/* Copyright (c) 2024 Barcelona Supercomputing Center (BSC) * SPDX-License-Identifier: GPL-3.0-or-later */ #include "common.h" #include "compat.h" #include #include static long fib(long index) { long a, b; if (index <= 1) return index; #pragma oss task shared(a) label("fibonacci") a = fib(index-1); #pragma oss task shared(b) label("fibonacci") b = fib(index-2); #pragma oss taskwait return a + b; } int main(void) { FILE *f = fopen("/proc/sys/kernel/perf_event_paranoid", "r"); if (f == NULL) die("cannot open /proc/sys/kernel/perf_event_paranoid:"); char buf[16] = {0}; if (fread(buf, 1, 16, f) <= 0) die("cannot read /proc/sys/kernel/perf_event_paranoid:"); fclose(f); /* We need the value 1 or less: * Access to performance monitoring and observability operations is limited. * Consider adjusting /proc/sys/kernel/perf_event_paranoid setting to open * access to performance monitoring and observability operations for processes * without CAP_PERFMON, CAP_SYS_PTRACE or CAP_SYS_ADMIN Linux capability. * More information can be found at 'Perf events and tool security' document: * https://www.kernel.org/doc/html/latest/admin-guide/perf-security.html * perf_event_paranoid setting is 3: * -1: Allow use of (almost) all events by all users * Ignore mlock limit after perf_event_mlock_kb without CAP_IPC_LOCK * >= 0: Disallow raw and ftrace function tracepoint access * >= 1: Disallow CPU event access * >= 2: Disallow kernel profiling * To make the adjusted perf_event_paranoid setting permanent preserve it * in /etc/sysctl.conf (e.g. kernel.perf_event_paranoid = ) */ int level = atoi(buf); if (level > 1) die("requires perf_event_paranoid = 1 or less"); for (int i = 0; i < 30; i++) fib(5); #pragma oss taskwait return 0; }