From owner-freebsd-hackers  Thu Apr 29 14:14:15 1999
Delivered-To: freebsd-hackers@freebsd.org
Received: from home.dragondata.com (home.dragondata.com [204.137.237.2])
	by hub.freebsd.org (Postfix) with ESMTP id B617214DB2
	for <hackers@freebsd.org>; Thu, 29 Apr 1999 14:14:11 -0700 (PDT)
	(envelope-from toasty@home.dragondata.com)
Received: (from toasty@localhost)
	by home.dragondata.com (8.9.2/8.9.2) id QAA00336
	for hackers@freebsd.org; Thu, 29 Apr 1999 16:14:09 -0500 (CDT)
From: Kevin Day <toasty@home.dragondata.com>
Message-Id: <199904292114.QAA00336@home.dragondata.com>
Subject: mmaping /dev/mem then spawning children
To: hackers@freebsd.org
Date: Thu, 29 Apr 1999 16:14:09 -0500 (CDT)
X-Mailer: ELM [version 2.4ME+ PL43 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-hackers@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG


I'm completely at a loss as to why this happens, but it just sprang up when
we upgraded from 3.0-RELEASE to 3.1-RELEASE.

I have a process that mmap's a device. (Our own custom driver). It then
spawns some children that don't really need access to it. However,
top/ps/etc show that the child is eating 32760 megabytes of ram.

This also seems to happen if you mmap /dev/mem, or a few other devices. Am I
just doing something stupid here, or is this an honest bug?

Here's sample code to reproduce this. Compile them both, then run '1'.


1.c:
#include <stdio.h>
#include <unistd.h>
#include <fcntl.h>
#include <sys/errno.h>
#include <sys/types.h>
#include <sys/mman.h>

void main(void) {
 int k;
 char *low;
 int devmem;

    devmem = open("/dev/mem",O_RDWR);   
    low1mb = (char *)mmap(0, 0x800000, (PROT_READ | PROT_WRITE), MAP_INHERIT | MAP_SHARED, devmem, 0);

   printf("Starting 2...\n");      
   if (rfork(RFCFDG|RFPROC)) {             
        usleep(100);    
   } else {
        k = execl("./2", "2", (char *) 0);              
        if (k) {
                printf("io returned %d errno %d\n",k, errno);           
        }
   _exit(0);       
   }

}




2.c:

void main(void) { 
     while(1)   
        sleep(10000);
}


I run ./1



top shows:

  PID USERNAME PRI NICE  SIZE    RES STATE    TIME   WCPU    CPU COMMAND
  899 root      10   0   748K 32760M nanslp   0:00  0.00%  0.00% 2

ps shows:

root     899  0.0 53622.9   748 -7960  p1  I     6:14AM   0:00.00 2                    0   899     1   0  10  0   748 -7960 nanslp I     p1    0:00.00 2

-7960k.... which is about 8M (from the mmap) minus how much '2' should really be taking up.




Is this a real bug?

Kevin


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message