/*
* syst.c - example of a simple system wide monitoring program
*
* Copyright (c) 2002-2006 Hewlett-Packard Development Company, L.P.
* Contributed by Stephane Eranian <eranian@hpl.hp.com>
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
* of the Software, and to permit persons to whom the Software is furnished to do so,
* subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,
* INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A
* PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
* HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
* CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE
* OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*/
#include <sys/types.h>
#include <inttypes.h>
#include <stdio.h>
#include <stdlib.h>
#include <stdarg.h>
#include <errno.h>
#include <unistd.h>
#include <string.h>
#include <sched.h>
#include <perfmon/pfmlib.h>
#include <perfmon/perfmon.h>
#include "detect_pmcs.h"
#define NUM_PMCS PFMLIB_MAX_PMCS
#define NUM_PMDS PFMLIB_MAX_PMDS
#define MAX_EVT_NAME_LEN 128
static void fatal_error(char *fmt,...) __attribute__((noreturn));
static void
fatal_error(char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
vfprintf(stderr, fmt, ap);
va_end(ap);
exit(1);
}
#define MAX_CPUS 2048
int
pin_cpu(pid_t pid, unsigned int cpu)
{
cpu_set_t my_set;
CPU_ZERO(&my_set);
if (cpu >= MAX_CPUS)
fatal_error("this program supports only up to %d CPUs\n", MAX_CPUS);
CPU_SET(cpu, &my_set);
return sched_setaffinity(pid, sizeof(cpu_set_t), &my_set);
}
int
main(int argc, char **argv)
{
char **p;
pfarg_pmc_t pc[NUM_PMCS];
pfarg_pmd_t pd[NUM_PMDS];
uint64_t pdo[NUM_PMDS];
pfarg_ctx_t ctx[1];
pfarg_load_t load_args;
pfmlib_input_param_t inp;
pfmlib_output_param_t outp;
pfmlib_options_t pfmlib_options;
unsigned int which_cpu;
int ret, ctx_fd;
unsigned int i, l;
unsigned int num_counters;
char name[MAX_EVT_NAME_LEN];
/*
* pass options to library (optional)
*/
memset(&pfmlib_options, 0, sizeof(pfmlib_options));
pfmlib_options.pfm_debug = 0; /* set to 1 for debug */
pfm_set_options(&pfmlib_options);
/*
* Initialize pfm library (required before we can use it)
*/
ret = pfm_initialize();
if (ret != PFMLIB_SUCCESS)
fatal_error("Cannot initialize library: %s\n", pfm_strerror(ret));
pfm_get_num_counters(&num_counters);
memset(pc, 0, sizeof(pc));
memset(pd, 0, sizeof(pd));
memset(pdo, 0, sizeof(pdo));
memset(ctx, 0, sizeof(ctx));
memset(&inp,0, sizeof(inp));
memset(&outp,0, sizeof(outp));
memset(&load_args,0, sizeof(load_args));
/*
* be nice to user!
*/
if (argc > 1) {
p = argv+1;
for (i=0; *p ; i++, p++) {
if (pfm_find_full_event(*p, &inp.pfp_events[i]) != PFMLIB_SUCCESS) {
fatal_error("Cannot find %s event\n", *p);
}
}
} else {
if (pfm_get_cycle_event(&inp.pfp_events[0]) != PFMLIB_SUCCESS) {
fatal_error("cannot find cycle event\n");
}
if (pfm_get_inst_retired_event(&inp.pfp_events[1]) != PFMLIB_SUCCESS) {
fatal_error("cannot find inst retired event\n");
}
i = 2;
}
/*
* set the privilege mode:
* PFM_PLM3 : user level
* PFM_PLM0 : kernel level
*/
inp.pfp_dfl_plm = PFM_PLM3|PFM_PLM0;
if (i > num_counters) {
i = num_counters;
printf("too many events provided (max=%d events), using first %d event(s)\n", num_counters, i);
}
/*
* how many counters we use
*/
inp.pfp_event_count = i;
/*
* indicate we are using the monitors for a system-wide session.
* This may impact the way the library sets up the PMC values.
*/
inp.pfp_flags = PFMLIB_PFP_SYSTEMWIDE;
/*
* In system wide mode, the perfmon context cannot be inherited.
* Also in this mode, we cannot use the blocking form of user level notification.
*/
ctx[0].ctx_flags = PFM_FL_SYSTEM_WIDE;
/*
* pick a random CPU. Assumes CPU are numbered with no holes
*/
srandom(getpid());
which_cpu = random() % sysconf(_SC_NPROCESSORS_ONLN);
/*
* The monitored CPU is determined by the processor core
* executing the PFM_LOAD_CONTEXT command. To ensure, we
* measure the right core, we pin the thread before making
* the call.
*/
ret = pin_cpu(getpid(), which_cpu);
if (ret == -1)
fatal_error("cannot set affinity to CPU%d: %s\n", which_cpu, strerror(errno));
/*
* after the call the task is pinned to which_cpu
*/
/*
* now create the context for self monitoring/per-task
*/
ctx_fd = pfm_create_context(ctx, NULL, NULL, 0);
if (ctx_fd == -1) {
if (errno == ENOSYS) {
fatal_error("Your kernel does not have performance monitoring support!\n");
}
fatal_error("Can't create PFM context %s\n", strerror(errno));
}
/*
* build the pfp_unavail_pmcs bitmask by looking
* at what perfmon has available. It is not always
* the case that all PMU registers are actually available
* to applications. For instance, on IA-32 platforms, some
* registers may be reserved for the NMI watchdog timer.
*
* With this bitmap, the library knows which registers NOT to
* use. Of source, it is possible that no valid assignement may
* be possible if certina PMU registers are not available.
*/
detect_unavail_pmcs(ctx_fd, &inp.pfp_unavail_pmcs);
/*
* let the library figure out the values for the PMCS
*/
if ((ret=pfm_dispatch_events(&inp, NULL, &outp, NULL)) != PFMLIB_SUCCESS)
fatal_error("cannot configure events: %s\n", pfm_strerror(ret));
/*
* Now prepare the argument to initialize the PMDs and PMCS.
* We use pfp_pmc_count to determine the number of PMC to intialize.
* We use pfp_pmd_count to determine the number of PMD to initialize.
* Some events/features may cause extra PMCs to be used, leading to:
* - pfp_pmc_count may be >= pfp_event_count
* - pfp_pmd_count may be >= pfp_event_count
*/
for (i=0; i < outp.pfp_pmc_count; i++) {
pc[i].reg_num = outp.pfp_pmcs[i].reg_num;
pc[i].reg_value = outp.pfp_pmcs[i].reg_value;
}
for (i=0; i < outp.pfp_pmd_count; i++)
pd[i].reg_num = outp.pfp_pmds[i].reg_num;
/*
* Now program the registers
*/
if (pfm_write_pmcs(ctx_fd, pc, outp.pfp_pmc_count) == -1)
fatal_error("pfm_write_pmcs error errno %d\n",errno);
if (pfm_write_pmds(ctx_fd, pd, outp.pfp_pmd_count) == -1)
fatal_error("pfm_write_pmds error errno %d\n",errno);
/*
* in system-wide mode, this field must provide the CPU the caller wants
* to monitor. The kernel checks and if calling from the wrong CPU, the
* call fails. The affinity is not affected.
*/
load_args.load_pid = which_cpu;
if (pfm_load_context(ctx_fd, &load_args) == -1)
fatal_error("pfm_load_context error errno %d\n",errno);
printf("<monitoring started on CPU%d, press CTRL-C to quit before 20s time limit>\n", which_cpu);
for(l=0; l < 10; l++) {
/*
* start monitoring
*/
if (pfm_start(ctx_fd, NULL) == -1)
fatal_error("pfm_start error errno %d\n",errno);
sleep(2);
/*
* stop monitoring.
* changed at the user level.
*/
if (pfm_stop(ctx_fd) == -1)
fatal_error("pfm_stop error errno %d\n",errno);
/*
* read the results
*/
if (pfm_read_pmds(ctx_fd, pd, inp.pfp_event_count) == -1)
fatal_error( "pfm_read_pmds error errno %d\n",errno);
/*
* print the results
*/
puts("------------------------");
for (i=0; i < inp.pfp_event_count; i++) {
pfm_get_full_event_name(&inp.pfp_events[i], name, MAX_EVT_NAME_LEN);
printf("CPU%-2d PMD%-3u raw=%-20"PRIu64" delta=%-20"PRIu64" %s\n",
which_cpu,
pd[i].reg_num,
pd[i].reg_value,
pd[i].reg_value - pdo[i],
name);
pdo[i] = pd[i].reg_value;
}
}
/*
* destroy everything
*/
close(ctx_fd);
return 0;
}