From b563b4e3f2dd601e19b46ada31bd176fc0a16efc Mon Sep 17 00:00:00 2001 From: Dirk Brandewie Date: Fri, 22 Mar 2013 01:29:28 +0100 Subject: [PATCH 1/5] cpufreq / intel_pstate: Add function to check that all MSRs are valid Some VMs seem to try to implement some MSRs but not all the registers the driver needs. Check to make sure all the MSR that we need are available. If any of the required MSRs are not available refuse to load. References: https://bugzilla.redhat.com/show_bug.cgi?id=922923 Reported-by: Josh Stone Signed-off-by: Dirk Brandewie Acked-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 26 ++++++++++++++++++++++++++ 1 file changed, 26 insertions(+) diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index f6dd1e761129..cd9c5f4f5805 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -752,6 +752,29 @@ static struct cpufreq_driver intel_pstate_driver = { static int __initdata no_load; +static int intel_pstate_msrs_not_valid(void) +{ + /* Check that all the msr's we are using are valid. */ + u64 aperf, mperf, tmp; + + rdmsrl(MSR_IA32_APERF, aperf); + rdmsrl(MSR_IA32_MPERF, mperf); + + if (!intel_pstate_min_pstate() || + !intel_pstate_max_pstate() || + !intel_pstate_turbo_pstate()) + return -ENODEV; + + rdmsrl(MSR_IA32_APERF, tmp); + if (!(tmp - aperf)) + return -ENODEV; + + rdmsrl(MSR_IA32_MPERF, tmp); + if (!(tmp - mperf)) + return -ENODEV; + + return 0; +} static int __init intel_pstate_init(void) { int cpu, rc = 0; @@ -764,6 +787,9 @@ static int __init intel_pstate_init(void) if (!id) return -ENODEV; + if (intel_pstate_msrs_not_valid()) + return -ENODEV; + pr_info("Intel P-state driver initializing.\n"); all_cpu_data = vmalloc(sizeof(void *) * num_possible_cpus()); From e6f3eb29be471c4b536a91daab76d4aeda72a261 Mon Sep 17 00:00:00 2001 From: Dirk Brandewie Date: Sun, 24 Mar 2013 00:54:39 +0100 Subject: [PATCH 2/5] cpufreq / intel_pstate: Fix calculation of current frequency Use the correct pstate value to calculate the effective frequency. References: https://bugzilla.redhat.com/show_bug.cgi?id=923942 Reported-by: Satish Balay Signed-off-by: Dirk Brandewie Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/intel_pstate.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index cd9c5f4f5805..b662529072f7 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -454,7 +454,7 @@ static inline void intel_pstate_calc_busy(struct cpudata *cpu, sample->idletime_us * 100, sample->duration_us); core_pct = div64_u64(sample->aperf * 100, sample->mperf); - sample->freq = cpu->pstate.turbo_pstate * core_pct * 1000; + sample->freq = cpu->pstate.max_pstate * core_pct * 1000; sample->core_pct_busy = div_s64((sample->pstate_pct_busy * core_pct), 100); From 05e99c8cf9d4e53ef6e016815db40a89a6156529 Mon Sep 17 00:00:00 2001 From: Konrad Rzeszutek Wilk Date: Wed, 20 Mar 2013 14:21:10 +0000 Subject: [PATCH 3/5] intel-pstate: Use #defines instead of hard-coded values. They are defined in coreboot (MSR_PLATFORM) and the other one is already defined in msr-index.h. Let's use those. Signed-off-by: Konrad Rzeszutek Wilk Acked-by: Viresh Kumar Acked-by: Dirk Brandewie Signed-off-by: Rafael J. Wysocki --- arch/x86/include/uapi/asm/msr-index.h | 1 + drivers/cpufreq/intel_pstate.c | 6 +++--- 2 files changed, 4 insertions(+), 3 deletions(-) diff --git a/arch/x86/include/uapi/asm/msr-index.h b/arch/x86/include/uapi/asm/msr-index.h index 892ce40a7470..7a060f4b411f 100644 --- a/arch/x86/include/uapi/asm/msr-index.h +++ b/arch/x86/include/uapi/asm/msr-index.h @@ -44,6 +44,7 @@ #define SNB_C1_AUTO_UNDEMOTE (1UL << 27) #define SNB_C3_AUTO_UNDEMOTE (1UL << 28) +#define MSR_PLATFORM_INFO 0x000000ce #define MSR_MTRRcap 0x000000fe #define MSR_IA32_BBL_CR_CTL 0x00000119 #define MSR_IA32_BBL_CR_CTL3 0x0000011e diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c index b662529072f7..ad72922919ed 100644 --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -358,14 +358,14 @@ static void intel_pstate_sysfs_expose_params(void) static int intel_pstate_min_pstate(void) { u64 value; - rdmsrl(0xCE, value); + rdmsrl(MSR_PLATFORM_INFO, value); return (value >> 40) & 0xFF; } static int intel_pstate_max_pstate(void) { u64 value; - rdmsrl(0xCE, value); + rdmsrl(MSR_PLATFORM_INFO, value); return (value >> 8) & 0xFF; } @@ -373,7 +373,7 @@ static int intel_pstate_turbo_pstate(void) { u64 value; int nont, ret; - rdmsrl(0x1AD, value); + rdmsrl(MSR_NHM_TURBO_RATIO_LIMIT, value); nont = intel_pstate_max_pstate(); ret = ((value) & 255); if (ret <= nont) From 187da1d97f3a949b967274d7ee2f95d3a4f39251 Mon Sep 17 00:00:00 2001 From: viresh kumar Date: Fri, 22 Mar 2013 10:13:52 +0000 Subject: [PATCH 4/5] cpufreq: stats: do cpufreq_cpu_put() corresponding to cpufreq_cpu_get() In cpufreq_stats_free_sysfs() we aren't balancing calls to cpufreq_cpu_get() with cpufreq_cpu_put(). This will never let us have ref count to policy->kobj as zero. We will get a hang if somehow cpufreq_driver_unregister() is called. And that can happen when we compile our driver as module and insmod/rmmod it. Signed-off-by: Viresh Kumar Acked-by: Amit Kucheria Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/cpufreq_stats.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) diff --git a/drivers/cpufreq/cpufreq_stats.c b/drivers/cpufreq/cpufreq_stats.c index 2fd779eb1ed1..bfd6273fd873 100644 --- a/drivers/cpufreq/cpufreq_stats.c +++ b/drivers/cpufreq/cpufreq_stats.c @@ -180,15 +180,19 @@ static void cpufreq_stats_free_sysfs(unsigned int cpu) { struct cpufreq_policy *policy = cpufreq_cpu_get(cpu); - if (!cpufreq_frequency_get_table(cpu)) + if (!policy) return; - if (policy && !policy_is_shared(policy)) { + if (!cpufreq_frequency_get_table(cpu)) + goto put_ref; + + if (!policy_is_shared(policy)) { pr_debug("%s: Free sysfs stat\n", __func__); sysfs_remove_group(&policy->kobj, &stats_attr_group); } - if (policy) - cpufreq_cpu_put(policy); + +put_ref: + cpufreq_cpu_put(policy); } static int cpufreq_stats_create_table(struct cpufreq_policy *policy, From aa77a52764a92216b61a6c8079b5c01937c046cd Mon Sep 17 00:00:00 2001 From: Viresh Kumar Date: Sun, 24 Mar 2013 15:58:12 +0000 Subject: [PATCH 5/5] cpufreq: acpi-cpufreq: Don't set policy->related_cpus from .init() With the addition of following patch: fcf8058 cpufreq: Simplify cpufreq_add_dev() cpufreq driver's .init() routine must initialize policy->cpus with mask of all possible CPUs (Online + Offline) that share the clock. Then the core would copy this mask onto policy->related_cpus and will reset policy->cpus to carry only online cpus. acpi-cpufreq driver wasn't updated with this assumption and so sometimes when we try to hot[un]plug CPUs at run time, sysfs directories get corrupted. This patch fixes acpi-cpufreq driver against this corruption. Reported-and-tested-by: Maciej Rutecki Tested-by: Borislav Petkov Signed-off-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki --- drivers/cpufreq/acpi-cpufreq.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/drivers/cpufreq/acpi-cpufreq.c b/drivers/cpufreq/acpi-cpufreq.c index 937bc286591f..57a8774f0b4e 100644 --- a/drivers/cpufreq/acpi-cpufreq.c +++ b/drivers/cpufreq/acpi-cpufreq.c @@ -730,7 +730,6 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy) policy->shared_type == CPUFREQ_SHARED_TYPE_ANY) { cpumask_copy(policy->cpus, perf->shared_cpu_map); } - cpumask_copy(policy->related_cpus, perf->shared_cpu_map); #ifdef CONFIG_SMP dmi_check_system(sw_any_bug_dmi_table); @@ -742,7 +741,6 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy) if (check_amd_hwpstate_cpu(cpu) && !acpi_pstate_strict) { cpumask_clear(policy->cpus); cpumask_set_cpu(cpu, policy->cpus); - cpumask_copy(policy->related_cpus, cpu_sibling_mask(cpu)); policy->shared_type = CPUFREQ_SHARED_TYPE_HW; pr_info_once(PFX "overriding BIOS provided _PSD data\n"); }