Following is an example. perf_example.c is a program modified from the perf_event_open man page to measure instruction count for a printf() call:

#include <stdlib.h> 
#include <stdio.h> 
#include <unistd.h> 
#include <string.h> 
#include <sys/ioctl.h> 
#include <linux/perf_event.h> 
#include <asm/unistd.h> 
  
static long 
perf_event_open(struct perf_event_attr *hw_event, pid_t pid, 
                int cpu, int group_fd, unsigned long flags) 
{ 
    int ret; 
 
    ret = syscall(__NR_perf_event_open, hw_event, pid, cpu, 
                   group_fd, flags); 
    return ret; 
} 
 
int 
main(int argc, char **argv) 
{ 
    struct perf_event_attr pe; 
    long long count; 
    int fd; 
 
    memset(&pe, 0, sizeof(struct perf_event_attr)); 
    pe.type = PERF_TYPE_HARDWARE; 
    pe.size = sizeof(struct perf_event_attr); 
    pe.config = PERF_COUNT_HW_INSTRUCTIONS; 
    pe.disabled = 1; 
    
    fd = perf_event_open(&pe, 0, -1, -1, 0); 
    if (fd == -1) { 
       fprintf(stderr, "Error opening leader %llx
", pe.config); 
       exit(EXIT_FAILURE); 
    } 
 
    ioctl(fd, PERF_EVENT_IOC_RESET, 0); 
    ioctl(fd, PERF_EVENT_IOC_ENABLE, 0); 
 
    printf("Measuring instruction count for this printf
"); 
 
    ioctl(fd, PERF_EVENT_IOC_DISABLE, 0); 
    read(fd, &count, sizeof(long long)); 
 
    printf("Used %lld instructions
", count); 
 
    close(fd); 
 
    return 0; 
} 

For compiling this program externally, we can use the following commands:

$ source /opt/poky/2.4/environment-setup-cortexa9hf-neon-poky-linux-gnueabi
$ ${CC} ${CFLAGS} ${LDFLAGS} perf_example.c -o perf_example  

After copying the binary to your target, you can then execute it with the help of the following steps:

# chmod a+x perf_example
# ./perf_example
Measuring instruction count for this printf
Used 0 instructions  

Obviously, using zero instructions for the printf() call can't be correct. Looking into possible causes, we find a documented erratum (ERR006259) on i.MX6 processors that states that in order for the PMU to be used, the SoC needs to receive at least 4 JTAG clock cycles after power on reset.

Rerunning the example with the JTAG connected and running:

# ./perf_example
Measuring instruction count for this printf
Used 3977 instructions