diff options
Diffstat (limited to 'kernel')
| -rw-r--r-- | kernel/cpu.c | 265 | ||||
| -rw-r--r-- | kernel/sched/core.c | 30 | ||||
| -rw-r--r-- | kernel/sched/fair.c | 1 |
3 files changed, 266 insertions, 30 deletions
diff --git a/kernel/cpu.c b/kernel/cpu.c index 0db8938fbb23..2bc3d2f5b2a5 100644 --- a/kernel/cpu.c +++ b/kernel/cpu.c @@ -60,6 +60,7 @@ struct cpuhp_cpu_state { bool rollback; bool single; bool bringup; + bool booted_once; struct hlist_node *node; struct hlist_node *last; enum cpuhp_state cb_state; @@ -342,6 +343,61 @@ void cpu_hotplug_enable(void) EXPORT_SYMBOL_GPL(cpu_hotplug_enable); #endif /* CONFIG_HOTPLUG_CPU */ +#ifdef CONFIG_HOTPLUG_SMT +enum cpuhp_smt_control cpu_smt_control __read_mostly = CPU_SMT_ENABLED; +EXPORT_SYMBOL_GPL(cpu_smt_control); + +void __init cpu_smt_disable(bool force) +{ + if (cpu_smt_control == CPU_SMT_FORCE_DISABLED || + cpu_smt_control == CPU_SMT_NOT_SUPPORTED) + return; + + if (force) { + pr_info("SMT: Force disabled\n"); + cpu_smt_control = CPU_SMT_FORCE_DISABLED; + } else { + cpu_smt_control = CPU_SMT_DISABLED; + } +} + +/* + * The decision whether SMT is supported can only be done after the full + * CPU identification. Called from architecture code. + */ +void __init cpu_smt_check_topology(void) +{ + if (!topology_smt_supported()) + cpu_smt_control = CPU_SMT_NOT_SUPPORTED; +} + +static int __init smt_cmdline_disable(char *str) +{ + cpu_smt_disable(str && !strcmp(str, "force")); + return 0; +} +early_param("nosmt", smt_cmdline_disable); + +static inline bool cpu_smt_allowed(unsigned int cpu) +{ + if (cpu_smt_control == CPU_SMT_ENABLED) + return true; + + if (topology_is_primary_thread(cpu)) + return true; + + /* + * On x86 it's required to boot all logical CPUs at least once so + * that the init code can get a chance to set CR4.MCE on each + * CPU. Otherwise, a broadacasted MCE observing CR4.MCE=0b on any + * core will shutdown the machine. + */ + return !per_cpu(cpuhp_state, cpu).booted_once; +} +#else +static inline bool cpu_smt_allowed(unsigned int cpu) { return true; } +#endif + static inline enum cpuhp_state cpuhp_set_state(struct cpuhp_cpu_state *st, enum cpuhp_state target) { @@ -422,6 +478,16 @@ static int bringup_wait_for_ap(unsigned int cpu) stop_machine_unpark(cpu); kthread_unpark(st->thread); + /* + * SMT soft disabling on X86 requires to bring the CPU out of the + * BIOS 'wait for SIPI' state in order to set the CR4.MCE bit. The + * CPU marked itself as booted_once in cpu_notify_starting() so the + * cpu_smt_allowed() check will now return false if this is not the + * primary sibling. + */ + if (!cpu_smt_allowed(cpu)) + return -ECANCELED; + if (st->target <= CPUHP_AP_ONLINE_IDLE) return 0; @@ -754,7 +820,6 @@ static int takedown_cpu(unsigned int cpu) /* Park the smpboot threads */ kthread_park(per_cpu_ptr(&cpuhp_state, cpu)->thread); - smpboot_park_threads(cpu); /* * Prevent irq alloc/free while the dying cpu reorganizes the @@ -907,20 +972,19 @@ out: return ret; } +static int cpu_down_maps_locked(unsigned int cpu, enum cpuhp_state target) +{ + if (cpu_hotplug_disabled) + return -EBUSY; + return _cpu_down(cpu, 0, target); +} + static int do_cpu_down(unsigned int cpu, enum cpuhp_state target) { int err; cpu_maps_update_begin(); - - if (cpu_hotplug_disabled) { - err = -EBUSY; - goto out; - } - - err = _cpu_down(cpu, 0, target); - -out: + err = cpu_down_maps_locked(cpu, target); cpu_maps_update_done(); return err; } @@ -949,6 +1013,7 @@ void notify_cpu_starting(unsigned int cpu) int ret; rcu_cpu_starting(cpu); /* Enables RCU usage on this CPU. */ + st->booted_once = true; while (st->state < target) { st->state++; ret = cpuhp_invoke_callback(cpu, st->state, true, NULL, NULL); @@ -1058,6 +1123,10 @@ static int do_cpu_up(unsigned int cpu, enum cpuhp_state target) err = -EBUSY; goto out; } + if (!cpu_smt_allowed(cpu)) { + err = -EPERM; + goto out; + } err = _cpu_up(cpu, 0, target); out: @@ -1332,7 +1401,7 @@ static struct cpuhp_step cpuhp_hp_states[] = { [CPUHP_AP_SMPBOOT_THREADS] = { .name = "smpboot/threads:online", .startup.single = smpboot_unpark_threads, - .teardown.single = NULL, + .teardown.single = smpboot_park_threads, }, [CPUHP_AP_IRQ_AFFINITY_ONLINE] = { .name = "irq/affinity:online", @@ -1906,10 +1975,181 @@ static const struct attribute_group cpuhp_cpu_root_attr_group = { NULL }; +#ifdef CONFIG_HOTPLUG_SMT + +static const char *smt_states[] = { + [CPU_SMT_ENABLED] = "on", + [CPU_SMT_DISABLED] = "off", + [CPU_SMT_FORCE_DISABLED] = "forceoff", + [CPU_SMT_NOT_SUPPORTED] = "notsupported", +}; + +static ssize_t +show_smt_control(struct device *dev, struct device_attribute *attr, char *buf) +{ + return snprintf(buf, PAGE_SIZE - 2, "%s\n", smt_states[cpu_smt_control]); +} + +static void cpuhp_offline_cpu_device(unsigned int cpu) +{ + struct device *dev = get_cpu_device(cpu); + + dev->offline = true; + /* Tell user space about the state change */ + kobject_uevent(&dev->kobj, KOBJ_OFFLINE); +} + +static void cpuhp_online_cpu_device(unsigned int cpu) +{ + struct device *dev = get_cpu_device(cpu); + + dev->offline = false; + /* Tell user space about the state change */ + kobject_uevent(&dev->kobj, KOBJ_ONLINE); +} + +static int cpuhp_smt_disable(enum cpuhp_smt_control ctrlval) +{ + int cpu, ret = 0; + + cpu_maps_update_begin(); + for_each_online_cpu(cpu) { + if (topology_is_primary_thread(cpu)) + continue; + ret = cpu_down_maps_locked(cpu, CPUHP_OFFLINE); + if (ret) + break; + /* + * As this needs to hold the cpu maps lock it's impossible + * to call device_offline() because that ends up calling + * cpu_down() which takes cpu maps lock. cpu maps lock + * needs to be held as this might race against in kernel + * abusers of the hotplug machinery (thermal management). + * + * So nothing would update device:offline state. That would + * leave the sysfs entry stale and prevent onlining after + * smt control has been changed to 'off' again. This is + * called under the sysfs hotplug lock, so it is properly + * serialized against the regular offline usage. + */ + cpuhp_offline_cpu_device(cpu); + } + if (!ret) + cpu_smt_control = ctrlval; + cpu_maps_update_done(); + return ret; +} + +static int cpuhp_smt_enable(void) +{ + int cpu, ret = 0; + + cpu_maps_update_begin(); + cpu_smt_control = CPU_SMT_ENABLED; + for_each_present_cpu(cpu) { + /* Skip online CPUs and CPUs on offline nodes */ + if (cpu_online(cpu) || !node_online(cpu_to_node(cpu))) + continue; + ret = _cpu_up(cpu, 0, CPUHP_ONLINE); + if (ret) + break; + /* See comment in cpuhp_smt_disable() */ + cpuhp_online_cpu_device(cpu); + } + cpu_maps_update_done(); + return ret; +} + +static ssize_t +store_smt_control(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + int ctrlval, ret; + + if (sysfs_streq(buf, "on")) + ctrlval = CPU_SMT_ENABLED; + else if (sysfs_streq(buf, "off")) + ctrlval = CPU_SMT_DISABLED; + else if (sysfs_streq(buf, "forceoff")) + ctrlval = CPU_SMT_FORCE_DISABLED; + else + return -EINVAL; + + if (cpu_smt_control == CPU_SMT_FORCE_DISABLED) + return -EPERM; + + if (cpu_smt_control == CPU_SMT_NOT_SUPPORTED) + return -ENODEV; + + ret = lock_device_hotplug_sysfs(); + if (ret) + return ret; + + if (ctrlval != cpu_smt_control) { + switch (ctrlval) { + case CPU_SMT_ENABLED: + ret = cpuhp_smt_enable(); + break; + case CPU_SMT_DISABLED: + case CPU_SMT_FORCE_DISABLED: + ret = cpuhp_smt_disable(ctrlval); + break; + } + } + + unlock_device_hotplug(); + return ret ? ret : count; +} +static DEVICE_ATTR(control, 0644, show_smt_control, store_smt_control); + +static ssize_t +show_smt_active(struct device *dev, struct device_attribute *attr, char *buf) +{ + bool active = topology_max_smt_threads() > 1; + + return snprintf(buf, PAGE_SIZE - 2, "%d\n", active); +} +static DEVICE_ATTR(active, 0444, show_smt_active, NULL); + +static struct attribute *cpuhp_smt_attrs[] = { + &dev_attr_control.attr, + &dev_attr_active.attr, + NULL +}; + +static const struct attribute_group cpuhp_smt_attr_group = { + .attrs = cpuhp_smt_attrs, + .name = "smt", + NULL +}; + +static int __init cpu_smt_state_init(void) +{ + /* + * If SMT was disabled by BIOS, detect it here, after the CPUs have + * been brought online. This ensures the smt/l1tf sysfs entries are + * consistent with reality. Note this may overwrite cpu_smt_control's + * previous setting. + */ + if (topology_max_smt_threads() == 1) + cpu_smt_control = CPU_SMT_NOT_SUPPORTED; + + return sysfs_create_group(&cpu_subsys.dev_root->kobj, + &cpuhp_smt_attr_group); +} + +#else +static inline int cpu_smt_state_init(void) { return 0; } +#endif + static int __init cpuhp_sysfs_init(void) { int cpu, ret; + ret = cpu_smt_state_init(); + if (ret) + return ret; + ret = sysfs_create_group(&cpu_subsys.dev_root->kobj, &cpuhp_cpu_root_attr_group); if (ret) @@ -2012,5 +2252,6 @@ void __init boot_cpu_init(void) */ void __init boot_cpu_state_init(void) { - per_cpu_ptr(&cpuhp_state, smp_processor_id())->state = CPUHP_ONLINE; + this_cpu_write(cpuhp_state.booted_once, true); + this_cpu_write(cpuhp_state.state, CPUHP_ONLINE); } diff --git a/kernel/sched/core.c b/kernel/sched/core.c index fe365c9a08e9..5ba96d9ddbde 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -5774,6 +5774,18 @@ int sched_cpu_activate(unsigned int cpu) struct rq *rq = cpu_rq(cpu); struct rq_flags rf; +#ifdef CONFIG_SCHED_SMT + /* + * The sched_smt_present static key needs to be evaluated on every + * hotplug event because at boot time SMT might be disabled when + * the number of booted CPUs is limited. + * + * If then later a sibling gets hotplugged, then the key would stay + * off and SMT scheduling would never be functional. + */ + if (cpumask_weight(cpu_smt_mask(cpu)) > 1) + static_branch_enable_cpuslocked(&sched_smt_present); +#endif set_cpu_active(cpu, true); if (sched_smp_initialized) { @@ -5871,22 +5883,6 @@ int sched_cpu_dying(unsigned int cpu) } #endif -#ifdef CONFIG_SCHED_SMT -DEFINE_STATIC_KEY_FALSE(sched_smt_present); - -static void sched_init_smt(void) -{ - /* - * We've enumerated all CPUs and will assume that if any CPU - * has SMT siblings, CPU0 will too. - */ - if (cpumask_weight(cpu_smt_mask(0)) > 1) - static_branch_enable(&sched_smt_present); -} -#else -static inline void sched_init_smt(void) { } -#endif - void __init sched_init_smp(void) { sched_init_numa(); @@ -5908,8 +5904,6 @@ void __init sched_init_smp(void) init_sched_rt_class(); init_sched_dl_class(); - sched_init_smt(); - sched_smp_initialized = true; } diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 2f0a0be4d344..9c219f7b0970 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -6237,6 +6237,7 @@ static inline int find_idlest_cpu(struct sched_domain *sd, struct task_struct *p } #ifdef CONFIG_SCHED_SMT +DEFINE_STATIC_KEY_FALSE(sched_smt_present); static inline void set_idle_cores(int cpu, int val) { |