Re: [PATCH 2.6.11-rc4-mm1] connector: Add a fork connector

[Evgeniy Polyakov <johnpol@2ka.mipt.ru>]

[-- Attachment #1: Type: text/plain, Size: 1437 bytes --]

Simple program to test fork() performance.
#include <sys/signal.h>
#include <sys/time.h>

int main(int argc, char *argv[])
{
        int pid;
        int i = 0, max = 100000;
        struct timeval tv0, tv1;
        struct timezone tz;
        long diff;

        if (argc >= 2)
                max = atoi(argv[1]);

        signal(SIGCHLD, SIG_IGN);

        gettimeofday(&tv0, &tz);
        while (i++ < max) {
                pid = fork();
                if (pid == 0) {
                        sleep(1);
                        exit (0);
                }
        }
        gettimeofday(&tv1, &tz);

        diff = (tv1.tv_sec - tv0.tv_sec)*1000000 + (tv1.tv_usec - tv0.tv_usec);

        printf("Average per process fork+exit time is %ld usecs [diff=%lu, max=%d].\n", diff/max, diff, max);
        return 0;
}

Creating 10k forks 100 times.
Results on 2-way SMP(1+1HT) Xeon for one fork()+exit():

2.6.11-rc4-mm1                                   494 usec
2.6.11-rc4-mm1-fork-connector-no_userspace       509 usec
2.6.11-rc4-mm1-fork-connector-userspace          520 usec

5% fork() degradation(connector with userspace vs. vanilla) with fork() connector.
On my test system global fork lock does not cost anything
(tested both with and without userspace listener), but it is only 2-way(pseudo).

-- 
        Evgeniy Polyakov

Crash is better than data corruption -- Arthur Grabowski

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]