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>

1 files changed, 56 insertions, 0 deletions

diff --git a/arch/alpha/lib/udelay.c b/arch/alpha/lib/udelay.c
new file mode 100644
index 000000000..873648263
--- /dev/null
+++ b/arch/alpha/lib/udelay.c
@@ -0,0 +1,56 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 1993, 2000 Linus Torvalds
+ *
+ * Delay routines, using a pre-computed "loops_per_jiffy" value.
+ */
+
+#include <linux/module.h>
+#include <linux/sched.h> /* for udelay's use of smp_processor_id */
+#include <asm/param.h>
+#include <asm/smp.h>
+#include <linux/delay.h>
+
+/*
+ * Use only for very small delays (< 1 msec). 
+ *
+ * The active part of our cycle counter is only 32-bits wide, and
+ * we're treating the difference between two marks as signed.  On
+ * a 1GHz box, that's about 2 seconds.
+ */
+
+void
+__delay(int loops)
+{
+	int tmp;
+	__asm__ __volatile__(
+		"	rpcc %0\n"
+		"	addl %1,%0,%1\n"
+		"1:	rpcc %0\n"
+		"	subl %1,%0,%0\n"
+		"	bgt %0,1b"
+		: "=&r" (tmp), "=r" (loops) : "1"(loops));
+}
+EXPORT_SYMBOL(__delay);
+
+#ifdef CONFIG_SMP
+#define LPJ	 cpu_data[smp_processor_id()].loops_per_jiffy
+#else
+#define LPJ	 loops_per_jiffy
+#endif
+
+void
+udelay(unsigned long usecs)
+{
+	usecs *= (((unsigned long)HZ << 32) / 1000000) * LPJ;
+	__delay((long)usecs >> 32);
+}
+EXPORT_SYMBOL(udelay);
+
+void
+ndelay(unsigned long nsecs)
+{
+	nsecs *= (((unsigned long)HZ << 32) / 1000000000) * LPJ;
+	__delay((long)nsecs >> 32);
+}
+EXPORT_SYMBOL(ndelay);