[PATCH v2 1/4] tools/perf: Fix perf bench futex to correct usage of affinity for machines with #CPUs > 1K

From: Athira Rajeev
Date: Wed Apr 06 2022 - 15:33:43 EST

Next message: Michael Straube: "[PATCH v2 0/5] staging: r8188eu: use round_up()"
Previous message: kernel test robot: "[ammarfaizi2-block:robh/linux/dt/pop-pci-nodes 2/2] drivers/pci/of.c:84: undefined reference to `.of_attach_node'"
In reply to: Athira Rajeev: "[PATCH v2 0/4] Fix perf bench numa, futex and epoll to work with machines having #CPUs > 1K"
Next in thread: Athira Rajeev: "[PATCH v2 4/4] tools/perf: Fix perf bench numa testcase to check if CPU used to bind task is online"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

perf bench futex testcase fails on systems with CPU's
more than 1K.

Testcase: perf bench futex all
Failure snippet:
<<>>Running futex/hash benchmark...

perf: pthread_create: No such file or directory
<<>>

All the futex benchmarks ( ie hash, lock-api, requeue, wake,
wake-parallel ), pthread_create is invoked in respective bench_futex_*
function. Though the logs shows direct failure from pthread_create,
strace logs showed that actual failure is from "sched_setaffinity"
returning EINVAL (invalid argument). This happens because the default
mask size in glibc is 1024. To overcome this 1024 CPUs mask size
limitation of cpu_set_t, change the mask size using the CPU_*_S macros.

Patch addresses this by fixing all the futex benchmarks to use
CPU_ALLOC to allocate cpumask, CPU_ALLOC_SIZE for size, and
CPU_SET_S to set the mask.

Tested-by: Disha Goel <disgoel@xxxxxxxxxxxxxxxxxx>
Signed-off-by: Athira Rajeev <atrajeev@xxxxxxxxxxxxxxxxxx>
Reported-by: Disha Goel <disgoel@xxxxxxxxxxxxxxxxxx>
---
tools/perf/bench/futex-hash.c | 26 +++++++++++++++++++-------
tools/perf/bench/futex-lock-pi.c | 21 ++++++++++++++++-----
tools/perf/bench/futex-requeue.c | 21 ++++++++++++++++-----
tools/perf/bench/futex-wake-parallel.c | 21 ++++++++++++++++-----
tools/perf/bench/futex-wake.c | 22 ++++++++++++++++------
5 files changed, 83 insertions(+), 28 deletions(-)

diff --git a/tools/perf/bench/futex-hash.c b/tools/perf/bench/futex-hash.c
index 9627b6ab8670..dfce64e551e2 100644
--- a/tools/perf/bench/futex-hash.c
+++ b/tools/perf/bench/futex-hash.c
@@ -122,12 +122,14 @@ static void print_summary(void)
int bench_futex_hash(int argc, const char **argv)
{
int ret = 0;
- cpu_set_t cpuset;
+ cpu_set_t *cpuset;
struct sigaction act;
unsigned int i;
pthread_attr_t thread_attr;
struct worker *worker = NULL;
struct perf_cpu_map *cpu;
+ int nrcpus;
+ size_t size;

argc = parse_options(argc, argv, options, bench_futex_hash_usage, 0);
if (argc) {
@@ -170,25 +172,35 @@ int bench_futex_hash(int argc, const char **argv)
threads_starting = params.nthreads;
pthread_attr_init(&thread_attr);
gettimeofday(&bench__start, NULL);
+
+ nrcpus = perf_cpu_map__nr(cpu);
+ cpuset = CPU_ALLOC(nrcpus);
+ BUG_ON(!cpuset);
+ size = CPU_ALLOC_SIZE(nrcpus);
+
for (i = 0; i < params.nthreads; i++) {
worker[i].tid = i;
worker[i].futex = calloc(params.nfutexes, sizeof(*worker[i].futex));
if (!worker[i].futex)
goto errmem;

- CPU_ZERO(&cpuset);
- CPU_SET(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, &cpuset);
+ CPU_ZERO_S(size, cpuset);

- ret = pthread_attr_setaffinity_np(&thread_attr, sizeof(cpu_set_t), &cpuset);
- if (ret)
+ CPU_SET_S(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, size, cpuset);
+ ret = pthread_attr_setaffinity_np(&thread_attr, size, cpuset);
+ if (ret) {
+ CPU_FREE(cpuset);
err(EXIT_FAILURE, "pthread_attr_setaffinity_np");
-
+ }
ret = pthread_create(&worker[i].thread, &thread_attr, workerfn,
(void *)(struct worker *) &worker[i]);
- if (ret)
+ if (ret) {
+ CPU_FREE(cpuset);
err(EXIT_FAILURE, "pthread_create");
+ }

}
+ CPU_FREE(cpuset);
pthread_attr_destroy(&thread_attr);

pthread_mutex_lock(&thread_lock);
diff --git a/tools/perf/bench/futex-lock-pi.c b/tools/perf/bench/futex-lock-pi.c
index a512a320df74..61c3bb80d4cf 100644
--- a/tools/perf/bench/futex-lock-pi.c
+++ b/tools/perf/bench/futex-lock-pi.c
@@ -120,11 +120,17 @@ static void *workerfn(void *arg)
static void create_threads(struct worker *w, pthread_attr_t thread_attr,
struct perf_cpu_map *cpu)
{
- cpu_set_t cpuset;
+ cpu_set_t *cpuset;
unsigned int i;
+ int nrcpus = perf_cpu_map__nr(cpu);
+ size_t size;

threads_starting = params.nthreads;

+ cpuset = CPU_ALLOC(nrcpus);
+ BUG_ON(!cpuset);
+ size = CPU_ALLOC_SIZE(nrcpus);
+
for (i = 0; i < params.nthreads; i++) {
worker[i].tid = i;

@@ -135,15 +141,20 @@ static void create_threads(struct worker *w, pthread_attr_t thread_attr,
} else
worker[i].futex = &global_futex;

- CPU_ZERO(&cpuset);
- CPU_SET(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, &cpuset);
+ CPU_ZERO_S(size, cpuset);
+ CPU_SET_S(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, size, cpuset);

- if (pthread_attr_setaffinity_np(&thread_attr, sizeof(cpu_set_t), &cpuset))
+ if (pthread_attr_setaffinity_np(&thread_attr, size, cpuset)) {
+ CPU_FREE(cpuset);
err(EXIT_FAILURE, "pthread_attr_setaffinity_np");
+ }

- if (pthread_create(&w[i].thread, &thread_attr, workerfn, &worker[i]))
+ if (pthread_create(&w[i].thread, &thread_attr, workerfn, &worker[i])) {
+ CPU_FREE(cpuset);
err(EXIT_FAILURE, "pthread_create");
+ }
}
+ CPU_FREE(cpuset);
}

int bench_futex_lock_pi(int argc, const char **argv)
diff --git a/tools/perf/bench/futex-requeue.c b/tools/perf/bench/futex-requeue.c
index aca47ce8b1e7..2cb013f7ffe5 100644
--- a/tools/perf/bench/futex-requeue.c
+++ b/tools/perf/bench/futex-requeue.c
@@ -123,22 +123,33 @@ static void *workerfn(void *arg __maybe_unused)
static void block_threads(pthread_t *w,
pthread_attr_t thread_attr, struct perf_cpu_map *cpu)
{
- cpu_set_t cpuset;
+ cpu_set_t *cpuset;
unsigned int i;
+ int nrcpus = perf_cpu_map__nr(cpu);
+ size_t size;

threads_starting = params.nthreads;

+ cpuset = CPU_ALLOC(nrcpus);
+ BUG_ON(!cpuset);
+ size = CPU_ALLOC_SIZE(nrcpus);
+
/* create and block all threads */
for (i = 0; i < params.nthreads; i++) {
- CPU_ZERO(&cpuset);
- CPU_SET(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, &cpuset);
+ CPU_ZERO_S(size, cpuset);
+ CPU_SET_S(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, size, cpuset);

- if (pthread_attr_setaffinity_np(&thread_attr, sizeof(cpu_set_t), &cpuset))
+ if (pthread_attr_setaffinity_np(&thread_attr, size, cpuset)) {
+ CPU_FREE(cpuset);
err(EXIT_FAILURE, "pthread_attr_setaffinity_np");
+ }

- if (pthread_create(&w[i], &thread_attr, workerfn, NULL))
+ if (pthread_create(&w[i], &thread_attr, workerfn, NULL)) {
+ CPU_FREE(cpuset);
err(EXIT_FAILURE, "pthread_create");
+ }
}
+ CPU_FREE(cpuset);
}

static void toggle_done(int sig __maybe_unused,
diff --git a/tools/perf/bench/futex-wake-parallel.c b/tools/perf/bench/futex-wake-parallel.c
index 888ee6037945..efa5070a5eb3 100644
--- a/tools/perf/bench/futex-wake-parallel.c
+++ b/tools/perf/bench/futex-wake-parallel.c
@@ -144,22 +144,33 @@ static void *blocked_workerfn(void *arg __maybe_unused)
static void block_threads(pthread_t *w, pthread_attr_t thread_attr,
struct perf_cpu_map *cpu)
{
- cpu_set_t cpuset;
+ cpu_set_t *cpuset;
unsigned int i;
+ int nrcpus = perf_cpu_map__nr(cpu);
+ size_t size;

threads_starting = params.nthreads;

+ cpuset = CPU_ALLOC(nrcpus);
+ BUG_ON(!cpuset);
+ size = CPU_ALLOC_SIZE(nrcpus);
+
/* create and block all threads */
for (i = 0; i < params.nthreads; i++) {
- CPU_ZERO(&cpuset);
- CPU_SET(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, &cpuset);
+ CPU_ZERO_S(size, cpuset);
+ CPU_SET_S(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, size, cpuset);

- if (pthread_attr_setaffinity_np(&thread_attr, sizeof(cpu_set_t), &cpuset))
+ if (pthread_attr_setaffinity_np(&thread_attr, size, cpuset)) {
+ CPU_FREE(cpuset);
err(EXIT_FAILURE, "pthread_attr_setaffinity_np");
+ }

- if (pthread_create(&w[i], &thread_attr, blocked_workerfn, NULL))
+ if (pthread_create(&w[i], &thread_attr, blocked_workerfn, NULL)) {
+ CPU_FREE(cpuset);
err(EXIT_FAILURE, "pthread_create");
+ }
}
+ CPU_FREE(cpuset);
}

static void print_run(struct thread_data *waking_worker, unsigned int run_num)
diff --git a/tools/perf/bench/futex-wake.c b/tools/perf/bench/futex-wake.c
index aa82db51c0ab..3a10f54900c1 100644
--- a/tools/perf/bench/futex-wake.c
+++ b/tools/perf/bench/futex-wake.c
@@ -97,22 +97,32 @@ static void print_summary(void)
static void block_threads(pthread_t *w,
pthread_attr_t thread_attr, struct perf_cpu_map *cpu)
{
- cpu_set_t cpuset;
+ cpu_set_t *cpuset;
unsigned int i;
-
+ size_t size;
+ int nrcpus = perf_cpu_map__nr(cpu);
threads_starting = params.nthreads;

+ cpuset = CPU_ALLOC(nrcpus);
+ BUG_ON(!cpuset);
+ size = CPU_ALLOC_SIZE(nrcpus);
+
/* create and block all threads */
for (i = 0; i < params.nthreads; i++) {
- CPU_ZERO(&cpuset);
- CPU_SET(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, &cpuset);
+ CPU_ZERO_S(size, cpuset);
+ CPU_SET_S(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, size, cpuset);

- if (pthread_attr_setaffinity_np(&thread_attr, sizeof(cpu_set_t), &cpuset))
+ if (pthread_attr_setaffinity_np(&thread_attr, size, cpuset)) {
+ CPU_FREE(cpuset);
err(EXIT_FAILURE, "pthread_attr_setaffinity_np");
+ }

- if (pthread_create(&w[i], &thread_attr, workerfn, NULL))
+ if (pthread_create(&w[i], &thread_attr, workerfn, NULL)) {
+ CPU_FREE(cpuset);
err(EXIT_FAILURE, "pthread_create");
+ }
}
+ CPU_FREE(cpuset);
}

static void toggle_done(int sig __maybe_unused,
--
2.35.1

Next message: Michael Straube: "[PATCH v2 0/5] staging: r8188eu: use round_up()"
Previous message: kernel test robot: "[ammarfaizi2-block:robh/linux/dt/pop-pci-nodes 2/2] drivers/pci/of.c:84: undefined reference to `.of_attach_node'"
In reply to: Athira Rajeev: "[PATCH v2 0/4] Fix perf bench numa, futex and epoll to work with machines having #CPUs > 1K"
Next in thread: Athira Rajeev: "[PATCH v2 4/4] tools/perf: Fix perf bench numa testcase to check if CPU used to bind task is online"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]