Skip to contentSkip to conversation

selftests: powerpc: Fix CPU affinity for child process

PATCH Applied Harish 4 participants 4 messages Jun 9, 2020 – Aug 2, 2020

Subsystems (2)
lore
Share & tools
lore

Harish

author
#0 lore raw

PATCH v3 1/1 selftests: powerpc: Fix CPU affinity for child process

On systems with large number of cpus, test fails trying to set affinity by calling sched_setaffinity() with smaller size for affinity mask. This patch fixes it by making sure that the size of allocated affinity mask is dependent on the number of CPUs as reported by get_nprocs().

Fixes 00b7ec5c9cf3 ("selftests/powerpc: Import Anton's context_switch2 benchmark")

---
v2: https://lore.kernel.org/linuxppc-dev/[email protected]/

Changes from v2:
- Interchanged size and ncpus as suggested by Satheesh
- Revert the exit code as suggested by Satheesh
- Added NULL check for the affinity mask as suggested by Kamalesh
- Freed the affinity mask allocation after affinity is set as suggested by Kamalesh
- Changed "cpu set" to "affinity mask" in the commit message

---
 .../powerpc/benchmarks/context_switch.c       | 21 ++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)
tools/testing/selftests/powerpc/benchmarks/context_switch.c +16 -5
index a2e8c9da7fa5..d50cc05df495 100644
@@ -19,6 +19,7 @@
 #include <limits.h>
 #include <sys/time.h>
 #include <sys/syscall.h>
+#include <sys/sysinfo.h>
 #include <sys/types.h>
 #include <sys/shm.h>
 #include <linux/futex.h>
@@ -104,8 +105,9 @@ static void start_thread_on(void *(*fn)(void *), void *arg, unsigned long cpu)
 
 static void start_process_on(void *(*fn)(void *), void *arg, unsigned long cpu)
 {
-	int pid;
-	cpu_set_t cpuset;
+	int pid, ncpus;
+	cpu_set_t *cpuset;
+	size_t size;
 
 	pid = fork();
 	if (pid == -1) {
@@ -116,14 +118,23 @@ static void start_process_on(void *(*fn)(void *), void *arg, unsigned long cpu)
 	if (pid)
 		return;
 
-	CPU_ZERO(&cpuset);
-	CPU_SET(cpu, &cpuset);
+	ncpus = get_nprocs();
+	size = CPU_ALLOC_SIZE(ncpus);
+	cpuset = CPU_ALLOC(ncpus);
+	if (!cpuset) {
+		perror("malloc");
+		exit(1);
+	}
+	CPU_ZERO_S(size, cpuset);
+	CPU_SET_S(cpu, size, cpuset);
 
-	if (sched_setaffinity(0, sizeof(cpuset), &cpuset)) {
+	if (sched_setaffinity(0, size, cpuset)) {
 		perror("sched_setaffinity");
+		CPU_FREE(cpuset);
 		exit(1);
 	}
 
+	CPU_FREE(cpuset);
 	fn(arg);
 
 	exit(0);
Signature · 1 line
2.24.1
Reviewed-by on v3 1/1
#1 lore raw
Quoted 10 lines · Harish

On systems with large number of cpus, test fails trying to set affinity by calling sched_setaffinity() with smaller size for affinity mask. This patch fixes it by making sure that the size of allocated affinity mask is dependent on the number of CPUs as reported by get_nprocs().

Fixes 00b7ec5c9cf3 ("selftests/powerpc: Import Anton's context_switch2 benchmark")

LGTM,

Signature · 1 line
Kamalesh
Reviewed-by on v3 1/1
#2 lore raw
In reply to Harish
Quoted 11 lines · Harish

On systems with large number of cpus, test fails trying to set affinity by calling sched_setaffinity() with smaller size for affinity mask. This patch fixes it by making sure that the size of allocated affinity mask is dependent on the number of CPUs as reported by get_nprocs().

Fixes 00b7ec5c9cf3 ("selftests/powerpc: Import Anton's context_switch2 benchmark")

---

Quoted 68 lines

v2: https://lore.kernel.org/linuxppc-dev/[email protected]/

Changes from v2:
- Interchanged size and ncpus as suggested by Satheesh
- Revert the exit code as suggested by Satheesh
- Added NULL check for the affinity mask as suggested by Kamalesh
- Freed the affinity mask allocation after affinity is set as suggested by Kamalesh
- Changed "cpu set" to "affinity mask" in the commit message

---
 .../powerpc/benchmarks/context_switch.c       | 21 ++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)
tools/testing/selftests/powerpc/benchmarks/context_switch.c +16 -5
index a2e8c9da7fa5..d50cc05df495 100644
@@ -19,6 +19,7 @@
 #include <limits.h>
 #include <sys/time.h>
 #include <sys/syscall.h>
+#include <sys/sysinfo.h>
 #include <sys/types.h>
 #include <sys/shm.h>
 #include <linux/futex.h>
@@ -104,8 +105,9 @@ static void start_thread_on(void *(*fn)(void *), void *arg, unsigned long cpu)
 
 static void start_process_on(void *(*fn)(void *), void *arg, unsigned long cpu)
 {
-	int pid;
-	cpu_set_t cpuset;
+	int pid, ncpus;
+	cpu_set_t *cpuset;
+	size_t size;
 
 	pid = fork();
 	if (pid == -1) {
@@ -116,14 +118,23 @@ static void start_process_on(void *(*fn)(void *), void *arg, unsigned long cpu)
 	if (pid)
 		return;
 
-	CPU_ZERO(&cpuset);
-	CPU_SET(cpu, &cpuset);
+	ncpus = get_nprocs();
+	size = CPU_ALLOC_SIZE(ncpus);
+	cpuset = CPU_ALLOC(ncpus);
+	if (!cpuset) {
+		perror("malloc");
+		exit(1);
+	}
+	CPU_ZERO_S(size, cpuset);
+	CPU_SET_S(cpu, size, cpuset);
 
-	if (sched_setaffinity(0, sizeof(cpuset), &cpuset)) {
+	if (sched_setaffinity(0, size, cpuset)) {
 		perror("sched_setaffinity");
+		CPU_FREE(cpuset);
 		exit(1);
 	}
 
+	CPU_FREE(cpuset);
 	fn(arg);
 
 	exit(0);
Signature · 1 line
2.24.1
#3 lore raw
In reply to Harish
Quoted 5 lines · Harish

On systems with large number of cpus, test fails trying to set affinity by calling sched_setaffinity() with smaller size for affinity mask. This patch fixes it by making sure that the size of allocated affinity mask is dependent on the number of CPUs as reported by get_nprocs().

Applied to powerpc/next.

[1/1] selftests/powerpc: Fix CPU affinity for child process

      https://git.kernel.org/powerpc/c/854eb5022be04f81e318765f089f41a57c8e5d83

cheers