQ_GOV_LIMITS: - return cpufreq_governor_limits(policy); - } - } - return -EINVAL; -} -EXPORT_SYMBOL_GPL(cpufreq_governor_dbs); +EXPORT_SYMBOL_GPL(cpufreq_dbs_governor_limits); diff --git a/drivers/cpufreq/cpufreq_governor.h b/drivers/cpufreq/cpufreq_governor.h index 34eb214b6d57..36f0d19dd869 100644 --- a/drivers/cpufreq/cpufreq_governor.h +++ b/drivers/cpufreq/cpufreq_governor.h @@ -148,6 +148,25 @@ static inline struct dbs_governor *dbs_governor_of(struct cpufreq_policy *policy return container_of(policy->governor, struct dbs_governor, gov); } +/* Governor callback routines */ +int cpufreq_dbs_governor_init(struct cpufreq_policy *policy); +void cpufreq_dbs_governor_exit(struct cpufreq_policy *policy); +int cpufreq_dbs_governor_start(struct cpufreq_policy *policy); +void cpufreq_dbs_governor_stop(struct cpufreq_policy *policy); +void cpufreq_dbs_governor_limits(struct cpufreq_policy *policy); + +#define CPUFREQ_DBS_GOVERNOR_INITIALIZER(_name_) \ + { \ + .name = _name_, \ + .max_transition_latency = TRANSITION_LATENCY_LIMIT, \ + .owner = THIS_MODULE, \ + .init = cpufreq_dbs_governor_init, \ + .exit = cpufreq_dbs_governor_exit, \ + .start = cpufreq_dbs_governor_start, \ + .stop = cpufreq_dbs_governor_stop, \ + .limits = cpufreq_dbs_governor_limits, \ + } + /* Governor specific operations */ struct od_ops { unsigned int (*powersave_bias_target)(struct cpufreq_policy *policy, @@ -155,7 +174,6 @@ struct od_ops { }; unsigned int dbs_update(struct cpufreq_policy *policy); -int cpufreq_governor_dbs(struct cpufreq_policy *policy, unsigned int event); void od_register_powersave_bias_handler(unsigned int (*f) (struct cpufreq_policy *, unsigned int, unsigned int), unsigned int powersave_bias); diff --git a/drivers/cpufreq/cpufreq_ondemand.c b/drivers/cpufreq/cpufreq_ondemand.c index 300163430516..09d6c0c405e4 100644 --- a/drivers/cpufreq/cpufreq_ondemand.c +++ b/drivers/cpufreq/cpufreq_ondemand.c @@ -420,12 +420,7 @@ static struct od_ops od_ops = { }; static struct dbs_governor od_dbs_gov = { - .gov = { - .name = "ondemand", - .governor = cpufreq_governor_dbs, - .max_transition_latency = TRANSITION_LATENCY_LIMIT, - .owner = THIS_MODULE, - }, + .gov = CPUFREQ_DBS_GOVERNOR_INITIALIZER("ondemand"), .kobj_type = { .default_attrs = od_attributes }, .gov_dbs_timer = od_dbs_timer, .alloc = od_alloc, diff --git a/drivers/cpufreq/cpufreq_performance.c b/drivers/cpufreq/cpufreq_performance.c index dd4dca3ab844..dafb679adc58 100644 --- a/drivers/cpufreq/cpufreq_performance.c +++ b/drivers/cpufreq/cpufreq_performance.c @@ -16,25 +16,16 @@ #include #include -static int cpufreq_governor_performance(struct cpufreq_policy *policy, - unsigned int event) +static void cpufreq_gov_performance_limits(struct cpufreq_policy *policy) { - switch (event) { - case CPUFREQ_GOV_LIMITS: - pr_debug("setting to %u kHz\n", policy->max); - __cpufreq_driver_target(policy, policy->max, - CPUFREQ_RELATION_H); - break; - default: - break; - } - return 0; + pr_debug("setting to %u kHz\n", policy->max); + __cpufreq_driver_target(policy, policy->max, CPUFREQ_RELATION_H); } static struct cpufreq_governor cpufreq_gov_performance = { .name = "performance", - .governor = cpufreq_governor_performance, .owner = THIS_MODULE, + .limits = cpufreq_gov_performance_limits, }; static int __init cpufreq_gov_performance_init(void) diff --git a/drivers/cpufreq/cpufreq_powersave.c b/drivers/cpufreq/cpufreq_powersave.c index 691e573e46ee..78a651038faf 100644 --- a/drivers/cpufreq/cpufreq_powersave.c +++ b/drivers/cpufreq/cpufreq_powersave.c @@ -16,24 +16,15 @@ #include #include -static int cpufreq_governor_powersave(struct cpufreq_policy *policy, - unsigned int event) +static void cpufreq_gov_powersave_limits(struct cpufreq_policy *policy) { - switch (event) { - case CPUFREQ_GOV_LIMITS: - pr_debug("setting to %u kHz\n", policy->min); - __cpufreq_driver_target(policy, policy->min, - CPUFREQ_RELATION_L); - break; - default: - break; - } - return 0; + pr_debug("setting to %u kHz\n", policy->min); + __cpufreq_driver_target(policy, policy->min, CPUFREQ_RELATION_L); } static struct cpufreq_governor cpufreq_gov_powersave = { .name = "powersave", - .governor = cpufreq_governor_powersave, + .limits = cpufreq_gov_powersave_limits, .owner = THIS_MODULE, }; diff --git a/drivers/cpufreq/cpufreq_userspace.c b/drivers/cpufreq/cpufreq_userspace.c index 9f3dec9a3f36..bd897e3e134d 100644 --- a/drivers/cpufreq/cpufreq_userspace.c +++ b/drivers/cpufreq/cpufreq_userspace.c @@ -65,66 +65,66 @@ static int cpufreq_userspace_policy_init(struct cpufreq_policy *policy) return 0; } -static int cpufreq_governor_userspace(struct cpufreq_policy *policy, - unsigned int event) +static void cpufreq_userspace_policy_exit(struct cpufreq_policy *policy) +{ + mutex_lock(&userspace_mutex); + kfree(policy->governor_data); + policy->governor_data = NULL; + mutex_unlock(&userspace_mutex); +} + +static int cpufreq_userspace_policy_start(struct cpufreq_policy *policy) { unsigned int *setspeed = policy->governor_data; - unsigned int cpu = policy->cpu; - int rc = 0; - if (event == CPUFREQ_GOV_POLICY_INIT) - return cpufreq_userspace_policy_init(policy); + BUG_ON(!policy->cur); + pr_debug("started managing cpu %u\n", policy->cpu); - if (!setspeed) - return -EINVAL; + mutex_lock(&userspace_mutex); + per_cpu(cpu_is_managed, policy->cpu) = 1; + *setspeed = policy->cur; + mutex_unlock(&userspace_mutex); + return 0; +} - switch (event) { - case CPUFREQ_GOV_POLICY_EXIT: - mutex_lock(&userspace_mutex); - policy->governor_data = NULL; - kfree(setspeed); - mutex_unlock(&userspace_mutex); - break; - case CPUFREQ_GOV_START: - BUG_ON(!policy->cur); - pr_debug("started managing cpu %u\n", cpu); +static void cpufreq_userspace_policy_stop(struct cpufreq_policy *policy) +{ + unsigned int *setspeed = policy->governor_data; - mutex_lock(&userspace_mutex); - per_cpu(cpu_is_managed, cpu) = 1; - *setspeed = policy->cur; - mutex_unlock(&userspace_mutex); - break; - case CPUFREQ_GOV_STOP: - pr_debug("managing cpu %u stopped\n", cpu); + pr_debug("managing cpu %u stopped\n", policy->cpu); - mutex_lock(&userspace_mutex); - per_cpu(cpu_is_managed, cpu) = 0; - *setspeed = 0; - mutex_unlock(&userspace_mutex); - break; - case CPUFREQ_GOV_LIMITS: - mutex_lock(&userspace_mutex); - pr_debug("limit event for cpu %u: %u - %u kHz, currently %u kHz, last set to %u kHz\n", - cpu, policy->min, policy->max, policy->cur, *setspeed); + mutex_lock(&userspace_mutex); + per_cpu(cpu_is_managed, policy->cpu) = 0; + *setspeed = 0; + mutex_unlock(&userspace_mutex); +} - if (policy->max < *setspeed) - __cpufreq_driver_target(policy, policy->max, - CPUFREQ_RELATION_H); - else if (policy->min > *setspeed) - __cpufreq_driver_target(policy, policy->min, - CPUFREQ_RELATION_L); - else - __cpufreq_driver_target(policy, *setspeed, - CPUFREQ_RELATION_L); - mutex_unlock(&userspace_mutex); - break; - } - return rc; +static void cpufreq_userspace_policy_limits(struct cpufreq_policy *policy) +{ + unsigned int *setspeed = policy->governor_data; + + mutex_lock(&userspace_mutex); + + pr_debug("limit event for cpu %u: %u - %u kHz, currently %u kHz, last set to %u kHz\n", + policy->cpu, policy->min, policy->max, policy->cur, *setspeed); + + if (policy->max < *setspeed) + __cpufreq_driver_target(policy, policy->max, CPUFREQ_RELATION_H); + else if (policy->min > *setspeed) + __cpufreq_driver_target(policy, policy->min, CPUFREQ_RELATION_L); + else + __cpufreq_driver_target(policy, *setspeed, CPUFREQ_RELATION_L); + + mutex_unlock(&userspace_mutex); } static struct cpufreq_governor cpufreq_gov_userspace = { .name = "userspace", - .governor = cpufreq_governor_userspace, + .init = cpufreq_userspace_policy_init, + .exit = cpufreq_userspace_policy_exit, + .start = cpufreq_userspace_policy_start, + .stop = cpufreq_userspace_policy_stop, + .limits = cpufreq_userspace_policy_limits, .store_setspeed = cpufreq_set, .show_setspeed = show_speed, .owner = THIS_MODULE, diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h index 4e81e08db752..5bcda3e6aa9e 100644 --- a/include/linux/cpufreq.h +++ b/include/linux/cpufreq.h @@ -455,18 +455,14 @@ static inline unsigned long cpufreq_scale(unsigned long old, u_int div, #define MIN_LATENCY_MULTIPLIER (20) #define TRANSITION_LATENCY_LIMIT (10 * 1000 * 1000) -/* Governor Events */ -#define CPUFREQ_GOV_START 1 -#define CPUFREQ_GOV_STOP 2 -#define CPUFREQ_GOV_LIMITS 3 -#define CPUFREQ_GOV_POLICY_INIT 4 -#define CPUFREQ_GOV_POLICY_EXIT 5 - struct cpufreq_governor { char name[CPUFREQ_NAME_LEN]; int initialized; - int (*governor) (struct cpufreq_policy *policy, - unsigned int event); + int (*init)(struct cpufreq_policy *policy); + void (*exit)(struct cpufreq_policy *policy); + int (*start)(struct cpufreq_policy *policy); + void (*stop)(struct cpufreq_policy *policy); + void (*limits)(struct cpufreq_policy *policy); ssize_t (*show_setspeed) (struct cpufreq_policy *policy, char *buf); int (*store_setspeed) (struct cpufreq_policy *policy, diff --git a/kernel/sched/cpufreq_schedutil.c b/kernel/sched/cpufreq_schedutil.c index 14c4aa25cc45..fdcee3cf38fc 100644 --- a/kernel/sched/cpufreq_schedutil.c +++ b/kernel/sched/cpufreq_schedutil.c @@ -394,7 +394,7 @@ static int sugov_init(struct cpufreq_policy *policy) return ret; } -static int sugov_exit(struct cpufreq_policy *policy) +static void sugov_exit(struct cpufreq_policy *policy) { struct sugov_policy *sg_policy = policy->governor_data; struct sugov_tunables *tunables = sg_policy->tunables; @@ -412,7 +412,6 @@ static int sugov_exit(struct cpufreq_policy *policy) mutex_unlock(&global_tunables_lock); sugov_policy_free(sg_policy); - return 0; } static int sugov_start(struct cpufreq_policy *policy) @@ -444,7 +443,7 @@ static int sugov_start(struct cpufreq_policy *policy) return 0; } -static int sugov_stop(struct cpufreq_policy *policy) +static void sugov_stop(struct cpufreq_policy *policy) { struct sugov_policy *sg_policy = policy->governor_data; unsigned int cpu; @@ -456,10 +455,9 @@ static int sugov_stop(struct cpufreq_policy *policy) irq_work_sync(&sg_policy->irq_work); cancel_work_sync(&sg_policy->work); - return 0; } -static int sugov_limits(struct cpufreq_policy *policy) +static void sugov_limits(struct cpufreq_policy *policy) { struct sugov_policy *sg_policy = policy->governor_data; @@ -477,32 +475,16 @@ static int sugov_limits(struct cpufreq_policy *policy) } sg_policy->need_freq_update = true; - return 0; -} - -int sugov_governor(struct cpufreq_policy *policy, unsigned int event) -{ - if (event == CPUFREQ_GOV_POLICY_INIT) { - return sugov_init(policy); - } else if (policy->governor_data) { - switch (event) { - case CPUFREQ_GOV_POLICY_EXIT: - return sugov_exit(policy); - case CPUFREQ_GOV_START: - return sugov_start(policy); - case CPUFREQ_GOV_STOP: - return sugov_stop(policy); - case CPUFREQ_GOV_LIMITS: - return sugov_limits(policy); - } - } - return -EINVAL; } static struct cpufreq_governor schedutil_gov = { .name = "schedutil", - .governor = sugov_governor, .owner = THIS_MODULE, + .init = sugov_init, + .exit = sugov_exit, + .start = sugov_start, + .stop = sugov_stop, + .limits = sugov_limits, }; static int __init sugov_module_init(void) From a69d6b2914865965acc6df95d1aac7ff8ca35094 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Wed, 18 May 2016 17:55:26 +0530 Subject: [PATCH 08/42] cpufreq: governor: Remove prints from allocation failures These aren't required anymore as the allocation core already prints such messages. Remove the redundant ones. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq_conservative.c | 4 +--- drivers/cpufreq/cpufreq_ondemand.c | 4 +--- 2 files changed, 2 insertions(+), 6 deletions(-) diff --git a/drivers/cpufreq/cpufreq_conservative.c b/drivers/cpufreq/cpufreq_conservative.c index 2568508fca38..78faa9fbc384 100644 --- a/drivers/cpufreq/cpufreq_conservative.c +++ b/drivers/cpufreq/cpufreq_conservative.c @@ -273,10 +273,8 @@ static int cs_init(struct dbs_data *dbs_data, bool notify) struct cs_dbs_tuners *tuners; tuners = kzalloc(sizeof(*tuners), GFP_KERNEL); - if (!tuners) { - pr_err("%s: kzalloc failed\n", __func__); + if (!tuners) return -ENOMEM; - } tuners->down_threshold = DEF_FREQUENCY_DOWN_THRESHOLD; tuners->freq_step = DEF_FREQUENCY_STEP; diff --git a/drivers/cpufreq/cpufreq_ondemand.c b/drivers/cpufreq/cpufreq_ondemand.c index 09d6c0c405e4..4441a11139f1 100644 --- a/drivers/cpufreq/cpufreq_ondemand.c +++ b/drivers/cpufreq/cpufreq_ondemand.c @@ -368,10 +368,8 @@ static int od_init(struct dbs_data *dbs_data, bool notify) int cpu; tuners = kzalloc(sizeof(*tuners), GFP_KERNEL); - if (!tuners) { - pr_err("%s: kzalloc failed\n", __func__); + if (!tuners) return -ENOMEM; - } cpu = get_cpu(); idle_time = get_cpu_idle_time_us(cpu, NULL); From 666f4ccc5d0f713500f5235eb3f12c1ea8d2278a Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Wed, 18 May 2016 17:55:27 +0530 Subject: [PATCH 09/42] cpufreq: governor: Remove unnecessary bits from print message pr_*() helpers already prefix the print messages with "cpufreq_governor:" and similar details aren't required in the actual message. For example, the print message getting fixed looks like this before this patch: cpufreq_governor: cpufreq: Governor initialization failed (dbs_data kobject init error 0) Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq_governor.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/cpufreq/cpufreq_governor.c b/drivers/cpufreq/cpufreq_governor.c index ca9927c15a71..18eab7f4ba4c 100644 --- a/drivers/cpufreq/cpufreq_governor.c +++ b/drivers/cpufreq/cpufreq_governor.c @@ -458,7 +458,7 @@ int cpufreq_dbs_governor_init(struct cpufreq_policy *policy) goto out; /* Failure, so roll back. */ - pr_err("cpufreq: Governor initialization failed (dbs_data kobject init error %d)\n", ret); + pr_err("initialization failed (dbs_data kobject init error %d)\n", ret); policy->governor_data = NULL; From bf2be2de8493dd5f86d6e0f0d4eecb5810ad035b Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Wed, 18 May 2016 17:55:31 +0530 Subject: [PATCH 10/42] cpufreq: governor: Create cpufreq_policy_apply_limits() Create a new helper to avoid code duplication across governors. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq_governor.c | 7 +------ include/linux/cpufreq.h | 8 ++++++++ kernel/sched/cpufreq_schedutil.c | 9 +-------- 3 files changed, 10 insertions(+), 14 deletions(-) diff --git a/drivers/cpufreq/cpufreq_governor.c b/drivers/cpufreq/cpufreq_governor.c index 18eab7f4ba4c..4b88b5bec4ef 100644 --- a/drivers/cpufreq/cpufreq_governor.c +++ b/drivers/cpufreq/cpufreq_governor.c @@ -553,12 +553,7 @@ void cpufreq_dbs_governor_limits(struct cpufreq_policy *policy) struct policy_dbs_info *policy_dbs = policy->governor_data; mutex_lock(&policy_dbs->timer_mutex); - - if (policy->max < policy->cur) - __cpufreq_driver_target(policy, policy->max, CPUFREQ_RELATION_H); - else if (policy->min > policy->cur) - __cpufreq_driver_target(policy, policy->min, CPUFREQ_RELATION_L); - + cpufreq_policy_apply_limits(policy); gov_update_sample_delay(policy_dbs, 0); mutex_unlock(&policy_dbs->timer_mutex); diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h index 5bcda3e6aa9e..3be54a0b4373 100644 --- a/include/linux/cpufreq.h +++ b/include/linux/cpufreq.h @@ -489,6 +489,14 @@ void cpufreq_unregister_governor(struct cpufreq_governor *governor); struct cpufreq_governor *cpufreq_default_governor(void); struct cpufreq_governor *cpufreq_fallback_governor(void); +static inline void cpufreq_policy_apply_limits(struct cpufreq_policy *policy) +{ + if (policy->max < policy->cur) + __cpufreq_driver_target(policy, policy->max, CPUFREQ_RELATION_H); + else if (policy->min > policy->cur) + __cpufreq_driver_target(policy, policy->min, CPUFREQ_RELATION_L); +} + /* Governor attribute set */ struct gov_attr_set { struct kobject kobj; diff --git a/kernel/sched/cpufreq_schedutil.c b/kernel/sched/cpufreq_schedutil.c index fdcee3cf38fc..758efd7f3abe 100644 --- a/kernel/sched/cpufreq_schedutil.c +++ b/kernel/sched/cpufreq_schedutil.c @@ -463,14 +463,7 @@ static void sugov_limits(struct cpufreq_policy *policy) if (!policy->fast_switch_enabled) { mutex_lock(&sg_policy->work_lock); - - if (policy->max < policy->cur) - __cpufreq_driver_target(policy, policy->max, - CPUFREQ_RELATION_H); - else if (policy->min > policy->cur) - __cpufreq_driver_target(policy, policy->min, - CPUFREQ_RELATION_L); - + cpufreq_policy_apply_limits(policy); mutex_unlock(&sg_policy->work_lock); } From 9a15fb2c797a15524e63eacb10bd6cd68a99e830 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Wed, 18 May 2016 22:59:49 +0200 Subject: [PATCH 11/42] cpufreq: Drop the 'initialized' field from struct cpufreq_governor The 'initialized' field in struct cpufreq_governor is only used by the conservative governor (as a usage counter) and the way that happens is far from straightforward and arguably incorrect. Namely, the value of 'initialized' is checked by cpufreq_dbs_governor_init() and cpufreq_dbs_governor_exit() and the results of those checks are passed (as the second argument) to the ->init() and ->exit() callbacks in struct dbs_governor. Those callbacks are only implemented by the ondemand and conservative governors and ondemand doesn't use their second argument at all. In turn, the conservative governor uses it to decide whether or not to either register or unregister a transition notifier. That whole mechanism is not only unnecessarily convoluted, but also racy, because the 'initialized' field of struct cpufreq_governor is updated in cpufreq_init_governor() and cpufreq_exit_governor() under policy->rwsem which doesn't help if one of these functions is run twice in parallel for different policies (which isn't impossible in principle), for example. Instead of it, add a proper usage counter to the conservative governor and update it from cs_init() and cs_exit() which is guaranteed to be non-racy, as those functions are only called under gov_dbs_data_mutex which is global. With that in place, drop the 'initialized' field from struct cpufreq_governor as it is not used any more. Signed-off-by: Rafael J. Wysocki Acked-by: Viresh Kumar --- drivers/cpufreq/cpufreq.c | 3 -- drivers/cpufreq/cpufreq_conservative.c | 44 +++++++++++++++++--------- drivers/cpufreq/cpufreq_governor.c | 6 ++-- drivers/cpufreq/cpufreq_governor.h | 4 +-- drivers/cpufreq/cpufreq_ondemand.c | 4 +-- include/linux/cpufreq.h | 1 - 6 files changed, 36 insertions(+), 26 deletions(-) diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index d98ff688b2f1..bdf1f280ae39 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -2031,7 +2031,6 @@ static int cpufreq_init_governor(struct cpufreq_policy *policy) } } - policy->governor->initialized++; return 0; } @@ -2045,7 +2044,6 @@ static void cpufreq_exit_governor(struct cpufreq_policy *policy) if (policy->governor->exit) policy->governor->exit(policy); - policy->governor->initialized--; module_put(policy->governor->owner); } @@ -2110,7 +2108,6 @@ int cpufreq_register_governor(struct cpufreq_governor *governor) mutex_lock(&cpufreq_governor_mutex); - governor->initialized = 0; err = -EBUSY; if (!find_governor(governor->name)) { err = 0; diff --git a/drivers/cpufreq/cpufreq_conservative.c b/drivers/cpufreq/cpufreq_conservative.c index 78faa9fbc384..f967ec6c5720 100644 --- a/drivers/cpufreq/cpufreq_conservative.c +++ b/drivers/cpufreq/cpufreq_conservative.c @@ -127,7 +127,6 @@ static struct notifier_block cs_cpufreq_notifier_block = { }; /************************** sysfs interface ************************/ -static struct dbs_governor cs_dbs_gov; static ssize_t store_sampling_down_factor(struct gov_attr_set *attr_set, const char *buf, size_t count) @@ -255,6 +254,13 @@ static struct attribute *cs_attributes[] = { /************************** sysfs end ************************/ +struct cs_governor { + struct dbs_governor dbs_gov; + unsigned int usage_count; +}; + +static struct cs_governor cs_gov; + static struct policy_dbs_info *cs_alloc(void) { struct cs_policy_dbs_info *dbs_info; @@ -268,7 +274,7 @@ static void cs_free(struct policy_dbs_info *policy_dbs) kfree(to_dbs_info(policy_dbs)); } -static int cs_init(struct dbs_data *dbs_data, bool notify) +static int cs_init(struct dbs_data *dbs_data) { struct cs_dbs_tuners *tuners; @@ -286,16 +292,22 @@ static int cs_init(struct dbs_data *dbs_data, bool notify) dbs_data->min_sampling_rate = MIN_SAMPLING_RATE_RATIO * jiffies_to_usecs(10); - if (notify) + /* + * This function and cs_exit() are only called under gov_dbs_data_mutex + * which is global, so the cs_gov.usage_count accesses are guaranteed + * to be serialized. + */ + if (!cs_gov.usage_count++) cpufreq_register_notifier(&cs_cpufreq_notifier_block, CPUFREQ_TRANSITION_NOTIFIER); return 0; } -static void cs_exit(struct dbs_data *dbs_data, bool notify) +static void cs_exit(struct dbs_data *dbs_data) { - if (notify) + /* Protected by gov_dbs_data_mutex - see the comment in cs_init(). */ + if (!--cs_gov.usage_count) cpufreq_unregister_notifier(&cs_cpufreq_notifier_block, CPUFREQ_TRANSITION_NOTIFIER); @@ -310,18 +322,20 @@ static void cs_start(struct cpufreq_policy *policy) dbs_info->requested_freq = policy->cur; } -static struct dbs_governor cs_dbs_gov = { - .gov = CPUFREQ_DBS_GOVERNOR_INITIALIZER("conservative"), - .kobj_type = { .default_attrs = cs_attributes }, - .gov_dbs_timer = cs_dbs_timer, - .alloc = cs_alloc, - .free = cs_free, - .init = cs_init, - .exit = cs_exit, - .start = cs_start, +static struct cs_governor cs_gov = { + .dbs_gov = { + .gov = CPUFREQ_DBS_GOVERNOR_INITIALIZER("conservative"), + .kobj_type = { .default_attrs = cs_attributes }, + .gov_dbs_timer = cs_dbs_timer, + .alloc = cs_alloc, + .free = cs_free, + .init = cs_init, + .exit = cs_exit, + .start = cs_start, + }, }; -#define CPU_FREQ_GOV_CONSERVATIVE (&cs_dbs_gov.gov) +#define CPU_FREQ_GOV_CONSERVATIVE (&cs_gov.dbs_gov.gov) static int dbs_cpufreq_notifier(struct notifier_block *nb, unsigned long val, void *data) diff --git a/drivers/cpufreq/cpufreq_governor.c b/drivers/cpufreq/cpufreq_governor.c index 4b88b5bec4ef..31369e247192 100644 --- a/drivers/cpufreq/cpufreq_governor.c +++ b/drivers/cpufreq/cpufreq_governor.c @@ -429,7 +429,7 @@ int cpufreq_dbs_governor_init(struct cpufreq_policy *policy) gov_attr_set_init(&dbs_data->attr_set, &policy_dbs->list); - ret = gov->init(dbs_data, !policy->governor->initialized); + ret = gov->init(dbs_data); if (ret) goto free_policy_dbs_info; @@ -464,7 +464,7 @@ int cpufreq_dbs_governor_init(struct cpufreq_policy *policy) if (!have_governor_per_policy()) gov->gdbs_data = NULL; - gov->exit(dbs_data, !policy->governor->initialized); + gov->exit(dbs_data); kfree(dbs_data); free_policy_dbs_info: @@ -494,7 +494,7 @@ void cpufreq_dbs_governor_exit(struct cpufreq_policy *policy) if (!have_governor_per_policy()) gov->gdbs_data = NULL; - gov->exit(dbs_data, policy->governor->initialized == 1); + gov->exit(dbs_data); kfree(dbs_data); } diff --git a/drivers/cpufreq/cpufreq_governor.h b/drivers/cpufreq/cpufreq_governor.h index 36f0d19dd869..ef1037e9c92b 100644 --- a/drivers/cpufreq/cpufreq_governor.h +++ b/drivers/cpufreq/cpufreq_governor.h @@ -138,8 +138,8 @@ struct dbs_governor { unsigned int (*gov_dbs_timer)(struct cpufreq_policy *policy); struct policy_dbs_info *(*alloc)(void); void (*free)(struct policy_dbs_info *policy_dbs); - int (*init)(struct dbs_data *dbs_data, bool notify); - void (*exit)(struct dbs_data *dbs_data, bool notify); + int (*init)(struct dbs_data *dbs_data); + void (*exit)(struct dbs_data *dbs_data); void (*start)(struct cpufreq_policy *policy); }; diff --git a/drivers/cpufreq/cpufreq_ondemand.c b/drivers/cpufreq/cpufreq_ondemand.c index 4441a11139f1..c84fc2240d49 100644 --- a/drivers/cpufreq/cpufreq_ondemand.c +++ b/drivers/cpufreq/cpufreq_ondemand.c @@ -361,7 +361,7 @@ static void od_free(struct policy_dbs_info *policy_dbs) kfree(to_dbs_info(policy_dbs)); } -static int od_init(struct dbs_data *dbs_data, bool notify) +static int od_init(struct dbs_data *dbs_data) { struct od_dbs_tuners *tuners; u64 idle_time; @@ -400,7 +400,7 @@ static int od_init(struct dbs_data *dbs_data, bool notify) return 0; } -static void od_exit(struct dbs_data *dbs_data, bool notify) +static void od_exit(struct dbs_data *dbs_data) { kfree(dbs_data->tuners); } diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h index 3be54a0b4373..fbd696edf5bd 100644 --- a/include/linux/cpufreq.h +++ b/include/linux/cpufreq.h @@ -457,7 +457,6 @@ static inline unsigned long cpufreq_scale(unsigned long old, u_int div, struct cpufreq_governor { char name[CPUFREQ_NAME_LEN]; - int initialized; int (*init)(struct cpufreq_policy *policy); void (*exit)(struct cpufreq_policy *policy); int (*start)(struct cpufreq_policy *policy); From 388612baba201bdcedfa66782f524dcc507d9f12 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Tue, 24 May 2016 10:26:50 +0530 Subject: [PATCH 12/42] cpufreq: Send START policy notification after sending CREATE The sequence got a bit wrong as we are sending CPUFREQ_START notifications even before we have sent CPUFREQ_CREATE_POLICY. Fix it. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index bdf1f280ae39..364922fda428 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -1258,9 +1258,6 @@ static int cpufreq_online(unsigned int cpu) } } - blocking_notifier_call_chain(&cpufreq_policy_notifier_list, - CPUFREQ_START, policy); - if (new_policy) { ret = cpufreq_add_dev_interface(policy); if (ret) @@ -1273,6 +1270,9 @@ static int cpufreq_online(unsigned int cpu) 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", From 910c6e881c3a610dfd7575b96975d4df21e3a920 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Thu, 26 May 2016 11:27:24 +0530 Subject: [PATCH 13/42] cpufreq: Use clamp_val() in __cpufreq_driver_target() Use clamp_val() instead of open coding it. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index 364922fda428..198416bc22de 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -1923,10 +1923,7 @@ int __cpufreq_driver_target(struct cpufreq_policy *policy, return -ENODEV; /* Make sure that target_freq is within supported range */ - if (target_freq > policy->max) - target_freq = policy->max; - if (target_freq < policy->min) - target_freq = policy->min; + target_freq = clamp_val(target_freq, policy->min, policy->max); pr_debug("target for CPU %u: %u kHz, relation %u, requested %u kHz\n", policy->cpu, target_freq, relation, old_target_freq); From 1aefc75b2449eb68a6fc3ca932e2a4ee353b748d Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Tue, 31 May 2016 22:14:44 +0200 Subject: [PATCH 14/42] cpufreq: stats: Make the stats code non-modular The modularity of cpufreq_stats is quite problematic. First off, the usage of policy notifiers for the initialization and cleanup in the cpufreq_stats module is inherently racy with respect to CPU offline/online and the initialization and cleanup of the cpufreq driver. Second, fast frequency switching (used by the schedutil governor) cannot be enabled if any transition notifiers are registered, so if the cpufreq_stats module (that registers a transition notifier for updating transition statistics) is loaded, the schedutil governor cannot use fast frequency switching. On the other hand, allowing cpufreq_stats to be built as a module doesn't really add much value. Arguably, there's not much reason for that code to be modular at all. For the above reasons, make the cpufreq stats code non-modular, modify the core to invoke functions provided by that code directly and drop the notifiers from it. Make the stats sysfs attributes appear empty if fast frequency switching is enabled as the statistics will not be updated in that case anyway (and returning -EBUSY from those attributes breaks powertop). While at it, clean up Kconfig help for the CPU_FREQ_STAT and CPU_FREQ_STAT_DETAILS options. Signed-off-by: Rafael J. Wysocki Acked-by: Viresh Kumar --- drivers/cpufreq/Kconfig | 13 +-- drivers/cpufreq/cpufreq.c | 4 + drivers/cpufreq/cpufreq_stats.c | 154 +++++--------------------------- include/linux/cpufreq.h | 12 +++ 4 files changed, 41 insertions(+), 142 deletions(-) diff --git a/drivers/cpufreq/Kconfig b/drivers/cpufreq/Kconfig index b7445b6ae5a4..c822d72629d5 100644 --- a/drivers/cpufreq/Kconfig +++ b/drivers/cpufreq/Kconfig @@ -31,23 +31,18 @@ config CPU_FREQ_BOOST_SW depends on THERMAL config CPU_FREQ_STAT - tristate "CPU frequency translation statistics" + bool "CPU frequency transition statistics" default y help - This driver exports CPU frequency statistics information through sysfs - file system. - - To compile this driver as a module, choose M here: the - module will be called cpufreq_stats. + Export CPU frequency statistics information through sysfs. If in doubt, say N. config CPU_FREQ_STAT_DETAILS - bool "CPU frequency translation statistics details" + bool "CPU frequency transition statistics details" depends on CPU_FREQ_STAT help - This will show detail CPU frequency translation table in sysfs file - system. + Show detailed CPU frequency transition table in sysfs. If in doubt, say N. diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index 198416bc22de..c6a14ba239a2 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -347,6 +347,7 @@ static void __cpufreq_notify_transition(struct cpufreq_policy *policy, pr_debug("FREQ: %lu - CPU: %lu\n", (unsigned long)freqs->new, (unsigned long)freqs->cpu); trace_cpu_frequency(freqs->new, freqs->cpu); + cpufreq_stats_record_transition(policy, freqs->new); srcu_notifier_call_chain(&cpufreq_transition_notifier_list, CPUFREQ_POSTCHANGE, freqs); if (likely(policy) && likely(policy->cpu == freqs->cpu)) @@ -1108,6 +1109,7 @@ static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy, bool notify) CPUFREQ_REMOVE_POLICY, policy); down_write(&policy->rwsem); + cpufreq_stats_free_table(policy); cpufreq_remove_dev_symlink(policy); kobj = &policy->kobj; cmp = &policy->kobj_unregister; @@ -1262,6 +1264,8 @@ static int cpufreq_online(unsigned int cpu) ret = cpufreq_add_dev_interface(policy); if (ret) goto out_exit_policy; + + cpufreq_stats_create_table(policy); blocking_notifier_call_chain(&cpufreq_policy_notifier_list, CPUFREQ_CREATE_POLICY, policy); diff --git a/drivers/cpufreq/cpufreq_stats.c b/drivers/cpufreq/cpufreq_stats.c index 5e370a30a964..c6e7f81a0397 100644 --- a/drivers/cpufreq/cpufreq_stats.c +++ b/drivers/cpufreq/cpufreq_stats.c @@ -15,7 +15,7 @@ #include #include -static spinlock_t cpufreq_stats_lock; +static DEFINE_SPINLOCK(cpufreq_stats_lock); struct cpufreq_stats { unsigned int total_trans; @@ -52,6 +52,9 @@ static ssize_t show_time_in_state(struct cpufreq_policy *policy, char *buf) ssize_t len = 0; int i; + if (policy->fast_switch_enabled) + return 0; + cpufreq_stats_update(stats); for (i = 0; i < stats->state_num; i++) { len += sprintf(buf + len, "%u %llu\n", stats->freq_table[i], @@ -68,6 +71,9 @@ static ssize_t show_trans_table(struct cpufreq_policy *policy, char *buf) ssize_t len = 0; int i, j; + if (policy->fast_switch_enabled) + return 0; + len += snprintf(buf + len, PAGE_SIZE - len, " From : To\n"); len += snprintf(buf + len, PAGE_SIZE - len, " : "); for (i = 0; i < stats->state_num; i++) { @@ -130,7 +136,7 @@ static int freq_table_get_index(struct cpufreq_stats *stats, unsigned int freq) return -1; } -static void __cpufreq_stats_free_table(struct cpufreq_policy *policy) +void cpufreq_stats_free_table(struct cpufreq_policy *policy) { struct cpufreq_stats *stats = policy->stats; @@ -146,20 +152,7 @@ static void __cpufreq_stats_free_table(struct cpufreq_policy *policy) policy->stats = NULL; } -static void cpufreq_stats_free_table(unsigned int cpu) -{ - struct cpufreq_policy *policy; - - policy = cpufreq_cpu_get(cpu); - if (!policy) - return; - - __cpufreq_stats_free_table(policy); - - cpufreq_cpu_put(policy); -} - -static int __cpufreq_stats_create_table(struct cpufreq_policy *policy) +void cpufreq_stats_create_table(struct cpufreq_policy *policy) { unsigned int i = 0, count = 0, ret = -ENOMEM; struct cpufreq_stats *stats; @@ -170,15 +163,15 @@ static int __cpufreq_stats_create_table(struct cpufreq_policy *policy) /* We need cpufreq table for creating stats table */ table = cpufreq_frequency_get_table(cpu); if (unlikely(!table)) - return 0; + return; /* stats already initialized */ if (policy->stats) - return -EEXIST; + return; stats = kzalloc(sizeof(*stats), GFP_KERNEL); if (!stats) - return -ENOMEM; + return; /* Find total allocation size */ cpufreq_for_each_valid_entry(pos, table) @@ -215,80 +208,32 @@ static int __cpufreq_stats_create_table(struct cpufreq_policy *policy) policy->stats = stats; ret = sysfs_create_group(&policy->kobj, &stats_attr_group); if (!ret) - return 0; + return; /* We failed, release resources */ policy->stats = NULL; kfree(stats->time_in_state); free_stat: kfree(stats); - - return ret; } -static void cpufreq_stats_create_table(unsigned int cpu) +void cpufreq_stats_record_transition(struct cpufreq_policy *policy, + unsigned int new_freq) { - struct cpufreq_policy *policy; - - /* - * "likely(!policy)" because normally cpufreq_stats will be registered - * before cpufreq driver - */ - policy = cpufreq_cpu_get(cpu); - if (likely(!policy)) - return; - - __cpufreq_stats_create_table(policy); - - cpufreq_cpu_put(policy); -} - -static int cpufreq_stat_notifier_policy(struct notifier_block *nb, - unsigned long val, void *data) -{ - int ret = 0; - struct cpufreq_policy *policy = data; - - if (val == CPUFREQ_CREATE_POLICY) - ret = __cpufreq_stats_create_table(policy); - else if (val == CPUFREQ_REMOVE_POLICY) - __cpufreq_stats_free_table(policy); - - return ret; -} - -static int cpufreq_stat_notifier_trans(struct notifier_block *nb, - unsigned long val, void *data) -{ - struct cpufreq_freqs *freq = data; - struct cpufreq_policy *policy = cpufreq_cpu_get(freq->cpu); - struct cpufreq_stats *stats; + struct cpufreq_stats *stats = policy->stats; int old_index, new_index; - if (!policy) { - pr_err("%s: No policy found\n", __func__); - return 0; - } - - if (val != CPUFREQ_POSTCHANGE) - goto put_policy; - - if (!policy->stats) { + if (!stats) { pr_debug("%s: No stats found\n", __func__); - goto put_policy; + return; } - stats = policy->stats; - old_index = stats->last_index; - new_index = freq_table_get_index(stats, freq->new); + new_index = freq_table_get_index(stats, new_freq); /* We can't do stats->time_in_state[-1]= .. */ - if (old_index == -1 || new_index == -1) - goto put_policy; - - if (old_index == new_index) - goto put_policy; + if (old_index == -1 || new_index == -1 || old_index == new_index) + return; cpufreq_stats_update(stats); @@ -297,61 +242,4 @@ static int cpufreq_stat_notifier_trans(struct notifier_block *nb, stats->trans_table[old_index * stats->max_state + new_index]++; #endif stats->total_trans++; - -put_policy: - cpufreq_cpu_put(policy); - return 0; } - -static struct notifier_block notifier_policy_block = { - .notifier_call = cpufreq_stat_notifier_policy -}; - -static struct notifier_block notifier_trans_block = { - .notifier_call = cpufreq_stat_notifier_trans -}; - -static int __init cpufreq_stats_init(void) -{ - int ret; - unsigned int cpu; - - spin_lock_init(&cpufreq_stats_lock); - ret = cpufreq_register_notifier(¬ifier_policy_block, - CPUFREQ_POLICY_NOTIFIER); - if (ret) - return ret; - - for_each_online_cpu(cpu) - cpufreq_stats_create_table(cpu); - - ret = cpufreq_register_notifier(¬ifier_trans_block, - CPUFREQ_TRANSITION_NOTIFIER); - if (ret) { - cpufreq_unregister_notifier(¬ifier_policy_block, - CPUFREQ_POLICY_NOTIFIER); - for_each_online_cpu(cpu) - cpufreq_stats_free_table(cpu); - return ret; - } - - return 0; -} -static void __exit cpufreq_stats_exit(void) -{ - unsigned int cpu; - - cpufreq_unregister_notifier(¬ifier_policy_block, - CPUFREQ_POLICY_NOTIFIER); - cpufreq_unregister_notifier(¬ifier_trans_block, - CPUFREQ_TRANSITION_NOTIFIER); - for_each_online_cpu(cpu) - cpufreq_stats_free_table(cpu); -} - -MODULE_AUTHOR("Zou Nan hai "); -MODULE_DESCRIPTION("Export cpufreq stats via sysfs"); -MODULE_LICENSE("GPL"); - -module_init(cpufreq_stats_init); -module_exit(cpufreq_stats_exit); diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h index fbd696edf5bd..7ed93c310c08 100644 --- a/include/linux/cpufreq.h +++ b/include/linux/cpufreq.h @@ -185,6 +185,18 @@ static inline unsigned int cpufreq_quick_get_max(unsigned int cpu) static inline void disable_cpufreq(void) { } #endif +#ifdef CONFIG_CPU_FREQ_STAT +void cpufreq_stats_create_table(struct cpufreq_policy *policy); +void cpufreq_stats_free_table(struct cpufreq_policy *policy); +void cpufreq_stats_record_transition(struct cpufreq_policy *policy, + unsigned int new_freq); +#else +static inline void cpufreq_stats_create_table(struct cpufreq_policy *policy) { } +static inline void cpufreq_stats_free_table(struct cpufreq_policy *policy) { } +static inline void cpufreq_stats_record_transition(struct cpufreq_policy *policy, + unsigned int new_freq) { } +#endif /* CONFIG_CPU_FREQ_STAT */ + /********************************************************************* * CPUFREQ DRIVER INTERFACE * *********************************************************************/ From 64bf55a72f550ad33f4053493717f51efe632e9f Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Tue, 31 May 2016 16:50:23 +0530 Subject: [PATCH 15/42] cpufreq: Unexport cpufreq_frequency_table_cpuinfo() All cpufreq drivers with a freq-table are migrated to use cpufreq_table_validate_and_show() long back and the routine cpufreq_frequency_table_cpuinfo() isn't used outside of cpufreq core now. Unexport it and update Documentation as well. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- Documentation/cpu-freq/core.txt | 4 ++-- Documentation/cpu-freq/cpu-drivers.txt | 2 +- drivers/cpufreq/freq_table.c | 2 -- 3 files changed, 3 insertions(+), 5 deletions(-) diff --git a/Documentation/cpu-freq/core.txt b/Documentation/cpu-freq/core.txt index ba78e7c2a069..4bc7287806de 100644 --- a/Documentation/cpu-freq/core.txt +++ b/Documentation/cpu-freq/core.txt @@ -96,7 +96,7 @@ new - new frequency For details about OPP, see Documentation/power/opp.txt dev_pm_opp_init_cpufreq_table - cpufreq framework typically is initialized with - cpufreq_frequency_table_cpuinfo which is provided with the list of + cpufreq_table_validate_and_show() which is provided with the list of frequencies that are available for operation. This function provides a ready to use conversion routine to translate the OPP layer's internal information about the available frequencies into a format readily @@ -110,7 +110,7 @@ dev_pm_opp_init_cpufreq_table - cpufreq framework typically is initialized with /* Do things */ r = dev_pm_opp_init_cpufreq_table(dev, &freq_table); if (!r) - cpufreq_frequency_table_cpuinfo(policy, freq_table); + cpufreq_table_validate_and_show(policy, freq_table); /* Do other things */ } diff --git a/Documentation/cpu-freq/cpu-drivers.txt b/Documentation/cpu-freq/cpu-drivers.txt index 14f4e6336d88..decbb8cb5573 100644 --- a/Documentation/cpu-freq/cpu-drivers.txt +++ b/Documentation/cpu-freq/cpu-drivers.txt @@ -231,7 +231,7 @@ 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_frequency_table_cpuinfo(struct cpufreq_policy *policy, +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 diff --git a/drivers/cpufreq/freq_table.c b/drivers/cpufreq/freq_table.c index a8f1daffc9bc..4e5c5dbfed7a 100644 --- a/drivers/cpufreq/freq_table.c +++ b/drivers/cpufreq/freq_table.c @@ -63,8 +63,6 @@ int cpufreq_frequency_table_cpuinfo(struct cpufreq_policy *policy, else return 0; } -EXPORT_SYMBOL_GPL(cpufreq_frequency_table_cpuinfo); - int cpufreq_frequency_table_verify(struct cpufreq_policy *policy, struct cpufreq_frequency_table *table) From 62d167330679994ec816a4fe6be22f589fcfdf76 Mon Sep 17 00:00:00 2001 From: Dave Hansen Date: Thu, 2 Jun 2016 17:19:36 -0700 Subject: [PATCH 16/42] x86, powercap, rapl: Use Intel model macros intead of open-coding Use the new macros to remove another large set of open-coded values. Signed-off-by: Dave Hansen Acked-by: Rafael J. Wysocki Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Dave Hansen Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: jacob.jun.pan@intel.com Cc: linux-pm@vger.kernel.org Link: http://lkml.kernel.org/r/20160603001936.F474F9D8@viggo.jf.intel.com Signed-off-by: Ingo Molnar --- drivers/powercap/intel_rapl.c | 43 ++++++++++++++++++----------------- 1 file changed, 22 insertions(+), 21 deletions(-) diff --git a/drivers/powercap/intel_rapl.c b/drivers/powercap/intel_rapl.c index b2766b867b0e..defa8d6c27bf 100644 --- a/drivers/powercap/intel_rapl.c +++ b/drivers/powercap/intel_rapl.c @@ -33,6 +33,7 @@ #include #include +#include /* Local defines */ #define MSR_PLATFORM_POWER_LIMIT 0x0000065C @@ -1096,27 +1097,27 @@ static const struct rapl_defaults rapl_defaults_cht = { } static const struct x86_cpu_id rapl_ids[] __initconst = { - RAPL_CPU(0x2a, rapl_defaults_core),/* Sandy Bridge */ - RAPL_CPU(0x2d, rapl_defaults_core),/* Sandy Bridge EP */ - RAPL_CPU(0x37, rapl_defaults_byt),/* Valleyview */ - RAPL_CPU(0x3a, rapl_defaults_core),/* Ivy Bridge */ - RAPL_CPU(0x3c, rapl_defaults_core),/* Haswell */ - RAPL_CPU(0x3d, rapl_defaults_core),/* Broadwell */ - RAPL_CPU(0x3f, rapl_defaults_hsw_server),/* Haswell servers */ - RAPL_CPU(0x4f, rapl_defaults_hsw_server),/* Broadwell servers */ - RAPL_CPU(0x45, rapl_defaults_core),/* Haswell ULT */ - RAPL_CPU(0x46, rapl_defaults_core),/* Haswell */ - RAPL_CPU(0x47, rapl_defaults_core),/* Broadwell-H */ - RAPL_CPU(0x4E, rapl_defaults_core),/* Skylake */ - RAPL_CPU(0x4C, rapl_defaults_cht),/* Braswell/Cherryview */ - RAPL_CPU(0x4A, rapl_defaults_tng),/* Tangier */ - RAPL_CPU(0x56, rapl_defaults_core),/* Future Xeon */ - RAPL_CPU(0x5A, rapl_defaults_ann),/* Annidale */ - RAPL_CPU(0X5C, rapl_defaults_core),/* Broxton */ - RAPL_CPU(0x5E, rapl_defaults_core),/* Skylake-H/S */ - RAPL_CPU(0x57, rapl_defaults_hsw_server),/* Knights Landing */ - RAPL_CPU(0x8E, rapl_defaults_core),/* Kabylake */ - RAPL_CPU(0x9E, rapl_defaults_core),/* Kabylake */ + RAPL_CPU(INTEL_FAM6_SANDYBRIDGE, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_SANDYBRIDGE_X, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_ATOM_SILVERMONT1, rapl_defaults_byt), + RAPL_CPU(INTEL_FAM6_IVYBRIDGE, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_HASWELL_CORE, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_BROADWELL_CORE, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_HASWELL_X, rapl_defaults_hsw_server), + RAPL_CPU(INTEL_FAM6_BROADWELL_X, rapl_defaults_hsw_server), + RAPL_CPU(INTEL_FAM6_HASWELL_ULT, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_HASWELL_GT3E, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_BROADWELL_GT3E, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_SKYLAKE_MOBILE, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_ATOM_AIRMONT, rapl_defaults_cht), + RAPL_CPU(INTEL_FAM6_ATOM_MERRIFIELD1, rapl_defaults_tng), + RAPL_CPU(INTEL_FAM6_BROADWELL_XEON_D, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_ATOM_MERRIFIELD2, rapl_defaults_ann), + RAPL_CPU(INTEL_FAM6_ATOM_GOLDMONT, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_SKYLAKE_DESKTOP, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_XEON_PHI_KNL, rapl_defaults_hsw_server), + RAPL_CPU(INTEL_FAM6_KABYLAKE_MOBILE, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_KABYLAKE_DESKTOP, rapl_defaults_core), {} }; MODULE_DEVICE_TABLE(x86cpu, rapl_ids); From 0bb04b5f2c2452fdf4e1e376421de6bb34485fbb Mon Sep 17 00:00:00 2001 From: Dave Hansen Date: Thu, 2 Jun 2016 17:19:37 -0700 Subject: [PATCH 17/42] x86, powercap, rapl: Reorder CPU detection table Let's make an effort to group these things by microarchitecture name. It makes it easier to see if something got missed. Signed-off-by: Dave Hansen Acked-by: Rafael J. Wysocki Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Dave Hansen Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: jacob.jun.pan@intel.com Cc: linux-pm@vger.kernel.org Link: http://lkml.kernel.org/r/20160603001937.B53A383A@viggo.jf.intel.com Signed-off-by: Ingo Molnar --- drivers/powercap/intel_rapl.c | 26 ++++++++++++++++---------- 1 file changed, 16 insertions(+), 10 deletions(-) diff --git a/drivers/powercap/intel_rapl.c b/drivers/powercap/intel_rapl.c index defa8d6c27bf..f4f85323ef78 100644 --- a/drivers/powercap/intel_rapl.c +++ b/drivers/powercap/intel_rapl.c @@ -1099,25 +1099,31 @@ static const struct rapl_defaults rapl_defaults_cht = { static const struct x86_cpu_id rapl_ids[] __initconst = { RAPL_CPU(INTEL_FAM6_SANDYBRIDGE, rapl_defaults_core), RAPL_CPU(INTEL_FAM6_SANDYBRIDGE_X, rapl_defaults_core), - RAPL_CPU(INTEL_FAM6_ATOM_SILVERMONT1, rapl_defaults_byt), + RAPL_CPU(INTEL_FAM6_IVYBRIDGE, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_HASWELL_CORE, rapl_defaults_core), - RAPL_CPU(INTEL_FAM6_BROADWELL_CORE, rapl_defaults_core), - RAPL_CPU(INTEL_FAM6_HASWELL_X, rapl_defaults_hsw_server), - RAPL_CPU(INTEL_FAM6_BROADWELL_X, rapl_defaults_hsw_server), RAPL_CPU(INTEL_FAM6_HASWELL_ULT, rapl_defaults_core), RAPL_CPU(INTEL_FAM6_HASWELL_GT3E, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_HASWELL_X, rapl_defaults_hsw_server), + + RAPL_CPU(INTEL_FAM6_BROADWELL_CORE, rapl_defaults_core), RAPL_CPU(INTEL_FAM6_BROADWELL_GT3E, rapl_defaults_core), - RAPL_CPU(INTEL_FAM6_SKYLAKE_MOBILE, rapl_defaults_core), - RAPL_CPU(INTEL_FAM6_ATOM_AIRMONT, rapl_defaults_cht), - RAPL_CPU(INTEL_FAM6_ATOM_MERRIFIELD1, rapl_defaults_tng), RAPL_CPU(INTEL_FAM6_BROADWELL_XEON_D, rapl_defaults_core), - RAPL_CPU(INTEL_FAM6_ATOM_MERRIFIELD2, rapl_defaults_ann), - RAPL_CPU(INTEL_FAM6_ATOM_GOLDMONT, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_BROADWELL_X, rapl_defaults_hsw_server), + RAPL_CPU(INTEL_FAM6_SKYLAKE_DESKTOP, rapl_defaults_core), - RAPL_CPU(INTEL_FAM6_XEON_PHI_KNL, rapl_defaults_hsw_server), + RAPL_CPU(INTEL_FAM6_SKYLAKE_MOBILE, rapl_defaults_core), RAPL_CPU(INTEL_FAM6_KABYLAKE_MOBILE, rapl_defaults_core), RAPL_CPU(INTEL_FAM6_KABYLAKE_DESKTOP, rapl_defaults_core), + + RAPL_CPU(INTEL_FAM6_ATOM_SILVERMONT1, rapl_defaults_byt), + RAPL_CPU(INTEL_FAM6_ATOM_AIRMONT, rapl_defaults_cht), + RAPL_CPU(INTEL_FAM6_ATOM_MERRIFIELD1, rapl_defaults_tng), + RAPL_CPU(INTEL_FAM6_ATOM_MERRIFIELD2, rapl_defaults_ann), + RAPL_CPU(INTEL_FAM6_ATOM_GOLDMONT, rapl_defaults_core), + + RAPL_CPU(INTEL_FAM6_XEON_PHI_KNL, rapl_defaults_hsw_server), {} }; MODULE_DEVICE_TABLE(x86cpu, rapl_ids); From d40671e30cb46e834651e0ce3d4590c915171414 Mon Sep 17 00:00:00 2001 From: Dave Hansen Date: Thu, 2 Jun 2016 17:19:55 -0700 Subject: [PATCH 18/42] x86, powercap, rapl: Add Skylake Server model number SKX uses similar RAPL interface as Broadwell server according to Jacob Pan. Signed-off-by: Dave Hansen Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Dave Hansen Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Jacob Pan Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Rafael J. Wysocki Cc: Thomas Gleixner Cc: jacob.jun.pan@intel.com Cc: linux-pm@vger.kernel.org Link: http://lkml.kernel.org/r/20160603001955.38E1E684@viggo.jf.intel.com Signed-off-by: Ingo Molnar --- drivers/powercap/intel_rapl.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/powercap/intel_rapl.c b/drivers/powercap/intel_rapl.c index f4f85323ef78..2e8f2be5b6f9 100644 --- a/drivers/powercap/intel_rapl.c +++ b/drivers/powercap/intel_rapl.c @@ -1114,6 +1114,7 @@ static const struct x86_cpu_id rapl_ids[] __initconst = { RAPL_CPU(INTEL_FAM6_SKYLAKE_DESKTOP, rapl_defaults_core), RAPL_CPU(INTEL_FAM6_SKYLAKE_MOBILE, rapl_defaults_core), + RAPL_CPU(INTEL_FAM6_SKYLAKE_X, rapl_defaults_hsw_server), RAPL_CPU(INTEL_FAM6_KABYLAKE_MOBILE, rapl_defaults_core), RAPL_CPU(INTEL_FAM6_KABYLAKE_DESKTOP, rapl_defaults_core), From db73c5a8c80decbb6ddf208e58f3865b4df5384d Mon Sep 17 00:00:00 2001 From: Dave Hansen Date: Thu, 2 Jun 2016 17:19:32 -0700 Subject: [PATCH 19/42] x86/intel_idle: Use Intel family macros for intel_idle Use the new INTEL_FAM6_* macros for intel_idle.c. Also fix up some of the macros to be consistent with how some of the intel_idle code refers to the model. There's on oddity here: model 0x1F is uniquely referred to here and nowhere else that I could find. 0x1E/0x1F are just spelled out as "Intel Core i7 and i5 Processors" in the SDM or as "Intel processors based on the Nehalem, Westmere microarchitectures" in the RDPMC section. Comments between tables 19-19 and 19-20 in the SDM seem to point to 0x1F being some kind of Westmere, so let's call it "WESTMERE2". Signed-off-by: Dave Hansen Acked-by: Rafael J. Wysocki Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Dave Hansen Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Len Brown Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: jacob.jun.pan@intel.com Cc: linux-pm@vger.kernel.org Link: http://lkml.kernel.org/r/20160603001932.EE978EB9@viggo.jf.intel.com Signed-off-by: Ingo Molnar --- drivers/idle/intel_idle.c | 71 ++++++++++++++++++++------------------- 1 file changed, 36 insertions(+), 35 deletions(-) diff --git a/drivers/idle/intel_idle.c b/drivers/idle/intel_idle.c index c96649292b55..b5dd41d13d3d 100644 --- a/drivers/idle/intel_idle.c +++ b/drivers/idle/intel_idle.c @@ -62,6 +62,7 @@ #include #include #include +#include #include #include @@ -1020,38 +1021,38 @@ static const struct idle_cpu idle_cpu_bxt = { { X86_VENDOR_INTEL, 6, model, X86_FEATURE_MWAIT, (unsigned long)&cpu } static const struct x86_cpu_id intel_idle_ids[] __initconst = { - ICPU(0x1a, idle_cpu_nehalem), - ICPU(0x1e, idle_cpu_nehalem), - ICPU(0x1f, idle_cpu_nehalem), - ICPU(0x25, idle_cpu_nehalem), - ICPU(0x2c, idle_cpu_nehalem), - ICPU(0x2e, idle_cpu_nehalem), - ICPU(0x1c, idle_cpu_atom), - ICPU(0x26, idle_cpu_lincroft), - ICPU(0x2f, idle_cpu_nehalem), - ICPU(0x2a, idle_cpu_snb), - ICPU(0x2d, idle_cpu_snb), - ICPU(0x36, idle_cpu_atom), - ICPU(0x37, idle_cpu_byt), - ICPU(0x4c, idle_cpu_cht), - ICPU(0x3a, idle_cpu_ivb), - ICPU(0x3e, idle_cpu_ivt), - ICPU(0x3c, idle_cpu_hsw), - ICPU(0x3f, idle_cpu_hsw), - ICPU(0x45, idle_cpu_hsw), - ICPU(0x46, idle_cpu_hsw), - ICPU(0x4d, idle_cpu_avn), - ICPU(0x3d, idle_cpu_bdw), - ICPU(0x47, idle_cpu_bdw), - ICPU(0x4f, idle_cpu_bdw), - ICPU(0x56, idle_cpu_bdw), - ICPU(0x4e, idle_cpu_skl), - ICPU(0x5e, idle_cpu_skl), - ICPU(0x8e, idle_cpu_skl), - ICPU(0x9e, idle_cpu_skl), - ICPU(0x55, idle_cpu_skx), - ICPU(0x57, idle_cpu_knl), - ICPU(0x5c, idle_cpu_bxt), + ICPU(INTEL_FAM6_NEHALEM_EP, idle_cpu_nehalem), + ICPU(INTEL_FAM6_NEHALEM, idle_cpu_nehalem), + ICPU(INTEL_FAM6_WESTMERE2, idle_cpu_nehalem), + ICPU(INTEL_FAM6_WESTMERE, idle_cpu_nehalem), + ICPU(INTEL_FAM6_WESTMERE_EP, idle_cpu_nehalem), + ICPU(INTEL_FAM6_NEHALEM_EX, idle_cpu_nehalem), + ICPU(INTEL_FAM6_ATOM_PINEVIEW, idle_cpu_atom), + ICPU(INTEL_FAM6_ATOM_LINCROFT, idle_cpu_lincroft), + ICPU(INTEL_FAM6_WESTMERE_EX, idle_cpu_nehalem), + ICPU(INTEL_FAM6_SANDYBRIDGE, idle_cpu_snb), + ICPU(INTEL_FAM6_SANDYBRIDGE_X, idle_cpu_snb), + ICPU(INTEL_FAM6_ATOM_CEDARVIEW, idle_cpu_atom), + ICPU(INTEL_FAM6_ATOM_SILVERMONT1, idle_cpu_byt), + ICPU(INTEL_FAM6_ATOM_AIRMONT, idle_cpu_cht), + ICPU(INTEL_FAM6_IVYBRIDGE, idle_cpu_ivb), + ICPU(INTEL_FAM6_IVYBRIDGE_X, idle_cpu_ivt), + ICPU(INTEL_FAM6_HASWELL_CORE, idle_cpu_hsw), + ICPU(INTEL_FAM6_HASWELL_X, idle_cpu_hsw), + ICPU(INTEL_FAM6_HASWELL_ULT, idle_cpu_hsw), + ICPU(INTEL_FAM6_HASWELL_GT3E, idle_cpu_hsw), + ICPU(INTEL_FAM6_ATOM_SILVERMONT2, idle_cpu_avn), + ICPU(INTEL_FAM6_BROADWELL_CORE, idle_cpu_bdw), + ICPU(INTEL_FAM6_BROADWELL_GT3E, idle_cpu_bdw), + ICPU(INTEL_FAM6_BROADWELL_X, idle_cpu_bdw), + ICPU(INTEL_FAM6_BROADWELL_XEON_D, idle_cpu_bdw), + ICPU(INTEL_FAM6_SKYLAKE_MOBILE, idle_cpu_skl), + ICPU(INTEL_FAM6_SKYLAKE_DESKTOP, idle_cpu_skl), + ICPU(INTEL_FAM6_KABYLAKE_MOBILE, idle_cpu_skl), + ICPU(INTEL_FAM6_KABYLAKE_DESKTOP, idle_cpu_skl), + ICPU(INTEL_FAM6_SKYLAKE_X, idle_cpu_skx), + ICPU(INTEL_FAM6_XEON_PHI_KNL, idle_cpu_knl), + ICPU(INTEL_FAM6_ATOM_GOLDMONT, idle_cpu_bxt), {} }; MODULE_DEVICE_TABLE(x86cpu, intel_idle_ids); @@ -1261,13 +1262,13 @@ static void intel_idle_state_table_update(void) { switch (boot_cpu_data.x86_model) { - case 0x3e: /* IVT */ + case INTEL_FAM6_IVYBRIDGE_X: ivt_idle_state_table_update(); break; - case 0x5c: /* BXT */ + case INTEL_FAM6_ATOM_GOLDMONT: bxt_idle_state_table_update(); break; - case 0x5e: /* SKL-H */ + case INTEL_FAM6_SKYLAKE_DESKTOP: sklh_idle_state_table_update(); break; } From d5e0c89a8ccde900c3245474915ea0f518abdb79 Mon Sep 17 00:00:00 2001 From: Dave Hansen Date: Thu, 2 Jun 2016 17:19:39 -0700 Subject: [PATCH 20/42] x86/platform: Use new Intel model number macros Remove the open-coded model numbers. Signed-off-by: Dave Hansen Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Dave Hansen Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Jacob Pan Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Rafael J. Wysocki Cc: Srinivas Pandruvada Cc: Thomas Gleixner Cc: jacob.jun.pan@intel.com Link: http://lkml.kernel.org/r/20160603001939.D1D7FC2F@viggo.jf.intel.com Signed-off-by: Ingo Molnar --- arch/x86/platform/atom/punit_atom_debug.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/arch/x86/platform/atom/punit_atom_debug.c b/arch/x86/platform/atom/punit_atom_debug.c index 81c769e80614..109782996867 100644 --- a/arch/x86/platform/atom/punit_atom_debug.c +++ b/arch/x86/platform/atom/punit_atom_debug.c @@ -23,6 +23,7 @@ #include #include #include +#include #include /* Power gate status reg */ @@ -143,8 +144,8 @@ static void punit_dbgfs_unregister(void) (kernel_ulong_t)&drv_data } static const struct x86_cpu_id intel_punit_cpu_ids[] = { - ICPU(55, punit_device_byt), /* Valleyview, Bay Trail */ - ICPU(76, punit_device_cht), /* Braswell, Cherry Trail */ + ICPU(INTEL_FAM6_ATOM_SILVERMONT1, punit_device_byt), + ICPU(INTEL_FAM6_ATOM_AIRMONT, punit_device_cht), {} }; From 5b20c944882ce35da0074b9eabe41a172aea030b Mon Sep 17 00:00:00 2001 From: Dave Hansen Date: Thu, 2 Jun 2016 17:19:45 -0700 Subject: [PATCH 21/42] x86/cpufreq: Use Intel family name macros for the intel_pstate cpufreq driver Another straightforward replacement of magic numbers. Signed-off-by: Dave Hansen Acked-by: Rafael J. Wysocki Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Dave Hansen Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Len Brown Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Srinivas Pandruvada Cc: Thomas Gleixner Cc: Viresh Kumar Cc: jacob.jun.pan@intel.com Cc: linux-pm@vger.kernel.org Link: http://lkml.kernel.org/r/20160603001945.0F5D02AA@viggo.jf.intel.com Signed-off-by: Ingo Molnar --- drivers/cpufreq/intel_pstate.c | 37 +++++++++++++++++----------------- 1 file changed, 19 insertions(+), 18 deletions(-) diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index 0d159b513469..9b1f5d73d306 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -35,6 +35,7 @@ #include #include #include +#include #define ATOM_RATIOS 0x66a #define ATOM_VIDS 0x66b @@ -1352,29 +1353,29 @@ static void intel_pstate_update_util(struct update_util_data *data, u64 time, (unsigned long)&policy } static const struct x86_cpu_id intel_pstate_cpu_ids[] = { - ICPU(0x2a, core_params), - ICPU(0x2d, core_params), - ICPU(0x37, silvermont_params), - ICPU(0x3a, core_params), - ICPU(0x3c, core_params), - ICPU(0x3d, core_params), - ICPU(0x3e, core_params), - ICPU(0x3f, core_params), - ICPU(0x45, core_params), - ICPU(0x46, core_params), - ICPU(0x47, core_params), - ICPU(0x4c, airmont_params), - ICPU(0x4e, core_params), - ICPU(0x4f, core_params), - ICPU(0x5e, core_params), - ICPU(0x56, core_params), - ICPU(0x57, knl_params), + ICPU(INTEL_FAM6_SANDYBRIDGE, core_params), + ICPU(INTEL_FAM6_SANDYBRIDGE_X, core_params), + ICPU(INTEL_FAM6_ATOM_SILVERMONT1, silvermont_params), + ICPU(INTEL_FAM6_IVYBRIDGE, core_params), + ICPU(INTEL_FAM6_HASWELL_CORE, core_params), + ICPU(INTEL_FAM6_BROADWELL_CORE, core_params), + ICPU(INTEL_FAM6_IVYBRIDGE_X, core_params), + ICPU(INTEL_FAM6_HASWELL_X, core_params), + ICPU(INTEL_FAM6_HASWELL_ULT, core_params), + ICPU(INTEL_FAM6_HASWELL_GT3E, core_params), + ICPU(INTEL_FAM6_BROADWELL_GT3E, core_params), + ICPU(INTEL_FAM6_ATOM_AIRMONT, airmont_params), + ICPU(INTEL_FAM6_SKYLAKE_MOBILE, core_params), + ICPU(INTEL_FAM6_BROADWELL_X, core_params), + ICPU(INTEL_FAM6_SKYLAKE_DESKTOP, core_params), + ICPU(INTEL_FAM6_BROADWELL_XEON_D, core_params), + ICPU(INTEL_FAM6_XEON_PHI_KNL, knl_params), {} }; MODULE_DEVICE_TABLE(x86cpu, intel_pstate_cpu_ids); static const struct x86_cpu_id intel_pstate_cpu_oob_ids[] = { - ICPU(0x56, core_params), + ICPU(INTEL_FAM6_BROADWELL_XEON_D, core_params), {} }; From 4626d840a1e0044e6f23d226ea8a5b96bd167636 Mon Sep 17 00:00:00 2001 From: Dave Hansen Date: Thu, 2 Jun 2016 17:19:46 -0700 Subject: [PATCH 22/42] x86/acpi/lss: Use Intel family name macros for the acpi_lpss driver Another straightforward replacement of magic numbers. Signed-off-by: Dave Hansen Acked-by: Rafael J. Wysocki Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Dave Hansen Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Len Brown Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: jacob.jun.pan@intel.com Cc: linux-acpi@vger.kernel.org Link: http://lkml.kernel.org/r/20160603001946.264CE704@viggo.jf.intel.com Signed-off-by: Ingo Molnar --- drivers/acpi/acpi_lpss.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/acpi/acpi_lpss.c b/drivers/acpi/acpi_lpss.c index 0872d5fecb82..357a0b8f860b 100644 --- a/drivers/acpi/acpi_lpss.c +++ b/drivers/acpi/acpi_lpss.c @@ -29,6 +29,7 @@ ACPI_MODULE_NAME("acpi_lpss"); #ifdef CONFIG_X86_INTEL_LPSS #include +#include #include #include @@ -229,8 +230,8 @@ static const struct lpss_device_desc bsw_spi_dev_desc = { #define ICPU(model) { X86_VENDOR_INTEL, 6, model, X86_FEATURE_ANY, } static const struct x86_cpu_id lpss_cpu_ids[] = { - ICPU(0x37), /* Valleyview, Bay Trail */ - ICPU(0x4c), /* Braswell, Cherry Trail */ + ICPU(INTEL_FAM6_ATOM_SILVERMONT1), /* Valleyview, Bay Trail */ + ICPU(INTEL_FAM6_ATOM_AIRMONT), /* Braswell, Cherry Trail */ {} }; From 678dec00a4753b74df8ad6fc5167429b614d1139 Mon Sep 17 00:00:00 2001 From: Dave Hansen Date: Thu, 2 Jun 2016 17:19:47 -0700 Subject: [PATCH 23/42] x86/intel_telemetry: Use Intel family name macros for telemetry driver Another straightforward replacement of magic numbers. Signed-off-by: Dave Hansen Acked-by: Darren Hart Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Dave Hansen Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Souvik Kumar Chakravarty Cc: Thomas Gleixner Cc: jacob.jun.pan@intel.com Cc: platform-driver-x86@vger.kernel.org Link: http://lkml.kernel.org/r/20160603001947.05102C3E@viggo.jf.intel.com Signed-off-by: Ingo Molnar --- drivers/platform/x86/intel_telemetry_debugfs.c | 3 ++- drivers/platform/x86/intel_telemetry_pltdrv.c | 3 ++- 2 files changed, 4 insertions(+), 2 deletions(-) diff --git a/drivers/platform/x86/intel_telemetry_debugfs.c b/drivers/platform/x86/intel_telemetry_debugfs.c index f5134acd6ff0..815a7c5e9566 100644 --- a/drivers/platform/x86/intel_telemetry_debugfs.c +++ b/drivers/platform/x86/intel_telemetry_debugfs.c @@ -32,6 +32,7 @@ #include #include +#include #include #include #include @@ -331,7 +332,7 @@ static struct telemetry_debugfs_conf telem_apl_debugfs_conf = { }; static const struct x86_cpu_id telemetry_debugfs_cpu_ids[] = { - TELEM_DEBUGFS_CPU(0x5c, telem_apl_debugfs_conf), + TELEM_DEBUGFS_CPU(INTEL_FAM6_ATOM_GOLDMONT, telem_apl_debugfs_conf), {} }; diff --git a/drivers/platform/x86/intel_telemetry_pltdrv.c b/drivers/platform/x86/intel_telemetry_pltdrv.c index 09c84a2b1c2c..6d884f7d1b9f 100644 --- a/drivers/platform/x86/intel_telemetry_pltdrv.c +++ b/drivers/platform/x86/intel_telemetry_pltdrv.c @@ -28,6 +28,7 @@ #include #include +#include #include #include #include @@ -163,7 +164,7 @@ static struct telemetry_plt_config telem_apl_config = { }; static const struct x86_cpu_id telemetry_cpu_ids[] = { - TELEM_CPU(0x5c, telem_apl_config), + TELEM_CPU(INTEL_FAM6_ATOM_GOLDMONT, telem_apl_config), {} }; From 8ba4cb53129c3089f248f1ebeb25128d93c8b5c5 Mon Sep 17 00:00:00 2001 From: Dave Hansen Date: Thu, 2 Jun 2016 17:19:51 -0700 Subject: [PATCH 24/42] x86, mmc: Use Intel family name macros for mmc driver Another straightforward replacement of magic numbers. Signed-off-by: Dave Hansen Acked-by: Adrian Hunter Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Dave Hansen Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: Ulf Hansson Cc: jacob.jun.pan@intel.com Cc: linux-mmc@vger.kernel.org Link: http://lkml.kernel.org/r/20160603001951.9EEA53D8@viggo.jf.intel.com Signed-off-by: Ingo Molnar --- drivers/mmc/host/sdhci-acpi.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/mmc/host/sdhci-acpi.c b/drivers/mmc/host/sdhci-acpi.c index 458ffb7637e5..008709c5cb09 100644 --- a/drivers/mmc/host/sdhci-acpi.c +++ b/drivers/mmc/host/sdhci-acpi.c @@ -43,6 +43,7 @@ #ifdef CONFIG_X86 #include +#include #include #endif @@ -126,7 +127,7 @@ static const struct sdhci_acpi_chip sdhci_acpi_chip_int = { static bool sdhci_acpi_byt(void) { static const struct x86_cpu_id byt[] = { - { X86_VENDOR_INTEL, 6, 0x37 }, + { X86_VENDOR_INTEL, 6, INTEL_FAM6_ATOM_SILVERMONT1 }, {} }; From ce53da02ebfbe93ec58dd6150b28b4606330ead5 Mon Sep 17 00:00:00 2001 From: Dave Hansen Date: Thu, 2 Jun 2016 17:19:52 -0700 Subject: [PATCH 25/42] x86, thermal: Clean up and fix CPU model detection for intel_soc_dts_thermal The X86_FAMILY_ANY in here is bogus. "BYT" and model 0x37 are family-6 only. Signed-off-by: Dave Hansen Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Dave Hansen Cc: Denys Vlasenko Cc: Eduardo Valentin Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: Zhang Rui Cc: jacob.jun.pan@intel.com Cc: linux-pm@vger.kernel.org Link: http://lkml.kernel.org/r/20160603001952.9B6E114D@viggo.jf.intel.com Signed-off-by: Ingo Molnar --- drivers/thermal/intel_soc_dts_thermal.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/thermal/intel_soc_dts_thermal.c b/drivers/thermal/intel_soc_dts_thermal.c index 4ebb31a35a64..b2bbaa1c60b0 100644 --- a/drivers/thermal/intel_soc_dts_thermal.c +++ b/drivers/thermal/intel_soc_dts_thermal.c @@ -18,6 +18,7 @@ #include #include #include +#include #include "intel_soc_dts_iosf.h" #define CRITICAL_OFFSET_FROM_TJ_MAX 5000 @@ -42,7 +43,8 @@ static irqreturn_t soc_irq_thread_fn(int irq, void *dev_data) } static const struct x86_cpu_id soc_thermal_ids[] = { - { X86_VENDOR_INTEL, X86_FAMILY_ANY, 0x37, 0, BYT_SOC_DTS_APIC_IRQ}, + { X86_VENDOR_INTEL, 6, INTEL_FAM6_ATOM_SILVERMONT1, 0, + BYT_SOC_DTS_APIC_IRQ}, {} }; MODULE_DEVICE_TABLE(x86cpu, soc_thermal_ids); From f0f879ba533ed29ca277c95d529fce703113271e Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Fri, 3 Jun 2016 10:58:46 +0530 Subject: [PATCH 26/42] cpufreq: s3c24xx: Remove useless checks These aren't required at all, remove them. Reviewed-by: Krzysztof Kozlowski Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/s3c24xx-cpufreq.c | 4 ---- 1 file changed, 4 deletions(-) diff --git a/drivers/cpufreq/s3c24xx-cpufreq.c b/drivers/cpufreq/s3c24xx-cpufreq.c index ae8eaed77b70..4567c3cab095 100644 --- a/drivers/cpufreq/s3c24xx-cpufreq.c +++ b/drivers/cpufreq/s3c24xx-cpufreq.c @@ -571,11 +571,7 @@ static int s3c_cpufreq_build_freq(void) { int size, ret; - if (!cpu_cur.info->calc_freqtable) - return -EINVAL; - kfree(ftab); - ftab = NULL; size = cpu_cur.info->calc_freqtable(&cpu_cur, NULL, 0); size++; From f8bfc116cacbdf7e0e655d8a798a242087ed70a5 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Fri, 3 Jun 2016 10:58:47 +0530 Subject: [PATCH 27/42] cpufreq: Remove cpufreq_frequency_get_table() Most of the callers of cpufreq_frequency_get_table() already have the pointer to a valid 'policy' structure and they don't really need to go through the per-cpu variable first and then a check to validate the frequency, in order to find the freq-table for the policy. Directly use the policy->freq_table field instead for them. Only one user of that API is left after above changes, cpu_cooling.c and it accesses the freq_table in a racy way as the policy can get freed in between. Fix it by using cpufreq_cpu_get() properly. Since there are no more users of cpufreq_frequency_get_table() left, get rid of it. Signed-off-by: Viresh Kumar Acked-by: Javi Merino (cpu_cooling.c) Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq.c | 38 ++++++++++----------------- drivers/cpufreq/cpufreq_ondemand.c | 2 +- drivers/cpufreq/cpufreq_stats.c | 3 +-- drivers/cpufreq/freq_table.c | 9 +++---- drivers/cpufreq/ppc_cbe_cpufreq_pmi.c | 3 +-- drivers/thermal/cpu_cooling.c | 26 +++++++++++++----- include/linux/cpufreq.h | 2 -- 7 files changed, 40 insertions(+), 43 deletions(-) diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index c6a14ba239a2..cc252eecc45a 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -126,15 +126,6 @@ struct kobject *get_governor_parent_kobj(struct cpufreq_policy *policy) } EXPORT_SYMBOL_GPL(get_governor_parent_kobj); -struct cpufreq_frequency_table *cpufreq_frequency_get_table(unsigned int cpu) -{ - struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu); - - return policy && !policy_is_inactive(policy) ? - policy->freq_table : NULL; -} -EXPORT_SYMBOL_GPL(cpufreq_frequency_get_table); - static inline u64 get_cpu_idle_time_jiffy(unsigned int cpu, u64 *wall) { u64 idle_time; @@ -1950,7 +1941,7 @@ int __cpufreq_driver_target(struct cpufreq_policy *policy, if (!cpufreq_driver->target_index) return -EINVAL; - freq_table = cpufreq_frequency_get_table(policy->cpu); + freq_table = policy->freq_table; if (unlikely(!freq_table)) { pr_err("%s: Unable to find freq_table\n", __func__); return -EINVAL; @@ -2345,26 +2336,25 @@ static struct notifier_block __refdata cpufreq_cpu_notifier = { *********************************************************************/ static int cpufreq_boost_set_sw(int state) { - struct cpufreq_frequency_table *freq_table; struct cpufreq_policy *policy; int ret = -EINVAL; for_each_active_policy(policy) { - freq_table = cpufreq_frequency_get_table(policy->cpu); - if (freq_table) { - ret = cpufreq_frequency_table_cpuinfo(policy, - freq_table); - if (ret) { - pr_err("%s: Policy frequency update failed\n", - __func__); - break; - } + if (!policy->freq_table) + continue; - down_write(&policy->rwsem); - policy->user_policy.max = policy->max; - cpufreq_governor_limits(policy); - up_write(&policy->rwsem); + ret = cpufreq_frequency_table_cpuinfo(policy, + policy->freq_table); + if (ret) { + pr_err("%s: Policy frequency update failed\n", + __func__); + break; } + + down_write(&policy->rwsem); + policy->user_policy.max = policy->max; + cpufreq_governor_limits(policy); + up_write(&policy->rwsem); } return ret; diff --git a/drivers/cpufreq/cpufreq_ondemand.c b/drivers/cpufreq/cpufreq_ondemand.c index c84fc2240d49..4d2fe2710c5d 100644 --- a/drivers/cpufreq/cpufreq_ondemand.c +++ b/drivers/cpufreq/cpufreq_ondemand.c @@ -113,7 +113,7 @@ static void ondemand_powersave_bias_init(struct cpufreq_policy *policy) { struct od_policy_dbs_info *dbs_info = to_dbs_info(policy->governor_data); - dbs_info->freq_table = cpufreq_frequency_get_table(policy->cpu); + dbs_info->freq_table = policy->freq_table; dbs_info->freq_lo = 0; } diff --git a/drivers/cpufreq/cpufreq_stats.c b/drivers/cpufreq/cpufreq_stats.c index c6e7f81a0397..06d3abdffd3a 100644 --- a/drivers/cpufreq/cpufreq_stats.c +++ b/drivers/cpufreq/cpufreq_stats.c @@ -157,11 +157,10 @@ void cpufreq_stats_create_table(struct cpufreq_policy *policy) unsigned int i = 0, count = 0, ret = -ENOMEM; struct cpufreq_stats *stats; unsigned int alloc_size; - unsigned int cpu = policy->cpu; struct cpufreq_frequency_table *pos, *table; /* We need cpufreq table for creating stats table */ - table = cpufreq_frequency_get_table(cpu); + table = policy->freq_table; if (unlikely(!table)) return; diff --git a/drivers/cpufreq/freq_table.c b/drivers/cpufreq/freq_table.c index 4e5c5dbfed7a..f52b5473b1f4 100644 --- a/drivers/cpufreq/freq_table.c +++ b/drivers/cpufreq/freq_table.c @@ -106,12 +106,10 @@ EXPORT_SYMBOL_GPL(cpufreq_frequency_table_verify); */ int cpufreq_generic_frequency_table_verify(struct cpufreq_policy *policy) { - struct cpufreq_frequency_table *table = - cpufreq_frequency_get_table(policy->cpu); - if (!table) + if (!policy->freq_table) return -ENODEV; - return cpufreq_frequency_table_verify(policy, table); + return cpufreq_frequency_table_verify(policy, policy->freq_table); } EXPORT_SYMBOL_GPL(cpufreq_generic_frequency_table_verify); @@ -210,9 +208,8 @@ EXPORT_SYMBOL_GPL(cpufreq_frequency_table_target); int cpufreq_frequency_table_get_index(struct cpufreq_policy *policy, unsigned int freq) { - struct cpufreq_frequency_table *pos, *table; + struct cpufreq_frequency_table *pos, *table = policy->freq_table; - table = cpufreq_frequency_get_table(policy->cpu); if (unlikely(!table)) { pr_debug("%s: Unable to find frequency table\n", __func__); return -ENOENT; diff --git a/drivers/cpufreq/ppc_cbe_cpufreq_pmi.c b/drivers/cpufreq/ppc_cbe_cpufreq_pmi.c index 7c4cd5c634f2..dc112481a408 100644 --- a/drivers/cpufreq/ppc_cbe_cpufreq_pmi.c +++ b/drivers/cpufreq/ppc_cbe_cpufreq_pmi.c @@ -94,7 +94,7 @@ static int pmi_notifier(struct notifier_block *nb, unsigned long event, void *data) { struct cpufreq_policy *policy = data; - struct cpufreq_frequency_table *cbe_freqs; + struct cpufreq_frequency_table *cbe_freqs = policy->freq_table; u8 node; /* Should this really be called for CPUFREQ_ADJUST and CPUFREQ_NOTIFY @@ -103,7 +103,6 @@ static int pmi_notifier(struct notifier_block *nb, if (event == CPUFREQ_START) return 0; - cbe_freqs = cpufreq_frequency_get_table(policy->cpu); node = cbe_cpu_to_node(policy->cpu); pr_debug("got notified, event=%lu, node=%u\n", event, node); diff --git a/drivers/thermal/cpu_cooling.c b/drivers/thermal/cpu_cooling.c index 6ceac4f2d4b2..4d678cfc81b1 100644 --- a/drivers/thermal/cpu_cooling.c +++ b/drivers/thermal/cpu_cooling.c @@ -787,22 +787,34 @@ __cpufreq_cooling_register(struct device_node *np, const struct cpumask *clip_cpus, u32 capacitance, get_static_t plat_static_func) { + struct cpufreq_policy *policy; struct thermal_cooling_device *cool_dev; struct cpufreq_cooling_device *cpufreq_dev; char dev_name[THERMAL_NAME_LENGTH]; struct cpufreq_frequency_table *pos, *table; + struct cpumask temp_mask; unsigned int freq, i, num_cpus; int ret; - table = cpufreq_frequency_get_table(cpumask_first(clip_cpus)); - if (!table) { - pr_debug("%s: CPUFreq table not found\n", __func__); + cpumask_and(&temp_mask, clip_cpus, cpu_online_mask); + policy = cpufreq_cpu_get(cpumask_first(&temp_mask)); + if (!policy) { + pr_debug("%s: CPUFreq policy not found\n", __func__); return ERR_PTR(-EPROBE_DEFER); } + table = policy->freq_table; + if (!table) { + pr_debug("%s: CPUFreq table not found\n", __func__); + cool_dev = ERR_PTR(-ENODEV); + goto put_policy; + } + cpufreq_dev = kzalloc(sizeof(*cpufreq_dev), GFP_KERNEL); - if (!cpufreq_dev) - return ERR_PTR(-ENOMEM); + if (!cpufreq_dev) { + cool_dev = ERR_PTR(-ENOMEM); + goto put_policy; + } num_cpus = cpumask_weight(clip_cpus); cpufreq_dev->time_in_idle = kcalloc(num_cpus, @@ -892,7 +904,7 @@ __cpufreq_cooling_register(struct device_node *np, CPUFREQ_POLICY_NOTIFIER); mutex_unlock(&cooling_cpufreq_lock); - return cool_dev; + goto put_policy; remove_idr: release_idr(&cpufreq_idr, cpufreq_dev->id); @@ -906,6 +918,8 @@ __cpufreq_cooling_register(struct device_node *np, kfree(cpufreq_dev->time_in_idle); free_cdev: kfree(cpufreq_dev); +put_policy: + cpufreq_cpu_put(policy); return cool_dev; } diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h index 7ed93c310c08..1342cbc0f25e 100644 --- a/include/linux/cpufreq.h +++ b/include/linux/cpufreq.h @@ -632,8 +632,6 @@ static inline bool policy_has_boost_freq(struct cpufreq_policy *policy) return false; } #endif -/* the following funtion is for cpufreq core use only */ -struct cpufreq_frequency_table *cpufreq_frequency_get_table(unsigned int cpu); /* the following are really really optional */ extern struct freq_attr cpufreq_freq_attr_scaling_available_freqs; From 34ac5d7a1d1da203008697e1a69cb4bdbff8684c Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Fri, 3 Jun 2016 10:58:48 +0530 Subject: [PATCH 28/42] cpufreq: ondemand: Don't keep a copy of freq_table pointer There is absolutely no need to keep a copy to the freq-table in 'struct od_policy_dbs_info'. Use policy->freq_table instead. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/amd_freq_sensitivity.c | 7 +++---- drivers/cpufreq/cpufreq_ondemand.c | 22 +++++++++++----------- drivers/cpufreq/cpufreq_ondemand.h | 1 - 3 files changed, 14 insertions(+), 16 deletions(-) diff --git a/drivers/cpufreq/amd_freq_sensitivity.c b/drivers/cpufreq/amd_freq_sensitivity.c index 404360cad25c..bc86816693a8 100644 --- a/drivers/cpufreq/amd_freq_sensitivity.c +++ b/drivers/cpufreq/amd_freq_sensitivity.c @@ -48,9 +48,8 @@ static unsigned int amd_powersave_bias_target(struct cpufreq_policy *policy, struct policy_dbs_info *policy_dbs = policy->governor_data; struct dbs_data *od_data = policy_dbs->dbs_data; struct od_dbs_tuners *od_tuners = od_data->tuners; - struct od_policy_dbs_info *od_info = to_dbs_info(policy_dbs); - if (!od_info->freq_table) + if (!policy->freq_table) return freq_next; rdmsr_on_cpu(policy->cpu, MSR_AMD64_FREQ_SENSITIVITY_ACTUAL, @@ -93,9 +92,9 @@ static unsigned int amd_powersave_bias_target(struct cpufreq_policy *policy, unsigned int index; cpufreq_frequency_table_target(policy, - od_info->freq_table, policy->cur - 1, + policy->freq_table, policy->cur - 1, CPUFREQ_RELATION_H, &index); - freq_next = od_info->freq_table[index].frequency; + freq_next = policy->freq_table[index].frequency; } data->freq_prev = freq_next; diff --git a/drivers/cpufreq/cpufreq_ondemand.c b/drivers/cpufreq/cpufreq_ondemand.c index 4d2fe2710c5d..528353f204fd 100644 --- a/drivers/cpufreq/cpufreq_ondemand.c +++ b/drivers/cpufreq/cpufreq_ondemand.c @@ -71,28 +71,29 @@ static unsigned int generic_powersave_bias_target(struct cpufreq_policy *policy, struct od_policy_dbs_info *dbs_info = to_dbs_info(policy_dbs); struct dbs_data *dbs_data = policy_dbs->dbs_data; struct od_dbs_tuners *od_tuners = dbs_data->tuners; + struct cpufreq_frequency_table *freq_table = policy->freq_table; - if (!dbs_info->freq_table) { + if (!freq_table) { dbs_info->freq_lo = 0; dbs_info->freq_lo_delay_us = 0; return freq_next; } - cpufreq_frequency_table_target(policy, dbs_info->freq_table, freq_next, - relation, &index); - freq_req = dbs_info->freq_table[index].frequency; + cpufreq_frequency_table_target(policy, freq_table, freq_next, relation, + &index); + freq_req = freq_table[index].frequency; freq_reduc = freq_req * od_tuners->powersave_bias / 1000; freq_avg = freq_req - freq_reduc; /* Find freq bounds for freq_avg in freq_table */ index = 0; - cpufreq_frequency_table_target(policy, dbs_info->freq_table, freq_avg, - CPUFREQ_RELATION_H, &index); - freq_lo = dbs_info->freq_table[index].frequency; + cpufreq_frequency_table_target(policy, freq_table, freq_avg, + CPUFREQ_RELATION_H, &index); + freq_lo = freq_table[index].frequency; index = 0; - cpufreq_frequency_table_target(policy, dbs_info->freq_table, freq_avg, - CPUFREQ_RELATION_L, &index); - freq_hi = dbs_info->freq_table[index].frequency; + cpufreq_frequency_table_target(policy, freq_table, freq_avg, + CPUFREQ_RELATION_L, &index); + freq_hi = freq_table[index].frequency; /* Find out how long we have to be in hi and lo freqs */ if (freq_hi == freq_lo) { @@ -113,7 +114,6 @@ static void ondemand_powersave_bias_init(struct cpufreq_policy *policy) { struct od_policy_dbs_info *dbs_info = to_dbs_info(policy->governor_data); - dbs_info->freq_table = policy->freq_table; dbs_info->freq_lo = 0; } diff --git a/drivers/cpufreq/cpufreq_ondemand.h b/drivers/cpufreq/cpufreq_ondemand.h index f0121db3cd9e..640ea4e97106 100644 --- a/drivers/cpufreq/cpufreq_ondemand.h +++ b/drivers/cpufreq/cpufreq_ondemand.h @@ -13,7 +13,6 @@ struct od_policy_dbs_info { struct policy_dbs_info policy_dbs; - struct cpufreq_frequency_table *freq_table; unsigned int freq_lo; unsigned int freq_lo_delay_us; unsigned int freq_hi_delay_us; From 7ab4aabbaa98fbf9f8967fd93162d675439337e1 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Fri, 3 Jun 2016 10:58:49 +0530 Subject: [PATCH 29/42] cpufreq: Drop freq-table param to cpufreq_frequency_table_target() The policy already has this pointer set, use it instead. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- Documentation/cpu-freq/cpu-drivers.txt | 1 - drivers/cpufreq/amd_freq_sensitivity.c | 3 +-- drivers/cpufreq/cpufreq.c | 4 ++-- drivers/cpufreq/cpufreq_ondemand.c | 11 +++++------ drivers/cpufreq/freq_table.c | 2 +- drivers/cpufreq/powernv-cpufreq.c | 3 +-- drivers/cpufreq/s3c24xx-cpufreq.c | 11 +++++------ drivers/cpufreq/s5pv210-cpufreq.c | 3 +-- include/linux/cpufreq.h | 1 - 9 files changed, 16 insertions(+), 23 deletions(-) diff --git a/Documentation/cpu-freq/cpu-drivers.txt b/Documentation/cpu-freq/cpu-drivers.txt index decbb8cb5573..74e812f0719c 100644 --- a/Documentation/cpu-freq/cpu-drivers.txt +++ b/Documentation/cpu-freq/cpu-drivers.txt @@ -244,7 +244,6 @@ policy->max, and all other criteria are met. This is helpful for the ->verify call. int cpufreq_frequency_table_target(struct cpufreq_policy *policy, - struct cpufreq_frequency_table *table, unsigned int target_freq, unsigned int relation, unsigned int *index); diff --git a/drivers/cpufreq/amd_freq_sensitivity.c b/drivers/cpufreq/amd_freq_sensitivity.c index bc86816693a8..3bea1bb791a9 100644 --- a/drivers/cpufreq/amd_freq_sensitivity.c +++ b/drivers/cpufreq/amd_freq_sensitivity.c @@ -92,8 +92,7 @@ static unsigned int amd_powersave_bias_target(struct cpufreq_policy *policy, unsigned int index; cpufreq_frequency_table_target(policy, - policy->freq_table, policy->cur - 1, - CPUFREQ_RELATION_H, &index); + policy->cur - 1, CPUFREQ_RELATION_H, &index); freq_next = policy->freq_table[index].frequency; } diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index cc252eecc45a..a28144697128 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -1947,8 +1947,8 @@ int __cpufreq_driver_target(struct cpufreq_policy *policy, return -EINVAL; } - retval = cpufreq_frequency_table_target(policy, freq_table, target_freq, - relation, &index); + retval = cpufreq_frequency_table_target(policy, target_freq, relation, + &index); if (unlikely(retval)) { pr_err("%s: Unable to find matching freq\n", __func__); return retval; diff --git a/drivers/cpufreq/cpufreq_ondemand.c b/drivers/cpufreq/cpufreq_ondemand.c index 528353f204fd..2ee476f5a2bd 100644 --- a/drivers/cpufreq/cpufreq_ondemand.c +++ b/drivers/cpufreq/cpufreq_ondemand.c @@ -79,20 +79,19 @@ static unsigned int generic_powersave_bias_target(struct cpufreq_policy *policy, return freq_next; } - cpufreq_frequency_table_target(policy, freq_table, freq_next, relation, - &index); + cpufreq_frequency_table_target(policy, freq_next, relation, &index); freq_req = freq_table[index].frequency; freq_reduc = freq_req * od_tuners->powersave_bias / 1000; freq_avg = freq_req - freq_reduc; /* Find freq bounds for freq_avg in freq_table */ index = 0; - cpufreq_frequency_table_target(policy, freq_table, freq_avg, - CPUFREQ_RELATION_H, &index); + cpufreq_frequency_table_target(policy, freq_avg, CPUFREQ_RELATION_H, + &index); freq_lo = freq_table[index].frequency; index = 0; - cpufreq_frequency_table_target(policy, freq_table, freq_avg, - CPUFREQ_RELATION_L, &index); + cpufreq_frequency_table_target(policy, freq_avg, CPUFREQ_RELATION_L, + &index); freq_hi = freq_table[index].frequency; /* Find out how long we have to be in hi and lo freqs */ diff --git a/drivers/cpufreq/freq_table.c b/drivers/cpufreq/freq_table.c index f52b5473b1f4..f145b64649ef 100644 --- a/drivers/cpufreq/freq_table.c +++ b/drivers/cpufreq/freq_table.c @@ -114,7 +114,6 @@ int cpufreq_generic_frequency_table_verify(struct cpufreq_policy *policy) EXPORT_SYMBOL_GPL(cpufreq_generic_frequency_table_verify); int cpufreq_frequency_table_target(struct cpufreq_policy *policy, - struct cpufreq_frequency_table *table, unsigned int target_freq, unsigned int relation, unsigned int *index) @@ -128,6 +127,7 @@ int cpufreq_frequency_table_target(struct cpufreq_policy *policy, .frequency = 0, }; struct cpufreq_frequency_table *pos; + struct cpufreq_frequency_table *table = policy->freq_table; unsigned int freq, diff, i = 0; pr_debug("request for target %u kHz (relation: %u) for cpu %u\n", diff --git a/drivers/cpufreq/powernv-cpufreq.c b/drivers/cpufreq/powernv-cpufreq.c index 54c45368e3f1..bf267c2dfe20 100644 --- a/drivers/cpufreq/powernv-cpufreq.c +++ b/drivers/cpufreq/powernv-cpufreq.c @@ -760,8 +760,7 @@ void powernv_cpufreq_work_fn(struct work_struct *work) struct cpufreq_policy policy; cpufreq_get_policy(&policy, cpu); - cpufreq_frequency_table_target(&policy, policy.freq_table, - policy.cur, + cpufreq_frequency_table_target(&policy, policy.cur, CPUFREQ_RELATION_C, &index); powernv_cpufreq_target_index(&policy, index); cpumask_andnot(&mask, &mask, policy.cpus); diff --git a/drivers/cpufreq/s3c24xx-cpufreq.c b/drivers/cpufreq/s3c24xx-cpufreq.c index 4567c3cab095..05a9737278f3 100644 --- a/drivers/cpufreq/s3c24xx-cpufreq.c +++ b/drivers/cpufreq/s3c24xx-cpufreq.c @@ -293,9 +293,8 @@ static int s3c_cpufreq_target(struct cpufreq_policy *policy, __func__, policy, target_freq, relation); if (ftab) { - if (cpufreq_frequency_table_target(policy, ftab, - target_freq, relation, - &index)) { + if (cpufreq_frequency_table_target(policy, target_freq, + relation, &index)) { s3c_freq_dbg("%s: table failed\n", __func__); return -EINVAL; } @@ -323,14 +322,14 @@ static int s3c_cpufreq_target(struct cpufreq_policy *policy, tmp_policy.min = policy->min * 1000; tmp_policy.max = policy->max * 1000; tmp_policy.cpu = policy->cpu; + tmp_policy.freq_table = pll_reg; /* cpufreq_frequency_table_target uses a pointer to 'index' * which is the number of the table entry, not the value of * the table entry's index field. */ - ret = cpufreq_frequency_table_target(&tmp_policy, pll_reg, - target_freq, relation, - &index); + ret = cpufreq_frequency_table_target(&tmp_policy, target_freq, + relation, &index); if (ret < 0) { pr_err("%s: no PLL available\n", __func__); diff --git a/drivers/cpufreq/s5pv210-cpufreq.c b/drivers/cpufreq/s5pv210-cpufreq.c index 06d85917b6d5..6b0cfc3b8c46 100644 --- a/drivers/cpufreq/s5pv210-cpufreq.c +++ b/drivers/cpufreq/s5pv210-cpufreq.c @@ -246,8 +246,7 @@ static int s5pv210_target(struct cpufreq_policy *policy, unsigned int index) new_freq = s5pv210_freq_table[index].frequency; /* Finding current running level index */ - if (cpufreq_frequency_table_target(policy, s5pv210_freq_table, - old_freq, CPUFREQ_RELATION_H, + if (cpufreq_frequency_table_target(policy, old_freq, CPUFREQ_RELATION_H, &priv_index)) { ret = -EINVAL; goto exit; diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h index 1342cbc0f25e..bdd7f0c035ae 100644 --- a/include/linux/cpufreq.h +++ b/include/linux/cpufreq.h @@ -598,7 +598,6 @@ int cpufreq_frequency_table_verify(struct cpufreq_policy *policy, int cpufreq_generic_frequency_table_verify(struct cpufreq_policy *policy); int cpufreq_frequency_table_target(struct cpufreq_policy *policy, - struct cpufreq_frequency_table *table, unsigned int target_freq, unsigned int relation, unsigned int *index); From 2372784542cb7a8b256a740840abf4503de0d669 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Fri, 3 Jun 2016 10:58:50 +0530 Subject: [PATCH 30/42] cpufreq: Drop 'freq_table' argument of __target_index() It is already present as part of the policy and so no need to pass it from the caller. Also, 'freq_table' is guaranteed to be valid in this function and so no need to check it. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq.c | 21 +++++++-------------- 1 file changed, 7 insertions(+), 14 deletions(-) diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index a28144697128..1685e930770f 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -1852,14 +1852,17 @@ static int __target_intermediate(struct cpufreq_policy *policy, return ret; } -static int __target_index(struct cpufreq_policy *policy, - struct cpufreq_frequency_table *freq_table, int index) +static int __target_index(struct cpufreq_policy *policy, int index) { struct cpufreq_freqs freqs = {.old = policy->cur, .flags = 0}; unsigned int intermediate_freq = 0; + unsigned int newfreq = policy->freq_table[index].frequency; int retval = -EINVAL; bool notify; + if (newfreq == policy->cur) + return 0; + notify = !(cpufreq_driver->flags & CPUFREQ_ASYNC_NOTIFICATION); if (notify) { /* Handle switching to intermediate frequency */ @@ -1874,7 +1877,7 @@ static int __target_index(struct cpufreq_policy *policy, freqs.old = freqs.new; } - freqs.new = freq_table[index].frequency; + freqs.new = newfreq; pr_debug("%s: cpu: %d, oldfreq: %u, new freq: %u\n", __func__, policy->cpu, freqs.old, freqs.new); @@ -1911,7 +1914,6 @@ int __cpufreq_driver_target(struct cpufreq_policy *policy, unsigned int relation) { unsigned int old_target_freq = target_freq; - struct cpufreq_frequency_table *freq_table; int index, retval; if (cpufreq_disabled()) @@ -1941,12 +1943,6 @@ int __cpufreq_driver_target(struct cpufreq_policy *policy, if (!cpufreq_driver->target_index) return -EINVAL; - freq_table = policy->freq_table; - if (unlikely(!freq_table)) { - pr_err("%s: Unable to find freq_table\n", __func__); - return -EINVAL; - } - retval = cpufreq_frequency_table_target(policy, target_freq, relation, &index); if (unlikely(retval)) { @@ -1954,10 +1950,7 @@ int __cpufreq_driver_target(struct cpufreq_policy *policy, return retval; } - if (freq_table[index].frequency == policy->cur) - return 0; - - return __target_index(policy, freq_table, index); + return __target_index(policy, index); } EXPORT_SYMBOL_GPL(__cpufreq_driver_target); From d218ed773915a2b762d132be6bb765637338c360 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Fri, 3 Jun 2016 10:58:51 +0530 Subject: [PATCH 31/42] cpufreq: Return index from cpufreq_frequency_table_target() This routine can't fail unless the frequency table is invalid and doesn't contain any valid entries. Make it return the index and WARN() in case it is used for an invalid table. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- Documentation/cpu-freq/cpu-drivers.txt | 7 +++---- drivers/cpufreq/amd_freq_sensitivity.c | 4 ++-- drivers/cpufreq/cpufreq.c | 9 ++------- drivers/cpufreq/cpufreq_ondemand.c | 14 ++++++-------- drivers/cpufreq/freq_table.c | 24 +++++++++++++----------- drivers/cpufreq/powernv-cpufreq.c | 4 ++-- drivers/cpufreq/s3c24xx-cpufreq.c | 26 ++++++-------------------- drivers/cpufreq/s5pv210-cpufreq.c | 7 ++----- include/linux/cpufreq.h | 3 +-- 9 files changed, 37 insertions(+), 61 deletions(-) diff --git a/Documentation/cpu-freq/cpu-drivers.txt b/Documentation/cpu-freq/cpu-drivers.txt index 74e812f0719c..772b94fde264 100644 --- a/Documentation/cpu-freq/cpu-drivers.txt +++ b/Documentation/cpu-freq/cpu-drivers.txt @@ -245,12 +245,11 @@ policy->max, and all other criteria are met. This is helpful for the int cpufreq_frequency_table_target(struct cpufreq_policy *policy, unsigned int target_freq, - unsigned int relation, - unsigned int *index); + unsigned int relation); is the corresponding frequency table helper for the ->target -stage. Just pass the values to this function, and the unsigned int -index returns the number of the frequency table entry which contains +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. The following macros can be used as iterators over cpufreq_frequency_table: diff --git a/drivers/cpufreq/amd_freq_sensitivity.c b/drivers/cpufreq/amd_freq_sensitivity.c index 3bea1bb791a9..6d5dc04c3a37 100644 --- a/drivers/cpufreq/amd_freq_sensitivity.c +++ b/drivers/cpufreq/amd_freq_sensitivity.c @@ -91,8 +91,8 @@ static unsigned int amd_powersave_bias_target(struct cpufreq_policy *policy, else { unsigned int index; - cpufreq_frequency_table_target(policy, - policy->cur - 1, CPUFREQ_RELATION_H, &index); + index = cpufreq_frequency_table_target(policy, + policy->cur - 1, CPUFREQ_RELATION_H); freq_next = policy->freq_table[index].frequency; } diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index 1685e930770f..07c933c6c29a 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -1914,7 +1914,7 @@ int __cpufreq_driver_target(struct cpufreq_policy *policy, unsigned int relation) { unsigned int old_target_freq = target_freq; - int index, retval; + int index; if (cpufreq_disabled()) return -ENODEV; @@ -1943,12 +1943,7 @@ int __cpufreq_driver_target(struct cpufreq_policy *policy, if (!cpufreq_driver->target_index) return -EINVAL; - retval = cpufreq_frequency_table_target(policy, target_freq, relation, - &index); - if (unlikely(retval)) { - pr_err("%s: Unable to find matching freq\n", __func__); - return retval; - } + index = cpufreq_frequency_table_target(policy, target_freq, relation); return __target_index(policy, index); } diff --git a/drivers/cpufreq/cpufreq_ondemand.c b/drivers/cpufreq/cpufreq_ondemand.c index 2ee476f5a2bd..0c93cd9dee99 100644 --- a/drivers/cpufreq/cpufreq_ondemand.c +++ b/drivers/cpufreq/cpufreq_ondemand.c @@ -65,7 +65,7 @@ static unsigned int generic_powersave_bias_target(struct cpufreq_policy *policy, { unsigned int freq_req, freq_reduc, freq_avg; unsigned int freq_hi, freq_lo; - unsigned int index = 0; + unsigned int index; unsigned int delay_hi_us; struct policy_dbs_info *policy_dbs = policy->governor_data; struct od_policy_dbs_info *dbs_info = to_dbs_info(policy_dbs); @@ -79,19 +79,17 @@ static unsigned int generic_powersave_bias_target(struct cpufreq_policy *policy, return freq_next; } - cpufreq_frequency_table_target(policy, freq_next, relation, &index); + index = cpufreq_frequency_table_target(policy, freq_next, relation); freq_req = freq_table[index].frequency; freq_reduc = freq_req * od_tuners->powersave_bias / 1000; freq_avg = freq_req - freq_reduc; /* Find freq bounds for freq_avg in freq_table */ - index = 0; - cpufreq_frequency_table_target(policy, freq_avg, CPUFREQ_RELATION_H, - &index); + index = cpufreq_frequency_table_target(policy, freq_avg, + CPUFREQ_RELATION_H); freq_lo = freq_table[index].frequency; - index = 0; - cpufreq_frequency_table_target(policy, freq_avg, CPUFREQ_RELATION_L, - &index); + index = cpufreq_frequency_table_target(policy, freq_avg, + CPUFREQ_RELATION_L); freq_hi = freq_table[index].frequency; /* Find out how long we have to be in hi and lo freqs */ diff --git a/drivers/cpufreq/freq_table.c b/drivers/cpufreq/freq_table.c index f145b64649ef..eac8bcbdaad1 100644 --- a/drivers/cpufreq/freq_table.c +++ b/drivers/cpufreq/freq_table.c @@ -114,9 +114,8 @@ int cpufreq_generic_frequency_table_verify(struct cpufreq_policy *policy) EXPORT_SYMBOL_GPL(cpufreq_generic_frequency_table_verify); int cpufreq_frequency_table_target(struct cpufreq_policy *policy, - unsigned int target_freq, - unsigned int relation, - unsigned int *index) + unsigned int target_freq, + unsigned int relation) { struct cpufreq_frequency_table optimal = { .driver_data = ~0, @@ -129,6 +128,7 @@ int cpufreq_frequency_table_target(struct cpufreq_policy *policy, struct cpufreq_frequency_table *pos; struct cpufreq_frequency_table *table = policy->freq_table; unsigned int freq, diff, i = 0; + int index; pr_debug("request for target %u kHz (relation: %u) for cpu %u\n", target_freq, relation, policy->cpu); @@ -192,16 +192,18 @@ int cpufreq_frequency_table_target(struct cpufreq_policy *policy, } } if (optimal.driver_data > i) { - if (suboptimal.driver_data > i) - return -EINVAL; - *index = suboptimal.driver_data; + if (suboptimal.driver_data > i) { + WARN(1, "Invalid frequency table: %d\n", policy->cpu); + return 0; + } + + index = suboptimal.driver_data; } else - *index = optimal.driver_data; + index = optimal.driver_data; - pr_debug("target index is %u, freq is:%u kHz\n", *index, - table[*index].frequency); - - return 0; + pr_debug("target index is %u, freq is:%u kHz\n", index, + table[index].frequency); + return index; } EXPORT_SYMBOL_GPL(cpufreq_frequency_table_target); diff --git a/drivers/cpufreq/powernv-cpufreq.c b/drivers/cpufreq/powernv-cpufreq.c index bf267c2dfe20..b29c5c20c3a1 100644 --- a/drivers/cpufreq/powernv-cpufreq.c +++ b/drivers/cpufreq/powernv-cpufreq.c @@ -760,8 +760,8 @@ void powernv_cpufreq_work_fn(struct work_struct *work) struct cpufreq_policy policy; cpufreq_get_policy(&policy, cpu); - cpufreq_frequency_table_target(&policy, policy.cur, - CPUFREQ_RELATION_C, &index); + index = cpufreq_frequency_table_target(&policy, policy.cur, + CPUFREQ_RELATION_C); powernv_cpufreq_target_index(&policy, index); cpumask_andnot(&mask, &mask, policy.cpus); } diff --git a/drivers/cpufreq/s3c24xx-cpufreq.c b/drivers/cpufreq/s3c24xx-cpufreq.c index 05a9737278f3..7b596fa38ad2 100644 --- a/drivers/cpufreq/s3c24xx-cpufreq.c +++ b/drivers/cpufreq/s3c24xx-cpufreq.c @@ -293,11 +293,8 @@ static int s3c_cpufreq_target(struct cpufreq_policy *policy, __func__, policy, target_freq, relation); if (ftab) { - if (cpufreq_frequency_table_target(policy, target_freq, - relation, &index)) { - s3c_freq_dbg("%s: table failed\n", __func__); - return -EINVAL; - } + index = cpufreq_frequency_table_target(policy, target_freq, + relation); s3c_freq_dbg("%s: adjust %d to entry %d (%u)\n", __func__, target_freq, index, ftab[index].frequency); @@ -314,7 +311,6 @@ static int s3c_cpufreq_target(struct cpufreq_policy *policy, pll = NULL; } else { struct cpufreq_policy tmp_policy; - int ret; /* we keep the cpu pll table in Hz, to ensure we get an * accurate value for the PLL output. */ @@ -324,18 +320,12 @@ static int s3c_cpufreq_target(struct cpufreq_policy *policy, tmp_policy.cpu = policy->cpu; tmp_policy.freq_table = pll_reg; - /* cpufreq_frequency_table_target uses a pointer to 'index' - * which is the number of the table entry, not the value of + /* cpufreq_frequency_table_target returns the index + * of the table entry, not the value of * the table entry's index field. */ - ret = cpufreq_frequency_table_target(&tmp_policy, target_freq, - relation, &index); - - if (ret < 0) { - pr_err("%s: no PLL available\n", __func__); - goto err_notpossible; - } - + index = cpufreq_frequency_table_target(&tmp_policy, target_freq, + relation); pll = pll_reg + index; s3c_freq_dbg("%s: target %u => %u\n", @@ -345,10 +335,6 @@ static int s3c_cpufreq_target(struct cpufreq_policy *policy, } return s3c_cpufreq_settarget(policy, target_freq, pll); - - err_notpossible: - pr_err("no compatible settings for %d\n", target_freq); - return -EINVAL; } struct clk *s3c_cpufreq_clk_get(struct device *dev, const char *name) diff --git a/drivers/cpufreq/s5pv210-cpufreq.c b/drivers/cpufreq/s5pv210-cpufreq.c index 6b0cfc3b8c46..4f4e9df9b7fc 100644 --- a/drivers/cpufreq/s5pv210-cpufreq.c +++ b/drivers/cpufreq/s5pv210-cpufreq.c @@ -246,11 +246,8 @@ static int s5pv210_target(struct cpufreq_policy *policy, unsigned int index) new_freq = s5pv210_freq_table[index].frequency; /* Finding current running level index */ - if (cpufreq_frequency_table_target(policy, old_freq, CPUFREQ_RELATION_H, - &priv_index)) { - ret = -EINVAL; - goto exit; - } + priv_index = cpufreq_frequency_table_target(policy, old_freq, + CPUFREQ_RELATION_H); arm_volt = dvs_conf[index].arm_volt; int_volt = dvs_conf[index].int_volt; diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h index bdd7f0c035ae..c378776628b4 100644 --- a/include/linux/cpufreq.h +++ b/include/linux/cpufreq.h @@ -599,8 +599,7 @@ int cpufreq_generic_frequency_table_verify(struct cpufreq_policy *policy); int cpufreq_frequency_table_target(struct cpufreq_policy *policy, unsigned int target_freq, - unsigned int relation, - unsigned int *index); + unsigned int relation); int cpufreq_frequency_table_get_index(struct cpufreq_policy *policy, unsigned int freq); From 9d4de2904ef0add3006df8896bfd07f534c44dd8 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Fri, 3 Jun 2016 10:58:52 +0530 Subject: [PATCH 32/42] cpufreq: davinci: Reuse cpufreq_generic_frequency_table_verify() policy->freq_table will always be valid for this platform, otherwise driver's probe() would fail. And so this routine will *always* return after calling cpufreq_frequency_table_verify(). This can be done using the generic callback provided by core, lets use it instead. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/davinci-cpufreq.c | 22 +--------------------- 1 file changed, 1 insertion(+), 21 deletions(-) diff --git a/drivers/cpufreq/davinci-cpufreq.c b/drivers/cpufreq/davinci-cpufreq.c index 7e336d20c184..b95a872800ec 100644 --- a/drivers/cpufreq/davinci-cpufreq.c +++ b/drivers/cpufreq/davinci-cpufreq.c @@ -38,26 +38,6 @@ struct davinci_cpufreq { }; static struct davinci_cpufreq cpufreq; -static int davinci_verify_speed(struct cpufreq_policy *policy) -{ - struct davinci_cpufreq_config *pdata = cpufreq.dev->platform_data; - struct cpufreq_frequency_table *freq_table = pdata->freq_table; - struct clk *armclk = cpufreq.armclk; - - if (freq_table) - return cpufreq_frequency_table_verify(policy, freq_table); - - if (policy->cpu) - return -EINVAL; - - cpufreq_verify_within_cpu_limits(policy); - policy->min = clk_round_rate(armclk, policy->min * 1000) / 1000; - policy->max = clk_round_rate(armclk, policy->max * 1000) / 1000; - cpufreq_verify_within_limits(policy, policy->cpuinfo.min_freq, - policy->cpuinfo.max_freq); - return 0; -} - static int davinci_target(struct cpufreq_policy *policy, unsigned int idx) { struct davinci_cpufreq_config *pdata = cpufreq.dev->platform_data; @@ -121,7 +101,7 @@ static int davinci_cpu_init(struct cpufreq_policy *policy) static struct cpufreq_driver davinci_driver = { .flags = CPUFREQ_STICKY | CPUFREQ_NEED_INITIAL_FREQ_CHECK, - .verify = davinci_verify_speed, + .verify = cpufreq_generic_frequency_table_verify, .target_index = davinci_target, .get = cpufreq_generic_get, .init = davinci_cpu_init, From f6709b8aa78fb6765c443ad6b70fdaf48b89d95d Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Thu, 9 Jun 2016 01:45:32 +0200 Subject: [PATCH 33/42] cpufreq: governor: Drop gov_cancel_work() There's no reason for gov_cancel_work() to exist at all, as it only has one caller and the only thing done by that caller is to invoke gov_cancel_work(). Accordingly, drop gov_cancel_work() and move its contents to the caller. Signed-off-by: Rafael J. Wysocki Acked-by: Viresh Kumar --- drivers/cpufreq/cpufreq_governor.c | 19 +++++++------------ 1 file changed, 7 insertions(+), 12 deletions(-) diff --git a/drivers/cpufreq/cpufreq_governor.c b/drivers/cpufreq/cpufreq_governor.c index 31369e247192..e415349ab31b 100644 --- a/drivers/cpufreq/cpufreq_governor.c +++ b/drivers/cpufreq/cpufreq_governor.c @@ -336,17 +336,6 @@ static inline void gov_clear_update_util(struct cpufreq_policy *policy) synchronize_sched(); } -static void gov_cancel_work(struct cpufreq_policy *policy) -{ - struct policy_dbs_info *policy_dbs = policy->governor_data; - - gov_clear_update_util(policy_dbs->policy); - irq_work_sync(&policy_dbs->irq_work); - cancel_work_sync(&policy_dbs->work); - atomic_set(&policy_dbs->work_count, 0); - policy_dbs->work_in_progress = false; -} - static struct policy_dbs_info *alloc_policy_dbs_info(struct cpufreq_policy *policy, struct dbs_governor *gov) { @@ -544,7 +533,13 @@ EXPORT_SYMBOL_GPL(cpufreq_dbs_governor_start); void cpufreq_dbs_governor_stop(struct cpufreq_policy *policy) { - gov_cancel_work(policy); + struct policy_dbs_info *policy_dbs = policy->governor_data; + + gov_clear_update_util(policy_dbs->policy); + irq_work_sync(&policy_dbs->irq_work); + cancel_work_sync(&policy_dbs->work); + atomic_set(&policy_dbs->work_count, 0); + policy_dbs->work_in_progress = false; } EXPORT_SYMBOL_GPL(cpufreq_dbs_governor_stop); From d352cf47d93e39494b44b792cca8d35a3a0bd9b3 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Mon, 13 Jun 2016 23:33:49 +0200 Subject: [PATCH 34/42] cpufreq: conservative: Do not use transition notifications The conservative governor registers a transition notifier so it can update its internal requested_freq value if it falls out of the policy->min...policy->max range, but requested_freq is not really necessary. That value is used to track the frequency requested by the governor previously, but policy->cur can be used instead of it and then the governor will not have to worry about updating the tracked value when the current frequency changes independently (for example, as a result of min or max changes). Accodringly, drop requested_freq from struct cs_policy_dbs_info and modify cs_dbs_timer() to use policy->cur instead of it. While at it, notice that __cpufreq_driver_target() clamps its target_freq argument between policy->min and policy->max, so the callers of it don't have to do that and make additional changes in cs_dbs_timer() in accordance with that. After these changes the transition notifier used by the conservative governor is not necessary any more, so drop it, which also makes it possible to drop the struct cs_governor definition and simplify the code accordingly. Signed-off-by: Rafael J. Wysocki Acked-by: Viresh Kumar --- drivers/cpufreq/cpufreq_conservative.c | 103 +++++-------------------- 1 file changed, 21 insertions(+), 82 deletions(-) diff --git a/drivers/cpufreq/cpufreq_conservative.c b/drivers/cpufreq/cpufreq_conservative.c index f967ec6c5720..18da4f8051d3 100644 --- a/drivers/cpufreq/cpufreq_conservative.c +++ b/drivers/cpufreq/cpufreq_conservative.c @@ -17,7 +17,6 @@ struct cs_policy_dbs_info { struct policy_dbs_info policy_dbs; unsigned int down_skip; - unsigned int requested_freq; }; static inline struct cs_policy_dbs_info *to_dbs_info(struct policy_dbs_info *policy_dbs) @@ -75,19 +74,17 @@ static unsigned int cs_dbs_timer(struct cpufreq_policy *policy) /* Check for frequency increase */ if (load > dbs_data->up_threshold) { + unsigned int requested_freq = policy->cur; + dbs_info->down_skip = 0; /* if we are already at full speed then break out early */ - if (dbs_info->requested_freq == policy->max) + if (requested_freq == policy->max) goto out; - dbs_info->requested_freq += get_freq_target(cs_tuners, policy); + requested_freq += get_freq_target(cs_tuners, policy); - if (dbs_info->requested_freq > policy->max) - dbs_info->requested_freq = policy->max; - - __cpufreq_driver_target(policy, dbs_info->requested_freq, - CPUFREQ_RELATION_H); + __cpufreq_driver_target(policy, requested_freq, CPUFREQ_RELATION_H); goto out; } @@ -98,34 +95,26 @@ static unsigned int cs_dbs_timer(struct cpufreq_policy *policy) /* Check for frequency decrease */ if (load < cs_tuners->down_threshold) { - unsigned int freq_target; + unsigned int freq_target, requested_freq = policy->cur; /* * if we cannot reduce the frequency anymore, break out early */ - if (policy->cur == policy->min) + if (requested_freq == policy->min) goto out; freq_target = get_freq_target(cs_tuners, policy); - if (dbs_info->requested_freq > freq_target) - dbs_info->requested_freq -= freq_target; + if (requested_freq > freq_target) + requested_freq -= freq_target; else - dbs_info->requested_freq = policy->min; + requested_freq = policy->min; - __cpufreq_driver_target(policy, dbs_info->requested_freq, - CPUFREQ_RELATION_L); + __cpufreq_driver_target(policy, requested_freq, CPUFREQ_RELATION_L); } out: return dbs_data->sampling_rate; } -static int dbs_cpufreq_notifier(struct notifier_block *nb, unsigned long val, - void *data); - -static struct notifier_block cs_cpufreq_notifier_block = { - .notifier_call = dbs_cpufreq_notifier, -}; - /************************** sysfs interface ************************/ static ssize_t store_sampling_down_factor(struct gov_attr_set *attr_set, @@ -254,13 +243,6 @@ static struct attribute *cs_attributes[] = { /************************** sysfs end ************************/ -struct cs_governor { - struct dbs_governor dbs_gov; - unsigned int usage_count; -}; - -static struct cs_governor cs_gov; - static struct policy_dbs_info *cs_alloc(void) { struct cs_policy_dbs_info *dbs_info; @@ -292,25 +274,11 @@ static int cs_init(struct dbs_data *dbs_data) dbs_data->min_sampling_rate = MIN_SAMPLING_RATE_RATIO * jiffies_to_usecs(10); - /* - * This function and cs_exit() are only called under gov_dbs_data_mutex - * which is global, so the cs_gov.usage_count accesses are guaranteed - * to be serialized. - */ - if (!cs_gov.usage_count++) - cpufreq_register_notifier(&cs_cpufreq_notifier_block, - CPUFREQ_TRANSITION_NOTIFIER); - return 0; } static void cs_exit(struct dbs_data *dbs_data) { - /* Protected by gov_dbs_data_mutex - see the comment in cs_init(). */ - if (!--cs_gov.usage_count) - cpufreq_unregister_notifier(&cs_cpufreq_notifier_block, - CPUFREQ_TRANSITION_NOTIFIER); - kfree(dbs_data->tuners); } @@ -319,49 +287,20 @@ static void cs_start(struct cpufreq_policy *policy) struct cs_policy_dbs_info *dbs_info = to_dbs_info(policy->governor_data); dbs_info->down_skip = 0; - dbs_info->requested_freq = policy->cur; } -static struct cs_governor cs_gov = { - .dbs_gov = { - .gov = CPUFREQ_DBS_GOVERNOR_INITIALIZER("conservative"), - .kobj_type = { .default_attrs = cs_attributes }, - .gov_dbs_timer = cs_dbs_timer, - .alloc = cs_alloc, - .free = cs_free, - .init = cs_init, - .exit = cs_exit, - .start = cs_start, - }, +static struct dbs_governor cs_governor = { + .gov = CPUFREQ_DBS_GOVERNOR_INITIALIZER("conservative"), + .kobj_type = { .default_attrs = cs_attributes }, + .gov_dbs_timer = cs_dbs_timer, + .alloc = cs_alloc, + .free = cs_free, + .init = cs_init, + .exit = cs_exit, + .start = cs_start, }; -#define CPU_FREQ_GOV_CONSERVATIVE (&cs_gov.dbs_gov.gov) - -static int dbs_cpufreq_notifier(struct notifier_block *nb, unsigned long val, - void *data) -{ - struct cpufreq_freqs *freq = data; - struct cpufreq_policy *policy = cpufreq_cpu_get_raw(freq->cpu); - struct cs_policy_dbs_info *dbs_info; - - if (!policy) - return 0; - - /* policy isn't governed by conservative governor */ - if (policy->governor != CPU_FREQ_GOV_CONSERVATIVE) - return 0; - - dbs_info = to_dbs_info(policy->governor_data); - /* - * we only care if our internally tracked freq moves outside the 'valid' - * ranges of frequency available to us otherwise we do not change it - */ - if (dbs_info->requested_freq > policy->max - || dbs_info->requested_freq < policy->min) - dbs_info->requested_freq = freq->new; - - return 0; -} +#define CPU_FREQ_GOV_CONSERVATIVE (&cs_governor.gov) static int __init cpufreq_gov_dbs_init(void) { From 41bad47f76481d8f55a555274062707e41e57e0e Mon Sep 17 00:00:00 2001 From: Srinivas Pandruvada Date: Thu, 9 Jun 2016 15:34:41 -0700 Subject: [PATCH 35/42] cpufreq: intel_pstate: Broxton support Add Broxton CPU model number. Broxton requires core_params to get performance limits via MSRs, but it is an Atom platform, which requires more power optimized algorithm. So the P state selection will use similar algorithm as other Atom platforms. Signed-off-by: Srinivas Pandruvada Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+) diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index e1b7a5e288f9..8a0e40591f1f 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -1110,6 +1110,26 @@ static struct cpu_defaults knl_params = { }, }; +static struct cpu_defaults bxt_params = { + .pid_policy = { + .sample_rate_ms = 10, + .deadband = 0, + .setpoint = 60, + .p_gain_pct = 14, + .d_gain_pct = 0, + .i_gain_pct = 4, + }, + .funcs = { + .get_max = core_get_max_pstate, + .get_max_physical = core_get_max_pstate_physical, + .get_min = core_get_min_pstate, + .get_turbo = core_get_turbo_pstate, + .get_scaling = core_get_scaling, + .get_val = core_get_val, + .get_target_pstate = get_target_pstate_use_cpu_load, + }, +}; + static void intel_pstate_get_min_max(struct cpudata *cpu, int *min, int *max) { int max_perf = cpu->pstate.turbo_pstate; @@ -1370,6 +1390,7 @@ static const struct x86_cpu_id intel_pstate_cpu_ids[] = { ICPU(INTEL_FAM6_SKYLAKE_DESKTOP, core_params), ICPU(INTEL_FAM6_BROADWELL_XEON_D, core_params), ICPU(INTEL_FAM6_XEON_PHI_KNL, knl_params), + ICPU(INTEL_FAM6_ATOM_GOLDMONT, bxt_params), {} }; MODULE_DEVICE_TABLE(x86cpu, intel_pstate_cpu_ids); From a0944d904f951466047cb7ae625fc17927084e59 Mon Sep 17 00:00:00 2001 From: Ben Dooks Date: Tue, 7 Jun 2016 12:30:13 +0100 Subject: [PATCH 36/42] cpufreq: mvebu: fix integer to pointer cast Fix the use of 0 instead of NULL to clk_get() call. This stops the following warning: drivers/cpufreq/mvebu-cpufreq.c:73:40: warning: Using plain integer as NULL pointer Signed-off-by: Ben Dooks Acked-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/mvebu-cpufreq.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/cpufreq/mvebu-cpufreq.c b/drivers/cpufreq/mvebu-cpufreq.c index e920889b9ac2..ed915ee85dd9 100644 --- a/drivers/cpufreq/mvebu-cpufreq.c +++ b/drivers/cpufreq/mvebu-cpufreq.c @@ -70,7 +70,7 @@ static int __init armada_xp_pmsu_cpufreq_init(void) continue; } - clk = clk_get(cpu_dev, 0); + clk = clk_get(cpu_dev, NULL); if (IS_ERR(clk)) { pr_err("Cannot get clock for CPU %d\n", cpu); return PTR_ERR(clk); From eed436095e62351c9f6d6b3ac6cdd18e89a852d8 Mon Sep 17 00:00:00 2001 From: Jisheng Zhang Date: Mon, 27 Jun 2016 18:07:16 +0800 Subject: [PATCH 37/42] intel_pstate: Fix incorrect placement of __initdata __initdata should be placed between the variable name and equal sign (if there is) for the variable to be placed in the intended section. Signed-off-by: Jisheng Zhang Acked-by: Viresh Kumar Acked-by: Srinivas Pandruvada Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index 10393fb4a955..6390d524a211 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -1596,9 +1596,9 @@ static struct cpufreq_driver intel_pstate_driver = { .name = "intel_pstate", }; -static int __initdata no_load; -static int __initdata no_hwp; -static int __initdata hwp_only; +static int no_load __initdata; +static int no_hwp __initdata; +static int hwp_only __initdata; static unsigned int force_load; static int intel_pstate_msrs_not_valid(void) From 29327c84baebec3549b9775dc8de3d26343720cb Mon Sep 17 00:00:00 2001 From: Jisheng Zhang Date: Mon, 27 Jun 2016 18:07:17 +0800 Subject: [PATCH 38/42] intel_pstate: add __init/__initdata marker to some functions/variables These functions/variables are not needed after booting, so mark them as __init or __initdata. Signed-off-by: Jisheng Zhang Acked-by: Viresh Kumar Acked-by: Srinivas Pandruvada Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index 6390d524a211..fc71f92cf016 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -1377,7 +1377,7 @@ static const struct x86_cpu_id intel_pstate_cpu_ids[] = { }; MODULE_DEVICE_TABLE(x86cpu, intel_pstate_cpu_ids); -static const struct x86_cpu_id intel_pstate_cpu_oob_ids[] = { +static const struct x86_cpu_id intel_pstate_cpu_oob_ids[] __initconst = { ICPU(INTEL_FAM6_BROADWELL_XEON_D, core_params), {} }; @@ -1599,9 +1599,9 @@ static struct cpufreq_driver intel_pstate_driver = { static int no_load __initdata; static int no_hwp __initdata; static int hwp_only __initdata; -static unsigned int force_load; +static unsigned int force_load __initdata; -static int intel_pstate_msrs_not_valid(void) +static int __init intel_pstate_msrs_not_valid(void) { if (!pstate_funcs.get_max() || !pstate_funcs.get_min() || @@ -1611,7 +1611,7 @@ static int intel_pstate_msrs_not_valid(void) return 0; } -static void copy_pid_params(struct pstate_adjust_policy *policy) +static void __init copy_pid_params(struct pstate_adjust_policy *policy) { pid_params.sample_rate_ms = policy->sample_rate_ms; pid_params.sample_rate_ns = pid_params.sample_rate_ms * NSEC_PER_MSEC; @@ -1622,7 +1622,7 @@ static void copy_pid_params(struct pstate_adjust_policy *policy) pid_params.setpoint = policy->setpoint; } -static void copy_cpu_funcs(struct pstate_funcs *funcs) +static void __init copy_cpu_funcs(struct pstate_funcs *funcs) { pstate_funcs.get_max = funcs->get_max; pstate_funcs.get_max_physical = funcs->get_max_physical; @@ -1637,7 +1637,7 @@ static void copy_cpu_funcs(struct pstate_funcs *funcs) #ifdef CONFIG_ACPI -static bool intel_pstate_no_acpi_pss(void) +static bool __init intel_pstate_no_acpi_pss(void) { int i; @@ -1666,7 +1666,7 @@ static bool intel_pstate_no_acpi_pss(void) return true; } -static bool intel_pstate_has_acpi_ppc(void) +static bool __init intel_pstate_has_acpi_ppc(void) { int i; @@ -1694,7 +1694,7 @@ struct hw_vendor_info { }; /* Hardware vendor-specific info that has its own power management modes */ -static struct hw_vendor_info vendor_info[] = { +static struct hw_vendor_info vendor_info[] __initdata = { {1, "HP ", "ProLiant", PSS}, {1, "ORACLE", "X4-2 ", PPC}, {1, "ORACLE", "X4-2L ", PPC}, @@ -1713,7 +1713,7 @@ static struct hw_vendor_info vendor_info[] = { {0, "", ""}, }; -static bool intel_pstate_platform_pwr_mgmt_exists(void) +static bool __init intel_pstate_platform_pwr_mgmt_exists(void) { struct acpi_table_header hdr; struct hw_vendor_info *v_info; From 4a7cb7a96aa32e34777b263fbf372543fb85ddea Mon Sep 17 00:00:00 2001 From: Jisheng Zhang Date: Mon, 27 Jun 2016 18:07:18 +0800 Subject: [PATCH 39/42] intel_pstate: Declare pid_params/pstate_funcs/hwp_active __read_mostly pid_params is written once by copy_pid_params() during initialization, and thereafter is mostly read by hot path intel_pstate_update_util(). The read of pid_params gets more after commit a4675fbc4a7a ("cpufreq: intel_pstate: Replace timers with utilization update callbacks") pstate_funcs is written once by copy_cpu_funcs() during initialization, and thereafter is mostly read by hot path intel_pstate_update_util() hwp_active is written to once during initialization and thereafter is mostly read by hot path intel_pstate_update_util(). The fact that they are mostly read and not written to makes them candidates for __read_mostly declarations. Signed-off-by: Jisheng Zhang Acked-by: Viresh Kumar Acked-by: Srinivas Pandruvada Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index fc71f92cf016..2ed7955f2698 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -282,9 +282,9 @@ struct cpu_defaults { static inline int32_t get_target_pstate_use_performance(struct cpudata *cpu); static inline int32_t get_target_pstate_use_cpu_load(struct cpudata *cpu); -static struct pstate_adjust_policy pid_params; -static struct pstate_funcs pstate_funcs; -static int hwp_active; +static struct pstate_adjust_policy pid_params __read_mostly; +static struct pstate_funcs pstate_funcs __read_mostly; +static int hwp_active __read_mostly; #ifdef CONFIG_ACPI static bool acpi_ppc; From 8d540ea79211dd272403ed79b2033bbded19ac42 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Tue, 28 Jun 2016 03:31:02 +0200 Subject: [PATCH 40/42] cpufreq: Drop redundant check from cpufreq_update_current_freq() Both callers of cpufreq_update_current_freq(), cpufreq_update_policy() and cpufreq_start_governor(), check cpufreq_suspended before calling that function, so drop the redundant cpufreq_suspended check from it. Signed-off-by: Rafael J. Wysocki Acked-by: Viresh Kumar --- drivers/cpufreq/cpufreq.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index 5918e954bb5d..118b4f30a406 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -1544,9 +1544,6 @@ static unsigned int cpufreq_update_current_freq(struct cpufreq_policy *policy) { unsigned int new_freq; - if (cpufreq_suspended) - return 0; - new_freq = cpufreq_driver->get(policy->cpu); if (!new_freq) return 0; From da0c6dc00c69d0adaae99c3b3d2ea0c869a3fb35 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Mon, 27 Jun 2016 16:04:07 +0530 Subject: [PATCH 41/42] cpufreq: Handle sorted frequency tables more efficiently cpufreq drivers aren't required to provide a sorted frequency table today, and even the ones which provide a sorted table aren't handled efficiently by cpufreq core. This patch adds infrastructure to verify if the freq-table provided by the drivers is sorted or not, and use efficient helpers if they are sorted. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/freq_table.c | 73 +++++++++-- include/linux/cpufreq.h | 234 ++++++++++++++++++++++++++++++++++- 2 files changed, 296 insertions(+), 11 deletions(-) diff --git a/drivers/cpufreq/freq_table.c b/drivers/cpufreq/freq_table.c index eac8bcbdaad1..3bbbf9e6960c 100644 --- a/drivers/cpufreq/freq_table.c +++ b/drivers/cpufreq/freq_table.c @@ -113,9 +113,9 @@ int cpufreq_generic_frequency_table_verify(struct cpufreq_policy *policy) } EXPORT_SYMBOL_GPL(cpufreq_generic_frequency_table_verify); -int cpufreq_frequency_table_target(struct cpufreq_policy *policy, - unsigned int target_freq, - unsigned int relation) +int cpufreq_table_index_unsorted(struct cpufreq_policy *policy, + unsigned int target_freq, + unsigned int relation) { struct cpufreq_frequency_table optimal = { .driver_data = ~0, @@ -205,7 +205,7 @@ int cpufreq_frequency_table_target(struct cpufreq_policy *policy, table[index].frequency); return index; } -EXPORT_SYMBOL_GPL(cpufreq_frequency_table_target); +EXPORT_SYMBOL_GPL(cpufreq_table_index_unsorted); int cpufreq_frequency_table_get_index(struct cpufreq_policy *policy, unsigned int freq) @@ -297,15 +297,72 @@ struct freq_attr *cpufreq_generic_attr[] = { }; EXPORT_SYMBOL_GPL(cpufreq_generic_attr); +static int set_freq_table_sorted(struct cpufreq_policy *policy) +{ + struct cpufreq_frequency_table *pos, *table = policy->freq_table; + struct cpufreq_frequency_table *prev = NULL; + int ascending = 0; + + policy->freq_table_sorted = CPUFREQ_TABLE_UNSORTED; + + cpufreq_for_each_valid_entry(pos, table) { + if (!prev) { + prev = pos; + continue; + } + + if (pos->frequency == prev->frequency) { + pr_warn("Duplicate freq-table entries: %u\n", + pos->frequency); + return -EINVAL; + } + + /* Frequency increased from prev to pos */ + if (pos->frequency > prev->frequency) { + /* But frequency was decreasing earlier */ + if (ascending < 0) { + pr_debug("Freq table is unsorted\n"); + return 0; + } + + ascending++; + } else { + /* Frequency decreased from prev to pos */ + + /* But frequency was increasing earlier */ + if (ascending > 0) { + pr_debug("Freq table is unsorted\n"); + return 0; + } + + ascending--; + } + + prev = pos; + } + + if (ascending > 0) + policy->freq_table_sorted = CPUFREQ_TABLE_SORTED_ASCENDING; + else + policy->freq_table_sorted = CPUFREQ_TABLE_SORTED_DESCENDING; + + pr_debug("Freq table is sorted in %s order\n", + ascending > 0 ? "ascending" : "descending"); + + return 0; +} + int cpufreq_table_validate_and_show(struct cpufreq_policy *policy, struct cpufreq_frequency_table *table) { - int ret = cpufreq_frequency_table_cpuinfo(policy, table); + int ret; - if (!ret) - policy->freq_table = table; + ret = cpufreq_frequency_table_cpuinfo(policy, table); + if (ret) + return ret; - return ret; + policy->freq_table = table; + return set_freq_table_sorted(policy); } EXPORT_SYMBOL_GPL(cpufreq_table_validate_and_show); diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h index c378776628b4..c6410b1b2490 100644 --- a/include/linux/cpufreq.h +++ b/include/linux/cpufreq.h @@ -36,6 +36,12 @@ struct cpufreq_governor; +enum cpufreq_table_sorting { + CPUFREQ_TABLE_UNSORTED, + CPUFREQ_TABLE_SORTED_ASCENDING, + CPUFREQ_TABLE_SORTED_DESCENDING +}; + struct cpufreq_freqs { unsigned int cpu; /* cpu nr */ unsigned int old; @@ -87,6 +93,7 @@ struct cpufreq_policy { struct cpufreq_user_policy user_policy; struct cpufreq_frequency_table *freq_table; + enum cpufreq_table_sorting freq_table_sorted; struct list_head policy_list; struct kobject kobj; @@ -597,9 +604,9 @@ int cpufreq_frequency_table_verify(struct cpufreq_policy *policy, struct cpufreq_frequency_table *table); int cpufreq_generic_frequency_table_verify(struct cpufreq_policy *policy); -int cpufreq_frequency_table_target(struct cpufreq_policy *policy, - unsigned int target_freq, - unsigned int relation); +int cpufreq_table_index_unsorted(struct cpufreq_policy *policy, + unsigned int target_freq, + unsigned int relation); int cpufreq_frequency_table_get_index(struct cpufreq_policy *policy, unsigned int freq); @@ -610,6 +617,227 @@ int cpufreq_boost_trigger_state(int state); int cpufreq_boost_enabled(void); int cpufreq_enable_boost_support(void); bool policy_has_boost_freq(struct cpufreq_policy *policy); + +/* Find lowest freq at or above target in a table in ascending order */ +static inline int cpufreq_table_find_index_al(struct cpufreq_policy *policy, + unsigned int target_freq) +{ + struct cpufreq_frequency_table *table = policy->freq_table; + unsigned int freq; + int i, best = -1; + + for (i = 0; table[i].frequency != CPUFREQ_TABLE_END; i++) { + freq = table[i].frequency; + + if (freq >= target_freq) + return i; + + best = i; + } + + return best; +} + +/* Find lowest freq at or above target in a table in descending order */ +static inline int cpufreq_table_find_index_dl(struct cpufreq_policy *policy, + unsigned int target_freq) +{ + struct cpufreq_frequency_table *table = policy->freq_table; + unsigned int freq; + int i, best = -1; + + for (i = 0; table[i].frequency != CPUFREQ_TABLE_END; i++) { + freq = table[i].frequency; + + if (freq == target_freq) + return i; + + if (freq > target_freq) { + best = i; + continue; + } + + /* No freq found above target_freq */ + if (best == -1) + return i; + + return best; + } + + return best; +} + +/* Works only on sorted freq-tables */ +static inline int cpufreq_table_find_index_l(struct cpufreq_policy *policy, + unsigned int target_freq) +{ + target_freq = clamp_val(target_freq, policy->min, policy->max); + + if (policy->freq_table_sorted == CPUFREQ_TABLE_SORTED_ASCENDING) + return cpufreq_table_find_index_al(policy, target_freq); + else + return cpufreq_table_find_index_dl(policy, target_freq); +} + +/* Find highest freq at or below target in a table in ascending order */ +static inline int cpufreq_table_find_index_ah(struct cpufreq_policy *policy, + unsigned int target_freq) +{ + struct cpufreq_frequency_table *table = policy->freq_table; + unsigned int freq; + int i, best = -1; + + for (i = 0; table[i].frequency != CPUFREQ_TABLE_END; i++) { + freq = table[i].frequency; + + if (freq == target_freq) + return i; + + if (freq < target_freq) { + best = i; + continue; + } + + /* No freq found below target_freq */ + if (best == -1) + return i; + + return best; + } + + return best; +} + +/* Find highest freq at or below target in a table in descending order */ +static inline int cpufreq_table_find_index_dh(struct cpufreq_policy *policy, + unsigned int target_freq) +{ + struct cpufreq_frequency_table *table = policy->freq_table; + unsigned int freq; + int i, best = -1; + + for (i = 0; table[i].frequency != CPUFREQ_TABLE_END; i++) { + freq = table[i].frequency; + + if (freq <= target_freq) + return i; + + best = i; + } + + return best; +} + +/* Works only on sorted freq-tables */ +static inline int cpufreq_table_find_index_h(struct cpufreq_policy *policy, + unsigned int target_freq) +{ + target_freq = clamp_val(target_freq, policy->min, policy->max); + + if (policy->freq_table_sorted == CPUFREQ_TABLE_SORTED_ASCENDING) + return cpufreq_table_find_index_ah(policy, target_freq); + else + return cpufreq_table_find_index_dh(policy, target_freq); +} + +/* Find closest freq to target in a table in ascending order */ +static inline int cpufreq_table_find_index_ac(struct cpufreq_policy *policy, + unsigned int target_freq) +{ + struct cpufreq_frequency_table *table = policy->freq_table; + unsigned int freq; + int i, best = -1; + + for (i = 0; table[i].frequency != CPUFREQ_TABLE_END; i++) { + freq = table[i].frequency; + + if (freq == target_freq) + return i; + + if (freq < target_freq) { + best = i; + continue; + } + + /* No freq found below target_freq */ + if (best == -1) + return i; + + /* Choose the closest freq */ + if (target_freq - table[best].frequency > freq - target_freq) + return i; + + return best; + } + + return best; +} + +/* Find closest freq to target in a table in descending order */ +static inline int cpufreq_table_find_index_dc(struct cpufreq_policy *policy, + unsigned int target_freq) +{ + struct cpufreq_frequency_table *table = policy->freq_table; + unsigned int freq; + int i, best = -1; + + for (i = 0; table[i].frequency != CPUFREQ_TABLE_END; i++) { + freq = table[i].frequency; + + if (freq == target_freq) + return i; + + if (freq > target_freq) { + best = i; + continue; + } + + /* No freq found above target_freq */ + if (best == -1) + return i; + + /* Choose the closest freq */ + if (table[best].frequency - target_freq > target_freq - freq) + return i; + + return best; + } + + return best; +} + +/* Works only on sorted freq-tables */ +static inline int cpufreq_table_find_index_c(struct cpufreq_policy *policy, + unsigned int target_freq) +{ + target_freq = clamp_val(target_freq, policy->min, policy->max); + + if (policy->freq_table_sorted == CPUFREQ_TABLE_SORTED_ASCENDING) + return cpufreq_table_find_index_ac(policy, target_freq); + else + return cpufreq_table_find_index_dc(policy, target_freq); +} + +static inline int cpufreq_frequency_table_target(struct cpufreq_policy *policy, + unsigned int target_freq, + unsigned int relation) +{ + if (unlikely(policy->freq_table_sorted == CPUFREQ_TABLE_UNSORTED)) + return cpufreq_table_index_unsorted(policy, target_freq, + relation); + + switch (relation) { + case CPUFREQ_RELATION_L: + return cpufreq_table_find_index_l(policy, target_freq); + case CPUFREQ_RELATION_H: + return cpufreq_table_find_index_h(policy, target_freq); + case CPUFREQ_RELATION_C: + return cpufreq_table_find_index_c(policy, target_freq); + default: + pr_err("%s: Invalid relation: %d\n", __func__, relation); + return -EINVAL; + } +} #else static inline int cpufreq_boost_trigger_state(int state) { From 825773609c8ae34a19d4b593555493ae2754eff7 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Mon, 27 Jun 2016 09:59:34 +0530 Subject: [PATCH 42/42] cpufreq: Reuse new freq-table helpers This patch migrates few users of cpufreq tables to the new helpers that work on sorted freq-tables. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/acpi-cpufreq.c | 14 ++++---------- drivers/cpufreq/amd_freq_sensitivity.c | 4 ++-- drivers/cpufreq/cpufreq_ondemand.c | 6 ++---- drivers/cpufreq/powernv-cpufreq.c | 3 +-- drivers/cpufreq/s5pv210-cpufreq.c | 3 +-- 5 files changed, 10 insertions(+), 20 deletions(-) diff --git a/drivers/cpufreq/acpi-cpufreq.c b/drivers/cpufreq/acpi-cpufreq.c index 32a15052f363..11c9a078e0fd 100644 --- a/drivers/cpufreq/acpi-cpufreq.c +++ b/drivers/cpufreq/acpi-cpufreq.c @@ -468,20 +468,14 @@ unsigned int acpi_cpufreq_fast_switch(struct cpufreq_policy *policy, struct acpi_cpufreq_data *data = policy->driver_data; struct acpi_processor_performance *perf; struct cpufreq_frequency_table *entry; - unsigned int next_perf_state, next_freq, freq; + unsigned int next_perf_state, next_freq, index; /* * Find the closest frequency above target_freq. - * - * The table is sorted in the reverse order with respect to the - * frequency and all of the entries are valid (see the initialization). */ - entry = policy->freq_table; - do { - entry++; - freq = entry->frequency; - } while (freq >= target_freq && freq != CPUFREQ_TABLE_END); - entry--; + index = cpufreq_table_find_index_dl(policy, target_freq); + + entry = &policy->freq_table[index]; next_freq = entry->frequency; next_perf_state = entry->driver_data; diff --git a/drivers/cpufreq/amd_freq_sensitivity.c b/drivers/cpufreq/amd_freq_sensitivity.c index 6d5dc04c3a37..042023bbbf62 100644 --- a/drivers/cpufreq/amd_freq_sensitivity.c +++ b/drivers/cpufreq/amd_freq_sensitivity.c @@ -91,8 +91,8 @@ static unsigned int amd_powersave_bias_target(struct cpufreq_policy *policy, else { unsigned int index; - index = cpufreq_frequency_table_target(policy, - policy->cur - 1, CPUFREQ_RELATION_H); + index = cpufreq_table_find_index_h(policy, + policy->cur - 1); freq_next = policy->freq_table[index].frequency; } diff --git a/drivers/cpufreq/cpufreq_ondemand.c b/drivers/cpufreq/cpufreq_ondemand.c index 0c93cd9dee99..3a1f49f5f4c6 100644 --- a/drivers/cpufreq/cpufreq_ondemand.c +++ b/drivers/cpufreq/cpufreq_ondemand.c @@ -85,11 +85,9 @@ static unsigned int generic_powersave_bias_target(struct cpufreq_policy *policy, freq_avg = freq_req - freq_reduc; /* Find freq bounds for freq_avg in freq_table */ - index = cpufreq_frequency_table_target(policy, freq_avg, - CPUFREQ_RELATION_H); + index = cpufreq_table_find_index_h(policy, freq_avg); freq_lo = freq_table[index].frequency; - index = cpufreq_frequency_table_target(policy, freq_avg, - CPUFREQ_RELATION_L); + index = cpufreq_table_find_index_l(policy, freq_avg); freq_hi = freq_table[index].frequency; /* Find out how long we have to be in hi and lo freqs */ diff --git a/drivers/cpufreq/powernv-cpufreq.c b/drivers/cpufreq/powernv-cpufreq.c index b29c5c20c3a1..2a2920c4fdf9 100644 --- a/drivers/cpufreq/powernv-cpufreq.c +++ b/drivers/cpufreq/powernv-cpufreq.c @@ -760,8 +760,7 @@ void powernv_cpufreq_work_fn(struct work_struct *work) struct cpufreq_policy policy; cpufreq_get_policy(&policy, cpu); - index = cpufreq_frequency_table_target(&policy, policy.cur, - CPUFREQ_RELATION_C); + index = cpufreq_table_find_index_c(&policy, policy.cur); powernv_cpufreq_target_index(&policy, index); cpumask_andnot(&mask, &mask, policy.cpus); } diff --git a/drivers/cpufreq/s5pv210-cpufreq.c b/drivers/cpufreq/s5pv210-cpufreq.c index 4f4e9df9b7fc..9e07588ea9f5 100644 --- a/drivers/cpufreq/s5pv210-cpufreq.c +++ b/drivers/cpufreq/s5pv210-cpufreq.c @@ -246,8 +246,7 @@ static int s5pv210_target(struct cpufreq_policy *policy, unsigned int index) new_freq = s5pv210_freq_table[index].frequency; /* Finding current running level index */ - priv_index = cpufreq_frequency_table_target(policy, old_freq, - CPUFREQ_RELATION_H); + priv_index = cpufreq_table_find_index_h(policy, old_freq); arm_volt = dvs_conf[index].arm_volt; int_volt = dvs_conf[index].int_volt;