Merge tag 'rcu.2026.03.31a' of git://git.kernel.org/pub/scm/linux/kernel/git/rcu/linux

Pull RCU updates from Joel Fernandes:
 "NOCB CPU management:

   - Consolidate rcu_nocb_cpu_offload() and rcu_nocb_cpu_deoffload() to
     reduce code duplication

   - Extract nocb_bypass_needs_flush() helper to reduce duplication in
     NOCB bypass path

  rcutorture/torture infrastructure:

   - Add NOCB01 config for RCU_LAZY torture testing

   - Add NOCB02 config for NOCB poll mode testing

   - Add TRIVIAL-PREEMPT config for textbook-style preemptible RCU
     torture

   - Test call_srcu() with preemption both disabled and enabled

   - Remove kvm-check-branches.sh in favor of kvm-series.sh

   - Make hangs more visible in torture.sh output

   - Add informative message for tests without a recheck file

   - Fix numeric test comparison in srcu_lockdep.sh

   - Use torture_shutdown_init() in refscale and rcuscale instead of
     open-coded shutdown functions

   - Fix modulo-zero error in torture_hrtimeout_ns().

  SRCU:

   - Fix SRCU read flavor macro comments

   - Fix s/they disables/they disable/ typo in srcu_read_unlock_fast()

  RCU Tasks:

   - Document that RCU Tasks Trace grace periods now imply RCU grace
     periods

   - Remove unnecessary smp_store_release() in cblist_init_generic()"

* tag 'rcu.2026.03.31a' of git://git.kernel.org/pub/scm/linux/kernel/git/rcu/linux:
  rcutorture: Test call_srcu() with preemption disabled and not
  rcu: Add BOOTPARAM_RCU_STALL_PANIC Kconfig option
  torture: Avoid modulo-zero error in torture_hrtimeout_ns()
  rcu/nocb: Extract nocb_bypass_needs_flush() to reduce duplication
  rcu/nocb: Consolidate rcu_nocb_cpu_offload/deoffload functions
  rcu-tasks: Remove unnecessary smp_store_release() in cblist_init_generic()
  rcutorture: Add NOCB02 config for nocb poll mode testing
  rcutorture: Add NOCB01 config for RCU_LAZY torture testing
  rcu-tasks: Document that RCU Tasks Trace grace periods now imply RCU grace periods
  srcu: Fix s/they disables/they disable/ typo in srcu_read_unlock_fast()
  srcu: Fix SRCU read flavor macro comments
  rcuscale: Ditch rcu_scale_shutdown in favor of torture_shutdown_init()
  refscale: Ditch ref_scale_shutdown in favor of torture_shutdown_init()
  rcutorture: Fix numeric "test" comparison in srcu_lockdep.sh
  torture: Print informative message for test without recheck file
  torture: Make hangs more visible in torture.sh output
  kvm-check-branches.sh: Remove in favor of kvm-series.sh
  rcutorture: Add a textbook-style trivial preemptible RCU

Author: torvalds

Documentation/RCU/Design/Requirements/Requirements.rst

@@ -2787,6 +2787,13 @@ which avoids the read-side memory barriers, at least for architectures
 that apply noinstr to kernel entry/exit code (or that build with
 ``CONFIG_TASKS_TRACE_RCU_NO_MB=y``.
 
+Now that the implementation is based on SRCU-fast, a call
+to synchronize_rcu_tasks_trace() implies at least one call to
+synchronize_rcu(), that is, every Tasks Trace RCU grace period contains
+at least one plain vanilla RCU grace period.  Should there ever
+be a synchronize_rcu_tasks_trace_expedited(), this guarantee would
+*not* necessarily apply to this hypothetical API member.
+
 The tasks-trace-RCU API is also reasonably compact,
 consisting of rcu_read_lock_trace(), rcu_read_unlock_trace(),
 rcu_read_lock_trace_held(), call_rcu_tasks_trace(),

include/linux/rcupdate.h

@@ -208,12 +208,9 @@ static inline void exit_tasks_rcu_finish(void) { }
 /**
  * rcu_trace_implies_rcu_gp - does an RCU Tasks Trace grace period imply an RCU grace period?
  *
- * As an accident of implementation, an RCU Tasks Trace grace period also
- * acts as an RCU grace period.  However, this could change at any time.
- * Code relying on this accident must call this function to verify that
- * this accident is still happening.
- *
- * You have been warned!
+ * Now that RCU Tasks Trace is implemented in terms of SRCU-fast, a
+ * call to synchronize_rcu_tasks_trace() is guaranteed to imply at least
+ * one call to synchronize_rcu().
  */
 static inline bool rcu_trace_implies_rcu_gp(void) { return true; }
 

include/linux/sched.h

@@ -949,6 +949,10 @@ struct task_struct {
 	struct srcu_ctr __percpu	*trc_reader_scp;
 #endif /* #ifdef CONFIG_TASKS_TRACE_RCU */
 
+#ifdef CONFIG_TRIVIAL_PREEMPT_RCU
+	int				rcu_trivial_preempt_nesting;
+#endif /* #ifdef CONFIG_TRIVIAL_PREEMPT_RCU */
+
 	struct sched_info		sched_info;
 
 	struct list_head		tasks;

include/linux/srcu.h

@@ -69,8 +69,8 @@ int init_srcu_struct_fast_updown(struct srcu_struct *ssp);
 #define SRCU_READ_FLAVOR_NORMAL		0x1		// srcu_read_lock().
 #define SRCU_READ_FLAVOR_NMI		0x2		// srcu_read_lock_nmisafe().
 //					0x4		// SRCU-lite is no longer with us.
-#define SRCU_READ_FLAVOR_FAST		0x4		// srcu_read_lock_fast().
-#define SRCU_READ_FLAVOR_FAST_UPDOWN	0x8		// srcu_read_lock_fast().
+#define SRCU_READ_FLAVOR_FAST		0x4		// srcu_read_lock_fast(), also NMI-safe.
+#define SRCU_READ_FLAVOR_FAST_UPDOWN	0x8		// srcu_read_lock_fast_updown().
 #define SRCU_READ_FLAVOR_ALL		(SRCU_READ_FLAVOR_NORMAL | SRCU_READ_FLAVOR_NMI | \
 					 SRCU_READ_FLAVOR_FAST | SRCU_READ_FLAVOR_FAST_UPDOWN)
 						// All of the above.

include/linux/srcutree.h

@@ -260,7 +260,7 @@ static inline struct srcu_ctr __percpu *__srcu_ctr_to_ptr(struct srcu_struct *ss
  * srcu_read_unlock_fast().
  *
  * Note that both this_cpu_inc() and atomic_long_inc() are RCU read-side
- * critical sections either because they disables interrupts, because
+ * critical sections either because they disable interrupts, because
  * they are a single instruction, or because they are read-modify-write
  * atomic operations, depending on the whims of the architecture.
  * This matters because the SRCU-fast grace-period mechanism uses either

kernel/rcu/Kconfig.debug

@@ -228,4 +228,15 @@ config RCU_DYNTICKS_TORTURE
 
 	  This has no value for production and is only for testing.
 
+config TRIVIAL_PREEMPT_RCU
+	bool "Textbook trivial preemptible RCU in rcutorture"
+	depends on RCU_EXPERT && RCU_TORTURE_TEST
+	default n
+	help
+	  This option enables a textbook preemptible RCU that is
+	  implemented in rcutorture.  Its sole purpose is to validate
+	  code used in books, papers, and presentations.
+
+	  This has no value for production and is only for testing.
+
 endmenu # "RCU Debugging"

kernel/rcu/rcu.h

@@ -691,4 +691,8 @@ int rcu_stall_notifier_call_chain(unsigned long val, void *v);
 static inline int rcu_stall_notifier_call_chain(unsigned long val, void *v) { return NOTIFY_DONE; }
 #endif // #else // #if defined(CONFIG_RCU_STALL_COMMON) && defined(CONFIG_RCU_CPU_STALL_NOTIFIER)
 
+#ifdef CONFIG_TRIVIAL_PREEMPT_RCU
+void synchronize_rcu_trivial_preempt(void);
+#endif // #ifdef CONFIG_TRIVIAL_PREEMPT_RCU
+
 #endif /* __LINUX_RCU_H */

kernel/rcu/rcuscale.c

@@ -79,21 +79,15 @@ MODULE_AUTHOR("Paul E. McKenney <paulmck@linux.ibm.com>");
  * test-end checks, and the pair of calls through pointers.
  */
 
-#ifdef MODULE
-# define RCUSCALE_SHUTDOWN 0
-#else
-# define RCUSCALE_SHUTDOWN 1
-#endif
-
 torture_param(bool, gp_async, false, "Use asynchronous GP wait primitives");
 torture_param(int, gp_async_max, 1000, "Max # outstanding waits per writer");
 torture_param(bool, gp_exp, false, "Use expedited GP wait primitives");
 torture_param(int, holdoff, 10, "Holdoff time before test start (s)");
 torture_param(int, minruntime, 0, "Minimum run time (s)");
 torture_param(int, nreaders, -1, "Number of RCU reader threads");
 torture_param(int, nwriters, -1, "Number of RCU updater threads");
-torture_param(bool, shutdown, RCUSCALE_SHUTDOWN,
-	      "Shutdown at end of scalability tests.");
+torture_param(int, shutdown_secs, !IS_MODULE(CONFIG_RCU_SCALE_TEST) * 300,
+	      "Shutdown at end of scalability tests or at specified timeout (s).");
 torture_param(int, verbose, 1, "Enable verbose debugging printk()s");
 torture_param(int, writer_holdoff, 0, "Holdoff (us) between GPs, zero to disable");
 torture_param(int, writer_holdoff_jiffies, 0, "Holdoff (jiffies) between GPs, zero to disable");
@@ -123,7 +117,6 @@ static int nrealreaders;
 static int nrealwriters;
 static struct task_struct **writer_tasks;
 static struct task_struct **reader_tasks;
-static struct task_struct *shutdown_task;
 
 static u64 **writer_durations;
 static bool *writer_done;
@@ -132,7 +125,6 @@ static int *writer_n_durations;
 static atomic_t n_rcu_scale_reader_started;
 static atomic_t n_rcu_scale_writer_started;
 static atomic_t n_rcu_scale_writer_finished;
-static wait_queue_head_t shutdown_wq;
 static u64 t_rcu_scale_writer_started;
 static u64 t_rcu_scale_writer_finished;
 static unsigned long b_rcu_gp_test_started;
@@ -519,6 +511,8 @@ static void rcu_scale_async_cb(struct rcu_head *rhp)
 	rcu_scale_free(wmbp);
 }
 
+static void rcu_scale_cleanup(void);
+
 /*
  * RCU scale writer kthread.  Repeatedly does a grace period.
  */
@@ -622,9 +616,11 @@ rcu_scale_writer(void *arg)
 					b_rcu_gp_test_finished =
 						cur_ops->get_gp_seq();
 				}
-				if (shutdown) {
+				if (shutdown_secs) {
+					writer_tasks[me] = NULL;
 					smp_mb(); /* Assign before wake. */
-					wake_up(&shutdown_wq);
+					rcu_scale_cleanup();
+					kernel_power_off();
 				}
 			}
 		}
@@ -668,8 +664,8 @@ static void
 rcu_scale_print_module_parms(struct rcu_scale_ops *cur_ops, const char *tag)
 {
 	pr_alert("%s" SCALE_FLAG
-		 "--- %s: gp_async=%d gp_async_max=%d gp_exp=%d holdoff=%d minruntime=%d nreaders=%d nwriters=%d writer_holdoff=%d writer_holdoff_jiffies=%d verbose=%d shutdown=%d\n",
-		 scale_type, tag, gp_async, gp_async_max, gp_exp, holdoff, minruntime, nrealreaders, nrealwriters, writer_holdoff, writer_holdoff_jiffies, verbose, shutdown);
+		 "--- %s: gp_async=%d gp_async_max=%d gp_exp=%d holdoff=%d minruntime=%d nreaders=%d nwriters=%d writer_holdoff=%d writer_holdoff_jiffies=%d verbose=%d shutdown_secs=%d\n",
+		 scale_type, tag, gp_async, gp_async_max, gp_exp, holdoff, minruntime, nrealreaders, nrealwriters, writer_holdoff, writer_holdoff_jiffies, verbose, shutdown_secs);
 }
 
 /*
@@ -722,6 +718,8 @@ static void kfree_call_rcu(struct rcu_head *rh)
 	kfree(obj);
 }
 
+static void kfree_scale_cleanup(void);
+
 static int
 kfree_scale_thread(void *arg)
 {
@@ -791,9 +789,11 @@ kfree_scale_thread(void *arg)
 		       rcuscale_seq_diff(b_rcu_gp_test_finished, b_rcu_gp_test_started),
 		       PAGES_TO_MB(mem_begin - mem_during));
 
-		if (shutdown) {
+		if (shutdown_secs) {
+			kfree_reader_tasks[me] = NULL;
 			smp_mb(); /* Assign before wake. */
-			wake_up(&shutdown_wq);
+			kfree_scale_cleanup();
+			kernel_power_off();
 		}
 	}
 
@@ -820,22 +820,6 @@ kfree_scale_cleanup(void)
 	torture_cleanup_end();
 }
 
-/*
- * shutdown kthread.  Just waits to be awakened, then shuts down system.
- */
-static int
-kfree_scale_shutdown(void *arg)
-{
-	wait_event_idle(shutdown_wq,
-			atomic_read(&n_kfree_scale_thread_ended) >= kfree_nrealthreads);
-
-	smp_mb(); /* Wake before output. */
-
-	kfree_scale_cleanup();
-	kernel_power_off();
-	return -EINVAL;
-}
-
 // Used if doing RCU-kfree'ing via call_rcu().
 static unsigned long jiffies_at_lazy_cb;
 static struct rcu_head lazy_test1_rh;
@@ -895,13 +879,10 @@ kfree_scale_init(void)
 
 	kfree_nrealthreads = compute_real(kfree_nthreads);
 	/* Start up the kthreads. */
-	if (shutdown) {
-		init_waitqueue_head(&shutdown_wq);
-		firsterr = torture_create_kthread(kfree_scale_shutdown, NULL,
-						  shutdown_task);
+	if (shutdown_secs) {
+		firsterr = torture_shutdown_init(shutdown_secs, kfree_scale_cleanup);
 		if (torture_init_error(firsterr))
 			goto unwind;
-		schedule_timeout_uninterruptible(1);
 	}
 
 	pr_alert("kfree object size=%zu, kfree_by_call_rcu=%d\n",
@@ -1058,20 +1039,6 @@ rcu_scale_cleanup(void)
 	torture_cleanup_end();
 }
 
-/*
- * RCU scalability shutdown kthread.  Just waits to be awakened, then shuts
- * down system.
- */
-static int
-rcu_scale_shutdown(void *arg)
-{
-	wait_event_idle(shutdown_wq, atomic_read(&n_rcu_scale_writer_finished) >= nrealwriters);
-	smp_mb(); /* Wake before output. */
-	rcu_scale_cleanup();
-	kernel_power_off();
-	return -EINVAL;
-}
-
 static int __init
 rcu_scale_init(void)
 {
@@ -1121,13 +1088,10 @@ rcu_scale_init(void)
 
 	/* Start up the kthreads. */
 
-	if (shutdown) {
-		init_waitqueue_head(&shutdown_wq);
-		firsterr = torture_create_kthread(rcu_scale_shutdown, NULL,
-						  shutdown_task);
+	if (shutdown_secs) {
+		firsterr = torture_shutdown_init(shutdown_secs, rcu_scale_cleanup);
 		if (torture_init_error(firsterr))
 			goto unwind;
-		schedule_timeout_uninterruptible(1);
 	}
 	reader_tasks = kzalloc_objs(reader_tasks[0], nrealreaders);
 	if (reader_tasks == NULL) {
@@ -1201,7 +1165,7 @@ rcu_scale_init(void)
 unwind:
 	torture_init_end();
 	rcu_scale_cleanup();
-	if (shutdown) {
+	if (shutdown_secs) {
 		WARN_ON(!IS_MODULE(CONFIG_RCU_SCALE_TEST));
 		kernel_power_off();
 	}

kernel/rcu/rcutorture.c

@@ -842,7 +842,14 @@ static unsigned long srcu_torture_completed(void)
 
 static void srcu_torture_deferred_free(struct rcu_torture *rp)
 {
+	unsigned long flags;
+	bool lockit = jiffies & 0x1;
+
+	if (lockit)
+		raw_spin_lock_irqsave(&current->pi_lock, flags);
 	call_srcu(srcu_ctlp, &rp->rtort_rcu, rcu_torture_cb);
+	if (lockit)
+		raw_spin_unlock_irqrestore(&current->pi_lock, flags);
 }
 
 static void srcu_torture_synchronize(void)
@@ -1061,6 +1068,61 @@ static struct rcu_torture_ops trivial_ops = {
 	.name		= "trivial"
 };
 
+#ifdef CONFIG_TRIVIAL_PREEMPT_RCU
+
+/*
+ * Definitions for trivial CONFIG_PREEMPT=y torture testing.  This
+ * implementation does not work well with large numbers of tasks or with
+ * long-term preemption.  Either or both get you RCU CPU stall warnings.
+ */
+
+static void rcu_sync_torture_init_trivial_preempt(void)
+{
+	rcu_sync_torture_init();
+	if (WARN_ONCE(onoff_interval || shuffle_interval, "%s: Non-zero onoff_interval (%d) or shuffle_interval (%d) breaks trivial RCU, resetting to zero", __func__, onoff_interval, shuffle_interval)) {
+		onoff_interval = 0;
+		shuffle_interval = 0;
+	}
+}
+
+static int rcu_torture_read_lock_trivial_preempt(void)
+{
+	struct task_struct *t = current;
+
+	WRITE_ONCE(t->rcu_trivial_preempt_nesting, t->rcu_trivial_preempt_nesting + 1);
+	smp_mb();
+	return 0;
+}
+
+static void rcu_torture_read_unlock_trivial_preempt(int idx)
+{
+	struct task_struct *t = current;
+
+	smp_store_release(&t->rcu_trivial_preempt_nesting, t->rcu_trivial_preempt_nesting - 1);
+}
+
+static struct rcu_torture_ops trivial_preempt_ops = {
+	.ttype		= RCU_TRIVIAL_FLAVOR,
+	.init		= rcu_sync_torture_init_trivial_preempt,
+	.readlock	= rcu_torture_read_lock_trivial_preempt,
+	.read_delay	= rcu_read_delay,  // just reuse rcu's version.
+	.readunlock	= rcu_torture_read_unlock_trivial_preempt,
+	.readlock_held	= torture_readlock_not_held,
+	.get_gp_seq	= rcu_no_completed,
+	.sync		= synchronize_rcu_trivial_preempt,
+	.exp_sync	= synchronize_rcu_trivial_preempt,
+	.irq_capable	= 0, // In theory it should be, but let's keep it trivial.
+	.name		= "trivial-preempt"
+};
+
+#define TRIVIAL_PREEMPT_OPS &trivial_preempt_ops,
+
+#else // #ifdef CONFIG_TRIVIAL_PREEMPT_RCU
+
+#define TRIVIAL_PREEMPT_OPS
+
+#endif // #else // #ifdef CONFIG_TRIVIAL_PREEMPT_RCU
+
 #ifdef CONFIG_TASKS_RCU
 
 /*
@@ -4449,7 +4511,7 @@ rcu_torture_init(void)
 	static struct rcu_torture_ops *torture_ops[] = {
 		&rcu_ops, &rcu_busted_ops, &srcu_ops, &srcud_ops, &busted_srcud_ops,
 		TASKS_OPS TASKS_RUDE_OPS TASKS_TRACING_OPS
-		&trivial_ops,
+		&trivial_ops, TRIVIAL_PREEMPT_OPS
 	};
 
 	if (!torture_init_begin(torture_type, verbose))