diff options
Diffstat (limited to 'Documentation/networking')
| -rw-r--r-- | Documentation/networking/.gitignore | 1 | ||||
| -rw-r--r-- | Documentation/networking/00-INDEX | 2 | ||||
| -rw-r--r-- | Documentation/networking/Makefile | 5 | ||||
| -rw-r--r-- | Documentation/networking/arcnet.txt | 7 | ||||
| -rw-r--r-- | Documentation/networking/bonding.txt | 79 | ||||
| -rw-r--r-- | Documentation/networking/ifenslave.c | 1105 | ||||
| -rw-r--r-- | Documentation/networking/ip-sysctl.txt | 11 | ||||
| -rw-r--r-- | Documentation/networking/ipvs-sysctl.txt | 13 | ||||
| -rw-r--r-- | Documentation/networking/netlink_mmap.txt | 6 | ||||
| -rw-r--r-- | Documentation/networking/packet_mmap.txt | 133 | ||||
| -rw-r--r-- | Documentation/networking/scaling.txt | 58 | ||||
| -rw-r--r-- | Documentation/networking/vortex.txt | 2 |
12 files changed, 152 insertions, 1270 deletions
diff --git a/Documentation/networking/.gitignore b/Documentation/networking/.gitignore index 286a5680f49..e69de29bb2d 100644 --- a/Documentation/networking/.gitignore +++ b/Documentation/networking/.gitignore @@ -1 +0,0 @@ -ifenslave diff --git a/Documentation/networking/00-INDEX b/Documentation/networking/00-INDEX index 258d9b92c36..32dfbd92412 100644 --- a/Documentation/networking/00-INDEX +++ b/Documentation/networking/00-INDEX @@ -88,8 +88,6 @@ gianfar.txt - Gianfar Ethernet Driver. ieee802154.txt - Linux IEEE 802.15.4 implementation, API and drivers -ifenslave.c - - Configure network interfaces for parallel routing (bonding). igb.txt - README for the Intel Gigabit Ethernet Driver (igb). igbvf.txt diff --git a/Documentation/networking/Makefile b/Documentation/networking/Makefile index 24c308dd3fd..0aa1ac98fc2 100644 --- a/Documentation/networking/Makefile +++ b/Documentation/networking/Makefile @@ -1,11 +1,6 @@ # kbuild trick to avoid linker error. Can be omitted if a module is built. obj- := dummy.o -# List of programs to build -hostprogs-y := ifenslave - -HOSTCFLAGS_ifenslave.o += -I$(objtree)/usr/include - # Tell kbuild to always build the programs always := $(hostprogs-y) diff --git a/Documentation/networking/arcnet.txt b/Documentation/networking/arcnet.txt index 9ff57950215..aff97f47c05 100644 --- a/Documentation/networking/arcnet.txt +++ b/Documentation/networking/arcnet.txt @@ -70,9 +70,10 @@ list, mail to linux-arcnet@tichy.ch.uj.edu.pl. There are archives of the mailing list at: http://epistolary.org/mailman/listinfo.cgi/arcnet -The people on linux-net@vger.kernel.org have also been known to be very -helpful, especially when we're talking about ALPHA Linux kernels that may or -may not work right in the first place. +The people on linux-net@vger.kernel.org (now defunct, replaced by +netdev@vger.kernel.org) have also been known to be very helpful, especially +when we're talking about ALPHA Linux kernels that may or may not work right +in the first place. Other Drivers and Info diff --git a/Documentation/networking/bonding.txt b/Documentation/networking/bonding.txt index 10a015c384b..87bbcfee2e0 100644 --- a/Documentation/networking/bonding.txt +++ b/Documentation/networking/bonding.txt @@ -104,8 +104,7 @@ Table of Contents ============================== Most popular distro kernels ship with the bonding driver -already available as a module and the ifenslave user level control -program installed and ready for use. If your distro does not, or you +already available as a module. If your distro does not, or you have need to compile bonding from source (e.g., configuring and installing a mainline kernel from kernel.org), you'll need to perform the following steps: @@ -124,46 +123,13 @@ device support" section. It is recommended that you configure the driver as module since it is currently the only way to pass parameters to the driver or configure more than one bonding device. - Build and install the new kernel and modules, then continue -below to install ifenslave. + Build and install the new kernel and modules. -1.2 Install ifenslave Control Utility +1.2 Bonding Control Utility ------------------------------------- - The ifenslave user level control program is included in the -kernel source tree, in the file Documentation/networking/ifenslave.c. -It is generally recommended that you use the ifenslave that -corresponds to the kernel that you are using (either from the same -source tree or supplied with the distro), however, ifenslave -executables from older kernels should function (but features newer -than the ifenslave release are not supported). Running an ifenslave -that is newer than the kernel is not supported, and may or may not -work. - - To install ifenslave, do the following: - -# gcc -Wall -O -I/usr/src/linux/include ifenslave.c -o ifenslave -# cp ifenslave /sbin/ifenslave - - If your kernel source is not in "/usr/src/linux," then replace -"/usr/src/linux/include" in the above with the location of your kernel -source include directory. - - You may wish to back up any existing /sbin/ifenslave, or, for -testing or informal use, tag the ifenslave to the kernel version -(e.g., name the ifenslave executable /sbin/ifenslave-2.6.10). - -IMPORTANT NOTE: - - If you omit the "-I" or specify an incorrect directory, you -may end up with an ifenslave that is incompatible with the kernel -you're trying to build it for. Some distros (e.g., Red Hat from 7.1 -onwards) do not have /usr/include/linux symbolically linked to the -default kernel source include directory. - -SECOND IMPORTANT NOTE: - If you plan to configure bonding using sysfs or using the -/etc/network/interfaces file, you do not need to use ifenslave. + It is recommended to configure bonding via iproute2 (netlink) +or sysfs, the old ifenslave control utility is obsolete. 2. Bonding Driver Options ========================= @@ -337,6 +303,12 @@ arp_validate such a situation, validation of backup slaves must be disabled. + The validation of ARP requests on backup slaves is mainly + helping bonding to decide which slaves are more likely to + work in case of the active slave failure, it doesn't really + guarantee that the backup slave will work if it's selected + as the next active slave. + This option is useful in network configurations in which multiple bonding hosts are concurrently issuing ARPs to one or more targets beyond a common switch. Should the link between @@ -349,6 +321,25 @@ arp_validate This option was added in bonding version 3.1.0. +arp_all_targets + + Specifies the quantity of arp_ip_targets that must be reachable + in order for the ARP monitor to consider a slave as being up. + This option affects only active-backup mode for slaves with + arp_validation enabled. + + Possible values are: + + any or 0 + + consider the slave up only when any of the arp_ip_targets + is reachable + + all or 1 + + consider the slave up only when all of the arp_ip_targets + are reachable + downdelay Specifies the time, in milliseconds, to wait before disabling @@ -851,7 +842,7 @@ resend_igmp ============================== You can configure bonding using either your distro's network -initialization scripts, or manually using either ifenslave or the +initialization scripts, or manually using either iproute2 or the sysfs interface. Distros generally use one of three packages for the network initialization scripts: initscripts, sysconfig or interfaces. Recent versions of these packages have support for bonding, while older @@ -1160,7 +1151,7 @@ not support this method for specifying multiple bonding interfaces; for those instances, see the "Configuring Multiple Bonds Manually" section, below. -3.3 Configuring Bonding Manually with Ifenslave +3.3 Configuring Bonding Manually with iproute2 ----------------------------------------------- This section applies to distros whose network initialization @@ -1171,7 +1162,7 @@ version 8. The general method for these systems is to place the bonding module parameters into a config file in /etc/modprobe.d/ (as appropriate for the installed distro), then add modprobe and/or -ifenslave commands to the system's global init script. The name of +`ip link` commands to the system's global init script. The name of the global init script differs; for sysconfig, it is /etc/init.d/boot.local and for initscripts it is /etc/rc.d/rc.local. @@ -1183,8 +1174,8 @@ reboots, edit the appropriate file (/etc/init.d/boot.local or modprobe bonding mode=balance-alb miimon=100 modprobe e100 ifconfig bond0 192.168.1.1 netmask 255.255.255.0 up -ifenslave bond0 eth0 -ifenslave bond0 eth1 +ip link set eth0 master bond0 +ip link set eth1 master bond0 Replace the example bonding module parameters and bond0 network configuration (IP address, netmask, etc) with the appropriate diff --git a/Documentation/networking/ifenslave.c b/Documentation/networking/ifenslave.c deleted file mode 100644 index ac5debb2f16..00000000000 --- a/Documentation/networking/ifenslave.c +++ /dev/null @@ -1,1105 +0,0 @@ -/* Mode: C; - * ifenslave.c: Configure network interfaces for parallel routing. - * - * This program controls the Linux implementation of running multiple - * network interfaces in parallel. - * - * Author: Donald Becker <becker@cesdis.gsfc.nasa.gov> - * Copyright 1994-1996 Donald Becker - * - * This program is free software; you can redistribute it - * and/or modify it under the terms of the GNU General Public - * License as published by the Free Software Foundation. - * - * The author may be reached as becker@CESDIS.gsfc.nasa.gov, or C/O - * Center of Excellence in Space Data and Information Sciences - * Code 930.5, Goddard Space Flight Center, Greenbelt MD 20771 - * - * Changes : - * - 2000/10/02 Willy Tarreau <willy at meta-x.org> : - * - few fixes. Master's MAC address is now correctly taken from - * the first device when not previously set ; - * - detach support : call BOND_RELEASE to detach an enslaved interface. - * - give a mini-howto from command-line help : # ifenslave -h - * - * - 2001/02/16 Chad N. Tindel <ctindel at ieee dot org> : - * - Master is now brought down before setting the MAC address. In - * the 2.4 kernel you can't change the MAC address while the device is - * up because you get EBUSY. - * - * - 2001/09/13 Takao Indoh <indou dot takao at jp dot fujitsu dot com> - * - Added the ability to change the active interface on a mode 1 bond - * at runtime. - * - * - 2001/10/23 Chad N. Tindel <ctindel at ieee dot org> : - * - No longer set the MAC address of the master. The bond device will - * take care of this itself - * - Try the SIOC*** versions of the bonding ioctls before using the - * old versions - * - 2002/02/18 Erik Habbinga <erik_habbinga @ hp dot com> : - * - ifr2.ifr_flags was not initialized in the hwaddr_notset case, - * SIOCGIFFLAGS now called before hwaddr_notset test - * - * - 2002/10/31 Tony Cureington <tony.cureington * hp_com> : - * - If the master does not have a hardware address when the first slave - * is enslaved, the master is assigned the hardware address of that - * slave - there is a comment in bonding.c stating "ifenslave takes - * care of this now." This corrects the problem of slaves having - * different hardware addresses in active-backup mode when - * multiple interfaces are specified on a single ifenslave command - * (ifenslave bond0 eth0 eth1). - * - * - 2003/03/18 - Tsippy Mendelson <tsippy.mendelson at intel dot com> and - * Shmulik Hen <shmulik.hen at intel dot com> - * - Moved setting the slave's mac address and openning it, from - * the application to the driver. This enables support of modes - * that need to use the unique mac address of each slave. - * The driver also takes care of closing the slave and restoring its - * original mac address upon release. - * In addition, block possibility of enslaving before the master is up. - * This prevents putting the system in an undefined state. - * - * - 2003/05/01 - Amir Noam <amir.noam at intel dot com> - * - Added ABI version control to restore compatibility between - * new/old ifenslave and new/old bonding. - * - Prevent adding an adapter that is already a slave. - * Fixes the problem of stalling the transmission and leaving - * the slave in a down state. - * - * - 2003/05/01 - Shmulik Hen <shmulik.hen at intel dot com> - * - Prevent enslaving if the bond device is down. - * Fixes the problem of leaving the system in unstable state and - * halting when trying to remove the module. - * - Close socket on all abnormal exists. - * - Add versioning scheme that follows that of the bonding driver. - * current version is 1.0.0 as a base line. - * - * - 2003/05/22 - Jay Vosburgh <fubar at us dot ibm dot com> - * - ifenslave -c was broken; it's now fixed - * - Fixed problem with routes vanishing from master during enslave - * processing. - * - * - 2003/05/27 - Amir Noam <amir.noam at intel dot com> - * - Fix backward compatibility issues: - * For drivers not using ABI versions, slave was set down while - * it should be left up before enslaving. - * Also, master was not set down and the default set_mac_address() - * would fail and generate an error message in the system log. - * - For opt_c: slave should not be set to the master's setting - * while it is running. It was already set during enslave. To - * simplify things, it is now handled separately. - * - * - 2003/12/01 - Shmulik Hen <shmulik.hen at intel dot com> - * - Code cleanup and style changes - * set version to 1.1.0 - */ - -#define APP_VERSION "1.1.0" -#define APP_RELDATE "December 1, 2003" -#define APP_NAME "ifenslave" - -static char *version = -APP_NAME ".c:v" APP_VERSION " (" APP_RELDATE ")\n" -"o Donald Becker (becker@cesdis.gsfc.nasa.gov).\n" -"o Detach support added on 2000/10/02 by Willy Tarreau (willy at meta-x.org).\n" -"o 2.4 kernel support added on 2001/02/16 by Chad N. Tindel\n" -" (ctindel at ieee dot org).\n"; - -static const char *usage_msg = -"Usage: ifenslave [-f] <master-if> <slave-if> [<slave-if>...]\n" -" ifenslave -d <master-if> <slave-if> [<slave-if>...]\n" -" ifenslave -c <master-if> <slave-if>\n" -" ifenslave --help\n"; - -static const char *help_msg = -"\n" -" To create a bond device, simply follow these three steps :\n" -" - ensure that the required drivers are properly loaded :\n" -" # modprobe bonding ; modprobe <3c59x|eepro100|pcnet32|tulip|...>\n" -" - assign an IP address to the bond device :\n" -" # ifconfig bond0 <addr> netmask <mask> broadcast <bcast>\n" -" - attach all the interfaces you need to the bond device :\n" -" # ifenslave [{-f|--force}] bond0 eth0 [eth1 [eth2]...]\n" -" If bond0 didn't have a MAC address, it will take eth0's. Then, all\n" -" interfaces attached AFTER this assignment will get the same MAC addr.\n" -" (except for ALB/TLB modes)\n" -"\n" -" To set the bond device down and automatically release all the slaves :\n" -" # ifconfig bond0 down\n" -"\n" -" To detach a dead interface without setting the bond device down :\n" -" # ifenslave {-d|--detach} bond0 eth0 [eth1 [eth2]...]\n" -"\n" -" To change active slave :\n" -" # ifenslave {-c|--change-active} bond0 eth0\n" -"\n" -" To show master interface info\n" -" # ifenslave bond0\n" -"\n" -" To show all interfaces info\n" -" # ifenslave {-a|--all-interfaces}\n" -"\n" -" To be more verbose\n" -" # ifenslave {-v|--verbose} ...\n" -"\n" -" # ifenslave {-u|--usage} Show usage\n" -" # ifenslave {-V|--version} Show version\n" -" # ifenslave {-h|--help} This message\n" -"\n"; - -#include <unistd.h> -#include <stdlib.h> -#include <stdio.h> -#include <ctype.h> -#include <string.h> -#include <errno.h> -#include <fcntl.h> -#include <getopt.h> -#include <sys/types.h> -#include <sys/socket.h> -#include <sys/ioctl.h> -#include <linux/if.h> -#include <net/if_arp.h> -#include <linux/if_ether.h> -#include <linux/if_bonding.h> -#include <linux/sockios.h> - -typedef unsigned long long u64; /* hack, so we may include kernel's ethtool.h */ -typedef __uint32_t u32; /* ditto */ -typedef __uint16_t u16; /* ditto */ -typedef __uint8_t u8; /* ditto */ -#include <linux/ethtool.h> - -struct option longopts[] = { - /* { name has_arg *flag val } */ - {"all-interfaces", 0, 0, 'a'}, /* Show all interfaces. */ - {"change-active", 0, 0, 'c'}, /* Change the active slave. */ - {"detach", 0, 0, 'd'}, /* Detach a slave interface. */ - {"force", 0, 0, 'f'}, /* Force the operation. */ - {"help", 0, 0, 'h'}, /* Give help */ - {"usage", 0, 0, 'u'}, /* Give usage */ - {"verbose", 0, 0, 'v'}, /* Report each action taken. */ - {"version", 0, 0, 'V'}, /* Emit version information. */ - { 0, 0, 0, 0} -}; - -/* Command-line flags. */ -unsigned int -opt_a = 0, /* Show-all-interfaces flag. */ -opt_c = 0, /* Change-active-slave flag. */ -opt_d = 0, /* Detach a slave interface. */ -opt_f = 0, /* Force the operation. */ -opt_h = 0, /* Help */ -opt_u = 0, /* Usage */ -opt_v = 0, /* Verbose flag. */ -opt_V = 0; /* Version */ - -int skfd = -1; /* AF_INET socket for ioctl() calls.*/ -int abi_ver = 0; /* userland - kernel ABI version */ -int hwaddr_set = 0; /* Master's hwaddr is set */ -int saved_errno; - -struct ifreq master_mtu, master_flags, master_hwaddr; -struct ifreq slave_mtu, slave_flags, slave_hwaddr; - -struct dev_ifr { - struct ifreq *req_ifr; - char *req_name; - int req_type; -}; - -struct dev_ifr master_ifra[] = { - {&master_mtu, "SIOCGIFMTU", SIOCGIFMTU}, - {&master_flags, "SIOCGIFFLAGS", SIOCGIFFLAGS}, - {&master_hwaddr, "SIOCGIFHWADDR", SIOCGIFHWADDR}, - {NULL, "", 0} -}; - -struct dev_ifr slave_ifra[] = { - {&slave_mtu, "SIOCGIFMTU", SIOCGIFMTU}, - {&slave_flags, "SIOCGIFFLAGS", SIOCGIFFLAGS}, - {&slave_hwaddr, "SIOCGIFHWADDR", SIOCGIFHWADDR}, - {NULL, "", 0} -}; - -static void if_print(char *ifname); -static int get_drv_info(char *master_ifname); -static int get_if_settings(char *ifname, struct dev_ifr ifra[]); -static int get_slave_flags(char *slave_ifname); -static int set_master_hwaddr(char *master_ifname, struct sockaddr *hwaddr); -static int set_slave_hwaddr(char *slave_ifname, struct sockaddr *hwaddr); -static int set_slave_mtu(char *slave_ifname, int mtu); -static int set_if_flags(char *ifname, short flags); -static int set_if_up(char *ifname, short flags); -static int set_if_down(char *ifname, short flags); -static int clear_if_addr(char *ifname); -static int set_if_addr(char *master_ifname, char *slave_ifname); -static int change_active(char *master_ifname, char *slave_ifname); -static int enslave(char *master_ifname, char *slave_ifname); -static int release(char *master_ifname, char *slave_ifname); -#define v_print(fmt, args...) \ - if (opt_v) \ - fprintf(stderr, fmt, ## args ) - -int main(int argc, char *argv[]) -{ - char **spp, *master_ifname, *slave_ifname; - int c, i, rv; - int res = 0; - int exclusive = 0; - - while ((c = getopt_long(argc, argv, "acdfhuvV", longopts, 0)) != EOF) { - switch (c) { - case 'a': opt_a++; exclusive++; break; - case 'c': opt_c++; exclusive++; break; - case 'd': opt_d++; exclusive++; break; - case 'f': opt_f++; exclusive++; break; - case 'h': opt_h++; exclusive++; break; - case 'u': opt_u++; exclusive++; break; - case 'v': opt_v++; break; - case 'V': opt_V++; exclusive++; break; - - case '?': - fprintf(stderr, "%s", usage_msg); - res = 2; - goto out; - } - } - - /* options check */ - if (exclusive > 1) { - fprintf(stderr, "%s", usage_msg); - res = 2; - goto out; - } - - if (opt_v || opt_V) { - printf("%s", version); - if (opt_V) { - res = 0; - goto out; - } - } - - if (opt_u) { - printf("%s", usage_msg); - res = 0; - goto out; - } - - if (opt_h) { - printf("%s", usage_msg); - printf("%s", help_msg); - res = 0; - goto out; - } - - /* Open a basic socket */ - if ((skfd = socket(AF_INET, SOCK_DGRAM, 0)) < 0) { - perror("socket"); - res = 1; - goto out; - } - - if (opt_a) { - if (optind == argc) { - /* No remaining args */ - /* show all interfaces */ - if_print((char *)NULL); - goto out; - } else { - /* Just show usage */ - fprintf(stderr, "%s", usage_msg); - res = 2; - goto out; - } - } - - /* Copy the interface name */ - spp = argv + optind; - master_ifname = *spp++; - - if (master_ifname == NULL) { - fprintf(stderr, "%s", usage_msg); - res = 2; - goto out; - } - - /* exchange abi version with bonding module */ - res = get_drv_info(master_ifname); - if (res) { - fprintf(stderr, - "Master '%s': Error: handshake with driver failed. " - "Aborting\n", - master_ifname); - goto out; - } - - slave_ifname = *spp++; - - if (slave_ifname == NULL) { - if (opt_d || opt_c) { - fprintf(stderr, "%s", usage_msg); - res = 2; - goto out; - } - - /* A single arg means show the - * configuration for this interface - */ - if_print(master_ifname); - goto out; - } - - res = get_if_settings(master_ifname, master_ifra); - if (res) { - /* Probably a good reason not to go on */ - fprintf(stderr, - "Master '%s': Error: get settings failed: %s. " - "Aborting\n", - master_ifname, strerror(res)); - goto out; - } - - /* check if master is indeed a master; - * if not then fail any operation - */ - if (!(master_flags.ifr_flags & IFF_MASTER)) { - fprintf(stderr, - "Illegal operation; the specified interface '%s' " - "is not a master. Aborting\n", - master_ifname); - res = 1; - goto out; - } - - /* check if master is up; if not then fail any operation */ - if (!(master_flags.ifr_flags & IFF_UP)) { - fprintf(stderr, - "Illegal operation; the specified master interface " - "'%s' is not up.\n", - master_ifname); - res = 1; - goto out; - } - - /* Only for enslaving */ - if (!opt_c && !opt_d) { - sa_family_t master_family = master_hwaddr.ifr_hwaddr.sa_family; - unsigned char *hwaddr = - (unsigned char *)master_hwaddr.ifr_hwaddr.sa_data; - - /* The family '1' is ARPHRD_ETHER for ethernet. */ - if (master_family != 1 && !opt_f) { - fprintf(stderr, - "Illegal operation: The specified master " - "interface '%s' is not ethernet-like.\n " - "This program is designed to work with " - "ethernet-like network interfaces.\n " - "Use the '-f' option to force the " - "operation.\n", - master_ifname); - res = 1; - goto out; - } - - /* Check master's hw addr */ - for (i = 0; i < 6; i++) { - if (hwaddr[i] != 0) { - hwaddr_set = 1; - break; - } - } - - if (hwaddr_set) { - v_print("current hardware address of master '%s' " - "is %2.2x:%2.2x:%2.2x:%2.2x:%2.2x:%2.2x, " - "type %d\n", - master_ifname, - hwaddr[0], hwaddr[1], - hwaddr[2], hwaddr[3], - hwaddr[4], hwaddr[5], - master_family); - } - } - - /* Accepts only one slave */ - if (opt_c) { - /* change active slave */ - res = get_slave_flags(slave_ifname); - if (res) { - fprintf(stderr, - "Slave '%s': Error: get flags failed. " - "Aborting\n", - slave_ifname); - goto out; - } - res = change_active(master_ifname, slave_ifname); - if (res) { - fprintf(stderr, - "Master '%s', Slave '%s': Error: " - "Change active failed\n", - master_ifname, slave_ifname); - } - } else { - /* Accept multiple slaves */ - do { - if (opt_d) { - /* detach a slave interface from the master */ - rv = get_slave_flags(slave_ifname); - if (rv) { - /* Can't work with this slave. */ - /* remember the error and skip it*/ - fprintf(stderr, - "Slave '%s': Error: get flags " - "failed. Skipping\n", - slave_ifname); - res = rv; - continue; - } - rv = release(master_ifname, slave_ifname); - if (rv) { - fprintf(stderr, - "Master '%s', Slave '%s': Error: " - "Release failed\n", - master_ifname, slave_ifname); - res = rv; - } - } else { - /* attach a slave interface to the master */ - rv = get_if_settings(slave_ifname, slave_ifra); - if (rv) { - /* Can't work with this slave. */ - /* remember the error and skip it*/ - fprintf(stderr, - "Slave '%s': Error: get " - "settings failed: %s. " - "Skipping\n", - slave_ifname, strerror(rv)); - res = rv; - continue; - } - rv = enslave(master_ifname, slave_ifname); - if (rv) { - fprintf(stderr, - "Master '%s', Slave '%s': Error: " - "Enslave failed\n", - master_ifname, slave_ifname); - res = rv; - } - } - } while ((slave_ifname = *spp++) != NULL); - } - -out: - if (skfd >= 0) { - close(skfd); - } - - return res; -} - -static short mif_flags; - -/* Get the inteface configuration from the kernel. */ -static int if_getconfig(char *ifname) -{ - struct ifreq ifr; - int metric, mtu; /* Parameters of the master interface. */ - struct sockaddr dstaddr, broadaddr, netmask; - unsigned char *hwaddr; - - strcpy(ifr.ifr_name, ifname); - if (ioctl(skfd, SIOCGIFFLAGS, &ifr) < 0) - return -1; - mif_flags = ifr.ifr_flags; - printf("The result of SIOCGIFFLAGS on %s is %x.\n", - ifname, ifr.ifr_flags); - - strcpy(ifr.ifr_name, ifname); - if (ioctl(skfd, SIOCGIFADDR, &ifr) < 0) - return -1; - printf("The result of SIOCGIFADDR is %2.2x.%2.2x.%2.2x.%2.2x.\n", - ifr.ifr_addr.sa_data[0], ifr.ifr_addr.sa_data[1], - ifr.ifr_addr.sa_data[2], ifr.ifr_addr.sa_data[3]); - - strcpy(ifr.ifr_name, ifname); - if (ioctl(skfd, SIOCGIFHWADDR, &ifr) < 0) - return -1; - - /* Gotta convert from 'char' to unsigned for printf(). */ - hwaddr = (unsigned char *)ifr.ifr_hwaddr.sa_data; - printf("The result of SIOCGIFHWADDR is type %d " - "%2.2x:%2.2x:%2.2x:%2.2x:%2.2x:%2.2x.\n", - ifr.ifr_hwaddr.sa_family, hwaddr[0], hwaddr[1], - hwaddr[2], hwaddr[3], hwaddr[4], hwaddr[5]); - - strcpy(ifr.ifr_name, ifname); - if (ioctl(skfd, SIOCGIFMETRIC, &ifr) < 0) { - metric = 0; - } else - metric = ifr.ifr_metric; - printf("The result of SIOCGIFMETRIC is %d\n", metric); - - strcpy(ifr.ifr_name, ifname); - if (ioctl(skfd, SIOCGIFMTU, &ifr) < 0) - mtu = 0; - else - mtu = ifr.ifr_mtu; - printf("The result of SIOCGIFMTU is %d\n", mtu); - - strcpy(ifr.ifr_name, ifname); - if (ioctl(skfd, SIOCGIFDSTADDR, &ifr) < 0) { - memset(&dstaddr, 0, sizeof(struct sockaddr)); - } else - dstaddr = ifr.ifr_dstaddr; - - strcpy(ifr.ifr_name, ifname); - if (ioctl(skfd, SIOCGIFBRDADDR, &ifr) < 0) { - memset(&broadaddr, 0, sizeof(struct sockaddr)); - } else - broadaddr = ifr.ifr_broadaddr; - - strcpy(ifr.ifr_name, ifname); - if (ioctl(skfd, SIOCGIFNETMASK, &ifr) < 0) { - memset(&netmask, 0, sizeof(struct sockaddr)); - } else - netmask = ifr.ifr_netmask; - - return 0; -} - -static void if_print(char *ifname) -{ - char buff[1024]; - struct ifconf ifc; - struct ifreq *ifr; - int i; - - if (ifname == (char *)NULL) { - ifc.ifc_len = sizeof(buff); - ifc.ifc_buf = buff; - if (ioctl(skfd, SIOCGIFCONF, &ifc) < 0) { - perror("SIOCGIFCONF failed"); - return; - } - - ifr = ifc.ifc_req; - for (i = ifc.ifc_len / sizeof(struct ifreq); --i >= 0; ifr++) { - if (if_getconfig(ifr->ifr_name) < 0) { - fprintf(stderr, - "%s: unknown interface.\n", - ifr->ifr_name); - continue; - } - - if (((mif_flags & IFF_UP) == 0) && !opt_a) continue; - /*ife_print(&ife);*/ - } - } else { - if (if_getconfig(ifname) < 0) { - fprintf(stderr, - "%s: unknown interface.\n", ifname); - } - } -} - -static int get_drv_info(char *master_ifname) -{ - struct ifreq ifr; - struct ethtool_drvinfo info; - char *endptr; - - memset(&ifr, 0, sizeof(ifr)); - strncpy(ifr.ifr_name, master_ifname, IFNAMSIZ); - ifr.ifr_data = (caddr_t)&info; - - info.cmd = ETHTOOL_GDRVINFO; - strncpy(info.driver, "ifenslave", 32); - snprintf(info.fw_version, 32, "%d", BOND_ABI_VERSION); - - if (ioctl(skfd, SIOCETHTOOL, &ifr) < 0) { - if (errno == EOPNOTSUPP) { - goto out; - } - - saved_errno = errno; - v_print("Master '%s': Error: get bonding info failed %s\n", - master_ifname, strerror(saved_errno)); - return 1; - } - - abi_ver = strtoul(info.fw_version, &endptr, 0); - if (*endptr) { - v_print("Master '%s': Error: got invalid string as an ABI " - "version from the bonding module\n", - master_ifname); - return 1; - } - -out: - v_print("ABI ver is %d\n", abi_ver); - - return 0; -} - -static int change_active(char *master_ifname, char *slave_ifname) -{ - struct ifreq ifr; - int res = 0; - - if (!(slave_flags.ifr_flags & IFF_SLAVE)) { - fprintf(stderr, - "Illegal operation: The specified slave interface " - "'%s' is not a slave\n", - slave_ifname); - return 1; - } - - strncpy(ifr.ifr_name, master_ifname, IFNAMSIZ); - strncpy(ifr.ifr_slave, slave_ifname, IFNAMSIZ); - if ((ioctl(skfd, SIOCBONDCHANGEACTIVE, &ifr) < 0) && - (ioctl(skfd, BOND_CHANGE_ACTIVE_OLD, &ifr) < 0)) { - saved_errno = errno; - v_print("Master '%s': Error: SIOCBONDCHANGEACTIVE failed: " - "%s\n", - master_ifname, strerror(saved_errno)); - res = 1; - } - - return res; -} - -static int enslave(char *master_ifname, char *slave_ifname) -{ - struct ifreq ifr; - int res = 0; - - if (slave_flags.ifr_flags & IFF_SLAVE) { - fprintf(stderr, - "Illegal operation: The specified slave interface " - "'%s' is already a slave\n", - slave_ifname); - return 1; - } - - res = set_if_down(slave_ifname, slave_flags.ifr_flags); - if (res) { - fprintf(stderr, - "Slave '%s': Error: bring interface down failed\n", - slave_ifname); - return res; - } - - if (abi_ver < 2) { - /* Older bonding versions would panic if the slave has no IP - * address, so get the IP setting from the master. - */ - set_if_addr(master_ifname, slave_ifname); - } else { - res = clear_if_addr(slave_ifname); - if (res) { - fprintf(stderr, - "Slave '%s': Error: clear address failed\n", - slave_ifname); - return res; - } - } - - if (master_mtu.ifr_mtu != slave_mtu.ifr_mtu) { - res = set_slave_mtu(slave_ifname, master_mtu.ifr_mtu); - if (res) { - fprintf(stderr, - "Slave '%s': Error: set MTU failed\n", - slave_ifname); - return res; - } - } - - if (hwaddr_set) { - /* Master already has an hwaddr - * so set it's hwaddr to the slave - */ - if (abi_ver < 1) { - /* The driver is using an old ABI, so - * the application sets the slave's - * hwaddr - */ - res = set_slave_hwaddr(slave_ifname, - &(master_hwaddr.ifr_hwaddr)); - if (res) { - fprintf(stderr, - "Slave '%s': Error: set hw address " - "failed\n", - slave_ifname); - goto undo_mtu; - } - - /* For old ABI the application needs to bring the - * slave back up - */ - res = set_if_up(slave_ifname, slave_flags.ifr_flags); - if (res) { - fprintf(stderr, - "Slave '%s': Error: bring interface " - "down failed\n", - slave_ifname); - goto undo_slave_mac; - } - } - /* The driver is using a new ABI, - * so the driver takes care of setting - * the slave's hwaddr and bringing - * it up again - */ - } else { - /* No hwaddr for master yet, so - * set the slave's hwaddr to it - */ - if (abi_ver < 1) { - /* For old ABI, the master needs to be - * down before setting its hwaddr - */ - res = set_if_down(master_ifname, master_flags.ifr_flags); - if (res) { - fprintf(stderr, - "Master '%s': Error: bring interface " - "down failed\n", - master_ifname); - goto undo_mtu; - } - } - - res = set_master_hwaddr(master_ifname, - &(slave_hwaddr.ifr_hwaddr)); - if (res) { - fprintf(stderr, - "Master '%s': Error: set hw address " - "failed\n", - master_ifname); - goto undo_mtu; - } - - if (abi_ver < 1) { - /* For old ABI, bring the master - * back up - */ - res = set_if_up(master_ifname, master_flags.ifr_flags); - if (res) { - fprintf(stderr, - "Master '%s': Error: bring interface " - "up failed\n", - master_ifname); - goto undo_master_mac; - } - } - - hwaddr_set = 1; - } - - /* Do the real thing */ - strncpy(ifr.ifr_name, master_ifname, IFNAMSIZ); - strncpy(ifr.ifr_slave, slave_ifname, IFNAMSIZ); - if ((ioctl(skfd, SIOCBONDENSLAVE, &ifr) < 0) && - (ioctl(skfd, BOND_ENSLAVE_OLD, &ifr) < 0)) { - saved_errno = errno; - v_print("Master '%s': Error: SIOCBONDENSLAVE failed: %s\n", - master_ifname, strerror(saved_errno)); - res = 1; - } - - if (res) { - goto undo_master_mac; - } - - return 0; - -/* rollback (best effort) */ -undo_master_mac: - set_master_hwaddr(master_ifname, &(master_hwaddr.ifr_hwaddr)); - hwaddr_set = 0; - goto undo_mtu; -undo_slave_mac: - set_slave_hwaddr(slave_ifname, &(slave_hwaddr.ifr_hwaddr)); -undo_mtu: - set_slave_mtu(slave_ifname, slave_mtu.ifr_mtu); - return res; -} - -static int release(char *master_ifname, char *slave_ifname) -{ - struct ifreq ifr; - int res = 0; - - if (!(slave_flags.ifr_flags & IFF_SLAVE)) { - fprintf(stderr, - "Illegal operation: The specified slave interface " - "'%s' is not a slave\n", - slave_ifname); - return 1; - } - - strncpy(ifr.ifr_name, master_ifname, IFNAMSIZ); - strncpy(ifr.ifr_slave, slave_ifname, IFNAMSIZ); - if ((ioctl(skfd, SIOCBONDRELEASE, &ifr) < 0) && - (ioctl(skfd, BOND_RELEASE_OLD, &ifr) < 0)) { - saved_errno = errno; - v_print("Master '%s': Error: SIOCBONDRELEASE failed: %s\n", - master_ifname, strerror(saved_errno)); - return 1; - } else if (abi_ver < 1) { - /* The driver is using an old ABI, so we'll set the interface - * down to avoid any conflicts due to same MAC/IP - */ - res = set_if_down(slave_ifname, slave_flags.ifr_flags); - if (res) { - fprintf(stderr, - "Slave '%s': Error: bring interface " - "down failed\n", - slave_ifname); - } - } - - /* set to default mtu */ - set_slave_mtu(slave_ifname, 1500); - - return res; -} - -static int get_if_settings(char *ifname, struct dev_ifr ifra[]) -{ - int i; - int res = 0; - - for (i = 0; ifra[i].req_ifr; i++) { - strncpy(ifra[i].req_ifr->ifr_name, ifname, IFNAMSIZ); - res = ioctl(skfd, ifra[i].req_type, ifra[i].req_ifr); - if (res < 0) { - saved_errno = errno; - v_print("Interface '%s': Error: %s failed: %s\n", - ifname, ifra[i].req_name, - strerror(saved_errno)); - - return saved_errno; - } - } - - return 0; -} - -static int get_slave_flags(char *slave_ifname) -{ - int res = 0; - - strncpy(slave_flags.ifr_name, slave_ifname, IFNAMSIZ); - res = ioctl(skfd, SIOCGIFFLAGS, &slave_flags); - if (res < 0) { - saved_errno = errno; - v_print("Slave '%s': Error: SIOCGIFFLAGS failed: %s\n", - slave_ifname, strerror(saved_errno)); - } else { - v_print("Slave %s: flags %04X.\n", - slave_ifname, slave_flags.ifr_flags); - } - - return res; -} - -static int set_master_hwaddr(char *master_ifname, struct sockaddr *hwaddr) -{ - unsigned char *addr = (unsigned char *)hwaddr->sa_data; - struct ifreq ifr; - int res = 0; - - strncpy(ifr.ifr_name, master_ifname, IFNAMSIZ); - memcpy(&(ifr.ifr_hwaddr), hwaddr, sizeof(struct sockaddr)); - res = ioctl(skfd, SIOCSIFHWADDR, &ifr); - if (res < 0) { - saved_errno = errno; - v_print("Master '%s': Error: SIOCSIFHWADDR failed: %s\n", - master_ifname, strerror(saved_errno)); - return res; - } else { - v_print("Master '%s': hardware address set to " - "%2.2x:%2.2x:%2.2x:%2.2x:%2.2x:%2.2x.\n", - master_ifname, addr[0], addr[1], addr[2], - addr[3], addr[4], addr[5]); - } - - return res; -} - -static int set_slave_hwaddr(char *slave_ifname, struct sockaddr *hwaddr) -{ - unsigned char *addr = (unsigned char *)hwaddr->sa_data; - struct ifreq ifr; - int res = 0; - - strncpy(ifr.ifr_name, slave_ifname, IFNAMSIZ); - memcpy(&(ifr.ifr_hwaddr), hwaddr, sizeof(struct sockaddr)); - res = ioctl(skfd, SIOCSIFHWADDR, &ifr); - if (res < 0) { - saved_errno = errno; - - v_print("Slave '%s': Error: SIOCSIFHWADDR failed: %s\n", - slave_ifname, strerror(saved_errno)); - - if (saved_errno == EBUSY) { - v_print(" The device is busy: it must be idle " - "before running this command.\n"); - } else if (saved_errno == EOPNOTSUPP) { - v_print(" The device does not support setting " - "the MAC address.\n" - " Your kernel likely does not support slave " - "devices.\n"); - } else if (saved_errno == EINVAL) { - v_print(" The device's address type does not match " - "the master's address type.\n"); - } - return res; - } else { - v_print("Slave '%s': hardware address set to " - "%2.2x:%2.2x:%2.2x:%2.2x:%2.2x:%2.2x.\n", - slave_ifname, addr[0], addr[1], addr[2], - addr[3], addr[4], addr[5]); - } - - return res; -} - -static int set_slave_mtu(char *slave_ifname, int mtu) -{ - struct ifreq ifr; - int res = 0; - - ifr.ifr_mtu = mtu; - strncpy(ifr.ifr_name, slave_ifname, IFNAMSIZ); - - res = ioctl(skfd, SIOCSIFMTU, &ifr); - if (res < 0) { - saved_errno = errno; - v_print("Slave '%s': Error: SIOCSIFMTU failed: %s\n", - slave_ifname, strerror(saved_errno)); - } else { - v_print("Slave '%s': MTU set to %d.\n", slave_ifname, mtu); - } - - return res; -} - -static int set_if_flags(char *ifname, short flags) -{ - struct ifreq ifr; - int res = 0; - - ifr.ifr_flags = flags; - strncpy(ifr.ifr_name, ifname, IFNAMSIZ); - - res = ioctl(skfd, SIOCSIFFLAGS, &ifr); - if (res < 0) { - saved_errno = errno; - v_print("Interface '%s': Error: SIOCSIFFLAGS failed: %s\n", - ifname, strerror(saved_errno)); - } else { - v_print("Interface '%s': flags set to %04X.\n", ifname, flags); - } - - return res; -} - -static int set_if_up(char *ifname, short flags) -{ - return set_if_flags(ifname, flags | IFF_UP); -} - -static int set_if_down(char *ifname, short flags) -{ - return set_if_flags(ifname, flags & ~IFF_UP); -} - -static int clear_if_addr(char *ifname) -{ - struct ifreq ifr; - int res = 0; - - strncpy(ifr.ifr_name, ifname, IFNAMSIZ); - ifr.ifr_addr.sa_family = AF_INET; - memset(ifr.ifr_addr.sa_data, 0, sizeof(ifr.ifr_addr.sa_data)); - - res = ioctl(skfd, SIOCSIFADDR, &ifr); - if (res < 0) { - saved_errno = errno; - v_print("Interface '%s': Error: SIOCSIFADDR failed: %s\n", - ifname, strerror(saved_errno)); - } else { - v_print("Interface '%s': address cleared\n", ifname); - } - - return res; -} - -static int set_if_addr(char *master_ifname, char *slave_ifname) -{ - struct ifreq ifr; - int res; - unsigned char *ipaddr; - int i; - struct { - char *req_name; - char *desc; - int g_ioctl; - int s_ioctl; - } ifra[] = { - {"IFADDR", "addr", SIOCGIFADDR, SIOCSIFADDR}, - {"DSTADDR", "destination addr", SIOCGIFDSTADDR, SIOCSIFDSTADDR}, - {"BRDADDR", "broadcast addr", SIOCGIFBRDADDR, SIOCSIFBRDADDR}, - {"NETMASK", "netmask", SIOCGIFNETMASK, SIOCSIFNETMASK}, - {NULL, NULL, 0, 0}, - }; - - for (i = 0; ifra[i].req_name; i++) { - strncpy(ifr.ifr_name, master_ifname, IFNAMSIZ); - res = ioctl(skfd, ifra[i].g_ioctl, &ifr); - if (res < 0) { - int saved_errno = errno; - - v_print("Interface '%s': Error: SIOCG%s failed: %s\n", - master_ifname, ifra[i].req_name, - strerror(saved_errno)); - - ifr.ifr_addr.sa_family = AF_INET; - memset(ifr.ifr_addr.sa_data, 0, - sizeof(ifr.ifr_addr.sa_data)); - } - - strncpy(ifr.ifr_name, slave_ifname, IFNAMSIZ); - res = ioctl(skfd, ifra[i].s_ioctl, &ifr); - if (res < 0) { - int saved_errno = errno; - - v_print("Interface '%s': Error: SIOCS%s failed: %s\n", - slave_ifname, ifra[i].req_name, - strerror(saved_errno)); - - } - - ipaddr = (unsigned char *)ifr.ifr_addr.sa_data; - v_print("Interface '%s': set IP %s to %d.%d.%d.%d\n", - slave_ifname, ifra[i].desc, - ipaddr[0], ipaddr[1], ipaddr[2], ipaddr[3]); - } - - return 0; -} - -/* - * Local variables: - * version-control: t - * kept-new-versions: 5 - * c-indent-level: 4 - * c-basic-offset: 4 - * tab-width: 4 - * compile-command: "gcc -Wall -Wstrict-prototypes -O -I/usr/src/linux/include ifenslave.c -o ifenslave" - * End: - */ - diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt index aa68f3c630c..10742902146 100644 --- a/Documentation/networking/ip-sysctl.txt +++ b/Documentation/networking/ip-sysctl.txt @@ -685,6 +685,15 @@ ip_dynaddr - BOOLEAN occurs. Default: 0 +ip_early_demux - BOOLEAN + Optimize input packet processing down to one demux for + certain kinds of local sockets. Currently we only do this + for established TCP sockets. + + It may add an additional cost for pure routing workloads that + reduces overall throughput, in such case you should disable it. + Default: 1 + icmp_echo_ignore_all - BOOLEAN If set non-zero, then the kernel will ignore all ICMP ECHO requests sent to it. @@ -729,7 +738,7 @@ icmp_ignore_bogus_error_responses - BOOLEAN frames. Such violations are normally logged via a kernel warning. If this is set to TRUE, the kernel will not give such warnings, which will avoid log file clutter. - Default: FALSE + Default: 1 icmp_errors_use_inbound_ifaddr - BOOLEAN diff --git a/Documentation/networking/ipvs-sysctl.txt b/Documentation/networking/ipvs-sysctl.txt index 9573d0c48c6..7a3c0472959 100644 --- a/Documentation/networking/ipvs-sysctl.txt +++ b/Documentation/networking/ipvs-sysctl.txt @@ -181,6 +181,19 @@ snat_reroute - BOOLEAN always be the same as the original route so it is an optimisation to disable snat_reroute and avoid the recalculation. +sync_persist_mode - INTEGER + default 0 + + Controls the synchronisation of connections when using persistence + + 0: All types of connections are synchronised + 1: Attempt to reduce the synchronisation traffic depending on + the connection type. For persistent services avoid synchronisation + for normal connections, do it only for persistence templates. + In such case, for TCP and SCTP it may need enabling sloppy_tcp and + sloppy_sctp flags on backup servers. For non-persistent services + such optimization is not applied, mode 0 is assumed. + sync_version - INTEGER default 1 diff --git a/Documentation/networking/netlink_mmap.txt b/Documentation/networking/netlink_mmap.txt index 9bd0f5211e9..53337883954 100644 --- a/Documentation/networking/netlink_mmap.txt +++ b/Documentation/networking/netlink_mmap.txt @@ -114,7 +114,7 @@ Some parameters are constrained, specifically: - nm_frame_nr must equal the actual number of frames as specified above. When the kernel can't allocate physically continuous memory for a ring block, -it will fall back to use physically discontinous memory. This might affect +it will fall back to use physically discontinuous memory. This might affect performance negatively, in order to avoid this the nm_frame_size parameter should be chosen to be as small as possible for the required frame size and the number of blocks should be increased instead. @@ -274,9 +274,9 @@ This example assumes some ring parameters of the ring setup are available. /* Get next frame header */ hdr = rx_ring + frame_offset; - if (hdr->nm_status == NL_MMAP_STATUS_VALID) + if (hdr->nm_status == NL_MMAP_STATUS_VALID) { /* Regular memory mapped frame */ - nlh = (void *hdr) + NL_MMAP_HDRLEN; + nlh = (void *)hdr + NL_MMAP_HDRLEN; len = hdr->nm_len; /* Release empty message immediately. May happen diff --git a/Documentation/networking/packet_mmap.txt b/Documentation/networking/packet_mmap.txt index 23dd80e82b8..8572796b1eb 100644 --- a/Documentation/networking/packet_mmap.txt +++ b/Documentation/networking/packet_mmap.txt @@ -704,6 +704,12 @@ So it seems to be a good candidate to be used with packet fanout. Minimal example code by Daniel Borkmann based on Chetan Loke's lolpcap (compile it with gcc -Wall -O2 blob.c, and try things like "./a.out eth0", etc.): +/* Written from scratch, but kernel-to-user space API usage + * dissected from lolpcap: + * Copyright 2011, Chetan Loke <loke.chetan@gmail.com> + * License: GPL, version 2.0 + */ + #include <stdio.h> #include <stdlib.h> #include <stdint.h> @@ -722,27 +728,6 @@ it with gcc -Wall -O2 blob.c, and try things like "./a.out eth0", etc.): #include <linux/if_ether.h> #include <linux/ip.h> -#define BLOCK_SIZE (1 << 22) -#define FRAME_SIZE 2048 - -#define NUM_BLOCKS 64 -#define NUM_FRAMES ((BLOCK_SIZE * NUM_BLOCKS) / FRAME_SIZE) - -#define BLOCK_RETIRE_TOV_IN_MS 64 -#define BLOCK_PRIV_AREA_SZ 13 - -#define ALIGN_8(x) (((x) + 8 - 1) & ~(8 - 1)) - -#define BLOCK_STATUS(x) ((x)->h1.block_status) -#define BLOCK_NUM_PKTS(x) ((x)->h1.num_pkts) -#define BLOCK_O2FP(x) ((x)->h1.offset_to_first_pkt) -#define BLOCK_LEN(x) ((x)->h1.blk_len) -#define BLOCK_SNUM(x) ((x)->h1.seq_num) -#define BLOCK_O2PRIV(x) ((x)->offset_to_priv) -#define BLOCK_PRIV(x) ((void *) ((uint8_t *) (x) + BLOCK_O2PRIV(x))) -#define BLOCK_HDR_LEN (ALIGN_8(sizeof(struct block_desc))) -#define BLOCK_PLUS_PRIV(sz_pri) (BLOCK_HDR_LEN + ALIGN_8((sz_pri))) - #ifndef likely # define likely(x) __builtin_expect(!!(x), 1) #endif @@ -765,7 +750,7 @@ struct ring { static unsigned long packets_total = 0, bytes_total = 0; static sig_atomic_t sigint = 0; -void sighandler(int num) +static void sighandler(int num) { sigint = 1; } @@ -774,6 +759,8 @@ static int setup_socket(struct ring *ring, char *netdev) { int err, i, fd, v = TPACKET_V3; struct sockaddr_ll ll; + unsigned int blocksiz = 1 << 22, framesiz = 1 << 11; + unsigned int blocknum = 64; fd = socket(AF_PACKET, SOCK_RAW, htons(ETH_P_ALL)); if (fd < 0) { @@ -788,13 +775,12 @@ static int setup_socket(struct ring *ring, char *netdev) } memset(&ring->req, 0, sizeof(ring->req)); - ring->req.tp_block_size = BLOCK_SIZE; - ring->req.tp_frame_size = FRAME_SIZE; - ring->req.tp_block_nr = NUM_BLOCKS; - ring->req.tp_frame_nr = NUM_FRAMES; - ring->req.tp_retire_blk_tov = BLOCK_RETIRE_TOV_IN_MS; - ring->req.tp_sizeof_priv = BLOCK_PRIV_AREA_SZ; - ring->req.tp_feature_req_word |= TP_FT_REQ_FILL_RXHASH; + ring->req.tp_block_size = blocksiz; + ring->req.tp_frame_size = framesiz; + ring->req.tp_block_nr = blocknum; + ring->req.tp_frame_nr = (blocksiz * blocknum) / framesiz; + ring->req.tp_retire_blk_tov = 60; + ring->req.tp_feature_req_word = TP_FT_REQ_FILL_RXHASH; err = setsockopt(fd, SOL_PACKET, PACKET_RX_RING, &ring->req, sizeof(ring->req)); @@ -804,8 +790,7 @@ static int setup_socket(struct ring *ring, char *netdev) } ring->map = mmap(NULL, ring->req.tp_block_size * ring->req.tp_block_nr, - PROT_READ | PROT_WRITE, MAP_SHARED | MAP_LOCKED, - fd, 0); + PROT_READ | PROT_WRITE, MAP_SHARED | MAP_LOCKED, fd, 0); if (ring->map == MAP_FAILED) { perror("mmap"); exit(1); @@ -835,58 +820,6 @@ static int setup_socket(struct ring *ring, char *netdev) return fd; } -#ifdef __checked -static uint64_t prev_block_seq_num = 0; - -void assert_block_seq_num(struct block_desc *pbd) -{ - if (unlikely(prev_block_seq_num + 1 != BLOCK_SNUM(pbd))) { - printf("prev_block_seq_num:%"PRIu64", expected seq:%"PRIu64" != " - "actual seq:%"PRIu64"\n", prev_block_seq_num, - prev_block_seq_num + 1, (uint64_t) BLOCK_SNUM(pbd)); - exit(1); - } - - prev_block_seq_num = BLOCK_SNUM(pbd); -} - -static void assert_block_len(struct block_desc *pbd, uint32_t bytes, int block_num) -{ - if (BLOCK_NUM_PKTS(pbd)) { - if (unlikely(bytes != BLOCK_LEN(pbd))) { - printf("block:%u with %upackets, expected len:%u != actual len:%u\n", - block_num, BLOCK_NUM_PKTS(pbd), bytes, BLOCK_LEN(pbd)); - exit(1); - } - } else { - if (unlikely(BLOCK_LEN(pbd) != BLOCK_PLUS_PRIV(BLOCK_PRIV_AREA_SZ))) { - printf("block:%u, expected len:%lu != actual len:%u\n", - block_num, BLOCK_HDR_LEN, BLOCK_LEN(pbd)); - exit(1); - } - } -} - -static void assert_block_header(struct block_desc *pbd, const int block_num) -{ - uint32_t block_status = BLOCK_STATUS(pbd); - - if (unlikely((block_status & TP_STATUS_USER) == 0)) { - printf("block:%u, not in TP_STATUS_USER\n", block_num); - exit(1); - } - - assert_block_seq_num(pbd); -} -#else -static inline void assert_block_header(struct block_desc *pbd, const int block_num) -{ -} -static void assert_block_len(struct block_desc *pbd, uint32_t bytes, int block_num) -{ -} -#endif - static void display(struct tpacket3_hdr *ppd) { struct ethhdr *eth = (struct ethhdr *) ((uint8_t *) ppd + ppd->tp_mac); @@ -916,37 +849,27 @@ static void display(struct tpacket3_hdr *ppd) static void walk_block(struct block_desc *pbd, const int block_num) { - int num_pkts = BLOCK_NUM_PKTS(pbd), i; + int num_pkts = pbd->h1.num_pkts, i; unsigned long bytes = 0; - unsigned long bytes_with_padding = BLOCK_PLUS_PRIV(BLOCK_PRIV_AREA_SZ); struct tpacket3_hdr *ppd; - assert_block_header(pbd, block_num); - - ppd = (struct tpacket3_hdr *) ((uint8_t *) pbd + BLOCK_O2FP(pbd)); + ppd = (struct tpacket3_hdr *) ((uint8_t *) pbd + + pbd->h1.offset_to_first_pkt); for (i = 0; i < num_pkts; ++i) { bytes += ppd->tp_snaplen; - if (ppd->tp_next_offset) - bytes_with_padding += ppd->tp_next_offset; - else - bytes_with_padding += ALIGN_8(ppd->tp_snaplen + ppd->tp_mac); - display(ppd); - ppd = (struct tpacket3_hdr *) ((uint8_t *) ppd + ppd->tp_next_offset); - __sync_synchronize(); + ppd = (struct tpacket3_hdr *) ((uint8_t *) ppd + + ppd->tp_next_offset); } - assert_block_len(pbd, bytes_with_padding, block_num); - packets_total += num_pkts; bytes_total += bytes; } -void flush_block(struct block_desc *pbd) +static void flush_block(struct block_desc *pbd) { - BLOCK_STATUS(pbd) = TP_STATUS_KERNEL; - __sync_synchronize(); + pbd->h1.block_status = TP_STATUS_KERNEL; } static void teardown_socket(struct ring *ring, int fd) @@ -962,7 +885,7 @@ int main(int argc, char **argp) socklen_t len; struct ring ring; struct pollfd pfd; - unsigned int block_num = 0; + unsigned int block_num = 0, blocks = 64; struct block_desc *pbd; struct tpacket_stats_v3 stats; @@ -984,15 +907,15 @@ int main(int argc, char **argp) while (likely(!sigint)) { pbd = (struct block_desc *) ring.rd[block_num].iov_base; -retry_block: - if ((BLOCK_STATUS(pbd) & TP_STATUS_USER) == 0) { + + if ((pbd->h1.block_status & TP_STATUS_USER) == 0) { poll(&pfd, 1, -1); - goto retry_block; + continue; } walk_block(pbd, block_num); flush_block(pbd); - block_num = (block_num + 1) % NUM_BLOCKS; + block_num = (block_num + 1) % blocks; } len = sizeof(stats); diff --git a/Documentation/networking/scaling.txt b/Documentation/networking/scaling.txt index 579994afbe0..ca6977f5b2e 100644 --- a/Documentation/networking/scaling.txt +++ b/Documentation/networking/scaling.txt @@ -163,6 +163,64 @@ and unnecessary. If there are fewer hardware queues than CPUs, then RPS might be beneficial if the rps_cpus for each queue are the ones that share the same memory domain as the interrupting CPU for that queue. +==== RPS Flow Limit + +RPS scales kernel receive processing across CPUs without introducing +reordering. The trade-off to sending all packets from the same flow +to the same CPU is CPU load imbalance if flows vary in packet rate. +In the extreme case a single flow dominates traffic. Especially on +common server workloads with many concurrent connections, such +behavior indicates a problem such as a misconfiguration or spoofed +source Denial of Service attack. + +Flow Limit is an optional RPS feature that prioritizes small flows +during CPU contention by dropping packets from large flows slightly +ahead of those from small flows. It is active only when an RPS or RFS +destination CPU approaches saturation. Once a CPU's input packet +queue exceeds half the maximum queue length (as set by sysctl +net.core.netdev_max_backlog), the kernel starts a per-flow packet +count over the last 256 packets. If a flow exceeds a set ratio (by +default, half) of these packets when a new packet arrives, then the +new packet is dropped. Packets from other flows are still only +dropped once the input packet queue reaches netdev_max_backlog. +No packets are dropped when the input packet queue length is below +the threshold, so flow limit does not sever connections outright: +even large flows maintain connectivity. + +== Interface + +Flow limit is compiled in by default (CONFIG_NET_FLOW_LIMIT), but not +turned on. It is implemented for each CPU independently (to avoid lock +and cache contention) and toggled per CPU by setting the relevant bit +in sysctl net.core.flow_limit_cpu_bitmap. It exposes the same CPU +bitmap interface as rps_cpus (see above) when called from procfs: + + /proc/sys/net/core/flow_limit_cpu_bitmap + +Per-flow rate is calculated by hashing each packet into a hashtable +bucket and incrementing a per-bucket counter. The hash function is +the same that selects a CPU in RPS, but as the number of buckets can +be much larger than the number of CPUs, flow limit has finer-grained +identification of large flows and fewer false positives. The default +table has 4096 buckets. This value can be modified through sysctl + + net.core.flow_limit_table_len + +The value is only consulted when a new table is allocated. Modifying +it does not update active tables. + +== Suggested Configuration + +Flow limit is useful on systems with many concurrent connections, +where a single connection taking up 50% of a CPU indicates a problem. +In such environments, enable the feature on all CPUs that handle +network rx interrupts (as set in /proc/irq/N/smp_affinity). + +The feature depends on the input packet queue length to exceed +the flow limit threshold (50%) + the flow history length (256). +Setting net.core.netdev_max_backlog to either 1000 or 10000 +performed well in experiments. + RFS: Receive Flow Steering ========================== diff --git a/Documentation/networking/vortex.txt b/Documentation/networking/vortex.txt index b4038ffb3bc..9a8041dcbb5 100644 --- a/Documentation/networking/vortex.txt +++ b/Documentation/networking/vortex.txt @@ -359,7 +359,7 @@ steps you should take: - OK, it's a driver problem. You need to generate a report. Typically this is an email to the - maintainer and/or linux-net@vger.kernel.org. The maintainer's + maintainer and/or netdev@vger.kernel.org. The maintainer's email address will be in the driver source or in the MAINTAINERS file. - The contents of your report will vary a lot depending upon the |