netlink/addr_linux.go
Dave Setzke 229a10237c Add ReceiveBufferSize and force option to *Subscribe
When there are a large number of existing results for the link, neighbor, and
address subscribe functions with ListExisting are likely to fail with ENOBUFS.
This takes the AddrSubscribeOptions ReceiveBufferSize, already applied
to LinkSubscribeOptions, and applies it to NeighSubscribeOptions and
RouteSubscribeOptions. The ReceiveTimeout option was also added to each.

Added a SetReceiveBufferSize to the nl_linux socket API.

The existing addr_linux subscribe function was modified so instead of setting
the ReceiveBufferSize on the netlink pkghandle, it is set on the socket
associated with the subscription. The new implementations also only change the
receive buffer size on the socket.

Lastly, a new ReceiveBufferForceSize option was applied to all four of the
modified Subscribe functions.
2023-09-21 19:50:55 -07:00

424 lines
12 KiB
Go

package netlink
import (
"fmt"
"net"
"strings"
"syscall"
"github.com/vishvananda/netlink/nl"
"github.com/vishvananda/netns"
"golang.org/x/sys/unix"
)
// AddrAdd will add an IP address to a link device.
//
// Equivalent to: `ip addr add $addr dev $link`
//
// If `addr` is an IPv4 address and the broadcast address is not given, it
// will be automatically computed based on the IP mask if /30 or larger.
func AddrAdd(link Link, addr *Addr) error {
return pkgHandle.AddrAdd(link, addr)
}
// AddrAdd will add an IP address to a link device.
//
// Equivalent to: `ip addr add $addr dev $link`
//
// If `addr` is an IPv4 address and the broadcast address is not given, it
// will be automatically computed based on the IP mask if /30 or larger.
func (h *Handle) AddrAdd(link Link, addr *Addr) error {
req := h.newNetlinkRequest(unix.RTM_NEWADDR, unix.NLM_F_CREATE|unix.NLM_F_EXCL|unix.NLM_F_ACK)
return h.addrHandle(link, addr, req)
}
// AddrReplace will replace (or, if not present, add) an IP address on a link device.
//
// Equivalent to: `ip addr replace $addr dev $link`
//
// If `addr` is an IPv4 address and the broadcast address is not given, it
// will be automatically computed based on the IP mask if /30 or larger.
func AddrReplace(link Link, addr *Addr) error {
return pkgHandle.AddrReplace(link, addr)
}
// AddrReplace will replace (or, if not present, add) an IP address on a link device.
//
// Equivalent to: `ip addr replace $addr dev $link`
//
// If `addr` is an IPv4 address and the broadcast address is not given, it
// will be automatically computed based on the IP mask if /30 or larger.
func (h *Handle) AddrReplace(link Link, addr *Addr) error {
req := h.newNetlinkRequest(unix.RTM_NEWADDR, unix.NLM_F_CREATE|unix.NLM_F_REPLACE|unix.NLM_F_ACK)
return h.addrHandle(link, addr, req)
}
// AddrDel will delete an IP address from a link device.
//
// Equivalent to: `ip addr del $addr dev $link`
//
// If `addr` is an IPv4 address and the broadcast address is not given, it
// will be automatically computed based on the IP mask if /30 or larger.
func AddrDel(link Link, addr *Addr) error {
return pkgHandle.AddrDel(link, addr)
}
// AddrDel will delete an IP address from a link device.
// Equivalent to: `ip addr del $addr dev $link`
//
// If `addr` is an IPv4 address and the broadcast address is not given, it
// will be automatically computed based on the IP mask if /30 or larger.
func (h *Handle) AddrDel(link Link, addr *Addr) error {
req := h.newNetlinkRequest(unix.RTM_DELADDR, unix.NLM_F_ACK)
return h.addrHandle(link, addr, req)
}
func (h *Handle) addrHandle(link Link, addr *Addr, req *nl.NetlinkRequest) error {
base := link.Attrs()
if addr.Label != "" && !strings.HasPrefix(addr.Label, base.Name) {
return fmt.Errorf("label must begin with interface name")
}
h.ensureIndex(base)
family := nl.GetIPFamily(addr.IP)
msg := nl.NewIfAddrmsg(family)
msg.Index = uint32(base.Index)
msg.Scope = uint8(addr.Scope)
mask := addr.Mask
if addr.Peer != nil {
mask = addr.Peer.Mask
}
prefixlen, masklen := mask.Size()
msg.Prefixlen = uint8(prefixlen)
req.AddData(msg)
var localAddrData []byte
if family == FAMILY_V4 {
localAddrData = addr.IP.To4()
} else {
localAddrData = addr.IP.To16()
}
localData := nl.NewRtAttr(unix.IFA_LOCAL, localAddrData)
req.AddData(localData)
var peerAddrData []byte
if addr.Peer != nil {
if family == FAMILY_V4 {
peerAddrData = addr.Peer.IP.To4()
} else {
peerAddrData = addr.Peer.IP.To16()
}
} else {
peerAddrData = localAddrData
}
addressData := nl.NewRtAttr(unix.IFA_ADDRESS, peerAddrData)
req.AddData(addressData)
if addr.Flags != 0 {
if addr.Flags <= 0xff {
msg.IfAddrmsg.Flags = uint8(addr.Flags)
} else {
b := make([]byte, 4)
native.PutUint32(b, uint32(addr.Flags))
flagsData := nl.NewRtAttr(unix.IFA_FLAGS, b)
req.AddData(flagsData)
}
}
if family == FAMILY_V4 {
// Automatically set the broadcast address if it is unset and the
// subnet is large enough to sensibly have one (/30 or larger).
// See: RFC 3021
if addr.Broadcast == nil && prefixlen < 31 {
calcBroadcast := make(net.IP, masklen/8)
for i := range localAddrData {
calcBroadcast[i] = localAddrData[i] | ^mask[i]
}
addr.Broadcast = calcBroadcast
}
if addr.Broadcast != nil {
req.AddData(nl.NewRtAttr(unix.IFA_BROADCAST, addr.Broadcast))
}
if addr.Label != "" {
labelData := nl.NewRtAttr(unix.IFA_LABEL, nl.ZeroTerminated(addr.Label))
req.AddData(labelData)
}
}
// 0 is the default value for these attributes. However, 0 means "expired", while the least-surprising default
// value should be "forever". To compensate for that, only add the attributes if at least one of the values is
// non-zero, which means the caller has explicitly set them
if addr.ValidLft > 0 || addr.PreferedLft > 0 {
cachedata := nl.IfaCacheInfo{unix.IfaCacheinfo{
Valid: uint32(addr.ValidLft),
Prefered: uint32(addr.PreferedLft),
}}
req.AddData(nl.NewRtAttr(unix.IFA_CACHEINFO, cachedata.Serialize()))
}
_, err := req.Execute(unix.NETLINK_ROUTE, 0)
return err
}
// AddrList gets a list of IP addresses in the system.
// Equivalent to: `ip addr show`.
// The list can be filtered by link and ip family.
func AddrList(link Link, family int) ([]Addr, error) {
return pkgHandle.AddrList(link, family)
}
// AddrList gets a list of IP addresses in the system.
// Equivalent to: `ip addr show`.
// The list can be filtered by link and ip family.
func (h *Handle) AddrList(link Link, family int) ([]Addr, error) {
req := h.newNetlinkRequest(unix.RTM_GETADDR, unix.NLM_F_DUMP)
msg := nl.NewIfAddrmsg(family)
req.AddData(msg)
msgs, err := req.Execute(unix.NETLINK_ROUTE, unix.RTM_NEWADDR)
if err != nil {
return nil, err
}
indexFilter := 0
if link != nil {
base := link.Attrs()
h.ensureIndex(base)
indexFilter = base.Index
}
var res []Addr
for _, m := range msgs {
addr, msgFamily, err := parseAddr(m)
if err != nil {
return res, err
}
if link != nil && addr.LinkIndex != indexFilter {
// Ignore messages from other interfaces
continue
}
if family != FAMILY_ALL && msgFamily != family {
continue
}
res = append(res, addr)
}
return res, nil
}
func parseAddr(m []byte) (addr Addr, family int, err error) {
msg := nl.DeserializeIfAddrmsg(m)
family = -1
addr.LinkIndex = -1
attrs, err1 := nl.ParseRouteAttr(m[msg.Len():])
if err1 != nil {
err = err1
return
}
family = int(msg.Family)
addr.LinkIndex = int(msg.Index)
var local, dst *net.IPNet
for _, attr := range attrs {
switch attr.Attr.Type {
case unix.IFA_ADDRESS:
dst = &net.IPNet{
IP: attr.Value,
Mask: net.CIDRMask(int(msg.Prefixlen), 8*len(attr.Value)),
}
case unix.IFA_LOCAL:
// iproute2 manual:
// If a peer address is specified, the local address
// cannot have a prefix length. The network prefix is
// associated with the peer rather than with the local
// address.
n := 8 * len(attr.Value)
local = &net.IPNet{
IP: attr.Value,
Mask: net.CIDRMask(n, n),
}
case unix.IFA_BROADCAST:
addr.Broadcast = attr.Value
case unix.IFA_LABEL:
addr.Label = string(attr.Value[:len(attr.Value)-1])
case unix.IFA_FLAGS:
addr.Flags = int(native.Uint32(attr.Value[0:4]))
case unix.IFA_CACHEINFO:
ci := nl.DeserializeIfaCacheInfo(attr.Value)
addr.PreferedLft = int(ci.Prefered)
addr.ValidLft = int(ci.Valid)
}
}
// libnl addr.c comment:
// IPv6 sends the local address as IFA_ADDRESS with no
// IFA_LOCAL, IPv4 sends both IFA_LOCAL and IFA_ADDRESS
// with IFA_ADDRESS being the peer address if they differ
//
// But obviously, as there are IPv6 PtP addresses, too,
// IFA_LOCAL should also be handled for IPv6.
if local != nil {
if family == FAMILY_V4 && dst != nil && local.IP.Equal(dst.IP) {
addr.IPNet = dst
} else {
addr.IPNet = local
addr.Peer = dst
}
} else {
addr.IPNet = dst
}
addr.Scope = int(msg.Scope)
return
}
type AddrUpdate struct {
LinkAddress net.IPNet
LinkIndex int
Flags int
Scope int
PreferedLft int
ValidLft int
NewAddr bool // true=added false=deleted
}
// AddrSubscribe takes a chan down which notifications will be sent
// when addresses change. Close the 'done' chan to stop subscription.
func AddrSubscribe(ch chan<- AddrUpdate, done <-chan struct{}) error {
return addrSubscribeAt(netns.None(), netns.None(), ch, done, nil, false, 0, nil, false)
}
// AddrSubscribeAt works like AddrSubscribe plus it allows the caller
// to choose the network namespace in which to subscribe (ns).
func AddrSubscribeAt(ns netns.NsHandle, ch chan<- AddrUpdate, done <-chan struct{}) error {
return addrSubscribeAt(ns, netns.None(), ch, done, nil, false, 0, nil, false)
}
// AddrSubscribeOptions contains a set of options to use with
// AddrSubscribeWithOptions.
type AddrSubscribeOptions struct {
Namespace *netns.NsHandle
ErrorCallback func(error)
ListExisting bool
ReceiveBufferSize int
ReceiveBufferForceSize bool
ReceiveTimeout *unix.Timeval
}
// AddrSubscribeWithOptions work like AddrSubscribe but enable to
// provide additional options to modify the behavior. Currently, the
// namespace can be provided as well as an error callback.
func AddrSubscribeWithOptions(ch chan<- AddrUpdate, done <-chan struct{}, options AddrSubscribeOptions) error {
if options.Namespace == nil {
none := netns.None()
options.Namespace = &none
}
return addrSubscribeAt(*options.Namespace, netns.None(), ch, done, options.ErrorCallback, options.ListExisting,
options.ReceiveBufferSize, options.ReceiveTimeout, options.ReceiveBufferForceSize)
}
func addrSubscribeAt(newNs, curNs netns.NsHandle, ch chan<- AddrUpdate, done <-chan struct{}, cberr func(error), listExisting bool,
rcvbuf int, rcvTimeout *unix.Timeval, rcvBufForce bool) error {
s, err := nl.SubscribeAt(newNs, curNs, unix.NETLINK_ROUTE, unix.RTNLGRP_IPV4_IFADDR, unix.RTNLGRP_IPV6_IFADDR)
if err != nil {
return err
}
if rcvTimeout != nil {
if err := s.SetReceiveTimeout(rcvTimeout); err != nil {
return err
}
}
if rcvbuf != 0 {
err = s.SetReceiveBufferSize(rcvbuf, rcvBufForce)
if err != nil {
return err
}
}
if done != nil {
go func() {
<-done
s.Close()
}()
}
if listExisting {
req := pkgHandle.newNetlinkRequest(unix.RTM_GETADDR,
unix.NLM_F_DUMP)
infmsg := nl.NewIfInfomsg(unix.AF_UNSPEC)
req.AddData(infmsg)
if err := s.Send(req); err != nil {
return err
}
}
go func() {
defer close(ch)
for {
msgs, from, err := s.Receive()
if err != nil {
if cberr != nil {
cberr(fmt.Errorf("Receive failed: %v",
err))
}
return
}
if from.Pid != nl.PidKernel {
if cberr != nil {
cberr(fmt.Errorf("Wrong sender portid %d, expected %d", from.Pid, nl.PidKernel))
}
continue
}
for _, m := range msgs {
if m.Header.Type == unix.NLMSG_DONE {
continue
}
if m.Header.Type == unix.NLMSG_ERROR {
error := int32(native.Uint32(m.Data[0:4]))
if error == 0 {
continue
}
if cberr != nil {
cberr(fmt.Errorf("error message: %v",
syscall.Errno(-error)))
}
continue
}
msgType := m.Header.Type
if msgType != unix.RTM_NEWADDR && msgType != unix.RTM_DELADDR {
if cberr != nil {
cberr(fmt.Errorf("bad message type: %d", msgType))
}
continue
}
addr, _, err := parseAddr(m.Data)
if err != nil {
if cberr != nil {
cberr(fmt.Errorf("could not parse address: %v", err))
}
continue
}
ch <- AddrUpdate{LinkAddress: *addr.IPNet,
LinkIndex: addr.LinkIndex,
NewAddr: msgType == unix.RTM_NEWADDR,
Flags: addr.Flags,
Scope: addr.Scope,
PreferedLft: addr.PreferedLft,
ValidLft: addr.ValidLft}
}
}
}()
return nil
}