diff options
23 files changed, 421 insertions, 86 deletions
diff --git a/Documentation/RCU/lockdep-splat.rst b/Documentation/RCU/lockdep-splat.rst index 2a5c79db57dc..bcbc4b3c88d7 100644 --- a/Documentation/RCU/lockdep-splat.rst +++ b/Documentation/RCU/lockdep-splat.rst @@ -10,7 +10,7 @@ misuses of the RCU API, most notably using one of the rcu_dereference() family to access an RCU-protected pointer without the proper protection. When such misuse is detected, an lockdep-RCU splat is emitted. -The usual cause of a lockdep-RCU slat is someone accessing an +The usual cause of a lockdep-RCU splat is someone accessing an RCU-protected data structure without either (1) being in the right kind of RCU read-side critical section or (2) holding the right update-side lock. This problem can therefore be serious: it might result in random memory diff --git a/Documentation/RCU/rculist_nulls.rst b/Documentation/RCU/rculist_nulls.rst index 9a734bf54b76..21e40fcc08de 100644 --- a/Documentation/RCU/rculist_nulls.rst +++ b/Documentation/RCU/rculist_nulls.rst @@ -18,7 +18,16 @@ to solve following problem. Without 'nulls', a typical RCU linked list managing objects which are allocated with SLAB_TYPESAFE_BY_RCU kmem_cache can use the following -algorithms: +algorithms. Following examples assume 'obj' is a pointer to such +objects, which is having below type. + +:: + + struct object { + struct hlist_node obj_node; + atomic_t refcnt; + unsigned int key; + }; 1) Lookup algorithm ------------------- @@ -26,11 +35,13 @@ algorithms: :: begin: - rcu_read_lock() + rcu_read_lock(); obj = lockless_lookup(key); if (obj) { - if (!try_get_ref(obj)) // might fail for free objects + if (!try_get_ref(obj)) { // might fail for free objects + rcu_read_unlock(); goto begin; + } /* * Because a writer could delete object, and a writer could * reuse these object before the RCU grace period, we @@ -54,7 +65,7 @@ but a version with an additional memory barrier (smp_rmb()) struct hlist_node *node, *next; for (pos = rcu_dereference((head)->first); pos && ({ next = pos->next; smp_rmb(); prefetch(next); 1; }) && - ({ tpos = hlist_entry(pos, typeof(*tpos), member); 1; }); + ({ obj = hlist_entry(pos, typeof(*obj), obj_node); 1; }); pos = rcu_dereference(next)) if (obj->key == key) return obj; @@ -66,10 +77,10 @@ And note the traditional hlist_for_each_entry_rcu() misses this smp_rmb():: struct hlist_node *node; for (pos = rcu_dereference((head)->first); pos && ({ prefetch(pos->next); 1; }) && - ({ tpos = hlist_entry(pos, typeof(*tpos), member); 1; }); + ({ obj = hlist_entry(pos, typeof(*obj), obj_node); 1; }); pos = rcu_dereference(pos->next)) - if (obj->key == key) - return obj; + if (obj->key == key) + return obj; return NULL; Quoting Corey Minyard:: @@ -86,7 +97,7 @@ Quoting Corey Minyard:: 2) Insertion algorithm ---------------------- -We need to make sure a reader cannot read the new 'obj->obj_next' value +We need to make sure a reader cannot read the new 'obj->obj_node.next' value and previous value of 'obj->key'. Otherwise, an item could be deleted from a chain, and inserted into another chain. If new chain was empty before the move, 'next' pointer is NULL, and lockless reader can not @@ -129,8 +140,7 @@ very very fast (before the end of RCU grace period) Avoiding extra smp_rmb() ======================== -With hlist_nulls we can avoid extra smp_rmb() in lockless_lookup() -and extra _release() in insert function. +With hlist_nulls we can avoid extra smp_rmb() in lockless_lookup(). For example, if we choose to store the slot number as the 'nulls' end-of-list marker for each slot of the hash table, we can detect @@ -142,6 +152,9 @@ the beginning. If the object was moved to the same chain, then the reader doesn't care: It might occasionally scan the list again without harm. +Note that using hlist_nulls means the type of 'obj_node' field of +'struct object' becomes 'struct hlist_nulls_node'. + 1) lookup algorithm ------------------- @@ -151,7 +164,7 @@ scan the list again without harm. head = &table[slot]; begin: rcu_read_lock(); - hlist_nulls_for_each_entry_rcu(obj, node, head, member) { + hlist_nulls_for_each_entry_rcu(obj, node, head, obj_node) { if (obj->key == key) { if (!try_get_ref(obj)) { // might fail for free objects rcu_read_unlock(); @@ -182,6 +195,9 @@ scan the list again without harm. 2) Insert algorithm ------------------- +Same to the above one, but uses hlist_nulls_add_head_rcu() instead of +hlist_add_head_rcu(). + :: /* diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt index a1457995fd41..3a6e30dc82f3 100644 --- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -2918,6 +2918,10 @@ locktorture.torture_type= [KNL] Specify the locking implementation to test. + locktorture.writer_fifo= [KNL] + Run the write-side locktorture kthreads at + sched_set_fifo() real-time priority. + locktorture.verbose= [KNL] Enable additional printk() statements. @@ -4928,6 +4932,15 @@ test until boot completes in order to avoid interference. + rcuscale.kfree_by_call_rcu= [KNL] + In kernels built with CONFIG_RCU_LAZY=y, test + call_rcu() instead of kfree_rcu(). + + rcuscale.kfree_mult= [KNL] + Instead of allocating an object of size kfree_obj, + allocate one of kfree_mult * sizeof(kfree_obj). + Defaults to 1. + rcuscale.kfree_rcu_test= [KNL] Set to measure performance of kfree_rcu() flooding. @@ -4953,6 +4966,12 @@ Number of loops doing rcuscale.kfree_alloc_num number of allocations and frees. + rcuscale.minruntime= [KNL] + Set the minimum test run time in seconds. This + does not affect the data-collection interval, + but instead allows better measurement of things + like CPU consumption. + rcuscale.nreaders= [KNL] Set number of RCU readers. The value -1 selects N, where N is the number of CPUs. A value @@ -4967,7 +4986,7 @@ the same as for rcuscale.nreaders. N, where N is the number of CPUs - rcuscale.perf_type= [KNL] + rcuscale.scale_type= [KNL] Specify the RCU implementation to test. rcuscale.shutdown= [KNL] @@ -4983,6 +5002,11 @@ in microseconds. The default of zero says no holdoff. + rcuscale.writer_holdoff_jiffies= [KNL] + Additional write-side holdoff between grace + periods, but in jiffies. The default of zero + says no holdoff. + rcutorture.fqs_duration= [KNL] Set duration of force_quiescent_state bursts in microseconds. @@ -5264,6 +5288,13 @@ number avoids disturbing real-time workloads, but lengthens grace periods. + rcupdate.rcu_task_lazy_lim= [KNL] + Number of callbacks on a given CPU that will + cancel laziness on that CPU. Use -1 to disable + cancellation of laziness, but be advised that + doing so increases the danger of OOM due to + callback flooding. + rcupdate.rcu_task_stall_info= [KNL] Set initial timeout in jiffies for RCU task stall informational messages, which give some indication @@ -5293,6 +5324,29 @@ A change in value does not take effect until the beginning of the next grace period. + rcupdate.rcu_tasks_lazy_ms= [KNL] + Set timeout in milliseconds RCU Tasks asynchronous + callback batching for call_rcu_tasks(). + A negative value will take the default. A value + 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 + call_rcu_tasks_trace(). A negative value + will take the default. A value of zero will + disable batching. Batching is always disabled + for synchronize_rcu_tasks_trace(). + rcupdate.rcu_self_test= [KNL] Run the RCU early boot self tests diff --git a/include/linux/notifier.h b/include/linux/notifier.h index 86544707236a..45702bdcbceb 100644 --- a/include/linux/notifier.h +++ b/include/linux/notifier.h @@ -73,9 +73,7 @@ struct raw_notifier_head { struct srcu_notifier_head { struct mutex mutex; -#ifdef CONFIG_TREE_SRCU struct srcu_usage srcuu; -#endif struct srcu_struct srcu; struct notifier_block __rcu *head; }; @@ -106,7 +104,6 @@ extern void srcu_init_notifier_head(struct srcu_notifier_head *nh); #define RAW_NOTIFIER_INIT(name) { \ .head = NULL } -#ifdef CONFIG_TREE_SRCU #define SRCU_NOTIFIER_INIT(name, pcpu) \ { \ .mutex = __MUTEX_INITIALIZER(name.mutex), \ @@ -114,14 +111,6 @@ extern void srcu_init_notifier_head(struct srcu_notifier_head *nh); .srcuu = __SRCU_USAGE_INIT(name.srcuu), \ .srcu = __SRCU_STRUCT_INIT(name.srcu, name.srcuu, pcpu), \ } -#else -#define SRCU_NOTIFIER_INIT(name, pcpu) \ - { \ - .mutex = __MUTEX_INITIALIZER(name.mutex), \ - .head = NULL, \ - .srcu = __SRCU_STRUCT_INIT(name.srcu, name.srcuu, pcpu), \ - } -#endif #define ATOMIC_NOTIFIER_HEAD(name) \ struct atomic_notifier_head name = \ diff --git a/include/linux/rculist_nulls.h b/include/linux/rculist_nulls.h index ba4c00dd8005..89186c499dd4 100644 --- a/include/linux/rculist_nulls.h +++ b/include/linux/rculist_nulls.h @@ -101,7 +101,7 @@ static inline void hlist_nulls_add_head_rcu(struct hlist_nulls_node *n, { struct hlist_nulls_node *first = h->first; - n->next = first; + WRITE_ONCE(n->next, first); WRITE_ONCE(n->pprev, &h->first); rcu_assign_pointer(hlist_nulls_first_rcu(h), n); if (!is_a_nulls(first)) @@ -137,7 +137,7 @@ static inline void hlist_nulls_add_tail_rcu(struct hlist_nulls_node *n, last = i; if (last) { - n->next = last->next; + WRITE_ONCE(n->next, last->next); n->pprev = &last->next; rcu_assign_pointer(hlist_nulls_next_rcu(last), n); } else { diff --git a/include/linux/rcupdate_trace.h b/include/linux/rcupdate_trace.h index 9bc8cbb33340..eda493200663 100644 --- a/include/linux/rcupdate_trace.h +++ b/include/linux/rcupdate_trace.h @@ -87,6 +87,7 @@ static inline void rcu_read_unlock_trace(void) void call_rcu_tasks_trace(struct rcu_head *rhp, rcu_callback_t func); void synchronize_rcu_tasks_trace(void); void rcu_barrier_tasks_trace(void); +struct task_struct *get_rcu_tasks_trace_gp_kthread(void); #else /* * The BPF JIT forms these addresses even when it doesn't call these diff --git a/include/linux/rcupdate_wait.h b/include/linux/rcupdate_wait.h index 699b938358bf..5e0f74f2f8ca 100644 --- a/include/linux/rcupdate_wait.h +++ b/include/linux/rcupdate_wait.h @@ -42,6 +42,11 @@ do { \ * call_srcu() function, with this wrapper supplying the pointer to the * corresponding srcu_struct. * + * Note that call_rcu_hurry() should be used instead of call_rcu() + * because in kernels built with CONFIG_RCU_LAZY=y the delay between the + * invocation of call_rcu() and that of the corresponding RCU callback + * can be multiple seconds. + * * The first argument tells Tiny RCU's _wait_rcu_gp() not to * bother waiting for RCU. The reason for this is because anywhere * synchronize_rcu_mult() can be called is automatically already a full diff --git a/include/linux/srcutiny.h b/include/linux/srcutiny.h index ebd72491af99..447133171d95 100644 --- a/include/linux/srcutiny.h +++ b/include/linux/srcutiny.h @@ -48,6 +48,10 @@ void srcu_drive_gp(struct work_struct *wp); #define DEFINE_STATIC_SRCU(name) \ static struct srcu_struct name = __SRCU_STRUCT_INIT(name, name, name) +// Dummy structure for srcu_notifier_head. +struct srcu_usage { }; +#define __SRCU_USAGE_INIT(name) { } + void synchronize_srcu(struct srcu_struct *ssp); /* diff --git a/include/linux/torture.h b/include/linux/torture.h index 7038104463e4..bb466eec01e4 100644 --- a/include/linux/torture.h +++ b/include/linux/torture.h @@ -108,12 +108,15 @@ bool torture_must_stop(void); bool torture_must_stop_irq(void); void torture_kthread_stopping(char *title); int _torture_create_kthread(int (*fn)(void *arg), void *arg, char *s, char *m, - char *f, struct task_struct **tp); + char *f, struct task_struct **tp, void (*cbf)(struct task_struct *tp)); void _torture_stop_kthread(char *m, struct task_struct **tp); #define torture_create_kthread(n, arg, tp) \ _torture_create_kthread(n, (arg), #n, "Creating " #n " task", \ - "Failed to create " #n, &(tp)) + "Failed to create " #n, &(tp), NULL) +#define torture_create_kthread_cb(n, arg, tp, cbf) \ + _torture_create_kthread(n, (arg), #n, "Creating " #n " task", \ + "Failed to create " #n, &(tp), cbf) #define torture_stop_kthread(n, tp) \ _torture_stop_kthread("Stopping " #n " task", &(tp)) diff --git a/kernel/locking/locktorture.c b/kernel/locking/locktorture.c index 949d3deae506..270c7f80ce84 100644 --- a/kernel/locking/locktorture.c +++ b/kernel/locking/locktorture.c @@ -45,6 +45,7 @@ torture_param(int, stutter, 5, "Number of jiffies to run/halt test, 0=disable"); torture_param(int, rt_boost, 2, "Do periodic rt-boost. 0=Disable, 1=Only for rt_mutex, 2=For all lock types."); torture_param(int, rt_boost_factor, 50, "A factor determining how often rt-boost happens."); +torture_param(int, writer_fifo, 0, "Run writers at sched_set_fifo() priority"); torture_param(int, verbose, 1, "Enable verbose debugging printk()s"); torture_param(int, nested_locks, 0, "Number of nested locks (max = 8)"); /* Going much higher trips "BUG: MAX_LOCKDEP_CHAIN_HLOCKS too low!" errors */ @@ -809,7 +810,8 @@ static int lock_torture_writer(void *arg) bool skip_main_lock; VERBOSE_TOROUT_STRING("lock_torture_writer task started"); - set_user_nice(current, MAX_NICE); + if (!rt_task(current)) + set_user_nice(current, MAX_NICE); do { if ((torture_random(&rand) & 0xfffff) == 0) @@ -1015,8 +1017,7 @@ static void lock_torture_cleanup(void) if (writer_tasks) { for (i = 0; i < cxt.nrealwriters_stress; i++) - torture_stop_kthread(lock_torture_writer, - writer_tasks[i]); + torture_stop_kthread(lock_torture_writer, writer_tasks[i]); kfree(writer_tasks); writer_tasks = NULL; } @@ -1244,8 +1245,9 @@ static int __init lock_torture_init(void) goto create_reader; /* Create writer. */ - firsterr = torture_create_kthread(lock_torture_writer, &cxt.lwsa[i], - writer_tasks[i]); + firsterr = torture_create_kthread_cb(lock_torture_writer, &cxt.lwsa[i], + writer_tasks[i], + writer_fifo ? sched_set_fifo : NULL); if (torture_init_error(firsterr)) goto unwind; diff --git a/kernel/rcu/rcu.h b/kernel/rcu/rcu.h index 98c1544cf572..5befd8780dcd 100644 --- a/kernel/rcu/rcu.h +++ b/kernel/rcu/rcu.h @@ -511,6 +511,14 @@ static inline void show_rcu_tasks_gp_kthreads(void) {} void rcu_request_urgent_qs_task(struct task_struct *t); #endif /* #else #ifdef CONFIG_TINY_RCU */ +#ifdef CONFIG_TASKS_RCU +struct task_struct *get_rcu_tasks_gp_kthread(void); +#endif // # ifdef CONFIG_TASKS_RCU + +#ifdef CONFIG_TASKS_RUDE_RCU +struct task_struct *get_rcu_tasks_rude_gp_kthread(void); +#endif // # ifdef CONFIG_TASKS_RUDE_RCU + #define RCU_SCHEDULER_INACTIVE 0 #define RCU_SCHEDULER_INIT 1 #define RCU_SCHEDULER_RUNNING 2 diff --git a/kernel/rcu/rcuscale.c b/kernel/rcu/rcuscale.c index d1221731c7cf..ffdb30495e3c 100644 --- a/kernel/rcu/rcuscale.c +++ b/kernel/rcu/rcuscale.c @@ -84,15 +84,17 @@ MODULE_AUTHOR("Paul E. McKenney <[email protected]>"); #endif torture_param(bool, gp_async, false, "Use asynchronous GP wait primitives"); -torture_param(int, gp_async_max, 1000, "Max # outstanding waits per reader"); +torture_param(int, gp_async_max, 1000, "Max # outstanding waits per writer"); torture_param(bool, gp_exp, false, "Use expedited GP wait primitives"); torture_param(int, holdoff, 10, "Holdoff time before test start (s)"); +torture_param(int, minruntime, 0, "Minimum run time (s)"); torture_param(int, nreaders, -1, "Number of RCU reader threads"); torture_param(int, nwriters, -1, "Number of RCU updater threads"); torture_param(bool, shutdown, RCUSCALE_SHUTDOWN, "Shutdown at end of scalability tests."); torture_param(int, verbose, 1, "Enable verbose debugging printk()s"); torture_param(int, writer_holdoff, 0, "Holdoff (us) between GPs, zero to disable"); +torture_param(int, writer_holdoff_jiffies, 0, "Holdoff (jiffies) between GPs, zero to disable"); torture_param(int, kfree_rcu_test, 0, "Do we run a kfree_rcu() scale test?"); torture_param(int, kfree_mult, 1, "Multiple of kfree_obj size to allocate."); torture_param(int, kfree_by_call_rcu, 0, "Use call_rcu() to emulate kfree_rcu()?"); @@ -139,6 +141,7 @@ struct rcu_scale_ops { void (*gp_barrier)(void); void (*sync)(void); void (*exp_sync)(void); + struct task_struct *(*rso_gp_kthread)(void); const char *name; }; @@ -295,6 +298,7 @@ static struct rcu_scale_ops tasks_ops = { .gp_barrier = rcu_barrier_tasks, .sync = synchronize_rcu_tasks, .exp_sync = synchronize_rcu_tasks, + .rso_gp_kthread = get_rcu_tasks_gp_kthread, .name = "tasks" }; @@ -306,6 +310,44 @@ static struct rcu_scale_ops tasks_ops = { #endif // #else // #ifdef CONFIG_TASKS_RCU +#ifdef CONFIG_TASKS_RUDE_RCU + +/* + * Definitions for RCU-tasks-rude scalability testing. + */ + +static int tasks_rude_scale_read_lock(void) +{ + return 0; +} + +static void tasks_rude_scale_read_unlock(int idx) +{ +} + +static struct rcu_scale_ops tasks_rude_ops = { + .ptype = RCU_TASKS_RUDE_FLAVOR, + .init = rcu_sync_scale_init, + .readlock = tasks_rude_scale_read_lock, + .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, + .name = "tasks-rude" +}; + +#define TASKS_RUDE_OPS &tasks_rude_ops, + +#else // #ifdef CONFIG_TASKS_RUDE_RCU + +#define TASKS_RUDE_OPS + +#endif // #else // #ifdef CONFIG_TASKS_RUDE_RCU + #ifdef CONFIG_TASKS_TRACE_RCU /* @@ -334,6 +376,7 @@ static struct rcu_scale_ops tasks_tracing_ops = { .gp_barrier = rcu_barrier_tasks_trace, .sync = synchronize_rcu_tasks_trace, .exp_sync = synchronize_rcu_tasks_trace, + .rso_gp_kthread = get_rcu_tasks_trace_gp_kthread, .name = "tasks-tracing" }; @@ -410,10 +453,12 @@ rcu_scale_writer(void *arg) { int i = 0; int i_max; + unsigned long jdone; long me = (long)arg; struct rcu_head *rhp = NULL; bool started = false, done = false, alldone = false; u64 t; + DEFINE_TORTURE_RANDOM(tr); u64 *wdp; u64 *wdpp = writer_durations[me]; @@ -424,7 +469,7 @@ rcu_scale_writer(void *arg) sched_set_fifo_low(current); if (holdoff) - schedule_timeout_uninterruptible(holdoff * HZ); + schedule_timeout_idle(holdoff * HZ); /* * Wait until rcu_end_inkernel_boot() is called for normal GP tests @@ -445,9 +490,12 @@ rcu_scale_writer(void *arg) } } + jdone = jiffies + minruntime * HZ; do { 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) { @@ -475,7 +523,7 @@ retry: if (!started && atomic_read(&n_rcu_scale_writer_started) >= nrealwriters) started = true; - if (!done && i >= MIN_MEAS) { + if (!done && i >= MIN_MEAS && time_after(jiffies, jdone)) { done = true; sched_set_normal(current, 0); pr_alert("%s%s rcu_scale_writer %ld has %d measurements\n", @@ -518,8 +566,8 @@ static void rcu_scale_print_module_parms(struct rcu_scale_ops *cur_ops, const char *tag) { pr_alert("%s" SCALE_FLAG - "--- %s: nreaders=%d nwriters=%d verbose=%d shutdown=%d\n", - scale_type, tag, nrealreaders, nrealwriters, verbose, shutdown); + "--- %s: gp_async=%d gp_async_max=%d gp_exp=%d holdoff=%d minruntime=%d nreaders=%d nwriters=%d writer_holdoff=%d writer_holdoff_jiffies=%d verbose=%d shutdown=%d\n", + scale_type, tag, gp_async, gp_async_max, gp_exp, holdoff, minruntime, nrealreaders, nrealwriters, writer_holdoff, writer_holdoff_jiffies, verbose, shutdown); } /* @@ -556,6 +604,8 @@ static struct task_struct **kfree_reader_tasks; static int kfree_nrealthreads; static atomic_t n_kfree_scale_thread_started; static atomic_t n_kfree_scale_thread_ended; +static struct task_struct *kthread_tp; +static u64 kthread_stime; struct kfree_obj { char kfree_obj[8]; @@ -701,6 +751,10 @@ kfree_scale_init(void) unsigned long jif_start; unsigned long orig_jif; + pr_alert("%s" SCALE_FLAG + "--- kfree_rcu_test: kfree_mult=%d kfree_by_call_rcu=%d kfree_nthreads=%d kfree_alloc_num=%d kfree_loops=%d kfree_rcu_test_double=%d kfree_rcu_test_single=%d\n", + scale_type, kfree_mult, kfree_by_call_rcu, kfree_nthreads, kfree_alloc_num, kfree_loops, kfree_rcu_test_double, kfree_rcu_test_single); + // Also, do a quick self-test to ensure laziness is as much as // expected. if (kfree_by_call_rcu && !IS_ENABLED(CONFIG_RCU_LAZY)) { @@ -797,6 +851,18 @@ rcu_scale_cleanup(void) if (gp_exp && gp_async) SCALEOUT_ERRSTRING("No expedited async GPs, so went with async!"); + // If built-in, just report all of the GP kthread's CPU time. + if (IS_BUILTIN(CONFIG_RCU_SCALE_TEST) && !kthread_tp && cur_ops->rso_gp_kthread) + kthread_tp = cur_ops->rso_gp_kthread(); + if (kthread_tp) { + u32 ns; + u64 us; + + kthread_stime = kthread_tp->stime - kthread_stime; + us = div_u64_rem(kthread_stime, 1000, &ns); + pr_info("rcu_scale: Grace-period kthread CPU time: %llu.%03u us\n", us, ns); + show_rcu_gp_kthreads(); + } if (kfree_rcu_test) { kfree_scale_cleanup(); return; @@ -885,7 +951,7 @@ rcu_scale_init(void) long i; int firsterr = 0; static struct rcu_scale_ops *scale_ops[] = { - &rcu_ops, &srcu_ops, &srcud_ops, TASKS_OPS TASKS_TRACING_OPS + &rcu_ops, &srcu_ops, &srcud_ops, TASKS_OPS TASKS_RUDE_OPS TASKS_TRACING_OPS }; if (!torture_init_begin(scale_type, verbose)) @@ -910,6 +976,11 @@ rcu_scale_init(void) if (cur_ops->init) cur_ops->init(); + if (cur_ops->rso_gp_kthread) { + kthread_tp = cur_ops->rso_gp_kthread(); + if (kthread_tp) + kthread_stime = kthread_tp->stime; + } if (kfree_rcu_test) return kfree_scale_init(); diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c index 147551c23baf..ade42d6a9d9b 100644 --- a/kernel/rcu/rcutorture.c +++ b/kernel/rcu/rcutorture.c @@ -1581,6 +1581,7 @@ rcu_torture_writer(void *arg) rcu_access_pointer(rcu_torture_current) != &rcu_tortures[i]) { tracing_off(); + show_rcu_gp_kthreads(); WARN(1, "%s: rtort_pipe_count: %d\n", __func__, rcu_tortures[i].rtort_pipe_count); rcu_ftrace_dump(DUMP_ALL); } @@ -1876,7 +1877,7 @@ static int rcutorture_extend_mask(int oldmask, struct torture_random_state *trsp) { int mask = rcutorture_extend_mask_max(); - unsigned long randmask1 = torture_random(trsp) >> 8; + unsigned long randmask1 = torture_random(trsp); unsigned long randmask2 = randmask1 >> 3; unsigned long preempts = RCUTORTURE_RDR_PREEMPT | RCUTORTURE_RDR_SCHED; unsigned long preempts_irq = preempts | RCUTORTURE_RDR_IRQ; @@ -1935,7 +1936,7 @@ rcutorture_loop_extend(int *readstate, struct torture_random_state *trsp, if (!((mask - 1) & mask)) return rtrsp; /* Current RCU reader not extendable. */ /* Bias towards larger numbers of loops. */ - i = (torture_random(trsp) >> 3); + i = torture_random(trsp); i = ((i | (i >> 3)) & RCUTORTURE_RDR_MAX_LOOPS) + 1; for (j = 0; j < i; j++) { mask = rcutorture_extend_mask(*readstate, trsp); @@ -2136,7 +2137,7 @@ static int rcu_nocb_toggle(void *arg) toggle_fuzz = NSEC_PER_USEC; do { r = torture_random(&rand); - cpu = (r >> 4) % (maxcpu + 1); + cpu = (r >> 1) % (maxcpu + 1); if (r & 0x1) { rcu_nocb_cpu_offload(cpu); atomic_long_inc(&n_nocb_offload); diff --git a/kernel/rcu/refscale.c b/kernel/rcu/refscale.c index 1970ce5f22d4..91a0fd0d4d9a 100644 --- a/kernel/rcu/refscale.c +++ b/kernel/rcu/refscale.c @@ -528,6 +528,38 @@ static struct ref_scale_ops clock_ops = { .name = "clock" }; +static void ref_jiffies_section(const int nloops) +{ + u64 x = 0; + int i; + + preempt_disable(); + for (i = nloops; i >= 0; i--) + x += jiffies; + preempt_enable(); + stopopts = x; +} + +static void ref_jiffies_delay_section(const int nloops, const int udl, const int ndl) +{ + u64 x = 0; + int i; + + preempt_disable(); + for (i = nloops; i >= 0; i--) { + x += jiffies; + un_delay(udl, ndl); + } + preempt_enable(); + stopopts = x; +} + +static struct ref_scale_ops jiffies_ops = { + .readsection = ref_jiffies_section, + .delaysection = ref_jiffies_delay_section, + .name = "jiffies" +}; + //////////////////////////////////////////////////////////////////////// // // Methods leveraging SLAB_TYPESAFE_BY_RCU. @@ -1047,7 +1079,7 @@ ref_scale_init(void) int firsterr = 0; static 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, + &rwsem_ops, &lock_ops, &lock_irq_ops, &acqrel_ops, &clock_ops, &jiffies_ops, &typesafe_ref_ops, &typesafe_lock_ops, &typesafe_seqlock_ops, }; @@ -1107,12 +1139,11 @@ ref_scale_init(void) VERBOSE_SCALEOUT("Starting %d reader threads", nreaders); for (i = 0; i < nreaders; i++) { + init_waitqueue_head(&reader_tasks[i].wq); firsterr = torture_create_kthread(ref_scale_reader, (void *)i, reader_tasks[i].task); if (torture_init_error(firsterr)) goto unwind; - - init_waitqueue_head(&(reader_tasks[i].wq)); } // Main Task diff --git a/kernel/rcu/tasks.h b/kernel/rcu/tasks.h index b770add3f843..8d65f7d576a3 100644 --- a/kernel/rcu/tasks.h +++ b/kernel/rcu/tasks.h @@ -25,6 +25,8 @@ typedef void (*postgp_func_t)(struct rcu_tasks *rtp); * @cblist: Callback list. * @lock: Lock protecting per-CPU callback list. * @rtp_jiffies: Jiffies counter value for statistics. + * @lazy_timer: Timer to unlazify callbacks. + * @urgent_gp: Number of additional non-lazy grace periods. * @rtp_n_lock_retries: Rough lock-contention statistic. * @rtp_work: Work queue for invoking callbacks. * @rtp_irq_work: IRQ work queue for deferred wakeups. @@ -38,6 +40,8 @@ struct rcu_tasks_percpu { raw_spinlock_t __private lock; unsigned long rtp_jiffies; unsigned long rtp_n_lock_retries; + struct timer_list lazy_timer; + unsigned int urgent_gp; struct work_struct rtp_work; struct irq_work rtp_irq_work; struct rcu_head barrier_q_head; @@ -51,7 +55,6 @@ struct rcu_tasks_percpu { * @cbs_wait: RCU wait allowing a new callback to get kthread's attention. * @cbs_gbl_lock: Lock protecting callback list. * @tasks_gp_mutex: Mutex protecting grace period, needed during mid-boot dead zone. - * @kthread_ptr: This flavor's grace-period/callback-invocation kthread. * @gp_func: This flavor's grace-period-wait function. * @gp_state: Grace period's most recent state transition (debugging). * @gp_sleep: Per-grace-period sleep to prevent CPU-bound looping. @@ -61,6 +64,8 @@ struct rcu_tasks_percpu { * @tasks_gp_seq: Number of grace periods completed since boot. * @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. + * @lazy_jiffies: Number of jiffies to allow callbacks to be lazy. * @pregp_func: This flavor's pre-grace-period function (optional). * @pertask_func: This flavor's per-task scan function (optional). * @postscan_func: This flavor's post-task scan function (optional). @@ -92,6 +97,7 @@ struct rcu_tasks { unsigned long n_ipis; unsigned long n_ipis_fails; struct task_struct *kthread_ptr; + unsigned long lazy_jiffies; rcu_tasks_gp_func_t gp_func; pregp_func_t pregp_func; pertask_func_t pertask_func; @@ -127,6 +133,7 @@ static struct rcu_tasks rt_name = \ .gp_func = gp, \ .call_func = call, \ .rtpcpu = &rt_name ## __percpu, \ + .lazy_jiffies = DIV_ROUND_UP(HZ, 4), \ .name = n, \ .percpu_enqueue_shift = order_base_2(CONFIG_NR_CPUS), \ .percpu_enqueue_lim = 1, \ @@ -139,9 +146,7 @@ static struct rcu_tasks rt_name = \ #ifdef CONFIG_TASKS_RCU /* Track exiting tasks in order to allow them to be waited for. */ DEFINE_STATIC_SRCU(tasks_rcu_exit_srcu); -#endif -#ifdef CONFIG_TASKS_RCU /* Report delay in synchronize_srcu() completion in rcu_tasks_postscan(). */ static void tasks_rcu_exit_srcu_stall(struct timer_list *unused); static DEFINE_TIMER(tasks_rcu_exit_srcu_stall_timer, tasks_rcu_exit_srcu_stall); @@ -171,6 +176,8 @@ static int rcu_task_contend_lim __read_mostly = 100; module_param(rcu_task_contend_lim, int, 0444); static int rcu_task_collapse_lim __read_mostly = 10; 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); /* RCU tasks grace-period state for debugging. */ #define RTGS_INIT 0 @@ -229,7 +236,7 @@ static const char *tasks_gp_state_getname(struct rcu_tasks *rtp) #endif /* #ifndef CONFIG_TINY_RCU */ // Initialize per-CPU callback lists for the specified flavor of -// Tasks RCU. +// Tasks RCU. Do not enqueue callbacks before this function is invoked. static void cblist_init_generic(struct rcu_tasks *rtp) { int cpu; @@ -237,7 +244,6 @@ static void cblist_init_generic(struct rcu_tasks *rtp) int lim; int shift; - raw_spin_lock_irqsave(&rtp->cbs_gbl_lock, flags); if (rcu_task_enqueue_lim < 0) { rcu_task_enqueue_lim = 1; rcu_task_cb_adjust = true; @@ -260,22 +266,48 @@ static void cblist_init_generic(struct rcu_tasks *rtp) WARN_ON_ONCE(!rtpcp); if (cpu) raw_spin_lock_init(&ACCESS_PRIVATE(rtpcp, lock)); - raw_spin_lock_rcu_node(rtpcp); // irqs already disabled. + local_irq_save(flags); // serialize initialization if (rcu_segcblist_empty(&rtpcp->cblist)) rcu_segcblist_init(&rtpcp->cblist); + local_irq_restore(flags); INIT_WORK(&rtpcp->rtp_work, rcu_tasks_invoke_cbs_wq); rtpcp->cpu = cpu; rtpcp->rtpp = rtp; if (!rtpcp->rtp_blkd_tasks.next) INIT_LIST_HEAD(&rtpcp->rtp_blkd_tasks); - raw_spin_unlock_rcu_node(rtpcp); // irqs remain disabled. } - raw_spin_unlock_irqrestore(&rtp->cbs_gbl_lock, flags); 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); } +// Compute wakeup time for lazy callback timer. +static unsigned long rcu_tasks_lazy_time(struct rcu_tasks *rtp) +{ + return jiffies + rtp->lazy_jiffies; +} + +// Timer handler that unlazifies lazy callbacks. +static void call_rcu_tasks_generic_timer(struct timer_list *tlp) +{ + unsigned long flags; + bool needwake = false; + struct rcu_tasks *rtp; + struct rcu_tasks_percpu *rtpcp = from_timer(rtpcp, tlp, lazy_timer); + + rtp = rtpcp->rtpp; + raw_spin_lock_irqsave_rcu_node(rtpcp, flags); + if (!rcu_segcblist_empty(&rtpcp->cblist) && rtp->lazy_jiffies) { + if (!rtpcp->urgent_gp) + rtpcp->urgent_gp = 1; + needwake = true; + mod_timer(&rtpcp->lazy_timer, rcu_tasks_lazy_time(rtp)); + } + raw_spin_unlock_irqrestore_rcu_node(rtpcp, flags); + if (needwake) + rcuwait_wake_up(&rtp->cbs_wait); +} + // IRQ-work handler that does deferred wakeup for call_rcu_tasks_generic(). static void call_rcu_tasks_iw_wakeup(struct irq_work *iwp) { @@ -292,6 +324,7 @@ static void call_rcu_tasks_generic(struct rcu_head *rhp, rcu_callback_t func, { int chosen_cpu; unsigned long flags; + bool havekthread = smp_load_acquire(&rtp->kthread_ptr); int ideal_cpu; unsigned long j; bool needadjust = false; @@ -316,12 +349,19 @@ static void call_rcu_tasks_generic(struct rcu_head *rhp, rcu_callback_t func, READ_ONCE(rtp->percpu_enqueue_lim) != nr_cpu_ids) needadjust = true; // Defer adjustment to avoid deadlock. } - if (!rcu_segcblist_is_enabled(&rtpcp->cblist)) { - raw_spin_unlock_rcu_node(rtpcp); // irqs remain disabled. - cblist_init_generic(rtp); - raw_spin_lock_rcu_node(rtpcp); // irqs already disabled. + // Queuing callbacks before initialization not yet supported. + if (WARN_ON_ONCE(!rcu_segcblist_is_enabled(&rtpcp->cblist))) + rcu_segcblist_init(&rtpcp->cblist); + needwake = (func == wakeme_after_rcu) || + (rcu_segcblist_n_cbs(&rtpcp->cblist) == rcu_task_lazy_lim); + if (havekthread && !needwake && !timer_pending(&rtpcp->lazy_timer)) { + if (rtp->lazy_jiffies) + mod_timer(&rtpcp->lazy_timer, rcu_tasks_lazy_time(rtp)); + else + needwake = rcu_segcblist_empty(&rtpcp->cblist); } - needwake = rcu_segcblist_empty(&rtpcp->cblist); + if (needwake) + rtpcp->urgent_gp = 3; rcu_segcblist_enqueue(&rtpcp->cblist, rhp); raw_spin_unlock_irqrestore_rcu_node(rtpcp, flags); if (unlikely(needadjust)) { @@ -415,9 +455,14 @@ static int rcu_tasks_need_gpcb(struct rcu_tasks *rtp) } rcu_segcblist_advance(&rtpcp->cblist, rcu_seq_current(&rtp->tasks_gp_seq)); (void)rcu_segcblist_accelerate(&rtpcp->cblist, rcu_seq_snap(&rtp->tasks_gp_seq)); - if (rcu_segcblist_pend_cbs(&rtpcp->cblist)) + if (rtpcp->urgent_gp > 0 && rcu_segcblist_pend_cbs(&rtpcp->cblist)) { + if (rtp->lazy_jiffies) + rtpcp->urgent_gp--; needgpcb |= 0x3; - if (!rcu_segcblist_empty(&rtpcp->cblist)) + } else if (rcu_segcblist_empty(&rtpcp->cblist)) { + rtpcp->urgent_gp = 0; + } + if (rcu_segcblist_ready_cbs(&rtpcp->cblist)) needgpcb |= 0x1; raw_spin_unlock_irqrestore_rcu_node(rtpcp, flags); } @@ -525,10 +570,12 @@ static void rcu_tasks_one_gp(struct rcu_tasks *rtp, bool midboot) if (unlikely(midboot)) { needgpcb = 0x2; } else { + mutex_unlock(&rtp->tasks_gp_mutex); set_tasks_gp_state(rtp, RTGS_WAIT_CBS); rcuwait_wait_event(&rtp->cbs_wait, (needgpcb = rcu_tasks_need_gpcb(rtp)), TASK_IDLE); + mutex_lock(&rtp->tasks_gp_mutex); } if (needgpcb & 0x2) { @@ -549,11 +596,19 @@ static void rcu_tasks_one_gp(struct rcu_tasks *rtp, bool midboot) // RCU-tasks kthread that detects grace periods and invokes callbacks. static int __noreturn rcu_tasks_kthread(void *arg) { + int cpu; struct rcu_tasks *rtp = arg; + for_each_possible_cpu(cpu) { + struct rcu_tasks_percpu *rtpcp = per_cpu_ptr(rtp->rtpcpu, cpu); + + timer_setup(&rtpcp->lazy_timer, call_rcu_tasks_generic_timer, 0); + rtpcp->urgent_gp = 1; + } + /* Run on housekeeping CPUs by default. Sysadm can move if desired. */ housekeeping_affine(current, HK_TYPE_RCU); - WRITE_ONCE(rtp->kthread_ptr, current); // Let GPs start! + smp_store_release(&rtp->kthread_ptr, current); // Let GPs start! /* * Each pass through the following loop makes one check for @@ -635,16 +690,22 @@ static void show_rcu_tasks_generic_gp_kthread(struct rcu_tasks *rtp, char *s) { int cpu; bool havecbs = false; + bool haveurgent = false; + bool haveurgentcbs = false; for_each_possible_cpu(cpu) { struct rcu_tasks_percpu *rtpcp = per_cpu_ptr(rtp->rtpcpu, cpu); - if (!data_race(rcu_segcblist_empty(&rtpcp->cblist))) { + if (!data_race(rcu_segcblist_empty(&rtpcp->cblist))) havecbs = true; + if (data_race(rtpcp->urgent_gp)) + haveurgent = true; + if (!data_race(rcu_segcblist_empty(&rtpcp->cblist)) && data_race(rtpcp->urgent_gp)) + haveurgentcbs = true; + if (havecbs && haveurgent && haveurgentcbs) break; - } } - pr_info("%s: %s(%d) since %lu g:%lu i:%lu/%lu %c%c %s\n", + pr_info("%s: %s(%d) since %lu g:%lu i:%lu/%lu %c%c%c%c l:%lu %s\n", rtp->kname, tasks_gp_state_getname(rtp), data_race(rtp->gp_state), jiffies - data_race(rtp->gp_jiffies), @@ -652,6 +713,9 @@ static void show_rcu_tasks_generic_gp_kthread(struct rcu_tasks *rtp, char *s) data_race(rtp->n_ipis_fails), data_race(rtp->n_ipis), ".k"[!!data_race(rtp->kthread_ptr)], ".C"[havecbs], + ".u"[haveurgent], + ".U"[haveurgentcbs], + rtp->lazy_jiffies, s); } #endif // #ifndef CONFIG_TINY_RCU @@ -1020,11 +1084,16 @@ void rcu_barrier_tasks(void) } EXPORT_SYMBOL_GPL(rcu_barrier_tasks); +int rcu_tasks_lazy_ms = -1; +module_param(rcu_tasks_lazy_ms, int, 0444); + static int __init rcu_spawn_tasks_kthread(void) { cblist_init_generic(&rcu_tasks); rcu_tasks.gp_sleep = HZ / 10; rcu_tasks.init_fract = HZ / 10; + if (rcu_tasks_lazy_ms >= 0) + rcu_tasks.lazy_jiffies = msecs_to_jiffies(rcu_tasks_lazy_ms); rcu_tasks.pregp_func = rcu_tasks_pregp_step; rcu_tasks.pertask_func = rcu_tasks_pertask; rcu_tasks.postscan_func = rcu_tasks_postscan; @@ -1042,6 +1111,12 @@ void show_rcu_tasks_classic_gp_kthread(void) EXPORT_SYMBOL_GPL(show_rcu_tasks_classic_gp_kthread); #endif // !defined(CONFIG_TINY_RCU) +struct task_struct *get_rcu_tasks_gp_kthread(void) +{ + return rcu_tasks.kthread_ptr; +} +EXPORT_SYMBOL_GPL(get_rcu_tasks_gp_kthread); + /* * Contribute to protect against tasklist scan blind spot while the * task is exiting and may be removed from the tasklist. See @@ -1173,10 +1248,15 @@ void rcu_barrier_tasks_rude(void) } 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) { cblist_init_generic(&rcu_tasks_rude); 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; } @@ -1188,6 +1268,13 @@ void show_rcu_tasks_rude_gp_kthread(void) } EXPORT_SYMBOL_GPL(show_rcu_tasks_rude_gp_kthread); #endif // !defined(CONFIG_TINY_RCU) + +struct task_struct *get_rcu_tasks_rude_gp_kthread(void) +{ + return rcu_tasks_rude.kthread_ptr; +} +EXPORT_SYMBOL_GPL(get_rcu_tasks_rude_gp_kthread); + #endif /* #ifdef CONFIG_TASKS_RUDE_RCU */ //////////////////////////////////////////////////////////////////////// @@ -1793,6 +1880,9 @@ void rcu_barrier_tasks_trace(void) } EXPORT_SYMBOL_GPL(rcu_barrier_tasks_trace); +int rcu_tasks_trace_lazy_ms = -1; +module_param(rcu_tasks_trace_lazy_ms, int, 0444); + static int __init rcu_spawn_tasks_trace_kthread(void) { cblist_init_generic(&rcu_tasks_trace); @@ -1807,6 +1897,8 @@ static int __init rcu_spawn_tasks_trace_kthread(void) if (rcu_tasks_trace.init_fract <= 0) rcu_tasks_trace.init_fract = 1; } + if (rcu_tasks_trace_lazy_ms >= 0) + rcu_tasks_trace.lazy_jiffies = msecs_to_jiffies(rcu_tasks_trace_lazy_ms); rcu_tasks_trace.pregp_func = rcu_tasks_trace_pregp_step; rcu_tasks_trace.postscan_func = rcu_tasks_trace_postscan; rcu_tasks_trace.holdouts_func = check_all_holdout_tasks_trace; @@ -1830,6 +1922,12 @@ void show_rcu_tasks_trace_gp_kthread(void) EXPORT_SYMBOL_GPL(show_rcu_tasks_trace_gp_kthread); #endif // !defined(CONFIG_TINY_RCU) +struct task_struct *get_rcu_tasks_trace_gp_kthread(void) +{ + return rcu_tasks_trace.kthread_ptr; +} +EXPORT_SYMBOL_GPL(get_rcu_tasks_trace_gp_kthread); + #else /* #ifdef CONFIG_TASKS_TRACE_RCU */ static void exit_tasks_rcu_finish_trace(struct task_struct *t) { } #endif /* #else #ifdef CONFIG_TASKS_TRACE_RCU */ diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 1449cb69a0e0..cb1caefa8bd0 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -632,7 +632,7 @@ void __rcu_irq_enter_check_tick(void) // prevents self-deadlock. So we can safely recheck under the lock. // Note that the nohz_full state currently cannot change. raw_spin_lock_rcu_node(rdp->mynode); - if (rdp->rcu_urgent_qs && !rdp->rcu_forced_tick) { + if (READ_ONCE(rdp->rcu_urgent_qs) && !rdp->rcu_forced_tick) { // A nohz_full CPU is in the kernel and RCU needs a // quiescent state. Turn on the tick! WRITE_ONCE(rdp->rcu_forced_tick, true); @@ -677,12 +677,16 @@ static void rcu_disable_urgency_upon_qs(struct rcu_data *rdp) } /** - * rcu_is_watching - see if RCU thinks that the current CPU is not idle + * rcu_is_watching - RCU read-side critical sections permitted on current CPU? * - * Return true if RCU is watching the running CPU, which means that this - * CPU can safely enter RCU read-side critical sections. In other words, - * if the current CPU is not in its idle loop or is in an interrupt or - * NMI handler, return true. + * Return @true if RCU is watching the running CPU and @false otherwise. + * An @true return means that this CPU can safely enter RCU read-side + * critical sections. + * + * Although calls to rcu_is_watching() from most parts of the kernel + * will return @true, there are important exceptions. For example, if the + * current CPU is deep within its idle loop, in kernel entry/exit code, + * or offline, rcu_is_watching() will return @false. * * Make notrace because it can be called by the internal functions of * ftrace, and making this notrace removes unnecessary recursion calls. diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h index 43229d2b0c44..5598212d1f27 100644 --- a/kernel/rcu/tree_nocb.h +++ b/kernel/rcu/tree_nocb.h @@ -77,9 +77,9 @@ __setup("rcu_nocbs", rcu_nocb_setup); static int __init parse_rcu_nocb_poll(char *arg) { rcu_nocb_poll = true; - return 0; + return 1; } -early_param("rcu_nocb_poll", parse_rcu_nocb_poll); +__setup("rcu_nocb_poll", parse_rcu_nocb_poll); /* * Don't bother bypassing ->cblist if the call_rcu() rate is low. diff --git a/kernel/torture.c b/kernel/torture.c index 1a0519b836ac..b28b05bbef02 100644 --- a/kernel/torture.c +++ b/kernel/torture.c @@ -37,6 +37,7 @@ #include <linux/ktime.h> #include <asm/byteorder.h> #include <linux/torture.h> +#include <linux/sched/rt.h> #include "rcu/rcu.h" MODULE_LICENSE("GPL"); @@ -54,6 +55,9 @@ module_param(verbose_sleep_frequency, int, 0444); static int verbose_sleep_duration = 1; module_param(verbose_sleep_duration, int, 0444); +static int random_shuffle; +module_param(random_shuffle, int, 0444); + static char *torture_type; static int verbose; @@ -88,8 +92,8 @@ int torture_hrtimeout_ns(ktime_t baset_ns, u32 fuzzt_ns, struct torture_random_s ktime_t hto = baset_ns; if (trsp) - hto += (torture_random(trsp) >> 3) % fuzzt_ns; - set_current_state(TASK_UNINTERRUPTIBLE); + hto += torture_random(trsp) % fuzzt_ns; + set_current_state(TASK_IDLE); return schedule_hrtimeout(&hto, HRTIMER_MODE_REL); } EXPORT_SYMBOL_GPL(torture_hrtimeout_ns); @@ -350,22 +354,22 @@ torture_onoff(void *arg) if (onoff_holdoff > 0) { VERBOSE_TOROUT_STRING("torture_onoff begin holdoff"); - schedule_timeout_interruptible(onoff_holdoff); + torture_hrtimeout_jiffies(onoff_holdoff, &rand); VERBOSE_TOROUT_STRING("torture_onoff end holdoff"); } while (!torture_must_stop()) { if (disable_onoff_at_boot && !rcu_inkernel_boot_has_ended()) { - schedule_timeout_interruptible(HZ / 10); + torture_hrtimeout_jiffies(HZ / 10, &rand); continue; } - cpu = (torture_random(&rand) >> 4) % (maxcpu + 1); + cpu = torture_random(&rand) % (maxcpu + 1); if (!torture_offline(cpu, &n_offline_attempts, &n_offline_successes, &sum_offline, &min_offline, &max_offline)) torture_online(cpu, &n_online_attempts, &n_online_successes, &sum_online, &min_online, &max_online); - schedule_timeout_interruptible(onoff_interval); + torture_hrtimeout_jiffies(onoff_interval, &rand); } stop: @@ -518,6 +522,7 @@ static void torture_shuffle_task_unregister_all(void) */ static void torture_shuffle_tasks(void) { + DEFINE_TORTURE_RANDOM(rand); struct shuffle_task *stp; cpumask_setall(shuffle_tmp_mask); @@ -537,8 +542,10 @@ static void torture_shuffle_tasks(void) cpumask_clear_cpu(shuffle_idle_cpu, shuffle_tmp_mask); mutex_lock(&shuffle_task_mutex); - list_for_each_entry(stp, &shuffle_task_list, st_l) - set_cpus_allowed_ptr(stp->st_t, shuffle_tmp_mask); + list_for_each_entry(stp, &shuffle_task_list, st_l) { + if (!random_shuffle || torture_random(&rand) & 0x1) + set_cpus_allowed_ptr(stp->st_t, shuffle_tmp_mask); + } mutex_unlock(&shuffle_task_mutex); cpus_read_unlock(); @@ -550,9 +557,11 @@ static void torture_shuffle_tasks(void) */ static int torture_shuffle(void *arg) { + DEFINE_TORTURE_RANDOM(rand); + VERBOSE_TOROUT_STRING("torture_shuffle task started"); do { - schedule_timeout_interruptible(shuffle_interval); + torture_hrtimeout_jiffies(shuffle_interval, &rand); torture_shuffle_tasks(); torture_shutdown_absorb("torture_shuffle"); } while (!torture_must_stop()); @@ -728,12 +737,12 @@ bool stutter_wait(const char *title) cond_resched_tasks_rcu_qs(); spt = READ_ONCE(stutter_pause_test); for (; spt; spt = READ_ONCE(stutter_pause_test)) { - if (!ret) { + if (!ret && !rt_task(current)) { sched_set_normal(current, MAX_NICE); ret = true; } if (spt == 1) { - schedule_timeout_interruptible(1); + torture_hrtimeout_jiffies(1, NULL); } else if (spt == 2) { while (READ_ONCE(stutter_pause_test)) { if (!(i++ & 0xffff)) @@ -741,7 +750,7 @@ bool stutter_wait(const char *title) cond_resched(); } } else { - schedule_timeout_interruptible(round_jiffies_relative(HZ)); + torture_hrtimeout_jiffies(round_jiffies_relative(HZ), NULL); } torture_shutdown_absorb(title); } @@ -926,7 +935,7 @@ EXPORT_SYMBOL_GPL(torture_kthread_stopping); * it starts, you will need to open-code your own. */ int _torture_create_kthread(int (*fn)(void *arg), void *arg, char *s, char *m, - char *f, struct task_struct **tp) + char *f, struct task_struct **tp, void (*cbf)(struct task_struct *tp)) { int ret = 0; @@ -938,6 +947,10 @@ int _torture_create_kthread(int (*fn)(void *arg), void *arg, char *s, char *m, *tp = NULL; return ret; } + + if (cbf) + cbf(*tp); + wake_up_process(*tp); // Process is sleeping, so ordering provided. torture_shuffle_task_register(*tp); return ret; diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl index 880fde13d9b8..a9841148cde2 100755 --- a/scripts/checkpatch.pl +++ b/scripts/checkpatch.pl @@ -7457,6 +7457,30 @@ sub process { } } +# Complain about RCU Tasks Trace used outside of BPF (and of course, RCU). + our $rcu_trace_funcs = qr{(?x: + rcu_read_lock_trace | + rcu_read_lock_trace_held | + rcu_read_unlock_trace | + call_rcu_tasks_trace | + synchronize_rcu_tasks_trace | + rcu_barrier_tasks_trace | + rcu_request_urgent_qs_task + )}; + our $rcu_trace_paths = qr{(?x: + kernel/bpf/ | + include/linux/bpf | + net/bpf/ | + kernel/rcu/ | + include/linux/rcu + )}; + if ($line =~ /\b($rcu_trace_funcs)\s*\(/) { + if ($realfile !~ m{^$rcu_trace_paths}) { + WARN("RCU_TASKS_TRACE", + "use of RCU tasks trace is incorrect outside BPF or core RCU code\n" . $herecurr); + } + } + # check for lockdep_set_novalidate_class if ($line =~ /^.\s*lockdep_set_novalidate_class\s*\(/ || $line =~ /__lockdep_no_validate__\s*\)/ ) { diff --git a/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuscale.sh b/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuscale.sh index b582113178ac..f683e424ddd5 100755 --- a/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuscale.sh +++ b/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuscale.sh @@ -40,6 +40,10 @@ awk ' sum += $5 / 1000.; } +/rcu_scale: Grace-period kthread CPU time/ { + cputime = $6; +} + END { newNR = asort(gptimes); if (newNR <= 0) { @@ -78,6 +82,8 @@ END { print "90th percentile grace-period duration: " gptimes[pct90]; print "99th percentile grace-period duration: " gptimes[pct99]; print "Maximum grace-period duration: " gptimes[newNR]; - print "Grace periods: " ngps + 0 " Batches: " nbatches + 0 " Ratio: " ngps / nbatches; + if (cputime != "") + cpustr = " CPU: " cputime; + print "Grace periods: " ngps + 0 " Batches: " nbatches + 0 " Ratio: " ngps / nbatches cpustr; print "Computed from rcuscale printk output."; }' diff --git a/tools/testing/selftests/rcutorture/configs/rcuscale/CFcommon b/tools/testing/selftests/rcutorture/configs/rcuscale/CFcommon index 6a00157bee5b..b1ffd7c67604 100644 --- a/tools/testing/selftests/rcutorture/configs/rcuscale/CFcommon +++ b/tools/testing/selftests/rcutorture/configs/rcuscale/CFcommon @@ -2,5 +2,7 @@ CONFIG_RCU_SCALE_TEST=y CONFIG_PRINTK_TIME=y CONFIG_FORCE_TASKS_RCU=y #CHECK#CONFIG_TASKS_RCU=y +CONFIG_FORCE_TASKS_RUDE_RCU=y +#CHECK#CONFIG_TASKS_RUDE_RCU=y CONFIG_FORCE_TASKS_TRACE_RCU=y #CHECK#CONFIG_TASKS_TRACE_RCU=y diff --git a/tools/testing/selftests/rcutorture/configs/rcuscale/TRACE01 b/tools/testing/selftests/rcutorture/configs/rcuscale/TRACE01 index 227aba7783af..0059592c7408 100644 --- a/tools/testing/selftests/rcutorture/configs/rcuscale/TRACE01 +++ b/tools/testing/selftests/rcutorture/configs/rcuscale/TRACE01 @@ -2,6 +2,8 @@ CONFIG_SMP=y CONFIG_PREEMPT_NONE=y CONFIG_PREEMPT_VOLUNTARY=n CONFIG_PREEMPT=n +CONFIG_PREEMPT_DYNAMIC=n +#CHECK#CONFIG_TREE_RCU=y CONFIG_HZ_PERIODIC=n CONFIG_NO_HZ_IDLE=y CONFIG_NO_HZ_FULL=n diff --git a/tools/testing/selftests/rcutorture/configs/refscale/NOPREEMPT b/tools/testing/selftests/rcutorture/configs/refscale/NOPREEMPT index ef2b501a6971..67f9d2998afd 100644 --- a/tools/testing/selftests/rcutorture/configs/refscale/NOPREEMPT +++ b/tools/testing/selftests/rcutorture/configs/refscale/NOPREEMPT @@ -2,6 +2,7 @@ CONFIG_SMP=y 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 |