diff options
Diffstat (limited to 'drivers/base')
| -rw-r--r-- | drivers/base/arch_topology.c | 2 | ||||
| -rw-r--r-- | drivers/base/cpu.c | 6 | ||||
| -rw-r--r-- | drivers/base/devcoredump.c | 138 | ||||
| -rw-r--r-- | drivers/base/node.c | 4 | ||||
| -rw-r--r-- | drivers/base/power/main.c | 14 | ||||
| -rw-r--r-- | drivers/base/power/runtime.c | 44 | ||||
| -rw-r--r-- | drivers/base/regmap/regmap.c | 2 |
7 files changed, 153 insertions, 57 deletions
diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c index 31bd6f4e5dc4..69dea9825f6a 100644 --- a/drivers/base/arch_topology.c +++ b/drivers/base/arch_topology.c @@ -290,7 +290,7 @@ bool __init topology_parse_cpu_capacity(struct device_node *cpu_node, int cpu) * frequency (by keeping the initial freq_factor value). */ cpu_clk = of_clk_get(cpu_node, 0); - if (!PTR_ERR_OR_ZERO(cpu_clk)) { + if (!IS_ERR_OR_NULL(cpu_clk)) { per_cpu(freq_factor, cpu) = clk_get_rate(cpu_clk) / 1000; clk_put(cpu_clk); diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c index 0e7f7f54665d..64d270ba759f 100644 --- a/drivers/base/cpu.c +++ b/drivers/base/cpu.c @@ -605,6 +605,10 @@ ssize_t __weak cpu_show_tsa(struct device *dev, struct device_attribute *attr, c { return sysfs_emit(buf, "Not affected\n"); } +ssize_t __weak cpu_show_vmscape(struct device *dev, struct device_attribute *attr, char *buf) +{ + return sysfs_emit(buf, "Not affected\n"); +} static DEVICE_ATTR(meltdown, 0444, cpu_show_meltdown, NULL); static DEVICE_ATTR(spectre_v1, 0444, cpu_show_spectre_v1, NULL); @@ -622,6 +626,7 @@ static DEVICE_ATTR(spec_rstack_overflow, 0444, cpu_show_spec_rstack_overflow, NU static DEVICE_ATTR(reg_file_data_sampling, 0444, cpu_show_reg_file_data_sampling, NULL); static DEVICE_ATTR(indirect_target_selection, 0444, cpu_show_indirect_target_selection, NULL); static DEVICE_ATTR(tsa, 0444, cpu_show_tsa, NULL); +static DEVICE_ATTR(vmscape, 0444, cpu_show_vmscape, NULL); static struct attribute *cpu_root_vulnerabilities_attrs[] = { &dev_attr_meltdown.attr, @@ -640,6 +645,7 @@ static struct attribute *cpu_root_vulnerabilities_attrs[] = { &dev_attr_reg_file_data_sampling.attr, &dev_attr_indirect_target_selection.attr, &dev_attr_tsa.attr, + &dev_attr_vmscape.attr, NULL }; diff --git a/drivers/base/devcoredump.c b/drivers/base/devcoredump.c index f3bd9f104bd1..48754841fd86 100644 --- a/drivers/base/devcoredump.c +++ b/drivers/base/devcoredump.c @@ -26,50 +26,46 @@ struct devcd_entry { void *data; size_t datalen; /* - * Here, mutex is required to serialize the calls to del_wk work between - * user/kernel space which happens when devcd is added with device_add() - * and that sends uevent to user space. User space reads the uevents, - * and calls to devcd_data_write() which try to modify the work which is - * not even initialized/queued from devcoredump. + * There are 2 races for which mutex is required. * + * The first race is between device creation and userspace writing to + * schedule immediately destruction. * + * This race is handled by arming the timer before device creation, but + * when device creation fails the timer still exists. * - * cpu0(X) cpu1(Y) + * To solve this, hold the mutex during device_add(), and set + * init_completed on success before releasing the mutex. * - * dev_coredump() uevent sent to user space - * device_add() ======================> user space process Y reads the - * uevents writes to devcd fd - * which results into writes to + * That way the timer will never fire until device_add() is called, + * it will do nothing if init_completed is not set. The timer is also + * cancelled in that case. * - * devcd_data_write() - * mod_delayed_work() - * try_to_grab_pending() - * del_timer() - * debug_assert_init() - * INIT_DELAYED_WORK() - * schedule_delayed_work() - * - * - * Also, mutex alone would not be enough to avoid scheduling of - * del_wk work after it get flush from a call to devcd_free() - * mentioned as below. - * - * disabled_store() - * devcd_free() - * mutex_lock() devcd_data_write() - * flush_delayed_work() - * mutex_unlock() - * mutex_lock() - * mod_delayed_work() - * mutex_unlock() - * So, delete_work flag is required. + * The second race involves multiple parallel invocations of devcd_free(), + * add a deleted flag so only 1 can call the destructor. */ struct mutex mutex; - bool delete_work; + bool init_completed, deleted; struct module *owner; ssize_t (*read)(char *buffer, loff_t offset, size_t count, void *data, size_t datalen); void (*free)(void *data); + /* + * If nothing interferes and device_add() was returns success, + * del_wk will destroy the device after the timer fires. + * + * Multiple userspace processes can interfere in the working of the timer: + * - Writing to the coredump will reschedule the timer to run immediately, + * if still armed. + * + * This is handled by using "if (cancel_delayed_work()) { + * schedule_delayed_work() }", to prevent re-arming after having + * been previously fired. + * - Writing to /sys/class/devcoredump/disabled will destroy the + * coredump synchronously. + * This is handled by using disable_delayed_work_sync(), and then + * checking if deleted flag is set with &devcd->mutex held. + */ struct delayed_work del_wk; struct device *failing_dev; }; @@ -98,14 +94,27 @@ static void devcd_dev_release(struct device *dev) kfree(devcd); } +static void __devcd_del(struct devcd_entry *devcd) +{ + devcd->deleted = true; + device_del(&devcd->devcd_dev); + put_device(&devcd->devcd_dev); +} + static void devcd_del(struct work_struct *wk) { struct devcd_entry *devcd; + bool init_completed; devcd = container_of(wk, struct devcd_entry, del_wk.work); - device_del(&devcd->devcd_dev); - put_device(&devcd->devcd_dev); + /* devcd->mutex serializes against dev_coredumpm_timeout */ + mutex_lock(&devcd->mutex); + init_completed = devcd->init_completed; + mutex_unlock(&devcd->mutex); + + if (init_completed) + __devcd_del(devcd); } static ssize_t devcd_data_read(struct file *filp, struct kobject *kobj, @@ -125,12 +134,12 @@ static ssize_t devcd_data_write(struct file *filp, struct kobject *kobj, struct device *dev = kobj_to_dev(kobj); struct devcd_entry *devcd = dev_to_devcd(dev); - mutex_lock(&devcd->mutex); - if (!devcd->delete_work) { - devcd->delete_work = true; - mod_delayed_work(system_wq, &devcd->del_wk, 0); - } - mutex_unlock(&devcd->mutex); + /* + * Although it's tempting to use mod_delayed work here, + * that will cause a reschedule if the timer already fired. + */ + if (cancel_delayed_work(&devcd->del_wk)) + schedule_delayed_work(&devcd->del_wk, 0); return count; } @@ -158,11 +167,21 @@ static int devcd_free(struct device *dev, void *data) { struct devcd_entry *devcd = dev_to_devcd(dev); + /* + * To prevent a race with devcd_data_write(), cancel work and + * complete manually instead. + * + * We cannot rely on the return value of + * cancel_delayed_work_sync() here, because it might be in the + * middle of a cancel_delayed_work + schedule_delayed_work pair. + * + * devcd->mutex here guards against multiple parallel invocations + * of devcd_free(). + */ + cancel_delayed_work_sync(&devcd->del_wk); mutex_lock(&devcd->mutex); - if (!devcd->delete_work) - devcd->delete_work = true; - - flush_delayed_work(&devcd->del_wk); + if (!devcd->deleted) + __devcd_del(devcd); mutex_unlock(&devcd->mutex); return 0; } @@ -186,12 +205,10 @@ static ssize_t disabled_show(struct class *class, struct class_attribute *attr, * put_device() <- last reference * error = fn(dev, data) devcd_dev_release() * devcd_free(dev, data) kfree(devcd) - * mutex_lock(&devcd->mutex); * * - * In the above diagram, It looks like disabled_store() would be racing with parallely - * running devcd_del() and result in memory abort while acquiring devcd->mutex which - * is called after kfree of devcd memory after dropping its last reference with + * In the above diagram, it looks like disabled_store() would be racing with parallelly + * running devcd_del() and result in memory abort after dropping its last reference with * put_device(). However, this will not happens as fn(dev, data) runs * with its own reference to device via klist_node so it is not its last reference. * so, above situation would not occur. @@ -353,7 +370,7 @@ void dev_coredumpm(struct device *dev, struct module *owner, devcd->read = read; devcd->free = free; devcd->failing_dev = get_device(dev); - devcd->delete_work = false; + devcd->deleted = false; mutex_init(&devcd->mutex); device_initialize(&devcd->devcd_dev); @@ -362,8 +379,14 @@ void dev_coredumpm(struct device *dev, struct module *owner, atomic_inc_return(&devcd_count)); devcd->devcd_dev.class = &devcd_class; - mutex_lock(&devcd->mutex); dev_set_uevent_suppress(&devcd->devcd_dev, true); + + /* devcd->mutex prevents devcd_del() completing until init finishes */ + mutex_lock(&devcd->mutex); + devcd->init_completed = false; + INIT_DELAYED_WORK(&devcd->del_wk, devcd_del); + schedule_delayed_work(&devcd->del_wk, DEVCD_TIMEOUT); + if (device_add(&devcd->devcd_dev)) goto put_device; @@ -380,13 +403,20 @@ void dev_coredumpm(struct device *dev, struct module *owner, dev_set_uevent_suppress(&devcd->devcd_dev, false); kobject_uevent(&devcd->devcd_dev.kobj, KOBJ_ADD); - INIT_DELAYED_WORK(&devcd->del_wk, devcd_del); - schedule_delayed_work(&devcd->del_wk, DEVCD_TIMEOUT); + + /* + * Safe to run devcd_del() now that we are done with devcd_dev. + * Alternatively we could have taken a ref on devcd_dev before + * dropping the lock. + */ + devcd->init_completed = true; mutex_unlock(&devcd->mutex); return; put_device: - put_device(&devcd->devcd_dev); mutex_unlock(&devcd->mutex); + cancel_delayed_work_sync(&devcd->del_wk); + put_device(&devcd->devcd_dev); + put_module: module_put(owner); free: diff --git a/drivers/base/node.c b/drivers/base/node.c index 5366d1b5359c..dd8c8fdfd158 100644 --- a/drivers/base/node.c +++ b/drivers/base/node.c @@ -983,6 +983,10 @@ int __register_one_node(int nid) return -ENOMEM; error = register_node(node_devices[nid], nid); + if (error) { + node_devices[nid] = NULL; + return error; + } /* link cpu under this node */ for_each_present_cpu(cpu) { diff --git a/drivers/base/power/main.c b/drivers/base/power/main.c index c784de10b494..9cd0a837af42 100644 --- a/drivers/base/power/main.c +++ b/drivers/base/power/main.c @@ -601,8 +601,20 @@ static void __device_resume_noirq(struct device *dev, pm_message_t state, bool a if (dev->power.syscore || dev->power.direct_complete) goto Out; - if (!dev->power.is_noirq_suspended) + if (!dev->power.is_noirq_suspended) { + /* + * This means that system suspend has been aborted in the noirq + * phase before invoking the noirq suspend callback for the + * device, so if device_suspend_late() has left it in suspend, + * device_resume_early() should leave it in suspend either in + * case the early resume of it depends on the noirq resume that + * has not run. + */ + if (dev_pm_skip_suspend(dev)) + dev->power.must_resume = false; + goto Out; + } if (!dpm_wait_for_superior(dev, async)) goto Out; diff --git a/drivers/base/power/runtime.c b/drivers/base/power/runtime.c index 26ea7f5c8d42..f048840ae284 100644 --- a/drivers/base/power/runtime.c +++ b/drivers/base/power/runtime.c @@ -1477,6 +1477,32 @@ void pm_runtime_enable(struct device *dev) } EXPORT_SYMBOL_GPL(pm_runtime_enable); +static void pm_runtime_set_suspended_action(void *data) +{ + pm_runtime_set_suspended(data); +} + +/** + * devm_pm_runtime_set_active_enabled - set_active version of devm_pm_runtime_enable. + * + * @dev: Device to handle. + */ +int devm_pm_runtime_set_active_enabled(struct device *dev) +{ + int err; + + err = pm_runtime_set_active(dev); + if (err) + return err; + + err = devm_add_action_or_reset(dev, pm_runtime_set_suspended_action, dev); + if (err) + return err; + + return devm_pm_runtime_enable(dev); +} +EXPORT_SYMBOL_GPL(devm_pm_runtime_set_active_enabled); + static void pm_runtime_disable_action(void *data) { pm_runtime_dont_use_autosuspend(data); @@ -1499,6 +1525,24 @@ int devm_pm_runtime_enable(struct device *dev) } EXPORT_SYMBOL_GPL(devm_pm_runtime_enable); +static void pm_runtime_put_noidle_action(void *data) +{ + pm_runtime_put_noidle(data); +} + +/** + * devm_pm_runtime_get_noresume - devres-enabled version of pm_runtime_get_noresume. + * + * @dev: Device to handle. + */ +int devm_pm_runtime_get_noresume(struct device *dev) +{ + pm_runtime_get_noresume(dev); + + return devm_add_action_or_reset(dev, pm_runtime_put_noidle_action, dev); +} +EXPORT_SYMBOL_GPL(devm_pm_runtime_get_noresume); + /** * pm_runtime_forbid - Block runtime PM of a device. * @dev: Device to handle. diff --git a/drivers/base/regmap/regmap.c b/drivers/base/regmap/regmap.c index 2843a5742946..a0ed8326aa2d 100644 --- a/drivers/base/regmap/regmap.c +++ b/drivers/base/regmap/regmap.c @@ -888,7 +888,7 @@ struct regmap *__regmap_init(struct device *dev, map->read_flag_mask = bus->read_flag_mask; } - if (config && config->read && config->write) { + if (config->read && config->write) { map->reg_read = _regmap_bus_read; /* Bulk read/write */ |