[PATCH v2 2/2] KVM: selftests: Use getcpu() instead of sched_getcpu() in rseq_test

[Gavin Shan]

sched_getcpu() is glibc dependent and it can simply return the CPU
ID from the registered rseq information, as Florian Weimer pointed.
In this case, it's pointless to compare the return value from
sched_getcpu() and that fetched from the registered rseq information.

Fix the issue by replacing sched_getcpu() with getcpu(), as Florian
suggested. The comments are modified accordingly by replacing
"sched_getcpu()" with "getcpu()".

Reported-by: Yihuang Yu <yihyu@xxxxxxxxxx>
Suggested-by: Florian Weimer <fweimer@xxxxxxxxxx>
Suggested-by: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
Suggested-by: Sean Christopherson <seanjc@xxxxxxxxxx>
Signed-off-by: Gavin Shan <gshan@xxxxxxxxxx>
---
 tools/testing/selftests/kvm/rseq_test.c | 42 ++++++++++++++++---------
 1 file changed, 27 insertions(+), 15 deletions(-)

diff --git a/tools/testing/selftests/kvm/rseq_test.c b/tools/testing/selftests/kvm/rseq_test.c
index 2cd5fe49ac8b..84e8425edc2c 100644
--- a/tools/testing/selftests/kvm/rseq_test.c
+++ b/tools/testing/selftests/kvm/rseq_test.c
@@ -42,6 +42,18 @@ static void guest_code(void)
 		GUEST_SYNC(0);
 }
 
+/*
+ * We have to perform direct system call for getcpu() because it's
+ * not available until glic 2.29.
+ */
+static void sys_getcpu(unsigned *cpu)
+{
+	int r;
+
+	r = syscall(__NR_getcpu, cpu, NULL, NULL);
+	TEST_ASSERT(!r, "getcpu failed, errno = %d (%s)", errno, strerror(errno));
+}
+
 static int next_cpu(int cpu)
 {
 	/*
@@ -86,7 +98,7 @@ static void *migration_worker(void *__rseq_tid)
 		atomic_inc(&seq_cnt);
 
 		/*
-		 * Ensure the odd count is visible while sched_getcpu() isn't
+		 * Ensure the odd count is visible while getcpu() isn't
 		 * stable, i.e. while changing affinity is in-progress.
 		 */
 		smp_wmb();
@@ -127,10 +139,10 @@ static void *migration_worker(void *__rseq_tid)
 		 *     check completes.
 		 *
 		 *  3. To ensure the read-side makes efficient forward progress,
-		 *     e.g. if sched_getcpu() involves a syscall.  Stalling the
-		 *     read-side means the test will spend more time waiting for
-		 *     sched_getcpu() to stabilize and less time trying to hit
-		 *     the timing-dependent bug.
+		 *     e.g. if getcpu() involves a syscall. Stalling the read-side
+		 *     means the test will spend more time waiting for getcpu()
+		 *     to stabilize and less time trying to hit the timing-dependent
+		 *     bug.
 		 *
 		 * Because any bug in this area is likely to be timing-dependent,
 		 * run with a range of delays at 1us intervals from 1us to 10us
@@ -226,9 +238,9 @@ int main(int argc, char *argv[])
 
 		/*
 		 * Verify rseq's CPU matches sched's CPU.  Ensure migration
-		 * doesn't occur between sched_getcpu() and reading the rseq
-		 * cpu_id by rereading both if the sequence count changes, or
-		 * if the count is odd (migration in-progress).
+		 * doesn't occur between getcpu() and reading the rseq cpu_id
+		 * by rereading both if the sequence count changes, or if the
+		 * count is odd (migration in-progress).
 		 */
 		do {
 			/*
@@ -238,12 +250,12 @@ int main(int argc, char *argv[])
 			snapshot = atomic_read(&seq_cnt) & ~1;
 
 			/*
-			 * Ensure reading sched_getcpu() and rseq.cpu_id
-			 * complete in a single "no migration" window, i.e. are
-			 * not reordered across the seq_cnt reads.
+			 * Ensure reading getcpu() and rseq.cpu_id complete in
+			 * a single "no migration" window, i.e. are not reordered
+			 * across the seq_cnt reads.
 			 */
 			smp_rmb();
-			cpu = sched_getcpu();
+			sys_getcpu(&cpu);
 			rseq_cpu = READ_ONCE(__rseq->cpu_id);
 			smp_rmb();
 		} while (snapshot != atomic_read(&seq_cnt));
@@ -255,9 +267,9 @@ int main(int argc, char *argv[])
 	/*
 	 * Sanity check that the test was able to enter the guest a reasonable
 	 * number of times, e.g. didn't get stalled too often/long waiting for
-	 * sched_getcpu() to stabilize.  A 2:1 migration:KVM_RUN ratio is a
-	 * fairly conservative ratio on x86-64, which can do _more_ KVM_RUNs
-	 * than migrations given the 1us+ delay in the migration task.
+	 * getcpu() to stabilize.  A 2:1 migration:KVM_RUN ratio is a fairly
+	 * conservative ratio on x86-64, which can do _more_ KVM_RUNs than
+	 * migrations given the 1us+ delay in the migration task.
 	 */
 	TEST_ASSERT(i > (NR_TASK_MIGRATIONS / 2),
 		    "Only performed %d KVM_RUNs, task stalled too much?\n", i);
-- 
2.23.0