From bb08be78721bed02b147448d2cb404babc369cfe Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Mon, 29 Apr 2013 13:24:44 +0000 Subject: cpufreq: ARM big LITTLE: Select PM_OPP The ARM big LITTLE cpufreq driver uses the OPP layer for its functionality. Select it in Kconfig. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/Kconfig.arm | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/Kconfig.arm b/drivers/cpufreq/Kconfig.arm index f3af18b9acc5..a78ce4490c61 100644 --- a/drivers/cpufreq/Kconfig.arm +++ b/drivers/cpufreq/Kconfig.arm @@ -5,6 +5,7 @@ config ARM_BIG_LITTLE_CPUFREQ tristate depends on ARM_CPU_TOPOLOGY + select PM_OPP config ARM_DT_BL_CPUFREQ tristate "Generic ARM big LITTLE CPUfreq driver probed via DT" -- cgit v1.2.3 From 996905f3334506296c8a4bec63695bcf7a9df159 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Mon, 29 Apr 2013 13:24:45 +0000 Subject: cpufreq: ARM big LITTLE DT: Return correct transition latency By mistake we are returning zero for successful call to dt_get_transition_latency(), whereas we should return transition_latency. Fix that. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/arm_big_little_dt.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/arm_big_little_dt.c b/drivers/cpufreq/arm_big_little_dt.c index 44be3115375c..d5ae4d2362e3 100644 --- a/drivers/cpufreq/arm_big_little_dt.c +++ b/drivers/cpufreq/arm_big_little_dt.c @@ -78,7 +78,7 @@ static int dt_get_transition_latency(struct device *cpu_dev) of_node_put(np); of_node_put(parent); - return 0; + return transition_latency; } return -ENODEV; -- cgit v1.2.3 From 3c792e0fe17858105b72516f709c48fc8e4a7523 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Mon, 29 Apr 2013 13:24:46 +0000 Subject: cpufreq: ARM big LITTLE DT: Return CPUFREQ_ETERNAL if clock-latency isn't found If "/cpus" node isn't present or "clock-latency" isn't defined we are returning error currently. Let's return CPUFREQ_ETERNAL instead, so that we don't fail. Flag appropriate messages to user in such cases. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/arm_big_little_dt.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/arm_big_little_dt.c b/drivers/cpufreq/arm_big_little_dt.c index d5ae4d2362e3..173ed059d95f 100644 --- a/drivers/cpufreq/arm_big_little_dt.c +++ b/drivers/cpufreq/arm_big_little_dt.c @@ -66,8 +66,8 @@ static int dt_get_transition_latency(struct device *cpu_dev) parent = of_find_node_by_path("/cpus"); if (!parent) { - pr_err("failed to find OF /cpus\n"); - return -ENOENT; + pr_info("Failed to find OF /cpus. Use CPUFREQ_ETERNAL transition latency\n"); + return CPUFREQ_ETERNAL; } for_each_child_of_node(parent, np) { @@ -81,7 +81,8 @@ static int dt_get_transition_latency(struct device *cpu_dev) return transition_latency; } - return -ENODEV; + pr_info("clock-latency isn't found, use CPUFREQ_ETERNAL transition latency\n"); + return CPUFREQ_ETERNAL; } static struct cpufreq_arm_bL_ops dt_bL_ops = { -- cgit v1.2.3 From 4521adf85cd18e4652b9285846835f74418bd88f Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Mon, 29 Apr 2013 13:24:47 +0000 Subject: cpufreq: ARM big LITTLE: Move cpu_to_cluster() to arm_big_little.h The cpu_to_cluster() function may be used by glue drivers, so it's better to keep it in arm_big_little.h. [rjw: Changelog] Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/arm_big_little.c | 5 ----- drivers/cpufreq/arm_big_little.h | 5 +++++ 2 files changed, 5 insertions(+), 5 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/arm_big_little.c b/drivers/cpufreq/arm_big_little.c index dbdf677d2f36..fd7beed801f8 100644 --- a/drivers/cpufreq/arm_big_little.c +++ b/drivers/cpufreq/arm_big_little.c @@ -40,11 +40,6 @@ static struct clk *clk[MAX_CLUSTERS]; static struct cpufreq_frequency_table *freq_table[MAX_CLUSTERS]; static atomic_t cluster_usage[MAX_CLUSTERS] = {ATOMIC_INIT(0), ATOMIC_INIT(0)}; -static int cpu_to_cluster(int cpu) -{ - return topology_physical_package_id(cpu); -} - static unsigned int bL_cpufreq_get(unsigned int cpu) { u32 cur_cluster = cpu_to_cluster(cpu); diff --git a/drivers/cpufreq/arm_big_little.h b/drivers/cpufreq/arm_big_little.h index 70f18fc12d4a..79b2ce17884d 100644 --- a/drivers/cpufreq/arm_big_little.h +++ b/drivers/cpufreq/arm_big_little.h @@ -34,6 +34,11 @@ struct cpufreq_arm_bL_ops { int (*init_opp_table)(struct device *cpu_dev); }; +static inline int cpu_to_cluster(int cpu) +{ + return topology_physical_package_id(cpu); +} + int bL_cpufreq_register(struct cpufreq_arm_bL_ops *ops); void bL_cpufreq_unregister(struct cpufreq_arm_bL_ops *ops); -- cgit v1.2.3 From 2b80f3138e8470194745a6b954b4905060ab4067 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Mon, 29 Apr 2013 13:24:48 +0000 Subject: cpufreq: ARM big LITTLE: Improve print message The message printed at the end of driver->init() doesn't include the "cpufreq" string at all and so is difficult to find in dmesg. Add function name to that message to clearly state where the message is coming from. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/arm_big_little.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/arm_big_little.c b/drivers/cpufreq/arm_big_little.c index fd7beed801f8..5d7f53fcd6f5 100644 --- a/drivers/cpufreq/arm_big_little.c +++ b/drivers/cpufreq/arm_big_little.c @@ -187,7 +187,7 @@ static int bL_cpufreq_init(struct cpufreq_policy *policy) cpumask_copy(policy->cpus, topology_core_cpumask(policy->cpu)); - dev_info(cpu_dev, "CPU %d initialized\n", policy->cpu); + dev_info(cpu_dev, "%s: CPU %d initialized\n", __func__, policy->cpu); return 0; } -- cgit v1.2.3 From a97c98adddbe98e824b69e6d7b320c8dc91fe581 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Tue, 30 Apr 2013 14:32:17 +0000 Subject: cpufreq: governors: Fix CPUFREQ_GOV_POLICY_{INIT|EXIT} notifiers There are two types of INIT/EXIT activities that we need to do for governors: - Done only once per governor (doesn't depend how many instances of the governor there are). eg: cpufreq_register_notifier() for conservative governor. - Done per governor instance, eg: sysfs_{create|remove}_group(). There were some corner cases where current code isn't able to handle them separately and so failing for some test cases. We use two separate variables now for keeping track of above two requirements. - governor->initialized for first one - dbs_data->usage_count for per governor instance Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq_governor.c | 11 +++++++---- drivers/cpufreq/cpufreq_governor.h | 1 + 2 files changed, 8 insertions(+), 4 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/cpufreq_governor.c b/drivers/cpufreq/cpufreq_governor.c index 443442df113b..5af40ad82d23 100644 --- a/drivers/cpufreq/cpufreq_governor.c +++ b/drivers/cpufreq/cpufreq_governor.c @@ -255,6 +255,7 @@ int cpufreq_governor_dbs(struct cpufreq_policy *policy, if (have_governor_per_policy()) { WARN_ON(dbs_data); } else if (dbs_data) { + dbs_data->usage_count++; policy->governor_data = dbs_data; return 0; } @@ -266,6 +267,7 @@ int cpufreq_governor_dbs(struct cpufreq_policy *policy, } dbs_data->cdata = cdata; + dbs_data->usage_count = 1; rc = cdata->init(dbs_data); if (rc) { pr_err("%s: POLICY_INIT: init() failed\n", __func__); @@ -294,7 +296,8 @@ int cpufreq_governor_dbs(struct cpufreq_policy *policy, set_sampling_rate(dbs_data, max(dbs_data->min_sampling_rate, latency * LATENCY_MULTIPLIER)); - if (dbs_data->cdata->governor == GOV_CONSERVATIVE) { + if ((cdata->governor == GOV_CONSERVATIVE) && + (!policy->governor->initialized)) { struct cs_ops *cs_ops = dbs_data->cdata->gov_ops; cpufreq_register_notifier(cs_ops->notifier_block, @@ -306,12 +309,12 @@ int cpufreq_governor_dbs(struct cpufreq_policy *policy, return 0; case CPUFREQ_GOV_POLICY_EXIT: - if ((policy->governor->initialized == 1) || - have_governor_per_policy()) { + if (!--dbs_data->usage_count) { sysfs_remove_group(get_governor_parent_kobj(policy), get_sysfs_attr(dbs_data)); - if (dbs_data->cdata->governor == GOV_CONSERVATIVE) { + if ((dbs_data->cdata->governor == GOV_CONSERVATIVE) && + (policy->governor->initialized == 1)) { struct cs_ops *cs_ops = dbs_data->cdata->gov_ops; cpufreq_unregister_notifier(cs_ops->notifier_block, diff --git a/drivers/cpufreq/cpufreq_governor.h b/drivers/cpufreq/cpufreq_governor.h index 8ac33538d0bd..e16a96130cb3 100644 --- a/drivers/cpufreq/cpufreq_governor.h +++ b/drivers/cpufreq/cpufreq_governor.h @@ -211,6 +211,7 @@ struct common_dbs_data { struct dbs_data { struct common_dbs_data *cdata; unsigned int min_sampling_rate; + int usage_count; void *tuners; /* dbs_mutex protects dbs_enable in governor start/stop */ -- cgit v1.2.3 From d96038e0fa00f42a5d6711884bef0a725cdc70bb Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Tue, 30 Apr 2013 14:32:18 +0000 Subject: cpufreq: Issue CPUFREQ_GOV_POLICY_EXIT notifier before dropping policy refcount We must call __cpufreq_governor(data, CPUFREQ_GOV_POLICY_EXIT) before calling cpufreq_cpu_put(data), so that policy kobject have valid fields. Otherwise, removing last online cpu of policy->cpus causes this crash for ondemand/conservative governor. [] (sysfs_find_dirent+0xe/0xa8) from [] (sysfs_get_dirent+0x21/0x58) [] (sysfs_get_dirent+0x21/0x58) from [] (sysfs_remove_group+0x85/0xbc) [] (sysfs_remove_group+0x85/0xbc) from [] (cpufreq_governor_dbs+0x369/0x4a0) [] (cpufreq_governor_dbs+0x369/0x4a0) from [] (__cpufreq_governor+0x2b/0x8c) [] (__cpufreq_governor+0x2b/0x8c) from [] (__cpufreq_remove_dev.isra.12+0x15b/0x250) [] (__cpufreq_remove_dev.isra.12+0x15b/0x250) from [] (cpufreq_cpu_callback+0x2f/0x3c) [] (cpufreq_cpu_callback+0x2f/0x3c) from [] (notifier_call_chain+0x45/0x54) [] (notifier_call_chain+0x45/0x54) from [] (__cpu_notify+0x1d/0x34) [] (__cpu_notify+0x1d/0x34) from [] (_cpu_down+0x63/0x1ac) [] (_cpu_down+0x63/0x1ac) from [] (cpu_down+0x1b/0x30) [] (cpu_down+0x1b/0x30) from [] (store_online+0x27/0x54) [] (store_online+0x27/0x54) from [] (dev_attr_store+0x11/0x18) [] (dev_attr_store+0x11/0x18) from [] (sysfs_write_file+0xed/0x114) [] (sysfs_write_file+0xed/0x114) from [] (vfs_write+0x65/0xd8) [] (vfs_write+0x65/0xd8) from [] (sys_write+0x2f/0x50) [] (sys_write+0x2f/0x50) from [] (ret_fast_syscall+0x1/0x52) Of course this only impacted drivers which have have_governor_per_policy set to true. i.e. big LITTLE cpufreq driver. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index 1b8a48eaf90f..b7acfd153bf9 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -1075,14 +1075,14 @@ static int __cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif __func__, cpu_dev->id, cpu); } + if ((cpus == 1) && (cpufreq_driver->target)) + __cpufreq_governor(data, CPUFREQ_GOV_POLICY_EXIT); + pr_debug("%s: removing link, cpu: %d\n", __func__, cpu); cpufreq_cpu_put(data); /* If cpu is last user of policy, free policy */ if (cpus == 1) { - if (cpufreq_driver->target) - __cpufreq_governor(data, CPUFREQ_GOV_POLICY_EXIT); - lock_policy_rwsem_read(cpu); kobj = &data->kobj; cmp = &data->kobj_unregister; -- cgit v1.2.3 From fc31d6f55908759530462998d0464a9f124b1c32 Mon Sep 17 00:00:00 2001 From: Nishanth Menon Date: Wed, 1 May 2013 13:38:12 +0000 Subject: cpufreq: cpufreq-cpu0: defer probe when regulator is not ready With commit 1e4b545, regulator_get will now return -EPROBE_DEFER when the cpu0-supply node is present, but the regulator is not yet registered. It is possible for this to occur when the regulator registration by itself might be defered due to some dependent interface not yet instantiated. For example: an regulator which uses I2C and GPIO might need both systems available before proceeding, in this case, the regulator might defer it's registration. However, the cpufreq-cpu0 driver assumes that any un-successful return result is equivalent of failure. When the regulator_get returns failure other than -EPROBE_DEFER, it makes sense to assume that supply node is not present and proceed with the assumption that only clock control is necessary in the platform. With this change, we can now handle the following conditions: a) cpu0-supply binding is not present, regulator_get will return appropriate error result, resulting in cpufreq-cpu0 driver controlling just the clock. b) cpu0-supply binding is present, regulator_get returns -EPROBE_DEFER, we retry resulting in cpufreq-cpu0 driver registering later once the regulator is available. c) cpu0-supply binding is present, regulator_get returns -EPROBE_DEFER, however, regulator never registers, we retry until cpufreq-cpu0 driver fails to register pointing at device tree information bug. However, in this case, the fact that cpufreq-cpu0 operates with clock only when the DT binding clearly indicates need of a supply is a bug of it's own. d) cpu0-supply gets an regulator at probe - cpufreq-cpu0 driver controls both the clock and regulator Signed-off-by: Nishanth Menon Acked-by: Shawn Guo Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq-cpu0.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/cpufreq-cpu0.c b/drivers/cpufreq/cpufreq-cpu0.c index 3ab8294eab04..ecd8af900432 100644 --- a/drivers/cpufreq/cpufreq-cpu0.c +++ b/drivers/cpufreq/cpufreq-cpu0.c @@ -195,6 +195,22 @@ static int cpu0_cpufreq_probe(struct platform_device *pdev) cpu_dev = &pdev->dev; cpu_dev->of_node = np; + cpu_reg = devm_regulator_get(cpu_dev, "cpu0"); + if (IS_ERR(cpu_reg)) { + /* + * If cpu0 regulator supply node is present, but regulator is + * not yet registered, we should try defering probe. + */ + if (PTR_ERR(cpu_reg) == -EPROBE_DEFER) { + dev_err(cpu_dev, "cpu0 regulator not ready, retry\n"); + ret = -EPROBE_DEFER; + goto out_put_node; + } + pr_warn("failed to get cpu0 regulator: %ld\n", + PTR_ERR(cpu_reg)); + cpu_reg = NULL; + } + cpu_clk = devm_clk_get(cpu_dev, NULL); if (IS_ERR(cpu_clk)) { ret = PTR_ERR(cpu_clk); @@ -202,12 +218,6 @@ static int cpu0_cpufreq_probe(struct platform_device *pdev) goto out_put_node; } - cpu_reg = devm_regulator_get(cpu_dev, "cpu0"); - if (IS_ERR(cpu_reg)) { - pr_warn("failed to get cpu0 regulator\n"); - cpu_reg = NULL; - } - ret = of_init_opp_table(cpu_dev); if (ret) { pr_err("failed to init OPP table: %d\n", ret); -- cgit v1.2.3 From 5aaa9cc7ab589893efe8e66bf02f7fc2175a1f5b Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Fri, 3 May 2013 04:44:25 +0000 Subject: cpufreq: cpufreq-cpu0: Free parent node for error cases We are freeing parent node in success cases but not in failure cases. Let's do it. Signed-off-by: Viresh Kumar Acked-by: Shawn Guo Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq-cpu0.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/cpufreq-cpu0.c b/drivers/cpufreq/cpufreq-cpu0.c index ecd8af900432..a64eb8b70444 100644 --- a/drivers/cpufreq/cpufreq-cpu0.c +++ b/drivers/cpufreq/cpufreq-cpu0.c @@ -189,7 +189,8 @@ static int cpu0_cpufreq_probe(struct platform_device *pdev) if (!np) { pr_err("failed to find cpu0 node\n"); - return -ENOENT; + ret = -ENOENT; + goto out_put_parent; } cpu_dev = &pdev->dev; @@ -274,6 +275,8 @@ out_free_table: opp_free_cpufreq_table(cpu_dev, &freq_table); out_put_node: of_node_put(np); +out_put_parent: + of_node_put(parent); return ret; } -- cgit v1.2.3 From 99af771115f7b2f86548ca2f762adb9032096702 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Sat, 4 May 2013 12:03:54 +0530 Subject: cpufreq: ARM big LITTLE: Fix Kconfig entries This fixes usage of "depends on" and "select" options in Kconfig for ARM big LITTLE cpufreq driver. Otherwise we get these warnings: warning: (ARM_DT_BL_CPUFREQ) selects ARM_BIG_LITTLE_CPUFREQ which has unmet direct dependencies (ARCH_HAS_CPUFREQ && CPU_FREQ && ARM && ARM_CPU_TOPOLOGY) Signed-off-by: Arnd Bergmann Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/Kconfig.arm | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/Kconfig.arm b/drivers/cpufreq/Kconfig.arm index a78ce4490c61..6e57543fe0b9 100644 --- a/drivers/cpufreq/Kconfig.arm +++ b/drivers/cpufreq/Kconfig.arm @@ -3,17 +3,17 @@ # config ARM_BIG_LITTLE_CPUFREQ - tristate - depends on ARM_CPU_TOPOLOGY - select PM_OPP + tristate "Generic ARM big LITTLE CPUfreq driver" + depends on ARM_CPU_TOPOLOGY && PM_OPP && HAVE_CLK + help + This enables the Generic CPUfreq driver for ARM big.LITTLE platforms. config ARM_DT_BL_CPUFREQ - tristate "Generic ARM big LITTLE CPUfreq driver probed via DT" - select ARM_BIG_LITTLE_CPUFREQ - depends on OF && HAVE_CLK + tristate "Generic probing via DT for ARM big LITTLE CPUfreq driver" + depends on ARM_BIG_LITTLE_CPUFREQ && OF help - This enables the Generic CPUfreq driver for ARM big.LITTLE platform. - This gets frequency tables from DT. + This enables probing via DT for Generic CPUfreq driver for ARM + big.LITTLE platform. This gets frequency tables from DT. config ARM_EXYNOS_CPUFREQ bool "SAMSUNG EXYNOS SoCs" -- cgit v1.2.3 From 559f56c70fc90bd9da8c9c9c36d86c5e582ac5b3 Mon Sep 17 00:00:00 2001 From: Alexander Shiyan Date: Sun, 5 May 2013 12:18:08 +0000 Subject: cpufreq: Fix incorrect dependecies for ARM SA11xx drivers Kconfig dependecies for ARM SA11xx drivers are incorrect, so fix them. [rjw: Changelog] Signed-off-by: Alexander Shiyan Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/Kconfig b/drivers/cpufreq/Kconfig index a1488f58f6ca..534fcb825153 100644 --- a/drivers/cpufreq/Kconfig +++ b/drivers/cpufreq/Kconfig @@ -47,7 +47,7 @@ config CPU_FREQ_STAT_DETAILS choice prompt "Default CPUFreq governor" - default CPU_FREQ_DEFAULT_GOV_USERSPACE if CPU_FREQ_SA1100 || CPU_FREQ_SA1110 + default CPU_FREQ_DEFAULT_GOV_USERSPACE if ARM_SA1100_CPUFREQ || ARM_SA1110_CPUFREQ default CPU_FREQ_DEFAULT_GOV_PERFORMANCE help This option sets which CPUFreq governor shall be loaded at -- cgit v1.2.3 From 1abc4b20b85b42e8573957e54b193385cf48b0d6 Mon Sep 17 00:00:00 2001 From: Dirk Brandewie Date: Tue, 7 May 2013 08:20:25 -0700 Subject: cpufreq / intel_pstate: remove idle time and duration from sample and calculations Idle time is taken into account in the APERF/MPERF ratio calculation there is no reason for the driver to track it seperately. This reduces the work in the driver and makes the code more readable. Removal of the tracking of sample duration removes the possibility of the divide by zero exception when the duration is sub 1us References: https://bugzilla.kernel.org/show_bug.cgi?id=56691 Reported-by: Mike Lothian Cc: 3.9+ Signed-off-by: Dirk Brandewie Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 45 ++++++++---------------------------------- 1 file changed, 8 insertions(+), 37 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index cc3a8e6c92be..c6e10d02b795 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -48,12 +48,7 @@ static inline int32_t div_fp(int32_t x, int32_t y) } struct sample { - ktime_t start_time; - ktime_t end_time; int core_pct_busy; - int pstate_pct_busy; - u64 duration_us; - u64 idletime_us; u64 aperf; u64 mperf; int freq; @@ -91,8 +86,6 @@ struct cpudata { int min_pstate_count; int idle_mode; - ktime_t prev_sample; - u64 prev_idle_time_us; u64 prev_aperf; u64 prev_mperf; int sample_ptr; @@ -450,48 +443,26 @@ static inline void intel_pstate_calc_busy(struct cpudata *cpu, struct sample *sample) { u64 core_pct; - sample->pstate_pct_busy = 100 - div64_u64( - sample->idletime_us * 100, - sample->duration_us); core_pct = div64_u64(sample->aperf * 100, sample->mperf); sample->freq = cpu->pstate.max_pstate * core_pct * 1000; - sample->core_pct_busy = div_s64((sample->pstate_pct_busy * core_pct), - 100); + sample->core_pct_busy = core_pct; } static inline void intel_pstate_sample(struct cpudata *cpu) { - ktime_t now; - u64 idle_time_us; u64 aperf, mperf; - now = ktime_get(); - idle_time_us = get_cpu_idle_time_us(cpu->cpu, NULL); - rdmsrl(MSR_IA32_APERF, aperf); rdmsrl(MSR_IA32_MPERF, mperf); - /* for the first sample, don't actually record a sample, just - * set the baseline */ - if (cpu->prev_idle_time_us > 0) { - cpu->sample_ptr = (cpu->sample_ptr + 1) % SAMPLE_COUNT; - cpu->samples[cpu->sample_ptr].start_time = cpu->prev_sample; - cpu->samples[cpu->sample_ptr].end_time = now; - cpu->samples[cpu->sample_ptr].duration_us = - ktime_us_delta(now, cpu->prev_sample); - cpu->samples[cpu->sample_ptr].idletime_us = - idle_time_us - cpu->prev_idle_time_us; - - cpu->samples[cpu->sample_ptr].aperf = aperf; - cpu->samples[cpu->sample_ptr].mperf = mperf; - cpu->samples[cpu->sample_ptr].aperf -= cpu->prev_aperf; - cpu->samples[cpu->sample_ptr].mperf -= cpu->prev_mperf; - - intel_pstate_calc_busy(cpu, &cpu->samples[cpu->sample_ptr]); - } + cpu->sample_ptr = (cpu->sample_ptr + 1) % SAMPLE_COUNT; + cpu->samples[cpu->sample_ptr].aperf = aperf; + cpu->samples[cpu->sample_ptr].mperf = mperf; + cpu->samples[cpu->sample_ptr].aperf -= cpu->prev_aperf; + cpu->samples[cpu->sample_ptr].mperf -= cpu->prev_mperf; + + intel_pstate_calc_busy(cpu, &cpu->samples[cpu->sample_ptr]); - cpu->prev_sample = now; - cpu->prev_idle_time_us = idle_time_us; cpu->prev_aperf = aperf; cpu->prev_mperf = mperf; } -- cgit v1.2.3 From d8f469e9cff3bc4a6317d923e9506be046aa7bdc Mon Sep 17 00:00:00 2001 From: Dirk Brandewie Date: Tue, 7 May 2013 08:20:26 -0700 Subject: cpufreq / intel_pstate: use lowest requested max performance There are two ways that the maximum p-state can be clamped, via a policy change and via the sysfs file. The acpi-thermal driver adjusts the p-state policy in response to thermal events. These changes override the users settings at the moment. Use the lowest of the two requested values this ensures that we will not exceed the requested pstate from either mechanism. Reported-by: Srinivas Pandruvada Cc: 3.9+ Signed-off-by: Dirk Brandewie Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index c6e10d02b795..4a437ffc5186 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -117,6 +117,8 @@ struct perf_limits { int min_perf_pct; int32_t max_perf; int32_t min_perf; + int max_policy_pct; + int max_sysfs_pct; }; static struct perf_limits limits = { @@ -125,6 +127,8 @@ static struct perf_limits limits = { .max_perf = int_tofp(1), .min_perf_pct = 0, .min_perf = 0, + .max_policy_pct = 100, + .max_sysfs_pct = 100, }; static inline void pid_reset(struct _pid *pid, int setpoint, int busy, @@ -295,7 +299,8 @@ static ssize_t store_max_perf_pct(struct kobject *a, struct attribute *b, if (ret != 1) return -EINVAL; - limits.max_perf_pct = clamp_t(int, input, 0 , 100); + limits.max_sysfs_pct = clamp_t(int, input, 0 , 100); + limits.max_perf_pct = min(limits.max_policy_pct, limits.max_sysfs_pct); limits.max_perf = div_fp(int_tofp(limits.max_perf_pct), int_tofp(100)); return count; } @@ -646,8 +651,9 @@ static int intel_pstate_set_policy(struct cpufreq_policy *policy) limits.min_perf_pct = clamp_t(int, limits.min_perf_pct, 0 , 100); limits.min_perf = div_fp(int_tofp(limits.min_perf_pct), int_tofp(100)); - limits.max_perf_pct = policy->max * 100 / policy->cpuinfo.max_freq; - limits.max_perf_pct = clamp_t(int, limits.max_perf_pct, 0 , 100); + limits.max_policy_pct = policy->max * 100 / policy->cpuinfo.max_freq; + limits.max_policy_pct = clamp_t(int, limits.max_policy_pct, 0 , 100); + limits.max_perf_pct = min(limits.max_policy_pct, limits.max_sysfs_pct); limits.max_perf = div_fp(int_tofp(limits.max_perf_pct), int_tofp(100)); return 0; -- cgit v1.2.3 From ca182aee389f8026401510f4c63841cb02c820e8 Mon Sep 17 00:00:00 2001 From: Dirk Brandewie Date: Tue, 7 May 2013 08:20:27 -0700 Subject: cpufreq / intel_pstate: fix ffmpeg regression The ffmpeg benchmark in the phoronix test suite has threads on multiple cores that rely on the progress on of threads on other cores and ping pong back and forth fast enough to make the core appear less busy than it "should" be. If the core has been at minimum p-state for a while bump the pstate up to kick the core to see if it is in this ping pong state. If the core is truly idle the p-state will be reduced at the next sample time. If the core makes more progress it will send more work to the thread bringing both threads out of the ping pong scenario and the p-state will be selected normally. This fixes a performance regression of approximately 30% Cc: 3.9+ Signed-off-by: Dirk Brandewie Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 10 ++-------- 1 file changed, 2 insertions(+), 8 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index 4a437ffc5186..a7f1946b3452 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -551,22 +551,16 @@ static void intel_pstate_timer_func(unsigned long __data) struct cpudata *cpu = (struct cpudata *) __data; intel_pstate_sample(cpu); + intel_pstate_adjust_busy_pstate(cpu); - if (!cpu->idle_mode) - intel_pstate_adjust_busy_pstate(cpu); - else - intel_pstate_adjust_idle_pstate(cpu); - -#if defined(XPERF_FIX) if (cpu->pstate.current_pstate == cpu->pstate.min_pstate) { cpu->min_pstate_count++; if (!(cpu->min_pstate_count % 5)) { intel_pstate_set_pstate(cpu, cpu->pstate.max_pstate); - intel_pstate_idle_mode(cpu); } } else cpu->min_pstate_count = 0; -#endif + intel_pstate_set_sample_time(cpu); } -- cgit v1.2.3 From a73108d578559c83e35fa386a4058142a019b8d4 Mon Sep 17 00:00:00 2001 From: Dirk Brandewie Date: Tue, 7 May 2013 08:20:28 -0700 Subject: cpufreq / intel_pstate: Remove idle mode PID Remove dead code from the driver. Signed-off-by: Dirk Brandewie Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 49 ------------------------------------------ 1 file changed, 49 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index a7f1946b3452..b93e3851b5d3 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -81,10 +81,8 @@ struct cpudata { struct pstate_adjust_policy *pstate_policy; struct pstate_data pstate; struct _pid pid; - struct _pid idle_pid; int min_pstate_count; - int idle_mode; u64 prev_aperf; u64 prev_mperf; @@ -199,19 +197,6 @@ static inline void intel_pstate_busy_pid_reset(struct cpudata *cpu) 0); } -static inline void intel_pstate_idle_pid_reset(struct cpudata *cpu) -{ - pid_p_gain_set(&cpu->idle_pid, cpu->pstate_policy->p_gain_pct); - pid_d_gain_set(&cpu->idle_pid, cpu->pstate_policy->d_gain_pct); - pid_i_gain_set(&cpu->idle_pid, cpu->pstate_policy->i_gain_pct); - - pid_reset(&cpu->idle_pid, - 75, - 50, - cpu->pstate_policy->deadband, - 0); -} - static inline void intel_pstate_reset_all_pid(void) { unsigned int cpu; @@ -481,16 +466,6 @@ static inline void intel_pstate_set_sample_time(struct cpudata *cpu) mod_timer_pinned(&cpu->timer, jiffies + delay); } -static inline void intel_pstate_idle_mode(struct cpudata *cpu) -{ - cpu->idle_mode = 1; -} - -static inline void intel_pstate_normal_mode(struct cpudata *cpu) -{ - cpu->idle_mode = 0; -} - static inline int intel_pstate_get_scaled_busy(struct cpudata *cpu) { int32_t busy_scaled; @@ -523,29 +498,6 @@ static inline void intel_pstate_adjust_busy_pstate(struct cpudata *cpu) intel_pstate_pstate_decrease(cpu, steps); } -static inline void intel_pstate_adjust_idle_pstate(struct cpudata *cpu) -{ - int busy_scaled; - struct _pid *pid; - int ctl = 0; - int steps; - - pid = &cpu->idle_pid; - - busy_scaled = intel_pstate_get_scaled_busy(cpu); - - ctl = pid_calc(pid, 100 - busy_scaled); - - steps = abs(ctl); - if (ctl < 0) - intel_pstate_pstate_decrease(cpu, steps); - else - intel_pstate_pstate_increase(cpu, steps); - - if (cpu->pstate.current_pstate == cpu->pstate.min_pstate) - intel_pstate_normal_mode(cpu); -} - static void intel_pstate_timer_func(unsigned long __data) { struct cpudata *cpu = (struct cpudata *) __data; @@ -601,7 +553,6 @@ static int intel_pstate_init_cpu(unsigned int cpunum) (unsigned long)cpu; cpu->timer.expires = jiffies + HZ/100; intel_pstate_busy_pid_reset(cpu); - intel_pstate_idle_pid_reset(cpu); intel_pstate_sample(cpu); intel_pstate_set_pstate(cpu, cpu->pstate.max_pstate); -- cgit v1.2.3 From 35363e943f2b0aa503e1dd55f894f736563e85a3 Mon Sep 17 00:00:00 2001 From: Dirk Brandewie Date: Tue, 7 May 2013 08:20:30 -0700 Subject: cpufreq / intel_pstate: remove #ifdef MODULE compile fence The driver can no longer be built as a module remove the compile fence around cpufreq tracing call. Signed-off-by: Dirk Brandewie Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index b93e3851b5d3..0cc7d60525ac 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -391,9 +391,8 @@ static void intel_pstate_set_pstate(struct cpudata *cpu, int pstate) if (pstate == cpu->pstate.current_pstate) return; -#ifndef MODULE trace_cpu_frequency(pstate * 100000, cpu->cpu); -#endif + cpu->pstate.current_pstate = pstate; wrmsrl(MSR_IA32_PERF_CTL, pstate << 8); -- cgit v1.2.3 From d96f7330176d69a9415162fb341b87e724e9ca74 Mon Sep 17 00:00:00 2001 From: Wolfram Sang Date: Fri, 10 May 2013 08:16:48 +0000 Subject: cpufreq / kirkwood: don't check resource with devm_ioremap_resource devm_ioremap_resource does sanity checks on the given resource. No need to duplicate this in the driver. Signed-off-by: Wolfram Sang Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/kirkwood-cpufreq.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/kirkwood-cpufreq.c b/drivers/cpufreq/kirkwood-cpufreq.c index d36ea8dc96eb..b2644af985ec 100644 --- a/drivers/cpufreq/kirkwood-cpufreq.c +++ b/drivers/cpufreq/kirkwood-cpufreq.c @@ -171,10 +171,6 @@ static int kirkwood_cpufreq_probe(struct platform_device *pdev) priv.dev = &pdev->dev; res = platform_get_resource(pdev, IORESOURCE_MEM, 0); - if (!res) { - dev_err(&pdev->dev, "Cannot get memory resource\n"); - return -ENODEV; - } priv.base = devm_ioremap_resource(&pdev->dev, res); if (IS_ERR(priv.base)) return PTR_ERR(priv.base); -- cgit v1.2.3 From 60e6726c7bdec7fedae278ee3bf973ba988f2abf Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Wed, 8 May 2013 21:34:48 +0200 Subject: cpufreq, ondemand: Remove leftover debug line I don't see how the virtual address of the tuners pointer would be of any help to anyone so remove it. Signed-off-by: Borislav Petkov Acked-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq_ondemand.c | 1 - 1 file changed, 1 deletion(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/cpufreq_ondemand.c b/drivers/cpufreq/cpufreq_ondemand.c index b0ffef96bf77..4b9bb5def6f1 100644 --- a/drivers/cpufreq/cpufreq_ondemand.c +++ b/drivers/cpufreq/cpufreq_ondemand.c @@ -547,7 +547,6 @@ static int od_init(struct dbs_data *dbs_data) tuners->io_is_busy = should_io_be_busy(); dbs_data->tuners = tuners; - pr_info("%s: tuners %p\n", __func__, tuners); mutex_init(&dbs_data->mutex); return 0; } -- cgit v1.2.3 From b57ffac5e57bff33dde3cff35dff5c41876a6d12 Mon Sep 17 00:00:00 2001 From: Wei Yongjun Date: Mon, 13 May 2013 08:03:43 +0000 Subject: cpufreq / intel_pstate: use vzalloc() instead of vmalloc()/memset(0) Use vzalloc() instead of vmalloc() and memset(0). Signed-off-by: Wei Yongjun Acked-by: Viresh Kumar Acked-by: Dirk Brandewie Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index 0cc7d60525ac..9c36ace92a39 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -709,10 +709,9 @@ static int __init intel_pstate_init(void) pr_info("Intel P-state driver initializing.\n"); - all_cpu_data = vmalloc(sizeof(void *) * num_possible_cpus()); + all_cpu_data = vzalloc(sizeof(void *) * num_possible_cpus()); if (!all_cpu_data) return -ENOMEM; - memset(all_cpu_data, 0, sizeof(void *) * num_possible_cpus()); rc = cpufreq_register_driver(&intel_pstate_driver); if (rc) -- cgit v1.2.3 From a66b2e503fc79fff6632d02ef5a0ee47c1d2553d Mon Sep 17 00:00:00 2001 From: "Srivatsa S. Bhat" Date: Wed, 15 May 2013 21:47:17 +0200 Subject: cpufreq: Preserve sysfs files across suspend/resume The file permissions of cpufreq per-cpu sysfs files are not preserved across suspend/resume because we internally go through the CPU Hotplug path which reinitializes the file permissions on CPU online. But the user is not supposed to know that we are using CPU hotplug internally within suspend/resume (IOW, the kernel should not silently wreck the user-set file permissions across a suspend cycle). Therefore, we need to preserve the file permissions as they are across suspend/resume. The simplest way to achieve that is to just not touch the sysfs files at all - ie., just ignore the CPU hotplug notifications in the suspend/resume path (_FROZEN) in the cpufreq hotplug callback. Reported-by: Robert Jarzmik Reported-by: Durgadoss R Signed-off-by: Srivatsa S. Bhat Acked-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq.c | 4 +--- drivers/cpufreq/cpufreq_stats.c | 7 ++++--- 2 files changed, 5 insertions(+), 6 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index b7acfd153bf9..4b8c7f297d74 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -1832,15 +1832,13 @@ static int __cpuinit cpufreq_cpu_callback(struct notifier_block *nfb, if (dev) { switch (action) { case CPU_ONLINE: - case CPU_ONLINE_FROZEN: cpufreq_add_dev(dev, NULL); break; case CPU_DOWN_PREPARE: - case CPU_DOWN_PREPARE_FROZEN: + case CPU_UP_CANCELED_FROZEN: __cpufreq_remove_dev(dev, NULL); break; case CPU_DOWN_FAILED: - case CPU_DOWN_FAILED_FROZEN: cpufreq_add_dev(dev, NULL); break; } diff --git a/drivers/cpufreq/cpufreq_stats.c b/drivers/cpufreq/cpufreq_stats.c index bfd6273fd873..fb65decffa28 100644 --- a/drivers/cpufreq/cpufreq_stats.c +++ b/drivers/cpufreq/cpufreq_stats.c @@ -349,15 +349,16 @@ static int __cpuinit cpufreq_stat_cpu_callback(struct notifier_block *nfb, switch (action) { case CPU_ONLINE: - case CPU_ONLINE_FROZEN: cpufreq_update_policy(cpu); break; case CPU_DOWN_PREPARE: - case CPU_DOWN_PREPARE_FROZEN: cpufreq_stats_free_sysfs(cpu); break; case CPU_DEAD: - case CPU_DEAD_FROZEN: + cpufreq_stats_free_table(cpu); + break; + case CPU_UP_CANCELED_FROZEN: + cpufreq_stats_free_sysfs(cpu); cpufreq_stats_free_table(cpu); break; } -- cgit v1.2.3 From 955ef4833574636819cd269cfbae12f79cbde63a Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Thu, 16 May 2013 05:09:58 +0000 Subject: cpufreq: Drop rwsem lock around CPUFREQ_GOV_POLICY_EXIT With the rwsem lock around __cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT), we get circular dependency when we call sysfs_remove_group(). ====================================================== [ INFO: possible circular locking dependency detected ] 3.9.0-rc7+ #15 Not tainted ------------------------------------------------------- cat/2387 is trying to acquire lock: (&per_cpu(cpu_policy_rwsem, cpu)){+++++.}, at: [] lock_policy_rwsem_read+0x25/0x34 but task is already holding lock: (s_active#41){++++.+}, at: [] sysfs_read_file+0x4f/0xcc which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #1 (s_active#41){++++.+}: [] lock_acquire+0x61/0xbc [] sysfs_addrm_finish+0xc1/0x128 [] sysfs_hash_and_remove+0x35/0x64 [] remove_files.isra.0+0x1b/0x24 [] sysfs_remove_group+0x2d/0xa8 [] cpufreq_governor_interactive+0x13b/0x35c [] __cpufreq_governor+0x2b/0x8c [] __cpufreq_set_policy+0xa9/0xf8 [] store_scaling_governor+0x61/0x100 [] store+0x39/0x60 [] sysfs_write_file+0xed/0x114 [] vfs_write+0x65/0xd8 [] sys_write+0x2f/0x50 [] ret_fast_syscall+0x1/0x52 -> #0 (&per_cpu(cpu_policy_rwsem, cpu)){+++++.}: [] __lock_acquire+0xef3/0x13dc [] lock_acquire+0x61/0xbc [] down_read+0x25/0x30 [] lock_policy_rwsem_read+0x25/0x34 [] show+0x21/0x58 [] sysfs_read_file+0x67/0xcc [] vfs_read+0x63/0xd8 [] sys_read+0x2f/0x50 [] ret_fast_syscall+0x1/0x52 other info that might help us debug this: Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(s_active#41); lock(&per_cpu(cpu_policy_rwsem, cpu)); lock(s_active#41); lock(&per_cpu(cpu_policy_rwsem, cpu)); *** DEADLOCK *** 2 locks held by cat/2387: #0: (&buffer->mutex){+.+.+.}, at: [] sysfs_read_file+0x25/0xcc #1: (s_active#41){++++.+}, at: [] sysfs_read_file+0x4f/0xcc stack backtrace: [] (unwind_backtrace+0x1/0x9c) from [] (print_circular_bug+0x19d/0x1e8) [] (print_circular_bug+0x19d/0x1e8) from [] (__lock_acquire+0xef3/0x13dc) [] (__lock_acquire+0xef3/0x13dc) from [] (lock_acquire+0x61/0xbc) [] (lock_acquire+0x61/0xbc) from [] (down_read+0x25/0x30) [] (down_read+0x25/0x30) from [] (lock_policy_rwsem_read+0x25/0x34) [] (lock_policy_rwsem_read+0x25/0x34) from [] (show+0x21/0x58) [] (show+0x21/0x58) from [] (sysfs_read_file+0x67/0xcc) [] (sysfs_read_file+0x67/0xcc) from [] (vfs_read+0x63/0xd8) [] (vfs_read+0x63/0xd8) from [] (sys_read+0x2f/0x50) [] (sys_read+0x2f/0x50) from [] (ret_fast_syscall+0x1/0x52) This lock isn't required while calling __cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT). Remove it. Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c index 4b8c7f297d74..2d53f47d1747 100644 --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -1729,18 +1729,23 @@ static int __cpufreq_set_policy(struct cpufreq_policy *data, /* end old governor */ if (data->governor) { __cpufreq_governor(data, CPUFREQ_GOV_STOP); + unlock_policy_rwsem_write(policy->cpu); __cpufreq_governor(data, CPUFREQ_GOV_POLICY_EXIT); + lock_policy_rwsem_write(policy->cpu); } /* start new governor */ data->governor = policy->governor; if (!__cpufreq_governor(data, CPUFREQ_GOV_POLICY_INIT)) { - if (!__cpufreq_governor(data, CPUFREQ_GOV_START)) + if (!__cpufreq_governor(data, CPUFREQ_GOV_START)) { failed = 0; - else + } else { + unlock_policy_rwsem_write(policy->cpu); __cpufreq_governor(data, CPUFREQ_GOV_POLICY_EXIT); + lock_policy_rwsem_write(policy->cpu); + } } if (failed) { -- cgit v1.2.3 From c96d53d600643ee0adfd1cb90814bd9510e62b71 Mon Sep 17 00:00:00 2001 From: Dirk Brandewie Date: Fri, 17 May 2013 16:10:24 +0000 Subject: cpufreq / intel_pstate: Add additional supported CPU ID Add CPU ID for Ivybrigde processor. Signed-off-by: Dirk Brandewie Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index 9c36ace92a39..07f2840ad805 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -521,6 +521,7 @@ static void intel_pstate_timer_func(unsigned long __data) static const struct x86_cpu_id intel_pstate_cpu_ids[] = { ICPU(0x2a, default_policy), ICPU(0x2d, default_policy), + ICPU(0x3a, default_policy), {} }; MODULE_DEVICE_TABLE(x86cpu, intel_pstate_cpu_ids); -- cgit v1.2.3 From fb40bc3e94933007d3e42e96daf1ec8044821cb8 Mon Sep 17 00:00:00 2001 From: Ralf Baechle Date: Tue, 21 May 2013 14:05:27 +0200 Subject: MIPS: Idle: Re-enable irqs at the end of r3081, au1k and loongson2 cpu_wait. Without this, the WARN_ON_ONCE(irqs_disabled()); in the idle loop will be triggered. Signed-off-by: Ralf Baechle --- arch/mips/kernel/idle.c | 2 ++ drivers/cpufreq/loongson2_cpufreq.c | 1 + 2 files changed, 3 insertions(+) (limited to 'drivers/cpufreq') diff --git a/arch/mips/kernel/idle.c b/arch/mips/kernel/idle.c index 36e79f528e89..78cc7d6fc845 100644 --- a/arch/mips/kernel/idle.c +++ b/arch/mips/kernel/idle.c @@ -34,6 +34,7 @@ static void r3081_wait(void) { unsigned long cfg = read_c0_conf(); write_c0_conf(cfg | R30XX_CONF_HALT); + local_irq_enable(); } static void r39xx_wait(void) @@ -109,6 +110,7 @@ static void au1k_wait(void) " nop \n" " .set mips0 \n" : : "r" (au1k_wait)); + local_irq_enable(); } static int __initdata nowait; diff --git a/drivers/cpufreq/loongson2_cpufreq.c b/drivers/cpufreq/loongson2_cpufreq.c index 84889573b566..868976d443a6 100644 --- a/drivers/cpufreq/loongson2_cpufreq.c +++ b/drivers/cpufreq/loongson2_cpufreq.c @@ -200,6 +200,7 @@ static void loongson2_cpu_wait(void) LOONGSON_CHIPCFG0 &= ~0x7; /* Put CPU into wait mode */ LOONGSON_CHIPCFG0 = cpu_freq; /* Restore CPU state */ spin_unlock_irqrestore(&loongson2_wait_lock, flags); + local_irq_enable(); } static int __init cpufreq_init(void) -- cgit v1.2.3 From bdc92d74e0ec95a8101447467c25f015105f2e5a Mon Sep 17 00:00:00 2001 From: Ralf Baechle Date: Tue, 21 May 2013 16:59:19 +0200 Subject: MIPS: Idle: Consolidate all declarations in . Signed-off-by: Ralf Baechle --- arch/mips/alchemy/board-gpr.c | 1 + arch/mips/alchemy/common/time.c | 1 + arch/mips/ath79/setup.c | 1 + arch/mips/cobalt/reset.c | 1 + arch/mips/include/asm/idle.h | 11 +++++++++++ arch/mips/include/asm/processor.h | 1 - arch/mips/kernel/idle.c | 3 +-- arch/mips/kernel/proc.c | 1 + arch/mips/kernel/smp.c | 1 + arch/mips/kernel/smtc.c | 2 +- arch/mips/kernel/traps.c | 2 +- arch/mips/loongson/common/reset.c | 1 + arch/mips/loongson1/common/reset.c | 1 + arch/mips/netlogic/xlp/setup.c | 1 + arch/mips/netlogic/xlr/setup.c | 1 + arch/mips/pmcs-msp71xx/msp_setup.c | 1 + arch/mips/txx9/generic/setup.c | 1 + arch/mips/vr41xx/common/pmu.c | 1 + arch/mips/wrppmc/reset.c | 1 + drivers/cpufreq/loongson2_cpufreq.c | 1 + 20 files changed, 29 insertions(+), 5 deletions(-) create mode 100644 arch/mips/include/asm/idle.h (limited to 'drivers/cpufreq') diff --git a/arch/mips/alchemy/board-gpr.c b/arch/mips/alchemy/board-gpr.c index cb0f6afb7389..9edc35ff8cf1 100644 --- a/arch/mips/alchemy/board-gpr.c +++ b/arch/mips/alchemy/board-gpr.c @@ -31,6 +31,7 @@ #include #include #include +#include #include #include #include diff --git a/arch/mips/alchemy/common/time.c b/arch/mips/alchemy/common/time.c index 38afb11ba2c4..93fa586d52e2 100644 --- a/arch/mips/alchemy/common/time.c +++ b/arch/mips/alchemy/common/time.c @@ -36,6 +36,7 @@ #include #include +#include #include #include #include diff --git a/arch/mips/ath79/setup.c b/arch/mips/ath79/setup.c index a0233a2c1988..8be4e856b8b8 100644 --- a/arch/mips/ath79/setup.c +++ b/arch/mips/ath79/setup.c @@ -19,6 +19,7 @@ #include #include +#include #include /* for mips_hpt_frequency */ #include /* for _machine_{restart,halt} */ #include diff --git a/arch/mips/cobalt/reset.c b/arch/mips/cobalt/reset.c index 516b4428df4e..4eedd481dd00 100644 --- a/arch/mips/cobalt/reset.c +++ b/arch/mips/cobalt/reset.c @@ -12,6 +12,7 @@ #include #include +#include #include #include diff --git a/arch/mips/include/asm/idle.h b/arch/mips/include/asm/idle.h new file mode 100644 index 000000000000..8b26ac5c854e --- /dev/null +++ b/arch/mips/include/asm/idle.h @@ -0,0 +1,11 @@ +#ifndef __ASM_IDLE_H +#define __ASM_IDLE_H + +#include + +extern void (*cpu_wait)(void); +extern asmlinkage void r4k_wait(void); +extern void r4k_wait_irqoff(void); +extern void __pastwait(void); + +#endif /* __ASM_IDLE_H */ diff --git a/arch/mips/include/asm/processor.h b/arch/mips/include/asm/processor.h index 71686c897dea..1470b7b68b0e 100644 --- a/arch/mips/include/asm/processor.h +++ b/arch/mips/include/asm/processor.h @@ -28,7 +28,6 @@ /* * System setup and hardware flags.. */ -extern void (*cpu_wait)(void); extern unsigned int vced_count, vcei_count; diff --git a/arch/mips/kernel/idle.c b/arch/mips/kernel/idle.c index 1f85dda03df4..985cc02786e3 100644 --- a/arch/mips/kernel/idle.c +++ b/arch/mips/kernel/idle.c @@ -18,6 +18,7 @@ #include #include #include +#include #include /* @@ -44,8 +45,6 @@ static void r39xx_wait(void) local_irq_enable(); } -extern void r4k_wait(void); - /* * This variant is preferable as it allows testing need_resched and going to * sleep depending on the outcome atomically. Unfortunately the "It is diff --git a/arch/mips/kernel/proc.c b/arch/mips/kernel/proc.c index a3e461408b7e..acb34373679e 100644 --- a/arch/mips/kernel/proc.c +++ b/arch/mips/kernel/proc.c @@ -10,6 +10,7 @@ #include #include #include +#include #include #include #include diff --git a/arch/mips/kernel/smp.c b/arch/mips/kernel/smp.c index c17619fe18e3..6e7862ab46cc 100644 --- a/arch/mips/kernel/smp.c +++ b/arch/mips/kernel/smp.c @@ -37,6 +37,7 @@ #include #include #include +#include #include #include #include diff --git a/arch/mips/kernel/smtc.c b/arch/mips/kernel/smtc.c index 7186222dc5bb..46303bc61364 100644 --- a/arch/mips/kernel/smtc.c +++ b/arch/mips/kernel/smtc.c @@ -34,6 +34,7 @@ #include #include #include +#include #include #include #include @@ -858,7 +859,6 @@ void smtc_send_ipi(int cpu, int type, unsigned int action) unsigned long flags; int mtflags; unsigned long tcrestart; - extern void r4k_wait_irqoff(void), __pastwait(void); int set_resched_flag = (type == LINUX_SMP_IPI && action == SMP_RESCHEDULE_YOURSELF); diff --git a/arch/mips/kernel/traps.c b/arch/mips/kernel/traps.c index cb14db3c5764..c6da4a905b98 100644 --- a/arch/mips/kernel/traps.c +++ b/arch/mips/kernel/traps.c @@ -41,6 +41,7 @@ #include #include #include +#include #include #include #include @@ -57,7 +58,6 @@ #include extern void check_wait(void); -extern asmlinkage void r4k_wait(void); extern asmlinkage void rollback_handle_int(void); extern asmlinkage void handle_int(void); extern u32 handle_tlbl[]; diff --git a/arch/mips/loongson/common/reset.c b/arch/mips/loongson/common/reset.c index 35c8c6468494..65bfbb5d06f4 100644 --- a/arch/mips/loongson/common/reset.c +++ b/arch/mips/loongson/common/reset.c @@ -12,6 +12,7 @@ #include #include +#include #include #include diff --git a/arch/mips/loongson1/common/reset.c b/arch/mips/loongson1/common/reset.c index d4f610f9604a..547f34b69e4c 100644 --- a/arch/mips/loongson1/common/reset.c +++ b/arch/mips/loongson1/common/reset.c @@ -9,6 +9,7 @@ #include #include +#include #include #include diff --git a/arch/mips/netlogic/xlp/setup.c b/arch/mips/netlogic/xlp/setup.c index af319143b591..eaa99d28cb8e 100644 --- a/arch/mips/netlogic/xlp/setup.c +++ b/arch/mips/netlogic/xlp/setup.c @@ -37,6 +37,7 @@ #include #include +#include #include #include #include diff --git a/arch/mips/netlogic/xlr/setup.c b/arch/mips/netlogic/xlr/setup.c index e3e094100e3e..89c8c1066632 100644 --- a/arch/mips/netlogic/xlr/setup.c +++ b/arch/mips/netlogic/xlr/setup.c @@ -36,6 +36,7 @@ #include #include +#include #include #include #include diff --git a/arch/mips/pmcs-msp71xx/msp_setup.c b/arch/mips/pmcs-msp71xx/msp_setup.c index 1651cfdbfe7b..396b2967ad85 100644 --- a/arch/mips/pmcs-msp71xx/msp_setup.c +++ b/arch/mips/pmcs-msp71xx/msp_setup.c @@ -12,6 +12,7 @@ #include #include +#include #include #include #include diff --git a/arch/mips/txx9/generic/setup.c b/arch/mips/txx9/generic/setup.c index 5364aabc2102..681e7f86c080 100644 --- a/arch/mips/txx9/generic/setup.c +++ b/arch/mips/txx9/generic/setup.c @@ -26,6 +26,7 @@ #include #include #include +#include #include #include #include diff --git a/arch/mips/vr41xx/common/pmu.c b/arch/mips/vr41xx/common/pmu.c index 70a3f90131d8..d7f755833c3f 100644 --- a/arch/mips/vr41xx/common/pmu.c +++ b/arch/mips/vr41xx/common/pmu.c @@ -27,6 +27,7 @@ #include #include +#include #include #include #include diff --git a/arch/mips/wrppmc/reset.c b/arch/mips/wrppmc/reset.c index cc5474b24f06..80beb188ed47 100644 --- a/arch/mips/wrppmc/reset.c +++ b/arch/mips/wrppmc/reset.c @@ -9,6 +9,7 @@ #include #include +#include #include #include diff --git a/drivers/cpufreq/loongson2_cpufreq.c b/drivers/cpufreq/loongson2_cpufreq.c index 868976d443a6..d53912768946 100644 --- a/drivers/cpufreq/loongson2_cpufreq.c +++ b/drivers/cpufreq/loongson2_cpufreq.c @@ -18,6 +18,7 @@ #include #include +#include #include -- cgit v1.2.3 From b5f14720a6421aab841d9f03f0129cfbe7db5133 Mon Sep 17 00:00:00 2001 From: Rafał Bilski Date: Sun, 19 May 2013 19:27:55 +0000 Subject: cpufreq / e_powersaver: Fix linker error when ACPI processor is a module on i386: CONFIG_ACPI_PROCESSOR=m CONFIG_X86_E_POWERSAVER=y drivers/built-in.o: In function `eps_cpu_init.part.8': e_powersaver.c:(.text.unlikely+0x2243): undefined reference to `acpi_processor_register_performance' e_powersaver.c:(.text.unlikely+0x22a2): undefined reference to `acpi_processor_unregister_performance' e_powersaver.c:(.text.unlikely+0x246b): undefined reference to `acpi_processor_get_bios_limit' X86_E_POWERSAVER should also depend on ACPI_PROCESSOR. Signed-off-by: Rafal Bilski Acked-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/Kconfig.x86 | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/Kconfig.x86 b/drivers/cpufreq/Kconfig.x86 index 2b8a8c374548..6bd63d63d356 100644 --- a/drivers/cpufreq/Kconfig.x86 +++ b/drivers/cpufreq/Kconfig.x86 @@ -272,7 +272,7 @@ config X86_LONGHAUL config X86_E_POWERSAVER tristate "VIA C7 Enhanced PowerSaver (DANGEROUS)" select CPU_FREQ_TABLE - depends on X86_32 + depends on X86_32 && ACPI_PROCESSOR help This adds the CPUFreq driver for VIA C7 processors. However, this driver does not have any safeguards to prevent operating the CPU out of spec -- cgit v1.2.3 From 92a9b5c291c72aa9899021699458f0b6e328b940 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Fri, 17 May 2013 11:25:11 +0000 Subject: cpufreq: arm_big_little_dt: Register driver only if DT has valid data If arm_big_little_dt driver is enabled, then it will always try to register with big LITTLE cpufreq core driver. In case DT doesn't have relevant data for cpu nodes, i.e. operating points aren't present, then we should exit early and shouldn't register with big LITTLE cpufreq core driver. Otherwise we will fail continuously from the driver->init() routine. This patch fixes this issue. Reported-and-tested-by: Jon Medhurst Reviewed-by: Jon Medhurst Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/arm_big_little_dt.c | 73 +++++++++++++++++++++---------------- 1 file changed, 42 insertions(+), 31 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/arm_big_little_dt.c b/drivers/cpufreq/arm_big_little_dt.c index 173ed059d95f..27e2f45ccdd5 100644 --- a/drivers/cpufreq/arm_big_little_dt.c +++ b/drivers/cpufreq/arm_big_little_dt.c @@ -19,6 +19,7 @@ #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt +#include #include #include #include @@ -29,60 +30,63 @@ #include #include "arm_big_little.h" -static int dt_init_opp_table(struct device *cpu_dev) +/* get cpu node with valid operating-points */ +static struct device_node *get_cpu_node_with_valid_op(int cpu) { - struct device_node *np, *parent; - int count = 0, ret; + struct device_node *np = NULL, *parent; + int count = 0; parent = of_find_node_by_path("/cpus"); if (!parent) { pr_err("failed to find OF /cpus\n"); - return -ENOENT; + return NULL; } for_each_child_of_node(parent, np) { - if (count++ != cpu_dev->id) + if (count++ != cpu) continue; if (!of_get_property(np, "operating-points", NULL)) { - ret = -ENODATA; - } else { - cpu_dev->of_node = np; - ret = of_init_opp_table(cpu_dev); + of_node_put(np); + np = NULL; } - of_node_put(np); - of_node_put(parent); - return ret; + break; } - return -ENODEV; + of_node_put(parent); + return np; +} + +static int dt_init_opp_table(struct device *cpu_dev) +{ + struct device_node *np; + int ret; + + np = get_cpu_node_with_valid_op(cpu_dev->id); + if (!np) + return -ENODATA; + + cpu_dev->of_node = np; + ret = of_init_opp_table(cpu_dev); + of_node_put(np); + + return ret; } static int dt_get_transition_latency(struct device *cpu_dev) { - struct device_node *np, *parent; + struct device_node *np; u32 transition_latency = CPUFREQ_ETERNAL; - int count = 0; - parent = of_find_node_by_path("/cpus"); - if (!parent) { - pr_info("Failed to find OF /cpus. Use CPUFREQ_ETERNAL transition latency\n"); + np = get_cpu_node_with_valid_op(cpu_dev->id); + if (!np) return CPUFREQ_ETERNAL; - } - - for_each_child_of_node(parent, np) { - if (count++ != cpu_dev->id) - continue; - - of_property_read_u32(np, "clock-latency", &transition_latency); - of_node_put(np); - of_node_put(parent); - return transition_latency; - } + of_property_read_u32(np, "clock-latency", &transition_latency); + of_node_put(np); - pr_info("clock-latency isn't found, use CPUFREQ_ETERNAL transition latency\n"); - return CPUFREQ_ETERNAL; + pr_debug("%s: clock-latency: %d\n", __func__, transition_latency); + return transition_latency; } static struct cpufreq_arm_bL_ops dt_bL_ops = { @@ -93,6 +97,13 @@ static struct cpufreq_arm_bL_ops dt_bL_ops = { static int generic_bL_init(void) { + struct device_node *np; + + np = get_cpu_node_with_valid_op(0); + if (!np) + return -ENODEV; + + of_node_put(np); return bL_cpufreq_register(&dt_bL_ops); } module_init(generic_bL_init); -- cgit v1.2.3 From 9076eaca60b3796b3b97d1914c4924c4bc39f066 Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Mon, 20 May 2013 09:57:17 +0530 Subject: cpufreq: arm_big_little_dt: Instantiate as platform_driver As multiplatform build is being adopted by more and more ARM platforms, initcall function should be used very carefully. For example, when both arm_big_little_dt and cpufreq-cpu0 drivers are compiled in, arm_big_little_dt driver may try to register even if we had platform device for cpufreq-cpu0 registered. To eliminate this undesired the effect, the patch changes arm_big_little_dt driver to have it instantiated as a platform_driver. Then it will only run on platforms that create the platform_device "arm-bL-cpufreq-dt". Reported-and-tested-by: Rob Herring Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/arm_big_little_dt.c | 20 +++++++++++++++----- 1 file changed, 15 insertions(+), 5 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/arm_big_little_dt.c b/drivers/cpufreq/arm_big_little_dt.c index 27e2f45ccdd5..fd9e3ea6a480 100644 --- a/drivers/cpufreq/arm_big_little_dt.c +++ b/drivers/cpufreq/arm_big_little_dt.c @@ -26,6 +26,7 @@ #include #include #include +#include #include #include #include "arm_big_little.h" @@ -95,7 +96,7 @@ static struct cpufreq_arm_bL_ops dt_bL_ops = { .init_opp_table = dt_init_opp_table, }; -static int generic_bL_init(void) +static int generic_bL_probe(struct platform_device *pdev) { struct device_node *np; @@ -106,13 +107,22 @@ static int generic_bL_init(void) of_node_put(np); return bL_cpufreq_register(&dt_bL_ops); } -module_init(generic_bL_init); -static void generic_bL_exit(void) +static int generic_bL_remove(struct platform_device *pdev) { - return bL_cpufreq_unregister(&dt_bL_ops); + bL_cpufreq_unregister(&dt_bL_ops); + return 0; } -module_exit(generic_bL_exit); + +static struct platform_driver generic_bL_platdrv = { + .driver = { + .name = "arm-bL-cpufreq-dt", + .owner = THIS_MODULE, + }, + .probe = generic_bL_probe, + .remove = generic_bL_remove, +}; +module_platform_driver(generic_bL_platdrv); MODULE_AUTHOR("Viresh Kumar "); MODULE_DESCRIPTION("Generic ARM big LITTLE cpufreq driver via DT"); -- cgit v1.2.3 From 8673b83bf2f013379453b4779047bf3c6ae387e4 Mon Sep 17 00:00:00 2001 From: Ross Lagerwall Date: Fri, 31 May 2013 20:45:17 +0100 Subject: acpi-cpufreq: set current frequency based on target P-State Commit 4b31e774 (Always set P-state on initialization) fixed bug #4634 and caused the driver to always set the target P-State at least once since the initial P-State may not be the desired one. Commit 5a1c0228 (cpufreq: Avoid calling cpufreq driver's target() routine if target_freq == policy->cur) caused a regression in this behavior. This fixes the regression by setting policy->cur based on the CPU's target frequency rather than the CPU's current reported frequency (which may be different). This means that the P-State will be set initially if the CPU's target frequency is different from the governor's target frequency. This fixes an issue where setting the default governor to performance wouldn't correctly enable turbo mode on all cores. Signed-off-by: Ross Lagerwall Reviewed-by: Len Brown Acked-by: Viresh Kumar Cc: 3.8+ Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/acpi-cpufreq.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/acpi-cpufreq.c b/drivers/cpufreq/acpi-cpufreq.c index 11b8b4b54ceb..edc089e9d0c4 100644 --- a/drivers/cpufreq/acpi-cpufreq.c +++ b/drivers/cpufreq/acpi-cpufreq.c @@ -347,11 +347,11 @@ static u32 get_cur_val(const struct cpumask *mask) switch (per_cpu(acfreq_data, cpumask_first(mask))->cpu_feature) { case SYSTEM_INTEL_MSR_CAPABLE: cmd.type = SYSTEM_INTEL_MSR_CAPABLE; - cmd.addr.msr.reg = MSR_IA32_PERF_STATUS; + cmd.addr.msr.reg = MSR_IA32_PERF_CTL; break; case SYSTEM_AMD_MSR_CAPABLE: cmd.type = SYSTEM_AMD_MSR_CAPABLE; - cmd.addr.msr.reg = MSR_AMD_PERF_STATUS; + cmd.addr.msr.reg = MSR_AMD_PERF_CTL; break; case SYSTEM_IO_CAPABLE: cmd.type = SYSTEM_IO_CAPABLE; -- cgit v1.2.3 From 2f7021a815f20f3481c10884fe9735ce2a56db35 Mon Sep 17 00:00:00 2001 From: Michael Wang Date: Wed, 5 Jun 2013 08:49:37 +0000 Subject: cpufreq: protect 'policy->cpus' from offlining during __gov_queue_work() Jiri Kosina and Borislav Petkov reported the warning: [ 51.616759] ------------[ cut here ]------------ [ 51.621460] WARNING: at arch/x86/kernel/smp.c:123 native_smp_send_reschedule+0x58/0x60() [ 51.629638] Modules linked in: ext2 vfat fat loop snd_hda_codec_hdmi usbhid snd_hda_codec_realtek coretemp kvm_intel kvm snd_hda_intel snd_hda_codec crc32_pclmul crc32c_intel ghash_clmulni_intel snd_hwdep snd_pcm aesni_intel sb_edac aes_x86_64 ehci_pci snd_page_alloc glue_helper snd_timer xhci_hcd snd iTCO_wdt iTCO_vendor_support ehci_hcd edac_core lpc_ich acpi_cpufreq lrw gf128mul ablk_helper cryptd mperf usbcore usb_common soundcore mfd_core dcdbas evdev pcspkr processor i2c_i801 button microcode [ 51.675581] CPU: 0 PID: 244 Comm: kworker/1:1 Tainted: G W 3.10.0-rc1+ #10 [ 51.683407] Hardware name: Dell Inc. Precision T3600/0PTTT9, BIOS A08 01/24/2013 [ 51.690901] Workqueue: events od_dbs_timer [ 51.695069] 0000000000000009 ffff88043a2f5b68 ffffffff8161441c ffff88043a2f5ba8 [ 51.702602] ffffffff8103e540 0000000000000033 0000000000000001 ffff88043d5f8000 [ 51.710136] 00000000ffff0ce1 0000000000000001 ffff88044fc4fc08 ffff88043a2f5bb8 [ 51.717691] Call Trace: [ 51.720191] [] dump_stack+0x19/0x1b [ 51.725396] [] warn_slowpath_common+0x70/0xa0 [ 51.731473] [] warn_slowpath_null+0x1a/0x20 [ 51.737378] [] native_smp_send_reschedule+0x58/0x60 [ 51.744013] [] wake_up_nohz_cpu+0x2d/0xa0 [ 51.749745] [] add_timer_on+0x8f/0x110 [ 51.755214] [] __queue_delayed_work+0x16e/0x1a0 [ 51.761470] [] ? try_to_grab_pending+0xd1/0x1a0 [ 51.767724] [] mod_delayed_work_on+0x5a/0xa0 [ 51.773719] [] gov_queue_work+0x4d/0xc0 [ 51.779271] [] od_dbs_timer+0xcb/0x170 [ 51.784734] [] process_one_work+0x1fd/0x540 [ 51.790634] [] ? process_one_work+0x192/0x540 [ 51.796711] [] worker_thread+0x122/0x380 [ 51.802350] [] ? rescuer_thread+0x320/0x320 [ 51.808264] [] kthread+0xea/0xf0 [ 51.813200] [] ? flush_kthread_worker+0x150/0x150 [ 51.819644] [] ret_from_fork+0x7c/0xb0 [ 51.918165] nouveau E[ DRM] GPU lockup - switching to software fbcon [ 51.930505] [] ? flush_kthread_worker+0x150/0x150 [ 51.936994] ---[ end trace f419538ada83b5c5 ]--- It was caused by the policy->cpus changed during the process of __gov_queue_work(), in other word, cpu offline happened. Use get/put_online_cpus() to prevent the offline from happening while __gov_queue_work() is running. [rjw: The problem has been present since recent commit 031299b (cpufreq: governors: Avoid unnecessary per cpu timer interrupts)] References: https://lkml.org/lkml/2013/6/5/88 Reported-by: Borislav Petkov Reported-and-tested-by: Jiri Kosina Signed-off-by: Michael Wang Acked-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq_governor.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/cpufreq_governor.c b/drivers/cpufreq/cpufreq_governor.c index 5af40ad82d23..dc9b72e25c1a 100644 --- a/drivers/cpufreq/cpufreq_governor.c +++ b/drivers/cpufreq/cpufreq_governor.c @@ -26,6 +26,7 @@ #include #include #include +#include #include "cpufreq_governor.h" @@ -180,8 +181,10 @@ void gov_queue_work(struct dbs_data *dbs_data, struct cpufreq_policy *policy, if (!all_cpus) { __gov_queue_work(smp_processor_id(), dbs_data, delay); } else { + get_online_cpus(); for_each_cpu(i, policy->cpus) __gov_queue_work(i, dbs_data, delay); + put_online_cpus(); } } EXPORT_SYMBOL_GPL(gov_queue_work); -- cgit v1.2.3 From 0ca684365547cd5f214b5739568dae3df5d6cec9 Mon Sep 17 00:00:00 2001 From: Guennadi Liakhovetski Date: Mon, 25 Feb 2013 18:22:37 +0100 Subject: cpufreq: cpufreq-cpu0: use the exact frequency for clk_set_rate() clk_set_rate() isn't supposed to accept approximate frequencies, instead a supported frequency should be obtained from clk_round_rate() and then used to set the clock. Signed-off-by: Guennadi Liakhovetski Acked-by: Shawn Guo Acked-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq-cpu0.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'drivers/cpufreq') diff --git a/drivers/cpufreq/cpufreq-cpu0.c b/drivers/cpufreq/cpufreq-cpu0.c index a64eb8b70444..ad1fde277661 100644 --- a/drivers/cpufreq/cpufreq-cpu0.c +++ b/drivers/cpufreq/cpufreq-cpu0.c @@ -45,7 +45,7 @@ static int cpu0_set_target(struct cpufreq_policy *policy, struct cpufreq_freqs freqs; struct opp *opp; unsigned long volt = 0, volt_old = 0, tol = 0; - long freq_Hz; + long freq_Hz, freq_exact; unsigned int index; int ret; @@ -60,6 +60,7 @@ static int cpu0_set_target(struct cpufreq_policy *policy, freq_Hz = clk_round_rate(cpu_clk, freq_table[index].frequency * 1000); if (freq_Hz < 0) freq_Hz = freq_table[index].frequency * 1000; + freq_exact = freq_Hz; freqs.new = freq_Hz / 1000; freqs.old = clk_get_rate(cpu_clk) / 1000; @@ -98,7 +99,7 @@ static int cpu0_set_target(struct cpufreq_policy *policy, } } - ret = clk_set_rate(cpu_clk, freqs.new * 1000); + ret = clk_set_rate(cpu_clk, freq_exact); if (ret) { pr_err("failed to set clock rate: %d\n", ret); if (cpu_reg) -- cgit v1.2.3