scftorture: Count single_rpc offline failures in statistics outputgrafted

scf_torture_stats_print() aggregates each invoker thread's counters
into a local scf_statistics structure before printing, but the
n_single_rpc_ofl field is missing from the aggregation loop.  As a
result, the "single_rpc_ofl" value printed in the statistics line is
always zero, even when smp_call_function_single() invocations for the
RPC test have failed due to offline CPUs and been counted by the
invoker threads.

Add the missing accumulation so that the printed value reflects the
actual counts.

Signed-off-by: Joel Fernandes <joelagnelf@nvidia.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>

2 files changed, 94 insertions, 0 deletions

diff --git a/tools/tracing/rtla/example/timerlat_bpf_action.c b/tools/tracing/rtla/example/timerlat_bpf_action.c
new file mode 100644
index 000000000..ac1be049a
--- /dev/null
+++ b/tools/tracing/rtla/example/timerlat_bpf_action.c
@@ -0,0 +1,16 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/bpf.h>
+#include <bpf/bpf_tracing.h>
+
+char LICENSE[] SEC("license") = "GPL";
+
+struct trace_event_raw_timerlat_sample {
+	unsigned long long timer_latency;
+} __attribute__((preserve_access_index));
+
+SEC("tp/timerlat_action")
+int action_handler(struct trace_event_raw_timerlat_sample *tp_args)
+{
+	bpf_printk("Latency: %lld\n", tp_args->timer_latency);
+	return 0;
+}
diff --git a/tools/tracing/rtla/example/timerlat_load.py b/tools/tracing/rtla/example/timerlat_load.py
new file mode 100644
index 000000000..a819c3588
--- /dev/null
+++ b/tools/tracing/rtla/example/timerlat_load.py
@@ -0,0 +1,78 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Copyright (C) 2024 Red Hat, Inc. Daniel Bristot de Oliveira <bristot@kernel.org>
+#
+# This is a sample code about how to use timerlat's timer by any workload
+# so rtla can measure and provide auto-analysis for the overall latency (IOW
+# the response time) for a task.
+#
+# Before running it, you need to dispatch timerlat with -U option in a terminal.
+# Then # run this script pinned to a CPU on another terminal. For example:
+#
+# timerlat_load.py 1 -p 95
+#
+# The "Timerlat IRQ" is the IRQ latency, The thread latency is the latency
+# for the python process to get the CPU. The Ret from user Timer Latency is
+# the overall latency. In other words, it is the response time for that
+# activation.
+#
+# This is just an example, the load is reading 20MB of data from /dev/full
+# It is in python because it is easy to read :-)
+
+import argparse
+import sys
+import os
+
+parser = argparse.ArgumentParser(description='user-space timerlat thread in Python')
+parser.add_argument("cpu", type=int, help='CPU to run timerlat thread')
+parser.add_argument("-p", "--prio", type=int, help='FIFO priority')
+args = parser.parse_args()
+
+try:
+    affinity_mask = {args.cpu}
+    os.sched_setaffinity(0, affinity_mask)
+except Exception as e:
+    print(f"Error setting affinity: {e}")
+    sys.exit(1)
+
+if args.prio:
+    try:
+        param = os.sched_param(args.prio)
+        os.sched_setscheduler(0, os.SCHED_FIFO, param)
+    except Exception as e:
+        print(f"Error setting priority: {e}")
+        sys.exit(1)
+
+try:
+    timerlat_path = f"/sys/kernel/tracing/osnoise/per_cpu/cpu{args.cpu}/timerlat_fd"
+    timerlat_fd = open(timerlat_path, 'r')
+except PermissionError:
+    print("Permission denied. Please check your access rights.")
+    sys.exit(1)
+except OSError:
+    print("Error opening timerlat fd, did you run timerlat -U?")
+    sys.exit(1)
+
+try:
+    data_fd = open("/dev/full", 'r')
+except Exception as e:
+    print(f"Error opening data fd: {e}")
+    sys.exit(1)
+
+while True:
+    try:
+        timerlat_fd.read(1)
+        data_fd.read(20 * 1024 * 1024)
+    except KeyboardInterrupt:
+        print("Leaving")
+        break
+    except IOError as e:
+        print(f"I/O error occurred: {e}")
+        break
+    except Exception as e:
+        print(f"Unexpected error: {e}")
+        break
+
+timerlat_fd.close()
+data_fd.close()