2001-07-06 22:59:22

by Sasha Pachev

[permalink] [raw]
Subject: memory allocation mystery

Hello,

I have been investigating kernel behavior ( I am running 2.4.3) in out of
memory conditions with swap completely disabled and discovered a rather
interesting behavior. If you run the following code:

#include <stdio.h>
#include <stdlib.h>

#define LEAK_BLOCK (1024*1024)
#define MB (1024*1024)

int main()
{
unsigned long total = 0;
for (;;)
{
char* p, *p_end;
if(!(p=malloc(LEAK_BLOCK)))
{
fprintf(stderr, "malloc() failed\n");
exit(1);
}
p_end = p + LEAK_BLOCK;
while(p < p_end)
*p++ = 0;
total += LEAK_BLOCK;
printf("Allocated %d MB\n", total/MB);
}

return 0;
}


the process eventually gets killed by the kernel, rather than getting an
error from malloc() as you would logically expect

I have straced the process and see just a bunch of old_mmap() calls like this:

old_mmap(NULL, 1052672, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0)
= 0x46b6a000

( in addition to writes to stdout, of course). So it looks like old_mmap()
never returns an error.

Can somebody explain this behavior? To me it looks like a bug...

--
MySQL Development Team
For technical support contracts, visit https://order.mysql.com/
__ ___ ___ ____ __
/ |/ /_ __/ __/ __ \/ / Sasha Pachev <[email protected]>
/ /|_/ / // /\ \/ /_/ / /__ MySQL AB, http://www.mysql.com/
/_/ /_/\_, /___/\___\_\___/ Provo, Utah, USA
<___/