From 70d932985757fbe978024db313001218e9f8fe5c Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 18 Aug 2020 15:57:36 +0200 Subject: notifier: Fix broken error handling pattern The current notifiers have the following error handling pattern all over the place: int err, nr; err = __foo_notifier_call_chain(&chain, val_up, v, -1, &nr); if (err & NOTIFIER_STOP_MASK) __foo_notifier_call_chain(&chain, val_down, v, nr-1, NULL) And aside from the endless repetition thereof, it is broken. Consider blocking notifiers; both calls take and drop the rwsem, this means that the notifier list can change in between the two calls, making @nr meaningless. Fix this by replacing all the __foo_notifier_call_chain() functions with foo_notifier_call_chain_robust() that embeds the above pattern, but ensures it is inside a single lock region. Note: I switched atomic_notifier_call_chain_robust() to use the spinlock, since RCU cannot provide the guarantee required for the recovery. Note: software_resume() error handling was broken afaict. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Acked-by: Rafael J. Wysocki Link: https://lore.kernel.org/r/20200818135804.325626653@infradead.org --- kernel/cpu_pm.c | 48 ++++++---------- kernel/notifier.c | 144 +++++++++++++++++++++++++++++------------------ kernel/power/hibernate.c | 39 ++++++------- kernel/power/main.c | 8 +-- kernel/power/power.h | 3 +- kernel/power/suspend.c | 14 ++--- kernel/power/user.c | 14 ++--- 7 files changed, 139 insertions(+), 131 deletions(-) (limited to 'kernel') diff --git a/kernel/cpu_pm.c b/kernel/cpu_pm.c index 44a259338e33..f7e1d0eccdbc 100644 --- a/kernel/cpu_pm.c +++ b/kernel/cpu_pm.c @@ -15,18 +15,28 @@ static ATOMIC_NOTIFIER_HEAD(cpu_pm_notifier_chain); -static int cpu_pm_notify(enum cpu_pm_event event, int nr_to_call, int *nr_calls) +static int cpu_pm_notify(enum cpu_pm_event event) { int ret; /* - * __atomic_notifier_call_chain has a RCU read critical section, which + * atomic_notifier_call_chain has a RCU read critical section, which * could be disfunctional in cpu idle. Copy RCU_NONIDLE code to let * RCU know this. */ rcu_irq_enter_irqson(); - ret = __atomic_notifier_call_chain(&cpu_pm_notifier_chain, event, NULL, - nr_to_call, nr_calls); + ret = atomic_notifier_call_chain(&cpu_pm_notifier_chain, event, NULL); + rcu_irq_exit_irqson(); + + return notifier_to_errno(ret); +} + +static int cpu_pm_notify_robust(enum cpu_pm_event event_up, enum cpu_pm_event event_down) +{ + int ret; + + rcu_irq_enter_irqson(); + ret = atomic_notifier_call_chain_robust(&cpu_pm_notifier_chain, event_up, event_down, NULL); rcu_irq_exit_irqson(); return notifier_to_errno(ret); @@ -80,18 +90,7 @@ EXPORT_SYMBOL_GPL(cpu_pm_unregister_notifier); */ int cpu_pm_enter(void) { - int nr_calls = 0; - int ret = 0; - - ret = cpu_pm_notify(CPU_PM_ENTER, -1, &nr_calls); - if (ret) - /* - * Inform listeners (nr_calls - 1) about failure of CPU PM - * PM entry who are notified earlier to prepare for it. - */ - cpu_pm_notify(CPU_PM_ENTER_FAILED, nr_calls - 1, NULL); - - return ret; + return cpu_pm_notify_robust(CPU_PM_ENTER, CPU_PM_ENTER_FAILED); } EXPORT_SYMBOL_GPL(cpu_pm_enter); @@ -109,7 +108,7 @@ EXPORT_SYMBOL_GPL(cpu_pm_enter); */ int cpu_pm_exit(void) { - return cpu_pm_notify(CPU_PM_EXIT, -1, NULL); + return cpu_pm_notify(CPU_PM_EXIT); } EXPORT_SYMBOL_GPL(cpu_pm_exit); @@ -131,18 +130,7 @@ EXPORT_SYMBOL_GPL(cpu_pm_exit); */ int cpu_cluster_pm_enter(void) { - int nr_calls = 0; - int ret = 0; - - ret = cpu_pm_notify(CPU_CLUSTER_PM_ENTER, -1, &nr_calls); - if (ret) - /* - * Inform listeners (nr_calls - 1) about failure of CPU cluster - * PM entry who are notified earlier to prepare for it. - */ - cpu_pm_notify(CPU_CLUSTER_PM_ENTER_FAILED, nr_calls - 1, NULL); - - return ret; + return cpu_pm_notify_robust(CPU_CLUSTER_PM_ENTER, CPU_CLUSTER_PM_ENTER_FAILED); } EXPORT_SYMBOL_GPL(cpu_cluster_pm_enter); @@ -163,7 +151,7 @@ EXPORT_SYMBOL_GPL(cpu_cluster_pm_enter); */ int cpu_cluster_pm_exit(void) { - return cpu_pm_notify(CPU_CLUSTER_PM_EXIT, -1, NULL); + return cpu_pm_notify(CPU_CLUSTER_PM_EXIT); } EXPORT_SYMBOL_GPL(cpu_cluster_pm_exit); diff --git a/kernel/notifier.c b/kernel/notifier.c index 84c987dfbe03..1b019cbca594 100644 --- a/kernel/notifier.c +++ b/kernel/notifier.c @@ -94,6 +94,34 @@ static int notifier_call_chain(struct notifier_block **nl, } NOKPROBE_SYMBOL(notifier_call_chain); +/** + * notifier_call_chain_robust - Inform the registered notifiers about an event + * and rollback on error. + * @nl: Pointer to head of the blocking notifier chain + * @val_up: Value passed unmodified to the notifier function + * @val_down: Value passed unmodified to the notifier function when recovering + * from an error on @val_up + * @v Pointer passed unmodified to the notifier function + * + * NOTE: It is important the @nl chain doesn't change between the two + * invocations of notifier_call_chain() such that we visit the + * exact same notifier callbacks; this rules out any RCU usage. + * + * Returns: the return value of the @val_up call. + */ +static int notifier_call_chain_robust(struct notifier_block **nl, + unsigned long val_up, unsigned long val_down, + void *v) +{ + int ret, nr = 0; + + ret = notifier_call_chain(nl, val_up, v, -1, &nr); + if (ret & NOTIFY_STOP_MASK) + notifier_call_chain(nl, val_down, v, nr-1, NULL); + + return ret; +} + /* * Atomic notifier chain routines. Registration and unregistration * use a spinlock, and call_chain is synchronized by RCU (no locks). @@ -144,13 +172,30 @@ int atomic_notifier_chain_unregister(struct atomic_notifier_head *nh, } EXPORT_SYMBOL_GPL(atomic_notifier_chain_unregister); +int atomic_notifier_call_chain_robust(struct atomic_notifier_head *nh, + unsigned long val_up, unsigned long val_down, void *v) +{ + unsigned long flags; + int ret; + + /* + * Musn't use RCU; because then the notifier list can + * change between the up and down traversal. + */ + spin_lock_irqsave(&nh->lock, flags); + ret = notifier_call_chain_robust(&nh->head, val_up, val_down, v); + spin_unlock_irqrestore(&nh->lock, flags); + + return ret; +} +EXPORT_SYMBOL_GPL(atomic_notifier_call_chain_robust); +NOKPROBE_SYMBOL(atomic_notifier_call_chain_robust); + /** - * __atomic_notifier_call_chain - Call functions in an atomic notifier chain + * atomic_notifier_call_chain - Call functions in an atomic notifier chain * @nh: Pointer to head of the atomic notifier chain * @val: Value passed unmodified to notifier function * @v: Pointer passed unmodified to notifier function - * @nr_to_call: See the comment for notifier_call_chain. - * @nr_calls: See the comment for notifier_call_chain. * * Calls each function in a notifier chain in turn. The functions * run in an atomic context, so they must not block. @@ -163,24 +208,16 @@ EXPORT_SYMBOL_GPL(atomic_notifier_chain_unregister); * Otherwise the return value is the return value * of the last notifier function called. */ -int __atomic_notifier_call_chain(struct atomic_notifier_head *nh, - unsigned long val, void *v, - int nr_to_call, int *nr_calls) +int atomic_notifier_call_chain(struct atomic_notifier_head *nh, + unsigned long val, void *v) { int ret; rcu_read_lock(); - ret = notifier_call_chain(&nh->head, val, v, nr_to_call, nr_calls); + ret = notifier_call_chain(&nh->head, val, v, -1, NULL); rcu_read_unlock(); - return ret; -} -EXPORT_SYMBOL_GPL(__atomic_notifier_call_chain); -NOKPROBE_SYMBOL(__atomic_notifier_call_chain); -int atomic_notifier_call_chain(struct atomic_notifier_head *nh, - unsigned long val, void *v) -{ - return __atomic_notifier_call_chain(nh, val, v, -1, NULL); + return ret; } EXPORT_SYMBOL_GPL(atomic_notifier_call_chain); NOKPROBE_SYMBOL(atomic_notifier_call_chain); @@ -250,13 +287,30 @@ int blocking_notifier_chain_unregister(struct blocking_notifier_head *nh, } EXPORT_SYMBOL_GPL(blocking_notifier_chain_unregister); +int blocking_notifier_call_chain_robust(struct blocking_notifier_head *nh, + unsigned long val_up, unsigned long val_down, void *v) +{ + int ret = NOTIFY_DONE; + + /* + * We check the head outside the lock, but if this access is + * racy then it does not matter what the result of the test + * is, we re-check the list after having taken the lock anyway: + */ + if (rcu_access_pointer(nh->head)) { + down_read(&nh->rwsem); + ret = notifier_call_chain_robust(&nh->head, val_up, val_down, v); + up_read(&nh->rwsem); + } + return ret; +} +EXPORT_SYMBOL_GPL(blocking_notifier_call_chain_robust); + /** - * __blocking_notifier_call_chain - Call functions in a blocking notifier chain + * blocking_notifier_call_chain - Call functions in a blocking notifier chain * @nh: Pointer to head of the blocking notifier chain * @val: Value passed unmodified to notifier function * @v: Pointer passed unmodified to notifier function - * @nr_to_call: See comment for notifier_call_chain. - * @nr_calls: See comment for notifier_call_chain. * * Calls each function in a notifier chain in turn. The functions * run in a process context, so they are allowed to block. @@ -268,9 +322,8 @@ EXPORT_SYMBOL_GPL(blocking_notifier_chain_unregister); * Otherwise the return value is the return value * of the last notifier function called. */ -int __blocking_notifier_call_chain(struct blocking_notifier_head *nh, - unsigned long val, void *v, - int nr_to_call, int *nr_calls) +int blocking_notifier_call_chain(struct blocking_notifier_head *nh, + unsigned long val, void *v) { int ret = NOTIFY_DONE; @@ -281,19 +334,11 @@ int __blocking_notifier_call_chain(struct blocking_notifier_head *nh, */ if (rcu_access_pointer(nh->head)) { down_read(&nh->rwsem); - ret = notifier_call_chain(&nh->head, val, v, nr_to_call, - nr_calls); + ret = notifier_call_chain(&nh->head, val, v, -1, NULL); up_read(&nh->rwsem); } return ret; } -EXPORT_SYMBOL_GPL(__blocking_notifier_call_chain); - -int blocking_notifier_call_chain(struct blocking_notifier_head *nh, - unsigned long val, void *v) -{ - return __blocking_notifier_call_chain(nh, val, v, -1, NULL); -} EXPORT_SYMBOL_GPL(blocking_notifier_call_chain); /* @@ -335,13 +380,18 @@ int raw_notifier_chain_unregister(struct raw_notifier_head *nh, } EXPORT_SYMBOL_GPL(raw_notifier_chain_unregister); +int raw_notifier_call_chain_robust(struct raw_notifier_head *nh, + unsigned long val_up, unsigned long val_down, void *v) +{ + return notifier_call_chain_robust(&nh->head, val_up, val_down, v); +} +EXPORT_SYMBOL_GPL(raw_notifier_call_chain_robust); + /** - * __raw_notifier_call_chain - Call functions in a raw notifier chain + * raw_notifier_call_chain - Call functions in a raw notifier chain * @nh: Pointer to head of the raw notifier chain * @val: Value passed unmodified to notifier function * @v: Pointer passed unmodified to notifier function - * @nr_to_call: See comment for notifier_call_chain. - * @nr_calls: See comment for notifier_call_chain * * Calls each function in a notifier chain in turn. The functions * run in an undefined context. @@ -354,18 +404,10 @@ EXPORT_SYMBOL_GPL(raw_notifier_chain_unregister); * Otherwise the return value is the return value * of the last notifier function called. */ -int __raw_notifier_call_chain(struct raw_notifier_head *nh, - unsigned long val, void *v, - int nr_to_call, int *nr_calls) -{ - return notifier_call_chain(&nh->head, val, v, nr_to_call, nr_calls); -} -EXPORT_SYMBOL_GPL(__raw_notifier_call_chain); - int raw_notifier_call_chain(struct raw_notifier_head *nh, unsigned long val, void *v) { - return __raw_notifier_call_chain(nh, val, v, -1, NULL); + return notifier_call_chain(&nh->head, val, v, -1, NULL); } EXPORT_SYMBOL_GPL(raw_notifier_call_chain); @@ -437,12 +479,10 @@ int srcu_notifier_chain_unregister(struct srcu_notifier_head *nh, EXPORT_SYMBOL_GPL(srcu_notifier_chain_unregister); /** - * __srcu_notifier_call_chain - Call functions in an SRCU notifier chain + * srcu_notifier_call_chain - Call functions in an SRCU notifier chain * @nh: Pointer to head of the SRCU notifier chain * @val: Value passed unmodified to notifier function * @v: Pointer passed unmodified to notifier function - * @nr_to_call: See comment for notifier_call_chain. - * @nr_calls: See comment for notifier_call_chain * * Calls each function in a notifier chain in turn. The functions * run in a process context, so they are allowed to block. @@ -454,25 +494,17 @@ EXPORT_SYMBOL_GPL(srcu_notifier_chain_unregister); * Otherwise the return value is the return value * of the last notifier function called. */ -int __srcu_notifier_call_chain(struct srcu_notifier_head *nh, - unsigned long val, void *v, - int nr_to_call, int *nr_calls) +int srcu_notifier_call_chain(struct srcu_notifier_head *nh, + unsigned long val, void *v) { int ret; int idx; idx = srcu_read_lock(&nh->srcu); - ret = notifier_call_chain(&nh->head, val, v, nr_to_call, nr_calls); + ret = notifier_call_chain(&nh->head, val, v, -1, NULL); srcu_read_unlock(&nh->srcu, idx); return ret; } -EXPORT_SYMBOL_GPL(__srcu_notifier_call_chain); - -int srcu_notifier_call_chain(struct srcu_notifier_head *nh, - unsigned long val, void *v) -{ - return __srcu_notifier_call_chain(nh, val, v, -1, NULL); -} EXPORT_SYMBOL_GPL(srcu_notifier_call_chain); /** diff --git a/kernel/power/hibernate.c b/kernel/power/hibernate.c index e7aa57fb2fdc..1dee70815f3c 100644 --- a/kernel/power/hibernate.c +++ b/kernel/power/hibernate.c @@ -706,8 +706,8 @@ static int load_image_and_restore(void) */ int hibernate(void) { - int error, nr_calls = 0; bool snapshot_test = false; + int error; if (!hibernation_available()) { pm_pr_dbg("Hibernation not available.\n"); @@ -723,11 +723,9 @@ int hibernate(void) pr_info("hibernation entry\n"); pm_prepare_console(); - error = __pm_notifier_call_chain(PM_HIBERNATION_PREPARE, -1, &nr_calls); - if (error) { - nr_calls--; - goto Exit; - } + error = pm_notifier_call_chain_robust(PM_HIBERNATION_PREPARE, PM_POST_HIBERNATION); + if (error) + goto Restore; ksys_sync_helper(); @@ -785,7 +783,8 @@ int hibernate(void) /* Don't bother checking whether freezer_test_done is true */ freezer_test_done = false; Exit: - __pm_notifier_call_chain(PM_POST_HIBERNATION, nr_calls, NULL); + pm_notifier_call_chain(PM_POST_HIBERNATION); + Restore: pm_restore_console(); hibernate_release(); Unlock: @@ -804,7 +803,7 @@ int hibernate(void) */ int hibernate_quiet_exec(int (*func)(void *data), void *data) { - int error, nr_calls = 0; + int error; lock_system_sleep(); @@ -815,11 +814,9 @@ int hibernate_quiet_exec(int (*func)(void *data), void *data) pm_prepare_console(); - error = __pm_notifier_call_chain(PM_HIBERNATION_PREPARE, -1, &nr_calls); - if (error) { - nr_calls--; - goto exit; - } + error = pm_notifier_call_chain_robust(PM_HIBERNATION_PREPARE, PM_POST_HIBERNATION); + if (error) + goto restore; error = freeze_processes(); if (error) @@ -880,8 +877,9 @@ thaw: thaw_processes(); exit: - __pm_notifier_call_chain(PM_POST_HIBERNATION, nr_calls, NULL); + pm_notifier_call_chain(PM_POST_HIBERNATION); +restore: pm_restore_console(); hibernate_release(); @@ -910,7 +908,7 @@ EXPORT_SYMBOL_GPL(hibernate_quiet_exec); */ static int software_resume(void) { - int error, nr_calls = 0; + int error; /* * If the user said "noresume".. bail out early. @@ -997,11 +995,9 @@ static int software_resume(void) pr_info("resume from hibernation\n"); pm_prepare_console(); - error = __pm_notifier_call_chain(PM_RESTORE_PREPARE, -1, &nr_calls); - if (error) { - nr_calls--; - goto Close_Finish; - } + error = pm_notifier_call_chain_robust(PM_RESTORE_PREPARE, PM_POST_RESTORE); + if (error) + goto Restore; pm_pr_dbg("Preparing processes for hibernation restore.\n"); error = freeze_processes(); @@ -1017,7 +1013,8 @@ static int software_resume(void) error = load_image_and_restore(); thaw_processes(); Finish: - __pm_notifier_call_chain(PM_POST_RESTORE, nr_calls, NULL); + pm_notifier_call_chain(PM_POST_RESTORE); + Restore: pm_restore_console(); pr_info("resume failed (%d)\n", error); hibernate_release(); diff --git a/kernel/power/main.c b/kernel/power/main.c index 40f86ec4ab30..0aefd6f57e0a 100644 --- a/kernel/power/main.c +++ b/kernel/power/main.c @@ -80,18 +80,18 @@ int unregister_pm_notifier(struct notifier_block *nb) } EXPORT_SYMBOL_GPL(unregister_pm_notifier); -int __pm_notifier_call_chain(unsigned long val, int nr_to_call, int *nr_calls) +int pm_notifier_call_chain_robust(unsigned long val_up, unsigned long val_down) { int ret; - ret = __blocking_notifier_call_chain(&pm_chain_head, val, NULL, - nr_to_call, nr_calls); + ret = blocking_notifier_call_chain_robust(&pm_chain_head, val_up, val_down, NULL); return notifier_to_errno(ret); } + int pm_notifier_call_chain(unsigned long val) { - return __pm_notifier_call_chain(val, -1, NULL); + return blocking_notifier_call_chain(&pm_chain_head, val, NULL); } /* If set, devices may be suspended and resumed asynchronously. */ diff --git a/kernel/power/power.h b/kernel/power/power.h index 32fc89ac96c3..24f12d534515 100644 --- a/kernel/power/power.h +++ b/kernel/power/power.h @@ -210,8 +210,7 @@ static inline void suspend_test_finish(const char *label) {} #ifdef CONFIG_PM_SLEEP /* kernel/power/main.c */ -extern int __pm_notifier_call_chain(unsigned long val, int nr_to_call, - int *nr_calls); +extern int pm_notifier_call_chain_robust(unsigned long val_up, unsigned long val_down); extern int pm_notifier_call_chain(unsigned long val); #endif diff --git a/kernel/power/suspend.c b/kernel/power/suspend.c index 8b1bb5ee7e5d..32391acc806b 100644 --- a/kernel/power/suspend.c +++ b/kernel/power/suspend.c @@ -342,18 +342,16 @@ static int suspend_test(int level) */ static int suspend_prepare(suspend_state_t state) { - int error, nr_calls = 0; + int error; if (!sleep_state_supported(state)) return -EPERM; pm_prepare_console(); - error = __pm_notifier_call_chain(PM_SUSPEND_PREPARE, -1, &nr_calls); - if (error) { - nr_calls--; - goto Finish; - } + error = pm_notifier_call_chain_robust(PM_SUSPEND_PREPARE, PM_POST_SUSPEND); + if (error) + goto Restore; trace_suspend_resume(TPS("freeze_processes"), 0, true); error = suspend_freeze_processes(); @@ -363,8 +361,8 @@ static int suspend_prepare(suspend_state_t state) suspend_stats.failed_freeze++; dpm_save_failed_step(SUSPEND_FREEZE); - Finish: - __pm_notifier_call_chain(PM_POST_SUSPEND, nr_calls, NULL); + pm_notifier_call_chain(PM_POST_SUSPEND); + Restore: pm_restore_console(); return error; } diff --git a/kernel/power/user.c b/kernel/power/user.c index d5eedc2baa2a..047f598f89a5 100644 --- a/kernel/power/user.c +++ b/kernel/power/user.c @@ -46,7 +46,7 @@ int is_hibernate_resume_dev(const struct inode *bd_inode) static int snapshot_open(struct inode *inode, struct file *filp) { struct snapshot_data *data; - int error, nr_calls = 0; + int error; if (!hibernation_available()) return -EPERM; @@ -73,9 +73,7 @@ static int snapshot_open(struct inode *inode, struct file *filp) swap_type_of(swsusp_resume_device, 0, NULL) : -1; data->mode = O_RDONLY; data->free_bitmaps = false; - error = __pm_notifier_call_chain(PM_HIBERNATION_PREPARE, -1, &nr_calls); - if (error) - __pm_notifier_call_chain(PM_POST_HIBERNATION, --nr_calls, NULL); + error = pm_notifier_call_chain_robust(PM_HIBERNATION_PREPARE, PM_POST_HIBERNATION); } else { /* * Resuming. We may need to wait for the image device to @@ -85,15 +83,11 @@ static int snapshot_open(struct inode *inode, struct file *filp) data->swap = -1; data->mode = O_WRONLY; - error = __pm_notifier_call_chain(PM_RESTORE_PREPARE, -1, &nr_calls); + error = pm_notifier_call_chain_robust(PM_RESTORE_PREPARE, PM_POST_RESTORE); if (!error) { error = create_basic_memory_bitmaps(); data->free_bitmaps = !error; - } else - nr_calls--; - - if (error) - __pm_notifier_call_chain(PM_POST_RESTORE, nr_calls, NULL); + } } if (error) hibernate_release(); -- cgit v1.2.3 From 0340a6b7fb767f7f296b9bacc9a215920519a644 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 18 Aug 2020 15:57:37 +0200 Subject: module: Fix up module_notifier return values While auditing all module notifiers I noticed a whole bunch of fail wrt the return value. Notifiers have a 'special' return semantics. As is; NOTIFY_DONE vs NOTIFY_OK is a bit vague; but notifier_from_errno(0) results in NOTIFY_OK and NOTIFY_DONE has a comment that says "Don't care". From this I've used NOTIFY_DONE when the function completely ignores the callback and notifier_to_error() isn't used. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Mathieu Desnoyers Reviewed-by: Joel Fernandes (Google) Reviewed-by: Robert Richter Acked-by: Steven Rostedt (VMware) Link: https://lore.kernel.org/r/20200818135804.385360407@infradead.org --- kernel/trace/bpf_trace.c | 8 ++++++-- kernel/trace/trace.c | 2 +- kernel/trace/trace_events.c | 2 +- kernel/trace/trace_printk.c | 4 ++-- kernel/tracepoint.c | 2 +- 5 files changed, 11 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c index a8d4f253ed77..2ecf7892a31b 100644 --- a/kernel/trace/bpf_trace.c +++ b/kernel/trace/bpf_trace.c @@ -2027,10 +2027,11 @@ static int bpf_event_notify(struct notifier_block *nb, unsigned long op, { struct bpf_trace_module *btm, *tmp; struct module *mod = module; + int ret = 0; if (mod->num_bpf_raw_events == 0 || (op != MODULE_STATE_COMING && op != MODULE_STATE_GOING)) - return 0; + goto out; mutex_lock(&bpf_module_mutex); @@ -2040,6 +2041,8 @@ static int bpf_event_notify(struct notifier_block *nb, unsigned long op, if (btm) { btm->module = module; list_add(&btm->list, &bpf_trace_modules); + } else { + ret = -ENOMEM; } break; case MODULE_STATE_GOING: @@ -2055,7 +2058,8 @@ static int bpf_event_notify(struct notifier_block *nb, unsigned long op, mutex_unlock(&bpf_module_mutex); - return 0; +out: + return notifier_from_errno(ret); } static struct notifier_block bpf_module_nb = { diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index f40d850ebabc..df499922b6a4 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -9072,7 +9072,7 @@ static int trace_module_notify(struct notifier_block *self, break; } - return 0; + return NOTIFY_OK; } static struct notifier_block trace_module_nb = { diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c index a85effb2373b..beebf2cd364b 100644 --- a/kernel/trace/trace_events.c +++ b/kernel/trace/trace_events.c @@ -2646,7 +2646,7 @@ static int trace_module_notify(struct notifier_block *self, mutex_unlock(&trace_types_lock); mutex_unlock(&event_mutex); - return 0; + return NOTIFY_OK; } static struct notifier_block trace_module_nb = { diff --git a/kernel/trace/trace_printk.c b/kernel/trace/trace_printk.c index d4e31e969206..bb7783b90361 100644 --- a/kernel/trace/trace_printk.c +++ b/kernel/trace/trace_printk.c @@ -96,7 +96,7 @@ static int module_trace_bprintk_format_notify(struct notifier_block *self, if (val == MODULE_STATE_COMING) hold_module_trace_bprintk_format(start, end); } - return 0; + return NOTIFY_OK; } /* @@ -174,7 +174,7 @@ __init static int module_trace_bprintk_format_notify(struct notifier_block *self, unsigned long val, void *data) { - return 0; + return NOTIFY_OK; } static inline const char ** find_next_mod_format(int start_index, void *v, const char **fmt, loff_t *pos) diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c index 73956eaff8a9..8e05ed2cd39e 100644 --- a/kernel/tracepoint.c +++ b/kernel/tracepoint.c @@ -521,7 +521,7 @@ static int tracepoint_module_notify(struct notifier_block *self, case MODULE_STATE_UNFORMED: break; } - return ret; + return notifier_from_errno(ret); } static struct notifier_block tracepoint_module_nb = { -- cgit v1.2.3 From 59cc8e0a906ea23190922e5e0252e5b5a60d70c2 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 18 Aug 2020 15:57:38 +0200 Subject: module: Properly propagate MODULE_STATE_COMING failure Now that notifiers got unbroken; use the proper interface to handle notifier errors and propagate them. There were already MODULE_STATE_COMING notifiers that failed; notably: - jump_label_module_notifier() - tracepoint_module_notify() - bpf_event_notify() By propagating this error, we fix those users. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Miroslav Benes Acked-by: Jessica Yu Acked-by: Josh Poimboeuf Link: https://lore.kernel.org/r/20200818135804.444372853@infradead.org --- kernel/module.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/module.c b/kernel/module.c index 1c5cff34d9f2..3c465cf31d08 100644 --- a/kernel/module.c +++ b/kernel/module.c @@ -3792,9 +3792,13 @@ static int prepare_coming_module(struct module *mod) if (err) return err; - blocking_notifier_call_chain(&module_notify_list, - MODULE_STATE_COMING, mod); - return 0; + err = blocking_notifier_call_chain_robust(&module_notify_list, + MODULE_STATE_COMING, MODULE_STATE_GOING, mod); + err = notifier_to_errno(err); + if (err) + klp_module_going(mod); + + return err; } static int unknown_module_param_cb(char *param, char *val, const char *modname, -- cgit v1.2.3 From 0db6e3734b130207026df1e78455fa98ca1d6f50 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 18 Aug 2020 15:57:39 +0200 Subject: jump_label,module: Fix module lifetime for __jump_label_mod_text_reserved() Nothing ensures the module exists while we're iterating mod->jump_entries in __jump_label_mod_text_reserved(), take a module reference to ensure the module sticks around. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Steven Rostedt (VMware) Link: https://lore.kernel.org/r/20200818135804.504501338@infradead.org --- kernel/jump_label.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/jump_label.c b/kernel/jump_label.c index cdb3ffab128b..e661c61b3d6b 100644 --- a/kernel/jump_label.c +++ b/kernel/jump_label.c @@ -539,19 +539,25 @@ static void static_key_set_mod(struct static_key *key, static int __jump_label_mod_text_reserved(void *start, void *end) { struct module *mod; + int ret; preempt_disable(); mod = __module_text_address((unsigned long)start); WARN_ON_ONCE(__module_text_address((unsigned long)end) != mod); + if (!try_module_get(mod)) + mod = NULL; preempt_enable(); if (!mod) return 0; - - return __jump_label_text_reserved(mod->jump_entries, + ret = __jump_label_text_reserved(mod->jump_entries, mod->jump_entries + mod->num_jump_entries, start, end); + + module_put(mod); + + return ret; } static void __jump_label_mod_update(struct static_key *key) -- cgit v1.2.3 From 9183c3f9ed710a8edf1a61e8a96d497258d26e08 Mon Sep 17 00:00:00 2001 From: Josh Poimboeuf Date: Tue, 18 Aug 2020 15:57:42 +0200 Subject: static_call: Add inline static call infrastructure Add infrastructure for an arch-specific CONFIG_HAVE_STATIC_CALL_INLINE option, which is a faster version of CONFIG_HAVE_STATIC_CALL. At runtime, the static call sites are patched directly, rather than using the out-of-line trampolines. Compared to out-of-line static calls, the performance benefits are more modest, but still measurable. Steven Rostedt did some tracepoint measurements: https://lkml.kernel.org/r/20181126155405.72b4f718@gandalf.local.home This code is heavily inspired by the jump label code (aka "static jumps"), as some of the concepts are very similar. For more details, see the comments in include/linux/static_call.h. [peterz: simplified interface; merged trampolines] Signed-off-by: Josh Poimboeuf Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Steven Rostedt (VMware) Cc: Linus Torvalds Link: https://lore.kernel.org/r/20200818135804.684334440@infradead.org --- kernel/Makefile | 1 + kernel/module.c | 5 + kernel/static_call.c | 303 +++++++++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 309 insertions(+) create mode 100644 kernel/static_call.c (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index 9a20016d4900..b74820d8b264 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -111,6 +111,7 @@ obj-$(CONFIG_CPU_PM) += cpu_pm.o obj-$(CONFIG_BPF) += bpf/ obj-$(CONFIG_KCSAN) += kcsan/ obj-$(CONFIG_SHADOW_CALL_STACK) += scs.o +obj-$(CONFIG_HAVE_STATIC_CALL_INLINE) += static_call.o obj-$(CONFIG_PERF_EVENTS) += events/ diff --git a/kernel/module.c b/kernel/module.c index 3c465cf31d08..c075a18103fb 100644 --- a/kernel/module.c +++ b/kernel/module.c @@ -3274,6 +3274,11 @@ static int find_module_sections(struct module *mod, struct load_info *info) mod->kprobe_blacklist = section_objs(info, "_kprobe_blacklist", sizeof(unsigned long), &mod->num_kprobe_blacklist); +#endif +#ifdef CONFIG_HAVE_STATIC_CALL_INLINE + mod->static_call_sites = section_objs(info, ".static_call_sites", + sizeof(*mod->static_call_sites), + &mod->num_static_call_sites); #endif mod->extable = section_objs(info, "__ex_table", sizeof(*mod->extable), &mod->num_exentries); diff --git a/kernel/static_call.c b/kernel/static_call.c new file mode 100644 index 000000000000..d24349244675 --- /dev/null +++ b/kernel/static_call.c @@ -0,0 +1,303 @@ +// SPDX-License-Identifier: GPL-2.0 +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +extern struct static_call_site __start_static_call_sites[], + __stop_static_call_sites[]; + +static bool static_call_initialized; + +#define STATIC_CALL_INIT 1UL + +/* mutex to protect key modules/sites */ +static DEFINE_MUTEX(static_call_mutex); + +static void static_call_lock(void) +{ + mutex_lock(&static_call_mutex); +} + +static void static_call_unlock(void) +{ + mutex_unlock(&static_call_mutex); +} + +static inline void *static_call_addr(struct static_call_site *site) +{ + return (void *)((long)site->addr + (long)&site->addr); +} + + +static inline struct static_call_key *static_call_key(const struct static_call_site *site) +{ + return (struct static_call_key *) + (((long)site->key + (long)&site->key) & ~STATIC_CALL_INIT); +} + +/* These assume the key is word-aligned. */ +static inline bool static_call_is_init(struct static_call_site *site) +{ + return ((long)site->key + (long)&site->key) & STATIC_CALL_INIT; +} + +static inline void static_call_set_init(struct static_call_site *site) +{ + site->key = ((long)static_call_key(site) | STATIC_CALL_INIT) - + (long)&site->key; +} + +static int static_call_site_cmp(const void *_a, const void *_b) +{ + const struct static_call_site *a = _a; + const struct static_call_site *b = _b; + const struct static_call_key *key_a = static_call_key(a); + const struct static_call_key *key_b = static_call_key(b); + + if (key_a < key_b) + return -1; + + if (key_a > key_b) + return 1; + + return 0; +} + +static void static_call_site_swap(void *_a, void *_b, int size) +{ + long delta = (unsigned long)_a - (unsigned long)_b; + struct static_call_site *a = _a; + struct static_call_site *b = _b; + struct static_call_site tmp = *a; + + a->addr = b->addr - delta; + a->key = b->key - delta; + + b->addr = tmp.addr + delta; + b->key = tmp.key + delta; +} + +static inline void static_call_sort_entries(struct static_call_site *start, + struct static_call_site *stop) +{ + sort(start, stop - start, sizeof(struct static_call_site), + static_call_site_cmp, static_call_site_swap); +} + +void __static_call_update(struct static_call_key *key, void *tramp, void *func) +{ + struct static_call_site *site, *stop; + struct static_call_mod *site_mod; + + cpus_read_lock(); + static_call_lock(); + + if (key->func == func) + goto done; + + key->func = func; + + arch_static_call_transform(NULL, tramp, func); + + /* + * If uninitialized, we'll not update the callsites, but they still + * point to the trampoline and we just patched that. + */ + if (WARN_ON_ONCE(!static_call_initialized)) + goto done; + + for (site_mod = key->mods; site_mod; site_mod = site_mod->next) { + struct module *mod = site_mod->mod; + + if (!site_mod->sites) { + /* + * This can happen if the static call key is defined in + * a module which doesn't use it. + */ + continue; + } + + stop = __stop_static_call_sites; + +#ifdef CONFIG_MODULES + if (mod) { + stop = mod->static_call_sites + + mod->num_static_call_sites; + } +#endif + + for (site = site_mod->sites; + site < stop && static_call_key(site) == key; site++) { + void *site_addr = static_call_addr(site); + + if (static_call_is_init(site)) { + /* + * Don't write to call sites which were in + * initmem and have since been freed. + */ + if (!mod && system_state >= SYSTEM_RUNNING) + continue; + if (mod && !within_module_init((unsigned long)site_addr, mod)) + continue; + } + + if (!kernel_text_address((unsigned long)site_addr)) { + WARN_ONCE(1, "can't patch static call site at %pS", + site_addr); + continue; + } + + arch_static_call_transform(site_addr, NULL, func); + } + } + +done: + static_call_unlock(); + cpus_read_unlock(); +} +EXPORT_SYMBOL_GPL(__static_call_update); + +static int __static_call_init(struct module *mod, + struct static_call_site *start, + struct static_call_site *stop) +{ + struct static_call_site *site; + struct static_call_key *key, *prev_key = NULL; + struct static_call_mod *site_mod; + + if (start == stop) + return 0; + + static_call_sort_entries(start, stop); + + for (site = start; site < stop; site++) { + void *site_addr = static_call_addr(site); + + if ((mod && within_module_init((unsigned long)site_addr, mod)) || + (!mod && init_section_contains(site_addr, 1))) + static_call_set_init(site); + + key = static_call_key(site); + if (key != prev_key) { + prev_key = key; + + site_mod = kzalloc(sizeof(*site_mod), GFP_KERNEL); + if (!site_mod) + return -ENOMEM; + + site_mod->mod = mod; + site_mod->sites = site; + site_mod->next = key->mods; + key->mods = site_mod; + } + + arch_static_call_transform(site_addr, NULL, key->func); + } + + return 0; +} + +#ifdef CONFIG_MODULES + +static int static_call_add_module(struct module *mod) +{ + return __static_call_init(mod, mod->static_call_sites, + mod->static_call_sites + mod->num_static_call_sites); +} + +static void static_call_del_module(struct module *mod) +{ + struct static_call_site *start = mod->static_call_sites; + struct static_call_site *stop = mod->static_call_sites + + mod->num_static_call_sites; + struct static_call_key *key, *prev_key = NULL; + struct static_call_mod *site_mod, **prev; + struct static_call_site *site; + + for (site = start; site < stop; site++) { + key = static_call_key(site); + if (key == prev_key) + continue; + + prev_key = key; + + for (prev = &key->mods, site_mod = key->mods; + site_mod && site_mod->mod != mod; + prev = &site_mod->next, site_mod = site_mod->next) + ; + + if (!site_mod) + continue; + + *prev = site_mod->next; + kfree(site_mod); + } +} + +static int static_call_module_notify(struct notifier_block *nb, + unsigned long val, void *data) +{ + struct module *mod = data; + int ret = 0; + + cpus_read_lock(); + static_call_lock(); + + switch (val) { + case MODULE_STATE_COMING: + ret = static_call_add_module(mod); + if (ret) { + WARN(1, "Failed to allocate memory for static calls"); + static_call_del_module(mod); + } + break; + case MODULE_STATE_GOING: + static_call_del_module(mod); + break; + } + + static_call_unlock(); + cpus_read_unlock(); + + return notifier_from_errno(ret); +} + +static struct notifier_block static_call_module_nb = { + .notifier_call = static_call_module_notify, +}; + +#endif /* CONFIG_MODULES */ + +static void __init static_call_init(void) +{ + int ret; + + if (static_call_initialized) + return; + + cpus_read_lock(); + static_call_lock(); + ret = __static_call_init(NULL, __start_static_call_sites, + __stop_static_call_sites); + static_call_unlock(); + cpus_read_unlock(); + + if (ret) { + pr_err("Failed to allocate memory for static_call!\n"); + BUG(); + } + + static_call_initialized = true; + +#ifdef CONFIG_MODULES + register_module_notifier(&static_call_module_nb); +#endif +} +early_initcall(static_call_init); -- cgit v1.2.3 From 6333e8f73b834f54e395a056e6002403f0862c51 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 18 Aug 2020 15:57:43 +0200 Subject: static_call: Avoid kprobes on inline static_call()s Similar to how we disallow kprobes on any other dynamic text (ftrace/jump_label) also disallow kprobes on inline static_call()s. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20200818135804.744920586@infradead.org --- kernel/kprobes.c | 2 ++ kernel/static_call.c | 68 ++++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 70 insertions(+) (limited to 'kernel') diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 287b263c9cb9..67e6a8c18007 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -36,6 +36,7 @@ #include #include #include +#include #include #include @@ -1634,6 +1635,7 @@ static int check_kprobe_address_safe(struct kprobe *p, if (!kernel_text_address((unsigned long) p->addr) || within_kprobe_blacklist((unsigned long) p->addr) || jump_label_text_reserved(p->addr, p->addr) || + static_call_text_reserved(p->addr, p->addr) || find_bug((unsigned long)p->addr)) { ret = -EINVAL; goto out; diff --git a/kernel/static_call.c b/kernel/static_call.c index d24349244675..753b2f1b4fb8 100644 --- a/kernel/static_call.c +++ b/kernel/static_call.c @@ -204,8 +204,58 @@ static int __static_call_init(struct module *mod, return 0; } +static int addr_conflict(struct static_call_site *site, void *start, void *end) +{ + unsigned long addr = (unsigned long)static_call_addr(site); + + if (addr <= (unsigned long)end && + addr + CALL_INSN_SIZE > (unsigned long)start) + return 1; + + return 0; +} + +static int __static_call_text_reserved(struct static_call_site *iter_start, + struct static_call_site *iter_stop, + void *start, void *end) +{ + struct static_call_site *iter = iter_start; + + while (iter < iter_stop) { + if (addr_conflict(iter, start, end)) + return 1; + iter++; + } + + return 0; +} + #ifdef CONFIG_MODULES +static int __static_call_mod_text_reserved(void *start, void *end) +{ + struct module *mod; + int ret; + + preempt_disable(); + mod = __module_text_address((unsigned long)start); + WARN_ON_ONCE(__module_text_address((unsigned long)end) != mod); + if (!try_module_get(mod)) + mod = NULL; + preempt_enable(); + + if (!mod) + return 0; + + ret = __static_call_text_reserved(mod->static_call_sites, + mod->static_call_sites + mod->num_static_call_sites, + start, end); + + module_put(mod); + + return ret; +} + static int static_call_add_module(struct module *mod) { return __static_call_init(mod, mod->static_call_sites, @@ -273,8 +323,26 @@ static struct notifier_block static_call_module_nb = { .notifier_call = static_call_module_notify, }; +#else + +static inline int __static_call_mod_text_reserved(void *start, void *end) +{ + return 0; +} + #endif /* CONFIG_MODULES */ +int static_call_text_reserved(void *start, void *end) +{ + int ret = __static_call_text_reserved(__start_static_call_sites, + __stop_static_call_sites, start, end); + + if (ret) + return ret; + + return __static_call_mod_text_reserved(start, end); +} + static void __init static_call_init(void) { int ret; -- cgit v1.2.3 From f03c412915f5f69f2d17bcd20ecdd69320bcbf7b Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 18 Aug 2020 15:57:46 +0200 Subject: static_call: Add simple self-test for static calls Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20200818135804.922581202@infradead.org --- kernel/static_call.c | 43 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 43 insertions(+) (limited to 'kernel') diff --git a/kernel/static_call.c b/kernel/static_call.c index 753b2f1b4fb8..97142cb6bfa6 100644 --- a/kernel/static_call.c +++ b/kernel/static_call.c @@ -369,3 +369,46 @@ static void __init static_call_init(void) #endif } early_initcall(static_call_init); + +#ifdef CONFIG_STATIC_CALL_SELFTEST + +static int func_a(int x) +{ + return x+1; +} + +static int func_b(int x) +{ + return x+2; +} + +DEFINE_STATIC_CALL(sc_selftest, func_a); + +static struct static_call_data { + int (*func)(int); + int val; + int expect; +} static_call_data [] __initdata = { + { NULL, 2, 3 }, + { func_b, 2, 4 }, + { func_a, 2, 3 } +}; + +static int __init test_static_call_init(void) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(static_call_data); i++ ) { + struct static_call_data *scd = &static_call_data[i]; + + if (scd->func) + static_call_update(sc_selftest, scd->func); + + WARN_ON(static_call(sc_selftest)(scd->val) != scd->expect); + } + + return 0; +} +early_initcall(test_static_call_init); + +#endif /* CONFIG_STATIC_CALL_SELFTEST */ -- cgit v1.2.3 From 5b06fd3bb9cdce4f3e731c48eb5b74c4acc47997 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 18 Aug 2020 15:57:49 +0200 Subject: static_call: Handle tail-calls GCC can turn our static_call(name)(args...) into a tail call, in which case we get a JMP.d32 into the trampoline (which then does a further tail-call). Teach objtool to recognise and mark these in .static_call_sites and adjust the code patching to deal with this. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Cc: Linus Torvalds Link: https://lore.kernel.org/r/20200818135805.101186767@infradead.org --- kernel/static_call.c | 21 +++++++++++++-------- 1 file changed, 13 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/static_call.c b/kernel/static_call.c index 97142cb6bfa6..d98e0e4272c1 100644 --- a/kernel/static_call.c +++ b/kernel/static_call.c @@ -15,8 +15,6 @@ extern struct static_call_site __start_static_call_sites[], static bool static_call_initialized; -#define STATIC_CALL_INIT 1UL - /* mutex to protect key modules/sites */ static DEFINE_MUTEX(static_call_mutex); @@ -39,18 +37,23 @@ static inline void *static_call_addr(struct static_call_site *site) static inline struct static_call_key *static_call_key(const struct static_call_site *site) { return (struct static_call_key *) - (((long)site->key + (long)&site->key) & ~STATIC_CALL_INIT); + (((long)site->key + (long)&site->key) & ~STATIC_CALL_SITE_FLAGS); } /* These assume the key is word-aligned. */ static inline bool static_call_is_init(struct static_call_site *site) { - return ((long)site->key + (long)&site->key) & STATIC_CALL_INIT; + return ((long)site->key + (long)&site->key) & STATIC_CALL_SITE_INIT; +} + +static inline bool static_call_is_tail(struct static_call_site *site) +{ + return ((long)site->key + (long)&site->key) & STATIC_CALL_SITE_TAIL; } static inline void static_call_set_init(struct static_call_site *site) { - site->key = ((long)static_call_key(site) | STATIC_CALL_INIT) - + site->key = ((long)static_call_key(site) | STATIC_CALL_SITE_INIT) - (long)&site->key; } @@ -104,7 +107,7 @@ void __static_call_update(struct static_call_key *key, void *tramp, void *func) key->func = func; - arch_static_call_transform(NULL, tramp, func); + arch_static_call_transform(NULL, tramp, func, false); /* * If uninitialized, we'll not update the callsites, but they still @@ -154,7 +157,8 @@ void __static_call_update(struct static_call_key *key, void *tramp, void *func) continue; } - arch_static_call_transform(site_addr, NULL, func); + arch_static_call_transform(site_addr, NULL, func, + static_call_is_tail(site)); } } @@ -198,7 +202,8 @@ static int __static_call_init(struct module *mod, key->mods = site_mod; } - arch_static_call_transform(site_addr, NULL, key->func); + arch_static_call_transform(site_addr, NULL, key->func, + static_call_is_tail(site)); } return 0; -- cgit v1.2.3 From a945c8345ec0decb2f1a7f19a8c5e60bcb1dd1eb Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 18 Aug 2020 15:57:51 +0200 Subject: static_call: Allow early init In order to use static_call() to wire up x86_pmu, we need to initialize earlier, specifically before memory allocation works; copy some of the tricks from jump_label to enable this. Primarily we overload key->next to store a sites pointer when there are no modules, this avoids having to use kmalloc() to initialize the sites and allows us to run much earlier. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Steven Rostedt (VMware) Link: https://lore.kernel.org/r/20200818135805.220737930@infradead.org --- kernel/static_call.c | 70 +++++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 66 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/static_call.c b/kernel/static_call.c index d98e0e4272c1..f8362b3f8fd5 100644 --- a/kernel/static_call.c +++ b/kernel/static_call.c @@ -94,10 +94,31 @@ static inline void static_call_sort_entries(struct static_call_site *start, static_call_site_cmp, static_call_site_swap); } +static inline bool static_call_key_has_mods(struct static_call_key *key) +{ + return !(key->type & 1); +} + +static inline struct static_call_mod *static_call_key_next(struct static_call_key *key) +{ + if (!static_call_key_has_mods(key)) + return NULL; + + return key->mods; +} + +static inline struct static_call_site *static_call_key_sites(struct static_call_key *key) +{ + if (static_call_key_has_mods(key)) + return NULL; + + return (struct static_call_site *)(key->type & ~1); +} + void __static_call_update(struct static_call_key *key, void *tramp, void *func) { struct static_call_site *site, *stop; - struct static_call_mod *site_mod; + struct static_call_mod *site_mod, first; cpus_read_lock(); static_call_lock(); @@ -116,13 +137,22 @@ void __static_call_update(struct static_call_key *key, void *tramp, void *func) if (WARN_ON_ONCE(!static_call_initialized)) goto done; - for (site_mod = key->mods; site_mod; site_mod = site_mod->next) { + first = (struct static_call_mod){ + .next = static_call_key_next(key), + .mod = NULL, + .sites = static_call_key_sites(key), + }; + + for (site_mod = &first; site_mod; site_mod = site_mod->next) { struct module *mod = site_mod->mod; if (!site_mod->sites) { /* * This can happen if the static call key is defined in * a module which doesn't use it. + * + * It also happens in the has_mods case, where the + * 'first' entry has no sites associated with it. */ continue; } @@ -192,16 +222,48 @@ static int __static_call_init(struct module *mod, if (key != prev_key) { prev_key = key; + /* + * For vmlinux (!mod) avoid the allocation by storing + * the sites pointer in the key itself. Also see + * __static_call_update()'s @first. + * + * This allows architectures (eg. x86) to call + * static_call_init() before memory allocation works. + */ + if (!mod) { + key->sites = site; + key->type |= 1; + goto do_transform; + } + site_mod = kzalloc(sizeof(*site_mod), GFP_KERNEL); if (!site_mod) return -ENOMEM; + /* + * When the key has a direct sites pointer, extract + * that into an explicit struct static_call_mod, so we + * can have a list of modules. + */ + if (static_call_key_sites(key)) { + site_mod->mod = NULL; + site_mod->next = NULL; + site_mod->sites = static_call_key_sites(key); + + key->mods = site_mod; + + site_mod = kzalloc(sizeof(*site_mod), GFP_KERNEL); + if (!site_mod) + return -ENOMEM; + } + site_mod->mod = mod; site_mod->sites = site; - site_mod->next = key->mods; + site_mod->next = static_call_key_next(key); key->mods = site_mod; } +do_transform: arch_static_call_transform(site_addr, NULL, key->func, static_call_is_tail(site)); } @@ -348,7 +410,7 @@ int static_call_text_reserved(void *start, void *end) return __static_call_mod_text_reserved(start, end); } -static void __init static_call_init(void) +void __init static_call_init(void) { int ret; -- cgit v1.2.3 From d25e37d89dd2f41d7acae0429039d2f0ae8b4a07 Mon Sep 17 00:00:00 2001 From: "Steven Rostedt (VMware)" Date: Tue, 18 Aug 2020 15:57:52 +0200 Subject: tracepoint: Optimize using static_call() Currently the tracepoint site will iterate a vector and issue indirect calls to however many handlers are registered (ie. the vector is long). Using static_call() it is possible to optimize this for the common case of only having a single handler registered. In this case the static_call() can directly call this handler. Otherwise, if the vector is longer than 1, call a function that iterates the whole vector like the current code. [peterz: updated to new interface] Signed-off-by: Steven Rostedt (VMware) Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Cc: Linus Torvalds Link: https://lore.kernel.org/r/20200818135805.279421092@infradead.org --- kernel/tracepoint.c | 25 +++++++++++++++++++++---- 1 file changed, 21 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c index 8e05ed2cd39e..e92f3fb8887a 100644 --- a/kernel/tracepoint.c +++ b/kernel/tracepoint.c @@ -221,6 +221,20 @@ static void *func_remove(struct tracepoint_func **funcs, return old; } +static void tracepoint_update_call(struct tracepoint *tp, struct tracepoint_func *tp_funcs) +{ + void *func = tp->iterator; + + /* Synthetic events do not have static call sites */ + if (!tp->static_call_key) + return; + + if (!tp_funcs[1].func) + func = tp_funcs[0].func; + + __static_call_update(tp->static_call_key, tp->static_call_tramp, func); +} + /* * Add the probe function to a tracepoint. */ @@ -251,8 +265,9 @@ static int tracepoint_add_func(struct tracepoint *tp, * include/linux/tracepoint.h using rcu_dereference_sched(). */ rcu_assign_pointer(tp->funcs, tp_funcs); - if (!static_key_enabled(&tp->key)) - static_key_slow_inc(&tp->key); + tracepoint_update_call(tp, tp_funcs); + static_key_enable(&tp->key); + release_probes(old); return 0; } @@ -281,9 +296,11 @@ static int tracepoint_remove_func(struct tracepoint *tp, if (tp->unregfunc && static_key_enabled(&tp->key)) tp->unregfunc(); - if (static_key_enabled(&tp->key)) - static_key_slow_dec(&tp->key); + static_key_disable(&tp->key); + } else { + tracepoint_update_call(tp, tp_funcs); } + rcu_assign_pointer(tp->funcs, tp_funcs); release_probes(old); return 0; -- cgit v1.2.3 From 547305a64632813286700cb6d768bfe773df7d19 Mon Sep 17 00:00:00 2001 From: "Steven Rostedt (VMware)" Date: Thu, 1 Oct 2020 21:27:57 -0400 Subject: tracepoint: Fix out of sync data passing by static caller Naresh reported a bug that appears to be a side effect of the static calls. It happens when going from more than one tracepoint callback to a single one, and removing the first callback on the list. The list of tracepoint callbacks holds data and a function to call with the parameters of that tracepoint and a handler to the associated data. old_list: 0: func = foo; data = NULL; 1: func = bar; data = &bar_struct; new_list: 0: func = bar; data = &bar_struct; CPU 0 CPU 1 ----- ----- tp_funcs = old_list; tp_static_caller = tp_interator __DO_TRACE() data = tp_funcs[0].data = NULL; tp_funcs = new_list; tracepoint_update_call() tp_static_caller = tp_funcs[0] = bar; tp_static_caller(data) bar(data) x = data->item = NULL->item BOOM! To solve this, add a tracepoint_synchronize_unregister() between changing tp_funcs and updating the static tracepoint, that does both a synchronize_rcu() and synchronize_srcu(). This will ensure that when the static call is updated to the single callback that it will be receiving the data that it registered with. Fixes: d25e37d89dd2f ("tracepoint: Optimize using static_call()") Reported-by: Naresh Kamboju Signed-off-by: Steven Rostedt (VMware) Signed-off-by: Peter Zijlstra (Intel) Link: https://lore.kernel.org/linux-next/CA+G9fYvPXVRO0NV7yL=FxCmFEMYkCwdz7R=9W+_votpT824YJA@mail.gmail.com --- kernel/tracepoint.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c index e92f3fb8887a..26efd22f0633 100644 --- a/kernel/tracepoint.c +++ b/kernel/tracepoint.c @@ -221,7 +221,7 @@ static void *func_remove(struct tracepoint_func **funcs, return old; } -static void tracepoint_update_call(struct tracepoint *tp, struct tracepoint_func *tp_funcs) +static void tracepoint_update_call(struct tracepoint *tp, struct tracepoint_func *tp_funcs, bool sync) { void *func = tp->iterator; @@ -229,8 +229,17 @@ static void tracepoint_update_call(struct tracepoint *tp, struct tracepoint_func if (!tp->static_call_key) return; - if (!tp_funcs[1].func) + if (!tp_funcs[1].func) { func = tp_funcs[0].func; + /* + * If going from the iterator back to a single caller, + * we need to synchronize with __DO_TRACE to make sure + * that the data passed to the callback is the one that + * belongs to that callback. + */ + if (sync) + tracepoint_synchronize_unregister(); + } __static_call_update(tp->static_call_key, tp->static_call_tramp, func); } @@ -265,7 +274,7 @@ static int tracepoint_add_func(struct tracepoint *tp, * include/linux/tracepoint.h using rcu_dereference_sched(). */ rcu_assign_pointer(tp->funcs, tp_funcs); - tracepoint_update_call(tp, tp_funcs); + tracepoint_update_call(tp, tp_funcs, false); static_key_enable(&tp->key); release_probes(old); @@ -297,11 +306,12 @@ static int tracepoint_remove_func(struct tracepoint *tp, tp->unregfunc(); static_key_disable(&tp->key); + rcu_assign_pointer(tp->funcs, tp_funcs); } else { - tracepoint_update_call(tp, tp_funcs); + rcu_assign_pointer(tp->funcs, tp_funcs); + tracepoint_update_call(tp, tp_funcs, + tp_funcs[0].func != old[0].func); } - - rcu_assign_pointer(tp->funcs, tp_funcs); release_probes(old); return 0; } -- cgit v1.2.3 From 69e0ad37c9f32d5aa1beb02aab4ec0cd055be013 Mon Sep 17 00:00:00 2001 From: Nathan Chancellor Date: Mon, 28 Sep 2020 16:09:39 -0700 Subject: static_call: Fix return type of static_call_init Functions that are passed to early_initcall should be of type initcall_t, which expects a return type of int. This is not currently an error but a patch in the Clang LTO series could change that in the future. Fixes: 9183c3f9ed71 ("static_call: Add inline static call infrastructure") Signed-off-by: Nathan Chancellor Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Sami Tolvanen Link: https://lore.kernel.org/lkml/20200903203053.3411268-17-samitolvanen@google.com/ --- kernel/static_call.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/static_call.c b/kernel/static_call.c index f8362b3f8fd5..84565c2a41b8 100644 --- a/kernel/static_call.c +++ b/kernel/static_call.c @@ -410,12 +410,12 @@ int static_call_text_reserved(void *start, void *end) return __static_call_mod_text_reserved(start, end); } -void __init static_call_init(void) +int __init static_call_init(void) { int ret; if (static_call_initialized) - return; + return 0; cpus_read_lock(); static_call_lock(); @@ -434,6 +434,7 @@ void __init static_call_init(void) #ifdef CONFIG_MODULES register_module_notifier(&static_call_module_nb); #endif + return 0; } early_initcall(static_call_init); -- cgit v1.2.3