diff --git a/Documentation/RCU/Design/Requirements/Requirements.rst b/Documentation/RCU/Design/Requirements/Requirements.rst index f511476b4550..6125e7068d2c 100644 --- a/Documentation/RCU/Design/Requirements/Requirements.rst +++ b/Documentation/RCU/Design/Requirements/Requirements.rst @@ -2649,8 +2649,7 @@ those that are idle from RCU's perspective) and then Tasks Rude RCU can be removed from the kernel. The tasks-rude-RCU API is also reader-marking-free and thus quite compact, -consisting of call_rcu_tasks_rude(), synchronize_rcu_tasks_rude(), -and rcu_barrier_tasks_rude(). +consisting solely of synchronize_rcu_tasks_rude(). Tasks Trace RCU ~~~~~~~~~~~~~~~ diff --git a/Documentation/RCU/checklist.rst b/Documentation/RCU/checklist.rst index 3e6407de231c..7de3e308f330 100644 --- a/Documentation/RCU/checklist.rst +++ b/Documentation/RCU/checklist.rst @@ -194,14 +194,13 @@ over a rather long period of time, but improvements are always welcome! when publicizing a pointer to a structure that can be traversed by an RCU read-side critical section. -5. If any of call_rcu(), call_srcu(), call_rcu_tasks(), - call_rcu_tasks_rude(), or call_rcu_tasks_trace() is used, - the callback function may be invoked from softirq context, - and in any case with bottom halves disabled. In particular, - this callback function cannot block. If you need the callback - to block, run that code in a workqueue handler scheduled from - the callback. The queue_rcu_work() function does this for you - in the case of call_rcu(). +5. If any of call_rcu(), call_srcu(), call_rcu_tasks(), or + call_rcu_tasks_trace() is used, the callback function may be + invoked from softirq context, and in any case with bottom halves + disabled. In particular, this callback function cannot block. + If you need the callback to block, run that code in a workqueue + handler scheduled from the callback. The queue_rcu_work() + function does this for you in the case of call_rcu(). 6. Since synchronize_rcu() can block, it cannot be called from any sort of irq context. The same rule applies @@ -254,10 +253,10 @@ over a rather long period of time, but improvements are always welcome! corresponding readers must use rcu_read_lock_trace() and rcu_read_unlock_trace(). - c. If an updater uses call_rcu_tasks_rude() or - synchronize_rcu_tasks_rude(), then the corresponding - readers must use anything that disables preemption, - for example, preempt_disable() and preempt_enable(). + c. If an updater uses synchronize_rcu_tasks_rude(), + then the corresponding readers must use anything that + disables preemption, for example, preempt_disable() + and preempt_enable(). Mixing things up will result in confusion and broken kernels, and has even resulted in an exploitable security issue. Therefore, @@ -326,11 +325,9 @@ over a rather long period of time, but improvements are always welcome! d. Periodically invoke rcu_barrier(), permitting a limited number of updates per grace period. - The same cautions apply to call_srcu(), call_rcu_tasks(), - call_rcu_tasks_rude(), and call_rcu_tasks_trace(). This is - why there is an srcu_barrier(), rcu_barrier_tasks(), - rcu_barrier_tasks_rude(), and rcu_barrier_tasks_rude(), - respectively. + The same cautions apply to call_srcu(), call_rcu_tasks(), and + call_rcu_tasks_trace(). This is why there is an srcu_barrier(), + rcu_barrier_tasks(), and rcu_barrier_tasks_trace(), respectively. Note that although these primitives do take action to avoid memory exhaustion when any given CPU has too many callbacks, @@ -383,17 +380,17 @@ over a rather long period of time, but improvements are always welcome! must use whatever locking or other synchronization is required to safely access and/or modify that data structure. - Do not assume that RCU callbacks will be executed on - the same CPU that executed the corresponding call_rcu(), - call_srcu(), call_rcu_tasks(), call_rcu_tasks_rude(), or - call_rcu_tasks_trace(). For example, if a given CPU goes offline - while having an RCU callback pending, then that RCU callback - will execute on some surviving CPU. (If this was not the case, - a self-spawning RCU callback would prevent the victim CPU from - ever going offline.) Furthermore, CPUs designated by rcu_nocbs= - might well *always* have their RCU callbacks executed on some - other CPUs, in fact, for some real-time workloads, this is the - whole point of using the rcu_nocbs= kernel boot parameter. + Do not assume that RCU callbacks will be executed on the same + CPU that executed the corresponding call_rcu(), call_srcu(), + call_rcu_tasks(), or call_rcu_tasks_trace(). For example, if + a given CPU goes offline while having an RCU callback pending, + then that RCU callback will execute on some surviving CPU. + (If this was not the case, a self-spawning RCU callback would + prevent the victim CPU from ever going offline.) Furthermore, + CPUs designated by rcu_nocbs= might well *always* have their + RCU callbacks executed on some other CPUs, in fact, for some + real-time workloads, this is the whole point of using the + rcu_nocbs= kernel boot parameter. In addition, do not assume that callbacks queued in a given order will be invoked in that order, even if they all are queued on the @@ -507,9 +504,9 @@ over a rather long period of time, but improvements are always welcome! These debugging aids can help you find problems that are otherwise extremely difficult to spot. -17. If you pass a callback function defined within a module to one of - call_rcu(), call_srcu(), call_rcu_tasks(), call_rcu_tasks_rude(), - or call_rcu_tasks_trace(), then it is necessary to wait for all +17. If you pass a callback function defined within a module + to one of call_rcu(), call_srcu(), call_rcu_tasks(), or + call_rcu_tasks_trace(), then it is necessary to wait for all pending callbacks to be invoked before unloading that module. Note that it is absolutely *not* sufficient to wait for a grace period! For example, synchronize_rcu() implementation is *not* @@ -522,7 +519,6 @@ over a rather long period of time, but improvements are always welcome! - call_rcu() -> rcu_barrier() - call_srcu() -> srcu_barrier() - call_rcu_tasks() -> rcu_barrier_tasks() - - call_rcu_tasks_rude() -> rcu_barrier_tasks_rude() - call_rcu_tasks_trace() -> rcu_barrier_tasks_trace() However, these barrier functions are absolutely *not* guaranteed @@ -539,7 +535,6 @@ over a rather long period of time, but improvements are always welcome! - Either synchronize_srcu() or synchronize_srcu_expedited(), together with and srcu_barrier() - synchronize_rcu_tasks() and rcu_barrier_tasks() - - synchronize_tasks_rude() and rcu_barrier_tasks_rude() - synchronize_tasks_trace() and rcu_barrier_tasks_trace() If necessary, you can use something like workqueues to execute diff --git a/Documentation/RCU/whatisRCU.rst b/Documentation/RCU/whatisRCU.rst index d585a5490aee..1ef5784c1b84 100644 --- a/Documentation/RCU/whatisRCU.rst +++ b/Documentation/RCU/whatisRCU.rst @@ -1103,7 +1103,7 @@ RCU-Tasks-Rude:: Critical sections Grace period Barrier - N/A call_rcu_tasks_rude rcu_barrier_tasks_rude + N/A N/A synchronize_rcu_tasks_rude diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt index f1384c7b59c9..d30b81e693f9 100644 --- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -4937,6 +4937,10 @@ Set maximum number of finished RCU callbacks to process in one batch. + rcutree.csd_lock_suppress_rcu_stall= [KNL] + Do only a one-line RCU CPU stall warning when + there is an ongoing too-long CSD-lock wait. + rcutree.do_rcu_barrier= [KNL] Request a call to rcu_barrier(). This is throttled so that userspace tests can safely @@ -5384,7 +5388,13 @@ Time to wait (s) after boot before inducing stall. rcutorture.stall_cpu_irqsoff= [KNL] - Disable interrupts while stalling if set. + Disable interrupts while stalling if set, but only + on the first stall in the set. + + rcutorture.stall_cpu_repeat= [KNL] + Number of times to repeat the stall sequence, + so that rcutorture.stall_cpu_repeat=3 will result + in four stall sequences. rcutorture.stall_gp_kthread= [KNL] Duration (s) of forced sleep within RCU @@ -5572,14 +5582,6 @@ of zero will disable batching. Batching is always disabled for synchronize_rcu_tasks(). - rcupdate.rcu_tasks_rude_lazy_ms= [KNL] - Set timeout in milliseconds RCU Tasks - Rude asynchronous callback batching for - call_rcu_tasks_rude(). A negative value - will take the default. A value of zero will - disable batching. Batching is always disabled - for synchronize_rcu_tasks_rude(). - rcupdate.rcu_tasks_trace_lazy_ms= [KNL] Set timeout in milliseconds RCU Tasks Trace asynchronous callback batching for diff --git a/include/linux/rcu_segcblist.h b/include/linux/rcu_segcblist.h index ba95c06675e1..2fdc2208f1ca 100644 --- a/include/linux/rcu_segcblist.h +++ b/include/linux/rcu_segcblist.h @@ -185,11 +185,7 @@ struct rcu_cblist { * ---------------------------------------------------------------------------- */ #define SEGCBLIST_ENABLED BIT(0) -#define SEGCBLIST_RCU_CORE BIT(1) -#define SEGCBLIST_LOCKING BIT(2) -#define SEGCBLIST_KTHREAD_CB BIT(3) -#define SEGCBLIST_KTHREAD_GP BIT(4) -#define SEGCBLIST_OFFLOADED BIT(5) +#define SEGCBLIST_OFFLOADED BIT(1) struct rcu_segcblist { struct rcu_head *head; diff --git a/include/linux/rculist.h b/include/linux/rculist.h index 3dc1e58865f7..14dfa6008467 100644 --- a/include/linux/rculist.h +++ b/include/linux/rculist.h @@ -191,7 +191,10 @@ static inline void hlist_del_init_rcu(struct hlist_node *n) * @old : the element to be replaced * @new : the new element to insert * - * The @old entry will be replaced with the @new entry atomically. + * The @old entry will be replaced with the @new entry atomically from + * the perspective of concurrent readers. It is the caller's responsibility + * to synchronize with concurrent updaters, if any. + * * Note: @old should not be empty. */ static inline void list_replace_rcu(struct list_head *old, @@ -519,7 +522,9 @@ static inline void hlist_del_rcu(struct hlist_node *n) * @old : the element to be replaced * @new : the new element to insert * - * The @old entry will be replaced with the @new entry atomically. + * The @old entry will be replaced with the @new entry atomically from + * the perspective of concurrent readers. It is the caller's responsibility + * to synchronize with concurrent updaters, if any. */ static inline void hlist_replace_rcu(struct hlist_node *old, struct hlist_node *new) diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h index 13f6f00aecf9..58d84c59f3dd 100644 --- a/include/linux/rcupdate.h +++ b/include/linux/rcupdate.h @@ -34,10 +34,12 @@ #define ULONG_CMP_GE(a, b) (ULONG_MAX / 2 >= (a) - (b)) #define ULONG_CMP_LT(a, b) (ULONG_MAX / 2 < (a) - (b)) +#define RCU_SEQ_CTR_SHIFT 2 +#define RCU_SEQ_STATE_MASK ((1 << RCU_SEQ_CTR_SHIFT) - 1) + /* Exported common interfaces */ void call_rcu(struct rcu_head *head, rcu_callback_t func); void rcu_barrier_tasks(void); -void rcu_barrier_tasks_rude(void); void synchronize_rcu(void); struct rcu_gp_oldstate; @@ -144,11 +146,18 @@ void rcu_init_nohz(void); int rcu_nocb_cpu_offload(int cpu); int rcu_nocb_cpu_deoffload(int cpu); void rcu_nocb_flush_deferred_wakeup(void); + +#define RCU_NOCB_LOCKDEP_WARN(c, s) RCU_LOCKDEP_WARN(c, s) + #else /* #ifdef CONFIG_RCU_NOCB_CPU */ + static inline void rcu_init_nohz(void) { } static inline int rcu_nocb_cpu_offload(int cpu) { return -EINVAL; } static inline int rcu_nocb_cpu_deoffload(int cpu) { return 0; } static inline void rcu_nocb_flush_deferred_wakeup(void) { } + +#define RCU_NOCB_LOCKDEP_WARN(c, s) + #endif /* #else #ifdef CONFIG_RCU_NOCB_CPU */ /* @@ -165,6 +174,7 @@ static inline void rcu_nocb_flush_deferred_wakeup(void) { } } while (0) void call_rcu_tasks(struct rcu_head *head, rcu_callback_t func); void synchronize_rcu_tasks(void); +void rcu_tasks_torture_stats_print(char *tt, char *tf); # else # define rcu_tasks_classic_qs(t, preempt) do { } while (0) # define call_rcu_tasks call_rcu @@ -191,6 +201,7 @@ void rcu_tasks_trace_qs_blkd(struct task_struct *t); rcu_tasks_trace_qs_blkd(t); \ } \ } while (0) +void rcu_tasks_trace_torture_stats_print(char *tt, char *tf); # else # define rcu_tasks_trace_qs(t) do { } while (0) # endif @@ -202,8 +213,8 @@ do { \ } while (0) # ifdef CONFIG_TASKS_RUDE_RCU -void call_rcu_tasks_rude(struct rcu_head *head, rcu_callback_t func); void synchronize_rcu_tasks_rude(void); +void rcu_tasks_rude_torture_stats_print(char *tt, char *tf); # endif #define rcu_note_voluntary_context_switch(t) rcu_tasks_qs(t, false) diff --git a/include/linux/smp.h b/include/linux/smp.h index fcd61dfe2af3..3871bd32018f 100644 --- a/include/linux/smp.h +++ b/include/linux/smp.h @@ -294,4 +294,10 @@ int smpcfd_prepare_cpu(unsigned int cpu); int smpcfd_dead_cpu(unsigned int cpu); int smpcfd_dying_cpu(unsigned int cpu); +#ifdef CONFIG_CSD_LOCK_WAIT_DEBUG +bool csd_lock_is_stuck(void); +#else +static inline bool csd_lock_is_stuck(void) { return false; } +#endif + #endif /* __LINUX_SMP_H */ diff --git a/include/linux/srcutree.h b/include/linux/srcutree.h index 8f3f72480e78..ed57598394de 100644 --- a/include/linux/srcutree.h +++ b/include/linux/srcutree.h @@ -129,10 +129,23 @@ struct srcu_struct { #define SRCU_STATE_SCAN1 1 #define SRCU_STATE_SCAN2 2 +/* + * Values for initializing gp sequence fields. Higher values allow wrap arounds to + * occur earlier. + * The second value with state is useful in the case of static initialization of + * srcu_usage where srcu_gp_seq_needed is expected to have some state value in its + * lower bits (or else it will appear to be already initialized within + * the call check_init_srcu_struct()). + */ +#define SRCU_GP_SEQ_INITIAL_VAL ((0UL - 100UL) << RCU_SEQ_CTR_SHIFT) +#define SRCU_GP_SEQ_INITIAL_VAL_WITH_STATE (SRCU_GP_SEQ_INITIAL_VAL - 1) + #define __SRCU_USAGE_INIT(name) \ { \ .lock = __SPIN_LOCK_UNLOCKED(name.lock), \ - .srcu_gp_seq_needed = -1UL, \ + .srcu_gp_seq = SRCU_GP_SEQ_INITIAL_VAL, \ + .srcu_gp_seq_needed = SRCU_GP_SEQ_INITIAL_VAL_WITH_STATE, \ + .srcu_gp_seq_needed_exp = SRCU_GP_SEQ_INITIAL_VAL, \ .work = __DELAYED_WORK_INITIALIZER(name.work, NULL, 0), \ } diff --git a/kernel/rcu/rcu.h b/kernel/rcu/rcu.h index 5564402af4cb..feb3ac1dc5d5 100644 --- a/kernel/rcu/rcu.h +++ b/kernel/rcu/rcu.h @@ -54,9 +54,6 @@ * grace-period sequence number. */ -#define RCU_SEQ_CTR_SHIFT 2 -#define RCU_SEQ_STATE_MASK ((1 << RCU_SEQ_CTR_SHIFT) - 1) - /* Low-order bit definition for polled grace-period APIs. */ #define RCU_GET_STATE_COMPLETED 0x1 @@ -255,6 +252,11 @@ static inline void debug_rcu_head_callback(struct rcu_head *rhp) kmem_dump_obj(rhp); } +static inline bool rcu_barrier_cb_is_done(struct rcu_head *rhp) +{ + return rhp->next == rhp; +} + extern int rcu_cpu_stall_suppress_at_boot; static inline bool rcu_stall_is_suppressed_at_boot(void) diff --git a/kernel/rcu/rcu_segcblist.c b/kernel/rcu/rcu_segcblist.c index 1693ea22ef1b..298a2c573f02 100644 --- a/kernel/rcu/rcu_segcblist.c +++ b/kernel/rcu/rcu_segcblist.c @@ -260,17 +260,6 @@ void rcu_segcblist_disable(struct rcu_segcblist *rsclp) rcu_segcblist_clear_flags(rsclp, SEGCBLIST_ENABLED); } -/* - * Mark the specified rcu_segcblist structure as offloaded (or not) - */ -void rcu_segcblist_offload(struct rcu_segcblist *rsclp, bool offload) -{ - if (offload) - rcu_segcblist_set_flags(rsclp, SEGCBLIST_LOCKING | SEGCBLIST_OFFLOADED); - else - rcu_segcblist_clear_flags(rsclp, SEGCBLIST_OFFLOADED); -} - /* * Does the specified rcu_segcblist structure contain callbacks that * are ready to be invoked? diff --git a/kernel/rcu/rcu_segcblist.h b/kernel/rcu/rcu_segcblist.h index 4fe877f5f654..259904075636 100644 --- a/kernel/rcu/rcu_segcblist.h +++ b/kernel/rcu/rcu_segcblist.h @@ -89,16 +89,7 @@ static inline bool rcu_segcblist_is_enabled(struct rcu_segcblist *rsclp) static inline bool rcu_segcblist_is_offloaded(struct rcu_segcblist *rsclp) { if (IS_ENABLED(CONFIG_RCU_NOCB_CPU) && - rcu_segcblist_test_flags(rsclp, SEGCBLIST_LOCKING)) - return true; - - return false; -} - -static inline bool rcu_segcblist_completely_offloaded(struct rcu_segcblist *rsclp) -{ - if (IS_ENABLED(CONFIG_RCU_NOCB_CPU) && - !rcu_segcblist_test_flags(rsclp, SEGCBLIST_RCU_CORE)) + rcu_segcblist_test_flags(rsclp, SEGCBLIST_OFFLOADED)) return true; return false; diff --git a/kernel/rcu/rcuscale.c b/kernel/rcu/rcuscale.c index b53a9e8f5904..6d37596deb1f 100644 --- a/kernel/rcu/rcuscale.c +++ b/kernel/rcu/rcuscale.c @@ -39,6 +39,7 @@ #include <linux/torture.h> #include <linux/vmalloc.h> #include <linux/rcupdate_trace.h> +#include <linux/sched/debug.h> #include "rcu.h" @@ -104,6 +105,20 @@ static char *scale_type = "rcu"; module_param(scale_type, charp, 0444); MODULE_PARM_DESC(scale_type, "Type of RCU to scalability-test (rcu, srcu, ...)"); +// Structure definitions for custom fixed-per-task allocator. +struct writer_mblock { + struct rcu_head wmb_rh; + struct llist_node wmb_node; + struct writer_freelist *wmb_wfl; +}; + +struct writer_freelist { + struct llist_head ws_lhg; + atomic_t ws_inflight; + struct llist_head ____cacheline_internodealigned_in_smp ws_lhp; + struct writer_mblock *ws_mblocks; +}; + static int nrealreaders; static int nrealwriters; static struct task_struct **writer_tasks; @@ -111,6 +126,8 @@ static struct task_struct **reader_tasks; static struct task_struct *shutdown_task; static u64 **writer_durations; +static bool *writer_done; +static struct writer_freelist *writer_freelists; static int *writer_n_durations; static atomic_t n_rcu_scale_reader_started; static atomic_t n_rcu_scale_writer_started; @@ -120,7 +137,6 @@ static u64 t_rcu_scale_writer_started; static u64 t_rcu_scale_writer_finished; static unsigned long b_rcu_gp_test_started; static unsigned long b_rcu_gp_test_finished; -static DEFINE_PER_CPU(atomic_t, n_async_inflight); #define MAX_MEAS 10000 #define MIN_MEAS 100 @@ -143,6 +159,7 @@ struct rcu_scale_ops { void (*sync)(void); void (*exp_sync)(void); struct task_struct *(*rso_gp_kthread)(void); + void (*stats)(void); const char *name; }; @@ -224,6 +241,11 @@ static void srcu_scale_synchronize(void) synchronize_srcu(srcu_ctlp); } +static void srcu_scale_stats(void) +{ + srcu_torture_stats_print(srcu_ctlp, scale_type, SCALE_FLAG); +} + static void srcu_scale_synchronize_expedited(void) { synchronize_srcu_expedited(srcu_ctlp); @@ -241,6 +263,7 @@ static struct rcu_scale_ops srcu_ops = { .gp_barrier = srcu_rcu_barrier, .sync = srcu_scale_synchronize, .exp_sync = srcu_scale_synchronize_expedited, + .stats = srcu_scale_stats, .name = "srcu" }; @@ -270,6 +293,7 @@ static struct rcu_scale_ops srcud_ops = { .gp_barrier = srcu_rcu_barrier, .sync = srcu_scale_synchronize, .exp_sync = srcu_scale_synchronize_expedited, + .stats = srcu_scale_stats, .name = "srcud" }; @@ -288,6 +312,11 @@ static void tasks_scale_read_unlock(int idx) { } +static void rcu_tasks_scale_stats(void) +{ + rcu_tasks_torture_stats_print(scale_type, SCALE_FLAG); +} + static struct rcu_scale_ops tasks_ops = { .ptype = RCU_TASKS_FLAVOR, .init = rcu_sync_scale_init, @@ -300,6 +329,7 @@ static struct rcu_scale_ops tasks_ops = { .sync = synchronize_rcu_tasks, .exp_sync = synchronize_rcu_tasks, .rso_gp_kthread = get_rcu_tasks_gp_kthread, + .stats = IS_ENABLED(CONFIG_TINY_RCU) ? NULL : rcu_tasks_scale_stats, .name = "tasks" }; @@ -326,6 +356,11 @@ static void tasks_rude_scale_read_unlock(int idx) { } +static void rcu_tasks_rude_scale_stats(void) +{ + rcu_tasks_rude_torture_stats_print(scale_type, SCALE_FLAG); +} + static struct rcu_scale_ops tasks_rude_ops = { .ptype = RCU_TASKS_RUDE_FLAVOR, .init = rcu_sync_scale_init, @@ -333,11 +368,10 @@ static struct rcu_scale_ops tasks_rude_ops = { .readunlock = tasks_rude_scale_read_unlock, .get_gp_seq = rcu_no_completed, .gp_diff = rcu_seq_diff, - .async = call_rcu_tasks_rude, - .gp_barrier = rcu_barrier_tasks_rude, .sync = synchronize_rcu_tasks_rude, .exp_sync = synchronize_rcu_tasks_rude, .rso_gp_kthread = get_rcu_tasks_rude_gp_kthread, + .stats = IS_ENABLED(CONFIG_TINY_RCU) ? NULL : rcu_tasks_rude_scale_stats, .name = "tasks-rude" }; @@ -366,6 +400,11 @@ static void tasks_trace_scale_read_unlock(int idx) rcu_read_unlock_trace(); } +static void rcu_tasks_trace_scale_stats(void) +{ + rcu_tasks_trace_torture_stats_print(scale_type, SCALE_FLAG); +} + static struct rcu_scale_ops tasks_tracing_ops = { .ptype = RCU_TASKS_FLAVOR, .init = rcu_sync_scale_init, @@ -378,6 +417,7 @@ static struct rcu_scale_ops tasks_tracing_ops = { .sync = synchronize_rcu_tasks_trace, .exp_sync = synchronize_rcu_tasks_trace, .rso_gp_kthread = get_rcu_tasks_trace_gp_kthread, + .stats = IS_ENABLED(CONFIG_TINY_RCU) ? NULL : rcu_tasks_trace_scale_stats, .name = "tasks-tracing" }; @@ -437,13 +477,53 @@ rcu_scale_reader(void *arg) return 0; } +/* + * Allocate a writer_mblock structure for the specified rcu_scale_writer + * task. + */ +static struct writer_mblock *rcu_scale_alloc(long me) +{ + struct llist_node *llnp; + struct writer_freelist *wflp; + struct writer_mblock *wmbp; + + if (WARN_ON_ONCE(!writer_freelists)) + return NULL; + wflp = &writer_freelists[me]; + if (llist_empty(&wflp->ws_lhp)) { + // ->ws_lhp is private to its rcu_scale_writer task. + wmbp = container_of(llist_del_all(&wflp->ws_lhg), struct writer_mblock, wmb_node); + wflp->ws_lhp.first = &wmbp->wmb_node; + } + llnp = llist_del_first(&wflp->ws_lhp); + if (!llnp) + return NULL; + return container_of(llnp, struct writer_mblock, wmb_node); +} + +/* + * Free a writer_mblock structure to its rcu_scale_writer task. + */ +static void rcu_scale_free(struct writer_mblock *wmbp) +{ + struct writer_freelist *wflp; + + if (!wmbp) + return; + wflp = wmbp->wmb_wfl; + llist_add(&wmbp->wmb_node, &wflp->ws_lhg); +} + /* * Callback function for asynchronous grace periods from rcu_scale_writer(). */ static void rcu_scale_async_cb(struct rcu_head *rhp) { - atomic_dec(this_cpu_ptr(&n_async_inflight)); - kfree(rhp); + struct writer_mblock *wmbp = container_of(rhp, struct writer_mblock, wmb_rh); + struct writer_freelist *wflp = wmbp->wmb_wfl; + + atomic_dec(&wflp->ws_inflight); + rcu_scale_free(wmbp); } /* @@ -456,12 +536,14 @@ rcu_scale_writer(void *arg) int i_max; unsigned long jdone; long me = (long)arg; - struct rcu_head *rhp = NULL; + bool selfreport = false; bool started = false, done = false, alldone = false; u64 t; DEFINE_TORTURE_RANDOM(tr); u64 *wdp; u64 *wdpp = writer_durations[me]; + struct writer_freelist *wflp = &writer_freelists[me]; + struct writer_mblock *wmbp = NULL; VERBOSE_SCALEOUT_STRING("rcu_scale_writer task started"); WARN_ON(!wdpp); @@ -493,30 +575,34 @@ rcu_scale_writer(void *arg) jdone = jiffies + minruntime * HZ; do { + bool gp_succeeded = false; + if (writer_holdoff) udelay(writer_holdoff); if (writer_holdoff_jiffies) schedule_timeout_idle(torture_random(&tr) % writer_holdoff_jiffies + 1); wdp = &wdpp[i]; *wdp = ktime_get_mono_fast_ns(); - if (gp_async) { -retry: - if (!rhp) - rhp = kmalloc(sizeof(*rhp), GFP_KERNEL); - if (rhp && atomic_read(this_cpu_ptr(&n_async_inflight)) < gp_async_max) { - atomic_inc(this_cpu_ptr(&n_async_inflight)); - cur_ops->async(rhp, rcu_scale_async_cb); - rhp = NULL; + if (gp_async && !WARN_ON_ONCE(!cur_ops->async)) { + if (!wmbp) + wmbp = rcu_scale_alloc(me); + if (wmbp && atomic_read(&wflp->ws_inflight) < gp_async_max) { + atomic_inc(&wflp->ws_inflight); + cur_ops->async(&wmbp->wmb_rh, rcu_scale_async_cb); + wmbp = NULL; + gp_succeeded = true; } else if (!kthread_should_stop()) { cur_ops->gp_barrier(); - goto retry; } else { - kfree(rhp); /* Because we are stopping. */ + rcu_scale_free(wmbp); /* Because we are stopping. */ + wmbp = NULL; } } else if (gp_exp) { cur_ops->exp_sync(); + gp_succeeded = true; } else { cur_ops->sync(); + gp_succeeded = true; } t = ktime_get_mono_fast_ns(); *wdp = t - *wdp; @@ -526,6 +612,7 @@ retry: started = true; if (!done && i >= MIN_MEAS && time_after(jiffies, jdone)) { done = true; + WRITE_ONCE(writer_done[me], true); sched_set_normal(current, 0); pr_alert("%s%s rcu_scale_writer %ld has %d measurements\n", scale_type, SCALE_FLAG, me, MIN_MEAS); @@ -551,11 +638,32 @@ retry: if (done && !alldone && atomic_read(&n_rcu_scale_writer_finished) >= nrealwriters) alldone = true; - if (started && !alldone && i < MAX_MEAS - 1) + if (done && !alldone && time_after(jiffies, jdone + HZ * 60)) { + static atomic_t dumped; + int i; + + if (!atomic_xchg(&dumped, 1)) { + for (i = 0; i < nrealwriters; i++) { + if (writer_done[i]) + continue; + pr_info("%s: Task %ld flags writer %d:\n", __func__, me, i); + sched_show_task(writer_tasks[i]); + } + if (cur_ops->stats) + cur_ops->stats(); + } + } + if (!selfreport && time_after(jiffies, jdone + HZ * (70 + me))) { + pr_info("%s: Writer %ld self-report: started %d done %d/%d->%d i %d jdone %lu.\n", + __func__, me, started, done, writer_done[me], atomic_read(&n_rcu_scale_writer_finished), i, jiffies - jdone); + selfreport = true; + } + if (gp_succeeded && started && !alldone && i < MAX_MEAS - 1) i++; rcu_scale_wait_shutdown(); } while (!torture_must_stop()); - if (gp_async) { + if (gp_async && cur_ops->async) { + rcu_scale_free(wmbp); cur_ops->gp_barrier(); } writer_n_durations[me] = i_max + 1; @@ -713,6 +821,7 @@ kfree_scale_cleanup(void) torture_stop_kthread(kfree_scale_thread, kfree_reader_tasks[i]); kfree(kfree_reader_tasks); + kfree_reader_tasks = NULL; } torture_cleanup_end(); @@ -881,6 +990,7 @@ rcu_scale_cleanup(void) torture_stop_kthread(rcu_scale_reader, reader_tasks[i]); kfree(reader_tasks); + reader_tasks = NULL; } if (writer_tasks) { @@ -919,10 +1029,33 @@ rcu_scale_cleanup(void) schedule_timeout_uninterruptible(1); } kfree(writer_durations[i]); + if (writer_freelists) { + int ctr = 0; + struct llist_node *llnp; + struct writer_freelist *wflp = &writer_freelists[i]; + + if (wflp->ws_mblocks) { + llist_for_each(llnp, wflp->ws_lhg.first) + ctr++; + llist_for_each(llnp, wflp->ws_lhp.first) + ctr++; + WARN_ONCE(ctr != gp_async_max, + "%s: ctr = %d gp_async_max = %d\n", + __func__, ctr, gp_async_max); + kfree(wflp->ws_mblocks); + } + } } kfree(writer_tasks); + writer_tasks = NULL; kfree(writer_durations); + writer_durations = NULL; kfree(writer_n_durations); + writer_n_durations = NULL; + kfree(writer_done); + writer_done = NULL; + kfree(writer_freelists); + writer_freelists = NULL; } /* Do torture-type-specific cleanup operations. */ @@ -949,8 +1082,9 @@ rcu_scale_shutdown(void *arg) static int __init rcu_scale_init(void) { - long i; int firsterr = 0; + long i; + long j; static struct rcu_scale_ops *scale_ops[] = { &rcu_ops, &srcu_ops, &srcud_ops, TASKS_OPS TASKS_RUDE_OPS TASKS_TRACING_OPS }; @@ -1017,14 +1151,22 @@ rcu_scale_init(void) } while (atomic_read(&n_rcu_scale_reader_started) < nrealreaders) schedule_timeout_uninterruptible(1); - writer_tasks = kcalloc(nrealwriters, sizeof(reader_tasks[0]), - GFP_KERNEL); - writer_durations = kcalloc(nrealwriters, sizeof(*writer_durations), - GFP_KERNEL); - writer_n_durations = - kcalloc(nrealwriters, sizeof(*writer_n_durations), - GFP_KERNEL); - if (!writer_tasks || !writer_durations || !writer_n_durations) { + writer_tasks = kcalloc(nrealwriters, sizeof(writer_tasks[0]), GFP_KERNEL); + writer_durations = kcalloc(nrealwriters, sizeof(*writer_durations), GFP_KERNEL); + writer_n_durations = kcalloc(nrealwriters, sizeof(*writer_n_durations), GFP_KERNEL); + writer_done = kcalloc(nrealwriters, sizeof(writer_done[0]), GFP_KERNEL); + if (gp_async) { + if (gp_async_max <= 0) { + pr_warn("%s: gp_async_max = %d must be greater than zero.\n", + __func__, gp_async_max); + WARN_ON_ONCE(IS_BUILTIN(CONFIG_RCU_TORTURE_TEST)); + firsterr = -EINVAL; + goto unwind; + } + writer_freelists = kcalloc(nrealwriters, sizeof(writer_freelists[0]), GFP_KERNEL); + } + if (!writer_tasks || !writer_durations || !writer_n_durations || !writer_done || + (gp_async && !writer_freelists)) { SCALEOUT_ERRSTRING("out of memory"); firsterr = -ENOMEM; goto unwind; @@ -1037,6 +1179,24 @@ rcu_scale_init(void) firsterr = -ENOMEM; goto unwind; } + if (writer_freelists) { + struct writer_freelist *wflp = &writer_freelists[i]; + + init_llist_head(&wflp->ws_lhg); + init_llist_head(&wflp->ws_lhp); + wflp->ws_mblocks = kcalloc(gp_async_max, sizeof(wflp->ws_mblocks[0]), + GFP_KERNEL); + if (!wflp->ws_mblocks) { + firsterr = -ENOMEM; + goto unwind; + } + for (j = 0; j < gp_async_max; j++) { + struct writer_mblock *wmbp = &wflp->ws_mblocks[j]; + + wmbp->wmb_wfl = wflp; + llist_add(&wmbp->wmb_node, &wflp->ws_lhp); + } + } firsterr = torture_create_kthread(rcu_scale_writer, (void *)i, writer_tasks[i]); if (torture_init_error(firsterr)) diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c index ef757c3e84a2..bb75dbf5c800 100644 --- a/kernel/rcu/rcutorture.c +++ b/kernel/rcu/rcutorture.c @@ -115,6 +115,7 @@ torture_param(int, stall_cpu_holdoff, 10, "Time to wait before starting stall (s torture_param(bool, stall_no_softlockup, false, "Avoid softlockup warning during cpu stall."); torture_param(int, stall_cpu_irqsoff, 0, "Disable interrupts while stalling."); torture_param(int, stall_cpu_block, 0, "Sleep while stalling."); +torture_param(int, stall_cpu_repeat, 0, "Number of additional stalls after the first one."); torture_param(int, stall_gp_kthread, 0, "Grace-period kthread stall duration (s)."); torture_param(int, stat_interval, 60, "Number of seconds between stats printk()s"); torture_param(int, stutter, 5, "Number of seconds to run/halt test"); @@ -366,8 +367,6 @@ struct rcu_torture_ops { bool (*same_gp_state_full)(struct rcu_gp_oldstate *rgosp1, struct rcu_gp_oldstate *rgosp2); unsigned long (*get_gp_state)(void); void (*get_gp_state_full)(struct rcu_gp_oldstate *rgosp); - unsigned long (*get_gp_completed)(void); - void (*get_gp_completed_full)(struct rcu_gp_oldstate *rgosp); unsigned long (*start_gp_poll)(void); void (*start_gp_poll_full)(struct rcu_gp_oldstate *rgosp); bool (*poll_gp_state)(unsigned long oldstate); @@ -375,6 +374,8 @@ struct rcu_torture_ops { bool (*poll_need_2gp)(bool poll, bool poll_full); void (*cond_sync)(unsigned long oldstate); void (*cond_sync_full)(struct rcu_gp_oldstate *rgosp); + int poll_active; + int poll_active_full; call_rcu_func_t call; void (*cb_barrier)(void); void (*fqs)(void); @@ -553,8 +554,6 @@ static struct rcu_torture_ops rcu_ops = { .get_comp_state_full = get_completed_synchronize_rcu_full, .get_gp_state = get_state_synchronize_rcu, .get_gp_state_full = get_state_synchronize_rcu_full, - .get_gp_completed = get_completed_synchronize_rcu, - .get_gp_completed_full = get_completed_synchronize_rcu_full, .start_gp_poll = start_poll_synchronize_rcu, .start_gp_poll_full = start_poll_synchronize_rcu_full, .poll_gp_state = poll_state_synchronize_rcu, @@ -562,6 +561,8 @@ static struct rcu_torture_ops rcu_ops = { .poll_need_2gp = rcu_poll_need_2gp, .cond_sync = cond_synchronize_rcu, .cond_sync_full = cond_synchronize_rcu_full, + .poll_active = NUM_ACTIVE_RCU_POLL_OLDSTATE, + .poll_active_full = NUM_ACTIVE_RCU_POLL_FULL_OLDSTATE, .get_gp_state_exp = get_state_synchronize_rcu, .start_gp_poll_exp = start_poll_synchronize_rcu_expedited, .start_gp_poll_exp_full = start_poll_synchronize_rcu_expedited_full, @@ -740,9 +741,12 @@ static struct rcu_torture_ops srcu_ops = { .deferred_free = srcu_torture_deferred_free, .sync = srcu_torture_synchronize, .exp_sync = srcu_torture_synchronize_expedited, + .same_gp_state = same_state_synchronize_srcu, + .get_comp_state = get_completed_synchronize_srcu, .get_gp_state = srcu_torture_get_gp_state, .start_gp_poll = srcu_torture_start_gp_poll, .poll_gp_state = srcu_torture_poll_gp_state, + .poll_active = NUM_ACTIVE_SRCU_POLL_OLDSTATE, .call = srcu_torture_call, .cb_barrier = srcu_torture_barrier, .stats = srcu_torture_stats, @@ -780,9 +784,12 @@ static struct rcu_torture_ops srcud_ops = { .deferred_free = srcu_torture_deferred_free, .sync = srcu_torture_synchronize, .exp_sync = srcu_torture_synchronize_expedited, + .same_gp_state = same_state_synchronize_srcu, + .get_comp_state = get_completed_synchronize_srcu, .get_gp_state = srcu_torture_get_gp_state, .start_gp_poll = srcu_torture_start_gp_poll, .poll_gp_state = srcu_torture_poll_gp_state, + .poll_active = NUM_ACTIVE_SRCU_POLL_OLDSTATE, .call = srcu_torture_call, .cb_barrier = srcu_torture_barrier, .stats = srcu_torture_stats, @@ -915,11 +922,6 @@ static struct rcu_torture_ops tasks_ops = { * Definitions for rude RCU-tasks torture testing. */ -static void rcu_tasks_rude_torture_deferred_free(struct rcu_torture *p) -{ - call_rcu_tasks_rude(&p->rtort_rcu, rcu_torture_cb); -} - static struct rcu_torture_ops tasks_rude_ops = { .ttype = RCU_TASKS_RUDE_FLAVOR, .init = rcu_sync_torture_init, @@ -927,11 +929,8 @@ static struct rcu_torture_ops tasks_rude_ops = { .read_delay = rcu_read_delay, /* just reuse rcu's version. */ .readunlock = rcu_torture_read_unlock_trivial, .get_gp_seq = rcu_no_completed, - .deferred_free = rcu_tasks_rude_torture_deferred_free, .sync = synchronize_rcu_tasks_rude, .exp_sync = synchronize_rcu_tasks_rude, - .call = call_rcu_tasks_rude, - .cb_barrier = rcu_barrier_tasks_rude, .gp_kthread_dbg = show_rcu_tasks_rude_gp_kthread, .get_gp_data = rcu_tasks_rude_get_gp_data, .cbflood_max = 50000, @@ -1318,6 +1317,7 @@ static void rcu_torture_write_types(void) } else if (gp_sync && !cur_ops->sync) { pr_alert("%s: gp_sync without primitives.\n", __func__); } + pr_alert("%s: Testing %d update types.\n", __func__, nsynctypes); } /* @@ -1374,17 +1374,20 @@ rcu_torture_writer(void *arg) int i; int idx; int oldnice = task_nice(current); - struct rcu_gp_oldstate rgo[NUM_ACTIVE_RCU_POLL_FULL_OLDSTATE]; + struct rcu_gp_oldstate *rgo = NULL; + int rgo_size = 0; struct rcu_torture *rp; struct rcu_torture *old_rp; static DEFINE_TORTURE_RANDOM(rand); unsigned long stallsdone = jiffies; bool stutter_waited; - unsigned long ulo[NUM_ACTIVE_RCU_POLL_OLDSTATE]; + unsigned long *ulo = NULL; + int ulo_size = 0; // If a new stall test is added, this must be adjusted. if (stall_cpu_holdoff + stall_gp_kthread + stall_cpu) - stallsdone += (stall_cpu_holdoff + stall_gp_kthread + stall_cpu + 60) * HZ; + stallsdone += (stall_cpu_holdoff + stall_gp_kthread + stall_cpu + 60) * + HZ * (stall_cpu_repeat + 1); VERBOSE_TOROUT_STRING("rcu_torture_writer task started"); if (!can_expedite) pr_alert("%s" TORTURE_FLAG @@ -1401,6 +1404,16 @@ rcu_torture_writer(void *arg) torture_kthread_stopping("rcu_torture_writer"); return 0; } + if (cur_ops->poll_active > 0) { + ulo = kzalloc(cur_ops->poll_active * sizeof(ulo[0]), GFP_KERNEL); + if (!WARN_ON(!ulo)) + ulo_size = cur_ops->poll_active; + } + if (cur_ops->poll_active_full > 0) { + rgo = kzalloc(cur_ops->poll_active_full * sizeof(rgo[0]), GFP_KERNEL); + if (!WARN_ON(!rgo)) + rgo_size = cur_ops->poll_active_full; + } do { rcu_torture_writer_state = RTWS_FIXED_DELAY; @@ -1437,8 +1450,8 @@ rcu_torture_writer(void *arg) rcu_torture_writer_state_getname(), rcu_torture_writer_state, cookie, cur_ops->get_gp_state()); - if (cur_ops->get_gp_completed) { - cookie = cur_ops->get_gp_completed(); + if (cur_ops->get_comp_state) { + cookie = cur_ops->get_comp_state(); WARN_ON_ONCE(!cur_ops->poll_gp_state(cookie)); } cur_ops->readunlock(idx); @@ -1452,8 +1465,8 @@ rcu_torture_writer(void *arg) rcu_torture_writer_state_getname(), rcu_torture_writer_state, cpumask_pr_args(cpu_online_mask)); - if (cur_ops->get_gp_completed_full) { - cur_ops->get_gp_completed_full(&cookie_full); + if (cur_ops->get_comp_state_full) { + cur_ops->get_comp_state_full(&cookie_full); WARN_ON_ONCE(!cur_ops->poll_gp_state_full(&cookie_full)); } cur_ops->readunlock(idx); @@ -1502,19 +1515,19 @@ rcu_torture_writer(void *arg) break; case RTWS_POLL_GET: rcu_torture_writer_state = RTWS_POLL_GET; - for (i = 0; i < ARRAY_SIZE(ulo); i++) + for (i = 0; i < ulo_size; i++) ulo[i] = cur_ops->get_comp_state(); gp_snap = cur_ops->start_gp_poll(); rcu_torture_writer_state = RTWS_POLL_WAIT; while (!cur_ops->poll_gp_state(gp_snap)) { gp_snap1 = cur_ops->get_gp_state(); - for (i = 0; i < ARRAY_SIZE(ulo); i++) + for (i = 0; i < ulo_size; i++) if (cur_ops->poll_gp_state(ulo[i]) || cur_ops->same_gp_state(ulo[i], gp_snap1)) { ulo[i] = gp_snap1; break; } - WARN_ON_ONCE(i >= ARRAY_SIZE(ulo)); + WARN_ON_ONCE(ulo_size > 0 && i >= ulo_size); torture_hrtimeout_jiffies(torture_random(&rand) % 16, &rand); } @@ -1522,20 +1535,20 @@ rcu_torture_writer(void *arg) break; case RTWS_POLL_GET_FULL: rcu_torture_writer_state = RTWS_POLL_GET_FULL; - for (i = 0; i < ARRAY_SIZE(rgo); i++) + for (i = 0; i < rgo_size; i++) cur_ops->get_comp_state_full(&rgo[i]); cur_ops->start_gp_poll_full(&gp_snap_full); rcu_torture_writer_state = RTWS_POLL_WAIT_FULL; while (!cur_ops->poll_gp_state_full(&gp_snap_full)) { cur_ops->get_gp_state_full(&gp_snap1_full); - for (i = 0; i < ARRAY_SIZE(rgo); i++) + for (i = 0; i < rgo_size; i++) if (cur_ops->poll_gp_state_full(&rgo[i]) || cur_ops->same_gp_state_full(&rgo[i], &gp_snap1_full)) { rgo[i] = gp_snap1_full; break; } - WARN_ON_ONCE(i >= ARRAY_SIZE(rgo)); + WARN_ON_ONCE(rgo_size > 0 && i >= rgo_size); torture_hrtimeout_jiffies(torture_random(&rand) % 16, &rand); } @@ -1617,6 +1630,8 @@ rcu_torture_writer(void *arg) pr_alert("%s" TORTURE_FLAG " Dynamic grace-period expediting was disabled.\n", torture_type); + kfree(ulo); + kfree(rgo); rcu_torture_writer_state = RTWS_STOPPING; torture_kthread_stopping("rcu_torture_writer"); return 0; @@ -2370,7 +2385,7 @@ rcu_torture_print_module_parms(struct rcu_torture_ops *cur_ops, const char *tag) "test_boost=%d/%d test_boost_interval=%d " "test_boost_duration=%d shutdown_secs=%d " "stall_cpu=%d stall_cpu_holdoff=%d stall_cpu_irqsoff=%d " - "stall_cpu_block=%d " + "stall_cpu_block=%d stall_cpu_repeat=%d " "n_barrier_cbs=%d " "onoff_interval=%d onoff_holdoff=%d " "read_exit_delay=%d read_exit_burst=%d " @@ -2382,7 +2397,7 @@ rcu_torture_print_module_parms(struct rcu_torture_ops *cur_ops, const char *tag) test_boost, cur_ops->can_boost, test_boost_interval, test_boost_duration, shutdown_secs, stall_cpu, stall_cpu_holdoff, stall_cpu_irqsoff, - stall_cpu_block, + stall_cpu_block, stall_cpu_repeat, n_barrier_cbs, onoff_interval, onoff_holdoff, read_exit_delay, read_exit_burst, @@ -2460,19 +2475,11 @@ static struct notifier_block rcu_torture_stall_block = { * induces a CPU stall for the time specified by stall_cpu. If a new * stall test is added, stallsdone in rcu_torture_writer() must be adjusted. */ -static int rcu_torture_stall(void *args) +static void rcu_torture_stall_one(int rep, int irqsoff) { int idx; - int ret; unsigned long stop_at; - VERBOSE_TOROUT_STRING("rcu_torture_stall task started"); - if (rcu_cpu_stall_notifiers) { - ret = rcu_stall_chain_notifier_register(&rcu_torture_stall_block); - if (ret) - pr_info("%s: rcu_stall_chain_notifier_register() returned %d, %sexpected.\n", - __func__, ret, !IS_ENABLED(CONFIG_RCU_STALL_COMMON) ? "un" : ""); - } if (stall_cpu_holdoff > 0) { VERBOSE_TOROUT_STRING("rcu_torture_stall begin holdoff"); schedule_timeout_interruptible(stall_cpu_holdoff * HZ); @@ -2492,12 +2499,12 @@ static int rcu_torture_stall(void *args) stop_at = ktime_get_seconds() + stall_cpu; /* RCU CPU stall is expected behavior in following code. */ idx = cur_ops->readlock(); - if (stall_cpu_irqsoff) + if (irqsoff) local_irq_disable(); else if (!stall_cpu_block) preempt_disable(); - pr_alert("%s start on CPU %d.\n", - __func__, raw_smp_processor_id()); + pr_alert("%s start stall episode %d on CPU %d.\n", + __func__, rep + 1, raw_smp_processor_id()); while (ULONG_CMP_LT((unsigned long)ktime_get_seconds(), stop_at) && !kthread_should_stop()) if (stall_cpu_block) { @@ -2509,12 +2516,42 @@ static int rcu_torture_stall(void *args) } else if (stall_no_softlockup) { touch_softlockup_watchdog(); } - if (stall_cpu_irqsoff) + if (irqsoff) local_irq_enable(); else if (!stall_cpu_block) preempt_enable(); cur_ops->readunlock(idx); } +} + +/* + * CPU-stall kthread. Invokes rcu_torture_stall_one() once, and then as many + * additional times as specified by the stall_cpu_repeat module parameter. + * Note that stall_cpu_irqsoff is ignored on the second and subsequent + * stall. + */ +static int rcu_torture_stall(void *args) +{ + int i; + int repeat = stall_cpu_repeat; + int ret; + + VERBOSE_TOROUT_STRING("rcu_torture_stall task started"); + if (repeat < 0) { + repeat = 0; + WARN_ON_ONCE(IS_BUILTIN(CONFIG_RCU_TORTURE_TEST)); + } + if (rcu_cpu_stall_notifiers) { + ret = rcu_stall_chain_notifier_register(&rcu_torture_stall_block); + if (ret) + pr_info("%s: rcu_stall_chain_notifier_register() returned %d, %sexpected.\n", + __func__, ret, !IS_ENABLED(CONFIG_RCU_STALL_COMMON) ? "un" : ""); + } + for (i = 0; i <= repeat; i++) { + if (kthread_should_stop()) + break; + rcu_torture_stall_one(i, i == 0 ? stall_cpu_irqsoff : 0); + } pr_alert("%s end.\n", __func__); if (rcu_cpu_stall_notifiers && !ret) { ret = rcu_stall_chain_notifier_unregister(&rcu_torture_stall_block); diff --git a/kernel/rcu/refscale.c b/kernel/rcu/refscale.c index f4ea5b1ec068..0db9db73f57f 100644 --- a/kernel/rcu/refscale.c +++ b/kernel/rcu/refscale.c @@ -28,6 +28,7 @@ #include <linux/rcupdate_trace.h> #include <linux/reboot.h> #include <linux/sched.h> +#include <linux/seq_buf.h> #include <linux/spinlock.h> #include <linux/smp.h> #include <linux/stat.h> @@ -134,7 +135,7 @@ struct ref_scale_ops { const char *name; }; -static struct ref_scale_ops *cur_ops; +static const struct ref_scale_ops *cur_ops; static void un_delay(const int udl, const int ndl) { @@ -170,7 +171,7 @@ static bool rcu_sync_scale_init(void) return true; } -static struct ref_scale_ops rcu_ops = { +static const struct ref_scale_ops rcu_ops = { .init = rcu_sync_scale_init, .readsection = ref_rcu_read_section, .delaysection = ref_rcu_delay_section, @@ -204,7 +205,7 @@ static void srcu_ref_scale_delay_section(const int nloops, const int udl, const } } -static struct ref_scale_ops srcu_ops = { +static const struct ref_scale_ops srcu_ops = { .init = rcu_sync_scale_init, .readsection = srcu_ref_scale_read_section, .delaysection = srcu_ref_scale_delay_section, @@ -231,7 +232,7 @@ static void rcu_tasks_ref_scale_delay_section(const int nloops, const int udl, c un_delay(udl, ndl); } -static struct ref_scale_ops rcu_tasks_ops = { +static const struct ref_scale_ops rcu_tasks_ops = { .init = rcu_sync_scale_init, .readsection = rcu_tasks_ref_scale_read_section, .delaysection = rcu_tasks_ref_scale_delay_section, @@ -270,7 +271,7 @@ static void rcu_trace_ref_scale_delay_section(const int nloops, const int udl, c } } -static struct ref_scale_ops rcu_trace_ops = { +static const struct ref_scale_ops rcu_trace_ops = { .init = rcu_sync_scale_init, .readsection = rcu_trace_ref_scale_read_section, .delaysection = rcu_trace_ref_scale_delay_section, @@ -309,7 +310,7 @@ static void ref_refcnt_delay_section(const int nloops, const int udl, const int } } -static struct ref_scale_ops refcnt_ops = { +static const struct ref_scale_ops refcnt_ops = { .init = rcu_sync_scale_init, .readsection = ref_refcnt_section, .delaysection = ref_refcnt_delay_section, @@ -346,7 +347,7 @@ static void ref_rwlock_delay_section(const int nloops, const int udl, const int } } -static struct ref_scale_ops rwlock_ops = { +static const struct ref_scale_ops rwlock_ops = { .init = ref_rwlock_init, .readsection = ref_rwlock_section, .delaysection = ref_rwlock_delay_section, @@ -383,7 +384,7 @@ static void ref_rwsem_delay_section(const int nloops, const int udl, const int n } } -static struct ref_scale_ops rwsem_ops = { +static const struct ref_scale_ops rwsem_ops = { .init = ref_rwsem_init, .readsection = ref_rwsem_section, .delaysection = ref_rwsem_delay_section, @@ -418,7 +419,7 @@ static void ref_lock_delay_section(const int nloops, const int udl, const int nd preempt_enable(); } -static struct ref_scale_ops lock_ops = { +static const struct ref_scale_ops lock_ops = { .readsection = ref_lock_section, .delaysection = ref_lock_delay_section, .name = "lock" @@ -453,7 +454,7 @@ static void ref_lock_irq_delay_section(const int nloops, const int udl, const in preempt_enable(); } -static struct ref_scale_ops lock_irq_ops = { +static const struct ref_scale_ops lock_irq_ops = { .readsection = ref_lock_irq_section, .delaysection = ref_lock_irq_delay_section, .name = "lock-irq" @@ -489,7 +490,7 @@ static void ref_acqrel_delay_section(const int nloops, const int udl, const int preempt_enable(); } -static struct ref_scale_ops acqrel_ops = { +static const struct ref_scale_ops acqrel_ops = { .readsection = ref_acqrel_section, .delaysection = ref_acqrel_delay_section, .name = "acqrel" @@ -523,7 +524,7 @@ static void ref_clock_delay_section(const int nloops, const int udl, const int n stopopts = x; } -static struct ref_scale_ops clock_ops = { +static const struct ref_scale_ops clock_ops = { .readsection = ref_clock_section, .delaysection = ref_clock_delay_section, .name = "clock" @@ -555,7 +556,7 @@ static void ref_jiffies_delay_section(const int nloops, const int udl, const int stopopts = x; } -static struct ref_scale_ops jiffies_ops = { +static const struct ref_scale_ops jiffies_ops = { .readsection = ref_jiffies_section, .delaysection = ref_jiffies_delay_section, .name = "jiffies" @@ -705,9 +706,9 @@ static void refscale_typesafe_ctor(void *rtsp_in) preempt_enable(); } -static struct ref_scale_ops typesafe_ref_ops; -static struct ref_scale_ops typesafe_lock_ops; -static struct ref_scale_ops typesafe_seqlock_ops; +static const struct ref_scale_ops typesafe_ref_ops; +static const struct ref_scale_ops typesafe_lock_ops; +static const struct ref_scale_ops typesafe_seqlock_ops; // Initialize for a typesafe test. static bool typesafe_init(void) @@ -768,7 +769,7 @@ static void typesafe_cleanup(void) } // The typesafe_init() function distinguishes these structures by address. -static struct ref_scale_ops typesafe_ref_ops = { +static const struct ref_scale_ops typesafe_ref_ops = { .init = typesafe_init, .cleanup = typesafe_cleanup, .readsection = typesafe_read_section, @@ -776,7 +777,7 @@ static struct ref_scale_ops typesafe_ref_ops = { .name = "typesafe_ref" }; -static struct ref_scale_ops typesafe_lock_ops = { +static const struct ref_scale_ops typesafe_lock_ops = { .init = typesafe_init, .cleanup = typesafe_cleanup, .readsection = typesafe_read_section, @@ -784,7 +785,7 @@ static struct ref_scale_ops typesafe_lock_ops = { .name = "typesafe_lock" }; -static struct ref_scale_ops typesafe_seqlock_ops = { +static const struct ref_scale_ops typesafe_seqlock_ops = { .init = typesafe_init, .cleanup = typesafe_cleanup, .readsection = typesafe_read_section, @@ -891,32 +892,34 @@ static u64 process_durations(int n) { int i; struct reader_task *rt; - char buf1[64]; + struct seq_buf s; char *buf; u64 sum = 0; buf = kmalloc(800 + 64, GFP_KERNEL); if (!buf) return 0; - buf[0] = 0; - sprintf(buf, "Experiment #%d (Format: <THREAD-NUM>:<Total loop time in ns>)", - exp_idx); + seq_buf_init(&s, buf, 800 + 64); + + seq_buf_printf(&s, "Experiment #%d (Format: <THREAD-NUM>:<Total loop time in ns>)", + exp_idx); for (i = 0; i < n && !torture_must_stop(); i++) { rt = &(reader_tasks[i]); - sprintf(buf1, "%d: %llu\t", i, rt->last_duration_ns); if (i % 5 == 0) - strcat(buf, "\n"); - if (strlen(buf) >= 800) { - pr_alert("%s", buf); - buf[0] = 0; + seq_buf_putc(&s, '\n'); + + if (seq_buf_used(&s) >= 800) { + pr_alert("%s", seq_buf_str(&s)); + seq_buf_clear(&s); } - strcat(buf, buf1); + + seq_buf_printf(&s, "%d: %llu\t", i, rt->last_duration_ns); sum += rt->last_duration_ns; } - pr_alert("%s\n", buf); + pr_alert("%s\n", seq_buf_str(&s)); kfree(buf); return sum; @@ -1023,7 +1026,7 @@ end: } static void -ref_scale_print_module_parms(struct ref_scale_ops *cur_ops, const char *tag) +ref_scale_print_module_parms(const struct ref_scale_ops *cur_ops, const char *tag) { pr_alert("%s" SCALE_FLAG "--- %s: verbose=%d verbose_batched=%d shutdown=%d holdoff=%d lookup_instances=%ld loops=%ld nreaders=%d nruns=%d readdelay=%d\n", scale_type, tag, @@ -1078,7 +1081,7 @@ ref_scale_init(void) { long i; int firsterr = 0; - static struct ref_scale_ops *scale_ops[] = { + static const struct ref_scale_ops *scale_ops[] = { &rcu_ops, &srcu_ops, RCU_TRACE_OPS RCU_TASKS_OPS &refcnt_ops, &rwlock_ops, &rwsem_ops, &lock_ops, &lock_irq_ops, &acqrel_ops, &clock_ops, &jiffies_ops, &typesafe_ref_ops, &typesafe_lock_ops, &typesafe_seqlock_ops, diff --git a/kernel/rcu/srcutree.c b/kernel/rcu/srcutree.c index b24db425f16d..31706e3293bc 100644 --- a/kernel/rcu/srcutree.c +++ b/kernel/rcu/srcutree.c @@ -137,6 +137,7 @@ static void init_srcu_struct_data(struct srcu_struct *ssp) sdp->srcu_cblist_invoking = false; sdp->srcu_gp_seq_needed = ssp->srcu_sup->srcu_gp_seq; sdp->srcu_gp_seq_needed_exp = ssp->srcu_sup->srcu_gp_seq; + sdp->srcu_barrier_head.next = &sdp->srcu_barrier_head; sdp->mynode = NULL; sdp->cpu = cpu; INIT_WORK(&sdp->work, srcu_invoke_callbacks); @@ -247,7 +248,7 @@ static int init_srcu_struct_fields(struct srcu_struct *ssp, bool is_static) mutex_init(&ssp->srcu_sup->srcu_cb_mutex); mutex_init(&ssp->srcu_sup->srcu_gp_mutex); ssp->srcu_idx = 0; - ssp->srcu_sup->srcu_gp_seq = 0; + ssp->srcu_sup->srcu_gp_seq = SRCU_GP_SEQ_INITIAL_VAL; ssp->srcu_sup->srcu_barrier_seq = 0; mutex_init(&ssp->srcu_sup->srcu_barrier_mutex); atomic_set(&ssp->srcu_sup->srcu_barrier_cpu_cnt, 0); @@ -258,7 +259,7 @@ static int init_srcu_struct_fields(struct srcu_struct *ssp, bool is_static) if (!ssp->sda) goto err_free_sup; init_srcu_struct_data(ssp); - ssp->srcu_sup->srcu_gp_seq_needed_exp = 0; + ssp->srcu_sup->srcu_gp_seq_needed_exp = SRCU_GP_SEQ_INITIAL_VAL; ssp->srcu_sup->srcu_last_gp_end = ktime_get_mono_fast_ns(); if (READ_ONCE(ssp->srcu_sup->srcu_size_state) == SRCU_SIZE_SMALL && SRCU_SIZING_IS_INIT()) { if (!init_srcu_struct_nodes(ssp, GFP_ATOMIC)) @@ -266,7 +267,8 @@ static int init_srcu_struct_fields(struct srcu_struct *ssp, bool is_static) WRITE_ONCE(ssp->srcu_sup->srcu_size_state, SRCU_SIZE_BIG); } ssp->srcu_sup->srcu_ssp = ssp; - smp_store_release(&ssp->srcu_sup->srcu_gp_seq_needed, 0); /* Init done. */ + smp_store_release(&ssp->srcu_sup->srcu_gp_seq_needed, + SRCU_GP_SEQ_INITIAL_VAL); /* Init done. */ return 0; err_free_sda: @@ -628,6 +630,7 @@ static unsigned long srcu_get_delay(struct srcu_struct *ssp) if (time_after(j, gpstart)) jbase += j - gpstart; if (!jbase) { + ASSERT_EXCLUSIVE_WRITER(sup->srcu_n_exp_nodelay); WRITE_ONCE(sup->srcu_n_exp_nodelay, READ_ONCE(sup->srcu_n_exp_nodelay) + 1); if (READ_ONCE(sup->srcu_n_exp_nodelay) > srcu_max_nodelay_phase) jbase = 1; @@ -1560,6 +1563,7 @@ static void srcu_barrier_cb(struct rcu_head *rhp) struct srcu_data *sdp; struct srcu_struct *ssp; + rhp->next = rhp; // Mark the callback as having been invoked. sdp = container_of(rhp, struct srcu_data, srcu_barrier_head); ssp = sdp->ssp; if (atomic_dec_and_test(&ssp->srcu_sup->srcu_barrier_cpu_cnt)) @@ -1818,6 +1822,7 @@ static void process_srcu(struct work_struct *work) } else { j = jiffies; if (READ_ONCE(sup->reschedule_jiffies) == j) { + ASSERT_EXCLUSIVE_WRITER(sup->reschedule_count); WRITE_ONCE(sup->reschedule_count, READ_ONCE(sup->reschedule_count) + 1); if (READ_ONCE(sup->reschedule_count) > srcu_max_nodelay) curdelay = 1; diff --git a/kernel/rcu/tasks.h b/kernel/rcu/tasks.h index 2484c1a8e051..6333f4ccf024 100644 --- a/kernel/rcu/tasks.h +++ b/kernel/rcu/tasks.h @@ -34,6 +34,7 @@ typedef void (*postgp_func_t)(struct rcu_tasks *rtp); * @rtp_blkd_tasks: List of tasks blocked as readers. * @rtp_exit_list: List of tasks in the latter portion of do_exit(). * @cpu: CPU number corresponding to this entry. + * @index: Index of this CPU in rtpcp_array of the rcu_tasks structure. * @rtpp: Pointer to the rcu_tasks structure. */ struct rcu_tasks_percpu { @@ -49,6 +50,7 @@ struct rcu_tasks_percpu { struct list_head rtp_blkd_tasks; struct list_head rtp_exit_list; int cpu; + int index; struct rcu_tasks *rtpp; }; @@ -63,7 +65,7 @@ struct rcu_tasks_percpu { * @init_fract: Initial backoff sleep interval. * @gp_jiffies: Time of last @gp_state transition. * @gp_start: Most recent grace-period start in jiffies. - * @tasks_gp_seq: Number of grace periods completed since boot. + * @tasks_gp_seq: Number of grace periods completed since boot in upper bits. * @n_ipis: Number of IPIs sent to encourage grace periods to end. * @n_ipis_fails: Number of IPI-send failures. * @kthread_ptr: This flavor's grace-period/callback-invocation kthread. @@ -76,6 +78,7 @@ struct rcu_tasks_percpu { * @call_func: This flavor's call_rcu()-equivalent function. * @wait_state: Task state for synchronous grace-period waits (default TASK_UNINTERRUPTIBLE). * @rtpcpu: This flavor's rcu_tasks_percpu structure. + * @rtpcp_array: Array of pointers to rcu_tasks_percpu structure of CPUs in cpu_possible_mask. * @percpu_enqueue_shift: Shift down CPU ID this much when enqueuing callbacks. * @percpu_enqueue_lim: Number of per-CPU callback queues in use for enqueuing. * @percpu_dequeue_lim: Number of per-CPU callback queues in use for dequeuing. @@ -84,6 +87,7 @@ struct rcu_tasks_percpu { * @barrier_q_count: Number of queues being waited on. * @barrier_q_completion: Barrier wait/wakeup mechanism. * @barrier_q_seq: Sequence number for barrier operations. + * @barrier_q_start: Most recent barrier start in jiffies. * @name: This flavor's textual name. * @kname: This flavor's kthread name. */ @@ -110,6 +114,7 @@ struct rcu_tasks { call_rcu_func_t call_func; unsigned int wait_state; struct rcu_tasks_percpu __percpu *rtpcpu; + struct rcu_tasks_percpu **rtpcp_array; int percpu_enqueue_shift; int percpu_enqueue_lim; int percpu_dequeue_lim; @@ -118,6 +123,7 @@ struct rcu_tasks { atomic_t barrier_q_count; struct completion barrier_q_completion; unsigned long barrier_q_seq; + unsigned long barrier_q_start; char *name; char *kname; }; @@ -182,6 +188,8 @@ module_param(rcu_task_collapse_lim, int, 0444); static int rcu_task_lazy_lim __read_mostly = 32; module_param(rcu_task_lazy_lim, int, 0444); +static int rcu_task_cpu_ids; + /* RCU tasks grace-period state for debugging. */ #define RTGS_INIT 0 #define RTGS_WAIT_WAIT_CBS 1 @@ -245,6 +253,8 @@ static void cblist_init_generic(struct rcu_tasks *rtp) int cpu; int lim; int shift; + int maxcpu; + int index = 0; if (rcu_task_enqueue_lim < 0) { rcu_task_enqueue_lim = 1; @@ -254,14 +264,9 @@ static void cblist_init_generic(struct rcu_tasks *rtp) } lim = rcu_task_enqueue_lim; - if (lim > nr_cpu_ids) - lim = nr_cpu_ids; - shift = ilog2(nr_cpu_ids / lim); - if (((nr_cpu_ids - 1) >> shift) >= lim) - shift++; - WRITE_ONCE(rtp->percpu_enqueue_shift, shift); - WRITE_ONCE(rtp->percpu_dequeue_lim, lim); - smp_store_release(&rtp->percpu_enqueue_lim, lim); + rtp->rtpcp_array = kcalloc(num_possible_cpus(), sizeof(struct rcu_tasks_percpu *), GFP_KERNEL); + BUG_ON(!rtp->rtpcp_array); + for_each_possible_cpu(cpu) { struct rcu_tasks_percpu *rtpcp = per_cpu_ptr(rtp->rtpcpu, cpu); @@ -273,14 +278,30 @@ static void cblist_init_generic(struct rcu_tasks *rtp) INIT_WORK(&rtpcp->rtp_work, rcu_tasks_invoke_cbs_wq); rtpcp->cpu = cpu; rtpcp->rtpp = rtp; + rtpcp->index = index; + rtp->rtpcp_array[index] = rtpcp; + index++; if (!rtpcp->rtp_blkd_tasks.next) INIT_LIST_HEAD(&rtpcp->rtp_blkd_tasks); if (!rtpcp->rtp_exit_list.next) INIT_LIST_HEAD(&rtpcp->rtp_exit_list); + rtpcp->barrier_q_head.next = &rtpcp->barrier_q_head; + maxcpu = cpu; } - pr_info("%s: Setting shift to %d and lim to %d rcu_task_cb_adjust=%d.\n", rtp->name, - data_race(rtp->percpu_enqueue_shift), data_race(rtp->percpu_enqueue_lim), rcu_task_cb_adjust); + rcu_task_cpu_ids = maxcpu + 1; + if (lim > rcu_task_cpu_ids) + lim = rcu_task_cpu_ids; + shift = ilog2(rcu_task_cpu_ids / lim); + if (((rcu_task_cpu_ids - 1) >> shift) >= lim) + shift++; + WRITE_ONCE(rtp->percpu_enqueue_shift, shift); + WRITE_ONCE(rtp->percpu_dequeue_lim, lim); + smp_store_release(&rtp->percpu_enqueue_lim, lim); + + pr_info("%s: Setting shift to %d and lim to %d rcu_task_cb_adjust=%d rcu_task_cpu_ids=%d.\n", + rtp->name, data_race(rtp->percpu_enqueue_shift), data_race(rtp->percpu_enqueue_lim), + rcu_task_cb_adjust, rcu_task_cpu_ids); } // Compute wakeup time for lazy callback timer. @@ -339,6 +360,7 @@ static void call_rcu_tasks_generic(struct rcu_head *rhp, rcu_callback_t func, rcu_read_lock(); ideal_cpu = smp_processor_id() >> READ_ONCE(rtp->percpu_enqueue_shift); chosen_cpu = cpumask_next(ideal_cpu - 1, cpu_possible_mask); + WARN_ON_ONCE(chosen_cpu >= rcu_task_cpu_ids); rtpcp = per_cpu_ptr(rtp->rtpcpu, chosen_cpu); if (!raw_spin_trylock_rcu_node(rtpcp)) { // irqs already disabled. raw_spin_lock_rcu_node(rtpcp); // irqs already disabled. @@ -348,7 +370,7 @@ static void call_rcu_tasks_generic(struct rcu_head *rhp, rcu_callback_t func, rtpcp->rtp_n_lock_retries = 0; } if (rcu_task_cb_adjust && ++rtpcp->rtp_n_lock_retries > rcu_task_contend_lim && - READ_ONCE(rtp->percpu_enqueue_lim) != nr_cpu_ids) + READ_ONCE(rtp->percpu_enqueue_lim) != rcu_task_cpu_ids) needadjust = true; // Defer adjustment to avoid deadlock. } // Queuing callbacks before initialization not yet supported. @@ -368,10 +390,10 @@ static void call_rcu_tasks_generic(struct rcu_head *rhp, rcu_callback_t func, raw_spin_unlock_irqrestore_rcu_node(rtpcp, flags); if (unlikely(needadjust)) { raw_spin_lock_irqsave(&rtp->cbs_gbl_lock, flags); - if (rtp->percpu_enqueue_lim != nr_cpu_ids) { + if (rtp->percpu_enqueue_lim != rcu_task_cpu_ids) { WRITE_ONCE(rtp->percpu_enqueue_shift, 0); - WRITE_ONCE(rtp->percpu_dequeue_lim, nr_cpu_ids); - smp_store_release(&rtp->percpu_enqueue_lim, nr_cpu_ids); + WRITE_ONCE(rtp->percpu_dequeue_lim, rcu_task_cpu_ids); + smp_store_release(&rtp->percpu_enqueue_lim, rcu_task_cpu_ids); pr_info("Switching %s to per-CPU callback queuing.\n", rtp->name); } raw_spin_unlock_irqrestore(&rtp->cbs_gbl_lock, flags); @@ -388,6 +410,7 @@ static void rcu_barrier_tasks_generic_cb(struct rcu_head *rhp) struct rcu_tasks *rtp; struct rcu_tasks_percpu *rtpcp; + rhp->next = rhp; // Mark the callback as having been invoked. rtpcp = container_of(rhp, struct rcu_tasks_percpu, barrier_q_head); rtp = rtpcp->rtpp; if (atomic_dec_and_test(&rtp->barrier_q_count)) @@ -396,7 +419,7 @@ static void rcu_barrier_tasks_generic_cb(struct rcu_head *rhp) // Wait for all in-flight callbacks for the specified RCU Tasks flavor. // Operates in a manner similar to rcu_barrier(). -static void rcu_barrier_tasks_generic(struct rcu_tasks *rtp) +static void __maybe_unused rcu_barrier_tasks_generic(struct rcu_tasks *rtp) { int cpu; unsigned long flags; @@ -409,6 +432,7 @@ static void rcu_barrier_tasks_generic(struct rcu_tasks *rtp) mutex_unlock(&rtp->barrier_q_mutex); return; } + rtp->barrier_q_start = jiffies; rcu_seq_start(&rtp->barrier_q_seq); init_completion(&rtp->barrier_q_completion); atomic_set(&rtp->barrier_q_count, 2); @@ -444,6 +468,8 @@ static int rcu_tasks_need_gpcb(struct rcu_tasks *rtp) dequeue_limit = smp_load_acquire(&rtp->percpu_dequeue_lim); for (cpu = 0; cpu < dequeue_limit; cpu++) { + if (!cpu_possible(cpu)) + continue; struct rcu_tasks_percpu *rtpcp = per_cpu_ptr(rtp->rtpcpu, cpu); /* Advance and accelerate any new callbacks. */ @@ -481,7 +507,7 @@ static int rcu_tasks_need_gpcb(struct rcu_tasks *rtp) if (rcu_task_cb_adjust && ncbs <= rcu_task_collapse_lim) { raw_spin_lock_irqsave(&rtp->cbs_gbl_lock, flags); if (rtp->percpu_enqueue_lim > 1) { - WRITE_ONCE(rtp->percpu_enqueue_shift, order_base_2(nr_cpu_ids)); + WRITE_ONCE(rtp->percpu_enqueue_shift, order_base_2(rcu_task_cpu_ids)); smp_store_release(&rtp->percpu_enqueue_lim, 1); rtp->percpu_dequeue_gpseq = get_state_synchronize_rcu(); gpdone = false; @@ -496,7 +522,9 @@ static int rcu_tasks_need_gpcb(struct rcu_tasks *rtp) pr_info("Completing switch %s to CPU-0 callback queuing.\n", rtp->name); } if (rtp->percpu_dequeue_lim == 1) { - for (cpu = rtp->percpu_dequeue_lim; cpu < nr_cpu_ids; cpu++) { + for (cpu = rtp->percpu_dequeue_lim; cpu < rcu_task_cpu_ids; cpu++) { + if (!cpu_possible(cpu)) + continue; struct rcu_tasks_percpu *rtpcp = per_cpu_ptr(rtp->rtpcpu, cpu); WARN_ON_ONCE(rcu_segcblist_n_cbs(&rtpcp->cblist)); @@ -511,30 +539,32 @@ static int rcu_tasks_need_gpcb(struct rcu_tasks *rtp) // Advance callbacks and invoke any that are ready. static void rcu_tasks_invoke_cbs(struct rcu_tasks *rtp, struct rcu_tasks_percpu *rtpcp) { - int cpu; - int cpunext; int cpuwq; unsigned long flags; int len; + int index; struct rcu_head *rhp; struct rcu_cblist rcl = RCU_CBLIST_INITIALIZER(rcl); struct rcu_tasks_percpu *rtpcp_next; - cpu = rtpcp->cpu; - cpunext = cpu * 2 + 1; - if (cpunext < smp_load_acquire(&rtp->percpu_dequeue_lim)) { - rtpcp_next = per_cpu_ptr(rtp->rtpcpu, cpunext); - cpuwq = rcu_cpu_beenfullyonline(cpunext) ? cpunext : WORK_CPU_UNBOUND; - queue_work_on(cpuwq, system_wq, &rtpcp_next->rtp_work); - cpunext++; - if (cpunext < smp_load_acquire(&rtp->percpu_dequeue_lim)) { - rtpcp_next = per_cpu_ptr(rtp->rtpcpu, cpunext); - cpuwq = rcu_cpu_beenfullyonline(cpunext) ? cpunext : WORK_CPU_UNBOUND; + index = rtpcp->index * 2 + 1; + if (index < num_possible_cpus()) { + rtpcp_next = rtp->rtpcp_array[index]; + if (rtpcp_next->cpu < smp_load_acquire(&rtp->percpu_dequeue_lim)) { + cpuwq = rcu_cpu_beenfullyonline(rtpcp_next->cpu) ? rtpcp_next->cpu : WORK_CPU_UNBOUND; queue_work_on(cpuwq, system_wq, &rtpcp_next->rtp_work); + index++; + if (index < num_possible_cpus()) { + rtpcp_next = rtp->rtpcp_array[index]; + if (rtpcp_next->cpu < smp_load_acquire(&rtp->percpu_dequeue_lim)) { + cpuwq = rcu_cpu_beenfullyonline(rtpcp_next->cpu) ? rtpcp_next->cpu : WORK_CPU_UNBOUND; + queue_work_on(cpuwq, system_wq, &rtpcp_next->rtp_work); + } + } } } - if (rcu_segcblist_empty(&rtpcp->cblist) || !cpu_possible(cpu)) + if (rcu_segcblist_empty(&rtpcp->cblist)) return; raw_spin_lock_irqsave_rcu_node(rtpcp, flags); rcu_segcblist_advance(&rtpcp->cblist, rcu_seq_current(&rtp->tasks_gp_seq)); @@ -687,9 +717,7 @@ static void __init rcu_tasks_bootup_oddness(void) #endif /* #ifdef CONFIG_TASKS_TRACE_RCU */ } -#endif /* #ifndef CONFIG_TINY_RCU */ -#ifndef CONFIG_TINY_RCU /* Dump out rcutorture-relevant state common to all RCU-tasks flavors. */ static void show_rcu_tasks_generic_gp_kthread(struct rcu_tasks *rtp, char *s) { @@ -723,6 +751,53 @@ static void show_rcu_tasks_generic_gp_kthread(struct rcu_tasks *rtp, char *s) rtp->lazy_jiffies, s); } + +/* Dump out more rcutorture-relevant state common to all RCU-tasks flavors. */ +static void rcu_tasks_torture_stats_print_generic(struct rcu_tasks *rtp, char *tt, + char *tf, char *tst) +{ + cpumask_var_t cm; + int cpu; + bool gotcb = false; + unsigned long j = jiffies; + + pr_alert("%s%s Tasks%s RCU g%ld gp_start %lu gp_jiffies %lu gp_state %d (%s).\n", + tt, tf, tst, data_race(rtp->tasks_gp_seq), + j - data_race(rtp->gp_start), j - data_race(rtp->gp_jiffies), + data_race(rtp->gp_state), tasks_gp_state_getname(rtp)); + pr_alert("\tEnqueue shift %d limit %d Dequeue limit %d gpseq %lu.\n", + data_race(rtp->percpu_enqueue_shift), + data_race(rtp->percpu_enqueue_lim), + data_race(rtp->percpu_dequeue_lim), + data_race(rtp->percpu_dequeue_gpseq)); + (void)zalloc_cpumask_var(&cm, GFP_KERNEL); + pr_alert("\tCallback counts:"); + for_each_possible_cpu(cpu) { + long n; + struct rcu_tasks_percpu *rtpcp = per_cpu_ptr(rtp->rtpcpu, cpu); + + if (cpumask_available(cm) && !rcu_barrier_cb_is_done(&rtpcp->barrier_q_head)) + cpumask_set_cpu(cpu, cm); + n = rcu_segcblist_n_cbs(&rtpcp->cblist); + if (!n) + continue; + pr_cont(" %d:%ld", cpu, n); + gotcb = true; + } + if (gotcb) + pr_cont(".\n"); + else + pr_cont(" (none).\n"); + pr_alert("\tBarrier seq %lu start %lu count %d holdout CPUs ", + data_race(rtp->barrier_q_seq), j - data_race(rtp->barrier_q_start), + atomic_read(&rtp->barrier_q_count)); + if (cpumask_available(cm) && !cpumask_empty(cm)) + pr_cont(" %*pbl.\n", cpumask_pr_args(cm)); + else + pr_cont("(none).\n"); + free_cpumask_var(cm); +} + #endif // #ifndef CONFIG_TINY_RCU static void exit_tasks_rcu_finish_trace(struct task_struct *t); @@ -1174,6 +1249,12 @@ void show_rcu_tasks_classic_gp_kthread(void) show_rcu_tasks_generic_gp_kthread(&rcu_tasks, ""); } EXPORT_SYMBOL_GPL(show_rcu_tasks_classic_gp_kthread); + +void rcu_tasks_torture_stats_print(char *tt, char *tf) +{ + rcu_tasks_torture_stats_print_generic(&rcu_tasks, tt, tf, ""); +} +EXPORT_SYMBOL_GPL(rcu_tasks_torture_stats_print); #endif // !defined(CONFIG_TINY_RCU) struct task_struct *get_rcu_tasks_gp_kthread(void) @@ -1244,13 +1325,12 @@ void exit_tasks_rcu_finish(void) { exit_tasks_rcu_finish_trace(current); } //////////////////////////////////////////////////////////////////////// // -// "Rude" variant of Tasks RCU, inspired by Steve Rostedt's trick of -// passing an empty function to schedule_on_each_cpu(). This approach -// provides an asynchronous call_rcu_tasks_rude() API and batching of -// concurrent calls to the synchronous synchronize_rcu_tasks_rude() API. -// This invokes schedule_on_each_cpu() in order to send IPIs far and wide -// and induces otherwise unnecessary context switches on all online CPUs, -// whether idle or not. +// "Rude" variant of Tasks RCU, inspired by Steve Rostedt's +// trick of passing an empty function to schedule_on_each_cpu(). +// This approach provides batching of concurrent calls to the synchronous +// synchronize_rcu_tasks_rude() API. This invokes schedule_on_each_cpu() +// in order to send IPIs far and wide and induces otherwise unnecessary +// context switches on all online CPUs, whether idle or not. // // Callback handling is provided by the rcu_tasks_kthread() function. // @@ -1268,11 +1348,11 @@ static void rcu_tasks_rude_wait_gp(struct rcu_tasks *rtp) schedule_on_each_cpu(rcu_tasks_be_rude); } -void call_rcu_tasks_rude(struct rcu_head *rhp, rcu_callback_t func); +static void call_rcu_tasks_rude(struct rcu_head *rhp, rcu_callback_t func); DEFINE_RCU_TASKS(rcu_tasks_rude, rcu_tasks_rude_wait_gp, call_rcu_tasks_rude, "RCU Tasks Rude"); -/** +/* * call_rcu_tasks_rude() - Queue a callback rude task-based grace period * @rhp: structure to be used for queueing the RCU updates. * @func: actual callback function to be invoked after the grace period @@ -1289,12 +1369,14 @@ DEFINE_RCU_TASKS(rcu_tasks_rude, rcu_tasks_rude_wait_gp, call_rcu_tasks_rude, * * See the description of call_rcu() for more detailed information on * memory ordering guarantees. + * + * This is no longer exported, and is instead reserved for use by + * synchronize_rcu_tasks_rude(). */ -void call_rcu_tasks_rude(struct rcu_head *rhp, rcu_callback_t func) +static void call_rcu_tasks_rude(struct rcu_head *rhp, rcu_callback_t func) { call_rcu_tasks_generic(rhp, func, &rcu_tasks_rude); } -EXPORT_SYMBOL_GPL(call_rcu_tasks_rude); /** * synchronize_rcu_tasks_rude - wait for a rude rcu-tasks grace period @@ -1320,26 +1402,9 @@ void synchronize_rcu_tasks_rude(void) } EXPORT_SYMBOL_GPL(synchronize_rcu_tasks_rude); -/** - * rcu_barrier_tasks_rude - Wait for in-flight call_rcu_tasks_rude() callbacks. - * - * Although the current implementation is guaranteed to wait, it is not - * obligated to, for example, if there are no pending callbacks. - */ -void rcu_barrier_tasks_rude(void) -{ - rcu_barrier_tasks_generic(&rcu_tasks_rude); -} -EXPORT_SYMBOL_GPL(rcu_barrier_tasks_rude); - -int rcu_tasks_rude_lazy_ms = -1; -module_param(rcu_tasks_rude_lazy_ms, int, 0444); - static int __init rcu_spawn_tasks_rude_kthread(void) { rcu_tasks_rude.gp_sleep = HZ / 10; - if (rcu_tasks_rude_lazy_ms >= 0) - rcu_tasks_rude.lazy_jiffies = msecs_to_jiffies(rcu_tasks_rude_lazy_ms); rcu_spawn_tasks_kthread_generic(&rcu_tasks_rude); return 0; } @@ -1350,6 +1415,12 @@ void show_rcu_tasks_rude_gp_kthread(void) show_rcu_tasks_generic_gp_kthread(&rcu_tasks_rude, ""); } EXPORT_SYMBOL_GPL(show_rcu_tasks_rude_gp_kthread); + +void rcu_tasks_rude_torture_stats_print(char *tt, char *tf) +{ + rcu_tasks_torture_stats_print_generic(&rcu_tasks_rude, tt, tf, ""); +} +EXPORT_SYMBOL_GPL(rcu_tasks_rude_torture_stats_print); #endif // !defined(CONFIG_TINY_RCU) struct task_struct *get_rcu_tasks_rude_gp_kthread(void) @@ -2027,6 +2098,12 @@ void show_rcu_tasks_trace_gp_kthread(void) show_rcu_tasks_generic_gp_kthread(&rcu_tasks_trace, buf); } EXPORT_SYMBOL_GPL(show_rcu_tasks_trace_gp_kthread); + +void rcu_tasks_trace_torture_stats_print(char *tt, char *tf) +{ + rcu_tasks_torture_stats_print_generic(&rcu_tasks_trace, tt, tf, ""); +} +EXPORT_SYMBOL_GPL(rcu_tasks_trace_torture_stats_print); #endif // !defined(CONFIG_TINY_RCU) struct task_struct *get_rcu_tasks_trace_gp_kthread(void) @@ -2069,11 +2146,6 @@ static struct rcu_tasks_test_desc tests[] = { /* If not defined, the test is skipped. */ .notrun = IS_ENABLED(CONFIG_TASKS_RCU), }, - { - .name = "call_rcu_tasks_rude()", - /* If not defined, the test is skipped. */ - .notrun = IS_ENABLED(CONFIG_TASKS_RUDE_RCU), - }, { .name = "call_rcu_tasks_trace()", /* If not defined, the test is skipped. */ @@ -2081,6 +2153,7 @@ static struct rcu_tasks_test_desc tests[] = { } }; +#if defined(CONFIG_TASKS_RCU) || defined(CONFIG_TASKS_TRACE_RCU) static void test_rcu_tasks_callback(struct rcu_head *rhp) { struct rcu_tasks_test_desc *rttd = @@ -2090,6 +2163,7 @@ static void test_rcu_tasks_callback(struct rcu_head *rhp) rttd->notrun = false; } +#endif // #if defined(CONFIG_TASKS_RCU) || defined(CONFIG_TASKS_TRACE_RCU) static void rcu_tasks_initiate_self_tests(void) { @@ -2102,16 +2176,14 @@ static void rcu_tasks_initiate_self_tests(void) #ifdef CONFIG_TASKS_RUDE_RCU pr_info("Running RCU Tasks Rude wait API self tests\n"); - tests[1].runstart = jiffies; synchronize_rcu_tasks_rude(); - call_rcu_tasks_rude(&tests[1].rh, test_rcu_tasks_callback); #endif #ifdef CONFIG_TASKS_TRACE_RCU pr_info("Running RCU Tasks Trace wait API self tests\n"); - tests[2].runstart = jiffies; + tests[1].runstart = jiffies; synchronize_rcu_tasks_trace(); - call_rcu_tasks_trace(&tests[2].rh, test_rcu_tasks_callback); + call_rcu_tasks_trace(&tests[1].rh, test_rcu_tasks_callback); #endif } diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index e0f6fc272906..0c8b56c4ee88 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -79,9 +79,6 @@ static void rcu_sr_normal_gp_cleanup_work(struct work_struct *); static DEFINE_PER_CPU_SHARED_ALIGNED(struct rcu_data, rcu_data) = { .gpwrap = true, -#ifdef CONFIG_RCU_NOCB_CPU - .cblist.flags = SEGCBLIST_RCU_CORE, -#endif }; static struct rcu_state rcu_state = { .level = { &rcu_state.node[0] }, @@ -97,6 +94,9 @@ static struct rcu_state rcu_state = { .srs_cleanup_work = __WORK_INITIALIZER(rcu_state.srs_cleanup_work, rcu_sr_normal_gp_cleanup_work), .srs_cleanups_pending = ATOMIC_INIT(0), +#ifdef CONFIG_RCU_NOCB_CPU + .nocb_mutex = __MUTEX_INITIALIZER(rcu_state.nocb_mutex), +#endif }; /* Dump rcu_node combining tree at boot to verify correct setup. */ @@ -1660,7 +1660,7 @@ static void rcu_sr_normal_gp_cleanup_work(struct work_struct *work) * the done tail list manipulations are protected here. */ done = smp_load_acquire(&rcu_state.srs_done_tail); - if (!done) + if (WARN_ON_ONCE(!done)) return; WARN_ON_ONCE(!rcu_sr_is_wait_head(done)); @@ -2394,7 +2394,6 @@ rcu_report_qs_rdp(struct rcu_data *rdp) { unsigned long flags; unsigned long mask; - bool needacc = false; struct rcu_node *rnp; WARN_ON_ONCE(rdp->cpu != smp_processor_id()); @@ -2431,23 +2430,11 @@ rcu_report_qs_rdp(struct rcu_data *rdp) * to return true. So complain, but don't awaken. */ WARN_ON_ONCE(rcu_accelerate_cbs(rnp, rdp)); - } else if (!rcu_segcblist_completely_offloaded(&rdp->cblist)) { - /* - * ...but NOCB kthreads may miss or delay callbacks acceleration - * if in the middle of a (de-)offloading process. - */ - needacc = true; } rcu_disable_urgency_upon_qs(rdp); rcu_report_qs_rnp(mask, rnp, rnp->gp_seq, flags); /* ^^^ Released rnp->lock */ - - if (needacc) { - rcu_nocb_lock_irqsave(rdp, flags); - rcu_accelerate_cbs_unlocked(rnp, rdp); - rcu_nocb_unlock_irqrestore(rdp, flags); - } } } @@ -2802,24 +2789,6 @@ static __latent_entropy void rcu_core(void) unsigned long flags; struct rcu_data *rdp = raw_cpu_ptr(&rcu_data); struct rcu_node *rnp = rdp->mynode; - /* - * On RT rcu_core() can be preempted when IRQs aren't disabled. - * Therefore this function can race with concurrent NOCB (de-)offloading - * on this CPU and the below condition must be considered volatile. - * However if we race with: - * - * _ Offloading: In the worst case we accelerate or process callbacks - * concurrently with NOCB kthreads. We are guaranteed to - * call rcu_nocb_lock() if that happens. - * - * _ Deoffloading: In the worst case we miss callbacks acceleration or - * processing. This is fine because the early stage - * of deoffloading invokes rcu_core() after setting - * SEGCBLIST_RCU_CORE. So we guarantee that we'll process - * what could have been dismissed without the need to wait - * for the next rcu_pending() check in the next jiffy. - */ - const bool do_batch = !rcu_segcblist_completely_offloaded(&rdp->cblist); if (cpu_is_offline(smp_processor_id())) return; @@ -2839,17 +2808,17 @@ static __latent_entropy void rcu_core(void) /* No grace period and unregistered callbacks? */ if (!rcu_gp_in_progress() && - rcu_segcblist_is_enabled(&rdp->cblist) && do_batch) { - rcu_nocb_lock_irqsave(rdp, flags); + rcu_segcblist_is_enabled(&rdp->cblist) && !rcu_rdp_is_offloaded(rdp)) { + local_irq_save(flags); if (!rcu_segcblist_restempty(&rdp->cblist, RCU_NEXT_READY_TAIL)) rcu_accelerate_cbs_unlocked(rnp, rdp); - rcu_nocb_unlock_irqrestore(rdp, flags); + local_irq_restore(flags); } rcu_check_gp_start_stall(rnp, rdp, rcu_jiffies_till_stall_check()); /* If there are callbacks ready, invoke them. */ - if (do_batch && rcu_segcblist_ready_cbs(&rdp->cblist) && + if (!rcu_rdp_is_offloaded(rdp) && rcu_segcblist_ready_cbs(&rdp->cblist) && likely(READ_ONCE(rcu_scheduler_fully_active))) { rcu_do_batch(rdp); /* Re-invoke RCU core processing if there are callbacks remaining. */ @@ -3238,7 +3207,7 @@ struct kvfree_rcu_bulk_data { struct list_head list; struct rcu_gp_oldstate gp_snap; unsigned long nr_records; - void *records[]; + void *records[] __counted_by(nr_records); }; /* @@ -3550,10 +3519,10 @@ schedule_delayed_monitor_work(struct kfree_rcu_cpu *krcp) if (delayed_work_pending(&krcp->monitor_work)) { delay_left = krcp->monitor_work.timer.expires - jiffies; if (delay < delay_left) - mod_delayed_work(system_wq, &krcp->monitor_work, delay); + mod_delayed_work(system_unbound_wq, &krcp->monitor_work, delay); return; } - queue_delayed_work(system_wq, &krcp->monitor_work, delay); + queue_delayed_work(system_unbound_wq, &krcp->monitor_work, delay); } static void @@ -3645,7 +3614,7 @@ static void kfree_rcu_monitor(struct work_struct *work) // be that the work is in the pending state when // channels have been detached following by each // other. - queue_rcu_work(system_wq, &krwp->rcu_work); + queue_rcu_work(system_unbound_wq, &krwp->rcu_work); } } @@ -3715,7 +3684,7 @@ run_page_cache_worker(struct kfree_rcu_cpu *krcp) if (rcu_scheduler_active == RCU_SCHEDULER_RUNNING && !atomic_xchg(&krcp->work_in_progress, 1)) { if (atomic_read(&krcp->backoff_page_cache_fill)) { - queue_delayed_work(system_wq, + queue_delayed_work(system_unbound_wq, &krcp->page_cache_work, msecs_to_jiffies(rcu_delay_page_cache_fill_msec)); } else { @@ -3778,7 +3747,8 @@ add_ptr_to_bulk_krc_lock(struct kfree_rcu_cpu **krcp, } // Finally insert and update the GP for this page. - bnode->records[bnode->nr_records++] = ptr; + bnode->nr_records++; + bnode->records[bnode->nr_records - 1] = ptr; get_state_synchronize_rcu_full(&bnode->gp_snap); atomic_inc(&(*krcp)->bulk_count[idx]); @@ -4414,6 +4384,7 @@ static void rcu_barrier_callback(struct rcu_head *rhp) { unsigned long __maybe_unused s = rcu_state.barrier_sequence; + rhp->next = rhp; // Mark the callback as having been invoked. if (atomic_dec_and_test(&rcu_state.barrier_cpu_count)) { rcu_barrier_trace(TPS("LastCB"), -1, s); complete(&rcu_state.barrier_completion); @@ -5435,6 +5406,8 @@ static void __init rcu_init_one(void) while (i > rnp->grphi) rnp++; per_cpu_ptr(&rcu_data, i)->mynode = rnp; + per_cpu_ptr(&rcu_data, i)->barrier_head.next = + &per_cpu_ptr(&rcu_data, i)->barrier_head; rcu_boot_init_percpu_data(i); } } diff --git a/kernel/rcu/tree.h b/kernel/rcu/tree.h index 13fdcc2aa081..a9a811d9d7a3 100644 --- a/kernel/rcu/tree.h +++ b/kernel/rcu/tree.h @@ -411,7 +411,6 @@ struct rcu_state { arch_spinlock_t ofl_lock ____cacheline_internodealigned_in_smp; /* Synchronize offline with */ /* GP pre-initialization. */ - int nocb_is_setup; /* nocb is setup from boot */ /* synchronize_rcu() part. */ struct llist_head srs_next; /* request a GP users. */ @@ -420,6 +419,11 @@ struct rcu_state { struct sr_wait_node srs_wait_nodes[SR_NORMAL_GP_WAIT_HEAD_MAX]; struct work_struct srs_cleanup_work; atomic_t srs_cleanups_pending; /* srs inflight worker cleanups. */ + +#ifdef CONFIG_RCU_NOCB_CPU + struct mutex nocb_mutex; /* Guards (de-)offloading */ + int nocb_is_setup; /* nocb is setup from boot */ +#endif }; /* Values for rcu_state structure's gp_flags field. */ diff --git a/kernel/rcu/tree_exp.h b/kernel/rcu/tree_exp.h index eec42df0c018..c3266bf709d5 100644 --- a/kernel/rcu/tree_exp.h +++ b/kernel/rcu/tree_exp.h @@ -542,6 +542,67 @@ static bool synchronize_rcu_expedited_wait_once(long tlimit) return false; } +/* + * Print out an expedited RCU CPU stall warning message. + */ +static void synchronize_rcu_expedited_stall(unsigned long jiffies_start, unsigned long j) +{ + int cpu; + unsigned long mask; + int ndetected; + struct rcu_node *rnp; + struct rcu_node *rnp_root = rcu_get_root(); + + if (READ_ONCE(csd_lock_suppress_rcu_stall) && csd_lock_is_stuck()) { + pr_err("INFO: %s detected expedited stalls, but suppressed full report due to a stuck CSD-lock.\n", rcu_state.name); + return; + } + pr_err("INFO: %s detected expedited stalls on CPUs/tasks: {", rcu_state.name); + ndetected = 0; + rcu_for_each_leaf_node(rnp) { + ndetected += rcu_print_task_exp_stall(rnp); + for_each_leaf_node_possible_cpu(rnp, cpu) { + struct rcu_data *rdp; + + mask = leaf_node_cpu_bit(rnp, cpu); + if (!(READ_ONCE(rnp->expmask) & mask)) + continue; + ndetected++; + rdp = per_cpu_ptr(&rcu_data, cpu); + pr_cont(" %d-%c%c%c%c", cpu, + "O."[!!cpu_online(cpu)], + "o."[!!(rdp->grpmask & rnp->expmaskinit)], + "N."[!!(rdp->grpmask & rnp->expmaskinitnext)], + "D."[!!data_race(rdp->cpu_no_qs.b.exp)]); + } + } + pr_cont(" } %lu jiffies s: %lu root: %#lx/%c\n", + j - jiffies_start, rcu_state.expedited_sequence, data_race(rnp_root->expmask), + ".T"[!!data_race(rnp_root->exp_tasks)]); + if (ndetected) { + pr_err("blocking rcu_node structures (internal RCU debug):"); + rcu_for_each_node_breadth_first(rnp) { + if (rnp == rnp_root) + continue; /* printed unconditionally */ + if (sync_rcu_exp_done_unlocked(rnp)) + continue; + pr_cont(" l=%u:%d-%d:%#lx/%c", + rnp->level, rnp->grplo, rnp->grphi, data_race(rnp->expmask), + ".T"[!!data_race(rnp->exp_tasks)]); + } + pr_cont("\n"); + } + rcu_for_each_leaf_node(rnp) { + for_each_leaf_node_possible_cpu(rnp, cpu) { + mask = leaf_node_cpu_bit(rnp, cpu); + if (!(READ_ONCE(rnp->expmask) & mask)) + continue; + dump_cpu_task(cpu); + } + rcu_exp_print_detail_task_stall_rnp(rnp); + } +} + /* * Wait for the expedited grace period to elapse, issuing any needed * RCU CPU stall warnings along the way. @@ -553,10 +614,8 @@ static void synchronize_rcu_expedited_wait(void) unsigned long jiffies_stall; unsigned long jiffies_start; unsigned long mask; - int ndetected; struct rcu_data *rdp; struct rcu_node *rnp; - struct rcu_node *rnp_root = rcu_get_root(); unsigned long flags; trace_rcu_exp_grace_period(rcu_state.name, rcu_exp_gp_seq_endval(), TPS("startwait")); @@ -593,55 +652,7 @@ static void synchronize_rcu_expedited_wait(void) j = jiffies; rcu_stall_notifier_call_chain(RCU_STALL_NOTIFY_EXP, (void *)(j - jiffies_start)); trace_rcu_stall_warning(rcu_state.name, TPS("ExpeditedStall")); - pr_err("INFO: %s detected expedited stalls on CPUs/tasks: {", - rcu_state.name); - ndetected = 0; - rcu_for_each_leaf_node(rnp) { - ndetected += rcu_print_task_exp_stall(rnp); - for_each_leaf_node_possible_cpu(rnp, cpu) { - struct rcu_data *rdp; - - mask = leaf_node_cpu_bit(rnp, cpu); - if (!(READ_ONCE(rnp->expmask) & mask)) - continue; - ndetected++; - rdp = per_cpu_ptr(&rcu_data, cpu); - pr_cont(" %d-%c%c%c%c", cpu, - "O."[!!cpu_online(cpu)], - "o."[!!(rdp->grpmask & rnp->expmaskinit)], - "N."[!!(rdp->grpmask & rnp->expmaskinitnext)], - "D."[!!data_race(rdp->cpu_no_qs.b.exp)]); - } - } - pr_cont(" } %lu jiffies s: %lu root: %#lx/%c\n", - j - jiffies_start, rcu_state.expedited_sequence, - data_race(rnp_root->expmask), - ".T"[!!data_race(rnp_root->exp_tasks)]); - if (ndetected) { - pr_err("blocking rcu_node structures (internal RCU debug):"); - rcu_for_each_node_breadth_first(rnp) { - if (rnp == rnp_root) - continue; /* printed unconditionally */ - if (sync_rcu_exp_done_unlocked(rnp)) - continue; - pr_cont(" l=%u:%d-%d:%#lx/%c", - rnp->level, rnp->grplo, rnp->grphi, - data_race(rnp->expmask), - ".T"[!!data_race(rnp->exp_tasks)]); - } - pr_cont("\n"); - } - rcu_for_each_leaf_node(rnp) { - for_each_leaf_node_possible_cpu(rnp, cpu) { - mask = leaf_node_cpu_bit(rnp, cpu); - if (!(READ_ONCE(rnp->expmask) & mask)) - continue; - preempt_disable(); // For smp_processor_id() in dump_cpu_task(). - dump_cpu_task(cpu); - preempt_enable(); - } - rcu_exp_print_detail_task_stall_rnp(rnp); - } + synchronize_rcu_expedited_stall(jiffies_start, j); jiffies_stall = 3 * rcu_exp_jiffies_till_stall_check() + 3; panic_on_rcu_stall(); } diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h index e629646ac6f5..97b99cd06923 100644 --- a/kernel/rcu/tree_nocb.h +++ b/kernel/rcu/tree_nocb.h @@ -16,10 +16,6 @@ #ifdef CONFIG_RCU_NOCB_CPU static cpumask_var_t rcu_nocb_mask; /* CPUs to have callbacks offloaded. */ static bool __read_mostly rcu_nocb_poll; /* Offload kthread are to poll. */ -static inline int rcu_lockdep_is_held_nocb(struct rcu_data *rdp) -{ - return lockdep_is_held(&rdp->nocb_lock); -} static inline bool rcu_current_is_nocb_kthread(struct rcu_data *rdp) { @@ -220,7 +216,7 @@ static bool __wake_nocb_gp(struct rcu_data *rdp_gp, raw_spin_unlock_irqrestore(&rdp_gp->nocb_gp_lock, flags); if (needwake) { trace_rcu_nocb_wake(rcu_state.name, rdp->cpu, TPS("DoWake")); - wake_up_process(rdp_gp->nocb_gp_kthread); + swake_up_one_online(&rdp_gp->nocb_gp_wq); } return needwake; @@ -413,14 +409,6 @@ static bool rcu_nocb_try_bypass(struct rcu_data *rdp, struct rcu_head *rhp, return false; } - // In the process of (de-)offloading: no bypassing, but - // locking. - if (!rcu_segcblist_completely_offloaded(&rdp->cblist)) { - rcu_nocb_lock(rdp); - *was_alldone = !rcu_segcblist_pend_cbs(&rdp->cblist); - return false; /* Not offloaded, no bypassing. */ - } - // Don't use ->nocb_bypass during early boot. if (rcu_scheduler_active != RCU_SCHEDULER_RUNNING) { rcu_nocb_lock(rdp); @@ -505,7 +493,7 @@ static bool rcu_nocb_try_bypass(struct rcu_data *rdp, struct rcu_head *rhp, trace_rcu_nocb_wake(rcu_state.name, rdp->cpu, TPS("FirstBQ")); } rcu_nocb_bypass_unlock(rdp); - smp_mb(); /* Order enqueue before wake. */ + // A wake up of the grace period kthread or timer adjustment // needs to be done only if: // 1. Bypass list was fully empty before (this is the first @@ -616,37 +604,33 @@ static void call_rcu_nocb(struct rcu_data *rdp, struct rcu_head *head, } } -static int nocb_gp_toggle_rdp(struct rcu_data *rdp) +static void nocb_gp_toggle_rdp(struct rcu_data *rdp_gp, struct rcu_data *rdp) { struct rcu_segcblist *cblist = &rdp->cblist; unsigned long flags; - int ret; - rcu_nocb_lock_irqsave(rdp, flags); - if (rcu_segcblist_test_flags(cblist, SEGCBLIST_OFFLOADED) && - !rcu_segcblist_test_flags(cblist, SEGCBLIST_KTHREAD_GP)) { + /* + * Locking orders future de-offloaded callbacks enqueue against previous + * handling of this rdp. Ie: Make sure rcuog is done with this rdp before + * deoffloaded callbacks can be enqueued. + */ + raw_spin_lock_irqsave(&rdp->nocb_lock, flags); + if (!rcu_segcblist_test_flags(cblist, SEGCBLIST_OFFLOADED)) { /* * Offloading. Set our flag and notify the offload worker. * We will handle this rdp until it ever gets de-offloaded. */ - rcu_segcblist_set_flags(cblist, SEGCBLIST_KTHREAD_GP); - ret = 1; - } else if (!rcu_segcblist_test_flags(cblist, SEGCBLIST_OFFLOADED) && - rcu_segcblist_test_flags(cblist, SEGCBLIST_KTHREAD_GP)) { + list_add_tail(&rdp->nocb_entry_rdp, &rdp_gp->nocb_head_rdp); + rcu_segcblist_set_flags(cblist, SEGCBLIST_OFFLOADED); + } else { /* * De-offloading. Clear our flag and notify the de-offload worker. * We will ignore this rdp until it ever gets re-offloaded. */ - rcu_segcblist_clear_flags(cblist, SEGCBLIST_KTHREAD_GP); - ret = 0; - } else { - WARN_ON_ONCE(1); - ret = -1; + list_del(&rdp->nocb_entry_rdp); + rcu_segcblist_clear_flags(cblist, SEGCBLIST_OFFLOADED); } - - rcu_nocb_unlock_irqrestore(rdp, flags); - - return ret; + raw_spin_unlock_irqrestore(&rdp->nocb_lock, flags); } static void nocb_gp_sleep(struct rcu_data *my_rdp, int cpu) @@ -853,14 +837,7 @@ static void nocb_gp_wait(struct rcu_data *my_rdp) } if (rdp_toggling) { - int ret; - - ret = nocb_gp_toggle_rdp(rdp_toggling); - if (ret == 1) - list_add_tail(&rdp_toggling->nocb_entry_rdp, &my_rdp->nocb_head_rdp); - else if (ret == 0) - list_del(&rdp_toggling->nocb_entry_rdp); - + nocb_gp_toggle_rdp(my_rdp, rdp_toggling); swake_up_one(&rdp_toggling->nocb_state_wq); } @@ -1030,16 +1007,11 @@ void rcu_nocb_flush_deferred_wakeup(void) } EXPORT_SYMBOL_GPL(rcu_nocb_flush_deferred_wakeup); -static int rdp_offload_toggle(struct rcu_data *rdp, - bool offload, unsigned long flags) - __releases(rdp->nocb_lock) +static int rcu_nocb_queue_toggle_rdp(struct rcu_data *rdp) { - struct rcu_segcblist *cblist = &rdp->cblist; struct rcu_data *rdp_gp = rdp->nocb_gp_rdp; bool wake_gp = false; - - rcu_segcblist_offload(cblist, offload); - rcu_nocb_unlock_irqrestore(rdp, flags); + unsigned long flags; raw_spin_lock_irqsave(&rdp_gp->nocb_gp_lock, flags); // Queue this rdp for add/del to/from the list to iterate on rcuog @@ -1053,89 +1025,74 @@ static int rdp_offload_toggle(struct rcu_data *rdp, return wake_gp; } -static long rcu_nocb_rdp_deoffload(void *arg) +static bool rcu_nocb_rdp_deoffload_wait_cond(struct rcu_data *rdp) +{ + unsigned long flags; + bool ret; + + /* + * Locking makes sure rcuog is done handling this rdp before deoffloaded + * enqueue can happen. Also it keeps the SEGCBLIST_OFFLOADED flag stable + * while the ->nocb_lock is held. + */ + raw_spin_lock_irqsave(&rdp->nocb_lock, flags); + ret = !rcu_segcblist_test_flags(&rdp->cblist, SEGCBLIST_OFFLOADED); + raw_spin_unlock_irqrestore(&rdp->nocb_lock, flags); + + return ret; +} + +static int rcu_nocb_rdp_deoffload(struct rcu_data *rdp) { - struct rcu_data *rdp = arg; - struct rcu_segcblist *cblist = &rdp->cblist; unsigned long flags; int wake_gp; struct rcu_data *rdp_gp = rdp->nocb_gp_rdp; - /* - * rcu_nocb_rdp_deoffload() may be called directly if - * rcuog/o[p] spawn failed, because at this time the rdp->cpu - * is not online yet. - */ - WARN_ON_ONCE((rdp->cpu != raw_smp_processor_id()) && cpu_online(rdp->cpu)); + /* CPU must be offline, unless it's early boot */ + WARN_ON_ONCE(cpu_online(rdp->cpu) && rdp->cpu != raw_smp_processor_id()); pr_info("De-offloading %d\n", rdp->cpu); + /* Flush all callbacks from segcblist and bypass */ + rcu_barrier(); + + /* + * Make sure the rcuoc kthread isn't in the middle of a nocb locked + * sequence while offloading is deactivated, along with nocb locking. + */ + if (rdp->nocb_cb_kthread) + kthread_park(rdp->nocb_cb_kthread); + rcu_nocb_lock_irqsave(rdp, flags); - /* - * Flush once and for all now. This suffices because we are - * running on the target CPU holding ->nocb_lock (thus having - * interrupts disabled), and because rdp_offload_toggle() - * invokes rcu_segcblist_offload(), which clears SEGCBLIST_OFFLOADED. - * Thus future calls to rcu_segcblist_completely_offloaded() will - * return false, which means that future calls to rcu_nocb_try_bypass() - * will refuse to put anything into the bypass. - */ - WARN_ON_ONCE(!rcu_nocb_flush_bypass(rdp, NULL, jiffies, false)); - /* - * Start with invoking rcu_core() early. This way if the current thread - * happens to preempt an ongoing call to rcu_core() in the middle, - * leaving some work dismissed because rcu_core() still thinks the rdp is - * completely offloaded, we are guaranteed a nearby future instance of - * rcu_core() to catch up. - */ - rcu_segcblist_set_flags(cblist, SEGCBLIST_RCU_CORE); - invoke_rcu_core(); - wake_gp = rdp_offload_toggle(rdp, false, flags); + WARN_ON_ONCE(rcu_cblist_n_cbs(&rdp->nocb_bypass)); + WARN_ON_ONCE(rcu_segcblist_n_cbs(&rdp->cblist)); + rcu_nocb_unlock_irqrestore(rdp, flags); + + wake_gp = rcu_nocb_queue_toggle_rdp(rdp); mutex_lock(&rdp_gp->nocb_gp_kthread_mutex); + if (rdp_gp->nocb_gp_kthread) { if (wake_gp) wake_up_process(rdp_gp->nocb_gp_kthread); swait_event_exclusive(rdp->nocb_state_wq, - !rcu_segcblist_test_flags(cblist, - SEGCBLIST_KTHREAD_GP)); - if (rdp->nocb_cb_kthread) - kthread_park(rdp->nocb_cb_kthread); + rcu_nocb_rdp_deoffload_wait_cond(rdp)); } else { /* * No kthread to clear the flags for us or remove the rdp from the nocb list * to iterate. Do it here instead. Locking doesn't look stricly necessary * but we stick to paranoia in this rare path. */ - rcu_nocb_lock_irqsave(rdp, flags); - rcu_segcblist_clear_flags(&rdp->cblist, SEGCBLIST_KTHREAD_GP); - rcu_nocb_unlock_irqrestore(rdp, flags); + raw_spin_lock_irqsave(&rdp->nocb_lock, flags); + rcu_segcblist_clear_flags(&rdp->cblist, SEGCBLIST_OFFLOADED); + raw_spin_unlock_irqrestore(&rdp->nocb_lock, flags); list_del(&rdp->nocb_entry_rdp); } + mutex_unlock(&rdp_gp->nocb_gp_kthread_mutex); - /* - * Lock one last time to acquire latest callback updates from kthreads - * so we can later handle callbacks locally without locking. - */ - rcu_nocb_lock_irqsave(rdp, flags); - /* - * Theoretically we could clear SEGCBLIST_LOCKING after the nocb - * lock is released but how about being paranoid for once? - */ - rcu_segcblist_clear_flags(cblist, SEGCBLIST_LOCKING); - /* - * Without SEGCBLIST_LOCKING, we can't use - * rcu_nocb_unlock_irqrestore() anymore. - */ - raw_spin_unlock_irqrestore(&rdp->nocb_lock, flags); - - /* Sanity check */ - WARN_ON_ONCE(rcu_cblist_n_cbs(&rdp->nocb_bypass)); - - return 0; } @@ -1145,33 +1102,42 @@ int rcu_nocb_cpu_deoffload(int cpu) int ret = 0; cpus_read_lock(); - mutex_lock(&rcu_state.barrier_mutex); + mutex_lock(&rcu_state.nocb_mutex); if (rcu_rdp_is_offloaded(rdp)) { - if (cpu_online(cpu)) { - ret = work_on_cpu(cpu, rcu_nocb_rdp_deoffload, rdp); + if (!cpu_online(cpu)) { + ret = rcu_nocb_rdp_deoffload(rdp); if (!ret) cpumask_clear_cpu(cpu, rcu_nocb_mask); } else { - pr_info("NOCB: Cannot CB-deoffload offline CPU %d\n", rdp->cpu); + pr_info("NOCB: Cannot CB-deoffload online CPU %d\n", rdp->cpu); ret = -EINVAL; } } - mutex_unlock(&rcu_state.barrier_mutex); + mutex_unlock(&rcu_state.nocb_mutex); cpus_read_unlock(); return ret; } EXPORT_SYMBOL_GPL(rcu_nocb_cpu_deoffload); -static long rcu_nocb_rdp_offload(void *arg) +static bool rcu_nocb_rdp_offload_wait_cond(struct rcu_data *rdp) { - struct rcu_data *rdp = arg; - struct rcu_segcblist *cblist = &rdp->cblist; unsigned long flags; + bool ret; + + raw_spin_lock_irqsave(&rdp->nocb_lock, flags); + ret = rcu_segcblist_test_flags(&rdp->cblist, SEGCBLIST_OFFLOADED); + raw_spin_unlock_irqrestore(&rdp->nocb_lock, flags); + + return ret; +} + +static int rcu_nocb_rdp_offload(struct rcu_data *rdp) +{ int wake_gp; struct rcu_data *rdp_gp = rdp->nocb_gp_rdp; - WARN_ON_ONCE(rdp->cpu != raw_smp_processor_id()); + WARN_ON_ONCE(cpu_online(rdp->cpu)); /* * For now we only support re-offload, ie: the rdp must have been * offloaded on boot first. @@ -1184,44 +1150,17 @@ static long rcu_nocb_rdp_offload(void *arg) pr_info("Offloading %d\n", rdp->cpu); - /* - * Can't use rcu_nocb_lock_irqsave() before SEGCBLIST_LOCKING - * is set. - */ - raw_spin_lock_irqsave(&rdp->nocb_lock, flags); + WARN_ON_ONCE(rcu_cblist_n_cbs(&rdp->nocb_bypass)); + WARN_ON_ONCE(rcu_segcblist_n_cbs(&rdp->cblist)); - /* - * We didn't take the nocb lock while working on the - * rdp->cblist with SEGCBLIST_LOCKING cleared (pure softirq/rcuc mode). - * Every modifications that have been done previously on - * rdp->cblist must be visible remotely by the nocb kthreads - * upon wake up after reading the cblist flags. - * - * The layout against nocb_lock enforces that ordering: - * - * __rcu_nocb_rdp_offload() nocb_cb_wait()/nocb_gp_wait() - * ------------------------- ---------------------------- - * WRITE callbacks rcu_nocb_lock() - * rcu_nocb_lock() READ flags - * WRITE flags READ callbacks - * rcu_nocb_unlock() rcu_nocb_unlock() - */ - wake_gp = rdp_offload_toggle(rdp, true, flags); + wake_gp = rcu_nocb_queue_toggle_rdp(rdp); if (wake_gp) wake_up_process(rdp_gp->nocb_gp_kthread); - kthread_unpark(rdp->nocb_cb_kthread); - swait_event_exclusive(rdp->nocb_state_wq, - rcu_segcblist_test_flags(cblist, SEGCBLIST_KTHREAD_GP)); + rcu_nocb_rdp_offload_wait_cond(rdp)); - /* - * All kthreads are ready to work, we can finally relieve rcu_core() and - * enable nocb bypass. - */ - rcu_nocb_lock_irqsave(rdp, flags); - rcu_segcblist_clear_flags(cblist, SEGCBLIST_RCU_CORE); - rcu_nocb_unlock_irqrestore(rdp, flags); + kthread_unpark(rdp->nocb_cb_kthread); return 0; } @@ -1232,18 +1171,18 @@ int rcu_nocb_cpu_offload(int cpu) int ret = 0; cpus_read_lock(); - mutex_lock(&rcu_state.barrier_mutex); + mutex_lock(&rcu_state.nocb_mutex); if (!rcu_rdp_is_offloaded(rdp)) { - if (cpu_online(cpu)) { - ret = work_on_cpu(cpu, rcu_nocb_rdp_offload, rdp); + if (!cpu_online(cpu)) { + ret = rcu_nocb_rdp_offload(rdp); if (!ret) cpumask_set_cpu(cpu, rcu_nocb_mask); } else { - pr_info("NOCB: Cannot CB-offload offline CPU %d\n", rdp->cpu); + pr_info("NOCB: Cannot CB-offload online CPU %d\n", rdp->cpu); ret = -EINVAL; } } - mutex_unlock(&rcu_state.barrier_mutex); + mutex_unlock(&rcu_state.nocb_mutex); cpus_read_unlock(); return ret; @@ -1261,7 +1200,7 @@ lazy_rcu_shrink_count(struct shrinker *shrink, struct shrink_control *sc) return 0; /* Protect rcu_nocb_mask against concurrent (de-)offloading. */ - if (!mutex_trylock(&rcu_state.barrier_mutex)) + if (!mutex_trylock(&rcu_state.nocb_mutex)) return 0; /* Snapshot count of all CPUs */ @@ -1271,7 +1210,7 @@ lazy_rcu_shrink_count(struct shrinker *shrink, struct shrink_control *sc) count += READ_ONCE(rdp->lazy_len); } - mutex_unlock(&rcu_state.barrier_mutex); + mutex_unlock(&rcu_state.nocb_mutex); return count ? count : SHRINK_EMPTY; } @@ -1289,9 +1228,9 @@ lazy_rcu_shrink_scan(struct shrinker *shrink, struct shrink_control *sc) * Protect against concurrent (de-)offloading. Otherwise nocb locking * may be ignored or imbalanced. */ - if (!mutex_trylock(&rcu_state.barrier_mutex)) { + if (!mutex_trylock(&rcu_state.nocb_mutex)) { /* - * But really don't insist if barrier_mutex is contended since we + * But really don't insist if nocb_mutex is contended since we * can't guarantee that it will never engage in a dependency * chain involving memory allocation. The lock is seldom contended * anyway. @@ -1330,7 +1269,7 @@ lazy_rcu_shrink_scan(struct shrinker *shrink, struct shrink_control *sc) break; } - mutex_unlock(&rcu_state.barrier_mutex); + mutex_unlock(&rcu_state.nocb_mutex); return count ? count : SHRINK_STOP; } @@ -1396,9 +1335,7 @@ void __init rcu_init_nohz(void) rdp = per_cpu_ptr(&rcu_data, cpu); if (rcu_segcblist_empty(&rdp->cblist)) rcu_segcblist_init(&rdp->cblist); - rcu_segcblist_offload(&rdp->cblist, true); - rcu_segcblist_set_flags(&rdp->cblist, SEGCBLIST_KTHREAD_GP); - rcu_segcblist_clear_flags(&rdp->cblist, SEGCBLIST_RCU_CORE); + rcu_segcblist_set_flags(&rdp->cblist, SEGCBLIST_OFFLOADED); } rcu_organize_nocb_kthreads(); } @@ -1446,7 +1383,7 @@ static void rcu_spawn_cpu_nocb_kthread(int cpu) "rcuog/%d", rdp_gp->cpu); if (WARN_ONCE(IS_ERR(t), "%s: Could not start rcuo GP kthread, OOM is now expected behavior\n", __func__)) { mutex_unlock(&rdp_gp->nocb_gp_kthread_mutex); - goto end; + goto err; } WRITE_ONCE(rdp_gp->nocb_gp_kthread, t); if (kthread_prio) @@ -1458,7 +1395,7 @@ static void rcu_spawn_cpu_nocb_kthread(int cpu) t = kthread_create(rcu_nocb_cb_kthread, rdp, "rcuo%c/%d", rcu_state.abbr, cpu); if (WARN_ONCE(IS_ERR(t), "%s: Could not start rcuo CB kthread, OOM is now expected behavior\n", __func__)) - goto end; + goto err; if (rcu_rdp_is_offloaded(rdp)) wake_up_process(t); @@ -1471,13 +1408,21 @@ static void rcu_spawn_cpu_nocb_kthread(int cpu) WRITE_ONCE(rdp->nocb_cb_kthread, t); WRITE_ONCE(rdp->nocb_gp_kthread, rdp_gp->nocb_gp_kthread); return; -end: - mutex_lock(&rcu_state.barrier_mutex); + +err: + /* + * No need to protect against concurrent rcu_barrier() + * because the number of callbacks should be 0 for a non-boot CPU, + * therefore rcu_barrier() shouldn't even try to grab the nocb_lock. + * But hold nocb_mutex to avoid nocb_lock imbalance from shrinker. + */ + WARN_ON_ONCE(system_state > SYSTEM_BOOTING && rcu_segcblist_n_cbs(&rdp->cblist)); + mutex_lock(&rcu_state.nocb_mutex); if (rcu_rdp_is_offloaded(rdp)) { rcu_nocb_rdp_deoffload(rdp); cpumask_clear_cpu(cpu, rcu_nocb_mask); } - mutex_unlock(&rcu_state.barrier_mutex); + mutex_unlock(&rcu_state.nocb_mutex); } /* How many CB CPU IDs per GP kthread? Default of -1 for sqrt(nr_cpu_ids). */ @@ -1653,16 +1598,6 @@ static void show_rcu_nocb_state(struct rcu_data *rdp) #else /* #ifdef CONFIG_RCU_NOCB_CPU */ -static inline int rcu_lockdep_is_held_nocb(struct rcu_data *rdp) -{ - return 0; -} - -static inline bool rcu_current_is_nocb_kthread(struct rcu_data *rdp) -{ - return false; -} - /* No ->nocb_lock to acquire. */ static void rcu_nocb_lock(struct rcu_data *rdp) { diff --git a/kernel/rcu/tree_plugin.h b/kernel/rcu/tree_plugin.h index 354fbf2e411c..1c7cbd145d5e 100644 --- a/kernel/rcu/tree_plugin.h +++ b/kernel/rcu/tree_plugin.h @@ -24,10 +24,11 @@ static bool rcu_rdp_is_offloaded(struct rcu_data *rdp) * timers have their own means of synchronization against the * offloaded state updaters. */ - RCU_LOCKDEP_WARN( + RCU_NOCB_LOCKDEP_WARN( !(lockdep_is_held(&rcu_state.barrier_mutex) || (IS_ENABLED(CONFIG_HOTPLUG_CPU) && lockdep_is_cpus_held()) || - rcu_lockdep_is_held_nocb(rdp) || + lockdep_is_held(&rdp->nocb_lock) || + lockdep_is_held(&rcu_state.nocb_mutex) || (!(IS_ENABLED(CONFIG_PREEMPT_COUNT) && preemptible()) && rdp == this_cpu_ptr(&rcu_data)) || rcu_current_is_nocb_kthread(rdp)), diff --git a/kernel/rcu/tree_stall.h b/kernel/rcu/tree_stall.h index e933c6a58d5f..2fb40ec4b2ae 100644 --- a/kernel/rcu/tree_stall.h +++ b/kernel/rcu/tree_stall.h @@ -9,6 +9,7 @@ #include <linux/kvm_para.h> #include <linux/rcu_notifier.h> +#include <linux/smp.h> ////////////////////////////////////////////////////////////////////////////// // @@ -370,6 +371,7 @@ static void rcu_dump_cpu_stacks(void) struct rcu_node *rnp; rcu_for_each_leaf_node(rnp) { + printk_deferred_enter(); raw_spin_lock_irqsave_rcu_node(rnp, flags); for_each_leaf_node_possible_cpu(rnp, cpu) if (rnp->qsmask & leaf_node_cpu_bit(rnp, cpu)) { @@ -379,6 +381,7 @@ static void rcu_dump_cpu_stacks(void) dump_cpu_task(cpu); } raw_spin_unlock_irqrestore_rcu_node(rnp, flags); + printk_deferred_exit(); } } @@ -719,6 +722,9 @@ static void print_cpu_stall(unsigned long gps) set_preempt_need_resched(); } +static bool csd_lock_suppress_rcu_stall; +module_param(csd_lock_suppress_rcu_stall, bool, 0644); + static void check_cpu_stall(struct rcu_data *rdp) { bool self_detected; @@ -791,7 +797,9 @@ static void check_cpu_stall(struct rcu_data *rdp) return; rcu_stall_notifier_call_chain(RCU_STALL_NOTIFY_NORM, (void *)j - gps); - if (self_detected) { + if (READ_ONCE(csd_lock_suppress_rcu_stall) && csd_lock_is_stuck()) { + pr_err("INFO: %s detected stall, but suppressed full report due to a stuck CSD-lock.\n", rcu_state.name); + } else if (self_detected) { /* We haven't checked in, so go dump stack. */ print_cpu_stall(gps); } else { diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 93a845fe8449..6569d036ddb4 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -9726,7 +9726,7 @@ struct cgroup_subsys cpu_cgrp_subsys = { void dump_cpu_task(int cpu) { - if (cpu == smp_processor_id() && in_hardirq()) { + if (in_hardirq() && cpu == smp_processor_id()) { struct pt_regs *regs; regs = get_irq_regs(); diff --git a/kernel/smp.c b/kernel/smp.c index aaffecdad319..f25e20617b7e 100644 --- a/kernel/smp.c +++ b/kernel/smp.c @@ -208,12 +208,25 @@ static int csd_lock_wait_getcpu(call_single_data_t *csd) return -1; } +static atomic_t n_csd_lock_stuck; + +/** + * csd_lock_is_stuck - Has a CSD-lock acquisition been stuck too long? + * + * Returns @true if a CSD-lock acquisition is stuck and has been stuck + * long enough for a "non-responsive CSD lock" message to be printed. + */ +bool csd_lock_is_stuck(void) +{ + return !!atomic_read(&n_csd_lock_stuck); +} + /* * Complain if too much time spent waiting. Note that only * the CSD_TYPE_SYNC/ASYNC types provide the destination CPU, * so waiting on other types gets much less information. */ -static bool csd_lock_wait_toolong(call_single_data_t *csd, u64 ts0, u64 *ts1, int *bug_id) +static bool csd_lock_wait_toolong(call_single_data_t *csd, u64 ts0, u64 *ts1, int *bug_id, unsigned long *nmessages) { int cpu = -1; int cpux; @@ -229,15 +242,26 @@ static bool csd_lock_wait_toolong(call_single_data_t *csd, u64 ts0, u64 *ts1, in cpu = csd_lock_wait_getcpu(csd); pr_alert("csd: CSD lock (#%d) got unstuck on CPU#%02d, CPU#%02d released the lock.\n", *bug_id, raw_smp_processor_id(), cpu); + atomic_dec(&n_csd_lock_stuck); return true; } ts2 = sched_clock(); /* How long since we last checked for a stuck CSD lock.*/ ts_delta = ts2 - *ts1; - if (likely(ts_delta <= csd_lock_timeout_ns || csd_lock_timeout_ns == 0)) + if (likely(ts_delta <= csd_lock_timeout_ns * (*nmessages + 1) * + (!*nmessages ? 1 : (ilog2(num_online_cpus()) / 2 + 1)) || + csd_lock_timeout_ns == 0)) return false; + if (ts0 > ts2) { + /* Our own sched_clock went backward; don't blame another CPU. */ + ts_delta = ts0 - ts2; + pr_alert("sched_clock on CPU %d went backward by %llu ns\n", raw_smp_processor_id(), ts_delta); + *ts1 = ts2; + return false; + } + firsttime = !*bug_id; if (firsttime) *bug_id = atomic_inc_return(&csd_bug_count); @@ -249,9 +273,12 @@ static bool csd_lock_wait_toolong(call_single_data_t *csd, u64 ts0, u64 *ts1, in cpu_cur_csd = smp_load_acquire(&per_cpu(cur_csd, cpux)); /* Before func and info. */ /* How long since this CSD lock was stuck. */ ts_delta = ts2 - ts0; - pr_alert("csd: %s non-responsive CSD lock (#%d) on CPU#%d, waiting %llu ns for CPU#%02d %pS(%ps).\n", - firsttime ? "Detected" : "Continued", *bug_id, raw_smp_processor_id(), ts_delta, + pr_alert("csd: %s non-responsive CSD lock (#%d) on CPU#%d, waiting %lld ns for CPU#%02d %pS(%ps).\n", + firsttime ? "Detected" : "Continued", *bug_id, raw_smp_processor_id(), (s64)ts_delta, cpu, csd->func, csd->info); + (*nmessages)++; + if (firsttime) + atomic_inc(&n_csd_lock_stuck); /* * If the CSD lock is still stuck after 5 minutes, it is unlikely * to become unstuck. Use a signed comparison to avoid triggering @@ -290,12 +317,13 @@ static bool csd_lock_wait_toolong(call_single_data_t *csd, u64 ts0, u64 *ts1, in */ static void __csd_lock_wait(call_single_data_t *csd) { + unsigned long nmessages = 0; int bug_id = 0; u64 ts0, ts1; ts1 = ts0 = sched_clock(); for (;;) { - if (csd_lock_wait_toolong(csd, ts0, &ts1, &bug_id)) + if (csd_lock_wait_toolong(csd, ts0, &ts1, &bug_id, &nmessages)) break; cpu_relax(); } diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug index a30c03a66172..4e5f61cba8e4 100644 --- a/lib/Kconfig.debug +++ b/lib/Kconfig.debug @@ -1614,6 +1614,7 @@ config SCF_TORTURE_TEST config CSD_LOCK_WAIT_DEBUG bool "Debugging for csd_lock_wait(), called from smp_call_function*()" depends on DEBUG_KERNEL + depends on SMP depends on 64BIT default n help diff --git a/tools/rcu/rcu-updaters.sh b/tools/rcu/rcu-updaters.sh index 4ef1397927bb..8a5df3f22550 100755 --- a/tools/rcu/rcu-updaters.sh +++ b/tools/rcu/rcu-updaters.sh @@ -21,12 +21,10 @@ fi bpftrace -e 'kprobe:kvfree_call_rcu, kprobe:call_rcu, kprobe:call_rcu_tasks, - kprobe:call_rcu_tasks_rude, kprobe:call_rcu_tasks_trace, kprobe:call_srcu, kprobe:rcu_barrier, kprobe:rcu_barrier_tasks, - kprobe:rcu_barrier_tasks_rude, kprobe:rcu_barrier_tasks_trace, kprobe:srcu_barrier, kprobe:synchronize_rcu, diff --git a/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh b/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh index b33cd8753689..ad79784e552d 100755 --- a/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh +++ b/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh @@ -68,6 +68,8 @@ config_override_param "--gdb options" KcList "$TORTURE_KCONFIG_GDB_ARG" config_override_param "--kasan options" KcList "$TORTURE_KCONFIG_KASAN_ARG" config_override_param "--kcsan options" KcList "$TORTURE_KCONFIG_KCSAN_ARG" config_override_param "--kconfig argument" KcList "$TORTURE_KCONFIG_ARG" +config_override_param "$config_dir/CFcommon.$(uname -m)" KcList \ + "`cat $config_dir/CFcommon.$(uname -m) 2> /dev/null`" cp $T/KcList $resdir/ConfigFragment base_resdir=`echo $resdir | sed -e 's/\.[0-9]\+$//'` diff --git a/tools/testing/selftests/rcutorture/bin/torture.sh b/tools/testing/selftests/rcutorture/bin/torture.sh index 990d24696fd3..0447c4a00cc4 100755 --- a/tools/testing/selftests/rcutorture/bin/torture.sh +++ b/tools/testing/selftests/rcutorture/bin/torture.sh @@ -19,10 +19,10 @@ PATH=${RCUTORTURE}/bin:$PATH; export PATH TORTURE_ALLOTED_CPUS="`identify_qemu_vcpus`" MAKE_ALLOTED_CPUS=$((TORTURE_ALLOTED_CPUS*2)) -HALF_ALLOTED_CPUS=$((TORTURE_ALLOTED_CPUS/2)) -if test "$HALF_ALLOTED_CPUS" -lt 1 +SCALE_ALLOTED_CPUS=$((TORTURE_ALLOTED_CPUS/2)) +if test "$SCALE_ALLOTED_CPUS" -lt 1 then - HALF_ALLOTED_CPUS=1 + SCALE_ALLOTED_CPUS=1 fi VERBOSE_BATCH_CPUS=$((TORTURE_ALLOTED_CPUS/16)) if test "$VERBOSE_BATCH_CPUS" -lt 2 @@ -90,6 +90,7 @@ usage () { echo " --do-scftorture / --do-no-scftorture / --no-scftorture" echo " --do-srcu-lockdep / --do-no-srcu-lockdep / --no-srcu-lockdep" echo " --duration [ <minutes> | <hours>h | <days>d ]" + echo " --guest-cpu-limit N" echo " --kcsan-kmake-arg kernel-make-arguments" exit 1 } @@ -203,6 +204,21 @@ do duration_base=$(($ts*mult)) shift ;; + --guest-cpu-limit|--guest-cpu-lim) + checkarg --guest-cpu-limit "(number)" "$#" "$2" '^[0-9]*$' '^--' + if (("$2" <= "$TORTURE_ALLOTED_CPUS" / 2)) + then + SCALE_ALLOTED_CPUS="$2" + VERBOSE_BATCH_CPUS="$((SCALE_ALLOTED_CPUS/8))" + if (("$VERBOSE_BATCH_CPUS" < 2)) + then + VERBOSE_BATCH_CPUS=0 + fi + else + echo "Ignoring value of $2 for --guest-cpu-limit which is greater than (("$TORTURE_ALLOTED_CPUS" / 2))." + fi + shift + ;; --kcsan-kmake-arg|--kcsan-kmake-args) checkarg --kcsan-kmake-arg "(kernel make arguments)" $# "$2" '.*' '^error$' kcsan_kmake_args="`echo "$kcsan_kmake_args $2" | sed -e 's/^ *//' -e 's/ *$//'`" @@ -425,9 +441,9 @@ fi if test "$do_scftorture" = "yes" then # Scale memory based on the number of CPUs. - scfmem=$((3+HALF_ALLOTED_CPUS/16)) - torture_bootargs="scftorture.nthreads=$HALF_ALLOTED_CPUS torture.disable_onoff_at_boot csdlock_debug=1" - torture_set "scftorture" tools/testing/selftests/rcutorture/bin/kvm.sh --torture scf --allcpus --duration "$duration_scftorture" --configs "$configs_scftorture" --kconfig "CONFIG_NR_CPUS=$HALF_ALLOTED_CPUS" --memory ${scfmem}G --trust-make + scfmem=$((3+SCALE_ALLOTED_CPUS/16)) + torture_bootargs="scftorture.nthreads=$SCALE_ALLOTED_CPUS torture.disable_onoff_at_boot csdlock_debug=1" + torture_set "scftorture" tools/testing/selftests/rcutorture/bin/kvm.sh --torture scf --allcpus --duration "$duration_scftorture" --configs "$configs_scftorture" --kconfig "CONFIG_NR_CPUS=$SCALE_ALLOTED_CPUS" --memory ${scfmem}G --trust-make fi if test "$do_rt" = "yes" @@ -471,8 +487,8 @@ for prim in $primlist do if test -n "$firsttime" then - torture_bootargs="refscale.scale_type="$prim" refscale.nreaders=$HALF_ALLOTED_CPUS refscale.loops=10000 refscale.holdoff=20 torture.disable_onoff_at_boot" - torture_set "refscale-$prim" tools/testing/selftests/rcutorture/bin/kvm.sh --torture refscale --allcpus --duration 5 --kconfig "CONFIG_TASKS_TRACE_RCU=y CONFIG_NR_CPUS=$HALF_ALLOTED_CPUS" --bootargs "refscale.verbose_batched=$VERBOSE_BATCH_CPUS torture.verbose_sleep_frequency=8 torture.verbose_sleep_duration=$VERBOSE_BATCH_CPUS" --trust-make + torture_bootargs="refscale.scale_type="$prim" refscale.nreaders=$SCALE_ALLOTED_CPUS refscale.loops=10000 refscale.holdoff=20 torture.disable_onoff_at_boot" + torture_set "refscale-$prim" tools/testing/selftests/rcutorture/bin/kvm.sh --torture refscale --allcpus --duration 5 --kconfig "CONFIG_TASKS_TRACE_RCU=y CONFIG_NR_CPUS=$SCALE_ALLOTED_CPUS" --bootargs "refscale.verbose_batched=$VERBOSE_BATCH_CPUS torture.verbose_sleep_frequency=8 torture.verbose_sleep_duration=$VERBOSE_BATCH_CPUS" --trust-make mv $T/last-resdir-nodebug $T/first-resdir-nodebug || : if test -f "$T/last-resdir-kasan" then @@ -520,8 +536,8 @@ for prim in $primlist do if test -n "$firsttime" then - torture_bootargs="rcuscale.scale_type="$prim" rcuscale.nwriters=$HALF_ALLOTED_CPUS rcuscale.holdoff=20 torture.disable_onoff_at_boot" - torture_set "rcuscale-$prim" tools/testing/selftests/rcutorture/bin/kvm.sh --torture rcuscale --allcpus --duration 5 --kconfig "CONFIG_TASKS_TRACE_RCU=y CONFIG_NR_CPUS=$HALF_ALLOTED_CPUS" --trust-make + torture_bootargs="rcuscale.scale_type="$prim" rcuscale.nwriters=$SCALE_ALLOTED_CPUS rcuscale.holdoff=20 torture.disable_onoff_at_boot" + torture_set "rcuscale-$prim" tools/testing/selftests/rcutorture/bin/kvm.sh --torture rcuscale --allcpus --duration 5 --kconfig "CONFIG_TASKS_TRACE_RCU=y CONFIG_NR_CPUS=$SCALE_ALLOTED_CPUS" --trust-make mv $T/last-resdir-nodebug $T/first-resdir-nodebug || : if test -f "$T/last-resdir-kasan" then @@ -559,7 +575,7 @@ do_kcsan="$do_kcsan_save" if test "$do_kvfree" = "yes" then torture_bootargs="rcuscale.kfree_rcu_test=1 rcuscale.kfree_nthreads=16 rcuscale.holdoff=20 rcuscale.kfree_loops=10000 torture.disable_onoff_at_boot" - torture_set "rcuscale-kvfree" tools/testing/selftests/rcutorture/bin/kvm.sh --torture rcuscale --allcpus --duration $duration_rcutorture --kconfig "CONFIG_NR_CPUS=$HALF_ALLOTED_CPUS" --memory 2G --trust-make + torture_set "rcuscale-kvfree" tools/testing/selftests/rcutorture/bin/kvm.sh --torture rcuscale --allcpus --duration $duration_rcutorture --kconfig "CONFIG_NR_CPUS=$SCALE_ALLOTED_CPUS" --memory 2G --trust-make fi if test "$do_clocksourcewd" = "yes" diff --git a/tools/testing/selftests/rcutorture/configs/rcu/CFcommon b/tools/testing/selftests/rcutorture/configs/rcu/CFcommon index 0e92d85313aa..217597e84905 100644 --- a/tools/testing/selftests/rcutorture/configs/rcu/CFcommon +++ b/tools/testing/selftests/rcutorture/configs/rcu/CFcommon @@ -1,7 +1,5 @@ CONFIG_RCU_TORTURE_TEST=y CONFIG_PRINTK_TIME=y -CONFIG_HYPERVISOR_GUEST=y CONFIG_PARAVIRT=y -CONFIG_KVM_GUEST=y CONFIG_KCSAN_ASSUME_PLAIN_WRITES_ATOMIC=n CONFIG_KCSAN_REPORT_VALUE_CHANGE_ONLY=n diff --git a/tools/testing/selftests/rcutorture/configs/rcu/CFcommon.i686 b/tools/testing/selftests/rcutorture/configs/rcu/CFcommon.i686 new file mode 100644 index 000000000000..d8b2f555686f --- /dev/null +++ b/tools/testing/selftests/rcutorture/configs/rcu/CFcommon.i686 @@ -0,0 +1,2 @@ +CONFIG_HYPERVISOR_GUEST=y +CONFIG_KVM_GUEST=y diff --git a/tools/testing/selftests/rcutorture/configs/rcu/CFcommon.ppc64le b/tools/testing/selftests/rcutorture/configs/rcu/CFcommon.ppc64le new file mode 100644 index 000000000000..133da04247ee --- /dev/null +++ b/tools/testing/selftests/rcutorture/configs/rcu/CFcommon.ppc64le @@ -0,0 +1 @@ +CONFIG_KVM_GUEST=y diff --git a/tools/testing/selftests/rcutorture/configs/rcu/CFcommon.x86_64 b/tools/testing/selftests/rcutorture/configs/rcu/CFcommon.x86_64 new file mode 100644 index 000000000000..d8b2f555686f --- /dev/null +++ b/tools/testing/selftests/rcutorture/configs/rcu/CFcommon.x86_64 @@ -0,0 +1,2 @@ +CONFIG_HYPERVISOR_GUEST=y +CONFIG_KVM_GUEST=y diff --git a/tools/testing/selftests/rcutorture/configs/rcu/TREE07.boot b/tools/testing/selftests/rcutorture/configs/rcu/TREE07.boot index 979edbf4c820..55ce305b2a3d 100644 --- a/tools/testing/selftests/rcutorture/configs/rcu/TREE07.boot +++ b/tools/testing/selftests/rcutorture/configs/rcu/TREE07.boot @@ -2,3 +2,4 @@ nohz_full=2-9 rcutorture.stall_cpu=14 rcutorture.stall_cpu_holdoff=90 rcutorture.fwd_progress=0 +rcutree.nohz_full_patience_delay=1000 diff --git a/tools/testing/selftests/rcutorture/configs/refscale/TINY b/tools/testing/selftests/rcutorture/configs/refscale/TINY new file mode 100644 index 000000000000..759343980b80 --- /dev/null +++ b/tools/testing/selftests/rcutorture/configs/refscale/TINY @@ -0,0 +1,20 @@ +CONFIG_SMP=n +CONFIG_PREEMPT_NONE=y +CONFIG_PREEMPT_VOLUNTARY=n +CONFIG_PREEMPT=n +CONFIG_PREEMPT_DYNAMIC=n +#CHECK#CONFIG_PREEMPT_RCU=n +CONFIG_HZ_PERIODIC=n +CONFIG_NO_HZ_IDLE=y +CONFIG_NO_HZ_FULL=n +CONFIG_HOTPLUG_CPU=n +CONFIG_SUSPEND=n +CONFIG_HIBERNATION=n +CONFIG_RCU_NOCB_CPU=n +CONFIG_DEBUG_LOCK_ALLOC=n +CONFIG_PROVE_LOCKING=n +CONFIG_RCU_BOOST=n +CONFIG_DEBUG_OBJECTS_RCU_HEAD=n +CONFIG_RCU_EXPERT=y +CONFIG_KPROBES=n +CONFIG_FTRACE=n