From 4aa35e0db6d758e8f952ed30d7ac3117c376562c Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Tue, 25 Jul 2023 12:19:01 +0100 Subject: context_tracking, rcu: Rename RCU_DYNTICKS_IDX into CT_RCU_WATCHING The symbols relating to the CT_STATE part of context_tracking.state are now all prefixed with CT_STATE. The RCU dynticks counter part of that atomic variable still involves symbols with different prefixes, align them all to be prefixed with CT_RCU_WATCHING. Suggested-by: "Paul E. McKenney" Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Acked-by: Thomas Gleixner Signed-off-by: Neeraj Upadhyay --- kernel/rcu/tree.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index e641cc681901..04f87c44385c 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -294,9 +294,9 @@ void rcu_softirq_qs(void) */ static void rcu_dynticks_eqs_online(void) { - if (ct_dynticks() & RCU_DYNTICKS_IDX) + if (ct_dynticks() & CT_RCU_WATCHING) return; - ct_state_inc(RCU_DYNTICKS_IDX); + ct_state_inc(CT_RCU_WATCHING); } /* @@ -305,7 +305,7 @@ static void rcu_dynticks_eqs_online(void) */ static bool rcu_dynticks_in_eqs(int snap) { - return !(snap & RCU_DYNTICKS_IDX); + return !(snap & CT_RCU_WATCHING); } /* @@ -335,7 +335,7 @@ bool rcu_dynticks_zero_in_eqs(int cpu, int *vp) int snap; // If not quiescent, force back to earlier extended quiescent state. - snap = ct_dynticks_cpu(cpu) & ~RCU_DYNTICKS_IDX; + snap = ct_dynticks_cpu(cpu) & ~CT_RCU_WATCHING; smp_rmb(); // Order ->dynticks and *vp reads. if (READ_ONCE(*vp)) return false; // Non-zero, so report failure; @@ -361,9 +361,9 @@ notrace void rcu_momentary_dyntick_idle(void) int seq; raw_cpu_write(rcu_data.rcu_need_heavy_qs, false); - seq = ct_state_inc(2 * RCU_DYNTICKS_IDX); + seq = ct_state_inc(2 * CT_RCU_WATCHING); /* It is illegal to call this from idle state. */ - WARN_ON_ONCE(!(seq & RCU_DYNTICKS_IDX)); + WARN_ON_ONCE(!(seq & CT_RCU_WATCHING)); rcu_preempt_deferred_qs(current); } EXPORT_SYMBOL_GPL(rcu_momentary_dyntick_idle); -- cgit v1.2.3 From a4a7921ec08d016f9d692752e2f82f66369c7ffa Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Tue, 16 Apr 2024 10:45:56 +0200 Subject: context_tracking, rcu: Rename ct_dynticks() into ct_rcu_watching() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, reflect that change in the related helpers. Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- include/linux/context_tracking_state.h | 2 +- kernel/context_tracking.c | 10 +++++----- kernel/rcu/tree.c | 2 +- 3 files changed, 7 insertions(+), 7 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/include/linux/context_tracking_state.h b/include/linux/context_tracking_state.h index 94d6a935af3b..cb90d8c17810 100644 --- a/include/linux/context_tracking_state.h +++ b/include/linux/context_tracking_state.h @@ -56,7 +56,7 @@ static __always_inline int __ct_state(void) #endif #ifdef CONFIG_CONTEXT_TRACKING_IDLE -static __always_inline int ct_dynticks(void) +static __always_inline int ct_rcu_watching(void) { return atomic_read(this_cpu_ptr(&context_tracking.state)) & CT_RCU_WATCHING_MASK; } diff --git a/kernel/context_tracking.c b/kernel/context_tracking.c index b2589bc59e18..868ae0bcd4be 100644 --- a/kernel/context_tracking.c +++ b/kernel/context_tracking.c @@ -137,7 +137,7 @@ static void noinstr ct_kernel_exit(bool user, int offset) instrumentation_begin(); lockdep_assert_irqs_disabled(); - trace_rcu_dyntick(TPS("Start"), ct_dynticks_nesting(), 0, ct_dynticks()); + trace_rcu_dyntick(TPS("Start"), ct_dynticks_nesting(), 0, ct_rcu_watching()); WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && !user && !is_idle_task(current)); rcu_preempt_deferred_qs(current); @@ -182,7 +182,7 @@ static void noinstr ct_kernel_enter(bool user, int offset) // instrumentation for the noinstr ct_kernel_enter_state() instrument_atomic_write(&ct->state, sizeof(ct->state)); - trace_rcu_dyntick(TPS("End"), ct_dynticks_nesting(), 1, ct_dynticks()); + trace_rcu_dyntick(TPS("End"), ct_dynticks_nesting(), 1, ct_rcu_watching()); WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && !user && !is_idle_task(current)); WRITE_ONCE(ct->dynticks_nesting, 1); WARN_ON_ONCE(ct_dynticks_nmi_nesting()); @@ -220,7 +220,7 @@ void noinstr ct_nmi_exit(void) */ if (ct_dynticks_nmi_nesting() != 1) { trace_rcu_dyntick(TPS("--="), ct_dynticks_nmi_nesting(), ct_dynticks_nmi_nesting() - 2, - ct_dynticks()); + ct_rcu_watching()); WRITE_ONCE(ct->dynticks_nmi_nesting, /* No store tearing. */ ct_dynticks_nmi_nesting() - 2); instrumentation_end(); @@ -228,7 +228,7 @@ void noinstr ct_nmi_exit(void) } /* This NMI interrupted an RCU-idle CPU, restore RCU-idleness. */ - trace_rcu_dyntick(TPS("Startirq"), ct_dynticks_nmi_nesting(), 0, ct_dynticks()); + trace_rcu_dyntick(TPS("Startirq"), ct_dynticks_nmi_nesting(), 0, ct_rcu_watching()); WRITE_ONCE(ct->dynticks_nmi_nesting, 0); /* Avoid store tearing. */ // instrumentation for the noinstr ct_kernel_exit_state() @@ -296,7 +296,7 @@ void noinstr ct_nmi_enter(void) trace_rcu_dyntick(incby == 1 ? TPS("Endirq") : TPS("++="), ct_dynticks_nmi_nesting(), - ct_dynticks_nmi_nesting() + incby, ct_dynticks()); + ct_dynticks_nmi_nesting() + incby, ct_rcu_watching()); instrumentation_end(); WRITE_ONCE(ct->dynticks_nmi_nesting, /* Prevent store tearing. */ ct_dynticks_nmi_nesting() + incby); diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 04f87c44385c..3a2f0325aa43 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -294,7 +294,7 @@ void rcu_softirq_qs(void) */ static void rcu_dynticks_eqs_online(void) { - if (ct_dynticks() & CT_RCU_WATCHING) + if (ct_rcu_watching() & CT_RCU_WATCHING) return; ct_state_inc(CT_RCU_WATCHING); } -- cgit v1.2.3 From a9fde9d1a5dd42edadf61cf4e64aa234b4c5bd3f Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Tue, 16 Apr 2024 10:47:14 +0200 Subject: context_tracking, rcu: Rename ct_dynticks_cpu() into ct_rcu_watching_cpu() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, reflect that change in the related helpers. Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- include/linux/context_tracking_state.h | 2 +- kernel/rcu/tree.c | 10 +++++----- kernel/rcu/tree_stall.h | 4 ++-- 3 files changed, 8 insertions(+), 8 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/include/linux/context_tracking_state.h b/include/linux/context_tracking_state.h index cb90d8c17810..ad5a06a42b4a 100644 --- a/include/linux/context_tracking_state.h +++ b/include/linux/context_tracking_state.h @@ -61,7 +61,7 @@ static __always_inline int ct_rcu_watching(void) return atomic_read(this_cpu_ptr(&context_tracking.state)) & CT_RCU_WATCHING_MASK; } -static __always_inline int ct_dynticks_cpu(int cpu) +static __always_inline int ct_rcu_watching_cpu(int cpu) { struct context_tracking *ct = per_cpu_ptr(&context_tracking, cpu); diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 3a2f0325aa43..e7f612e9f7e5 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -335,14 +335,14 @@ bool rcu_dynticks_zero_in_eqs(int cpu, int *vp) int snap; // If not quiescent, force back to earlier extended quiescent state. - snap = ct_dynticks_cpu(cpu) & ~CT_RCU_WATCHING; - smp_rmb(); // Order ->dynticks and *vp reads. + snap = ct_rcu_watching_cpu(cpu) & ~CT_RCU_WATCHING; + smp_rmb(); // Order CT state and *vp reads. if (READ_ONCE(*vp)) return false; // Non-zero, so report failure; - smp_rmb(); // Order *vp read and ->dynticks re-read. + smp_rmb(); // Order *vp read and CT state re-read. // If still in the same extended quiescent state, we are good! - return snap == ct_dynticks_cpu(cpu); + return snap == ct_rcu_watching_cpu(cpu); } /* @@ -4805,7 +4805,7 @@ rcu_boot_init_percpu_data(int cpu) rdp->grpmask = leaf_node_cpu_bit(rdp->mynode, cpu); INIT_WORK(&rdp->strict_work, strict_work_handler); WARN_ON_ONCE(ct->dynticks_nesting != 1); - WARN_ON_ONCE(rcu_dynticks_in_eqs(ct_dynticks_cpu(cpu))); + WARN_ON_ONCE(rcu_dynticks_in_eqs(ct_rcu_watching_cpu(cpu))); rdp->barrier_seq_snap = rcu_state.barrier_sequence; rdp->rcu_ofl_gp_seq = rcu_state.gp_seq; rdp->rcu_ofl_gp_state = RCU_GP_CLEANED; diff --git a/kernel/rcu/tree_stall.h b/kernel/rcu/tree_stall.h index 4b0e9d7c4c68..d65974448e81 100644 --- a/kernel/rcu/tree_stall.h +++ b/kernel/rcu/tree_stall.h @@ -501,7 +501,7 @@ static void print_cpu_stall_info(int cpu) } delta = rcu_seq_ctr(rdp->mynode->gp_seq - rdp->rcu_iw_gp_seq); falsepositive = rcu_is_gp_kthread_starving(NULL) && - rcu_dynticks_in_eqs(ct_dynticks_cpu(cpu)); + rcu_dynticks_in_eqs(ct_rcu_watching_cpu(cpu)); rcuc_starved = rcu_is_rcuc_kthread_starving(rdp, &j); if (rcuc_starved) // Print signed value, as negative values indicate a probable bug. @@ -515,7 +515,7 @@ static void print_cpu_stall_info(int cpu) rdp->rcu_iw_pending ? (int)min(delta, 9UL) + '0' : "!."[!delta], ticks_value, ticks_title, - ct_dynticks_cpu(cpu) & 0xffff, + ct_rcu_watching_cpu(cpu) & 0xffff, ct_dynticks_nesting_cpu(cpu), ct_dynticks_nmi_nesting_cpu(cpu), rdp->softirq_snap, kstat_softirqs_cpu(RCU_SOFTIRQ, cpu), data_race(rcu_state.n_force_qs) - rcu_state.n_force_qs_gpstart, -- cgit v1.2.3 From 125716c393889f9035567d4d78da239483f63ece Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Tue, 16 Apr 2024 14:47:12 +0200 Subject: context_tracking, rcu: Rename ct_dynticks_cpu_acquire() into ct_rcu_watching_cpu_acquire() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, reflect that change in the related helpers. Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- Documentation/RCU/Design/Memory-Ordering/Tree-RCU-Memory-Ordering.rst | 2 +- include/linux/context_tracking_state.h | 2 +- kernel/rcu/tree.c | 4 ++-- kernel/rcu/tree_exp.h | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/Documentation/RCU/Design/Memory-Ordering/Tree-RCU-Memory-Ordering.rst b/Documentation/RCU/Design/Memory-Ordering/Tree-RCU-Memory-Ordering.rst index 728b1e690c64..2d7036ad7476 100644 --- a/Documentation/RCU/Design/Memory-Ordering/Tree-RCU-Memory-Ordering.rst +++ b/Documentation/RCU/Design/Memory-Ordering/Tree-RCU-Memory-Ordering.rst @@ -149,7 +149,7 @@ This case is handled by calls to the strongly ordered ``atomic_add_return()`` read-modify-write atomic operation that is invoked within ``rcu_dynticks_eqs_enter()`` at idle-entry time and within ``rcu_dynticks_eqs_exit()`` at idle-exit time. -The grace-period kthread invokes first ``ct_dynticks_cpu_acquire()`` +The grace-period kthread invokes first ``ct_rcu_watching_cpu_acquire()`` (preceded by a full memory barrier) and ``rcu_dynticks_in_eqs_since()`` (both of which rely on acquire semantics) to detect idle CPUs. diff --git a/include/linux/context_tracking_state.h b/include/linux/context_tracking_state.h index ad5a06a42b4a..ad6570ffeff3 100644 --- a/include/linux/context_tracking_state.h +++ b/include/linux/context_tracking_state.h @@ -68,7 +68,7 @@ static __always_inline int ct_rcu_watching_cpu(int cpu) return atomic_read(&ct->state) & CT_RCU_WATCHING_MASK; } -static __always_inline int ct_dynticks_cpu_acquire(int cpu) +static __always_inline int ct_rcu_watching_cpu_acquire(int cpu) { struct context_tracking *ct = per_cpu_ptr(&context_tracking, cpu); diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index e7f612e9f7e5..45a9f3667c2a 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -323,7 +323,7 @@ static bool rcu_dynticks_in_eqs_since(struct rcu_data *rdp, int snap) * performed by the remote CPU prior to entering idle and therefore can * rely solely on acquire semantics. */ - return snap != ct_dynticks_cpu_acquire(rdp->cpu); + return snap != ct_rcu_watching_cpu_acquire(rdp->cpu); } /* @@ -782,7 +782,7 @@ static int dyntick_save_progress_counter(struct rcu_data *rdp) * Ordering between remote CPU's pre idle accesses and post grace period * updater's accesses is enforced by the below acquire semantic. */ - rdp->dynticks_snap = ct_dynticks_cpu_acquire(rdp->cpu); + rdp->dynticks_snap = ct_rcu_watching_cpu_acquire(rdp->cpu); if (rcu_dynticks_in_eqs(rdp->dynticks_snap)) { trace_rcu_fqs(rcu_state.name, rdp->gp_seq, rdp->cpu, TPS("dti")); rcu_gpnum_ovf(rdp->mynode, rdp); diff --git a/kernel/rcu/tree_exp.h b/kernel/rcu/tree_exp.h index 4acd29d16fdb..daa87fec703f 100644 --- a/kernel/rcu/tree_exp.h +++ b/kernel/rcu/tree_exp.h @@ -376,7 +376,7 @@ static void __sync_rcu_exp_select_node_cpus(struct rcu_exp_work *rewp) * post grace period updater's accesses is enforced by the * below acquire semantic. */ - snap = ct_dynticks_cpu_acquire(cpu); + snap = ct_rcu_watching_cpu_acquire(cpu); if (rcu_dynticks_in_eqs(snap)) mask_ofl_test |= mask; else -- cgit v1.2.3 From 7121dd915a262aee1f5a88cf86b1543f3a9439d7 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 30 May 2024 15:45:45 +0200 Subject: rcu/nocb: Introduce nocb mutex The barrier_mutex is used currently to protect (de-)offloading operations and prevent from nocb_lock locking imbalance in rcu_barrier() and shrinker, and also from misordered RCU barrier invocation. Now since RCU (de-)offloading is going to happen on offline CPUs, an RCU barrier will have to be executed while transitionning from offloaded to de-offloaded state. And this can't happen while holding the barrier_mutex. Introduce a NOCB mutex to protect (de-)offloading transitions. The barrier_mutex is still held for now when necessary to avoid barrier callbacks reordering and nocb_lock imbalance. Signed-off-by: Frederic Weisbecker Signed-off-by: Paul E. McKenney Reviewed-by: Paul E. McKenney Signed-off-by: Neeraj Upadhyay --- kernel/rcu/tree.c | 3 +++ kernel/rcu/tree.h | 1 + kernel/rcu/tree_nocb.h | 20 ++++++++++++-------- kernel/rcu/tree_plugin.h | 1 + 4 files changed, 17 insertions(+), 8 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index e641cc681901..2b9e713854b0 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -97,6 +97,9 @@ 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. */ diff --git a/kernel/rcu/tree.h b/kernel/rcu/tree.h index a297dc89a09c..16e6fe63d93c 100644 --- a/kernel/rcu/tree.h +++ b/kernel/rcu/tree.h @@ -421,6 +421,7 @@ struct rcu_state { 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 }; diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h index fdd0616f2fd1..16bcb8b13a5e 100644 --- a/kernel/rcu/tree_nocb.h +++ b/kernel/rcu/tree_nocb.h @@ -1141,6 +1141,7 @@ int rcu_nocb_cpu_deoffload(int cpu) int ret = 0; cpus_read_lock(); + mutex_lock(&rcu_state.nocb_mutex); mutex_lock(&rcu_state.barrier_mutex); if (rcu_rdp_is_offloaded(rdp)) { if (cpu_online(cpu)) { @@ -1153,6 +1154,7 @@ int rcu_nocb_cpu_deoffload(int cpu) } } mutex_unlock(&rcu_state.barrier_mutex); + mutex_unlock(&rcu_state.nocb_mutex); cpus_read_unlock(); return ret; @@ -1228,6 +1230,7 @@ int rcu_nocb_cpu_offload(int cpu) int ret = 0; cpus_read_lock(); + mutex_lock(&rcu_state.nocb_mutex); mutex_lock(&rcu_state.barrier_mutex); if (!rcu_rdp_is_offloaded(rdp)) { if (cpu_online(cpu)) { @@ -1240,6 +1243,7 @@ int rcu_nocb_cpu_offload(int cpu) } } mutex_unlock(&rcu_state.barrier_mutex); + mutex_unlock(&rcu_state.nocb_mutex); cpus_read_unlock(); return ret; @@ -1257,7 +1261,7 @@ 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 */ @@ -1267,7 +1271,7 @@ 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; } @@ -1285,9 +1289,9 @@ 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. @@ -1326,7 +1330,7 @@ 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; } @@ -1473,15 +1477,15 @@ 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 barrier_mutex to avoid nocb_lock imbalance from shrinker. + * 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.barrier_mutex); + 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). */ diff --git a/kernel/rcu/tree_plugin.h b/kernel/rcu/tree_plugin.h index f752b2a1d887..c662376c8af0 100644 --- a/kernel/rcu/tree_plugin.h +++ b/kernel/rcu/tree_plugin.h @@ -28,6 +28,7 @@ static bool rcu_rdp_is_offloaded(struct rcu_data *rdp) !(lockdep_is_held(&rcu_state.barrier_mutex) || (IS_ENABLED(CONFIG_HOTPLUG_CPU) && lockdep_is_cpus_held()) || 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)), -- cgit v1.2.3 From 5a4f9059a8f46fc1fbbae74781a5f7c8d74c732b Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 30 May 2024 15:45:48 +0200 Subject: rcu/nocb: Remove halfway (de-)offloading handling from rcu_core()'s QS reporting RCU core can't be running anymore while in the middle of (de-)offloading since this sort of transition now only applies to offline CPUs. The locked callback acceleration handling during the transition can therefore be removed. Signed-off-by: Frederic Weisbecker Signed-off-by: Paul E. McKenney Reviewed-by: Paul E. McKenney Signed-off-by: Neeraj Upadhyay --- kernel/rcu/tree.c | 13 ------------- 1 file changed, 13 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 2b9e713854b0..60f271f5c079 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -2386,7 +2386,6 @@ 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()); @@ -2423,23 +2422,11 @@ 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); - } } } -- cgit v1.2.3 From df7c249a0ed464b3f690c6f04b8cbc0fdbf30afc Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 30 May 2024 15:45:49 +0200 Subject: rcu/nocb: Remove halfway (de-)offloading handling from rcu_core RCU core can't be running anymore while in the middle of (de-)offloading since this sort of transition now only applies to offline CPUs. The locked callback acceleration handling during the transition can therefore be removed, along with concurrent batch execution. Signed-off-by: Frederic Weisbecker Signed-off-by: Paul E. McKenney Reviewed-by: Paul E. McKenney Signed-off-by: Neeraj Upadhyay --- kernel/rcu/tree.c | 26 ++++---------------------- 1 file changed, 4 insertions(+), 22 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 60f271f5c079..1a272c678533 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -2781,24 +2781,6 @@ 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; @@ -2818,17 +2800,17 @@ 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. */ -- cgit v1.2.3 From bae6076ebbd14cc1f1bd4de65c2db21d3ab109d7 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 30 May 2024 15:45:50 +0200 Subject: rcu/nocb: Remove SEGCBLIST_RCU_CORE RCU core can't be running anymore while in the middle of (de-)offloading since this sort of transition now only applies to offline CPUs. The SEGCBLIST_RCU_CORE state can therefore be removed. Signed-off-by: Frederic Weisbecker Signed-off-by: Paul E. McKenney Reviewed-by: Paul E. McKenney Signed-off-by: Neeraj Upadhyay --- include/linux/rcu_segcblist.h | 9 ++++----- kernel/rcu/rcu_segcblist.h | 9 --------- kernel/rcu/tree.c | 3 --- kernel/rcu/tree_nocb.h | 9 --------- 4 files changed, 4 insertions(+), 26 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/include/linux/rcu_segcblist.h b/include/linux/rcu_segcblist.h index ba95c06675e1..5469c54cd778 100644 --- a/include/linux/rcu_segcblist.h +++ b/include/linux/rcu_segcblist.h @@ -185,11 +185,10 @@ 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_LOCKING BIT(1) +#define SEGCBLIST_KTHREAD_CB BIT(2) +#define SEGCBLIST_KTHREAD_GP BIT(3) +#define SEGCBLIST_OFFLOADED BIT(4) struct rcu_segcblist { struct rcu_head *head; diff --git a/kernel/rcu/rcu_segcblist.h b/kernel/rcu/rcu_segcblist.h index 4fe877f5f654..7a0962dfee86 100644 --- a/kernel/rcu/rcu_segcblist.h +++ b/kernel/rcu/rcu_segcblist.h @@ -95,15 +95,6 @@ static inline bool rcu_segcblist_is_offloaded(struct rcu_segcblist *rsclp) 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)) - return true; - - return false; -} - /* * Are all segments following the specified segment of the specified * rcu_segcblist structure empty of callbacks? (The specified diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 1a272c678533..82e831b969e4 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] }, diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h index af44e75eb0cd..24daf606de0c 100644 --- a/kernel/rcu/tree_nocb.h +++ b/kernel/rcu/tree_nocb.h @@ -1060,7 +1060,6 @@ static int rcu_nocb_rdp_deoffload(struct rcu_data *rdp) WARN_ON_ONCE(rcu_cblist_n_cbs(&rdp->nocb_bypass)); WARN_ON_ONCE(rcu_segcblist_n_cbs(&rdp->cblist)); - rcu_segcblist_set_flags(cblist, SEGCBLIST_RCU_CORE); wake_gp = rdp_offload_toggle(rdp, false, flags); mutex_lock(&rdp_gp->nocb_gp_kthread_mutex); @@ -1168,13 +1167,6 @@ static int rcu_nocb_rdp_offload(struct rcu_data *rdp) swait_event_exclusive(rdp->nocb_state_wq, rcu_segcblist_test_flags(cblist, SEGCBLIST_KTHREAD_GP)); - /* - * All kthreads are ready to work, we can finally enable nocb bypass. - */ - rcu_nocb_lock_irqsave(rdp, flags); - rcu_segcblist_clear_flags(cblist, SEGCBLIST_RCU_CORE); - rcu_nocb_unlock_irqrestore(rdp, flags); - return 0; } @@ -1350,7 +1342,6 @@ void __init rcu_init_nohz(void) 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_organize_nocb_kthreads(); } -- cgit v1.2.3 From 3471e96bcf53731639587580fd02e5a297940b91 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Tue, 18 Jun 2024 10:40:30 -0700 Subject: rcu/kfree: Warn on unexpected tail state Within the rcu_sr_normal_gp_cleanup_work() function, there is an acquire load from rcu_state.srs_done_tail, which is expected to be non-NULL. This commit adds a WARN_ON_ONCE() to check this expectation. Signed-off-by: Paul E. McKenney Signed-off-by: Neeraj Upadhyay --- kernel/rcu/tree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel/rcu/tree.c') diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index e641cc681901..0f41a81138dc 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -1649,7 +1649,7 @@ 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)); -- cgit v1.2.3 From bf66471987b4bd537bc800353ca7d39bfd1d1022 Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Tue, 16 Apr 2024 10:51:10 +0200 Subject: context_tracking, rcu: Rename struct context_tracking .dynticks_nesting into .nesting The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, reflect that change in the related helpers. [ neeraj.upadhyay: Fix htmldocs build error reported by Stephen Rothwell ] Suggested-by: Frederic Weisbecker Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- Documentation/RCU/Design/Data-Structures/Data-Structures.rst | 10 +++++----- include/linux/context_tracking_state.h | 6 +++--- include/trace/events/rcu.h | 2 +- kernel/context_tracking.c | 10 +++++----- kernel/rcu/tree.c | 8 ++++---- 5 files changed, 18 insertions(+), 18 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/Documentation/RCU/Design/Data-Structures/Data-Structures.rst b/Documentation/RCU/Design/Data-Structures/Data-Structures.rst index b34990c7c377..69860bbec202 100644 --- a/Documentation/RCU/Design/Data-Structures/Data-Structures.rst +++ b/Documentation/RCU/Design/Data-Structures/Data-Structures.rst @@ -935,7 +935,7 @@ This portion of the rcu_data structure is declared as follows: :: - 1 long dynticks_nesting; + 1 long nesting; 2 long dynticks_nmi_nesting; 3 atomic_t dynticks; 4 bool rcu_need_heavy_qs; @@ -945,7 +945,7 @@ These fields in the rcu_data structure maintain the per-CPU dyntick-idle state for the corresponding CPU. The fields may be accessed only from the corresponding CPU (and from tracing) unless otherwise stated. -The ``->dynticks_nesting`` field counts the nesting depth of process +The ``->nesting`` field counts the nesting depth of process execution, so that in normal circumstances this counter has value zero or one. NMIs, irqs, and tracers are counted by the ``->dynticks_nmi_nesting`` field. Because NMIs cannot be masked, changes @@ -960,9 +960,9 @@ process-level transitions. However, it turns out that when running in non-idle kernel context, the Linux kernel is fully capable of entering interrupt handlers that never exit and perhaps also vice versa. Therefore, whenever the -``->dynticks_nesting`` field is incremented up from zero, the +``->nesting`` field is incremented up from zero, the ``->dynticks_nmi_nesting`` field is set to a large positive number, and -whenever the ``->dynticks_nesting`` field is decremented down to zero, +whenever the ``->nesting`` field is decremented down to zero, the ``->dynticks_nmi_nesting`` field is set to zero. Assuming that the number of misnested interrupts is not sufficient to overflow the counter, this approach corrects the ``->dynticks_nmi_nesting`` field @@ -992,7 +992,7 @@ code. +-----------------------------------------------------------------------+ | **Quick Quiz**: | +-----------------------------------------------------------------------+ -| Why not simply combine the ``->dynticks_nesting`` and | +| Why not simply combine the ``->nesting`` and | | ``->dynticks_nmi_nesting`` counters into a single counter that just | | counts the number of reasons that the corresponding CPU is non-idle? | +-----------------------------------------------------------------------+ diff --git a/include/linux/context_tracking_state.h b/include/linux/context_tracking_state.h index ad6570ffeff3..65290e7677e6 100644 --- a/include/linux/context_tracking_state.h +++ b/include/linux/context_tracking_state.h @@ -39,7 +39,7 @@ struct context_tracking { atomic_t state; #endif #ifdef CONFIG_CONTEXT_TRACKING_IDLE - long dynticks_nesting; /* Track process nesting level. */ + long nesting; /* Track process nesting level. */ long dynticks_nmi_nesting; /* Track irq/NMI nesting level. */ #endif }; @@ -77,14 +77,14 @@ static __always_inline int ct_rcu_watching_cpu_acquire(int cpu) static __always_inline long ct_dynticks_nesting(void) { - return __this_cpu_read(context_tracking.dynticks_nesting); + return __this_cpu_read(context_tracking.nesting); } static __always_inline long ct_dynticks_nesting_cpu(int cpu) { struct context_tracking *ct = per_cpu_ptr(&context_tracking, cpu); - return ct->dynticks_nesting; + return ct->nesting; } static __always_inline long ct_dynticks_nmi_nesting(void) diff --git a/include/trace/events/rcu.h b/include/trace/events/rcu.h index 31b3e0d3e65f..4066b6d51e46 100644 --- a/include/trace/events/rcu.h +++ b/include/trace/events/rcu.h @@ -469,7 +469,7 @@ TRACE_EVENT(rcu_stall_warning, * polarity: "Start", "End", "StillNonIdle" for entering, exiting or still not * being in dyntick-idle mode. * context: "USER" or "IDLE" or "IRQ". - * NMIs nested in IRQs are inferred with dynticks_nesting > 1 in IRQ context. + * NMIs nested in IRQs are inferred with nesting > 1 in IRQ context. * * These events also take a pair of numbers, which indicate the nesting * depth before and after the event of interest, and a third number that is diff --git a/kernel/context_tracking.c b/kernel/context_tracking.c index 868ae0bcd4be..5cfdfc03b401 100644 --- a/kernel/context_tracking.c +++ b/kernel/context_tracking.c @@ -28,7 +28,7 @@ DEFINE_PER_CPU(struct context_tracking, context_tracking) = { #ifdef CONFIG_CONTEXT_TRACKING_IDLE - .dynticks_nesting = 1, + .nesting = 1, .dynticks_nmi_nesting = DYNTICK_IRQ_NONIDLE, #endif .state = ATOMIC_INIT(CT_RCU_WATCHING), @@ -131,7 +131,7 @@ static void noinstr ct_kernel_exit(bool user, int offset) ct_dynticks_nesting() == 0); if (ct_dynticks_nesting() != 1) { // RCU will still be watching, so just do accounting and leave. - ct->dynticks_nesting--; + ct->nesting--; return; } @@ -145,7 +145,7 @@ static void noinstr ct_kernel_exit(bool user, int offset) instrument_atomic_write(&ct->state, sizeof(ct->state)); instrumentation_end(); - WRITE_ONCE(ct->dynticks_nesting, 0); /* Avoid irq-access tearing. */ + WRITE_ONCE(ct->nesting, 0); /* Avoid irq-access tearing. */ // RCU is watching here ... ct_kernel_exit_state(offset); // ... but is no longer watching here. @@ -170,7 +170,7 @@ static void noinstr ct_kernel_enter(bool user, int offset) WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && oldval < 0); if (oldval) { // RCU was already watching, so just do accounting and leave. - ct->dynticks_nesting++; + ct->nesting++; return; } rcu_dynticks_task_exit(); @@ -184,7 +184,7 @@ static void noinstr ct_kernel_enter(bool user, int offset) trace_rcu_dyntick(TPS("End"), ct_dynticks_nesting(), 1, ct_rcu_watching()); WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && !user && !is_idle_task(current)); - WRITE_ONCE(ct->dynticks_nesting, 1); + WRITE_ONCE(ct->nesting, 1); WARN_ON_ONCE(ct_dynticks_nmi_nesting()); WRITE_ONCE(ct->dynticks_nmi_nesting, DYNTICK_IRQ_NONIDLE); instrumentation_end(); diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 45a9f3667c2a..0d5a539acd58 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -389,7 +389,7 @@ static int rcu_is_cpu_rrupt_from_idle(void) /* Check for counter underflows */ RCU_LOCKDEP_WARN(ct_dynticks_nesting() < 0, - "RCU dynticks_nesting counter underflow!"); + "RCU nesting counter underflow!"); RCU_LOCKDEP_WARN(ct_dynticks_nmi_nesting() <= 0, "RCU dynticks_nmi_nesting counter underflow/zero!"); @@ -597,7 +597,7 @@ void rcu_irq_exit_check_preempt(void) lockdep_assert_irqs_disabled(); RCU_LOCKDEP_WARN(ct_dynticks_nesting() <= 0, - "RCU dynticks_nesting counter underflow/zero!"); + "RCU nesting counter underflow/zero!"); RCU_LOCKDEP_WARN(ct_dynticks_nmi_nesting() != DYNTICK_IRQ_NONIDLE, "Bad RCU dynticks_nmi_nesting counter\n"); @@ -4804,7 +4804,7 @@ rcu_boot_init_percpu_data(int cpu) /* Set up local state, ensuring consistent view of global state. */ rdp->grpmask = leaf_node_cpu_bit(rdp->mynode, cpu); INIT_WORK(&rdp->strict_work, strict_work_handler); - WARN_ON_ONCE(ct->dynticks_nesting != 1); + WARN_ON_ONCE(ct->nesting != 1); WARN_ON_ONCE(rcu_dynticks_in_eqs(ct_rcu_watching_cpu(cpu))); rdp->barrier_seq_snap = rcu_state.barrier_sequence; rdp->rcu_ofl_gp_seq = rcu_state.gp_seq; @@ -4898,7 +4898,7 @@ int rcutree_prepare_cpu(unsigned int cpu) rdp->qlen_last_fqs_check = 0; rdp->n_force_qs_snap = READ_ONCE(rcu_state.n_force_qs); rdp->blimit = blimit; - ct->dynticks_nesting = 1; /* CPU not up, no tearing. */ + ct->nesting = 1; /* CPU not up, no tearing. */ raw_spin_unlock_rcu_node(rnp); /* irqs remain disabled. */ /* -- cgit v1.2.3 From 1089c0078b69bce0c2d540e3cc43470ba9e5dcfd Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Tue, 16 Apr 2024 14:45:30 +0200 Subject: context_tracking, rcu: Rename ct_dynticks_nesting() into ct_nesting() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, reflect that change in the related helpers. Suggested-by: Frederic Weisbecker Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- include/linux/context_tracking_state.h | 2 +- kernel/context_tracking.c | 10 +++++----- kernel/rcu/tree.c | 6 +++--- 3 files changed, 9 insertions(+), 9 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/include/linux/context_tracking_state.h b/include/linux/context_tracking_state.h index 65290e7677e6..586c1ff22c2e 100644 --- a/include/linux/context_tracking_state.h +++ b/include/linux/context_tracking_state.h @@ -75,7 +75,7 @@ static __always_inline int ct_rcu_watching_cpu_acquire(int cpu) return atomic_read_acquire(&ct->state) & CT_RCU_WATCHING_MASK; } -static __always_inline long ct_dynticks_nesting(void) +static __always_inline long ct_nesting(void) { return __this_cpu_read(context_tracking.nesting); } diff --git a/kernel/context_tracking.c b/kernel/context_tracking.c index 5cfdfc03b401..a951bde0bbcb 100644 --- a/kernel/context_tracking.c +++ b/kernel/context_tracking.c @@ -128,8 +128,8 @@ static void noinstr ct_kernel_exit(bool user, int offset) WARN_ON_ONCE(ct_dynticks_nmi_nesting() != DYNTICK_IRQ_NONIDLE); WRITE_ONCE(ct->dynticks_nmi_nesting, 0); WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && - ct_dynticks_nesting() == 0); - if (ct_dynticks_nesting() != 1) { + ct_nesting() == 0); + if (ct_nesting() != 1) { // RCU will still be watching, so just do accounting and leave. ct->nesting--; return; @@ -137,7 +137,7 @@ static void noinstr ct_kernel_exit(bool user, int offset) instrumentation_begin(); lockdep_assert_irqs_disabled(); - trace_rcu_dyntick(TPS("Start"), ct_dynticks_nesting(), 0, ct_rcu_watching()); + trace_rcu_dyntick(TPS("Start"), ct_nesting(), 0, ct_rcu_watching()); WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && !user && !is_idle_task(current)); rcu_preempt_deferred_qs(current); @@ -166,7 +166,7 @@ static void noinstr ct_kernel_enter(bool user, int offset) long oldval; WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && !raw_irqs_disabled()); - oldval = ct_dynticks_nesting(); + oldval = ct_nesting(); WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && oldval < 0); if (oldval) { // RCU was already watching, so just do accounting and leave. @@ -182,7 +182,7 @@ static void noinstr ct_kernel_enter(bool user, int offset) // instrumentation for the noinstr ct_kernel_enter_state() instrument_atomic_write(&ct->state, sizeof(ct->state)); - trace_rcu_dyntick(TPS("End"), ct_dynticks_nesting(), 1, ct_rcu_watching()); + trace_rcu_dyntick(TPS("End"), ct_nesting(), 1, ct_rcu_watching()); WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && !user && !is_idle_task(current)); WRITE_ONCE(ct->nesting, 1); WARN_ON_ONCE(ct_dynticks_nmi_nesting()); diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 0d5a539acd58..cf69a234080f 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -388,7 +388,7 @@ static int rcu_is_cpu_rrupt_from_idle(void) lockdep_assert_irqs_disabled(); /* Check for counter underflows */ - RCU_LOCKDEP_WARN(ct_dynticks_nesting() < 0, + RCU_LOCKDEP_WARN(ct_nesting() < 0, "RCU nesting counter underflow!"); RCU_LOCKDEP_WARN(ct_dynticks_nmi_nesting() <= 0, "RCU dynticks_nmi_nesting counter underflow/zero!"); @@ -404,7 +404,7 @@ static int rcu_is_cpu_rrupt_from_idle(void) WARN_ON_ONCE(!nesting && !is_idle_task(current)); /* Does CPU appear to be idle from an RCU standpoint? */ - return ct_dynticks_nesting() == 0; + return ct_nesting() == 0; } #define DEFAULT_RCU_BLIMIT (IS_ENABLED(CONFIG_RCU_STRICT_GRACE_PERIOD) ? 1000 : 10) @@ -596,7 +596,7 @@ void rcu_irq_exit_check_preempt(void) { lockdep_assert_irqs_disabled(); - RCU_LOCKDEP_WARN(ct_dynticks_nesting() <= 0, + RCU_LOCKDEP_WARN(ct_nesting() <= 0, "RCU nesting counter underflow/zero!"); RCU_LOCKDEP_WARN(ct_dynticks_nmi_nesting() != DYNTICK_IRQ_NONIDLE, -- cgit v1.2.3 From dc5fface4b30508933b515a4985401c8c8f6bcfd Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Tue, 16 Apr 2024 10:52:03 +0200 Subject: context_tracking, rcu: Rename struct context_tracking .dynticks_nmi_nesting into .nmi_nesting The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, and the 'dynticks' prefix can be dropped without losing any meaning. [ neeraj.upadhyay: Fix htmldocs build error reported by Stephen Rothwell ] Suggested-by: Frederic Weisbecker Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- .../RCU/Design/Data-Structures/Data-Structures.rst | 14 +++++----- include/linux/context_tracking_state.h | 6 ++--- kernel/context_tracking.c | 30 +++++++++++----------- kernel/rcu/tree.c | 4 +-- 4 files changed, 27 insertions(+), 27 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/Documentation/RCU/Design/Data-Structures/Data-Structures.rst b/Documentation/RCU/Design/Data-Structures/Data-Structures.rst index 69860bbec202..28d9f9e1783f 100644 --- a/Documentation/RCU/Design/Data-Structures/Data-Structures.rst +++ b/Documentation/RCU/Design/Data-Structures/Data-Structures.rst @@ -936,7 +936,7 @@ This portion of the rcu_data structure is declared as follows: :: 1 long nesting; - 2 long dynticks_nmi_nesting; + 2 long nmi_nesting; 3 atomic_t dynticks; 4 bool rcu_need_heavy_qs; 5 bool rcu_urgent_qs; @@ -948,11 +948,11 @@ the corresponding CPU (and from tracing) unless otherwise stated. The ``->nesting`` field counts the nesting depth of process execution, so that in normal circumstances this counter has value zero or one. NMIs, irqs, and tracers are counted by the -``->dynticks_nmi_nesting`` field. Because NMIs cannot be masked, changes +``->nmi_nesting`` field. Because NMIs cannot be masked, changes to this variable have to be undertaken carefully using an algorithm provided by Andy Lutomirski. The initial transition from idle adds one, and nested transitions add two, so that a nesting level of five is -represented by a ``->dynticks_nmi_nesting`` value of nine. This counter +represented by a ``->nmi_nesting`` value of nine. This counter can therefore be thought of as counting the number of reasons why this CPU cannot be permitted to enter dyntick-idle mode, aside from process-level transitions. @@ -961,11 +961,11 @@ However, it turns out that when running in non-idle kernel context, the Linux kernel is fully capable of entering interrupt handlers that never exit and perhaps also vice versa. Therefore, whenever the ``->nesting`` field is incremented up from zero, the -``->dynticks_nmi_nesting`` field is set to a large positive number, and +``->nmi_nesting`` field is set to a large positive number, and whenever the ``->nesting`` field is decremented down to zero, -the ``->dynticks_nmi_nesting`` field is set to zero. Assuming that +the ``->nmi_nesting`` field is set to zero. Assuming that the number of misnested interrupts is not sufficient to overflow the -counter, this approach corrects the ``->dynticks_nmi_nesting`` field +counter, this approach corrects the ``->nmi_nesting`` field every time the corresponding CPU enters the idle loop from process context. @@ -993,7 +993,7 @@ code. | **Quick Quiz**: | +-----------------------------------------------------------------------+ | Why not simply combine the ``->nesting`` and | -| ``->dynticks_nmi_nesting`` counters into a single counter that just | +| ``->nmi_nesting`` counters into a single counter that just | | counts the number of reasons that the corresponding CPU is non-idle? | +-----------------------------------------------------------------------+ | **Answer**: | diff --git a/include/linux/context_tracking_state.h b/include/linux/context_tracking_state.h index fd42d8120ac2..12d00adf29e1 100644 --- a/include/linux/context_tracking_state.h +++ b/include/linux/context_tracking_state.h @@ -40,7 +40,7 @@ struct context_tracking { #endif #ifdef CONFIG_CONTEXT_TRACKING_IDLE long nesting; /* Track process nesting level. */ - long dynticks_nmi_nesting; /* Track irq/NMI nesting level. */ + long nmi_nesting; /* Track irq/NMI nesting level. */ #endif }; @@ -89,14 +89,14 @@ static __always_inline long ct_nesting_cpu(int cpu) static __always_inline long ct_dynticks_nmi_nesting(void) { - return __this_cpu_read(context_tracking.dynticks_nmi_nesting); + return __this_cpu_read(context_tracking.nmi_nesting); } static __always_inline long ct_dynticks_nmi_nesting_cpu(int cpu) { struct context_tracking *ct = per_cpu_ptr(&context_tracking, cpu); - return ct->dynticks_nmi_nesting; + return ct->nmi_nesting; } #endif /* #ifdef CONFIG_CONTEXT_TRACKING_IDLE */ diff --git a/kernel/context_tracking.c b/kernel/context_tracking.c index a951bde0bbcb..ae94215aa132 100644 --- a/kernel/context_tracking.c +++ b/kernel/context_tracking.c @@ -29,7 +29,7 @@ DEFINE_PER_CPU(struct context_tracking, context_tracking) = { #ifdef CONFIG_CONTEXT_TRACKING_IDLE .nesting = 1, - .dynticks_nmi_nesting = DYNTICK_IRQ_NONIDLE, + .nmi_nesting = DYNTICK_IRQ_NONIDLE, #endif .state = ATOMIC_INIT(CT_RCU_WATCHING), }; @@ -117,7 +117,7 @@ static noinstr void ct_kernel_enter_state(int offset) * Enter an RCU extended quiescent state, which can be either the * idle loop or adaptive-tickless usermode execution. * - * We crowbar the ->dynticks_nmi_nesting field to zero to allow for + * We crowbar the ->nmi_nesting field to zero to allow for * the possibility of usermode upcalls having messed up our count * of interrupt nesting level during the prior busy period. */ @@ -126,7 +126,7 @@ static void noinstr ct_kernel_exit(bool user, int offset) struct context_tracking *ct = this_cpu_ptr(&context_tracking); WARN_ON_ONCE(ct_dynticks_nmi_nesting() != DYNTICK_IRQ_NONIDLE); - WRITE_ONCE(ct->dynticks_nmi_nesting, 0); + WRITE_ONCE(ct->nmi_nesting, 0); WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && ct_nesting() == 0); if (ct_nesting() != 1) { @@ -156,7 +156,7 @@ static void noinstr ct_kernel_exit(bool user, int offset) * Exit an RCU extended quiescent state, which can be either the * idle loop or adaptive-tickless usermode execution. * - * We crowbar the ->dynticks_nmi_nesting field to DYNTICK_IRQ_NONIDLE to + * We crowbar the ->nmi_nesting field to DYNTICK_IRQ_NONIDLE to * allow for the possibility of usermode upcalls messing up our count of * interrupt nesting level during the busy period that is just now starting. */ @@ -186,7 +186,7 @@ static void noinstr ct_kernel_enter(bool user, int offset) WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && !user && !is_idle_task(current)); WRITE_ONCE(ct->nesting, 1); WARN_ON_ONCE(ct_dynticks_nmi_nesting()); - WRITE_ONCE(ct->dynticks_nmi_nesting, DYNTICK_IRQ_NONIDLE); + WRITE_ONCE(ct->nmi_nesting, DYNTICK_IRQ_NONIDLE); instrumentation_end(); } @@ -194,7 +194,7 @@ static void noinstr ct_kernel_enter(bool user, int offset) * ct_nmi_exit - inform RCU of exit from NMI context * * If we are returning from the outermost NMI handler that interrupted an - * RCU-idle period, update ct->state and ct->dynticks_nmi_nesting + * RCU-idle period, update ct->state and ct->nmi_nesting * to let the RCU grace-period handling know that the CPU is back to * being RCU-idle. * @@ -207,7 +207,7 @@ void noinstr ct_nmi_exit(void) instrumentation_begin(); /* - * Check for ->dynticks_nmi_nesting underflow and bad ->dynticks. + * Check for ->nmi_nesting underflow and bad ->dynticks. * (We are exiting an NMI handler, so RCU better be paying attention * to us!) */ @@ -221,7 +221,7 @@ void noinstr ct_nmi_exit(void) if (ct_dynticks_nmi_nesting() != 1) { trace_rcu_dyntick(TPS("--="), ct_dynticks_nmi_nesting(), ct_dynticks_nmi_nesting() - 2, ct_rcu_watching()); - WRITE_ONCE(ct->dynticks_nmi_nesting, /* No store tearing. */ + WRITE_ONCE(ct->nmi_nesting, /* No store tearing. */ ct_dynticks_nmi_nesting() - 2); instrumentation_end(); return; @@ -229,7 +229,7 @@ void noinstr ct_nmi_exit(void) /* This NMI interrupted an RCU-idle CPU, restore RCU-idleness. */ trace_rcu_dyntick(TPS("Startirq"), ct_dynticks_nmi_nesting(), 0, ct_rcu_watching()); - WRITE_ONCE(ct->dynticks_nmi_nesting, 0); /* Avoid store tearing. */ + WRITE_ONCE(ct->nmi_nesting, 0); /* Avoid store tearing. */ // instrumentation for the noinstr ct_kernel_exit_state() instrument_atomic_write(&ct->state, sizeof(ct->state)); @@ -247,7 +247,7 @@ void noinstr ct_nmi_exit(void) * ct_nmi_enter - inform RCU of entry to NMI context * * If the CPU was idle from RCU's viewpoint, update ct->state and - * ct->dynticks_nmi_nesting to let the RCU grace-period handling know + * ct->nmi_nesting to let the RCU grace-period handling know * that the CPU is active. This implementation permits nested NMIs, as * long as the nesting level does not overflow an int. (You will probably * run out of stack space first.) @@ -264,10 +264,10 @@ void noinstr ct_nmi_enter(void) WARN_ON_ONCE(ct_dynticks_nmi_nesting() < 0); /* - * If idle from RCU viewpoint, atomically increment ->dynticks - * to mark non-idle and increment ->dynticks_nmi_nesting by one. - * Otherwise, increment ->dynticks_nmi_nesting by two. This means - * if ->dynticks_nmi_nesting is equal to one, we are guaranteed + * If idle from RCU viewpoint, atomically increment CT state + * to mark non-idle and increment ->nmi_nesting by one. + * Otherwise, increment ->nmi_nesting by two. This means + * if ->nmi_nesting is equal to one, we are guaranteed * to be in the outermost NMI handler that interrupted an RCU-idle * period (observation due to Andy Lutomirski). */ @@ -298,7 +298,7 @@ void noinstr ct_nmi_enter(void) ct_dynticks_nmi_nesting(), ct_dynticks_nmi_nesting() + incby, ct_rcu_watching()); instrumentation_end(); - WRITE_ONCE(ct->dynticks_nmi_nesting, /* Prevent store tearing. */ + WRITE_ONCE(ct->nmi_nesting, /* Prevent store tearing. */ ct_dynticks_nmi_nesting() + incby); barrier(); } diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index cf69a234080f..934f6b34a551 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -391,7 +391,7 @@ static int rcu_is_cpu_rrupt_from_idle(void) RCU_LOCKDEP_WARN(ct_nesting() < 0, "RCU nesting counter underflow!"); RCU_LOCKDEP_WARN(ct_dynticks_nmi_nesting() <= 0, - "RCU dynticks_nmi_nesting counter underflow/zero!"); + "RCU nmi_nesting counter underflow/zero!"); /* Are we at first interrupt nesting level? */ nesting = ct_dynticks_nmi_nesting(); @@ -600,7 +600,7 @@ void rcu_irq_exit_check_preempt(void) "RCU nesting counter underflow/zero!"); RCU_LOCKDEP_WARN(ct_dynticks_nmi_nesting() != DYNTICK_IRQ_NONIDLE, - "Bad RCU dynticks_nmi_nesting counter\n"); + "Bad RCU nmi_nesting counter\n"); RCU_LOCKDEP_WARN(rcu_dynticks_curr_cpu_in_eqs(), "RCU in extended quiescent state!"); } -- cgit v1.2.3 From 8375cb260d7e43610934af63748d1a51201449f8 Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Tue, 16 Apr 2024 14:46:14 +0200 Subject: context_tracking, rcu: Rename ct_dynticks_nmi_nesting() into ct_nmi_nesting() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, and the 'dynticks' prefix can be dropped without losing any meaning. Suggested-by: Frederic Weisbecker Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- include/linux/context_tracking_state.h | 2 +- kernel/context_tracking.c | 24 ++++++++++++------------ kernel/rcu/tree.c | 6 +++--- 3 files changed, 16 insertions(+), 16 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/include/linux/context_tracking_state.h b/include/linux/context_tracking_state.h index 12d00adf29e1..8f32fe599c5c 100644 --- a/include/linux/context_tracking_state.h +++ b/include/linux/context_tracking_state.h @@ -87,7 +87,7 @@ static __always_inline long ct_nesting_cpu(int cpu) return ct->nesting; } -static __always_inline long ct_dynticks_nmi_nesting(void) +static __always_inline long ct_nmi_nesting(void) { return __this_cpu_read(context_tracking.nmi_nesting); } diff --git a/kernel/context_tracking.c b/kernel/context_tracking.c index ae94215aa132..115843eeb030 100644 --- a/kernel/context_tracking.c +++ b/kernel/context_tracking.c @@ -125,7 +125,7 @@ static void noinstr ct_kernel_exit(bool user, int offset) { struct context_tracking *ct = this_cpu_ptr(&context_tracking); - WARN_ON_ONCE(ct_dynticks_nmi_nesting() != DYNTICK_IRQ_NONIDLE); + WARN_ON_ONCE(ct_nmi_nesting() != DYNTICK_IRQ_NONIDLE); WRITE_ONCE(ct->nmi_nesting, 0); WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && ct_nesting() == 0); @@ -185,7 +185,7 @@ static void noinstr ct_kernel_enter(bool user, int offset) trace_rcu_dyntick(TPS("End"), ct_nesting(), 1, ct_rcu_watching()); WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && !user && !is_idle_task(current)); WRITE_ONCE(ct->nesting, 1); - WARN_ON_ONCE(ct_dynticks_nmi_nesting()); + WARN_ON_ONCE(ct_nmi_nesting()); WRITE_ONCE(ct->nmi_nesting, DYNTICK_IRQ_NONIDLE); instrumentation_end(); } @@ -207,28 +207,28 @@ void noinstr ct_nmi_exit(void) instrumentation_begin(); /* - * Check for ->nmi_nesting underflow and bad ->dynticks. + * Check for ->nmi_nesting underflow and bad CT state. * (We are exiting an NMI handler, so RCU better be paying attention * to us!) */ - WARN_ON_ONCE(ct_dynticks_nmi_nesting() <= 0); + WARN_ON_ONCE(ct_nmi_nesting() <= 0); WARN_ON_ONCE(rcu_dynticks_curr_cpu_in_eqs()); /* * If the nesting level is not 1, the CPU wasn't RCU-idle, so * leave it in non-RCU-idle state. */ - if (ct_dynticks_nmi_nesting() != 1) { - trace_rcu_dyntick(TPS("--="), ct_dynticks_nmi_nesting(), ct_dynticks_nmi_nesting() - 2, + if (ct_nmi_nesting() != 1) { + trace_rcu_dyntick(TPS("--="), ct_nmi_nesting(), ct_nmi_nesting() - 2, ct_rcu_watching()); WRITE_ONCE(ct->nmi_nesting, /* No store tearing. */ - ct_dynticks_nmi_nesting() - 2); + ct_nmi_nesting() - 2); instrumentation_end(); return; } /* This NMI interrupted an RCU-idle CPU, restore RCU-idleness. */ - trace_rcu_dyntick(TPS("Startirq"), ct_dynticks_nmi_nesting(), 0, ct_rcu_watching()); + trace_rcu_dyntick(TPS("Startirq"), ct_nmi_nesting(), 0, ct_rcu_watching()); WRITE_ONCE(ct->nmi_nesting, 0); /* Avoid store tearing. */ // instrumentation for the noinstr ct_kernel_exit_state() @@ -261,7 +261,7 @@ void noinstr ct_nmi_enter(void) struct context_tracking *ct = this_cpu_ptr(&context_tracking); /* Complain about underflow. */ - WARN_ON_ONCE(ct_dynticks_nmi_nesting() < 0); + WARN_ON_ONCE(ct_nmi_nesting() < 0); /* * If idle from RCU viewpoint, atomically increment CT state @@ -295,11 +295,11 @@ void noinstr ct_nmi_enter(void) } trace_rcu_dyntick(incby == 1 ? TPS("Endirq") : TPS("++="), - ct_dynticks_nmi_nesting(), - ct_dynticks_nmi_nesting() + incby, ct_rcu_watching()); + ct_nmi_nesting(), + ct_nmi_nesting() + incby, ct_rcu_watching()); instrumentation_end(); WRITE_ONCE(ct->nmi_nesting, /* Prevent store tearing. */ - ct_dynticks_nmi_nesting() + incby); + ct_nmi_nesting() + incby); barrier(); } diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 934f6b34a551..14cc314eedad 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -390,11 +390,11 @@ static int rcu_is_cpu_rrupt_from_idle(void) /* Check for counter underflows */ RCU_LOCKDEP_WARN(ct_nesting() < 0, "RCU nesting counter underflow!"); - RCU_LOCKDEP_WARN(ct_dynticks_nmi_nesting() <= 0, + RCU_LOCKDEP_WARN(ct_nmi_nesting() <= 0, "RCU nmi_nesting counter underflow/zero!"); /* Are we at first interrupt nesting level? */ - nesting = ct_dynticks_nmi_nesting(); + nesting = ct_nmi_nesting(); if (nesting > 1) return false; @@ -598,7 +598,7 @@ void rcu_irq_exit_check_preempt(void) RCU_LOCKDEP_WARN(ct_nesting() <= 0, "RCU nesting counter underflow/zero!"); - RCU_LOCKDEP_WARN(ct_dynticks_nmi_nesting() != + RCU_LOCKDEP_WARN(ct_nmi_nesting() != DYNTICK_IRQ_NONIDLE, "Bad RCU nmi_nesting counter\n"); RCU_LOCKDEP_WARN(rcu_dynticks_curr_cpu_in_eqs(), -- cgit v1.2.3 From e1de438336222dbe27f2d4baa8c01074fcac2bef Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Mon, 29 Apr 2024 15:52:32 +0200 Subject: context_tracking, rcu: Rename DYNTICK_IRQ_NONIDLE into CT_NESTING_IRQ_NONIDLE The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, and the 'dynticks' prefix can be dropped without losing any meaning. Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- include/linux/context_tracking_state.h | 2 +- kernel/context_tracking.c | 8 ++++---- kernel/rcu/tree.c | 2 +- 3 files changed, 6 insertions(+), 6 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/include/linux/context_tracking_state.h b/include/linux/context_tracking_state.h index 34fd504e53a8..0dbda59c9f37 100644 --- a/include/linux/context_tracking_state.h +++ b/include/linux/context_tracking_state.h @@ -7,7 +7,7 @@ #include /* Offset to allow distinguishing irq vs. task-based idle entry/exit. */ -#define DYNTICK_IRQ_NONIDLE ((LONG_MAX / 2) + 1) +#define CT_NESTING_IRQ_NONIDLE ((LONG_MAX / 2) + 1) enum ctx_state { CT_STATE_DISABLED = -1, /* returned by ct_state() if unknown */ diff --git a/kernel/context_tracking.c b/kernel/context_tracking.c index 115843eeb030..8262f57a4363 100644 --- a/kernel/context_tracking.c +++ b/kernel/context_tracking.c @@ -29,7 +29,7 @@ DEFINE_PER_CPU(struct context_tracking, context_tracking) = { #ifdef CONFIG_CONTEXT_TRACKING_IDLE .nesting = 1, - .nmi_nesting = DYNTICK_IRQ_NONIDLE, + .nmi_nesting = CT_NESTING_IRQ_NONIDLE, #endif .state = ATOMIC_INIT(CT_RCU_WATCHING), }; @@ -125,7 +125,7 @@ static void noinstr ct_kernel_exit(bool user, int offset) { struct context_tracking *ct = this_cpu_ptr(&context_tracking); - WARN_ON_ONCE(ct_nmi_nesting() != DYNTICK_IRQ_NONIDLE); + WARN_ON_ONCE(ct_nmi_nesting() != CT_NESTING_IRQ_NONIDLE); WRITE_ONCE(ct->nmi_nesting, 0); WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && ct_nesting() == 0); @@ -156,7 +156,7 @@ static void noinstr ct_kernel_exit(bool user, int offset) * Exit an RCU extended quiescent state, which can be either the * idle loop or adaptive-tickless usermode execution. * - * We crowbar the ->nmi_nesting field to DYNTICK_IRQ_NONIDLE to + * We crowbar the ->nmi_nesting field to CT_NESTING_IRQ_NONIDLE to * allow for the possibility of usermode upcalls messing up our count of * interrupt nesting level during the busy period that is just now starting. */ @@ -186,7 +186,7 @@ static void noinstr ct_kernel_enter(bool user, int offset) WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) && !user && !is_idle_task(current)); WRITE_ONCE(ct->nesting, 1); WARN_ON_ONCE(ct_nmi_nesting()); - WRITE_ONCE(ct->nmi_nesting, DYNTICK_IRQ_NONIDLE); + WRITE_ONCE(ct->nmi_nesting, CT_NESTING_IRQ_NONIDLE); instrumentation_end(); } diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 14cc314eedad..4778c873f4ac 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -599,7 +599,7 @@ void rcu_irq_exit_check_preempt(void) RCU_LOCKDEP_WARN(ct_nesting() <= 0, "RCU nesting counter underflow/zero!"); RCU_LOCKDEP_WARN(ct_nmi_nesting() != - DYNTICK_IRQ_NONIDLE, + CT_NESTING_IRQ_NONIDLE, "Bad RCU nmi_nesting counter\n"); RCU_LOCKDEP_WARN(rcu_dynticks_curr_cpu_in_eqs(), "RCU in extended quiescent state!"); -- cgit v1.2.3 From fb579e6656a8d5e042e65062f68ca41321817237 Mon Sep 17 00:00:00 2001 From: Thorsten Blum Date: Wed, 7 Aug 2024 11:55:00 +0200 Subject: rcu: Annotate struct kvfree_rcu_bulk_data with __counted_by() Add the __counted_by compiler attribute to the flexible array member records to improve access bounds-checking via CONFIG_UBSAN_BOUNDS and CONFIG_FORTIFY_SOURCE. Increment nr_records before adding a new pointer to the records array. Signed-off-by: Thorsten Blum Reviewed-by: "Gustavo A. R. Silva" Reviewed-by: "Uladzislau Rezki (Sony)" Reviewed-by: Paul E. McKenney Signed-off-by: Neeraj Upadhyay --- kernel/rcu/tree.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 0f41a81138dc..d5bf824159da 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -3227,7 +3227,7 @@ 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); }; /* @@ -3767,7 +3767,8 @@ 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]); -- cgit v1.2.3 From 0aac9daef6763e6efef398faff71f8c593651cce Mon Sep 17 00:00:00 2001 From: Waiman Long Date: Tue, 23 Jul 2024 14:10:25 -0400 Subject: rcu: Use system_unbound_wq to avoid disturbing isolated CPUs It was discovered that isolated CPUs could sometimes be disturbed by kworkers processing kfree_rcu() works causing higher than expected latency. It is because the RCU core uses "system_wq" which doesn't have the WQ_UNBOUND flag to handle all its work items. Fix this violation of latency limits by using "system_unbound_wq" in the RCU core instead. This will ensure that those work items will not be run on CPUs marked as isolated. Beside the WQ_UNBOUND flag, the other major difference between system_wq and system_unbound_wq is their max_active count. The system_unbound_wq has a max_active of WQ_MAX_ACTIVE (512) while system_wq's max_active is WQ_DFL_ACTIVE (256) which is half of WQ_MAX_ACTIVE. Reported-by: Vratislav Bendel Closes: https://issues.redhat.com/browse/RHEL-50220 Signed-off-by: Waiman Long Reviewed-by: "Uladzislau Rezki (Sony)" Tested-by: Breno Leitao Signed-off-by: Neeraj Upadhyay --- kernel/rcu/tree.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index e641cc681901..494aa9513d0b 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -3539,10 +3539,10 @@ 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 @@ -3634,7 +3634,7 @@ 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); } } @@ -3704,7 +3704,7 @@ 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 { -- cgit v1.2.3 From 0616f7e981968743edb1ef758ba1a0e984240419 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Thu, 1 Aug 2024 17:43:01 -0700 Subject: rcu: Mark callbacks not currently participating in barrier operation RCU keeps a count of the number of callbacks that the current rcu_barrier() is waiting on, but there is currently no easy way to work out which callback is stuck. One way to do this is to mark idle RCU-barrier callbacks by making the ->next pointer point to the callback itself, and this commit does just that. Later commits will use this for debug output. Signed-off-by: "Paul E. McKenney" Signed-off-by: Neeraj Upadhyay --- kernel/rcu/tree.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel/rcu/tree.c') diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index e641cc681901..f931171daecd 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -4403,6 +4403,7 @@ 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); @@ -5424,6 +5425,8 @@ 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); } } -- cgit v1.2.3 From fda70207135b60dd1a7c8f16cc036bb1cc344490 Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Tue, 16 Apr 2024 11:01:23 +0200 Subject: context_tracking, rcu: Rename rcu_dynticks_curr_cpu_in_eqs() into rcu_is_watching_curr_cpu() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, reflect that change in the related helpers. Note that "watching" is the opposite of "in EQS", so the negation is lifted out of the helper and into the callsites. Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- include/linux/context_tracking.h | 12 ++++++++---- kernel/context_tracking.c | 6 +++--- kernel/rcu/tree.c | 6 +++--- 3 files changed, 14 insertions(+), 10 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/include/linux/context_tracking.h b/include/linux/context_tracking.h index a6c36780cc3b..d53092ffa9db 100644 --- a/include/linux/context_tracking.h +++ b/include/linux/context_tracking.h @@ -113,13 +113,17 @@ extern void ct_idle_enter(void); extern void ct_idle_exit(void); /* - * Is the current CPU in an extended quiescent state? + * Is RCU watching the current CPU (IOW, it is not in an extended quiescent state)? + * + * Note that this returns the actual boolean data (watching / not watching), + * whereas ct_rcu_watching() returns the RCU_WATCHING subvariable of + * context_tracking.state. * * No ordering, as we are sampling CPU-local information. */ -static __always_inline bool rcu_dynticks_curr_cpu_in_eqs(void) +static __always_inline bool rcu_is_watching_curr_cpu(void) { - return !(raw_atomic_read(this_cpu_ptr(&context_tracking.state)) & CT_RCU_WATCHING); + return raw_atomic_read(this_cpu_ptr(&context_tracking.state)) & CT_RCU_WATCHING; } /* @@ -140,7 +144,7 @@ static __always_inline bool warn_rcu_enter(void) * lots of the actual reporting also relies on RCU. */ preempt_disable_notrace(); - if (rcu_dynticks_curr_cpu_in_eqs()) { + if (!rcu_is_watching_curr_cpu()) { ret = true; ct_state_inc(CT_RCU_WATCHING); } diff --git a/kernel/context_tracking.c b/kernel/context_tracking.c index 00c8f128885a..31c6d0c1b79d 100644 --- a/kernel/context_tracking.c +++ b/kernel/context_tracking.c @@ -212,7 +212,7 @@ void noinstr ct_nmi_exit(void) * to us!) */ WARN_ON_ONCE(ct_nmi_nesting() <= 0); - WARN_ON_ONCE(rcu_dynticks_curr_cpu_in_eqs()); + WARN_ON_ONCE(!rcu_is_watching_curr_cpu()); /* * If the nesting level is not 1, the CPU wasn't RCU-idle, so @@ -271,7 +271,7 @@ void noinstr ct_nmi_enter(void) * to be in the outermost NMI handler that interrupted an RCU-idle * period (observation due to Andy Lutomirski). */ - if (rcu_dynticks_curr_cpu_in_eqs()) { + if (!rcu_is_watching_curr_cpu()) { if (!in_nmi()) rcu_task_enter(); @@ -281,7 +281,7 @@ void noinstr ct_nmi_enter(void) // ... but is watching here. instrumentation_begin(); - // instrumentation for the noinstr rcu_dynticks_curr_cpu_in_eqs() + // instrumentation for the noinstr rcu_is_watching_curr_cpu() instrument_atomic_read(&ct->state, sizeof(ct->state)); // instrumentation for the noinstr ct_kernel_enter_state() instrument_atomic_write(&ct->state, sizeof(ct->state)); diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 4778c873f4ac..68eca7d3fdd1 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -601,7 +601,7 @@ void rcu_irq_exit_check_preempt(void) RCU_LOCKDEP_WARN(ct_nmi_nesting() != CT_NESTING_IRQ_NONIDLE, "Bad RCU nmi_nesting counter\n"); - RCU_LOCKDEP_WARN(rcu_dynticks_curr_cpu_in_eqs(), + RCU_LOCKDEP_WARN(!rcu_is_watching_curr_cpu(), "RCU in extended quiescent state!"); } #endif /* #ifdef CONFIG_PROVE_RCU */ @@ -641,7 +641,7 @@ void __rcu_irq_enter_check_tick(void) if (in_nmi()) return; - RCU_LOCKDEP_WARN(rcu_dynticks_curr_cpu_in_eqs(), + RCU_LOCKDEP_WARN(!rcu_is_watching_curr_cpu(), "Illegal rcu_irq_enter_check_tick() from extended quiescent state"); if (!tick_nohz_full_cpu(rdp->cpu) || @@ -723,7 +723,7 @@ notrace bool rcu_is_watching(void) bool ret; preempt_disable_notrace(); - ret = !rcu_dynticks_curr_cpu_in_eqs(); + ret = rcu_is_watching_curr_cpu(); preempt_enable_notrace(); return ret; } -- cgit v1.2.3 From 654b578e4db04a68f9bfd0532edaeb7a87abc945 Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Tue, 16 Apr 2024 12:25:31 +0200 Subject: rcu: Rename rcu_dynticks_eqs_online() into rcu_watching_online() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, reflect that change in the related helpers. Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- kernel/rcu/tree.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 68eca7d3fdd1..47ec9f2daaed 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -283,16 +283,16 @@ void rcu_softirq_qs(void) } /* - * Reset the current CPU's ->dynticks counter to indicate that the + * Reset the current CPU's RCU_WATCHING counter to indicate that the * newly onlined CPU is no longer in an extended quiescent state. * This will either leave the counter unchanged, or increment it * to the next non-quiescent value. * * The non-atomic test/increment sequence works because the upper bits - * of the ->dynticks counter are manipulated only by the corresponding CPU, + * of the ->state variable are manipulated only by the corresponding CPU, * or when the corresponding CPU is offline. */ -static void rcu_dynticks_eqs_online(void) +static void rcu_watching_online(void) { if (ct_rcu_watching() & CT_RCU_WATCHING) return; @@ -5058,7 +5058,7 @@ void rcutree_report_cpu_starting(unsigned int cpu) rnp = rdp->mynode; mask = rdp->grpmask; arch_spin_lock(&rcu_state.ofl_lock); - rcu_dynticks_eqs_online(); + rcu_watching_online(); raw_spin_lock(&rcu_state.barrier_lock); raw_spin_lock_rcu_node(rnp); WRITE_ONCE(rnp->qsmaskinitnext, rnp->qsmaskinitnext | mask); -- cgit v1.2.3 From 9629936d06d05540d4dc83cee1167422a477a818 Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Tue, 16 Apr 2024 17:18:18 +0200 Subject: rcu: Rename rcu_dynticks_in_eqs() into rcu_watching_snap_in_eqs() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, reflect that change in the related helpers. While at it, update a comment that still refers to rcu_dynticks_snap(), which was removed by commit: 7be2e6323b9b ("rcu: Remove full memory barrier on RCU stall printout") Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- kernel/rcu/tree.c | 8 ++++---- kernel/rcu/tree_exp.h | 2 +- kernel/rcu/tree_stall.h | 2 +- 3 files changed, 6 insertions(+), 6 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 47ec9f2daaed..73516d76b70a 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -300,10 +300,10 @@ static void rcu_watching_online(void) } /* - * Return true if the snapshot returned from rcu_dynticks_snap() + * Return true if the snapshot returned from ct_rcu_watching() * indicates that RCU is in an extended quiescent state. */ -static bool rcu_dynticks_in_eqs(int snap) +static bool rcu_watching_snap_in_eqs(int snap) { return !(snap & CT_RCU_WATCHING); } @@ -783,7 +783,7 @@ static int dyntick_save_progress_counter(struct rcu_data *rdp) * updater's accesses is enforced by the below acquire semantic. */ rdp->dynticks_snap = ct_rcu_watching_cpu_acquire(rdp->cpu); - if (rcu_dynticks_in_eqs(rdp->dynticks_snap)) { + if (rcu_watching_snap_in_eqs(rdp->dynticks_snap)) { trace_rcu_fqs(rcu_state.name, rdp->gp_seq, rdp->cpu, TPS("dti")); rcu_gpnum_ovf(rdp->mynode, rdp); return 1; @@ -4805,7 +4805,7 @@ rcu_boot_init_percpu_data(int cpu) rdp->grpmask = leaf_node_cpu_bit(rdp->mynode, cpu); INIT_WORK(&rdp->strict_work, strict_work_handler); WARN_ON_ONCE(ct->nesting != 1); - WARN_ON_ONCE(rcu_dynticks_in_eqs(ct_rcu_watching_cpu(cpu))); + WARN_ON_ONCE(rcu_watching_snap_in_eqs(ct_rcu_watching_cpu(cpu))); rdp->barrier_seq_snap = rcu_state.barrier_sequence; rdp->rcu_ofl_gp_seq = rcu_state.gp_seq; rdp->rcu_ofl_gp_state = RCU_GP_CLEANED; diff --git a/kernel/rcu/tree_exp.h b/kernel/rcu/tree_exp.h index daa87fec703f..137559c84309 100644 --- a/kernel/rcu/tree_exp.h +++ b/kernel/rcu/tree_exp.h @@ -377,7 +377,7 @@ static void __sync_rcu_exp_select_node_cpus(struct rcu_exp_work *rewp) * below acquire semantic. */ snap = ct_rcu_watching_cpu_acquire(cpu); - if (rcu_dynticks_in_eqs(snap)) + if (rcu_watching_snap_in_eqs(snap)) mask_ofl_test |= mask; else rdp->exp_dynticks_snap = snap; diff --git a/kernel/rcu/tree_stall.h b/kernel/rcu/tree_stall.h index ec49f0155bec..e933c6a58d5f 100644 --- a/kernel/rcu/tree_stall.h +++ b/kernel/rcu/tree_stall.h @@ -501,7 +501,7 @@ static void print_cpu_stall_info(int cpu) } delta = rcu_seq_ctr(rdp->mynode->gp_seq - rdp->rcu_iw_gp_seq); falsepositive = rcu_is_gp_kthread_starving(NULL) && - rcu_dynticks_in_eqs(ct_rcu_watching_cpu(cpu)); + rcu_watching_snap_in_eqs(ct_rcu_watching_cpu(cpu)); rcuc_starved = rcu_is_rcuc_kthread_starving(rdp, &j); if (rcuc_starved) // Print signed value, as negative values indicate a probable bug. -- cgit v1.2.3 From 3116a32eb40464de61828787c21f0629a1c26cfa Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Tue, 16 Apr 2024 17:21:25 +0200 Subject: rcu: Rename rcu_dynticks_in_eqs_since() into rcu_watching_snap_stopped_since() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, the dynticks prefix can go. While at it, this helper is only meant to be called after failing an earlier call to rcu_watching_snap_in_eqs(), document this in the comments and add a WARN_ON_ONCE() for good measure. Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- .../Memory-Ordering/Tree-RCU-Memory-Ordering.rst | 2 +- kernel/rcu/tree.c | 23 ++++++++++++++++------ kernel/rcu/tree_exp.h | 2 +- 3 files changed, 19 insertions(+), 8 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/Documentation/RCU/Design/Memory-Ordering/Tree-RCU-Memory-Ordering.rst b/Documentation/RCU/Design/Memory-Ordering/Tree-RCU-Memory-Ordering.rst index 2d7036ad7476..7163d0def34e 100644 --- a/Documentation/RCU/Design/Memory-Ordering/Tree-RCU-Memory-Ordering.rst +++ b/Documentation/RCU/Design/Memory-Ordering/Tree-RCU-Memory-Ordering.rst @@ -150,7 +150,7 @@ This case is handled by calls to the strongly ordered is invoked within ``rcu_dynticks_eqs_enter()`` at idle-entry time and within ``rcu_dynticks_eqs_exit()`` at idle-exit time. The grace-period kthread invokes first ``ct_rcu_watching_cpu_acquire()`` -(preceded by a full memory barrier) and ``rcu_dynticks_in_eqs_since()`` +(preceded by a full memory barrier) and ``rcu_watching_snap_stopped_since()`` (both of which rely on acquire semantics) to detect idle CPUs. +-----------------------------------------------------------------------+ diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 73516d76b70a..93570cb5e99e 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -308,12 +308,20 @@ static bool rcu_watching_snap_in_eqs(int snap) return !(snap & CT_RCU_WATCHING); } -/* - * Return true if the CPU corresponding to the specified rcu_data - * structure has spent some time in an extended quiescent state since - * rcu_dynticks_snap() returned the specified snapshot. +/** + * rcu_watching_snap_stopped_since() - Has RCU stopped watching a given CPU + * since the specified @snap? + * + * @rdp: The rcu_data corresponding to the CPU for which to check EQS. + * @snap: rcu_watching snapshot taken when the CPU wasn't in an EQS. + * + * Returns true if the CPU corresponding to @rdp has spent some time in an + * extended quiescent state since @snap. Note that this doesn't check if it + * /still/ is in an EQS, just that it went through one since @snap. + * + * This is meant to be used in a loop waiting for a CPU to go through an EQS. */ -static bool rcu_dynticks_in_eqs_since(struct rcu_data *rdp, int snap) +static bool rcu_watching_snap_stopped_since(struct rcu_data *rdp, int snap) { /* * The first failing snapshot is already ordered against the accesses @@ -323,6 +331,9 @@ static bool rcu_dynticks_in_eqs_since(struct rcu_data *rdp, int snap) * performed by the remote CPU prior to entering idle and therefore can * rely solely on acquire semantics. */ + if (WARN_ON_ONCE(rcu_watching_snap_in_eqs(snap))) + return true; + return snap != ct_rcu_watching_cpu_acquire(rdp->cpu); } @@ -815,7 +826,7 @@ static int rcu_implicit_dynticks_qs(struct rcu_data *rdp) * read-side critical section that started before the beginning * of the current RCU grace period. */ - if (rcu_dynticks_in_eqs_since(rdp, rdp->dynticks_snap)) { + if (rcu_watching_snap_stopped_since(rdp, rdp->dynticks_snap)) { trace_rcu_fqs(rcu_state.name, rdp->gp_seq, rdp->cpu, TPS("dti")); rcu_gpnum_ovf(rnp, rdp); return 1; diff --git a/kernel/rcu/tree_exp.h b/kernel/rcu/tree_exp.h index 137559c84309..48ad8b868d83 100644 --- a/kernel/rcu/tree_exp.h +++ b/kernel/rcu/tree_exp.h @@ -400,7 +400,7 @@ static void __sync_rcu_exp_select_node_cpus(struct rcu_exp_work *rewp) unsigned long mask = rdp->grpmask; retry_ipi: - if (rcu_dynticks_in_eqs_since(rdp, rdp->exp_dynticks_snap)) { + if (rcu_watching_snap_stopped_since(rdp, rdp->exp_dynticks_snap)) { mask_ofl_test |= mask; continue; } -- cgit v1.2.3 From fc1096ab1f318fd7cabb842279e88978bb31768e Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Mon, 29 Apr 2024 13:11:34 +0200 Subject: rcu: Rename rcu_dynticks_zero_in_eqs() into rcu_watching_zero_in_eqs() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, reflect that change in the related helpers. Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- kernel/rcu/rcu.h | 4 ++-- kernel/rcu/tasks.h | 2 +- kernel/rcu/tree.c | 2 +- 3 files changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/kernel/rcu/rcu.h b/kernel/rcu/rcu.h index 38238e595a61..5564402af4cb 100644 --- a/kernel/rcu/rcu.h +++ b/kernel/rcu/rcu.h @@ -606,7 +606,7 @@ void srcutorture_get_gp_data(struct srcu_struct *sp, int *flags, #endif #ifdef CONFIG_TINY_RCU -static inline bool rcu_dynticks_zero_in_eqs(int cpu, int *vp) { return false; } +static inline bool rcu_watching_zero_in_eqs(int cpu, int *vp) { return false; } static inline unsigned long rcu_get_gp_seq(void) { return 0; } static inline unsigned long rcu_exp_batches_completed(void) { return 0; } static inline unsigned long @@ -619,7 +619,7 @@ static inline void rcu_fwd_progress_check(unsigned long j) { } static inline void rcu_gp_slow_register(atomic_t *rgssp) { } static inline void rcu_gp_slow_unregister(atomic_t *rgssp) { } #else /* #ifdef CONFIG_TINY_RCU */ -bool rcu_dynticks_zero_in_eqs(int cpu, int *vp); +bool rcu_watching_zero_in_eqs(int cpu, int *vp); unsigned long rcu_get_gp_seq(void); unsigned long rcu_exp_batches_completed(void); unsigned long srcu_batches_completed(struct srcu_struct *sp); diff --git a/kernel/rcu/tasks.h b/kernel/rcu/tasks.h index ba3440a45b6d..2484c1a8e051 100644 --- a/kernel/rcu/tasks.h +++ b/kernel/rcu/tasks.h @@ -1613,7 +1613,7 @@ static int trc_inspect_reader(struct task_struct *t, void *bhp_in) // However, we cannot safely change its state. n_heavy_reader_attempts++; // Check for "running" idle tasks on offline CPUs. - if (!rcu_dynticks_zero_in_eqs(cpu, &t->trc_reader_nesting)) + if (!rcu_watching_zero_in_eqs(cpu, &t->trc_reader_nesting)) return -EINVAL; // No quiescent state, do it the hard way. n_heavy_reader_updates++; nesting = 0; diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 93570cb5e99e..ac3577ac10bd 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -341,7 +341,7 @@ static bool rcu_watching_snap_stopped_since(struct rcu_data *rdp, int snap) * Return true if the referenced integer is zero while the specified * CPU remains within a single extended quiescent state. */ -bool rcu_dynticks_zero_in_eqs(int cpu, int *vp) +bool rcu_watching_zero_in_eqs(int cpu, int *vp) { int snap; -- cgit v1.2.3 From 2dba2230f9e26ad492fcd70e9ef2edc28313ea50 Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Mon, 29 Apr 2024 14:05:29 +0200 Subject: rcu: Rename struct rcu_data .dynticks_snap into .watching_snap The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, and the snapshot helpers are now prefix by "rcu_watching". Reflect that change into the storage variables for these snapshots. Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- Documentation/RCU/Design/Data-Structures/Data-Structures.rst | 4 ++-- kernel/rcu/tree.c | 6 +++--- kernel/rcu/tree.h | 2 +- 3 files changed, 6 insertions(+), 6 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/Documentation/RCU/Design/Data-Structures/Data-Structures.rst b/Documentation/RCU/Design/Data-Structures/Data-Structures.rst index 28d9f9e1783f..04e16775c752 100644 --- a/Documentation/RCU/Design/Data-Structures/Data-Structures.rst +++ b/Documentation/RCU/Design/Data-Structures/Data-Structures.rst @@ -921,10 +921,10 @@ This portion of the ``rcu_data`` structure is declared as follows: :: - 1 int dynticks_snap; + 1 int watching_snap; 2 unsigned long dynticks_fqs; -The ``->dynticks_snap`` field is used to take a snapshot of the +The ``->watching_snap`` field is used to take a snapshot of the corresponding CPU's dyntick-idle state when forcing quiescent states, and is therefore accessed from other CPUs. Finally, the ``->dynticks_fqs`` field is used to count the number of times this CPU diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index ac3577ac10bd..a969d8ddc58e 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -793,8 +793,8 @@ static int dyntick_save_progress_counter(struct rcu_data *rdp) * Ordering between remote CPU's pre idle accesses and post grace period * updater's accesses is enforced by the below acquire semantic. */ - rdp->dynticks_snap = ct_rcu_watching_cpu_acquire(rdp->cpu); - if (rcu_watching_snap_in_eqs(rdp->dynticks_snap)) { + rdp->watching_snap = ct_rcu_watching_cpu_acquire(rdp->cpu); + if (rcu_watching_snap_in_eqs(rdp->watching_snap)) { trace_rcu_fqs(rcu_state.name, rdp->gp_seq, rdp->cpu, TPS("dti")); rcu_gpnum_ovf(rdp->mynode, rdp); return 1; @@ -826,7 +826,7 @@ static int rcu_implicit_dynticks_qs(struct rcu_data *rdp) * read-side critical section that started before the beginning * of the current RCU grace period. */ - if (rcu_watching_snap_stopped_since(rdp, rdp->dynticks_snap)) { + if (rcu_watching_snap_stopped_since(rdp, rdp->watching_snap)) { trace_rcu_fqs(rcu_state.name, rdp->gp_seq, rdp->cpu, TPS("dti")); rcu_gpnum_ovf(rnp, rdp); return 1; diff --git a/kernel/rcu/tree.h b/kernel/rcu/tree.h index fcf2b4aa3441..f5361a7d7269 100644 --- a/kernel/rcu/tree.h +++ b/kernel/rcu/tree.h @@ -206,7 +206,7 @@ struct rcu_data { long blimit; /* Upper limit on a processed batch */ /* 3) dynticks interface. */ - int dynticks_snap; /* Per-GP tracking for dynticks. */ + int watching_snap; /* Per-GP tracking for dynticks. */ bool rcu_need_heavy_qs; /* GP old, so heavy quiescent state! */ bool rcu_urgent_qs; /* GP old need light quiescent state. */ bool rcu_forced_tick; /* Forced tick to provide QS. */ -- cgit v1.2.3 From 49f82c64fdc66447d839e77d2bd1c5b5b797590c Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Mon, 29 Apr 2024 16:38:19 +0200 Subject: rcu: Rename dyntick_save_progress_counter() into rcu_watching_snap_save() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, and the 'dynticks' prefix can be dropped without losing any meaning. Suggested-by: Frederic Weisbecker Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- Documentation/RCU/Design/Memory-Ordering/TreeRCU-dyntick.svg | 2 +- Documentation/RCU/Design/Memory-Ordering/TreeRCU-gp-fqs.svg | 2 +- Documentation/RCU/Design/Memory-Ordering/TreeRCU-gp.svg | 2 +- Documentation/RCU/Design/Memory-Ordering/TreeRCU-hotplug.svg | 2 +- kernel/rcu/tree.c | 8 ++++---- 5 files changed, 8 insertions(+), 8 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/Documentation/RCU/Design/Memory-Ordering/TreeRCU-dyntick.svg b/Documentation/RCU/Design/Memory-Ordering/TreeRCU-dyntick.svg index 423df00c4df9..13956baf748a 100644 --- a/Documentation/RCU/Design/Memory-Ordering/TreeRCU-dyntick.svg +++ b/Documentation/RCU/Design/Memory-Ordering/TreeRCU-dyntick.svg @@ -528,7 +528,7 @@ font-style="normal" y="-8652.5312" x="2466.7822" - xml:space="preserve">dyntick_save_progress_counter() + xml:space="preserve">rcu_watching_snap_save() dyntick_save_progress_counter() + xml:space="preserve">rcu_watching_snap_save() dyntick_save_progress_counter() + xml:space="preserve">rcu_watching_snap_save() dyntick_save_progress_counter() + xml:space="preserve">rcu_watching_snap_save() Date: Mon, 29 Apr 2024 16:41:37 +0200 Subject: rcu: Rename rcu_implicit_dynticks_qs() into rcu_watching_snap_recheck() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, drop the dyntick reference and update the name of this helper to express that it rechecks rdp->watching_snap after an earlier rcu_watching_snap_save(). Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- Documentation/RCU/Design/Memory-Ordering/TreeRCU-dyntick.svg | 2 +- Documentation/RCU/Design/Memory-Ordering/TreeRCU-gp-fqs.svg | 2 +- Documentation/RCU/Design/Memory-Ordering/TreeRCU-gp.svg | 2 +- Documentation/RCU/Design/Memory-Ordering/TreeRCU-hotplug.svg | 2 +- kernel/rcu/tree.c | 4 ++-- 5 files changed, 6 insertions(+), 6 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/Documentation/RCU/Design/Memory-Ordering/TreeRCU-dyntick.svg b/Documentation/RCU/Design/Memory-Ordering/TreeRCU-dyntick.svg index 13956baf748a..ab9707f04e66 100644 --- a/Documentation/RCU/Design/Memory-Ordering/TreeRCU-dyntick.svg +++ b/Documentation/RCU/Design/Memory-Ordering/TreeRCU-dyntick.svg @@ -537,7 +537,7 @@ font-style="normal" y="-8368.1475" x="2463.3262" - xml:space="preserve">rcu_implicit_dynticks_qs() + xml:space="preserve">rcu_watching_snap_recheck() rcu_implicit_dynticks_qs() + xml:space="preserve">rcu_watching_snap_recheck() rcu_implicit_dynticks_qs() + xml:space="preserve">rcu_watching_snap_recheck() rcu_implicit_dynticks_qs() + xml:space="preserve">rcu_watching_snap_recheck() Date: Mon, 29 Apr 2024 16:43:36 +0200 Subject: rcu: Rename rcu_momentary_dyntick_idle() into rcu_momentary_eqs() The context_tracking.state RCU_DYNTICKS subvariable has been renamed to RCU_WATCHING, replace "dyntick_idle" into "eqs" to drop the dyntick reference. Signed-off-by: Valentin Schneider Reviewed-by: Frederic Weisbecker Signed-off-by: Neeraj Upadhyay --- include/linux/rcutiny.h | 2 +- include/linux/rcutree.h | 2 +- kernel/rcu/rcutorture.c | 4 ++-- kernel/rcu/tree.c | 4 ++-- kernel/rcu/tree_nocb.h | 2 +- kernel/rcu/tree_plugin.h | 6 +++--- kernel/stop_machine.c | 2 +- kernel/trace/trace_osnoise.c | 4 ++-- 8 files changed, 13 insertions(+), 13 deletions(-) (limited to 'kernel/rcu/tree.c') diff --git a/include/linux/rcutiny.h b/include/linux/rcutiny.h index d9ac7b136aea..cf2b5a188f78 100644 --- a/include/linux/rcutiny.h +++ b/include/linux/rcutiny.h @@ -158,7 +158,7 @@ void rcu_scheduler_starting(void); static inline void rcu_end_inkernel_boot(void) { } static inline bool rcu_inkernel_boot_has_ended(void) { return true; } static inline bool rcu_is_watching(void) { return true; } -static inline void rcu_momentary_dyntick_idle(void) { } +static inline void rcu_momentary_eqs(void) { } static inline void kfree_rcu_scheduler_running(void) { } static inline bool rcu_gp_might_be_stalled(void) { return false; } diff --git a/include/linux/rcutree.h b/include/linux/rcutree.h index 254244202ea9..7dbde2b6f714 100644 --- a/include/linux/rcutree.h +++ b/include/linux/rcutree.h @@ -37,7 +37,7 @@ void synchronize_rcu_expedited(void); void kvfree_call_rcu(struct rcu_head *head, void *ptr); void rcu_barrier(void); -void rcu_momentary_dyntick_idle(void); +void rcu_momentary_eqs(void); void kfree_rcu_scheduler_running(void); bool rcu_gp_might_be_stalled(void); diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c index 08bf7c669dd3..ef757c3e84a2 100644 --- a/kernel/rcu/rcutorture.c +++ b/kernel/rcu/rcutorture.c @@ -2680,7 +2680,7 @@ static unsigned long rcu_torture_fwd_prog_cbfree(struct rcu_fwd *rfp) rcu_torture_fwd_prog_cond_resched(freed); if (tick_nohz_full_enabled()) { local_irq_save(flags); - rcu_momentary_dyntick_idle(); + rcu_momentary_eqs(); local_irq_restore(flags); } } @@ -2830,7 +2830,7 @@ static void rcu_torture_fwd_prog_cr(struct rcu_fwd *rfp) rcu_torture_fwd_prog_cond_resched(n_launders + n_max_cbs); if (tick_nohz_full_enabled()) { local_irq_save(flags); - rcu_momentary_dyntick_idle(); + rcu_momentary_eqs(); local_irq_restore(flags); } } diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index b53b354417ae..e0f6fc272906 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -367,7 +367,7 @@ bool rcu_watching_zero_in_eqs(int cpu, int *vp) * * The caller must have disabled interrupts and must not be idle. */ -notrace void rcu_momentary_dyntick_idle(void) +notrace void rcu_momentary_eqs(void) { int seq; @@ -377,7 +377,7 @@ notrace void rcu_momentary_dyntick_idle(void) WARN_ON_ONCE(!(seq & CT_RCU_WATCHING)); rcu_preempt_deferred_qs(current); } -EXPORT_SYMBOL_GPL(rcu_momentary_dyntick_idle); +EXPORT_SYMBOL_GPL(rcu_momentary_eqs); /** * rcu_is_cpu_rrupt_from_idle - see if 'interrupted' from idle diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h index 3ce30841119a..e629646ac6f5 100644 --- a/kernel/rcu/tree_nocb.h +++ b/kernel/rcu/tree_nocb.h @@ -917,7 +917,7 @@ static void nocb_cb_wait(struct rcu_data *rdp) WARN_ON_ONCE(!rcu_rdp_is_offloaded(rdp)); local_irq_save(flags); - rcu_momentary_dyntick_idle(); + rcu_momentary_eqs(); local_irq_restore(flags); /* * Disable BH to provide the expected environment. Also, when diff --git a/kernel/rcu/tree_plugin.h b/kernel/rcu/tree_plugin.h index c569da65b421..354fbf2e411c 100644 --- a/kernel/rcu/tree_plugin.h +++ b/kernel/rcu/tree_plugin.h @@ -869,7 +869,7 @@ static void rcu_qs(void) /* * Register an urgently needed quiescent state. If there is an - * emergency, invoke rcu_momentary_dyntick_idle() to do a heavy-weight + * emergency, invoke rcu_momentary_eqs() to do a heavy-weight * dyntick-idle quiescent state visible to other CPUs, which will in * some cases serve for expedited as well as normal grace periods. * Either way, register a lightweight quiescent state. @@ -889,7 +889,7 @@ void rcu_all_qs(void) this_cpu_write(rcu_data.rcu_urgent_qs, false); if (unlikely(raw_cpu_read(rcu_data.rcu_need_heavy_qs))) { local_irq_save(flags); - rcu_momentary_dyntick_idle(); + rcu_momentary_eqs(); local_irq_restore(flags); } rcu_qs(); @@ -909,7 +909,7 @@ void rcu_note_context_switch(bool preempt) goto out; this_cpu_write(rcu_data.rcu_urgent_qs, false); if (unlikely(raw_cpu_read(rcu_data.rcu_need_heavy_qs))) - rcu_momentary_dyntick_idle(); + rcu_momentary_eqs(); out: rcu_tasks_qs(current, preempt); trace_rcu_utilization(TPS("End context switch")); diff --git a/kernel/stop_machine.c b/kernel/stop_machine.c index cedb17ba158a..da821ce258ea 100644 --- a/kernel/stop_machine.c +++ b/kernel/stop_machine.c @@ -251,7 +251,7 @@ static int multi_cpu_stop(void *data) */ touch_nmi_watchdog(); } - rcu_momentary_dyntick_idle(); + rcu_momentary_eqs(); } while (curstate != MULTI_STOP_EXIT); local_irq_restore(flags); diff --git a/kernel/trace/trace_osnoise.c b/kernel/trace/trace_osnoise.c index 66a871553d4a..ad86b1ba2e43 100644 --- a/kernel/trace/trace_osnoise.c +++ b/kernel/trace/trace_osnoise.c @@ -1535,7 +1535,7 @@ static int run_osnoise(void) * This will eventually cause unwarranted noise as PREEMPT_RCU * will force preemption as the means of ending the current * grace period. We avoid this problem by calling - * rcu_momentary_dyntick_idle(), which performs a zero duration + * rcu_momentary_eqs(), which performs a zero duration * EQS allowing PREEMPT_RCU to end the current grace period. * This call shouldn't be wrapped inside an RCU critical * section. @@ -1547,7 +1547,7 @@ static int run_osnoise(void) if (!disable_irq) local_irq_disable(); - rcu_momentary_dyntick_idle(); + rcu_momentary_eqs(); if (!disable_irq) local_irq_enable(); -- cgit v1.2.3