You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Mark Rowe <ro...@farpointer.net> on 2004/03/11 21:08:28 UTC
apache 1.3.29 pool problem with clearing pools in a loop
(allocates 100's of MB instead of 100KB)
Check out this pool test,
Using apache 1.3.29
I defined tenbyte_string as the
constant string "0123456789"
which is 10 bytes long.
Here is a test where I used
ap_pstrcat( ... ) to
in a loop 10,000 times
so the resulting string
should only take up
10 * 10,000 = 100,000 bytes.
It took nearly 400 MB in redhat
7.2 and over 1 GB in windows XP
I ran this code on apache 1.3.29
redhat version 7.2 and it used nearly
400 Mega Bytes of RAM, with physical
RAM of only 128 MB
When I ran it on windows XP built with
MS Visual Studio.NET and it used
over 1 Giga Bytes of RAM on a
machine with only 384 MB of physical RAM
in windows it persisted with over 86 MB
of Mem Usage according to Task Manager and
over 1 Gigabyte in VM Size according to Task Manager
Windows XP never released either the Mem Usage or
VM Size. Eventually I got a warning from XP saying
I should increase my RAM
I simplified and altered the mod_example source code from
the apache 1.3.29 distribution for the test.
Does anybody know why this little loop uses so much
RAM? Is there a problem with the way the code is
written? Is it possible that apache has a pool problem?
Does this repeat in Apache version 2 pools?
The source code follows:
-------------------------------------------------------------------------------------------
#define tenbyte_string "0123456789"
static int example_handler(request_rec *r)
{
int i;
pool *subp1, *subp2, *p = r->pool;
char *s = "";
r->content_type = "text/html";
ap_soft_timeout("send example call trace", r);
ap_send_http_header(r);
/*
* If we're only supposed to send header information (HEAD request), we're
* already there.
*/
if (r->header_only) {
ap_kill_timeout(r);
return OK;
}
//ap_rputs(DOCTYPE_HTML_3_2, r);
ap_rprintf(r, "<h2>hello world</h2>");
subp1 = ap_make_sub_pool(p);
subp2 = ap_make_sub_pool(p);
for (i = 0; i < 10000; i++) {
ap_clear_pool(subp1);
s = ap_pstrcat(subp1, s, tenbyte_string, NULL);
ap_clear_pool(subp2);
s = ap_pstrdup(subp2, s);
}
ap_destroy_pool(subp1);
ap_destroy_pool(subp2);
ap_kill_timeout(r);
return OK;
}
-------------------------------------------------------------------------------------------
Let me know what everyone thinks.
Mark R. Rowe, MSEE