freebsd-dev/usr.bin/systat/mbufs.c

227 lines
5.6 KiB
C
Raw Normal View History

1994-05-27 12:33:43 +00:00
/*-
* Copyright (c) 1980, 1992, 1993
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#ifndef lint
static char sccsid[] = "@(#)mbufs.c 8.1 (Berkeley) 6/6/93";
static const char rcsid[] =
1999-08-28 01:08:13 +00:00
"$FreeBSD$";
1994-05-27 12:33:43 +00:00
#endif /* not lint */
#include <sys/param.h>
#include <sys/types.h>
#include <sys/mbuf.h>
#include <sys/sysctl.h>
1994-05-27 12:33:43 +00:00
#include <errno.h>
1994-05-27 12:33:43 +00:00
#include <stdlib.h>
#include <string.h>
#include <paths.h>
#include "systat.h"
#include "extern.h"
Introduce numerous SMP friendly changes to the mbuf allocator. Namely, introduce a modified allocation mechanism for mbufs and mbuf clusters; one which can scale under SMP and which offers the possibility of resource reclamation to be implemented in the future. Notable advantages: o Reduce contention for SMP by offering per-CPU pools and locks. o Better use of data cache due to per-CPU pools. o Much less code cache pollution due to excessively large allocation macros. o Framework for `grouping' objects from same page together so as to be able to possibly free wired-down pages back to the system if they are no longer needed by the network stacks. Additional things changed with this addition: - Moved some mbuf specific declarations and initializations from sys/conf/param.c into mbuf-specific code where they belong. - m_getclr() has been renamed to m_get_clrd() because the old name is really confusing. m_getclr() HAS been preserved though and is defined to the new name. No tree sweep has been done "to change the interface," as the old name will continue to be supported and is not depracated. The change was merely done because m_getclr() sounds too much like "m_get a cluster." - TEMPORARILY disabled mbtypes statistics displaying in netstat(1) and systat(1) (see TODO below). - Fixed systat(1) to display number of "free mbufs" based on new per-CPU stat structures. - Fixed netstat(1) to display new per-CPU stats based on sysctl-exported per-CPU stat structures. All infos are fetched via sysctl. TODO (in order of priority): - Re-enable mbtypes statistics in both netstat(1) and systat(1) after introducing an SMP friendly way to collect the mbtypes stats under the already introduced per-CPU locks (i.e. hopefully don't use atomic() - it seems too costly for a mere stat update, especially when other locks are already present). - Optionally have systat(1) display not only "total free mbufs" but also "total free mbufs per CPU pool." - Fix minor length-fetching issues in netstat(1) related to recently re-enabled option to read mbuf stats from a core file. - Move reference counters at least for mbuf clusters into an unused portion of the cluster itself, to save space and need to allocate a counter. - Look into introducing resource freeing possibly from a kproc. Reviewed by (in parts): jlemon, jake, silby, terry Tested by: jlemon (Intel & Alpha), mjacob (Intel & Alpha) Preliminary performance measurements: jlemon (and me, obviously) URL: http://people.freebsd.org/~bmilekic/mb_alloc/
2001-06-22 06:35:32 +00:00
static struct mbpstat **mbpstat;
static struct mbstat *mbstat;
static int num_objs;
static long *m_mbtypes;
static short nmbtypes;
Introduce numerous SMP friendly changes to the mbuf allocator. Namely, introduce a modified allocation mechanism for mbufs and mbuf clusters; one which can scale under SMP and which offers the possibility of resource reclamation to be implemented in the future. Notable advantages: o Reduce contention for SMP by offering per-CPU pools and locks. o Better use of data cache due to per-CPU pools. o Much less code cache pollution due to excessively large allocation macros. o Framework for `grouping' objects from same page together so as to be able to possibly free wired-down pages back to the system if they are no longer needed by the network stacks. Additional things changed with this addition: - Moved some mbuf specific declarations and initializations from sys/conf/param.c into mbuf-specific code where they belong. - m_getclr() has been renamed to m_get_clrd() because the old name is really confusing. m_getclr() HAS been preserved though and is defined to the new name. No tree sweep has been done "to change the interface," as the old name will continue to be supported and is not depracated. The change was merely done because m_getclr() sounds too much like "m_get a cluster." - TEMPORARILY disabled mbtypes statistics displaying in netstat(1) and systat(1) (see TODO below). - Fixed systat(1) to display number of "free mbufs" based on new per-CPU stat structures. - Fixed netstat(1) to display new per-CPU stats based on sysctl-exported per-CPU stat structures. All infos are fetched via sysctl. TODO (in order of priority): - Re-enable mbtypes statistics in both netstat(1) and systat(1) after introducing an SMP friendly way to collect the mbtypes stats under the already introduced per-CPU locks (i.e. hopefully don't use atomic() - it seems too costly for a mere stat update, especially when other locks are already present). - Optionally have systat(1) display not only "total free mbufs" but also "total free mbufs per CPU pool." - Fix minor length-fetching issues in netstat(1) related to recently re-enabled option to read mbuf stats from a core file. - Move reference counters at least for mbuf clusters into an unused portion of the cluster itself, to save space and need to allocate a counter. - Look into introducing resource freeing possibly from a kproc. Reviewed by (in parts): jlemon, jake, silby, terry Tested by: jlemon (Intel & Alpha), mjacob (Intel & Alpha) Preliminary performance measurements: jlemon (and me, obviously) URL: http://people.freebsd.org/~bmilekic/mb_alloc/
2001-06-22 06:35:32 +00:00
#define GENLST (num_objs - 1)
static struct mtnames {
short mt_type;
char *mt_name;
} mtnames[] = {
{ MT_DATA, "data"},
{ MT_HEADER, "headers"},
{ MT_SONAME, "socknames"},
{ MT_FTABLE, "frags"},
{ MT_CONTROL, "control"},
{ MT_OOBDATA, "oobdata"}
1994-05-27 12:33:43 +00:00
};
#define NNAMES (sizeof (mtnames) / sizeof (mtnames[0]))
WINDOW *
openmbufs()
{
return (subwin(stdscr, LINES-5-1, 0, 5, 0));
}
void
closembufs(w)
WINDOW *w;
{
if (w == NULL)
return;
wclear(w);
wrefresh(w);
delwin(w);
}
void
labelmbufs()
{
wmove(wnd, 0, 0); wclrtoeol(wnd);
mvwaddstr(wnd, 0, 10,
"/0 /5 /10 /15 /20 /25 /30 /35 /40 /45 /50 /55 /60");
}
void
showmbufs()
{
Introduce numerous SMP friendly changes to the mbuf allocator. Namely, introduce a modified allocation mechanism for mbufs and mbuf clusters; one which can scale under SMP and which offers the possibility of resource reclamation to be implemented in the future. Notable advantages: o Reduce contention for SMP by offering per-CPU pools and locks. o Better use of data cache due to per-CPU pools. o Much less code cache pollution due to excessively large allocation macros. o Framework for `grouping' objects from same page together so as to be able to possibly free wired-down pages back to the system if they are no longer needed by the network stacks. Additional things changed with this addition: - Moved some mbuf specific declarations and initializations from sys/conf/param.c into mbuf-specific code where they belong. - m_getclr() has been renamed to m_get_clrd() because the old name is really confusing. m_getclr() HAS been preserved though and is defined to the new name. No tree sweep has been done "to change the interface," as the old name will continue to be supported and is not depracated. The change was merely done because m_getclr() sounds too much like "m_get a cluster." - TEMPORARILY disabled mbtypes statistics displaying in netstat(1) and systat(1) (see TODO below). - Fixed systat(1) to display number of "free mbufs" based on new per-CPU stat structures. - Fixed netstat(1) to display new per-CPU stats based on sysctl-exported per-CPU stat structures. All infos are fetched via sysctl. TODO (in order of priority): - Re-enable mbtypes statistics in both netstat(1) and systat(1) after introducing an SMP friendly way to collect the mbtypes stats under the already introduced per-CPU locks (i.e. hopefully don't use atomic() - it seems too costly for a mere stat update, especially when other locks are already present). - Optionally have systat(1) display not only "total free mbufs" but also "total free mbufs per CPU pool." - Fix minor length-fetching issues in netstat(1) related to recently re-enabled option to read mbuf stats from a core file. - Move reference counters at least for mbuf clusters into an unused portion of the cluster itself, to save space and need to allocate a counter. - Look into introducing resource freeing possibly from a kproc. Reviewed by (in parts): jlemon, jake, silby, terry Tested by: jlemon (Intel & Alpha), mjacob (Intel & Alpha) Preliminary performance measurements: jlemon (and me, obviously) URL: http://people.freebsd.org/~bmilekic/mb_alloc/
2001-06-22 06:35:32 +00:00
int i, j, max, index;
u_long totfree;
1994-05-27 12:33:43 +00:00
char buf[10];
char *mtname;
1994-05-27 12:33:43 +00:00
totfree = mbpstat[GENLST]->mb_mbfree;
for (i = 1; i < nmbtypes; i++)
m_mbtypes[i] += mbpstat[GENLST]->mb_mbtypes[i];
for (i = 0; i < GENLST; i++) {
if (mbpstat[i]->mb_active == 0)
continue;
totfree += mbpstat[i]->mb_mbfree;
for (j = 1; j < nmbtypes; j++)
m_mbtypes[j] += mbpstat[i]->mb_mbtypes[j];
}
/*
* Print totals for different mbuf types.
*/
for (j = 0; j < wnd->_maxy; j++) {
1995-05-30 06:41:30 +00:00
max = 0, index = -1;
for (i = 0; i < wnd->_maxy; i++) {
if (i == MT_NOTMBUF)
continue;
if (i >= nmbtypes)
break;
if (m_mbtypes[i] > max) {
max = m_mbtypes[i];
1994-05-27 12:33:43 +00:00
index = i;
}
}
1994-05-27 12:33:43 +00:00
if (max == 0)
break;
mtname = NULL;
for (i = 0; i < NNAMES; i++)
if (mtnames[i].mt_type == index)
mtname = mtnames[i].mt_name;
if (mtname == NULL)
1994-05-27 12:33:43 +00:00
mvwprintw(wnd, 1+j, 0, "%10d", index);
else
mvwprintw(wnd, 1+j, 0, "%-10.10s", mtname);
1994-05-27 12:33:43 +00:00
wmove(wnd, 1 + j, 10);
if (max > 60) {
snprintf(buf, sizeof(buf), " %d", max);
1994-05-27 12:33:43 +00:00
max = 60;
while (max--)
waddch(wnd, 'X');
waddstr(wnd, buf);
} else
1994-05-27 12:33:43 +00:00
while (max--)
waddch(wnd, 'X');
wclrtoeol(wnd);
m_mbtypes[index] = 0;
1994-05-27 12:33:43 +00:00
}
Introduce numerous SMP friendly changes to the mbuf allocator. Namely, introduce a modified allocation mechanism for mbufs and mbuf clusters; one which can scale under SMP and which offers the possibility of resource reclamation to be implemented in the future. Notable advantages: o Reduce contention for SMP by offering per-CPU pools and locks. o Better use of data cache due to per-CPU pools. o Much less code cache pollution due to excessively large allocation macros. o Framework for `grouping' objects from same page together so as to be able to possibly free wired-down pages back to the system if they are no longer needed by the network stacks. Additional things changed with this addition: - Moved some mbuf specific declarations and initializations from sys/conf/param.c into mbuf-specific code where they belong. - m_getclr() has been renamed to m_get_clrd() because the old name is really confusing. m_getclr() HAS been preserved though and is defined to the new name. No tree sweep has been done "to change the interface," as the old name will continue to be supported and is not depracated. The change was merely done because m_getclr() sounds too much like "m_get a cluster." - TEMPORARILY disabled mbtypes statistics displaying in netstat(1) and systat(1) (see TODO below). - Fixed systat(1) to display number of "free mbufs" based on new per-CPU stat structures. - Fixed netstat(1) to display new per-CPU stats based on sysctl-exported per-CPU stat structures. All infos are fetched via sysctl. TODO (in order of priority): - Re-enable mbtypes statistics in both netstat(1) and systat(1) after introducing an SMP friendly way to collect the mbtypes stats under the already introduced per-CPU locks (i.e. hopefully don't use atomic() - it seems too costly for a mere stat update, especially when other locks are already present). - Optionally have systat(1) display not only "total free mbufs" but also "total free mbufs per CPU pool." - Fix minor length-fetching issues in netstat(1) related to recently re-enabled option to read mbuf stats from a core file. - Move reference counters at least for mbuf clusters into an unused portion of the cluster itself, to save space and need to allocate a counter. - Look into introducing resource freeing possibly from a kproc. Reviewed by (in parts): jlemon, jake, silby, terry Tested by: jlemon (Intel & Alpha), mjacob (Intel & Alpha) Preliminary performance measurements: jlemon (and me, obviously) URL: http://people.freebsd.org/~bmilekic/mb_alloc/
2001-06-22 06:35:32 +00:00
/*
* Print total number of free mbufs.
*/
if (totfree > 0) {
mvwprintw(wnd, 1+j, 0, "%-10.10s", "free");
Introduce numerous SMP friendly changes to the mbuf allocator. Namely, introduce a modified allocation mechanism for mbufs and mbuf clusters; one which can scale under SMP and which offers the possibility of resource reclamation to be implemented in the future. Notable advantages: o Reduce contention for SMP by offering per-CPU pools and locks. o Better use of data cache due to per-CPU pools. o Much less code cache pollution due to excessively large allocation macros. o Framework for `grouping' objects from same page together so as to be able to possibly free wired-down pages back to the system if they are no longer needed by the network stacks. Additional things changed with this addition: - Moved some mbuf specific declarations and initializations from sys/conf/param.c into mbuf-specific code where they belong. - m_getclr() has been renamed to m_get_clrd() because the old name is really confusing. m_getclr() HAS been preserved though and is defined to the new name. No tree sweep has been done "to change the interface," as the old name will continue to be supported and is not depracated. The change was merely done because m_getclr() sounds too much like "m_get a cluster." - TEMPORARILY disabled mbtypes statistics displaying in netstat(1) and systat(1) (see TODO below). - Fixed systat(1) to display number of "free mbufs" based on new per-CPU stat structures. - Fixed netstat(1) to display new per-CPU stats based on sysctl-exported per-CPU stat structures. All infos are fetched via sysctl. TODO (in order of priority): - Re-enable mbtypes statistics in both netstat(1) and systat(1) after introducing an SMP friendly way to collect the mbtypes stats under the already introduced per-CPU locks (i.e. hopefully don't use atomic() - it seems too costly for a mere stat update, especially when other locks are already present). - Optionally have systat(1) display not only "total free mbufs" but also "total free mbufs per CPU pool." - Fix minor length-fetching issues in netstat(1) related to recently re-enabled option to read mbuf stats from a core file. - Move reference counters at least for mbuf clusters into an unused portion of the cluster itself, to save space and need to allocate a counter. - Look into introducing resource freeing possibly from a kproc. Reviewed by (in parts): jlemon, jake, silby, terry Tested by: jlemon (Intel & Alpha), mjacob (Intel & Alpha) Preliminary performance measurements: jlemon (and me, obviously) URL: http://people.freebsd.org/~bmilekic/mb_alloc/
2001-06-22 06:35:32 +00:00
if (totfree > 60) {
snprintf(buf, sizeof(buf), " %lu", totfree);
totfree = 60;
while(totfree--)
waddch(wnd, 'X');
waddstr(wnd, buf);
} else {
Introduce numerous SMP friendly changes to the mbuf allocator. Namely, introduce a modified allocation mechanism for mbufs and mbuf clusters; one which can scale under SMP and which offers the possibility of resource reclamation to be implemented in the future. Notable advantages: o Reduce contention for SMP by offering per-CPU pools and locks. o Better use of data cache due to per-CPU pools. o Much less code cache pollution due to excessively large allocation macros. o Framework for `grouping' objects from same page together so as to be able to possibly free wired-down pages back to the system if they are no longer needed by the network stacks. Additional things changed with this addition: - Moved some mbuf specific declarations and initializations from sys/conf/param.c into mbuf-specific code where they belong. - m_getclr() has been renamed to m_get_clrd() because the old name is really confusing. m_getclr() HAS been preserved though and is defined to the new name. No tree sweep has been done "to change the interface," as the old name will continue to be supported and is not depracated. The change was merely done because m_getclr() sounds too much like "m_get a cluster." - TEMPORARILY disabled mbtypes statistics displaying in netstat(1) and systat(1) (see TODO below). - Fixed systat(1) to display number of "free mbufs" based on new per-CPU stat structures. - Fixed netstat(1) to display new per-CPU stats based on sysctl-exported per-CPU stat structures. All infos are fetched via sysctl. TODO (in order of priority): - Re-enable mbtypes statistics in both netstat(1) and systat(1) after introducing an SMP friendly way to collect the mbtypes stats under the already introduced per-CPU locks (i.e. hopefully don't use atomic() - it seems too costly for a mere stat update, especially when other locks are already present). - Optionally have systat(1) display not only "total free mbufs" but also "total free mbufs per CPU pool." - Fix minor length-fetching issues in netstat(1) related to recently re-enabled option to read mbuf stats from a core file. - Move reference counters at least for mbuf clusters into an unused portion of the cluster itself, to save space and need to allocate a counter. - Look into introducing resource freeing possibly from a kproc. Reviewed by (in parts): jlemon, jake, silby, terry Tested by: jlemon (Intel & Alpha), mjacob (Intel & Alpha) Preliminary performance measurements: jlemon (and me, obviously) URL: http://people.freebsd.org/~bmilekic/mb_alloc/
2001-06-22 06:35:32 +00:00
while(totfree--)
waddch(wnd, 'X');
}
wclrtoeol(wnd);
j++;
}
1994-05-27 12:33:43 +00:00
wmove(wnd, 1+j, 0); wclrtobot(wnd);
}
int
initmbufs()
{
Introduce numerous SMP friendly changes to the mbuf allocator. Namely, introduce a modified allocation mechanism for mbufs and mbuf clusters; one which can scale under SMP and which offers the possibility of resource reclamation to be implemented in the future. Notable advantages: o Reduce contention for SMP by offering per-CPU pools and locks. o Better use of data cache due to per-CPU pools. o Much less code cache pollution due to excessively large allocation macros. o Framework for `grouping' objects from same page together so as to be able to possibly free wired-down pages back to the system if they are no longer needed by the network stacks. Additional things changed with this addition: - Moved some mbuf specific declarations and initializations from sys/conf/param.c into mbuf-specific code where they belong. - m_getclr() has been renamed to m_get_clrd() because the old name is really confusing. m_getclr() HAS been preserved though and is defined to the new name. No tree sweep has been done "to change the interface," as the old name will continue to be supported and is not depracated. The change was merely done because m_getclr() sounds too much like "m_get a cluster." - TEMPORARILY disabled mbtypes statistics displaying in netstat(1) and systat(1) (see TODO below). - Fixed systat(1) to display number of "free mbufs" based on new per-CPU stat structures. - Fixed netstat(1) to display new per-CPU stats based on sysctl-exported per-CPU stat structures. All infos are fetched via sysctl. TODO (in order of priority): - Re-enable mbtypes statistics in both netstat(1) and systat(1) after introducing an SMP friendly way to collect the mbtypes stats under the already introduced per-CPU locks (i.e. hopefully don't use atomic() - it seems too costly for a mere stat update, especially when other locks are already present). - Optionally have systat(1) display not only "total free mbufs" but also "total free mbufs per CPU pool." - Fix minor length-fetching issues in netstat(1) related to recently re-enabled option to read mbuf stats from a core file. - Move reference counters at least for mbuf clusters into an unused portion of the cluster itself, to save space and need to allocate a counter. - Look into introducing resource freeing possibly from a kproc. Reviewed by (in parts): jlemon, jake, silby, terry Tested by: jlemon (Intel & Alpha), mjacob (Intel & Alpha) Preliminary performance measurements: jlemon (and me, obviously) URL: http://people.freebsd.org/~bmilekic/mb_alloc/
2001-06-22 06:35:32 +00:00
int i;
size_t len;
len = sizeof *mbstat;
if ((mbstat = malloc(len)) == NULL) {
error("malloc mbstat failed");
return 0;
}
if (sysctlbyname("kern.ipc.mbstat", mbstat, &len, NULL, 0) < 0) {
error("sysctl retrieving mbstat");
return 0;
}
nmbtypes = mbstat->m_numtypes;
if ((m_mbtypes = calloc(nmbtypes, sizeof(long *))) == NULL) {
error("calloc m_mbtypes failed");
return 0;
}
Introduce numerous SMP friendly changes to the mbuf allocator. Namely, introduce a modified allocation mechanism for mbufs and mbuf clusters; one which can scale under SMP and which offers the possibility of resource reclamation to be implemented in the future. Notable advantages: o Reduce contention for SMP by offering per-CPU pools and locks. o Better use of data cache due to per-CPU pools. o Much less code cache pollution due to excessively large allocation macros. o Framework for `grouping' objects from same page together so as to be able to possibly free wired-down pages back to the system if they are no longer needed by the network stacks. Additional things changed with this addition: - Moved some mbuf specific declarations and initializations from sys/conf/param.c into mbuf-specific code where they belong. - m_getclr() has been renamed to m_get_clrd() because the old name is really confusing. m_getclr() HAS been preserved though and is defined to the new name. No tree sweep has been done "to change the interface," as the old name will continue to be supported and is not depracated. The change was merely done because m_getclr() sounds too much like "m_get a cluster." - TEMPORARILY disabled mbtypes statistics displaying in netstat(1) and systat(1) (see TODO below). - Fixed systat(1) to display number of "free mbufs" based on new per-CPU stat structures. - Fixed netstat(1) to display new per-CPU stats based on sysctl-exported per-CPU stat structures. All infos are fetched via sysctl. TODO (in order of priority): - Re-enable mbtypes statistics in both netstat(1) and systat(1) after introducing an SMP friendly way to collect the mbtypes stats under the already introduced per-CPU locks (i.e. hopefully don't use atomic() - it seems too costly for a mere stat update, especially when other locks are already present). - Optionally have systat(1) display not only "total free mbufs" but also "total free mbufs per CPU pool." - Fix minor length-fetching issues in netstat(1) related to recently re-enabled option to read mbuf stats from a core file. - Move reference counters at least for mbuf clusters into an unused portion of the cluster itself, to save space and need to allocate a counter. - Look into introducing resource freeing possibly from a kproc. Reviewed by (in parts): jlemon, jake, silby, terry Tested by: jlemon (Intel & Alpha), mjacob (Intel & Alpha) Preliminary performance measurements: jlemon (and me, obviously) URL: http://people.freebsd.org/~bmilekic/mb_alloc/
2001-06-22 06:35:32 +00:00
if (sysctlbyname("kern.ipc.mb_statpcpu", NULL, &len, NULL, 0) < 0) {
error("sysctl getting mbpstat total size failed");
return 0;
1994-05-27 12:33:43 +00:00
}
Introduce numerous SMP friendly changes to the mbuf allocator. Namely, introduce a modified allocation mechanism for mbufs and mbuf clusters; one which can scale under SMP and which offers the possibility of resource reclamation to be implemented in the future. Notable advantages: o Reduce contention for SMP by offering per-CPU pools and locks. o Better use of data cache due to per-CPU pools. o Much less code cache pollution due to excessively large allocation macros. o Framework for `grouping' objects from same page together so as to be able to possibly free wired-down pages back to the system if they are no longer needed by the network stacks. Additional things changed with this addition: - Moved some mbuf specific declarations and initializations from sys/conf/param.c into mbuf-specific code where they belong. - m_getclr() has been renamed to m_get_clrd() because the old name is really confusing. m_getclr() HAS been preserved though and is defined to the new name. No tree sweep has been done "to change the interface," as the old name will continue to be supported and is not depracated. The change was merely done because m_getclr() sounds too much like "m_get a cluster." - TEMPORARILY disabled mbtypes statistics displaying in netstat(1) and systat(1) (see TODO below). - Fixed systat(1) to display number of "free mbufs" based on new per-CPU stat structures. - Fixed netstat(1) to display new per-CPU stats based on sysctl-exported per-CPU stat structures. All infos are fetched via sysctl. TODO (in order of priority): - Re-enable mbtypes statistics in both netstat(1) and systat(1) after introducing an SMP friendly way to collect the mbtypes stats under the already introduced per-CPU locks (i.e. hopefully don't use atomic() - it seems too costly for a mere stat update, especially when other locks are already present). - Optionally have systat(1) display not only "total free mbufs" but also "total free mbufs per CPU pool." - Fix minor length-fetching issues in netstat(1) related to recently re-enabled option to read mbuf stats from a core file. - Move reference counters at least for mbuf clusters into an unused portion of the cluster itself, to save space and need to allocate a counter. - Look into introducing resource freeing possibly from a kproc. Reviewed by (in parts): jlemon, jake, silby, terry Tested by: jlemon (Intel & Alpha), mjacob (Intel & Alpha) Preliminary performance measurements: jlemon (and me, obviously) URL: http://people.freebsd.org/~bmilekic/mb_alloc/
2001-06-22 06:35:32 +00:00
num_objs = (int)(len / sizeof(struct mbpstat));
if ((mbpstat = calloc(num_objs, sizeof(struct mbpstat *))) == NULL) {
error("calloc mbpstat pointers failed");
return 0;
}
if ((mbpstat[0] = calloc(num_objs, sizeof(struct mbpstat))) == NULL) {
error("calloc mbpstat structures failed");
return 0;
}
Introduce numerous SMP friendly changes to the mbuf allocator. Namely, introduce a modified allocation mechanism for mbufs and mbuf clusters; one which can scale under SMP and which offers the possibility of resource reclamation to be implemented in the future. Notable advantages: o Reduce contention for SMP by offering per-CPU pools and locks. o Better use of data cache due to per-CPU pools. o Much less code cache pollution due to excessively large allocation macros. o Framework for `grouping' objects from same page together so as to be able to possibly free wired-down pages back to the system if they are no longer needed by the network stacks. Additional things changed with this addition: - Moved some mbuf specific declarations and initializations from sys/conf/param.c into mbuf-specific code where they belong. - m_getclr() has been renamed to m_get_clrd() because the old name is really confusing. m_getclr() HAS been preserved though and is defined to the new name. No tree sweep has been done "to change the interface," as the old name will continue to be supported and is not depracated. The change was merely done because m_getclr() sounds too much like "m_get a cluster." - TEMPORARILY disabled mbtypes statistics displaying in netstat(1) and systat(1) (see TODO below). - Fixed systat(1) to display number of "free mbufs" based on new per-CPU stat structures. - Fixed netstat(1) to display new per-CPU stats based on sysctl-exported per-CPU stat structures. All infos are fetched via sysctl. TODO (in order of priority): - Re-enable mbtypes statistics in both netstat(1) and systat(1) after introducing an SMP friendly way to collect the mbtypes stats under the already introduced per-CPU locks (i.e. hopefully don't use atomic() - it seems too costly for a mere stat update, especially when other locks are already present). - Optionally have systat(1) display not only "total free mbufs" but also "total free mbufs per CPU pool." - Fix minor length-fetching issues in netstat(1) related to recently re-enabled option to read mbuf stats from a core file. - Move reference counters at least for mbuf clusters into an unused portion of the cluster itself, to save space and need to allocate a counter. - Look into introducing resource freeing possibly from a kproc. Reviewed by (in parts): jlemon, jake, silby, terry Tested by: jlemon (Intel & Alpha), mjacob (Intel & Alpha) Preliminary performance measurements: jlemon (and me, obviously) URL: http://people.freebsd.org/~bmilekic/mb_alloc/
2001-06-22 06:35:32 +00:00
for (i = 0; i < num_objs; i++)
mbpstat[i] = mbpstat[0] + i;
return 1;
1994-05-27 12:33:43 +00:00
}
void
fetchmbufs()
{
size_t len;
Introduce numerous SMP friendly changes to the mbuf allocator. Namely, introduce a modified allocation mechanism for mbufs and mbuf clusters; one which can scale under SMP and which offers the possibility of resource reclamation to be implemented in the future. Notable advantages: o Reduce contention for SMP by offering per-CPU pools and locks. o Better use of data cache due to per-CPU pools. o Much less code cache pollution due to excessively large allocation macros. o Framework for `grouping' objects from same page together so as to be able to possibly free wired-down pages back to the system if they are no longer needed by the network stacks. Additional things changed with this addition: - Moved some mbuf specific declarations and initializations from sys/conf/param.c into mbuf-specific code where they belong. - m_getclr() has been renamed to m_get_clrd() because the old name is really confusing. m_getclr() HAS been preserved though and is defined to the new name. No tree sweep has been done "to change the interface," as the old name will continue to be supported and is not depracated. The change was merely done because m_getclr() sounds too much like "m_get a cluster." - TEMPORARILY disabled mbtypes statistics displaying in netstat(1) and systat(1) (see TODO below). - Fixed systat(1) to display number of "free mbufs" based on new per-CPU stat structures. - Fixed netstat(1) to display new per-CPU stats based on sysctl-exported per-CPU stat structures. All infos are fetched via sysctl. TODO (in order of priority): - Re-enable mbtypes statistics in both netstat(1) and systat(1) after introducing an SMP friendly way to collect the mbtypes stats under the already introduced per-CPU locks (i.e. hopefully don't use atomic() - it seems too costly for a mere stat update, especially when other locks are already present). - Optionally have systat(1) display not only "total free mbufs" but also "total free mbufs per CPU pool." - Fix minor length-fetching issues in netstat(1) related to recently re-enabled option to read mbuf stats from a core file. - Move reference counters at least for mbuf clusters into an unused portion of the cluster itself, to save space and need to allocate a counter. - Look into introducing resource freeing possibly from a kproc. Reviewed by (in parts): jlemon, jake, silby, terry Tested by: jlemon (Intel & Alpha), mjacob (Intel & Alpha) Preliminary performance measurements: jlemon (and me, obviously) URL: http://people.freebsd.org/~bmilekic/mb_alloc/
2001-06-22 06:35:32 +00:00
len = num_objs * sizeof(struct mbpstat);
if (sysctlbyname("kern.ipc.mb_statpcpu", mbpstat[0], &len, NULL, 0) < 0)
printw("sysctl: mbpstat: %s", strerror(errno));
1994-05-27 12:33:43 +00:00
}