trace,smp: Add tracepoints around remotelly called functions

The recently added ipi_send_{cpu,cpumask} tracepoints allow finding sources
of IPIs targeting CPUs running latency-sensitive applications.

For NOHZ_FULL CPUs, all IPIs are interference, and those tracepoints are
sufficient to find them and work on getting rid of them. In some setups
however, not *all* IPIs are to be suppressed, but long-running IPI
callbacks can still be problematic.

Add a pair of tracepoints to mark the start and end of processing a CSD IPI
callback, similar to what exists for softirq, workqueue or timer callbacks.

Signed-off-by: Leonardo Bras <leobras@redhat.com>
Tested-and-reviewed-by: Valentin Schneider <vschneid@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230615065944.188876-5-leobras@redhat.com

include/trace/events/csd.h

@@ -0,0 +1,45 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM csd
+
+#if !defined(_TRACE_CSD_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_CSD_H
+
+#include <linux/tracepoint.h>
+
+/*
+ * Tracepoints for a function which is called as an effect of smp_call_function.*
+ */
+DECLARE_EVENT_CLASS(csd_function,
+
+	TP_PROTO(smp_call_func_t func, struct __call_single_data *csd),
+
+	TP_ARGS(func, csd),
+
+	TP_STRUCT__entry(
+		__field(void *,	func)
+		__field(void *,	csd)
+	),
+
+	TP_fast_assign(
+		__entry->func	= func;
+		__entry->csd	= csd;
+	),
+
+	TP_printk("func=%ps, csd=%p", __entry->func, __entry->csd)
+);
+
+DEFINE_EVENT(csd_function, csd_function_entry,
+	TP_PROTO(smp_call_func_t func, struct __call_single_data *csd),
+	TP_ARGS(func, csd)
+);
+
+DEFINE_EVENT(csd_function, csd_function_exit,
+	TP_PROTO(smp_call_func_t func, struct __call_single_data *csd),
+	TP_ARGS(func, csd)
+);
+
+#endif /* _TRACE_CSD_H */
+
+/* This part must be outside protection */
+#include <trace/define_trace.h>

kernel/smp.c

@@ -27,6 +27,9 @@
 #include <linux/jump_label.h>
 
 #include <trace/events/ipi.h>
+#define CREATE_TRACE_POINTS
+#include <trace/events/csd.h>
+#undef CREATE_TRACE_POINTS
 
 #include "smpboot.h"
 #include "sched/smp.h"
@@ -121,6 +124,14 @@ send_call_function_ipi_mask(struct cpumask *mask)
 	arch_send_call_function_ipi_mask(mask);
 }
 
+static __always_inline void
+csd_do_func(smp_call_func_t func, void *info, struct __call_single_data *csd)
+{
+	trace_csd_function_entry(func, csd);
+	func(info);
+	trace_csd_function_exit(func, csd);
+}
+
 #ifdef CONFIG_CSD_LOCK_WAIT_DEBUG
 
 static DEFINE_STATIC_KEY_MAYBE(CONFIG_CSD_LOCK_WAIT_DEBUG_DEFAULT, csdlock_debug_enabled);
@@ -375,7 +386,7 @@ static int generic_exec_single(int cpu, struct __call_single_data *csd)
 		csd_lock_record(csd);
 		csd_unlock(csd);
 		local_irq_save(flags);
-		func(info);
+		csd_do_func(func, info, NULL);
 		csd_lock_record(NULL);
 		local_irq_restore(flags);
 		return 0;
@@ -477,7 +488,7 @@ static void __flush_smp_call_function_queue(bool warn_cpu_offline)
 			}
 
 			csd_lock_record(csd);
-			func(info);
+			csd_do_func(func, info, csd);
 			csd_unlock(csd);
 			csd_lock_record(NULL);
 		} else {
@@ -508,7 +519,7 @@ static void __flush_smp_call_function_queue(bool warn_cpu_offline)
 
 				csd_lock_record(csd);
 				csd_unlock(csd);
-				func(info);
+				csd_do_func(func, info, csd);
 				csd_lock_record(NULL);
 			} else if (type == CSD_TYPE_IRQ_WORK) {
 				irq_work_single(csd);
@@ -522,8 +533,10 @@ static void __flush_smp_call_function_queue(bool warn_cpu_offline)
 	/*
 	 * Third; only CSD_TYPE_TTWU is left, issue those.
 	 */
-	if (entry)
+	if (entry) {
-		sched_ttwu_pending(entry);
+		csd = llist_entry(entry, typeof(*csd), node.llist);
+		csd_do_func(sched_ttwu_pending, entry, csd);
+	}
 }
 
 
@@ -816,7 +829,7 @@ static void smp_call_function_many_cond(const struct cpumask *mask,
 		unsigned long flags;
 
 		local_irq_save(flags);
-		func(info);
+		csd_do_func(func, info, NULL);
 		local_irq_restore(flags);
 	}