Merge branches 'fixes.2024.12.14a', 'rcutorture.2024.12.14a', 'srcu.2024.12.14a' and 'torture-test.2024.12.14a' into rcu-merge.2024.12.14a

fixes.2024.12.14a: RCU fixes
rcutorture.2024.12.14a: Torture-test updates
srcu.2024.12.14a: SRCU updates
torture-test.2024.12.14a: Adding an extra test, fixes

Author: urezki

Documentation/admin-guide/kernel-parameters.txt

@@ -5362,7 +5362,42 @@
 
 	rcutorture.gp_cond= [KNL]
 			Use conditional/asynchronous update-side
-			primitives, if available.
+			normal-grace-period primitives, if available.
+
+	rcutorture.gp_cond_exp= [KNL]
+			Use conditional/asynchronous update-side
+			expedited-grace-period primitives, if available.
+
+	rcutorture.gp_cond_full= [KNL]
+			Use conditional/asynchronous update-side
+			normal-grace-period primitives that also take
+			concurrent expedited grace periods into account,
+			if available.
+
+	rcutorture.gp_cond_exp_full= [KNL]
+			Use conditional/asynchronous update-side
+			expedited-grace-period primitives that also take
+			concurrent normal grace periods into account,
+			if available.
+
+	rcutorture.gp_cond_wi= [KNL]
+			Nominal wait interval for normal conditional
+			grace periods (specified by rcutorture's
+			gp_cond and gp_cond_full module parameters),
+			in microseconds.  The actual wait interval will
+			be randomly selected to nanosecond granularity up
+			to this wait interval.	Defaults to 16 jiffies,
+			for example, 16,000 microseconds on a system
+			with HZ=1000.
+
+	rcutorture.gp_cond_wi_exp= [KNL]
+			Nominal wait interval for expedited conditional
+			grace periods (specified by rcutorture's
+			gp_cond_exp and gp_cond_exp_full module
+			parameters), in microseconds.  The actual wait
+			interval will be randomly selected to nanosecond
+			granularity up to this wait interval.  Defaults to
+			128 microseconds.
 
 	rcutorture.gp_exp= [KNL]
 			Use expedited update-side primitives, if available.
@@ -5371,6 +5406,43 @@
 			Use normal (non-expedited) asynchronous
 			update-side primitives, if available.
 
+	rcutorture.gp_poll= [KNL]
+			Use polled update-side normal-grace-period
+			primitives, if available.
+
+	rcutorture.gp_poll_exp= [KNL]
+			Use polled update-side expedited-grace-period
+			primitives, if available.
+
+	rcutorture.gp_poll_full= [KNL]
+			Use polled update-side normal-grace-period
+			primitives that also take concurrent expedited
+			grace periods into account, if available.
+
+	rcutorture.gp_poll_exp_full= [KNL]
+			Use polled update-side expedited-grace-period
+			primitives that also take concurrent normal
+			grace periods into account, if available.
+
+	rcutorture.gp_poll_wi= [KNL]
+			Nominal wait interval for normal conditional
+			grace periods (specified by rcutorture's
+			gp_poll and gp_poll_full module parameters),
+			in microseconds.  The actual wait interval will
+			be randomly selected to nanosecond granularity up
+			to this wait interval.	Defaults to 16 jiffies,
+			for example, 16,000 microseconds on a system
+			with HZ=1000.
+
+	rcutorture.gp_poll_wi_exp= [KNL]
+			Nominal wait interval for expedited conditional
+			grace periods (specified by rcutorture's
+			gp_poll_exp and gp_poll_exp_full module
+			parameters), in microseconds.  The actual wait
+			interval will be randomly selected to nanosecond
+			granularity up to this wait interval.  Defaults to
+			128 microseconds.
+
 	rcutorture.gp_sync= [KNL]
 			Use normal (non-expedited) synchronous
 			update-side primitives, if available.  If all
@@ -5424,6 +5496,22 @@
 			Set time (jiffies) between CPU-hotplug operations,
 			or zero to disable CPU-hotplug testing.
 
+	rcutorture.preempt_duration= [KNL]
+			Set duration (in milliseconds) of preemptions
+			by a high-priority FIFO real-time task.  Set to
+			zero (the default) to disable.	The CPUs to
+			preempt are selected randomly from the set that
+			are online at a given point in time.  Races with
+			CPUs going offline are ignored, with that attempt
+			at preemption skipped.
+
+	rcutorture.preempt_interval= [KNL]
+			Set interval (in milliseconds, defaulting to one
+			second) between preemptions by a high-priority
+			FIFO real-time task.  This delay is mediated
+			by an hrtimer and is further fuzzed to avoid
+			inadvertent synchronizations.
+
 	rcutorture.read_exit_burst= [KNL]
 			The number of times in a given read-then-exit
 			episode that a set of read-then-exit kthreads

include/linux/rcupdate_wait.h

@@ -65,4 +65,15 @@ static inline void cond_resched_rcu(void)
 #endif
 }
 
+// Has the current task blocked within its current RCU read-side
+// critical section?
+static inline bool has_rcu_reader_blocked(void)
+{
+#ifdef CONFIG_PREEMPT_RCU
+	return !list_empty(&current->rcu_node_entry);
+#else
+	return false;
+#endif
+}
+
 #endif /* _LINUX_SCHED_RCUPDATE_WAIT_H */

include/linux/srcu.h

@@ -43,6 +43,12 @@ int init_srcu_struct(struct srcu_struct *ssp);
 #define __SRCU_DEP_MAP_INIT(srcu_name)
 #endif /* #else #ifdef CONFIG_DEBUG_LOCK_ALLOC */
 
+/* Values for SRCU Tree srcu_data ->srcu_reader_flavor, but also used by rcutorture. */
+#define SRCU_READ_FLAVOR_NORMAL	0x1		// srcu_read_lock().
+#define SRCU_READ_FLAVOR_NMI	0x2		// srcu_read_lock_nmisafe().
+#define SRCU_READ_FLAVOR_LITE	0x4		// srcu_read_lock_lite().
+#define SRCU_READ_FLAVOR_ALL	0x7		// All of the above.
+
 #ifdef CONFIG_TINY_SRCU
 #include <linux/srcutiny.h>
 #elif defined(CONFIG_TREE_SRCU)
@@ -232,13 +238,14 @@ static inline int srcu_read_lock_held(const struct srcu_struct *ssp)
  * a mutex that is held elsewhere while calling synchronize_srcu() or
  * synchronize_srcu_expedited().
  *
- * The return value from srcu_read_lock() must be passed unaltered
- * to the matching srcu_read_unlock().  Note that srcu_read_lock() and
- * the matching srcu_read_unlock() must occur in the same context, for
- * example, it is illegal to invoke srcu_read_unlock() in an irq handler
- * if the matching srcu_read_lock() was invoked in process context.  Or,
- * for that matter to invoke srcu_read_unlock() from one task and the
- * matching srcu_read_lock() from another.
+ * The return value from srcu_read_lock() is guaranteed to be
+ * non-negative.  This value must be passed unaltered to the matching
+ * srcu_read_unlock().  Note that srcu_read_lock() and the matching
+ * srcu_read_unlock() must occur in the same context, for example, it is
+ * illegal to invoke srcu_read_unlock() in an irq handler if the matching
+ * srcu_read_lock() was invoked in process context.  Or, for that matter to
+ * invoke srcu_read_unlock() from one task and the matching srcu_read_lock()
+ * from another.
  */
 static inline int srcu_read_lock(struct srcu_struct *ssp) __acquires(ssp)
 {

include/linux/srcutree.h

@@ -26,6 +26,7 @@ struct srcu_data {
 	atomic_long_t srcu_lock_count[2];	/* Locks per CPU. */
 	atomic_long_t srcu_unlock_count[2];	/* Unlocks per CPU. */
 	int srcu_reader_flavor;			/* Reader flavor for srcu_struct structure? */
+						/* Values: SRCU_READ_FLAVOR_.*  */
 
 	/* Update-side state. */
 	spinlock_t __private lock ____cacheline_internodealigned_in_smp;
@@ -43,11 +44,6 @@ struct srcu_data {
 	struct srcu_struct *ssp;
 };
 
-/* Values for ->srcu_reader_flavor. */
-#define SRCU_READ_FLAVOR_NORMAL	0x1		// srcu_read_lock().
-#define SRCU_READ_FLAVOR_NMI	0x2		// srcu_read_lock_nmisafe().
-#define SRCU_READ_FLAVOR_LITE	0x4		// srcu_read_lock_lite().
-
 /*
  * Node in SRCU combining tree, similar in function to rcu_data.
  */
@@ -258,7 +254,7 @@ static inline void srcu_check_read_flavor_lite(struct srcu_struct *ssp)
 	if (likely(READ_ONCE(sdp->srcu_reader_flavor) & SRCU_READ_FLAVOR_LITE))
 		return;
 
-	// Note that the cmpxchg() in srcu_check_read_flavor() is fully ordered.
+	// Note that the cmpxchg() in __srcu_check_read_flavor() is fully ordered.
 	__srcu_check_read_flavor(ssp, SRCU_READ_FLAVOR_LITE);
 }
 

include/linux/torture.h

@@ -130,7 +130,7 @@ void _torture_stop_kthread(char *m, struct task_struct **tp);
 #endif
 
 #if IS_ENABLED(CONFIG_RCU_TORTURE_TEST) || IS_MODULE(CONFIG_RCU_TORTURE_TEST) || IS_ENABLED(CONFIG_LOCK_TORTURE_TEST) || IS_MODULE(CONFIG_LOCK_TORTURE_TEST)
-long torture_sched_setaffinity(pid_t pid, const struct cpumask *in_mask);
+long torture_sched_setaffinity(pid_t pid, const struct cpumask *in_mask, bool dowarn);
 #endif
 
 #endif /* __LINUX_TORTURE_H */

kernel/locking/locktorture.c

@@ -106,7 +106,7 @@ static const struct kernel_param_ops lt_bind_ops = {
 module_param_cb(bind_readers, &lt_bind_ops, &bind_readers, 0644);
 module_param_cb(bind_writers, &lt_bind_ops, &bind_writers, 0644);
 
-long torture_sched_setaffinity(pid_t pid, const struct cpumask *in_mask);
+long torture_sched_setaffinity(pid_t pid, const struct cpumask *in_mask, bool dowarn);
 
 static struct task_struct *stats_task;
 static struct task_struct **writer_tasks;
@@ -1358,7 +1358,7 @@ static int __init lock_torture_init(void)
 		if (torture_init_error(firsterr))
 			goto unwind;
 		if (cpumask_nonempty(bind_writers))
-			torture_sched_setaffinity(writer_tasks[i]->pid, bind_writers);
+			torture_sched_setaffinity(writer_tasks[i]->pid, bind_writers, true);
 
 	create_reader:
 		if (cxt.cur_ops->readlock == NULL || (j >= cxt.nrealreaders_stress))
@@ -1369,7 +1369,7 @@ static int __init lock_torture_init(void)
 		if (torture_init_error(firsterr))
 			goto unwind;
 		if (cpumask_nonempty(bind_readers))
-			torture_sched_setaffinity(reader_tasks[j]->pid, bind_readers);
+			torture_sched_setaffinity(reader_tasks[j]->pid, bind_readers, true);
 	}
 	if (stat_interval > 0) {
 		firsterr = torture_create_kthread(lock_torture_stats, NULL,

kernel/rcu/Kconfig.debug

@@ -53,6 +53,37 @@ config RCU_TORTURE_TEST
 	  Say M if you want the RCU torture tests to build as a module.
 	  Say N if you are unsure.
 
+config RCU_TORTURE_TEST_CHK_RDR_STATE
+	tristate "Check rcutorture reader state"
+	depends on RCU_TORTURE_TEST
+	default n
+	help
+	  This option causes rcutorture to check the desired rcutorture
+	  reader state for each segment against the actual context.
+	  Note that PREEMPT_COUNT must be enabled if the preempt-disabled
+	  and bh-disabled checks are to take effect, and that PREEMPT_RCU
+	  must be enabled for the RCU-nesting checks to take effect.
+	  These checks add overhead, and this Kconfig options is therefore
+	  disabled by default.
+
+	  Say Y here if you want rcutorture reader contexts checked.
+	  Say N if you are unsure.
+
+config RCU_TORTURE_TEST_LOG_CPU
+	tristate "Log CPU for rcutorture failures"
+	depends on RCU_TORTURE_TEST
+	default n
+	help
+	  This option causes rcutorture to decorate each entry of its
+	  log of failure/close-call rcutorture reader segments with the
+	  number of the CPU that the reader was running on at the time.
+	  This information can be useful, but it does incur additional
+	  overhead, overhead that can make both failures and close calls
+	  less probable.
+
+	  Say Y here if you want CPU IDs logged.
+	  Say N if you are unsure.
+
 config RCU_REF_SCALE_TEST
 	tristate "Scalability tests for read-side synchronization (RCU and others)"
 	depends on DEBUG_KERNEL