[PATCH v6 0/6] Switch __DECLARE_TRACE() to new notrace variant of SRCU-fast for v6.18

[PATCH v6 0/6] Switch __DECLARE_TRACE() to new notrace variant of SRCU-fast for v6.18

[Paul E. McKenney]

Hello!

This is version 6 of a patch series creating a new notrace variant of
SRCU-fast and introducing it to the __DECLARE_TRACE() macro in place of
the current preemption disabling.  This change enables preemption of BPF
programs attached to tracepoints, as is required for runtime use of BPF
in real-time systems.

The patches are as follows:

1.	Move rcu_is_watching() checks to srcu_read_{,un}lock_fast().

2.	Add srcu_read_lock_fast_notrace() and
	srcu_read_unlock_fast_notrace().

3.	Add guards for notrace variants of SRCU-fast readers.

4.	Guard __DECLARE_TRACE() use of __DO_TRACE_CALL() with SRCU-fast.

5.	Document __srcu_read_{,un}lock_fast() implicit RCU readers.

6.	Document srcu_flip() memory-barrier D relation to SRCU-fast.

Changes since v5:

o	Forward-port to v6.17-rc1.

Changes since v4:

o	Remove the redundant "notrace" tags per Boqun Feng feedback.

o	Disable preemption for perf and ftrace per Steven Rostedt
	feedback, which addresses the kernel test robot reports
	provoked by earlier versions.

	https://lore.kernel.org/all/45397494-544e-41c0-bf48-c66d213fce05@paulmck-laptop/

Changes since v3:

o	Add "notrace" per Joel, Steven, and Matthew feedback.

o	Upgrade explanatory comments and add new ones per Joel feedback.

	https://lore.kernel.org/all/20250721162433.10454-1-paulmck@kernel.org/

Changes since v2:

o	Posting standalone as opposed to a reply.

	https://lore.kernel.org/all/3cecf6c9-b2ee-4f34-9d1b-ca4cfb8e56a7@paulmck-laptop/

Changes since RFC version:

o	RFC patch 6/4 has been pulled into the shared RCU tree:
	e88c632a8698 ("srcu: Add guards for SRCU-fast readers")

o	RFC patch 5/4 (which removed the now-unnecessary special boot-time
	avoidance of SRCU) has been folded into patch 4/4 shown above,
	as suggested by Steven Rostedt.

	https://lore.kernel.org/all/bb20a575-235b-499e-aa1d-70fe9e2c7617@paulmck-laptop/

						Thanx, Paul

------------------------------------------------------------------------

 b/include/linux/srcu.h         |    4 +++
 b/include/linux/srcutree.h     |    2 -
 b/include/linux/tracepoint.h   |    6 +++--
 b/include/trace/perf.h         |    2 +
 b/include/trace/trace_events.h |    2 +
 b/kernel/rcu/srcutree.c        |   10 ++++++++
 b/kernel/tracepoint.c          |   21 +++++++++++++++++-
 include/linux/srcu.h           |   30 ++++++++++++++++++++++++++
 include/linux/srcutree.h       |   47 ++++++++++++++++++++++++++---------------
 9 files changed, 102 insertions(+), 22 deletions(-)

[PATCH v6 1/6] srcu: Move rcu_is_watching() checks to srcu_read_{,un}lock_fast()

[Paul E. McKenney]

The rcu_is_watching() warnings are currently in the SRCU-tree
implementations of __srcu_read_lock_fast() and __srcu_read_unlock_fast().
However, this makes it difficult to create _notrace variants of
srcu_read_lock_fast() and srcu_read_unlock_fast().  This commit therefore
moves these checks to srcu_read_lock_fast(), srcu_read_unlock_fast(),
srcu_down_read_fast(), and srcu_up_read_fast().

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Reviewed-by: Joel Fernandes <joelagnelf@nvidia.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: <bpf@vger.kernel.org>
---
 include/linux/srcu.h     | 4 ++++
 include/linux/srcutree.h | 2 --
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/include/linux/srcu.h b/include/linux/srcu.h
index f179700fecafb8..478c73d067f7d3 100644
--- a/include/linux/srcu.h
+++ b/include/linux/srcu.h
@@ -275,6 +275,7 @@ static inline struct srcu_ctr __percpu *srcu_read_lock_fast(struct srcu_struct *
 {
 	struct srcu_ctr __percpu *retval;
 
+	RCU_LOCKDEP_WARN(!rcu_is_watching(), "RCU must be watching srcu_read_lock_fast().");
 	srcu_check_read_flavor_force(ssp, SRCU_READ_FLAVOR_FAST);
 	retval = __srcu_read_lock_fast(ssp);
 	rcu_try_lock_acquire(&ssp->dep_map);
@@ -295,6 +296,7 @@ static inline struct srcu_ctr __percpu *srcu_read_lock_fast(struct srcu_struct *
 static inline struct srcu_ctr __percpu *srcu_down_read_fast(struct srcu_struct *ssp) __acquires(ssp)
 {
 	WARN_ON_ONCE(IS_ENABLED(CONFIG_PROVE_RCU) && in_nmi());
+	RCU_LOCKDEP_WARN(!rcu_is_watching(), "RCU must be watching srcu_down_read_fast().");
 	srcu_check_read_flavor_force(ssp, SRCU_READ_FLAVOR_FAST);
 	return __srcu_read_lock_fast(ssp);
 }
@@ -389,6 +391,7 @@ static inline void srcu_read_unlock_fast(struct srcu_struct *ssp, struct srcu_ct
 	srcu_check_read_flavor(ssp, SRCU_READ_FLAVOR_FAST);
 	srcu_lock_release(&ssp->dep_map);
 	__srcu_read_unlock_fast(ssp, scp);
+	RCU_LOCKDEP_WARN(!rcu_is_watching(), "RCU must be watching srcu_read_unlock_fast().");
 }
 
 /**
@@ -405,6 +408,7 @@ static inline void srcu_up_read_fast(struct srcu_struct *ssp, struct srcu_ctr __
 	WARN_ON_ONCE(IS_ENABLED(CONFIG_PROVE_RCU) && in_nmi());
 	srcu_check_read_flavor(ssp, SRCU_READ_FLAVOR_FAST);
 	__srcu_read_unlock_fast(ssp, scp);
+	RCU_LOCKDEP_WARN(!rcu_is_watching(), "RCU must be watching srcu_up_read_fast().");
 }
 
 /**
diff --git a/include/linux/srcutree.h b/include/linux/srcutree.h
index bf44d8d1e69eab..043b5a67ef71eb 100644
--- a/include/linux/srcutree.h
+++ b/include/linux/srcutree.h
@@ -244,7 +244,6 @@ static inline struct srcu_ctr __percpu *__srcu_read_lock_fast(struct srcu_struct
 {
 	struct srcu_ctr __percpu *scp = READ_ONCE(ssp->srcu_ctrp);
 
-	RCU_LOCKDEP_WARN(!rcu_is_watching(), "RCU must be watching srcu_read_lock_fast().");
 	if (!IS_ENABLED(CONFIG_NEED_SRCU_NMI_SAFE))
 		this_cpu_inc(scp->srcu_locks.counter); /* Y */
 	else
@@ -275,7 +274,6 @@ static inline void __srcu_read_unlock_fast(struct srcu_struct *ssp, struct srcu_
 		this_cpu_inc(scp->srcu_unlocks.counter);  /* Z */
 	else
 		atomic_long_inc(raw_cpu_ptr(&scp->srcu_unlocks));  /* Z */
-	RCU_LOCKDEP_WARN(!rcu_is_watching(), "RCU must be watching srcu_read_unlock_fast().");
 }
 
 void __srcu_check_read_flavor(struct srcu_struct *ssp, int read_flavor);
-- 
2.40.1

[PATCH v6 2/6] srcu: Add srcu_read_lock_fast_notrace() and srcu_read_unlock_fast_notrace()

[Paul E. McKenney]

This commit adds no-trace variants of the srcu_read_lock_fast() and
srcu_read_unlock_fast() functions for tracing use.

[ paulmck: Apply notrace feedback from Joel Fernandes, Steven Rostedt, and Mathieu Desnoyers. ]
[ paulmck: Apply excess-notrace feedback from Boqun Feng. ]

Link: https://lore.kernel.org/all/20250721162433.10454-1-paulmck@kernel.org
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Reviewed-by: Joel Fernandes <joelagnelf@nvidia.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: <bpf@vger.kernel.org>
---
 include/linux/srcu.h     | 25 +++++++++++++++++++++++++
 include/linux/srcutree.h |  5 +++--
 2 files changed, 28 insertions(+), 2 deletions(-)

diff --git a/include/linux/srcu.h b/include/linux/srcu.h
index 478c73d067f7d3..7a692bf8f99b96 100644
--- a/include/linux/srcu.h
+++ b/include/linux/srcu.h
@@ -282,6 +282,20 @@ static inline struct srcu_ctr __percpu *srcu_read_lock_fast(struct srcu_struct *
 	return retval;
 }
 
+/*
+ * Used by tracing, cannot be traced and cannot call lockdep.
+ * See srcu_read_lock_fast() for more information.
+ */
+static inline struct srcu_ctr __percpu *srcu_read_lock_fast_notrace(struct srcu_struct *ssp)
+	__acquires(ssp)
+{
+	struct srcu_ctr __percpu *retval;
+
+	srcu_check_read_flavor_force(ssp, SRCU_READ_FLAVOR_FAST);
+	retval = __srcu_read_lock_fast(ssp);
+	return retval;
+}
+
 /**
  * srcu_down_read_fast - register a new reader for an SRCU-protected structure.
  * @ssp: srcu_struct in which to register the new reader.
@@ -394,6 +408,17 @@ static inline void srcu_read_unlock_fast(struct srcu_struct *ssp, struct srcu_ct
 	RCU_LOCKDEP_WARN(!rcu_is_watching(), "RCU must be watching srcu_read_unlock_fast().");
 }
 
+/*
+ * Used by tracing, cannot be traced and cannot call lockdep.
+ * See srcu_read_unlock_fast() for more information.
+ */
+static inline void srcu_read_unlock_fast_notrace(struct srcu_struct *ssp,
+						 struct srcu_ctr __percpu *scp) __releases(ssp)
+{
+	srcu_check_read_flavor(ssp, SRCU_READ_FLAVOR_FAST);
+	__srcu_read_unlock_fast(ssp, scp);
+}
+
 /**
  * srcu_up_read_fast - unregister a old reader from an SRCU-protected structure.
  * @ssp: srcu_struct in which to unregister the old reader.
diff --git a/include/linux/srcutree.h b/include/linux/srcutree.h
index 043b5a67ef71eb..4d2fee4d38289f 100644
--- a/include/linux/srcutree.h
+++ b/include/linux/srcutree.h
@@ -240,7 +240,7 @@ static inline struct srcu_ctr __percpu *__srcu_ctr_to_ptr(struct srcu_struct *ss
  * on architectures that support NMIs but do not supply NMI-safe
  * implementations of this_cpu_inc().
  */
-static inline struct srcu_ctr __percpu *__srcu_read_lock_fast(struct srcu_struct *ssp)
+static inline struct srcu_ctr __percpu notrace *__srcu_read_lock_fast(struct srcu_struct *ssp)
 {
 	struct srcu_ctr __percpu *scp = READ_ONCE(ssp->srcu_ctrp);
 
@@ -267,7 +267,8 @@ static inline struct srcu_ctr __percpu *__srcu_read_lock_fast(struct srcu_struct
  * on architectures that support NMIs but do not supply NMI-safe
  * implementations of this_cpu_inc().
  */
-static inline void __srcu_read_unlock_fast(struct srcu_struct *ssp, struct srcu_ctr __percpu *scp)
+static inline void notrace
+__srcu_read_unlock_fast(struct srcu_struct *ssp, struct srcu_ctr __percpu *scp)
 {
 	barrier();  /* Avoid leaking the critical section. */
 	if (!IS_ENABLED(CONFIG_NEED_SRCU_NMI_SAFE))
-- 
2.40.1

[PATCH v6 3/6] srcu: Add guards for notrace variants of SRCU-fast readers

[Paul E. McKenney]

This adds the usual scoped_guard(srcu_fast_notrace, &my_srcu) and
guard(srcu_fast_notrace)(&my_srcu).

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Reviewed-by: Joel Fernandes <joelagnelf@nvidia.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: <bpf@vger.kernel.org>
---
 include/linux/srcu.h | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/include/linux/srcu.h b/include/linux/srcu.h
index 7a692bf8f99b96..ada65b58bc4c5f 100644
--- a/include/linux/srcu.h
+++ b/include/linux/srcu.h
@@ -515,4 +515,9 @@ DEFINE_LOCK_GUARD_1(srcu_fast, struct srcu_struct,
 		    srcu_read_unlock_fast(_T->lock, _T->scp),
 		    struct srcu_ctr __percpu *scp)
 
+DEFINE_LOCK_GUARD_1(srcu_fast_notrace, struct srcu_struct,
+		    _T->scp = srcu_read_lock_fast_notrace(_T->lock),
+		    srcu_read_unlock_fast_notrace(_T->lock, _T->scp),
+		    struct srcu_ctr __percpu *scp)
+
 #endif
-- 
2.40.1

[PATCH v6 4/6] tracing: Guard __DECLARE_TRACE() use of __DO_TRACE_CALL() with SRCU-fast

[Paul E. McKenney]

The current use of guard(preempt_notrace)() within __DECLARE_TRACE()
to protect invocation of __DO_TRACE_CALL() means that BPF programs
attached to tracepoints are non-preemptible.  This is unhelpful in
real-time systems, whose users apparently wish to use BPF while also
achieving low latencies.  (Who knew?)

One option would be to use preemptible RCU, but this introduces
many opportunities for infinite recursion, which many consider to
be counterproductive, especially given the relatively small stacks
provided by the Linux kernel.  These opportunities could be shut down
by sufficiently energetic duplication of code, but this sort of thing
is considered impolite in some circles.

Therefore, use the shiny new SRCU-fast API, which provides somewhat faster
readers than those of preemptible RCU, at least on my laptop, where
task_struct access is more expensive than access to per-CPU variables.
And SRCU fast provides way faster readers than does SRCU, courtesy of
being able to avoid the read-side use of smp_mb().  Also, it is quite
straightforward to create srcu_read_{,un}lock_fast_notrace() functions.

While in the area, SRCU now supports early boot call_srcu().  Therefore,
remove the checks that used to avoid such use from rcu_free_old_probes()
before this commit was applied:

e53244e2c893 ("tracepoint: Remove SRCU protection")

The current commit can be thought of as an approximate revert of that
commit, with some compensating additions of preemption disabling pointed
out by Steven Rostedt (thank you, Steven!).

[ Apply kernel test robot feedback. ]

Link: https://lore.kernel.org/all/20250613152218.1924093-1-bigeasy@linutronix.de/
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: <bpf@vger.kernel.org>
---
 include/linux/tracepoint.h   |  6 ++++--
 include/trace/perf.h         |  2 ++
 include/trace/trace_events.h |  2 ++
 kernel/tracepoint.c          | 21 ++++++++++++++++++++-
 4 files changed, 28 insertions(+), 3 deletions(-)

diff --git a/include/linux/tracepoint.h b/include/linux/tracepoint.h
index 826ce3f8e1f851..a22c1ab88560b8 100644
--- a/include/linux/tracepoint.h
+++ b/include/linux/tracepoint.h
@@ -33,6 +33,8 @@ struct trace_eval_map {
 
 #define TRACEPOINT_DEFAULT_PRIO	10
 
+extern struct srcu_struct tracepoint_srcu;
+
 extern int
 tracepoint_probe_register(struct tracepoint *tp, void *probe, void *data);
 extern int
@@ -115,7 +117,7 @@ void for_each_tracepoint_in_module(struct module *mod,
 static inline void tracepoint_synchronize_unregister(void)
 {
 	synchronize_rcu_tasks_trace();
-	synchronize_rcu();
+	synchronize_srcu(&tracepoint_srcu);
 }
 static inline bool tracepoint_is_faultable(struct tracepoint *tp)
 {
@@ -271,7 +273,7 @@ static inline struct tracepoint *tracepoint_ptr_deref(tracepoint_ptr_t *p)
 	static inline void __do_trace_##name(proto)			\
 	{								\
 		if (cond) {						\
-			guard(preempt_notrace)();			\
+			guard(srcu_fast_notrace)(&tracepoint_srcu);	\
 			__DO_TRACE_CALL(name, TP_ARGS(args));		\
 		}							\
 	}								\
diff --git a/include/trace/perf.h b/include/trace/perf.h
index a1754b73a8f55b..1b7925a859665f 100644
--- a/include/trace/perf.h
+++ b/include/trace/perf.h
@@ -71,7 +71,9 @@ perf_trace_##call(void *__data, proto)					\
 	u64 __count __attribute__((unused));				\
 	struct task_struct *__task __attribute__((unused));		\
 									\
+	preempt_disable_notrace();					\
 	do_perf_trace_##call(__data, args);				\
+	preempt_enable_notrace();					\
 }
 
 #undef DECLARE_EVENT_SYSCALL_CLASS
diff --git a/include/trace/trace_events.h b/include/trace/trace_events.h
index 4f22136fd4656c..0504a423ca2539 100644
--- a/include/trace/trace_events.h
+++ b/include/trace/trace_events.h
@@ -436,7 +436,9 @@ __DECLARE_EVENT_CLASS(call, PARAMS(proto), PARAMS(args), PARAMS(tstruct), \
 static notrace void							\
 trace_event_raw_event_##call(void *__data, proto)			\
 {									\
+	preempt_disable_notrace();					\
 	do_trace_event_raw_event_##call(__data, args);			\
+	preempt_enable_notrace();					\
 }
 
 #undef DECLARE_EVENT_SYSCALL_CLASS
diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c
index 62719d2941c900..e19973015cbd73 100644
--- a/kernel/tracepoint.c
+++ b/kernel/tracepoint.c
@@ -25,6 +25,9 @@ enum tp_func_state {
 extern tracepoint_ptr_t __start___tracepoints_ptrs[];
 extern tracepoint_ptr_t __stop___tracepoints_ptrs[];
 
+DEFINE_SRCU(tracepoint_srcu);
+EXPORT_SYMBOL_GPL(tracepoint_srcu);
+
 enum tp_transition_sync {
 	TP_TRANSITION_SYNC_1_0_1,
 	TP_TRANSITION_SYNC_N_2_1,
@@ -34,6 +37,7 @@ enum tp_transition_sync {
 
 struct tp_transition_snapshot {
 	unsigned long rcu;
+	unsigned long srcu_gp;
 	bool ongoing;
 };
 
@@ -46,6 +50,7 @@ static void tp_rcu_get_state(enum tp_transition_sync sync)
 
 	/* Keep the latest get_state snapshot. */
 	snapshot->rcu = get_state_synchronize_rcu();
+	snapshot->srcu_gp = start_poll_synchronize_srcu(&tracepoint_srcu);
 	snapshot->ongoing = true;
 }
 
@@ -56,6 +61,8 @@ static void tp_rcu_cond_sync(enum tp_transition_sync sync)
 	if (!snapshot->ongoing)
 		return;
 	cond_synchronize_rcu(snapshot->rcu);
+	if (!poll_state_synchronize_srcu(&tracepoint_srcu, snapshot->srcu_gp))
+		synchronize_srcu(&tracepoint_srcu);
 	snapshot->ongoing = false;
 }
 
@@ -101,17 +108,29 @@ static inline void *allocate_probes(int count)
 	return p == NULL ? NULL : p->probes;
 }
 
-static void rcu_free_old_probes(struct rcu_head *head)
+static void srcu_free_old_probes(struct rcu_head *head)
 {
 	kfree(container_of(head, struct tp_probes, rcu));
 }
 
+static void rcu_free_old_probes(struct rcu_head *head)
+{
+	call_srcu(&tracepoint_srcu, head, srcu_free_old_probes);
+}
+
 static inline void release_probes(struct tracepoint *tp, struct tracepoint_func *old)
 {
 	if (old) {
 		struct tp_probes *tp_probes = container_of(old,
 			struct tp_probes, probes[0]);
 
+		/*
+		 * Tracepoint probes are protected by either RCU or
+		 * Tasks Trace RCU and also by SRCU.  By calling the SRCU
+		 * callback in the [Tasks Trace] RCU callback we cover
+		 * both cases. So let us chain the SRCU and [Tasks Trace]
+		 * RCU callbacks to wait for both grace periods.
+		 */
 		if (tracepoint_is_faultable(tp))
 			call_rcu_tasks_trace(&tp_probes->rcu, rcu_free_old_probes);
 		else
-- 
2.40.1

[PATCH v6 5/6] srcu: Document __srcu_read_{,un}lock_fast() implicit RCU readers

[Paul E. McKenney]

This commit documents the implicit RCU readers that are implied by the
this_cpu_inc() and atomic_long_inc() operations in __srcu_read_lock_fast()
and __srcu_read_unlock_fast().  While in the area, fix the documentation
of the memory pairing of atomic_long_inc() in __srcu_read_lock_fast().

[ paulmck: Apply Joel Fernandes feedback. ]

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: <bpf@vger.kernel.org>
---
 include/linux/srcutree.h | 42 ++++++++++++++++++++++++++--------------
 1 file changed, 27 insertions(+), 15 deletions(-)

diff --git a/include/linux/srcutree.h b/include/linux/srcutree.h
index 4d2fee4d38289f..42098e0fa0b7dd 100644
--- a/include/linux/srcutree.h
+++ b/include/linux/srcutree.h
@@ -232,9 +232,27 @@ 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 they
- * are a single instruction, or because they are a read-modify-write atomic
- * operation, depending on the whims of the architecture.
+ * critical sections either because they disables 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
+ * synchronize_rcu() or synchronize_rcu_expedited(), that is, RCU,
+ * *not* SRCU, in order to eliminate the need for the read-side smp_mb()
+ * invocations that are used by srcu_read_lock() and srcu_read_unlock().
+ * The __srcu_read_unlock_fast() function also relies on this same RCU
+ * (again, *not* SRCU) trick to eliminate the need for smp_mb().
+ *
+ * The key point behind this RCU trick is that if any part of a given
+ * RCU reader precedes the beginning of a given RCU grace period, then
+ * the entirety of that RCU reader and everything preceding it happens
+ * before the end of that same RCU grace period.  Similarly, if any part
+ * of a given RCU reader follows the end of a given RCU grace period,
+ * then the entirety of that RCU reader and everything following it
+ * happens after the beginning of that same RCU grace period.  Therefore,
+ * the operations labeled Y in __srcu_read_lock_fast() and those labeled Z
+ * in __srcu_read_unlock_fast() are ordered against the corresponding SRCU
+ * read-side critical section from the viewpoint of the SRCU grace period.
+ * This is all the ordering that is required, hence no calls to smp_mb().
  *
  * This means that __srcu_read_lock_fast() is not all that fast
  * on architectures that support NMIs but do not supply NMI-safe
@@ -245,9 +263,9 @@ static inline struct srcu_ctr __percpu notrace *__srcu_read_lock_fast(struct src
 	struct srcu_ctr __percpu *scp = READ_ONCE(ssp->srcu_ctrp);
 
 	if (!IS_ENABLED(CONFIG_NEED_SRCU_NMI_SAFE))
-		this_cpu_inc(scp->srcu_locks.counter); /* Y */
+		this_cpu_inc(scp->srcu_locks.counter); // Y, and implicit RCU reader.
 	else
-		atomic_long_inc(raw_cpu_ptr(&scp->srcu_locks));  /* Z */
+		atomic_long_inc(raw_cpu_ptr(&scp->srcu_locks));  // Y, and implicit RCU reader.
 	barrier(); /* Avoid leaking the critical section. */
 	return scp;
 }
@@ -258,23 +276,17 @@ static inline struct srcu_ctr __percpu notrace *__srcu_read_lock_fast(struct src
  * different CPU than that which was incremented by the corresponding
  * srcu_read_lock_fast(), but it must be within the same task.
  *
- * Note that both this_cpu_inc() and atomic_long_inc() are RCU read-side
- * critical sections either because they disables interrupts, because they
- * are a single instruction, or because they are a read-modify-write atomic
- * operation, depending on the whims of the architecture.
- *
- * This means that __srcu_read_unlock_fast() is not all that fast
- * on architectures that support NMIs but do not supply NMI-safe
- * implementations of this_cpu_inc().
+ * Please see the __srcu_read_lock_fast() function's header comment for
+ * information on implicit RCU readers and NMI safety.
  */
 static inline void notrace
 __srcu_read_unlock_fast(struct srcu_struct *ssp, struct srcu_ctr __percpu *scp)
 {
 	barrier();  /* Avoid leaking the critical section. */
 	if (!IS_ENABLED(CONFIG_NEED_SRCU_NMI_SAFE))
-		this_cpu_inc(scp->srcu_unlocks.counter);  /* Z */
+		this_cpu_inc(scp->srcu_unlocks.counter);  // Z, and implicit RCU reader.
 	else
-		atomic_long_inc(raw_cpu_ptr(&scp->srcu_unlocks));  /* Z */
+		atomic_long_inc(raw_cpu_ptr(&scp->srcu_unlocks));  // Z, and implicit RCU reader.
 }
 
 void __srcu_check_read_flavor(struct srcu_struct *ssp, int read_flavor);
-- 
2.40.1

[PATCH v6 6/6] srcu: Document srcu_flip() memory-barrier D relation to SRCU-fast

[Paul E. McKenney]

The smp_mb() memory barrier at the end of srcu_flip() has a comment,
but that comment does not make it clear that this memory barrier is an
optimization, as opposed to being needed for correctness.  This commit
therefore adds this information and points out that it is omitted
for SRCU-fast, where a much heavier weight synchronize_srcu() would
be required.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: <bpf@vger.kernel.org>
---
 kernel/rcu/srcutree.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/kernel/rcu/srcutree.c b/kernel/rcu/srcutree.c
index c5e8ebc493d5ee..1ff94b76d91f15 100644
--- a/kernel/rcu/srcutree.c
+++ b/kernel/rcu/srcutree.c
@@ -1168,6 +1168,16 @@ static void srcu_flip(struct srcu_struct *ssp)
 	 * counter update.  Note that both this memory barrier and the
 	 * one in srcu_readers_active_idx_check() provide the guarantee
 	 * for __srcu_read_lock().
+	 *
+	 * Note that this is a performance optimization, in which we spend
+	 * an otherwise unnecessary smp_mb() in order to reduce the number
+	 * of full per-CPU-variable scans in srcu_readers_lock_idx() and
+	 * srcu_readers_unlock_idx().  But this performance optimization
+	 * is not so optimal for SRCU-fast, where we would be spending
+	 * not smp_mb(), but rather synchronize_rcu().  At the same time,
+	 * the overhead of the smp_mb() is in the noise, so there is no
+	 * point in omitting it in the SRCU-fast case.  So the same code
+	 * is executed either way.
 	 */
 	smp_mb(); /* D */  /* Pairs with C. */
 }
-- 
2.40.1