summaryrefslogtreecommitdiffstats
path: root/tools/perf/util/affinity.c
diff options
context:
space:
mode:
authorDaniel Baumann <daniel.baumann@progress-linux.org>2024-04-11 08:27:49 +0000
committerDaniel Baumann <daniel.baumann@progress-linux.org>2024-04-11 08:27:49 +0000
commitace9429bb58fd418f0c81d4c2835699bddf6bde6 (patch)
treeb2d64bc10158fdd5497876388cd68142ca374ed3 /tools/perf/util/affinity.c
parentInitial commit. (diff)
downloadlinux-ace9429bb58fd418f0c81d4c2835699bddf6bde6.tar.xz
linux-ace9429bb58fd418f0c81d4c2835699bddf6bde6.zip
Adding upstream version 6.6.15.upstream/6.6.15
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'tools/perf/util/affinity.c')
-rw-r--r--tools/perf/util/affinity.c85
1 files changed, 85 insertions, 0 deletions
diff --git a/tools/perf/util/affinity.c b/tools/perf/util/affinity.c
new file mode 100644
index 0000000000..38dc4524b7
--- /dev/null
+++ b/tools/perf/util/affinity.c
@@ -0,0 +1,85 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Manage affinity to optimize IPIs inside the kernel perf API. */
+#define _GNU_SOURCE 1
+#include <sched.h>
+#include <stdlib.h>
+#include <linux/bitmap.h>
+#include <linux/zalloc.h>
+#include "perf.h"
+#include "cpumap.h"
+#include "affinity.h"
+
+static int get_cpu_set_size(void)
+{
+ int sz = cpu__max_cpu().cpu + 8 - 1;
+ /*
+ * sched_getaffinity doesn't like masks smaller than the kernel.
+ * Hopefully that's big enough.
+ */
+ if (sz < 4096)
+ sz = 4096;
+ return sz / 8;
+}
+
+int affinity__setup(struct affinity *a)
+{
+ int cpu_set_size = get_cpu_set_size();
+
+ a->orig_cpus = bitmap_zalloc(cpu_set_size * 8);
+ if (!a->orig_cpus)
+ return -1;
+ sched_getaffinity(0, cpu_set_size, (cpu_set_t *)a->orig_cpus);
+ a->sched_cpus = bitmap_zalloc(cpu_set_size * 8);
+ if (!a->sched_cpus) {
+ zfree(&a->orig_cpus);
+ return -1;
+ }
+ bitmap_zero((unsigned long *)a->sched_cpus, cpu_set_size);
+ a->changed = false;
+ return 0;
+}
+
+/*
+ * perf_event_open does an IPI internally to the target CPU.
+ * It is more efficient to change perf's affinity to the target
+ * CPU and then set up all events on that CPU, so we amortize
+ * CPU communication.
+ */
+void affinity__set(struct affinity *a, int cpu)
+{
+ int cpu_set_size = get_cpu_set_size();
+
+ /*
+ * Return:
+ * - if cpu is -1
+ * - restrict out of bound access to sched_cpus
+ */
+ if (cpu == -1 || ((cpu >= (cpu_set_size * 8))))
+ return;
+
+ a->changed = true;
+ __set_bit(cpu, a->sched_cpus);
+ /*
+ * We ignore errors because affinity is just an optimization.
+ * This could happen for example with isolated CPUs or cpusets.
+ * In this case the IPIs inside the kernel's perf API still work.
+ */
+ sched_setaffinity(0, cpu_set_size, (cpu_set_t *)a->sched_cpus);
+ __clear_bit(cpu, a->sched_cpus);
+}
+
+static void __affinity__cleanup(struct affinity *a)
+{
+ int cpu_set_size = get_cpu_set_size();
+
+ if (a->changed)
+ sched_setaffinity(0, cpu_set_size, (cpu_set_t *)a->orig_cpus);
+ zfree(&a->sched_cpus);
+ zfree(&a->orig_cpus);
+}
+
+void affinity__cleanup(struct affinity *a)
+{
+ if (a != NULL)
+ __affinity__cleanup(a);
+}