CNI: bump to 1.0.1 (#297)

* CNI: bump to 1.0.1

This commit bumps the declared version of CNI in the Kilo manifests to
1.0.1. This is possible with no changes to the configuration lists
because our simple configuration is not affected by any of the
deprecations, and there was effectively no change between 0.4.0 and
1.0.0, other than the declaration of a stable API. Similarly, this
commit also bumps the version of the CNI library and the plugins
package.

Bumping to CNI 1.0.0 will help ensure that Kilo stays compatible with
container runtimes in the future.

Signed-off-by: Lucas Servén Marín <lserven@gmail.com>

* vendor: revendor

Signed-off-by: Lucas Servén Marín <lserven@gmail.com>
This commit is contained in:
Lucas Servén Marín
2022-04-18 19:00:37 +02:00
committed by GitHub
parent 6862274e8e
commit 8cadff2b79
129 changed files with 10937 additions and 1892 deletions

1
vendor/github.com/vishvananda/netlink/.gitignore generated vendored Normal file
View File

@@ -0,0 +1 @@
.idea/

View File

@@ -1,4 +1,8 @@
language: go
go:
- "1.12.x"
- "1.13.x"
- "1.14.x"
before_script:
# make sure we keep path in tact when we sudo
- sudo sed -i -e 's/^Defaults\tsecure_path.*$//' /etc/sudoers
@@ -9,5 +13,8 @@ before_script:
- sudo modprobe nf_conntrack_netlink
- sudo modprobe nf_conntrack_ipv4
- sudo modprobe nf_conntrack_ipv6
- sudo modprobe sch_hfsc
- sudo modprobe sch_sfq
install:
- go get github.com/vishvananda/netns
- go get -v -t ./...
go_import_path: github.com/vishvananda/netlink

View File

@@ -17,6 +17,7 @@ type Addr struct {
Broadcast net.IP
PreferedLft int
ValidLft int
LinkIndex int
}
// String returns $ip/$netmask $label

View File

@@ -11,43 +11,63 @@ import (
"golang.org/x/sys/unix"
)
// IFA_FLAGS is a u32 attribute.
const IFA_FLAGS = 0x8
// 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)
@@ -65,7 +85,11 @@ func (h *Handle) addrHandle(link Link, addr *Addr, req *nl.NetlinkRequest) error
msg := nl.NewIfAddrmsg(family)
msg.Index = uint32(base.Index)
msg.Scope = uint8(addr.Scope)
prefixlen, masklen := addr.Mask.Size()
mask := addr.Mask
if addr.Peer != nil {
mask = addr.Peer.Mask
}
prefixlen, masklen := mask.Size()
msg.Prefixlen = uint8(prefixlen)
req.AddData(msg)
@@ -98,20 +122,26 @@ func (h *Handle) addrHandle(link Link, addr *Addr, req *nl.NetlinkRequest) error
} else {
b := make([]byte, 4)
native.PutUint32(b, uint32(addr.Flags))
flagsData := nl.NewRtAttr(IFA_FLAGS, b)
flagsData := nl.NewRtAttr(unix.IFA_FLAGS, b)
req.AddData(flagsData)
}
}
if family == FAMILY_V4 {
if addr.Broadcast == nil {
// 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] | ^addr.Mask[i]
calcBroadcast[i] = localAddrData[i] | ^mask[i]
}
addr.Broadcast = calcBroadcast
}
req.AddData(nl.NewRtAttr(unix.IFA_BROADCAST, addr.Broadcast))
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))
@@ -123,10 +153,10 @@ func (h *Handle) addrHandle(link Link, addr *Addr, req *nl.NetlinkRequest) error
// 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{
IfaValid: uint32(addr.ValidLft),
IfaPrefered: uint32(addr.PreferedLft),
}
cachedata := nl.IfaCacheInfo{unix.IfaCacheinfo{
Valid: uint32(addr.ValidLft),
Prefered: uint32(addr.PreferedLft),
}}
req.AddData(nl.NewRtAttr(unix.IFA_CACHEINFO, cachedata.Serialize()))
}
@@ -163,12 +193,12 @@ func (h *Handle) AddrList(link Link, family int) ([]Addr, error) {
var res []Addr
for _, m := range msgs {
addr, msgFamily, ifindex, err := parseAddr(m)
addr, msgFamily, err := parseAddr(m)
if err != nil {
return res, err
}
if link != nil && ifindex != indexFilter {
if link != nil && addr.LinkIndex != indexFilter {
// Ignore messages from other interfaces
continue
}
@@ -183,11 +213,11 @@ func (h *Handle) AddrList(link Link, family int) ([]Addr, error) {
return res, nil
}
func parseAddr(m []byte) (addr Addr, family, index int, err error) {
func parseAddr(m []byte) (addr Addr, family int, err error) {
msg := nl.DeserializeIfAddrmsg(m)
family = -1
index = -1
addr.LinkIndex = -1
attrs, err1 := nl.ParseRouteAttr(m[msg.Len():])
if err1 != nil {
@@ -196,7 +226,7 @@ func parseAddr(m []byte) (addr Addr, family, index int, err error) {
}
family = int(msg.Family)
index = int(msg.Index)
addr.LinkIndex = int(msg.Index)
var local, dst *net.IPNet
for _, attr := range attrs {
@@ -206,32 +236,48 @@ func parseAddr(m []byte) (addr Addr, family, index int, err error) {
IP: attr.Value,
Mask: net.CIDRMask(int(msg.Prefixlen), 8*len(attr.Value)),
}
addr.Peer = dst
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(int(msg.Prefixlen), 8*len(attr.Value)),
Mask: net.CIDRMask(n, n),
}
addr.IPNet = local
case unix.IFA_BROADCAST:
addr.Broadcast = attr.Value
case unix.IFA_LABEL:
addr.Label = string(attr.Value[:len(attr.Value)-1])
case IFA_FLAGS:
case unix.IFA_FLAGS:
addr.Flags = int(native.Uint32(attr.Value[0:4]))
case nl.IFA_CACHEINFO:
case unix.IFA_CACHEINFO:
ci := nl.DeserializeIfaCacheInfo(attr.Value)
addr.PreferedLft = int(ci.IfaPrefered)
addr.ValidLft = int(ci.IfaValid)
addr.PreferedLft = int(ci.Prefered)
addr.ValidLft = int(ci.Valid)
}
}
// IFA_LOCAL should be there but if not, fall back to IFA_ADDRESS
// 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 {
addr.IPNet = local
if family == FAMILY_V4 && 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
@@ -250,21 +296,22 @@ type AddrUpdate struct {
// 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)
return addrSubscribeAt(netns.None(), netns.None(), ch, done, nil, false, 0)
}
// 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)
return addrSubscribeAt(ns, netns.None(), ch, done, nil, false, 0)
}
// AddrSubscribeOptions contains a set of options to use with
// AddrSubscribeWithOptions.
type AddrSubscribeOptions struct {
Namespace *netns.NsHandle
ErrorCallback func(error)
ListExisting bool
Namespace *netns.NsHandle
ErrorCallback func(error)
ListExisting bool
ReceiveBufferSize int
}
// AddrSubscribeWithOptions work like AddrSubscribe but enable to
@@ -275,10 +322,10 @@ func AddrSubscribeWithOptions(ch chan<- AddrUpdate, done <-chan struct{}, option
none := netns.None()
options.Namespace = &none
}
return addrSubscribeAt(*options.Namespace, netns.None(), ch, done, options.ErrorCallback, options.ListExisting)
return addrSubscribeAt(*options.Namespace, netns.None(), ch, done, options.ErrorCallback, options.ListExisting, options.ReceiveBufferSize)
}
func addrSubscribeAt(newNs, curNs netns.NsHandle, ch chan<- AddrUpdate, done <-chan struct{}, cberr func(error), listExisting bool) error {
func addrSubscribeAt(newNs, curNs netns.NsHandle, ch chan<- AddrUpdate, done <-chan struct{}, cberr func(error), listExisting bool, rcvbuf int) error {
s, err := nl.SubscribeAt(newNs, curNs, unix.NETLINK_ROUTE, unix.RTNLGRP_IPV4_IFADDR, unix.RTNLGRP_IPV6_IFADDR)
if err != nil {
return err
@@ -289,6 +336,12 @@ func addrSubscribeAt(newNs, curNs netns.NsHandle, ch chan<- AddrUpdate, done <-c
s.Close()
}()
}
if rcvbuf != 0 {
err = pkgHandle.SetSocketReceiveBufferSize(rcvbuf, false)
if err != nil {
return err
}
}
if listExisting {
req := pkgHandle.newNetlinkRequest(unix.RTM_GETADDR,
unix.NLM_F_DUMP)
@@ -301,13 +354,19 @@ func addrSubscribeAt(newNs, curNs netns.NsHandle, ch chan<- AddrUpdate, done <-c
go func() {
defer close(ch)
for {
msgs, err := s.Receive()
msgs, from, err := s.Receive()
if err != nil {
if cberr != nil {
cberr(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
@@ -319,28 +378,29 @@ func addrSubscribeAt(newNs, curNs netns.NsHandle, ch chan<- AddrUpdate, done <-c
continue
}
if cberr != nil {
cberr(syscall.Errno(-error))
cberr(fmt.Errorf("error message: %v",
syscall.Errno(-error)))
}
return
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))
}
return
continue
}
addr, _, ifindex, err := parseAddr(m.Data)
addr, _, err := parseAddr(m.Data)
if err != nil {
if cberr != nil {
cberr(fmt.Errorf("could not parse address: %v", err))
}
return
continue
}
ch <- AddrUpdate{LinkAddress: *addr.IPNet,
LinkIndex: ifindex,
LinkIndex: addr.LinkIndex,
NewAddr: msgType == unix.RTM_NEWADDR,
Flags: addr.Flags,
Scope: addr.Scope,

View File

@@ -96,7 +96,7 @@ func (h *Handle) bridgeVlanModify(cmd int, link Link, vid uint16, pvid, untagged
flags |= nl.BRIDGE_FLAGS_MASTER
}
if flags > 0 {
nl.NewRtAttrChild(br, nl.IFLA_BRIDGE_FLAGS, nl.Uint16Attr(flags))
br.AddRtAttr(nl.IFLA_BRIDGE_FLAGS, nl.Uint16Attr(flags))
}
vlanInfo := &nl.BridgeVlanInfo{Vid: vid}
if pvid {
@@ -105,11 +105,8 @@ func (h *Handle) bridgeVlanModify(cmd int, link Link, vid uint16, pvid, untagged
if untagged {
vlanInfo.Flags |= nl.BRIDGE_VLAN_INFO_UNTAGGED
}
nl.NewRtAttrChild(br, nl.IFLA_BRIDGE_VLAN_INFO, vlanInfo.Serialize())
br.AddRtAttr(nl.IFLA_BRIDGE_VLAN_INFO, vlanInfo.Serialize())
req.AddData(br)
_, err := req.Execute(unix.NETLINK_ROUTE, 0)
if err != nil {
return err
}
return nil
return err
}

View File

@@ -4,25 +4,76 @@ import (
"fmt"
)
// Class interfaces for all classes
type Class interface {
Attrs() *ClassAttrs
Type() string
}
// Generic networking statistics for netlink users.
// This file contains "gnet_" prefixed structs and relevant functions.
// See Documentation/networking/getn_stats.txt in Linux source code for more details.
// GnetStatsBasic Ref: struct gnet_stats_basic { ... }
type GnetStatsBasic struct {
Bytes uint64 // number of seen bytes
Packets uint32 // number of seen packets
}
// GnetStatsRateEst Ref: struct gnet_stats_rate_est { ... }
type GnetStatsRateEst struct {
Bps uint32 // current byte rate
Pps uint32 // current packet rate
}
// GnetStatsRateEst64 Ref: struct gnet_stats_rate_est64 { ... }
type GnetStatsRateEst64 struct {
Bps uint64 // current byte rate
Pps uint64 // current packet rate
}
// GnetStatsQueue Ref: struct gnet_stats_queue { ... }
type GnetStatsQueue struct {
Qlen uint32 // queue length
Backlog uint32 // backlog size of queue
Drops uint32 // number of dropped packets
Requeues uint32 // number of requues
Overlimits uint32 // number of enqueues over the limit
}
// ClassStatistics representation based on generic networking statistics for netlink.
// See Documentation/networking/gen_stats.txt in Linux source code for more details.
type ClassStatistics struct {
Basic *GnetStatsBasic
Queue *GnetStatsQueue
RateEst *GnetStatsRateEst
}
// NewClassStatistics Construct a ClassStatistics struct which fields are all initialized by 0.
func NewClassStatistics() *ClassStatistics {
return &ClassStatistics{
Basic: &GnetStatsBasic{},
Queue: &GnetStatsQueue{},
RateEst: &GnetStatsRateEst{},
}
}
// ClassAttrs represents a netlink class. A filter is associated with a link,
// has a handle and a parent. The root filter of a device should have a
// parent == HANDLE_ROOT.
type ClassAttrs struct {
LinkIndex int
Handle uint32
Parent uint32
Leaf uint32
LinkIndex int
Handle uint32
Parent uint32
Leaf uint32
Statistics *ClassStatistics
}
func (q ClassAttrs) String() string {
return fmt.Sprintf("{LinkIndex: %d, Handle: %s, Parent: %s, Leaf: %d}", q.LinkIndex, HandleStr(q.Handle), HandleStr(q.Parent), q.Leaf)
}
// HtbClassAttrs stores the attributes of HTB class
type HtbClassAttrs struct {
// TODO handle all attributes
Rate uint64
@@ -54,10 +105,12 @@ func (q HtbClass) String() string {
return fmt.Sprintf("{Rate: %d, Ceil: %d, Buffer: %d, Cbuffer: %d}", q.Rate, q.Ceil, q.Buffer, q.Cbuffer)
}
// Attrs returns the class attributes
func (q *HtbClass) Attrs() *ClassAttrs {
return &q.ClassAttrs
}
// Type return the class type
func (q *HtbClass) Type() string {
return "htb"
}
@@ -69,10 +122,118 @@ type GenericClass struct {
ClassType string
}
// Attrs return the class attributes
func (class *GenericClass) Attrs() *ClassAttrs {
return &class.ClassAttrs
}
// Type return the class type
func (class *GenericClass) Type() string {
return class.ClassType
}
// ServiceCurve is a nondecreasing function of some time unit, returning the amount of service
// (an allowed or allocated amount of bandwidth) at some specific point in time. The purpose of it
// should be subconsciously obvious: if a class was allowed to transfer not less than the amount
// specified by its service curve, then the service curve is not violated.
type ServiceCurve struct {
m1 uint32
d uint32
m2 uint32
}
// Attrs return the parameters of the service curve
func (c *ServiceCurve) Attrs() (uint32, uint32, uint32) {
return c.m1, c.d, c.m2
}
// Burst returns the burst rate (m1) of the curve
func (c *ServiceCurve) Burst() uint32 {
return c.m1
}
// Delay return the delay (d) of the curve
func (c *ServiceCurve) Delay() uint32 {
return c.d
}
// Rate returns the rate (m2) of the curve
func (c *ServiceCurve) Rate() uint32 {
return c.m2
}
// HfscClass is a representation of the HFSC class
type HfscClass struct {
ClassAttrs
Rsc ServiceCurve
Fsc ServiceCurve
Usc ServiceCurve
}
// SetUsc sets the USC curve. The bandwidth (m1 and m2) is specified in bits and the delay in
// seconds.
func (hfsc *HfscClass) SetUsc(m1 uint32, d uint32, m2 uint32) {
hfsc.Usc = ServiceCurve{m1: m1, d: d, m2: m2}
}
// SetFsc sets the Fsc curve. The bandwidth (m1 and m2) is specified in bits and the delay in
// seconds.
func (hfsc *HfscClass) SetFsc(m1 uint32, d uint32, m2 uint32) {
hfsc.Fsc = ServiceCurve{m1: m1, d: d, m2: m2}
}
// SetRsc sets the Rsc curve. The bandwidth (m1 and m2) is specified in bits and the delay in
// seconds.
func (hfsc *HfscClass) SetRsc(m1 uint32, d uint32, m2 uint32) {
hfsc.Rsc = ServiceCurve{m1: m1, d: d, m2: m2}
}
// SetSC implements the SC from the `tc` CLI. This function behaves the same as if one would set the
// USC through the `tc` command-line tool. This means bandwidth (m1 and m2) is specified in bits and
// the delay in ms.
func (hfsc *HfscClass) SetSC(m1 uint32, d uint32, m2 uint32) {
hfsc.SetRsc(m1, d, m2)
hfsc.SetFsc(m1, d, m2)
}
// SetUL implements the UL from the `tc` CLI. This function behaves the same as if one would set the
// USC through the `tc` command-line tool. This means bandwidth (m1 and m2) is specified in bits and
// the delay in ms.
func (hfsc *HfscClass) SetUL(m1 uint32, d uint32, m2 uint32) {
hfsc.SetUsc(m1, d, m2)
}
// SetLS implements the LS from the `tc` CLI. This function behaves the same as if one would set the
// USC through the `tc` command-line tool. This means bandwidth (m1 and m2) is specified in bits and
// the delay in ms.
func (hfsc *HfscClass) SetLS(m1 uint32, d uint32, m2 uint32) {
hfsc.SetFsc(m1, d, m2)
}
// NewHfscClass returns a new HFSC struct with the set parameters
func NewHfscClass(attrs ClassAttrs) *HfscClass {
return &HfscClass{
ClassAttrs: attrs,
Rsc: ServiceCurve{},
Fsc: ServiceCurve{},
Usc: ServiceCurve{},
}
}
// String() returns a string that contains the information and attributes of the HFSC class
func (hfsc *HfscClass) String() string {
return fmt.Sprintf(
"{%s -- {RSC: {m1=%d d=%d m2=%d}} {FSC: {m1=%d d=%d m2=%d}} {USC: {m1=%d d=%d m2=%d}}}",
hfsc.Attrs(), hfsc.Rsc.m1*8, hfsc.Rsc.d, hfsc.Rsc.m2*8, hfsc.Fsc.m1*8, hfsc.Fsc.d, hfsc.Fsc.m2*8, hfsc.Usc.m1*8, hfsc.Usc.d, hfsc.Usc.m2*8,
)
}
// Attrs return the Hfsc parameters
func (hfsc *HfscClass) Attrs() *ClassAttrs {
return &hfsc.ClassAttrs
}
// Type return the type of the class
func (hfsc *HfscClass) Type() string {
return "hfsc"
}

View File

@@ -1,14 +1,34 @@
package netlink
import (
"bytes"
"encoding/binary"
"encoding/hex"
"errors"
"fmt"
"syscall"
"github.com/vishvananda/netlink/nl"
"golang.org/x/sys/unix"
)
// NOTE: function is in here because it uses other linux functions
// Internal tc_stats representation in Go struct.
// This is for internal uses only to deserialize the payload of rtattr.
// After the deserialization, this should be converted into the canonical stats
// struct, ClassStatistics, in case of statistics of a class.
// Ref: struct tc_stats { ... }
type tcStats struct {
Bytes uint64 // Number of enqueued bytes
Packets uint32 // Number of enqueued packets
Drops uint32 // Packets dropped because of lack of resources
Overlimits uint32 // Number of throttle events when this flow goes out of allocated bandwidth
Bps uint32 // Current flow byte rate
Pps uint32 // Current flow packet rate
Qlen uint32
Backlog uint32
}
// NewHtbClass NOTE: function is in here because it uses other linux functions
func NewHtbClass(attrs ClassAttrs, cattrs HtbClassAttrs) *HtbClass {
mtu := 1600
rate := cattrs.Rate / 8
@@ -23,12 +43,12 @@ func NewHtbClass(attrs ClassAttrs, cattrs HtbClassAttrs) *HtbClass {
if buffer == 0 {
buffer = uint32(float64(rate)/Hz() + float64(mtu))
}
buffer = uint32(Xmittime(rate, buffer))
buffer = Xmittime(rate, buffer)
if cbuffer == 0 {
cbuffer = uint32(float64(ceil)/Hz() + float64(mtu))
}
cbuffer = uint32(Xmittime(ceil, cbuffer))
cbuffer = Xmittime(ceil, cbuffer)
return &HtbClass{
ClassAttrs: attrs,
@@ -36,9 +56,9 @@ func NewHtbClass(attrs ClassAttrs, cattrs HtbClassAttrs) *HtbClass {
Ceil: ceil,
Buffer: buffer,
Cbuffer: cbuffer,
Quantum: 10,
Level: 0,
Prio: 0,
Prio: cattrs.Prio,
Quantum: cattrs.Quantum,
}
}
@@ -126,7 +146,9 @@ func classPayload(req *nl.NetlinkRequest, class Class) error {
req.AddData(nl.NewRtAttr(nl.TCA_KIND, nl.ZeroTerminated(class.Type())))
options := nl.NewRtAttr(nl.TCA_OPTIONS, nil)
if htb, ok := class.(*HtbClass); ok {
switch class.Type() {
case "htb":
htb := class.(*HtbClass)
opt := nl.TcHtbCopt{}
opt.Buffer = htb.Buffer
opt.Cbuffer = htb.Cbuffer
@@ -151,9 +173,27 @@ func classPayload(req *nl.NetlinkRequest, class Class) error {
return errors.New("HTB: failed to calculate ceil rate table")
}
opt.Ceil = tcceil
nl.NewRtAttrChild(options, nl.TCA_HTB_PARMS, opt.Serialize())
nl.NewRtAttrChild(options, nl.TCA_HTB_RTAB, SerializeRtab(rtab))
nl.NewRtAttrChild(options, nl.TCA_HTB_CTAB, SerializeRtab(ctab))
options.AddRtAttr(nl.TCA_HTB_PARMS, opt.Serialize())
options.AddRtAttr(nl.TCA_HTB_RTAB, SerializeRtab(rtab))
options.AddRtAttr(nl.TCA_HTB_CTAB, SerializeRtab(ctab))
if htb.Rate >= uint64(1<<32) {
options.AddRtAttr(nl.TCA_HTB_RATE64, nl.Uint64Attr(htb.Rate))
}
if htb.Ceil >= uint64(1<<32) {
options.AddRtAttr(nl.TCA_HTB_CEIL64, nl.Uint64Attr(htb.Ceil))
}
case "hfsc":
hfsc := class.(*HfscClass)
opt := nl.HfscCopt{}
rm1, rd, rm2 := hfsc.Rsc.Attrs()
opt.Rsc.Set(rm1/8, rd, rm2/8)
fm1, fd, fm2 := hfsc.Fsc.Attrs()
opt.Fsc.Set(fm1/8, fd, fm2/8)
um1, ud, um2 := hfsc.Usc.Attrs()
opt.Usc.Set(um1/8, ud, um2/8)
nl.NewRtAttrChild(options, nl.TCA_HFSC_RSC, nl.SerializeHfscCurve(&opt.Rsc))
nl.NewRtAttrChild(options, nl.TCA_HFSC_FSC, nl.SerializeHfscCurve(&opt.Fsc))
nl.NewRtAttrChild(options, nl.TCA_HFSC_USC, nl.SerializeHfscCurve(&opt.Usc))
}
req.AddData(options)
return nil
@@ -197,9 +237,10 @@ func (h *Handle) ClassList(link Link, parent uint32) ([]Class, error) {
}
base := ClassAttrs{
LinkIndex: int(msg.Ifindex),
Handle: msg.Handle,
Parent: msg.Parent,
LinkIndex: int(msg.Ifindex),
Handle: msg.Handle,
Parent: msg.Parent,
Statistics: nil,
}
var class Class
@@ -211,6 +252,8 @@ func (h *Handle) ClassList(link Link, parent uint32) ([]Class, error) {
switch classType {
case "htb":
class = &HtbClass{}
case "hfsc":
class = &HfscClass{}
default:
class = &GenericClass{ClassType: classType}
}
@@ -225,6 +268,26 @@ func (h *Handle) ClassList(link Link, parent uint32) ([]Class, error) {
if err != nil {
return nil, err
}
case "hfsc":
data, err := nl.ParseRouteAttr(attr.Value)
if err != nil {
return nil, err
}
_, err = parseHfscClassData(class, data)
if err != nil {
return nil, err
}
}
// For backward compatibility.
case nl.TCA_STATS:
base.Statistics, err = parseTcStats(attr.Value)
if err != nil {
return nil, err
}
case nl.TCA_STATS2:
base.Statistics, err = parseTcStats2(attr.Value)
if err != nil {
return nil, err
}
}
}
@@ -249,7 +312,86 @@ func parseHtbClassData(class Class, data []syscall.NetlinkRouteAttr) (bool, erro
htb.Quantum = opt.Quantum
htb.Level = opt.Level
htb.Prio = opt.Prio
case nl.TCA_HTB_RATE64:
htb.Rate = native.Uint64(datum.Value[0:8])
case nl.TCA_HTB_CEIL64:
htb.Ceil = native.Uint64(datum.Value[0:8])
}
}
return detailed, nil
}
func parseHfscClassData(class Class, data []syscall.NetlinkRouteAttr) (bool, error) {
hfsc := class.(*HfscClass)
detailed := false
for _, datum := range data {
m1, d, m2 := nl.DeserializeHfscCurve(datum.Value).Attrs()
switch datum.Attr.Type {
case nl.TCA_HFSC_RSC:
hfsc.Rsc = ServiceCurve{m1: m1 * 8, d: d, m2: m2 * 8}
case nl.TCA_HFSC_FSC:
hfsc.Fsc = ServiceCurve{m1: m1 * 8, d: d, m2: m2 * 8}
case nl.TCA_HFSC_USC:
hfsc.Usc = ServiceCurve{m1: m1 * 8, d: d, m2: m2 * 8}
}
}
return detailed, nil
}
func parseTcStats(data []byte) (*ClassStatistics, error) {
buf := &bytes.Buffer{}
buf.Write(data)
native := nl.NativeEndian()
tcStats := &tcStats{}
if err := binary.Read(buf, native, tcStats); err != nil {
return nil, err
}
stats := NewClassStatistics()
stats.Basic.Bytes = tcStats.Bytes
stats.Basic.Packets = tcStats.Packets
stats.Queue.Qlen = tcStats.Qlen
stats.Queue.Backlog = tcStats.Backlog
stats.Queue.Drops = tcStats.Drops
stats.Queue.Overlimits = tcStats.Overlimits
stats.RateEst.Bps = tcStats.Bps
stats.RateEst.Pps = tcStats.Pps
return stats, nil
}
func parseGnetStats(data []byte, gnetStats interface{}) error {
buf := &bytes.Buffer{}
buf.Write(data)
native := nl.NativeEndian()
return binary.Read(buf, native, gnetStats)
}
func parseTcStats2(data []byte) (*ClassStatistics, error) {
rtAttrs, err := nl.ParseRouteAttr(data)
if err != nil {
return nil, err
}
stats := NewClassStatistics()
for _, datum := range rtAttrs {
switch datum.Attr.Type {
case nl.TCA_STATS_BASIC:
if err := parseGnetStats(datum.Value, stats.Basic); err != nil {
return nil, fmt.Errorf("Failed to parse ClassStatistics.Basic with: %v\n%s",
err, hex.Dump(datum.Value))
}
case nl.TCA_STATS_QUEUE:
if err := parseGnetStats(datum.Value, stats.Queue); err != nil {
return nil, fmt.Errorf("Failed to parse ClassStatistics.Queue with: %v\n%s",
err, hex.Dump(datum.Value))
}
case nl.TCA_STATS_RATE_EST:
if err := parseGnetStats(datum.Value, stats.RateEst); err != nil {
return nil, fmt.Errorf("Failed to parse ClassStatistics.RateEst with: %v\n%s",
err, hex.Dump(datum.Value))
}
}
}
return stats, nil
}

View File

@@ -22,11 +22,7 @@ const (
// https://github.com/torvalds/linux/blob/master/include/uapi/linux/netfilter/nfnetlink.h -> #define NFNL_SUBSYS_CTNETLINK_EXP 2
ConntrackExpectTable = 2
)
const (
// For Parsing Mark
TCP_PROTO = 6
UDP_PROTO = 17
)
const (
// backward compatibility with golang 1.6 which does not have io.SeekCurrent
seekCurrent = 1
@@ -135,11 +131,13 @@ func (h *Handle) dumpConntrackTable(table ConntrackTableType, family InetFamily)
// http://git.netfilter.org/libnetfilter_conntrack/tree/include/internal/object.h
// For the time being, the structure below allows to parse and extract the base information of a flow
type ipTuple struct {
SrcIP net.IP
Bytes uint64
DstIP net.IP
Protocol uint8
SrcPort uint16
DstPort uint16
Packets uint64
Protocol uint8
SrcIP net.IP
SrcPort uint16
}
type ConntrackFlow struct {
@@ -151,11 +149,12 @@ type ConntrackFlow struct {
func (s *ConntrackFlow) String() string {
// conntrack cmd output:
// udp 17 src=127.0.0.1 dst=127.0.0.1 sport=4001 dport=1234 [UNREPLIED] src=127.0.0.1 dst=127.0.0.1 sport=1234 dport=4001 mark=0
return fmt.Sprintf("%s\t%d src=%s dst=%s sport=%d dport=%d\tsrc=%s dst=%s sport=%d dport=%d mark=%d",
// udp 17 src=127.0.0.1 dst=127.0.0.1 sport=4001 dport=1234 packets=5 bytes=532 [UNREPLIED] src=127.0.0.1 dst=127.0.0.1 sport=1234 dport=4001 packets=10 bytes=1078 mark=0
return fmt.Sprintf("%s\t%d src=%s dst=%s sport=%d dport=%d packets=%d bytes=%d\tsrc=%s dst=%s sport=%d dport=%d packets=%d bytes=%d mark=%d",
nl.L4ProtoMap[s.Forward.Protocol], s.Forward.Protocol,
s.Forward.SrcIP.String(), s.Forward.DstIP.String(), s.Forward.SrcPort, s.Forward.DstPort,
s.Reverse.SrcIP.String(), s.Reverse.DstIP.String(), s.Reverse.SrcPort, s.Reverse.DstPort, s.Mark)
s.Forward.SrcIP.String(), s.Forward.DstIP.String(), s.Forward.SrcPort, s.Forward.DstPort, s.Forward.Packets, s.Forward.Bytes,
s.Reverse.SrcIP.String(), s.Reverse.DstIP.String(), s.Reverse.SrcPort, s.Reverse.DstPort, s.Reverse.Packets, s.Reverse.Bytes,
s.Mark)
}
// This method parse the ip tuple structure
@@ -220,9 +219,35 @@ func parseBERaw16(r *bytes.Reader, v *uint16) {
binary.Read(r, binary.BigEndian, v)
}
func parseBERaw32(r *bytes.Reader, v *uint32) {
binary.Read(r, binary.BigEndian, v)
}
func parseBERaw64(r *bytes.Reader, v *uint64) {
binary.Read(r, binary.BigEndian, v)
}
func parseByteAndPacketCounters(r *bytes.Reader) (bytes, packets uint64) {
for i := 0; i < 2; i++ {
switch _, t, _ := parseNfAttrTL(r); t {
case nl.CTA_COUNTERS_BYTES:
parseBERaw64(r, &bytes)
case nl.CTA_COUNTERS_PACKETS:
parseBERaw64(r, &packets)
default:
return
}
}
return
}
func parseConnectionMark(r *bytes.Reader) (mark uint32) {
parseBERaw32(r, &mark)
return
}
func parseRawData(data []byte) *ConntrackFlow {
s := &ConntrackFlow{}
var proto uint8
// First there is the Nfgenmsg header
// consume only the family field
reader := bytes.NewReader(data)
@@ -238,36 +263,31 @@ func parseRawData(data []byte) *ConntrackFlow {
// <len, NLA_F_NESTED|CTA_TUPLE_IP> 4 bytes
// flow information of the reverse flow
for reader.Len() > 0 {
nested, t, l := parseNfAttrTL(reader)
if nested && t == nl.CTA_TUPLE_ORIG {
if nested, t, _ = parseNfAttrTL(reader); nested && t == nl.CTA_TUPLE_IP {
proto = parseIpTuple(reader, &s.Forward)
if nested, t, l := parseNfAttrTL(reader); nested {
switch t {
case nl.CTA_TUPLE_ORIG:
if nested, t, _ = parseNfAttrTL(reader); nested && t == nl.CTA_TUPLE_IP {
parseIpTuple(reader, &s.Forward)
}
case nl.CTA_TUPLE_REPLY:
if nested, t, _ = parseNfAttrTL(reader); nested && t == nl.CTA_TUPLE_IP {
parseIpTuple(reader, &s.Reverse)
} else {
// Header not recognized skip it
reader.Seek(int64(l), seekCurrent)
}
case nl.CTA_COUNTERS_ORIG:
s.Forward.Bytes, s.Forward.Packets = parseByteAndPacketCounters(reader)
case nl.CTA_COUNTERS_REPLY:
s.Reverse.Bytes, s.Reverse.Packets = parseByteAndPacketCounters(reader)
}
} else if nested && t == nl.CTA_TUPLE_REPLY {
if nested, t, _ = parseNfAttrTL(reader); nested && t == nl.CTA_TUPLE_IP {
parseIpTuple(reader, &s.Reverse)
// Got all the useful information stop parsing
break
} else {
// Header not recognized skip it
reader.Seek(int64(l), seekCurrent)
} else {
switch t {
case nl.CTA_MARK:
s.Mark = parseConnectionMark(reader)
}
}
}
if proto == TCP_PROTO {
reader.Seek(64, seekCurrent)
_, t, _, v := parseNfAttrTLV(reader)
if t == nl.CTA_MARK {
s.Mark = uint32(v[3])
}
} else if proto == UDP_PROTO {
reader.Seek(16, seekCurrent)
_, t, _, v := parseNfAttrTLV(reader)
if t == nl.CTA_MARK {
s.Mark = uint32(v[3])
}
}
return s
}
@@ -285,7 +305,7 @@ func parseRawData(data []byte) *ConntrackFlow {
// Common parameters and options:
// -s, --src, --orig-src ip Source address from original direction
// -d, --dst, --orig-dst ip Destination address from original direction
// -r, --reply-src ip Source addres from reply direction
// -r, --reply-src ip Source address from reply direction
// -q, --reply-dst ip Destination address from reply direction
// -p, --protonum proto Layer 4 Protocol, eg. 'tcp'
// -f, --family proto Layer 3 Protocol, eg. 'ipv6'
@@ -298,15 +318,25 @@ func parseRawData(data []byte) *ConntrackFlow {
// --mask-src ip Source mask address
// --mask-dst ip Destination mask address
// Layer 4 Protocol common parameters and options:
// TCP, UDP, SCTP, UDPLite and DCCP
// --sport, --orig-port-src port Source port in original direction
// --dport, --orig-port-dst port Destination port in original direction
// Filter types
type ConntrackFilterType uint8
const (
ConntrackOrigSrcIP = iota // -orig-src ip Source address from original direction
ConntrackOrigDstIP // -orig-dst ip Destination address from original direction
ConntrackNatSrcIP // -src-nat ip Source NAT ip
ConntrackNatDstIP // -dst-nat ip Destination NAT ip
ConntrackNatAnyIP // -any-nat ip Source or destination NAT ip
ConntrackOrigSrcIP = iota // -orig-src ip Source address from original direction
ConntrackOrigDstIP // -orig-dst ip Destination address from original direction
ConntrackReplySrcIP // --reply-src ip Reply Source IP
ConntrackReplyDstIP // --reply-dst ip Reply Destination IP
ConntrackReplyAnyIP // Match source or destination reply IP
ConntrackOrigSrcPort // --orig-port-src port Source port in original direction
ConntrackOrigDstPort // --orig-port-dst port Destination port in original direction
ConntrackNatSrcIP = ConntrackReplySrcIP // deprecated use instead ConntrackReplySrcIP
ConntrackNatDstIP = ConntrackReplyDstIP // deprecated use instead ConntrackReplyDstIP
ConntrackNatAnyIP = ConntrackReplyAnyIP // deprecated use instead ConntrackReplyAnyIP
)
type CustomConntrackFilter interface {
@@ -316,7 +346,9 @@ type CustomConntrackFilter interface {
}
type ConntrackFilter struct {
ipFilter map[ConntrackFilterType]net.IP
ipFilter map[ConntrackFilterType]net.IP
portFilter map[ConntrackFilterType]uint16
protoFilter uint8
}
// AddIP adds an IP to the conntrack filter
@@ -331,38 +363,89 @@ func (f *ConntrackFilter) AddIP(tp ConntrackFilterType, ip net.IP) error {
return nil
}
// AddPort adds a Port to the conntrack filter if the Layer 4 protocol allows it
func (f *ConntrackFilter) AddPort(tp ConntrackFilterType, port uint16) error {
switch f.protoFilter {
// TCP, UDP, DCCP, SCTP, UDPLite
case 6, 17, 33, 132, 136:
default:
return fmt.Errorf("Filter attribute not available without a valid Layer 4 protocol: %d", f.protoFilter)
}
if f.portFilter == nil {
f.portFilter = make(map[ConntrackFilterType]uint16)
}
if _, ok := f.portFilter[tp]; ok {
return errors.New("Filter attribute already present")
}
f.portFilter[tp] = port
return nil
}
// AddProtocol adds the Layer 4 protocol to the conntrack filter
func (f *ConntrackFilter) AddProtocol(proto uint8) error {
if f.protoFilter != 0 {
return errors.New("Filter attribute already present")
}
f.protoFilter = proto
return nil
}
// MatchConntrackFlow applies the filter to the flow and returns true if the flow matches the filter
// false otherwise
func (f *ConntrackFilter) MatchConntrackFlow(flow *ConntrackFlow) bool {
if len(f.ipFilter) == 0 {
if len(f.ipFilter) == 0 && len(f.portFilter) == 0 && f.protoFilter == 0 {
// empty filter always not match
return false
}
// -p, --protonum proto Layer 4 Protocol, eg. 'tcp'
if f.protoFilter != 0 && flow.Forward.Protocol != f.protoFilter {
// different Layer 4 protocol always not match
return false
}
match := true
// -orig-src ip Source address from original direction
if elem, found := f.ipFilter[ConntrackOrigSrcIP]; found {
match = match && elem.Equal(flow.Forward.SrcIP)
// IP conntrack filter
if len(f.ipFilter) > 0 {
// -orig-src ip Source address from original direction
if elem, found := f.ipFilter[ConntrackOrigSrcIP]; found {
match = match && elem.Equal(flow.Forward.SrcIP)
}
// -orig-dst ip Destination address from original direction
if elem, found := f.ipFilter[ConntrackOrigDstIP]; match && found {
match = match && elem.Equal(flow.Forward.DstIP)
}
// -src-nat ip Source NAT ip
if elem, found := f.ipFilter[ConntrackReplySrcIP]; match && found {
match = match && elem.Equal(flow.Reverse.SrcIP)
}
// -dst-nat ip Destination NAT ip
if elem, found := f.ipFilter[ConntrackReplyDstIP]; match && found {
match = match && elem.Equal(flow.Reverse.DstIP)
}
// Match source or destination reply IP
if elem, found := f.ipFilter[ConntrackReplyAnyIP]; match && found {
match = match && (elem.Equal(flow.Reverse.SrcIP) || elem.Equal(flow.Reverse.DstIP))
}
}
// -orig-dst ip Destination address from original direction
if elem, found := f.ipFilter[ConntrackOrigDstIP]; match && found {
match = match && elem.Equal(flow.Forward.DstIP)
}
// Layer 4 Port filter
if len(f.portFilter) > 0 {
// -orig-port-src port Source port from original direction
if elem, found := f.portFilter[ConntrackOrigSrcPort]; match && found {
match = match && elem == flow.Forward.SrcPort
}
// -src-nat ip Source NAT ip
if elem, found := f.ipFilter[ConntrackNatSrcIP]; match && found {
match = match && elem.Equal(flow.Reverse.SrcIP)
}
// -dst-nat ip Destination NAT ip
if elem, found := f.ipFilter[ConntrackNatDstIP]; match && found {
match = match && elem.Equal(flow.Reverse.DstIP)
}
// -any-nat ip Source or destination NAT ip
if elem, found := f.ipFilter[ConntrackNatAnyIP]; match && found {
match = match && (elem.Equal(flow.Reverse.SrcIP) || elem.Equal(flow.Reverse.DstIP))
// -orig-port-dst port Destination port from original direction
if elem, found := f.portFilter[ConntrackOrigDstPort]; match && found {
match = match && elem == flow.Forward.DstPort
}
}
return match

393
vendor/github.com/vishvananda/netlink/devlink_linux.go generated vendored Normal file
View File

@@ -0,0 +1,393 @@
package netlink
import (
"syscall"
"fmt"
"github.com/vishvananda/netlink/nl"
"golang.org/x/sys/unix"
)
// DevlinkDevEswitchAttr represents device's eswitch attributes
type DevlinkDevEswitchAttr struct {
Mode string
InlineMode string
EncapMode string
}
// DevlinkDevAttrs represents device attributes
type DevlinkDevAttrs struct {
Eswitch DevlinkDevEswitchAttr
}
// DevlinkDevice represents device and its attributes
type DevlinkDevice struct {
BusName string
DeviceName string
Attrs DevlinkDevAttrs
}
// DevlinkPort represents port and its attributes
type DevlinkPort struct {
BusName string
DeviceName string
PortIndex uint32
PortType uint16
NetdeviceName string
NetdevIfIndex uint32
RdmaDeviceName string
PortFlavour uint16
}
func parseDevLinkDeviceList(msgs [][]byte) ([]*DevlinkDevice, error) {
devices := make([]*DevlinkDevice, 0, len(msgs))
for _, m := range msgs {
attrs, err := nl.ParseRouteAttr(m[nl.SizeofGenlmsg:])
if err != nil {
return nil, err
}
dev := &DevlinkDevice{}
if err = dev.parseAttributes(attrs); err != nil {
return nil, err
}
devices = append(devices, dev)
}
return devices, nil
}
func eswitchStringToMode(modeName string) (uint16, error) {
if modeName == "legacy" {
return nl.DEVLINK_ESWITCH_MODE_LEGACY, nil
} else if modeName == "switchdev" {
return nl.DEVLINK_ESWITCH_MODE_SWITCHDEV, nil
} else {
return 0xffff, fmt.Errorf("invalid switchdev mode")
}
}
func parseEswitchMode(mode uint16) string {
var eswitchMode = map[uint16]string{
nl.DEVLINK_ESWITCH_MODE_LEGACY: "legacy",
nl.DEVLINK_ESWITCH_MODE_SWITCHDEV: "switchdev",
}
if eswitchMode[mode] == "" {
return "unknown"
} else {
return eswitchMode[mode]
}
}
func parseEswitchInlineMode(inlinemode uint8) string {
var eswitchInlineMode = map[uint8]string{
nl.DEVLINK_ESWITCH_INLINE_MODE_NONE: "none",
nl.DEVLINK_ESWITCH_INLINE_MODE_LINK: "link",
nl.DEVLINK_ESWITCH_INLINE_MODE_NETWORK: "network",
nl.DEVLINK_ESWITCH_INLINE_MODE_TRANSPORT: "transport",
}
if eswitchInlineMode[inlinemode] == "" {
return "unknown"
} else {
return eswitchInlineMode[inlinemode]
}
}
func parseEswitchEncapMode(encapmode uint8) string {
var eswitchEncapMode = map[uint8]string{
nl.DEVLINK_ESWITCH_ENCAP_MODE_NONE: "disable",
nl.DEVLINK_ESWITCH_ENCAP_MODE_BASIC: "enable",
}
if eswitchEncapMode[encapmode] == "" {
return "unknown"
} else {
return eswitchEncapMode[encapmode]
}
}
func (d *DevlinkDevice) parseAttributes(attrs []syscall.NetlinkRouteAttr) error {
for _, a := range attrs {
switch a.Attr.Type {
case nl.DEVLINK_ATTR_BUS_NAME:
d.BusName = string(a.Value)
case nl.DEVLINK_ATTR_DEV_NAME:
d.DeviceName = string(a.Value)
case nl.DEVLINK_ATTR_ESWITCH_MODE:
d.Attrs.Eswitch.Mode = parseEswitchMode(native.Uint16(a.Value))
case nl.DEVLINK_ATTR_ESWITCH_INLINE_MODE:
d.Attrs.Eswitch.InlineMode = parseEswitchInlineMode(uint8(a.Value[0]))
case nl.DEVLINK_ATTR_ESWITCH_ENCAP_MODE:
d.Attrs.Eswitch.EncapMode = parseEswitchEncapMode(uint8(a.Value[0]))
}
}
return nil
}
func (dev *DevlinkDevice) parseEswitchAttrs(msgs [][]byte) {
m := msgs[0]
attrs, err := nl.ParseRouteAttr(m[nl.SizeofGenlmsg:])
if err != nil {
return
}
dev.parseAttributes(attrs)
}
func (h *Handle) getEswitchAttrs(family *GenlFamily, dev *DevlinkDevice) {
msg := &nl.Genlmsg{
Command: nl.DEVLINK_CMD_ESWITCH_GET,
Version: nl.GENL_DEVLINK_VERSION,
}
req := h.newNetlinkRequest(int(family.ID), unix.NLM_F_REQUEST|unix.NLM_F_ACK)
req.AddData(msg)
b := make([]byte, len(dev.BusName))
copy(b, dev.BusName)
data := nl.NewRtAttr(nl.DEVLINK_ATTR_BUS_NAME, b)
req.AddData(data)
b = make([]byte, len(dev.DeviceName))
copy(b, dev.DeviceName)
data = nl.NewRtAttr(nl.DEVLINK_ATTR_DEV_NAME, b)
req.AddData(data)
msgs, err := req.Execute(unix.NETLINK_GENERIC, 0)
if err != nil {
return
}
dev.parseEswitchAttrs(msgs)
}
// DevLinkGetDeviceList provides a pointer to devlink devices and nil error,
// otherwise returns an error code.
func (h *Handle) DevLinkGetDeviceList() ([]*DevlinkDevice, error) {
f, err := h.GenlFamilyGet(nl.GENL_DEVLINK_NAME)
if err != nil {
return nil, err
}
msg := &nl.Genlmsg{
Command: nl.DEVLINK_CMD_GET,
Version: nl.GENL_DEVLINK_VERSION,
}
req := h.newNetlinkRequest(int(f.ID),
unix.NLM_F_REQUEST|unix.NLM_F_ACK|unix.NLM_F_DUMP)
req.AddData(msg)
msgs, err := req.Execute(unix.NETLINK_GENERIC, 0)
if err != nil {
return nil, err
}
devices, err := parseDevLinkDeviceList(msgs)
if err != nil {
return nil, err
}
for _, d := range devices {
h.getEswitchAttrs(f, d)
}
return devices, nil
}
// DevLinkGetDeviceList provides a pointer to devlink devices and nil error,
// otherwise returns an error code.
func DevLinkGetDeviceList() ([]*DevlinkDevice, error) {
return pkgHandle.DevLinkGetDeviceList()
}
func parseDevlinkDevice(msgs [][]byte) (*DevlinkDevice, error) {
m := msgs[0]
attrs, err := nl.ParseRouteAttr(m[nl.SizeofGenlmsg:])
if err != nil {
return nil, err
}
dev := &DevlinkDevice{}
if err = dev.parseAttributes(attrs); err != nil {
return nil, err
}
return dev, nil
}
func (h *Handle) createCmdReq(cmd uint8, bus string, device string) (*GenlFamily, *nl.NetlinkRequest, error) {
f, err := h.GenlFamilyGet(nl.GENL_DEVLINK_NAME)
if err != nil {
return nil, nil, err
}
msg := &nl.Genlmsg{
Command: cmd,
Version: nl.GENL_DEVLINK_VERSION,
}
req := h.newNetlinkRequest(int(f.ID),
unix.NLM_F_REQUEST|unix.NLM_F_ACK)
req.AddData(msg)
b := make([]byte, len(bus)+1)
copy(b, bus)
data := nl.NewRtAttr(nl.DEVLINK_ATTR_BUS_NAME, b)
req.AddData(data)
b = make([]byte, len(device)+1)
copy(b, device)
data = nl.NewRtAttr(nl.DEVLINK_ATTR_DEV_NAME, b)
req.AddData(data)
return f, req, nil
}
// DevlinkGetDeviceByName provides a pointer to devlink device and nil error,
// otherwise returns an error code.
func (h *Handle) DevLinkGetDeviceByName(Bus string, Device string) (*DevlinkDevice, error) {
f, req, err := h.createCmdReq(nl.DEVLINK_CMD_GET, Bus, Device)
if err != nil {
return nil, err
}
respmsg, err := req.Execute(unix.NETLINK_GENERIC, 0)
if err != nil {
return nil, err
}
dev, err := parseDevlinkDevice(respmsg)
if err == nil {
h.getEswitchAttrs(f, dev)
}
return dev, err
}
// DevlinkGetDeviceByName provides a pointer to devlink device and nil error,
// otherwise returns an error code.
func DevLinkGetDeviceByName(Bus string, Device string) (*DevlinkDevice, error) {
return pkgHandle.DevLinkGetDeviceByName(Bus, Device)
}
// DevLinkSetEswitchMode sets eswitch mode if able to set successfully or
// returns an error code.
// Equivalent to: `devlink dev eswitch set $dev mode switchdev`
// Equivalent to: `devlink dev eswitch set $dev mode legacy`
func (h *Handle) DevLinkSetEswitchMode(Dev *DevlinkDevice, NewMode string) error {
mode, err := eswitchStringToMode(NewMode)
if err != nil {
return err
}
_, req, err := h.createCmdReq(nl.DEVLINK_CMD_ESWITCH_SET, Dev.BusName, Dev.DeviceName)
if err != nil {
return err
}
req.AddData(nl.NewRtAttr(nl.DEVLINK_ATTR_ESWITCH_MODE, nl.Uint16Attr(mode)))
_, err = req.Execute(unix.NETLINK_GENERIC, 0)
return err
}
// DevLinkSetEswitchMode sets eswitch mode if able to set successfully or
// returns an error code.
// Equivalent to: `devlink dev eswitch set $dev mode switchdev`
// Equivalent to: `devlink dev eswitch set $dev mode legacy`
func DevLinkSetEswitchMode(Dev *DevlinkDevice, NewMode string) error {
return pkgHandle.DevLinkSetEswitchMode(Dev, NewMode)
}
func (port *DevlinkPort) parseAttributes(attrs []syscall.NetlinkRouteAttr) error {
for _, a := range attrs {
switch a.Attr.Type {
case nl.DEVLINK_ATTR_BUS_NAME:
port.BusName = string(a.Value)
case nl.DEVLINK_ATTR_DEV_NAME:
port.DeviceName = string(a.Value)
case nl.DEVLINK_ATTR_PORT_INDEX:
port.PortIndex = native.Uint32(a.Value)
case nl.DEVLINK_ATTR_PORT_TYPE:
port.PortType = native.Uint16(a.Value)
case nl.DEVLINK_ATTR_PORT_NETDEV_NAME:
port.NetdeviceName = string(a.Value)
case nl.DEVLINK_ATTR_PORT_NETDEV_IFINDEX:
port.NetdevIfIndex = native.Uint32(a.Value)
case nl.DEVLINK_ATTR_PORT_IBDEV_NAME:
port.RdmaDeviceName = string(a.Value)
case nl.DEVLINK_ATTR_PORT_FLAVOUR:
port.PortFlavour = native.Uint16(a.Value)
}
}
return nil
}
func parseDevLinkAllPortList(msgs [][]byte) ([]*DevlinkPort, error) {
ports := make([]*DevlinkPort, 0, len(msgs))
for _, m := range msgs {
attrs, err := nl.ParseRouteAttr(m[nl.SizeofGenlmsg:])
if err != nil {
return nil, err
}
port := &DevlinkPort{}
if err = port.parseAttributes(attrs); err != nil {
return nil, err
}
ports = append(ports, port)
}
return ports, nil
}
// DevLinkGetPortList provides a pointer to devlink ports and nil error,
// otherwise returns an error code.
func (h *Handle) DevLinkGetAllPortList() ([]*DevlinkPort, error) {
f, err := h.GenlFamilyGet(nl.GENL_DEVLINK_NAME)
if err != nil {
return nil, err
}
msg := &nl.Genlmsg{
Command: nl.DEVLINK_CMD_PORT_GET,
Version: nl.GENL_DEVLINK_VERSION,
}
req := h.newNetlinkRequest(int(f.ID),
unix.NLM_F_REQUEST|unix.NLM_F_ACK|unix.NLM_F_DUMP)
req.AddData(msg)
msgs, err := req.Execute(unix.NETLINK_GENERIC, 0)
if err != nil {
return nil, err
}
ports, err := parseDevLinkAllPortList(msgs)
if err != nil {
return nil, err
}
return ports, nil
}
// DevLinkGetPortList provides a pointer to devlink ports and nil error,
// otherwise returns an error code.
func DevLinkGetAllPortList() ([]*DevlinkPort, error) {
return pkgHandle.DevLinkGetAllPortList()
}
func parseDevlinkPortMsg(msgs [][]byte) (*DevlinkPort, error) {
m := msgs[0]
attrs, err := nl.ParseRouteAttr(m[nl.SizeofGenlmsg:])
if err != nil {
return nil, err
}
port := &DevlinkPort{}
if err = port.parseAttributes(attrs); err != nil {
return nil, err
}
return port, nil
}
// DevLinkGetPortByIndexprovides a pointer to devlink device and nil error,
// otherwise returns an error code.
func (h *Handle) DevLinkGetPortByIndex(Bus string, Device string, PortIndex uint32) (*DevlinkPort, error) {
_, req, err := h.createCmdReq(nl.DEVLINK_CMD_PORT_GET, Bus, Device)
if err != nil {
return nil, err
}
req.AddData(nl.NewRtAttr(nl.DEVLINK_ATTR_PORT_INDEX, nl.Uint32Attr(PortIndex)))
respmsg, err := req.Execute(unix.NETLINK_GENERIC, 0)
if err != nil {
return nil, err
}
port, err := parseDevlinkPortMsg(respmsg)
return port, err
}
// DevLinkGetPortByIndex provides a pointer to devlink portand nil error,
// otherwise returns an error code.
func DevLinkGetPortByIndex(Bus string, Device string, PortIndex uint32) (*DevlinkPort, error) {
return pkgHandle.DevLinkGetPortByIndex(Bus, Device, PortIndex)
}

View File

@@ -2,6 +2,7 @@ package netlink
import (
"fmt"
"net"
)
type Filter interface {
@@ -135,6 +136,27 @@ func (action *BpfAction) Attrs() *ActionAttrs {
return &action.ActionAttrs
}
type ConnmarkAction struct {
ActionAttrs
Zone uint16
}
func (action *ConnmarkAction) Type() string {
return "connmark"
}
func (action *ConnmarkAction) Attrs() *ActionAttrs {
return &action.ActionAttrs
}
func NewConnmarkAction() *ConnmarkAction {
return &ConnmarkAction{
ActionAttrs: ActionAttrs{
Action: TC_ACT_PIPE,
},
}
}
type MirredAct uint8
func (a MirredAct) String() string {
@@ -182,47 +204,60 @@ func NewMirredAction(redirIndex int) *MirredAction {
}
}
// Sel of the U32 filters that contains multiple TcU32Key. This is the copy
// and the frontend representation of nl.TcU32Sel. It is serialized into canonical
// nl.TcU32Sel with the appropriate endianness.
type TcU32Sel struct {
Flags uint8
Offshift uint8
Nkeys uint8
Pad uint8
Offmask uint16
Off uint16
Offoff int16
Hoff int16
Hmask uint32
Keys []TcU32Key
type TunnelKeyAct int8
const (
TCA_TUNNEL_KEY_SET TunnelKeyAct = 1 // set tunnel key
TCA_TUNNEL_KEY_UNSET TunnelKeyAct = 2 // unset tunnel key
)
type TunnelKeyAction struct {
ActionAttrs
Action TunnelKeyAct
SrcAddr net.IP
DstAddr net.IP
KeyID uint32
DestPort uint16
}
// TcU32Key contained of Sel in the U32 filters. This is the copy and the frontend
// representation of nl.TcU32Key. It is serialized into chanonical nl.TcU32Sel
// with the appropriate endianness.
type TcU32Key struct {
Mask uint32
Val uint32
Off int32
OffMask int32
func (action *TunnelKeyAction) Type() string {
return "tunnel_key"
}
// U32 filters on many packet related properties
type U32 struct {
FilterAttrs
ClassId uint32
RedirIndex int
Sel *TcU32Sel
Actions []Action
func (action *TunnelKeyAction) Attrs() *ActionAttrs {
return &action.ActionAttrs
}
func (filter *U32) Attrs() *FilterAttrs {
return &filter.FilterAttrs
func NewTunnelKeyAction() *TunnelKeyAction {
return &TunnelKeyAction{
ActionAttrs: ActionAttrs{
Action: TC_ACT_PIPE,
},
}
}
func (filter *U32) Type() string {
return "u32"
type SkbEditAction struct {
ActionAttrs
QueueMapping *uint16
PType *uint16
Priority *uint32
Mark *uint32
}
func (action *SkbEditAction) Type() string {
return "skbedit"
}
func (action *SkbEditAction) Attrs() *ActionAttrs {
return &action.ActionAttrs
}
func NewSkbEditAction() *SkbEditAction {
return &SkbEditAction{
ActionAttrs: ActionAttrs{
Action: TC_ACT_PIPE,
},
}
}
// MatchAll filters match all packets
@@ -262,6 +297,8 @@ type BpfFilter struct {
Fd int
Name string
DirectAction bool
Id int
Tag string
}
func (filter *BpfFilter) Type() string {

View File

@@ -3,10 +3,10 @@ package netlink
import (
"bytes"
"encoding/binary"
"encoding/hex"
"errors"
"fmt"
"syscall"
"unsafe"
"github.com/vishvananda/netlink/nl"
"golang.org/x/sys/unix"
@@ -20,6 +20,36 @@ const (
TC_U32_EAT = nl.TC_U32_EAT
)
// Sel of the U32 filters that contains multiple TcU32Key. This is the type
// alias and the frontend representation of nl.TcU32Sel. It is serialized into
// canonical nl.TcU32Sel with the appropriate endianness.
type TcU32Sel = nl.TcU32Sel
// TcU32Key contained of Sel in the U32 filters. This is the type alias and the
// frontend representation of nl.TcU32Key. It is serialized into chanonical
// nl.TcU32Sel with the appropriate endianness.
type TcU32Key = nl.TcU32Key
// U32 filters on many packet related properties
type U32 struct {
FilterAttrs
ClassId uint32
Divisor uint32 // Divisor MUST be power of 2.
Hash uint32
Link uint32
RedirIndex int
Sel *TcU32Sel
Actions []Action
}
func (filter *U32) Attrs() *FilterAttrs {
return &filter.FilterAttrs
}
func (filter *U32) Type() string {
return "u32"
}
// Fw filter filters on firewall marks
// NOTE: this is in filter_linux because it refers to nl.TcPolice which
// is defined in nl/tc_linux.go
@@ -59,7 +89,7 @@ func NewFw(attrs FilterAttrs, fattrs FilterFwAttrs) (*Fw, error) {
if CalcRtable(&police.Rate, rtab[:], rcellLog, fattrs.Mtu, linklayer) < 0 {
return nil, errors.New("TBF: failed to calculate rate table")
}
police.Burst = uint32(Xmittime(uint64(police.Rate.Rate), uint32(buffer)))
police.Burst = Xmittime(uint64(police.Rate.Rate), uint32(buffer))
}
police.Mtu = fattrs.Mtu
if police.PeakRate.Rate != 0 {
@@ -123,8 +153,24 @@ func FilterAdd(filter Filter) error {
// FilterAdd will add a filter to the system.
// Equivalent to: `tc filter add $filter`
func (h *Handle) FilterAdd(filter Filter) error {
return h.filterModify(filter, unix.NLM_F_CREATE|unix.NLM_F_EXCL)
}
// FilterReplace will replace a filter.
// Equivalent to: `tc filter replace $filter`
func FilterReplace(filter Filter) error {
return pkgHandle.FilterReplace(filter)
}
// FilterReplace will replace a filter.
// Equivalent to: `tc filter replace $filter`
func (h *Handle) FilterReplace(filter Filter) error {
return h.filterModify(filter, unix.NLM_F_CREATE)
}
func (h *Handle) filterModify(filter Filter, flags int) error {
native = nl.NativeEndian()
req := h.newNetlinkRequest(unix.RTM_NEWTFILTER, unix.NLM_F_CREATE|unix.NLM_F_EXCL|unix.NLM_F_ACK)
req := h.newNetlinkRequest(unix.RTM_NEWTFILTER, flags|unix.NLM_F_ACK)
base := filter.Attrs()
msg := &nl.TcMsg{
Family: nl.FAMILY_ALL,
@@ -140,8 +186,7 @@ func (h *Handle) FilterAdd(filter Filter) error {
switch filter := filter.(type) {
case *U32:
// Convert TcU32Sel into nl.TcU32Sel as it is without copy.
sel := (*nl.TcU32Sel)(unsafe.Pointer(filter.Sel))
sel := filter.Sel
if sel == nil {
// match all
sel = &nl.TcU32Sel{
@@ -168,11 +213,23 @@ func (h *Handle) FilterAdd(filter Filter) error {
}
}
sel.Nkeys = uint8(len(sel.Keys))
nl.NewRtAttrChild(options, nl.TCA_U32_SEL, sel.Serialize())
options.AddRtAttr(nl.TCA_U32_SEL, sel.Serialize())
if filter.ClassId != 0 {
nl.NewRtAttrChild(options, nl.TCA_U32_CLASSID, nl.Uint32Attr(filter.ClassId))
options.AddRtAttr(nl.TCA_U32_CLASSID, nl.Uint32Attr(filter.ClassId))
}
actionsAttr := nl.NewRtAttrChild(options, nl.TCA_U32_ACT, nil)
if filter.Divisor != 0 {
if (filter.Divisor-1)&filter.Divisor != 0 {
return fmt.Errorf("illegal divisor %d. Must be a power of 2", filter.Divisor)
}
options.AddRtAttr(nl.TCA_U32_DIVISOR, nl.Uint32Attr(filter.Divisor))
}
if filter.Hash != 0 {
options.AddRtAttr(nl.TCA_U32_HASH, nl.Uint32Attr(filter.Hash))
}
if filter.Link != 0 {
options.AddRtAttr(nl.TCA_U32_LINK, nl.Uint32Attr(filter.Link))
}
actionsAttr := options.AddRtAttr(nl.TCA_U32_ACT, nil)
// backwards compatibility
if filter.RedirIndex != 0 {
filter.Actions = append([]Action{NewMirredAction(filter.RedirIndex)}, filter.Actions...)
@@ -184,51 +241,51 @@ func (h *Handle) FilterAdd(filter Filter) error {
if filter.Mask != 0 {
b := make([]byte, 4)
native.PutUint32(b, filter.Mask)
nl.NewRtAttrChild(options, nl.TCA_FW_MASK, b)
options.AddRtAttr(nl.TCA_FW_MASK, b)
}
if filter.InDev != "" {
nl.NewRtAttrChild(options, nl.TCA_FW_INDEV, nl.ZeroTerminated(filter.InDev))
options.AddRtAttr(nl.TCA_FW_INDEV, nl.ZeroTerminated(filter.InDev))
}
if (filter.Police != nl.TcPolice{}) {
police := nl.NewRtAttrChild(options, nl.TCA_FW_POLICE, nil)
nl.NewRtAttrChild(police, nl.TCA_POLICE_TBF, filter.Police.Serialize())
police := options.AddRtAttr(nl.TCA_FW_POLICE, nil)
police.AddRtAttr(nl.TCA_POLICE_TBF, filter.Police.Serialize())
if (filter.Police.Rate != nl.TcRateSpec{}) {
payload := SerializeRtab(filter.Rtab)
nl.NewRtAttrChild(police, nl.TCA_POLICE_RATE, payload)
police.AddRtAttr(nl.TCA_POLICE_RATE, payload)
}
if (filter.Police.PeakRate != nl.TcRateSpec{}) {
payload := SerializeRtab(filter.Ptab)
nl.NewRtAttrChild(police, nl.TCA_POLICE_PEAKRATE, payload)
police.AddRtAttr(nl.TCA_POLICE_PEAKRATE, payload)
}
}
if filter.ClassId != 0 {
b := make([]byte, 4)
native.PutUint32(b, filter.ClassId)
nl.NewRtAttrChild(options, nl.TCA_FW_CLASSID, b)
options.AddRtAttr(nl.TCA_FW_CLASSID, b)
}
case *BpfFilter:
var bpfFlags uint32
if filter.ClassId != 0 {
nl.NewRtAttrChild(options, nl.TCA_BPF_CLASSID, nl.Uint32Attr(filter.ClassId))
options.AddRtAttr(nl.TCA_BPF_CLASSID, nl.Uint32Attr(filter.ClassId))
}
if filter.Fd >= 0 {
nl.NewRtAttrChild(options, nl.TCA_BPF_FD, nl.Uint32Attr((uint32(filter.Fd))))
options.AddRtAttr(nl.TCA_BPF_FD, nl.Uint32Attr((uint32(filter.Fd))))
}
if filter.Name != "" {
nl.NewRtAttrChild(options, nl.TCA_BPF_NAME, nl.ZeroTerminated(filter.Name))
options.AddRtAttr(nl.TCA_BPF_NAME, nl.ZeroTerminated(filter.Name))
}
if filter.DirectAction {
bpfFlags |= nl.TCA_BPF_FLAG_ACT_DIRECT
}
nl.NewRtAttrChild(options, nl.TCA_BPF_FLAGS, nl.Uint32Attr(bpfFlags))
options.AddRtAttr(nl.TCA_BPF_FLAGS, nl.Uint32Attr(bpfFlags))
case *MatchAll:
actionsAttr := nl.NewRtAttrChild(options, nl.TCA_MATCHALL_ACT, nil)
actionsAttr := options.AddRtAttr(nl.TCA_MATCHALL_ACT, nil)
if err := EncodeActions(actionsAttr, filter.Actions); err != nil {
return err
}
if filter.ClassId != 0 {
nl.NewRtAttrChild(options, nl.TCA_MATCHALL_CLASSID, nl.Uint32Attr(filter.ClassId))
options.AddRtAttr(nl.TCA_MATCHALL_CLASSID, nl.Uint32Attr(filter.ClassId))
}
}
@@ -366,34 +423,94 @@ func EncodeActions(attr *nl.RtAttr, actions []Action) error {
default:
return fmt.Errorf("unknown action type %s", action.Type())
case *MirredAction:
table := nl.NewRtAttrChild(attr, tabIndex, nil)
table := attr.AddRtAttr(tabIndex, nil)
tabIndex++
nl.NewRtAttrChild(table, nl.TCA_ACT_KIND, nl.ZeroTerminated("mirred"))
aopts := nl.NewRtAttrChild(table, nl.TCA_ACT_OPTIONS, nil)
table.AddRtAttr(nl.TCA_ACT_KIND, nl.ZeroTerminated("mirred"))
aopts := table.AddRtAttr(nl.TCA_ACT_OPTIONS, nil)
mirred := nl.TcMirred{
Eaction: int32(action.MirredAction),
Ifindex: uint32(action.Ifindex),
}
toTcGen(action.Attrs(), &mirred.TcGen)
nl.NewRtAttrChild(aopts, nl.TCA_MIRRED_PARMS, mirred.Serialize())
aopts.AddRtAttr(nl.TCA_MIRRED_PARMS, mirred.Serialize())
case *TunnelKeyAction:
table := attr.AddRtAttr(tabIndex, nil)
tabIndex++
table.AddRtAttr(nl.TCA_ACT_KIND, nl.ZeroTerminated("tunnel_key"))
aopts := table.AddRtAttr(nl.TCA_ACT_OPTIONS, nil)
tun := nl.TcTunnelKey{
Action: int32(action.Action),
}
toTcGen(action.Attrs(), &tun.TcGen)
aopts.AddRtAttr(nl.TCA_TUNNEL_KEY_PARMS, tun.Serialize())
if action.Action == TCA_TUNNEL_KEY_SET {
aopts.AddRtAttr(nl.TCA_TUNNEL_KEY_ENC_KEY_ID, htonl(action.KeyID))
if v4 := action.SrcAddr.To4(); v4 != nil {
aopts.AddRtAttr(nl.TCA_TUNNEL_KEY_ENC_IPV4_SRC, v4[:])
} else if v6 := action.SrcAddr.To16(); v6 != nil {
aopts.AddRtAttr(nl.TCA_TUNNEL_KEY_ENC_IPV6_SRC, v6[:])
} else {
return fmt.Errorf("invalid src addr %s for tunnel_key action", action.SrcAddr)
}
if v4 := action.DstAddr.To4(); v4 != nil {
aopts.AddRtAttr(nl.TCA_TUNNEL_KEY_ENC_IPV4_DST, v4[:])
} else if v6 := action.DstAddr.To16(); v6 != nil {
aopts.AddRtAttr(nl.TCA_TUNNEL_KEY_ENC_IPV6_DST, v6[:])
} else {
return fmt.Errorf("invalid dst addr %s for tunnel_key action", action.DstAddr)
}
if action.DestPort != 0 {
aopts.AddRtAttr(nl.TCA_TUNNEL_KEY_ENC_DST_PORT, htons(action.DestPort))
}
}
case *SkbEditAction:
table := attr.AddRtAttr(tabIndex, nil)
tabIndex++
table.AddRtAttr(nl.TCA_ACT_KIND, nl.ZeroTerminated("skbedit"))
aopts := table.AddRtAttr(nl.TCA_ACT_OPTIONS, nil)
skbedit := nl.TcSkbEdit{}
toTcGen(action.Attrs(), &skbedit.TcGen)
aopts.AddRtAttr(nl.TCA_SKBEDIT_PARMS, skbedit.Serialize())
if action.QueueMapping != nil {
aopts.AddRtAttr(nl.TCA_SKBEDIT_QUEUE_MAPPING, nl.Uint16Attr(*action.QueueMapping))
}
if action.Priority != nil {
aopts.AddRtAttr(nl.TCA_SKBEDIT_PRIORITY, nl.Uint32Attr(*action.Priority))
}
if action.PType != nil {
aopts.AddRtAttr(nl.TCA_SKBEDIT_PTYPE, nl.Uint16Attr(*action.PType))
}
if action.Mark != nil {
aopts.AddRtAttr(nl.TCA_SKBEDIT_MARK, nl.Uint32Attr(*action.Mark))
}
case *ConnmarkAction:
table := attr.AddRtAttr(tabIndex, nil)
tabIndex++
table.AddRtAttr(nl.TCA_ACT_KIND, nl.ZeroTerminated("connmark"))
aopts := table.AddRtAttr(nl.TCA_ACT_OPTIONS, nil)
connmark := nl.TcConnmark{
Zone: action.Zone,
}
toTcGen(action.Attrs(), &connmark.TcGen)
aopts.AddRtAttr(nl.TCA_CONNMARK_PARMS, connmark.Serialize())
case *BpfAction:
table := nl.NewRtAttrChild(attr, tabIndex, nil)
table := attr.AddRtAttr(tabIndex, nil)
tabIndex++
nl.NewRtAttrChild(table, nl.TCA_ACT_KIND, nl.ZeroTerminated("bpf"))
aopts := nl.NewRtAttrChild(table, nl.TCA_ACT_OPTIONS, nil)
table.AddRtAttr(nl.TCA_ACT_KIND, nl.ZeroTerminated("bpf"))
aopts := table.AddRtAttr(nl.TCA_ACT_OPTIONS, nil)
gen := nl.TcGen{}
toTcGen(action.Attrs(), &gen)
nl.NewRtAttrChild(aopts, nl.TCA_ACT_BPF_PARMS, gen.Serialize())
nl.NewRtAttrChild(aopts, nl.TCA_ACT_BPF_FD, nl.Uint32Attr(uint32(action.Fd)))
nl.NewRtAttrChild(aopts, nl.TCA_ACT_BPF_NAME, nl.ZeroTerminated(action.Name))
aopts.AddRtAttr(nl.TCA_ACT_BPF_PARMS, gen.Serialize())
aopts.AddRtAttr(nl.TCA_ACT_BPF_FD, nl.Uint32Attr(uint32(action.Fd)))
aopts.AddRtAttr(nl.TCA_ACT_BPF_NAME, nl.ZeroTerminated(action.Name))
case *GenericAction:
table := nl.NewRtAttrChild(attr, tabIndex, nil)
table := attr.AddRtAttr(tabIndex, nil)
tabIndex++
nl.NewRtAttrChild(table, nl.TCA_ACT_KIND, nl.ZeroTerminated("gact"))
aopts := nl.NewRtAttrChild(table, nl.TCA_ACT_OPTIONS, nil)
table.AddRtAttr(nl.TCA_ACT_KIND, nl.ZeroTerminated("gact"))
aopts := table.AddRtAttr(nl.TCA_ACT_OPTIONS, nil)
gen := nl.TcGen{}
toTcGen(action.Attrs(), &gen)
nl.NewRtAttrChild(aopts, nl.TCA_GACT_PARMS, gen.Serialize())
aopts.AddRtAttr(nl.TCA_GACT_PARMS, gen.Serialize())
}
}
return nil
@@ -419,8 +536,14 @@ func parseActions(tables []syscall.NetlinkRouteAttr) ([]Action, error) {
action = &MirredAction{}
case "bpf":
action = &BpfAction{}
case "connmark":
action = &ConnmarkAction{}
case "gact":
action = &GenericAction{}
case "tunnel_key":
action = &TunnelKeyAction{}
case "skbedit":
action = &SkbEditAction{}
default:
break nextattr
}
@@ -435,11 +558,46 @@ func parseActions(tables []syscall.NetlinkRouteAttr) ([]Action, error) {
switch adatum.Attr.Type {
case nl.TCA_MIRRED_PARMS:
mirred := *nl.DeserializeTcMirred(adatum.Value)
toAttrs(&mirred.TcGen, action.Attrs())
action.(*MirredAction).ActionAttrs = ActionAttrs{}
toAttrs(&mirred.TcGen, action.Attrs())
action.(*MirredAction).Ifindex = int(mirred.Ifindex)
action.(*MirredAction).MirredAction = MirredAct(mirred.Eaction)
}
case "tunnel_key":
switch adatum.Attr.Type {
case nl.TCA_TUNNEL_KEY_PARMS:
tun := *nl.DeserializeTunnelKey(adatum.Value)
action.(*TunnelKeyAction).ActionAttrs = ActionAttrs{}
toAttrs(&tun.TcGen, action.Attrs())
action.(*TunnelKeyAction).Action = TunnelKeyAct(tun.Action)
case nl.TCA_TUNNEL_KEY_ENC_KEY_ID:
action.(*TunnelKeyAction).KeyID = networkOrder.Uint32(adatum.Value[0:4])
case nl.TCA_TUNNEL_KEY_ENC_IPV6_SRC, nl.TCA_TUNNEL_KEY_ENC_IPV4_SRC:
action.(*TunnelKeyAction).SrcAddr = adatum.Value[:]
case nl.TCA_TUNNEL_KEY_ENC_IPV6_DST, nl.TCA_TUNNEL_KEY_ENC_IPV4_DST:
action.(*TunnelKeyAction).DstAddr = adatum.Value[:]
case nl.TCA_TUNNEL_KEY_ENC_DST_PORT:
action.(*TunnelKeyAction).DestPort = ntohs(adatum.Value)
}
case "skbedit":
switch adatum.Attr.Type {
case nl.TCA_SKBEDIT_PARMS:
skbedit := *nl.DeserializeSkbEdit(adatum.Value)
action.(*SkbEditAction).ActionAttrs = ActionAttrs{}
toAttrs(&skbedit.TcGen, action.Attrs())
case nl.TCA_SKBEDIT_MARK:
mark := native.Uint32(adatum.Value[0:4])
action.(*SkbEditAction).Mark = &mark
case nl.TCA_SKBEDIT_PRIORITY:
priority := native.Uint32(adatum.Value[0:4])
action.(*SkbEditAction).Priority = &priority
case nl.TCA_SKBEDIT_PTYPE:
ptype := native.Uint16(adatum.Value[0:2])
action.(*SkbEditAction).PType = &ptype
case nl.TCA_SKBEDIT_QUEUE_MAPPING:
mapping := native.Uint16(adatum.Value[0:2])
action.(*SkbEditAction).QueueMapping = &mapping
}
case "bpf":
switch adatum.Attr.Type {
case nl.TCA_ACT_BPF_PARMS:
@@ -450,6 +608,14 @@ func parseActions(tables []syscall.NetlinkRouteAttr) ([]Action, error) {
case nl.TCA_ACT_BPF_NAME:
action.(*BpfAction).Name = string(adatum.Value[:len(adatum.Value)-1])
}
case "connmark":
switch adatum.Attr.Type {
case nl.TCA_CONNMARK_PARMS:
connmark := *nl.DeserializeTcConnmark(adatum.Value)
action.(*ConnmarkAction).ActionAttrs = ActionAttrs{}
toAttrs(&connmark.TcGen, action.Attrs())
action.(*ConnmarkAction).Zone = connmark.Zone
}
case "gact":
switch adatum.Attr.Type {
case nl.TCA_GACT_PARMS:
@@ -474,7 +640,7 @@ func parseU32Data(filter Filter, data []syscall.NetlinkRouteAttr) (bool, error)
case nl.TCA_U32_SEL:
detailed = true
sel := nl.DeserializeTcU32Sel(datum.Value)
u32.Sel = (*TcU32Sel)(unsafe.Pointer(sel))
u32.Sel = sel
if native != networkOrder {
// Handle the endianness of attributes
u32.Sel.Offmask = native.Uint16(htons(sel.Offmask))
@@ -500,6 +666,12 @@ func parseU32Data(filter Filter, data []syscall.NetlinkRouteAttr) (bool, error)
}
case nl.TCA_U32_CLASSID:
u32.ClassId = native.Uint32(datum.Value)
case nl.TCA_U32_DIVISOR:
u32.Divisor = native.Uint32(datum.Value)
case nl.TCA_U32_HASH:
u32.Hash = native.Uint32(datum.Value)
case nl.TCA_U32_LINK:
u32.Link = native.Uint32(datum.Value)
}
}
return detailed, nil
@@ -551,6 +723,10 @@ func parseBpfData(filter Filter, data []syscall.NetlinkRouteAttr) (bool, error)
if (flags & nl.TCA_BPF_FLAG_ACT_DIRECT) != 0 {
bpf.DirectAction = true
}
case nl.TCA_BPF_ID:
bpf.Id = int(native.Uint32(datum.Value[0:4]))
case nl.TCA_BPF_TAG:
bpf.Tag = hex.EncodeToString(datum.Value[:len(datum.Value)-1])
}
}
return detailed, nil
@@ -615,7 +791,7 @@ func CalcRtable(rate *nl.TcRateSpec, rtab []uint32, cellLog int, mtu uint32, lin
}
for i := 0; i < 256; i++ {
sz = AdjustSize(uint((i+1)<<uint32(cellLog)), uint(mpu), linklayer)
rtab[i] = uint32(Xmittime(uint64(bps), uint32(sz)))
rtab[i] = Xmittime(uint64(bps), uint32(sz))
}
rate.CellAlign = -1
rate.CellLog = uint8(cellLog)

View File

@@ -90,11 +90,7 @@ func (h *Handle) FouAdd(f Fou) error {
req.AddRawData(raw)
_, err = req.Execute(unix.NETLINK_GENERIC, 0)
if err != nil {
return err
}
return nil
return err
}
func FouDel(f Fou) error {

View File

@@ -157,6 +157,9 @@ func (h *Handle) GenlFamilyGet(name string) (*GenlFamily, error) {
return nil, err
}
families, err := parseFamilies(msgs)
if err != nil {
return nil, err
}
if len(families) != 1 {
return nil, fmt.Errorf("invalid response for GENL_CTRL_CMD_GETFAMILY")
}

View File

@@ -21,6 +21,22 @@ type Handle struct {
lookupByDump bool
}
// SetSocketTimeout configures timeout for default netlink sockets
func SetSocketTimeout(to time.Duration) error {
if to < time.Microsecond {
return fmt.Errorf("invalid timeout, minimul value is %s", time.Microsecond)
}
nl.SocketTimeoutTv = unix.NsecToTimeval(to.Nanoseconds())
return nil
}
// GetSocketTimeout returns the timeout value used by default netlink sockets
func GetSocketTimeout() time.Duration {
nsec := unix.TimevalToNsec(nl.SocketTimeoutTv)
return time.Duration(nsec) * time.Nanosecond
}
// SupportsNetlinkFamily reports whether the passed netlink family is supported by this Handle
func (h *Handle) SupportsNetlinkFamily(nlFamily int) bool {
_, ok := h.sockets[nlFamily]
@@ -91,7 +107,7 @@ func (h *Handle) GetSocketReceiveBufferSize() ([]int, error) {
return results, nil
}
// NewHandle returns a netlink handle on the network namespace
// NewHandleAt returns a netlink handle on the network namespace
// specified by ns. If ns=netns.None(), current network namespace
// will be assumed
func NewHandleAt(ns netns.NsHandle, nlFamilies ...int) (*Handle, error) {

View File

@@ -73,11 +73,19 @@ func (h *Handle) LinkSetVfVlan(link Link, vf, vlan int) error {
return ErrNotImplemented
}
func (h *Handle) LinkSetVfVlanQos(link Link, vf, vlan, qos int) error {
return ErrNotImplemented
}
func (h *Handle) LinkSetVfTxRate(link Link, vf, rate int) error {
return ErrNotImplemented
}
func (h *Handle) LinkSetMaster(link Link, master *Bridge) error {
func (h *Handle) LinkSetVfRate(link Link, vf, minRate, maxRate int) error {
return ErrNotImplemented
}
func (h *Handle) LinkSetMaster(link Link, master Link) error {
return ErrNotImplemented
}
@@ -149,6 +157,10 @@ func (h *Handle) LinkSetTxQLen(link Link, qlen int) error {
return ErrNotImplemented
}
func (h *Handle) LinkSetGroup(link Link, group int) error {
return ErrNotImplemented
}
func (h *Handle) setProtinfoAttr(link Link, mode bool, attr int) error {
return ErrNotImplemented
}
@@ -225,6 +237,10 @@ func (h *Handle) RouteAdd(route *Route) error {
return ErrNotImplemented
}
func (h *Handle) RouteAppend(route *Route) error {
return ErrNotImplemented
}
func (h *Handle) RouteDel(route *Route) error {
return ErrNotImplemented
}

31
vendor/github.com/vishvananda/netlink/inet_diag.go generated vendored Normal file
View File

@@ -0,0 +1,31 @@
package netlink
// INET_DIAG constatns
const (
INET_DIAG_NONE = iota
INET_DIAG_MEMINFO
INET_DIAG_INFO
INET_DIAG_VEGASINFO
INET_DIAG_CONG
INET_DIAG_TOS
INET_DIAG_TCLASS
INET_DIAG_SKMEMINFO
INET_DIAG_SHUTDOWN
INET_DIAG_DCTCPINFO
INET_DIAG_PROTOCOL
INET_DIAG_SKV6ONLY
INET_DIAG_LOCALS
INET_DIAG_PEERS
INET_DIAG_PAD
INET_DIAG_MARK
INET_DIAG_BBRINFO
INET_DIAG_CLASS_ID
INET_DIAG_MD5SIG
INET_DIAG_MAX
)
type InetDiagTCPInfoResp struct {
InetDiagMsg *Socket
TCPInfo *TCPInfo
TCPBBRInfo *TCPBBRInfo
}

View File

@@ -56,18 +56,10 @@ type ethtoolSset struct {
data [1]uint32
}
// ethtoolGstrings is string set for data tagging
type ethtoolGstrings struct {
cmd uint32
stringSet uint32
length uint32
data [32]byte
}
type ethtoolStats struct {
cmd uint32
nStats uint32
data [1]uint64
// Followed by nStats * []uint64.
}
// newIocltSlaveReq returns filled IfreqSlave with proper interface names

365
vendor/github.com/vishvananda/netlink/ipset_linux.go generated vendored Normal file
View File

@@ -0,0 +1,365 @@
package netlink
import (
"log"
"net"
"syscall"
"github.com/vishvananda/netlink/nl"
"golang.org/x/sys/unix"
)
// IPSetEntry is used for adding, updating, retreiving and deleting entries
type IPSetEntry struct {
Comment string
MAC net.HardwareAddr
IP net.IP
Timeout *uint32
Packets *uint64
Bytes *uint64
Replace bool // replace existing entry
}
// IPSetResult is the result of a dump request for a set
type IPSetResult struct {
Nfgenmsg *nl.Nfgenmsg
Protocol uint8
ProtocolMinVersion uint8
Revision uint8
Family uint8
Flags uint8
SetName string
TypeName string
Comment string
HashSize uint32
NumEntries uint32
MaxElements uint32
References uint32
SizeInMemory uint32
CadtFlags uint32
Timeout *uint32
LineNo uint32
Entries []IPSetEntry
}
// IpsetCreateOptions is the options struct for creating a new ipset
type IpsetCreateOptions struct {
Replace bool // replace existing ipset
Timeout *uint32
Counters bool
Comments bool
Skbinfo bool
}
// IpsetProtocol returns the ipset protocol version from the kernel
func IpsetProtocol() (uint8, uint8, error) {
return pkgHandle.IpsetProtocol()
}
// IpsetCreate creates a new ipset
func IpsetCreate(setname, typename string, options IpsetCreateOptions) error {
return pkgHandle.IpsetCreate(setname, typename, options)
}
// IpsetDestroy destroys an existing ipset
func IpsetDestroy(setname string) error {
return pkgHandle.IpsetDestroy(setname)
}
// IpsetFlush flushes an existing ipset
func IpsetFlush(setname string) error {
return pkgHandle.IpsetFlush(setname)
}
// IpsetList dumps an specific ipset.
func IpsetList(setname string) (*IPSetResult, error) {
return pkgHandle.IpsetList(setname)
}
// IpsetListAll dumps all ipsets.
func IpsetListAll() ([]IPSetResult, error) {
return pkgHandle.IpsetListAll()
}
// IpsetAdd adds an entry to an existing ipset.
func IpsetAdd(setname string, entry *IPSetEntry) error {
return pkgHandle.ipsetAddDel(nl.IPSET_CMD_ADD, setname, entry)
}
// IpsetDel deletes an entry from an existing ipset.
func IpsetDel(setname string, entry *IPSetEntry) error {
return pkgHandle.ipsetAddDel(nl.IPSET_CMD_DEL, setname, entry)
}
func (h *Handle) IpsetProtocol() (protocol uint8, minVersion uint8, err error) {
req := h.newIpsetRequest(nl.IPSET_CMD_PROTOCOL)
msgs, err := req.Execute(unix.NETLINK_NETFILTER, 0)
if err != nil {
return 0, 0, err
}
response := ipsetUnserialize(msgs)
return response.Protocol, response.ProtocolMinVersion, nil
}
func (h *Handle) IpsetCreate(setname, typename string, options IpsetCreateOptions) error {
req := h.newIpsetRequest(nl.IPSET_CMD_CREATE)
if !options.Replace {
req.Flags |= unix.NLM_F_EXCL
}
req.AddData(nl.NewRtAttr(nl.IPSET_ATTR_SETNAME, nl.ZeroTerminated(setname)))
req.AddData(nl.NewRtAttr(nl.IPSET_ATTR_TYPENAME, nl.ZeroTerminated(typename)))
req.AddData(nl.NewRtAttr(nl.IPSET_ATTR_REVISION, nl.Uint8Attr(0)))
req.AddData(nl.NewRtAttr(nl.IPSET_ATTR_FAMILY, nl.Uint8Attr(2))) // 2 == inet
data := nl.NewRtAttr(nl.IPSET_ATTR_DATA|int(nl.NLA_F_NESTED), nil)
if timeout := options.Timeout; timeout != nil {
data.AddChild(&nl.Uint32Attribute{Type: nl.IPSET_ATTR_TIMEOUT | nl.NLA_F_NET_BYTEORDER, Value: *timeout})
}
var cadtFlags uint32
if options.Comments {
cadtFlags |= nl.IPSET_FLAG_WITH_COMMENT
}
if options.Counters {
cadtFlags |= nl.IPSET_FLAG_WITH_COUNTERS
}
if options.Skbinfo {
cadtFlags |= nl.IPSET_FLAG_WITH_SKBINFO
}
if cadtFlags != 0 {
data.AddChild(&nl.Uint32Attribute{Type: nl.IPSET_ATTR_CADT_FLAGS | nl.NLA_F_NET_BYTEORDER, Value: cadtFlags})
}
req.AddData(data)
_, err := ipsetExecute(req)
return err
}
func (h *Handle) IpsetDestroy(setname string) error {
req := h.newIpsetRequest(nl.IPSET_CMD_DESTROY)
req.AddData(nl.NewRtAttr(nl.IPSET_ATTR_SETNAME, nl.ZeroTerminated(setname)))
_, err := ipsetExecute(req)
return err
}
func (h *Handle) IpsetFlush(setname string) error {
req := h.newIpsetRequest(nl.IPSET_CMD_FLUSH)
req.AddData(nl.NewRtAttr(nl.IPSET_ATTR_SETNAME, nl.ZeroTerminated(setname)))
_, err := ipsetExecute(req)
return err
}
func (h *Handle) IpsetList(name string) (*IPSetResult, error) {
req := h.newIpsetRequest(nl.IPSET_CMD_LIST)
req.AddData(nl.NewRtAttr(nl.IPSET_ATTR_SETNAME, nl.ZeroTerminated(name)))
msgs, err := ipsetExecute(req)
if err != nil {
return nil, err
}
result := ipsetUnserialize(msgs)
return &result, nil
}
func (h *Handle) IpsetListAll() ([]IPSetResult, error) {
req := h.newIpsetRequest(nl.IPSET_CMD_LIST)
msgs, err := ipsetExecute(req)
if err != nil {
return nil, err
}
result := make([]IPSetResult, len(msgs))
for i, msg := range msgs {
result[i].unserialize(msg)
}
return result, nil
}
func (h *Handle) ipsetAddDel(nlCmd int, setname string, entry *IPSetEntry) error {
req := h.newIpsetRequest(nlCmd)
req.AddData(nl.NewRtAttr(nl.IPSET_ATTR_SETNAME, nl.ZeroTerminated(setname)))
if entry.Comment != "" {
req.AddData(nl.NewRtAttr(nl.IPSET_ATTR_COMMENT, nl.ZeroTerminated(entry.Comment)))
}
data := nl.NewRtAttr(nl.IPSET_ATTR_DATA|int(nl.NLA_F_NESTED), nil)
if !entry.Replace {
req.Flags |= unix.NLM_F_EXCL
}
if entry.Timeout != nil {
data.AddChild(&nl.Uint32Attribute{Type: nl.IPSET_ATTR_TIMEOUT | nl.NLA_F_NET_BYTEORDER, Value: *entry.Timeout})
}
if entry.MAC != nil {
nestedData := nl.NewRtAttr(nl.IPSET_ATTR_ETHER|int(nl.NLA_F_NET_BYTEORDER), entry.MAC)
data.AddChild(nl.NewRtAttr(nl.IPSET_ATTR_ETHER|int(nl.NLA_F_NESTED), nestedData.Serialize()))
}
if entry.IP != nil {
nestedData := nl.NewRtAttr(nl.IPSET_ATTR_IP|int(nl.NLA_F_NET_BYTEORDER), entry.IP)
data.AddChild(nl.NewRtAttr(nl.IPSET_ATTR_IP|int(nl.NLA_F_NESTED), nestedData.Serialize()))
}
data.AddChild(&nl.Uint32Attribute{Type: nl.IPSET_ATTR_LINENO | nl.NLA_F_NET_BYTEORDER, Value: 0})
req.AddData(data)
_, err := ipsetExecute(req)
return err
}
func (h *Handle) newIpsetRequest(cmd int) *nl.NetlinkRequest {
req := h.newNetlinkRequest(cmd|(unix.NFNL_SUBSYS_IPSET<<8), nl.GetIpsetFlags(cmd))
// Add the netfilter header
msg := &nl.Nfgenmsg{
NfgenFamily: uint8(unix.AF_NETLINK),
Version: nl.NFNETLINK_V0,
ResId: 0,
}
req.AddData(msg)
req.AddData(nl.NewRtAttr(nl.IPSET_ATTR_PROTOCOL, nl.Uint8Attr(nl.IPSET_PROTOCOL)))
return req
}
func ipsetExecute(req *nl.NetlinkRequest) (msgs [][]byte, err error) {
msgs, err = req.Execute(unix.NETLINK_NETFILTER, 0)
if err != nil {
if errno := int(err.(syscall.Errno)); errno >= nl.IPSET_ERR_PRIVATE {
err = nl.IPSetError(uintptr(errno))
}
}
return
}
func ipsetUnserialize(msgs [][]byte) (result IPSetResult) {
for _, msg := range msgs {
result.unserialize(msg)
}
return result
}
func (result *IPSetResult) unserialize(msg []byte) {
result.Nfgenmsg = nl.DeserializeNfgenmsg(msg)
for attr := range nl.ParseAttributes(msg[4:]) {
switch attr.Type {
case nl.IPSET_ATTR_PROTOCOL:
result.Protocol = attr.Value[0]
case nl.IPSET_ATTR_SETNAME:
result.SetName = nl.BytesToString(attr.Value)
case nl.IPSET_ATTR_COMMENT:
result.Comment = nl.BytesToString(attr.Value)
case nl.IPSET_ATTR_TYPENAME:
result.TypeName = nl.BytesToString(attr.Value)
case nl.IPSET_ATTR_REVISION:
result.Revision = attr.Value[0]
case nl.IPSET_ATTR_FAMILY:
result.Family = attr.Value[0]
case nl.IPSET_ATTR_FLAGS:
result.Flags = attr.Value[0]
case nl.IPSET_ATTR_DATA | nl.NLA_F_NESTED:
result.parseAttrData(attr.Value)
case nl.IPSET_ATTR_ADT | nl.NLA_F_NESTED:
result.parseAttrADT(attr.Value)
case nl.IPSET_ATTR_PROTOCOL_MIN:
result.ProtocolMinVersion = attr.Value[0]
default:
log.Printf("unknown ipset attribute from kernel: %+v %v", attr, attr.Type&nl.NLA_TYPE_MASK)
}
}
}
func (result *IPSetResult) parseAttrData(data []byte) {
for attr := range nl.ParseAttributes(data) {
switch attr.Type {
case nl.IPSET_ATTR_HASHSIZE | nl.NLA_F_NET_BYTEORDER:
result.HashSize = attr.Uint32()
case nl.IPSET_ATTR_MAXELEM | nl.NLA_F_NET_BYTEORDER:
result.MaxElements = attr.Uint32()
case nl.IPSET_ATTR_TIMEOUT | nl.NLA_F_NET_BYTEORDER:
val := attr.Uint32()
result.Timeout = &val
case nl.IPSET_ATTR_ELEMENTS | nl.NLA_F_NET_BYTEORDER:
result.NumEntries = attr.Uint32()
case nl.IPSET_ATTR_REFERENCES | nl.NLA_F_NET_BYTEORDER:
result.References = attr.Uint32()
case nl.IPSET_ATTR_MEMSIZE | nl.NLA_F_NET_BYTEORDER:
result.SizeInMemory = attr.Uint32()
case nl.IPSET_ATTR_CADT_FLAGS | nl.NLA_F_NET_BYTEORDER:
result.CadtFlags = attr.Uint32()
case nl.IPSET_ATTR_IP | nl.NLA_F_NESTED:
for nested := range nl.ParseAttributes(attr.Value) {
switch nested.Type {
case nl.IPSET_ATTR_IP | nl.NLA_F_NET_BYTEORDER:
result.Entries = append(result.Entries, IPSetEntry{IP: nested.Value})
}
}
case nl.IPSET_ATTR_CADT_LINENO | nl.NLA_F_NET_BYTEORDER:
result.LineNo = attr.Uint32()
case nl.IPSET_ATTR_COMMENT:
result.Comment = nl.BytesToString(attr.Value)
default:
log.Printf("unknown ipset data attribute from kernel: %+v %v", attr, attr.Type&nl.NLA_TYPE_MASK)
}
}
}
func (result *IPSetResult) parseAttrADT(data []byte) {
for attr := range nl.ParseAttributes(data) {
switch attr.Type {
case nl.IPSET_ATTR_DATA | nl.NLA_F_NESTED:
result.Entries = append(result.Entries, parseIPSetEntry(attr.Value))
default:
log.Printf("unknown ADT attribute from kernel: %+v %v", attr, attr.Type&nl.NLA_TYPE_MASK)
}
}
}
func parseIPSetEntry(data []byte) (entry IPSetEntry) {
for attr := range nl.ParseAttributes(data) {
switch attr.Type {
case nl.IPSET_ATTR_TIMEOUT | nl.NLA_F_NET_BYTEORDER:
val := attr.Uint32()
entry.Timeout = &val
case nl.IPSET_ATTR_BYTES | nl.NLA_F_NET_BYTEORDER:
val := attr.Uint64()
entry.Bytes = &val
case nl.IPSET_ATTR_PACKETS | nl.NLA_F_NET_BYTEORDER:
val := attr.Uint64()
entry.Packets = &val
case nl.IPSET_ATTR_ETHER:
entry.MAC = net.HardwareAddr(attr.Value)
case nl.IPSET_ATTR_IP:
entry.IP = net.IP(attr.Value)
case nl.IPSET_ATTR_COMMENT:
entry.Comment = nl.BytesToString(attr.Value)
case nl.IPSET_ATTR_IP | nl.NLA_F_NESTED:
for attr := range nl.ParseAttributes(attr.Value) {
switch attr.Type {
case nl.IPSET_ATTR_IP:
entry.IP = net.IP(attr.Value)
default:
log.Printf("unknown nested ADT attribute from kernel: %+v", attr)
}
}
default:
log.Printf("unknown ADT attribute from kernel: %+v", attr)
}
}
return
}

View File

@@ -4,6 +4,7 @@ import (
"fmt"
"net"
"os"
"strconv"
)
// Link represents a link device from netlink. Shared link attributes
@@ -41,6 +42,40 @@ type LinkAttrs struct {
NetNsID int
NumTxQueues int
NumRxQueues int
GSOMaxSize uint32
GSOMaxSegs uint32
Vfs []VfInfo // virtual functions available on link
Group uint32
Slave LinkSlave
}
// LinkSlave represents a slave device.
type LinkSlave interface {
SlaveType() string
}
// VfInfo represents configuration of virtual function
type VfInfo struct {
ID int
Mac net.HardwareAddr
Vlan int
Qos int
TxRate int // IFLA_VF_TX_RATE Max TxRate
Spoofchk bool
LinkState uint32
MaxTxRate uint32 // IFLA_VF_RATE Max TxRate
MinTxRate uint32 // IFLA_VF_RATE Min TxRate
RxPackets uint64
TxPackets uint64
RxBytes uint64
TxBytes uint64
Multicast uint64
Broadcast uint64
RxDropped uint64
TxDropped uint64
RssQuery uint32
Trust uint32
}
// LinkOperState represents the values of the IFLA_OPERSTATE link
@@ -79,7 +114,8 @@ func (s LinkOperState) String() string {
// NewLinkAttrs returns LinkAttrs structure filled with default values
func NewLinkAttrs() LinkAttrs {
return LinkAttrs{
TxQLen: -1,
NetNsID: -1,
TxQLen: -1,
}
}
@@ -172,10 +208,11 @@ type LinkStatistics64 struct {
}
type LinkXdp struct {
Fd int
Attached bool
Flags uint32
ProgId uint32
Fd int
Attached bool
AttachMode uint32
Flags uint32
ProgId uint32
}
// Device links cannot be created via netlink. These links
@@ -222,7 +259,9 @@ func (ifb *Ifb) Type() string {
type Bridge struct {
LinkAttrs
MulticastSnooping *bool
AgeingTime *uint32
HelloTime *uint32
VlanFiltering *bool
}
func (bridge *Bridge) Attrs() *LinkAttrs {
@@ -236,7 +275,8 @@ func (bridge *Bridge) Type() string {
// Vlan links have ParentIndex set in their Attrs()
type Vlan struct {
LinkAttrs
VlanId int
VlanId int
VlanProtocol VlanProtocol
}
func (vlan *Vlan) Attrs() *LinkAttrs {
@@ -290,10 +330,13 @@ type TuntapFlag uint16
// Tuntap links created via /dev/tun/tap, but can be destroyed via netlink
type Tuntap struct {
LinkAttrs
Mode TuntapMode
Flags TuntapFlag
Queues int
Fds []*os.File
Mode TuntapMode
Flags TuntapFlag
NonPersist bool
Queues int
Fds []*os.File
Owner uint32
Group uint32
}
func (tuntap *Tuntap) Attrs() *LinkAttrs {
@@ -307,7 +350,9 @@ func (tuntap *Tuntap) Type() string {
// Veth devices must specify PeerName on create
type Veth struct {
LinkAttrs
PeerName string // veth on create only
PeerName string // veth on create only
PeerHardwareAddr net.HardwareAddr
PeerNamespace interface{}
}
func (veth *Veth) Attrs() *LinkAttrs {
@@ -318,6 +363,19 @@ func (veth *Veth) Type() string {
return "veth"
}
// Wireguard represent links of type "wireguard", see https://www.wireguard.com/
type Wireguard struct {
LinkAttrs
}
func (wg *Wireguard) Attrs() *LinkAttrs {
return &wg.LinkAttrs
}
func (wg *Wireguard) Type() string {
return "wireguard"
}
// GenericLink links represent types that are not currently understood
// by this netlink library.
type GenericLink struct {
@@ -376,9 +434,18 @@ const (
IPVLAN_MODE_MAX
)
type IPVlanFlag uint16
const (
IPVLAN_FLAG_BRIDGE IPVlanFlag = iota
IPVLAN_FLAG_PRIVATE
IPVLAN_FLAG_VEPA
)
type IPVlan struct {
LinkAttrs
Mode IPVlanMode
Flag IPVlanFlag
}
func (ipvlan *IPVlan) Attrs() *LinkAttrs {
@@ -389,6 +456,43 @@ func (ipvlan *IPVlan) Type() string {
return "ipvlan"
}
// VlanProtocol type
type VlanProtocol int
func (p VlanProtocol) String() string {
s, ok := VlanProtocolToString[p]
if !ok {
return fmt.Sprintf("VlanProtocol(%d)", p)
}
return s
}
// StringToVlanProtocol returns vlan protocol, or unknown is the s is invalid.
func StringToVlanProtocol(s string) VlanProtocol {
mode, ok := StringToVlanProtocolMap[s]
if !ok {
return VLAN_PROTOCOL_UNKNOWN
}
return mode
}
// VlanProtocol possible values
const (
VLAN_PROTOCOL_UNKNOWN VlanProtocol = 0
VLAN_PROTOCOL_8021Q VlanProtocol = 0x8100
VLAN_PROTOCOL_8021AD VlanProtocol = 0x88A8
)
var VlanProtocolToString = map[VlanProtocol]string{
VLAN_PROTOCOL_8021Q: "802.1q",
VLAN_PROTOCOL_8021AD: "802.1ad",
}
var StringToVlanProtocolMap = map[string]VlanProtocol{
"802.1q": VLAN_PROTOCOL_8021Q,
"802.1ad": VLAN_PROTOCOL_8021AD,
}
// BondMode type
type BondMode int
@@ -400,7 +504,7 @@ func (b BondMode) String() string {
return s
}
// StringToBondMode returns bond mode, or uknonw is the s is invalid.
// StringToBondMode returns bond mode, or unknown is the s is invalid.
func StringToBondMode(s string) BondMode {
mode, ok := StringToBondModeMap[s]
if !ok {
@@ -451,6 +555,27 @@ const (
BOND_ARP_VALIDATE_ALL
)
var bondArpValidateToString = map[BondArpValidate]string{
BOND_ARP_VALIDATE_NONE: "none",
BOND_ARP_VALIDATE_ACTIVE: "active",
BOND_ARP_VALIDATE_BACKUP: "backup",
BOND_ARP_VALIDATE_ALL: "none",
}
var StringToBondArpValidateMap = map[string]BondArpValidate{
"none": BOND_ARP_VALIDATE_NONE,
"active": BOND_ARP_VALIDATE_ACTIVE,
"backup": BOND_ARP_VALIDATE_BACKUP,
"all": BOND_ARP_VALIDATE_ALL,
}
func (b BondArpValidate) String() string {
s, ok := bondArpValidateToString[b]
if !ok {
return fmt.Sprintf("BondArpValidate(%d)", b)
}
return s
}
// BondPrimaryReselect type
type BondPrimaryReselect int
@@ -461,6 +586,25 @@ const (
BOND_PRIMARY_RESELECT_FAILURE
)
var bondPrimaryReselectToString = map[BondPrimaryReselect]string{
BOND_PRIMARY_RESELECT_ALWAYS: "always",
BOND_PRIMARY_RESELECT_BETTER: "better",
BOND_PRIMARY_RESELECT_FAILURE: "failure",
}
var StringToBondPrimaryReselectMap = map[string]BondPrimaryReselect{
"always": BOND_PRIMARY_RESELECT_ALWAYS,
"better": BOND_PRIMARY_RESELECT_BETTER,
"failure": BOND_PRIMARY_RESELECT_FAILURE,
}
func (b BondPrimaryReselect) String() string {
s, ok := bondPrimaryReselectToString[b]
if !ok {
return fmt.Sprintf("BondPrimaryReselect(%d)", b)
}
return s
}
// BondArpAllTargets type
type BondArpAllTargets int
@@ -470,6 +614,23 @@ const (
BOND_ARP_ALL_TARGETS_ALL
)
var bondArpAllTargetsToString = map[BondArpAllTargets]string{
BOND_ARP_ALL_TARGETS_ANY: "any",
BOND_ARP_ALL_TARGETS_ALL: "all",
}
var StringToBondArpAllTargetsMap = map[string]BondArpAllTargets{
"any": BOND_ARP_ALL_TARGETS_ANY,
"all": BOND_ARP_ALL_TARGETS_ALL,
}
func (b BondArpAllTargets) String() string {
s, ok := bondArpAllTargetsToString[b]
if !ok {
return fmt.Sprintf("BondArpAllTargets(%d)", b)
}
return s
}
// BondFailOverMac type
type BondFailOverMac int
@@ -480,6 +641,25 @@ const (
BOND_FAIL_OVER_MAC_FOLLOW
)
var bondFailOverMacToString = map[BondFailOverMac]string{
BOND_FAIL_OVER_MAC_NONE: "none",
BOND_FAIL_OVER_MAC_ACTIVE: "active",
BOND_FAIL_OVER_MAC_FOLLOW: "follow",
}
var StringToBondFailOverMacMap = map[string]BondFailOverMac{
"none": BOND_FAIL_OVER_MAC_NONE,
"active": BOND_FAIL_OVER_MAC_ACTIVE,
"follow": BOND_FAIL_OVER_MAC_FOLLOW,
}
func (b BondFailOverMac) String() string {
s, ok := bondFailOverMacToString[b]
if !ok {
return fmt.Sprintf("BondFailOverMac(%d)", b)
}
return s
}
// BondXmitHashPolicy type
type BondXmitHashPolicy int
@@ -491,7 +671,7 @@ func (b BondXmitHashPolicy) String() string {
return s
}
// StringToBondXmitHashPolicy returns bond lacp arte, or uknonw is the s is invalid.
// StringToBondXmitHashPolicy returns bond lacp arte, or unknown is the s is invalid.
func StringToBondXmitHashPolicy(s string) BondXmitHashPolicy {
lacp, ok := StringToBondXmitHashPolicyMap[s]
if !ok {
@@ -536,7 +716,7 @@ func (b BondLacpRate) String() string {
return s
}
// StringToBondLacpRate returns bond lacp arte, or uknonw is the s is invalid.
// StringToBondLacpRate returns bond lacp arte, or unknown is the s is invalid.
func StringToBondLacpRate(s string) BondLacpRate {
lacp, ok := StringToBondLacpRateMap[s]
if !ok {
@@ -571,6 +751,25 @@ const (
BOND_AD_SELECT_COUNT
)
var bondAdSelectToString = map[BondAdSelect]string{
BOND_AD_SELECT_STABLE: "stable",
BOND_AD_SELECT_BANDWIDTH: "bandwidth",
BOND_AD_SELECT_COUNT: "count",
}
var StringToBondAdSelectMap = map[string]BondAdSelect{
"stable": BOND_AD_SELECT_STABLE,
"bandwidth": BOND_AD_SELECT_BANDWIDTH,
"count": BOND_AD_SELECT_COUNT,
}
func (b BondAdSelect) String() string {
s, ok := bondAdSelectToString[b]
if !ok {
return fmt.Sprintf("BondAdSelect(%d)", b)
}
return s
}
// BondAdInfo represents ad info for bond
type BondAdInfo struct {
AggregatorId int
@@ -602,7 +801,7 @@ type Bond struct {
AllSlavesActive int
MinLinks int
LpInterval int
PackersPerSlave int
PacketsPerSlave int
LacpRate BondLacpRate
AdSelect BondAdSelect
// looking at iproute tool AdInfo can only be retrived. It can't be set.
@@ -635,7 +834,7 @@ func NewLinkBond(atr LinkAttrs) *Bond {
AllSlavesActive: -1,
MinLinks: -1,
LpInterval: -1,
PackersPerSlave: -1,
PacketsPerSlave: -1,
LacpRate: -1,
AdSelect: -1,
AdActorSysPrio: -1,
@@ -680,6 +879,97 @@ func (bond *Bond) Type() string {
return "bond"
}
// BondSlaveState represents the values of the IFLA_BOND_SLAVE_STATE bond slave
// attribute, which contains the state of the bond slave.
type BondSlaveState uint8
const (
//BondStateActive Link is active.
BondStateActive BondSlaveState = iota
//BondStateBackup Link is backup.
BondStateBackup
)
func (s BondSlaveState) String() string {
switch s {
case BondStateActive:
return "ACTIVE"
case BondStateBackup:
return "BACKUP"
default:
return strconv.Itoa(int(s))
}
}
// BondSlaveMiiStatus represents the values of the IFLA_BOND_SLAVE_MII_STATUS bond slave
// attribute, which contains the status of MII link monitoring
type BondSlaveMiiStatus uint8
const (
//BondLinkUp link is up and running.
BondLinkUp BondSlaveMiiStatus = iota
//BondLinkFail link has just gone down.
BondLinkFail
//BondLinkDown link has been down for too long time.
BondLinkDown
//BondLinkBack link is going back.
BondLinkBack
)
func (s BondSlaveMiiStatus) String() string {
switch s {
case BondLinkUp:
return "UP"
case BondLinkFail:
return "GOING_DOWN"
case BondLinkDown:
return "DOWN"
case BondLinkBack:
return "GOING_BACK"
default:
return strconv.Itoa(int(s))
}
}
type BondSlave struct {
State BondSlaveState
MiiStatus BondSlaveMiiStatus
LinkFailureCount uint32
PermHardwareAddr net.HardwareAddr
QueueId uint16
AggregatorId uint16
AdActorOperPortState uint8
AdPartnerOperPortState uint16
}
func (b *BondSlave) SlaveType() string {
return "bond"
}
// Geneve devices must specify RemoteIP and ID (VNI) on create
// https://github.com/torvalds/linux/blob/47ec5303d73ea344e84f46660fff693c57641386/drivers/net/geneve.c#L1209-L1223
type Geneve struct {
LinkAttrs
ID uint32 // vni
Remote net.IP
Ttl uint8
Tos uint8
Dport uint16
UdpCsum uint8
UdpZeroCsum6Tx uint8
UdpZeroCsum6Rx uint8
Link uint32
FlowBased bool
}
func (geneve *Geneve) Attrs() *LinkAttrs {
return &geneve.LinkAttrs
}
func (geneve *Geneve) Type() string {
return "geneve"
}
// Gretap devices must specify LocalIP and RemoteIP on create
type Gretap struct {
LinkAttrs
@@ -734,14 +1024,41 @@ func (iptun *Iptun) Type() string {
return "ipip"
}
type Sittun struct {
type Ip6tnl struct {
LinkAttrs
Link uint32
Local net.IP
Remote net.IP
Ttl uint8
Tos uint8
Flags uint32
Proto uint8
FlowInfo uint32
EncapLimit uint8
EncapType uint16
EncapFlags uint16
EncapSport uint16
EncapDport uint16
}
func (ip6tnl *Ip6tnl) Attrs() *LinkAttrs {
return &ip6tnl.LinkAttrs
}
func (ip6tnl *Ip6tnl) Type() string {
return "ip6tnl"
}
type Sittun struct {
LinkAttrs
Link uint32
Ttl uint8
Tos uint8
PMtuDisc uint8
Proto uint8
Local net.IP
Remote net.IP
EncapLimit uint8
EncapType uint16
EncapFlags uint16
EncapSport uint16
@@ -769,7 +1086,10 @@ func (vti *Vti) Attrs() *LinkAttrs {
return &vti.LinkAttrs
}
func (iptun *Vti) Type() string {
func (vti *Vti) Type() string {
if vti.Local.To4() == nil {
return "vti6"
}
return "vti"
}
@@ -831,11 +1151,120 @@ func (gtp *GTP) Type() string {
return "gtp"
}
// Virtual XFRM Interfaces
// Named "xfrmi" to prevent confusion with XFRM objects
type Xfrmi struct {
LinkAttrs
Ifid uint32
}
func (xfrm *Xfrmi) Attrs() *LinkAttrs {
return &xfrm.LinkAttrs
}
func (xfrm *Xfrmi) Type() string {
return "xfrm"
}
// IPoIB interface
type IPoIBMode uint16
func (m *IPoIBMode) String() string {
str, ok := iPoIBModeToString[*m]
if !ok {
return fmt.Sprintf("mode(%d)", *m)
}
return str
}
const (
IPOIB_MODE_DATAGRAM = iota
IPOIB_MODE_CONNECTED
)
var iPoIBModeToString = map[IPoIBMode]string{
IPOIB_MODE_DATAGRAM: "datagram",
IPOIB_MODE_CONNECTED: "connected",
}
var StringToIPoIBMode = map[string]IPoIBMode{
"datagram": IPOIB_MODE_DATAGRAM,
"connected": IPOIB_MODE_CONNECTED,
}
const (
CAN_STATE_ERROR_ACTIVE = iota
CAN_STATE_ERROR_WARNING
CAN_STATE_ERROR_PASSIVE
CAN_STATE_BUS_OFF
CAN_STATE_STOPPED
CAN_STATE_SLEEPING
)
type Can struct {
LinkAttrs
BitRate uint32
SamplePoint uint32
TimeQuanta uint32
PropagationSegment uint32
PhaseSegment1 uint32
PhaseSegment2 uint32
SyncJumpWidth uint32
BitRatePreScaler uint32
Name string
TimeSegment1Min uint32
TimeSegment1Max uint32
TimeSegment2Min uint32
TimeSegment2Max uint32
SyncJumpWidthMax uint32
BitRatePreScalerMin uint32
BitRatePreScalerMax uint32
BitRatePreScalerInc uint32
ClockFrequency uint32
State uint32
Mask uint32
Flags uint32
TxError uint16
RxError uint16
RestartMs uint32
}
func (can *Can) Attrs() *LinkAttrs {
return &can.LinkAttrs
}
func (can *Can) Type() string {
return "can"
}
type IPoIB struct {
LinkAttrs
Pkey uint16
Mode IPoIBMode
Umcast uint16
}
func (ipoib *IPoIB) Attrs() *LinkAttrs {
return &ipoib.LinkAttrs
}
func (ipoib *IPoIB) Type() string {
return "ipoib"
}
// iproute2 supported devices;
// vlan | veth | vcan | dummy | ifb | macvlan | macvtap |
// bridge | bond | ipoib | ip6tnl | ipip | sit | vxlan |
// gre | gretap | ip6gre | ip6gretap | vti | nlmon |
// bond_slave | ipvlan
// gre | gretap | ip6gre | ip6gretap | vti | vti6 | nlmon |
// bond_slave | ipvlan | xfrm
// LinkNotFoundError wraps the various not found errors when
// getting/reading links. This is intended for better error

File diff suppressed because it is too large Load Diff

View File

@@ -17,9 +17,16 @@ type Neigh struct {
LLIPAddr net.IP //Used in the case of NHRP
Vlan int
VNI int
MasterIndex int
}
// String returns $ip/$hwaddr $label
func (neigh *Neigh) String() string {
return fmt.Sprintf("%s %s", neigh.IP, neigh.HardwareAddr)
}
// NeighUpdate is sent when a neighbor changes - type is RTM_NEWNEIGH or RTM_DELNEIGH.
type NeighUpdate struct {
Type uint16
Neigh
}

View File

@@ -1,10 +1,13 @@
package netlink
import (
"fmt"
"net"
"syscall"
"unsafe"
"github.com/vishvananda/netlink/nl"
"github.com/vishvananda/netns"
"golang.org/x/sys/unix"
)
@@ -18,7 +21,10 @@ const (
NDA_PORT
NDA_VNI
NDA_IFINDEX
NDA_MAX = NDA_IFINDEX
NDA_MASTER
NDA_LINK_NETNSID
NDA_SRC_VNI
NDA_MAX = NDA_SRC_VNI
)
// Neighbor Cache Entry States.
@@ -43,6 +49,7 @@ const (
NTF_ROUTER = 0x80
)
// Ndmsg is for adding, removing or receiving information about a neighbor table entry
type Ndmsg struct {
Family uint8
Index uint32
@@ -170,45 +177,58 @@ func neighHandle(neigh *Neigh, req *nl.NetlinkRequest) error {
req.AddData(vniData)
}
if neigh.MasterIndex != 0 {
masterData := nl.NewRtAttr(NDA_MASTER, nl.Uint32Attr(uint32(neigh.MasterIndex)))
req.AddData(masterData)
}
_, err := req.Execute(unix.NETLINK_ROUTE, 0)
return err
}
// NeighList gets a list of IP-MAC mappings in the system (ARP table).
// NeighList returns a list of IP-MAC mappings in the system (ARP table).
// Equivalent to: `ip neighbor show`.
// The list can be filtered by link and ip family.
func NeighList(linkIndex, family int) ([]Neigh, error) {
return pkgHandle.NeighList(linkIndex, family)
}
// NeighProxyList gets a list of neighbor proxies in the system.
// NeighProxyList returns a list of neighbor proxies in the system.
// Equivalent to: `ip neighbor show proxy`.
// The list can be filtered by link and ip family.
func NeighProxyList(linkIndex, family int) ([]Neigh, error) {
return pkgHandle.NeighProxyList(linkIndex, family)
}
// NeighList gets a list of IP-MAC mappings in the system (ARP table).
// NeighList returns a list of IP-MAC mappings in the system (ARP table).
// Equivalent to: `ip neighbor show`.
// The list can be filtered by link and ip family.
func (h *Handle) NeighList(linkIndex, family int) ([]Neigh, error) {
return h.neighList(linkIndex, family, 0)
return h.NeighListExecute(Ndmsg{
Family: uint8(family),
Index: uint32(linkIndex),
})
}
// NeighProxyList gets a list of neighbor proxies in the system.
// NeighProxyList returns a list of neighbor proxies in the system.
// Equivalent to: `ip neighbor show proxy`.
// The list can be filtered by link, ip family.
func (h *Handle) NeighProxyList(linkIndex, family int) ([]Neigh, error) {
return h.neighList(linkIndex, family, NTF_PROXY)
}
func (h *Handle) neighList(linkIndex, family, flags int) ([]Neigh, error) {
req := h.newNetlinkRequest(unix.RTM_GETNEIGH, unix.NLM_F_DUMP)
msg := Ndmsg{
return h.NeighListExecute(Ndmsg{
Family: uint8(family),
Index: uint32(linkIndex),
Flags: uint8(flags),
}
Flags: NTF_PROXY,
})
}
// NeighListExecute returns a list of neighbour entries filtered by link, ip family, flag and state.
func NeighListExecute(msg Ndmsg) ([]Neigh, error) {
return pkgHandle.NeighListExecute(msg)
}
// NeighListExecute returns a list of neighbour entries filtered by link, ip family, flag and state.
func (h *Handle) NeighListExecute(msg Ndmsg) ([]Neigh, error) {
req := h.newNetlinkRequest(unix.RTM_GETNEIGH, unix.NLM_F_DUMP)
req.AddData(&msg)
msgs, err := req.Execute(unix.NETLINK_ROUTE, unix.RTM_NEWNEIGH)
@@ -219,10 +239,22 @@ func (h *Handle) neighList(linkIndex, family, flags int) ([]Neigh, error) {
var res []Neigh
for _, m := range msgs {
ndm := deserializeNdmsg(m)
if linkIndex != 0 && int(ndm.Index) != linkIndex {
if msg.Index != 0 && ndm.Index != msg.Index {
// Ignore messages from other interfaces
continue
}
if msg.Family != 0 && ndm.Family != msg.Family {
continue
}
if msg.State != 0 && ndm.State != msg.State {
continue
}
if msg.Type != 0 && ndm.Type != msg.Type {
continue
}
if msg.Flags != 0 && ndm.Flags != msg.Flags {
continue
}
neigh, err := NeighDeserialize(m)
if err != nil {
@@ -251,14 +283,6 @@ func NeighDeserialize(m []byte) (*Neigh, error) {
return nil, err
}
// This should be cached for perfomance
// once per table dump
link, err := LinkByIndex(neigh.LinkIndex)
if err != nil {
return nil, err
}
encapType := link.Attrs().EncapType
for _, attr := range attrs {
switch attr.Attr.Type {
case NDA_DST:
@@ -268,13 +292,16 @@ func NeighDeserialize(m []byte) (*Neigh, error) {
// #define RTA_LENGTH(len) (RTA_ALIGN(sizeof(struct rtattr)) + (len))
// #define RTA_PAYLOAD(rta) ((int)((rta)->rta_len) - RTA_LENGTH(0))
attrLen := attr.Attr.Len - unix.SizeofRtAttr
if attrLen == 4 && (encapType == "ipip" ||
encapType == "sit" ||
encapType == "gre") {
if attrLen == 4 {
neigh.LLIPAddr = net.IP(attr.Value)
} else if attrLen == 16 &&
encapType == "tunnel6" {
neigh.IP = net.IP(attr.Value)
} else if attrLen == 16 {
// Can be IPv6 or FireWire HWAddr
link, err := LinkByIndex(neigh.LinkIndex)
if err == nil && link.Attrs().EncapType == "tunnel6" {
neigh.IP = net.IP(attr.Value)
} else {
neigh.HardwareAddr = net.HardwareAddr(attr.Value)
}
} else {
neigh.HardwareAddr = net.HardwareAddr(attr.Value)
}
@@ -282,8 +309,126 @@ func NeighDeserialize(m []byte) (*Neigh, error) {
neigh.Vlan = int(native.Uint16(attr.Value[0:2]))
case NDA_VNI:
neigh.VNI = int(native.Uint32(attr.Value[0:4]))
case NDA_MASTER:
neigh.MasterIndex = int(native.Uint32(attr.Value[0:4]))
}
}
return &neigh, nil
}
// NeighSubscribe takes a chan down which notifications will be sent
// when neighbors are added or deleted. Close the 'done' chan to stop subscription.
func NeighSubscribe(ch chan<- NeighUpdate, done <-chan struct{}) error {
return neighSubscribeAt(netns.None(), netns.None(), ch, done, nil, false)
}
// NeighSubscribeAt works like NeighSubscribe plus it allows the caller
// to choose the network namespace in which to subscribe (ns).
func NeighSubscribeAt(ns netns.NsHandle, ch chan<- NeighUpdate, done <-chan struct{}) error {
return neighSubscribeAt(ns, netns.None(), ch, done, nil, false)
}
// NeighSubscribeOptions contains a set of options to use with
// NeighSubscribeWithOptions.
type NeighSubscribeOptions struct {
Namespace *netns.NsHandle
ErrorCallback func(error)
ListExisting bool
}
// NeighSubscribeWithOptions work like NeighSubscribe but enable to
// provide additional options to modify the behavior. Currently, the
// namespace can be provided as well as an error callback.
func NeighSubscribeWithOptions(ch chan<- NeighUpdate, done <-chan struct{}, options NeighSubscribeOptions) error {
if options.Namespace == nil {
none := netns.None()
options.Namespace = &none
}
return neighSubscribeAt(*options.Namespace, netns.None(), ch, done, options.ErrorCallback, options.ListExisting)
}
func neighSubscribeAt(newNs, curNs netns.NsHandle, ch chan<- NeighUpdate, done <-chan struct{}, cberr func(error), listExisting bool) error {
s, err := nl.SubscribeAt(newNs, curNs, unix.NETLINK_ROUTE, unix.RTNLGRP_NEIGH)
makeRequest := func(family int) error {
req := pkgHandle.newNetlinkRequest(unix.RTM_GETNEIGH,
unix.NLM_F_DUMP)
infmsg := nl.NewIfInfomsg(family)
req.AddData(infmsg)
if err := s.Send(req); err != nil {
return err
}
return nil
}
if err != nil {
return err
}
if done != nil {
go func() {
<-done
s.Close()
}()
}
if listExisting {
if err := makeRequest(unix.AF_UNSPEC); err != nil {
return err
}
// We have to wait for NLMSG_DONE before making AF_BRIDGE request
}
go func() {
defer close(ch)
for {
msgs, from, err := s.Receive()
if err != nil {
if cberr != nil {
cberr(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 {
if listExisting {
// This will be called after handling AF_UNSPEC
// list request, we have to wait for NLMSG_DONE
// before making another request
if err := makeRequest(unix.AF_BRIDGE); err != nil {
if cberr != nil {
cberr(err)
}
return
}
listExisting = false
}
continue
}
if m.Header.Type == unix.NLMSG_ERROR {
native := nl.NativeEndian()
error := int32(native.Uint32(m.Data[0:4]))
if error == 0 {
continue
}
if cberr != nil {
cberr(syscall.Errno(-error))
}
return
}
neigh, err := NeighDeserialize(m.Data)
if err != nil {
if cberr != nil {
cberr(err)
}
return
}
ch <- NeighUpdate{Type: m.Header.Type, Neigh: *neigh}
}
}
}()
return nil
}

View File

@@ -27,7 +27,8 @@ func ParseIPNet(s string) (*net.IPNet, error) {
if err != nil {
return nil, err
}
return &net.IPNet{IP: ip, Mask: ipNet.Mask}, nil
ipNet.IP = ip
return ipNet, nil
}
// NewIPNet generates an IPNet from an ip address using a netmask of 32 or 128.

View File

@@ -16,7 +16,7 @@ func LinkSetMTU(link Link, mtu int) error {
return ErrNotImplemented
}
func LinkSetMaster(link Link, master *Bridge) error {
func LinkSetMaster(link Link, master Link) error {
return ErrNotImplemented
}
@@ -48,10 +48,18 @@ func LinkSetVfVlan(link Link, vf, vlan int) error {
return ErrNotImplemented
}
func LinkSetVfVlanQos(link Link, vf, vlan, qos int) error {
return ErrNotImplemented
}
func LinkSetVfTxRate(link Link, vf, rate int) error {
return ErrNotImplemented
}
func LinkSetVfRate(link Link, vf, minRate, maxRate int) error {
return ErrNotImplemented
}
func LinkSetNoMaster(link Link) error {
return ErrNotImplemented
}
@@ -64,6 +72,10 @@ func LinkSetXdpFd(link Link, fd int) error {
return ErrNotImplemented
}
func LinkSetXdpFdWithFlags(link Link, fd, flags int) error {
return ErrNotImplemented
}
func LinkSetARPOff(link Link) error {
return ErrNotImplemented
}
@@ -152,6 +164,10 @@ func AddrAdd(link Link, addr *Addr) error {
return ErrNotImplemented
}
func AddrReplace(link Link, addr *Addr) error {
return ErrNotImplemented
}
func AddrDel(link Link, addr *Addr) error {
return ErrNotImplemented
}

141
vendor/github.com/vishvananda/netlink/netns_linux.go generated vendored Normal file
View File

@@ -0,0 +1,141 @@
package netlink
// Network namespace ID functions
//
// The kernel has a weird concept called the network namespace ID.
// This is different from the file reference in proc (and any bind-mounted
// namespaces, etc.)
//
// Instead, namespaces can be assigned a numeric ID at any time. Once set,
// the ID is fixed. The ID can either be set manually by the user, or
// automatically, triggered by certain kernel actions. The most common kernel
// action that triggers namespace ID creation is moving one end of a veth pair
// in to that namespace.
import (
"fmt"
"github.com/vishvananda/netlink/nl"
"golang.org/x/sys/unix"
)
// These can be replaced by the values from sys/unix when it is next released.
const (
_ = iota
NETNSA_NSID
NETNSA_PID
NETNSA_FD
)
// GetNetNsIdByPid looks up the network namespace ID for a given pid (really thread id).
// Returns -1 if the namespace does not have an ID set.
func (h *Handle) GetNetNsIdByPid(pid int) (int, error) {
return h.getNetNsId(NETNSA_PID, uint32(pid))
}
// GetNetNsIdByPid looks up the network namespace ID for a given pid (really thread id).
// Returns -1 if the namespace does not have an ID set.
func GetNetNsIdByPid(pid int) (int, error) {
return pkgHandle.GetNetNsIdByPid(pid)
}
// SetNetNSIdByPid sets the ID of the network namespace for a given pid (really thread id).
// The ID can only be set for namespaces without an ID already set.
func (h *Handle) SetNetNsIdByPid(pid, nsid int) error {
return h.setNetNsId(NETNSA_PID, uint32(pid), uint32(nsid))
}
// SetNetNSIdByPid sets the ID of the network namespace for a given pid (really thread id).
// The ID can only be set for namespaces without an ID already set.
func SetNetNsIdByPid(pid, nsid int) error {
return pkgHandle.SetNetNsIdByPid(pid, nsid)
}
// GetNetNsIdByFd looks up the network namespace ID for a given fd.
// fd must be an open file descriptor to a namespace file.
// Returns -1 if the namespace does not have an ID set.
func (h *Handle) GetNetNsIdByFd(fd int) (int, error) {
return h.getNetNsId(NETNSA_FD, uint32(fd))
}
// GetNetNsIdByFd looks up the network namespace ID for a given fd.
// fd must be an open file descriptor to a namespace file.
// Returns -1 if the namespace does not have an ID set.
func GetNetNsIdByFd(fd int) (int, error) {
return pkgHandle.GetNetNsIdByFd(fd)
}
// SetNetNSIdByFd sets the ID of the network namespace for a given fd.
// fd must be an open file descriptor to a namespace file.
// The ID can only be set for namespaces without an ID already set.
func (h *Handle) SetNetNsIdByFd(fd, nsid int) error {
return h.setNetNsId(NETNSA_FD, uint32(fd), uint32(nsid))
}
// SetNetNSIdByFd sets the ID of the network namespace for a given fd.
// fd must be an open file descriptor to a namespace file.
// The ID can only be set for namespaces without an ID already set.
func SetNetNsIdByFd(fd, nsid int) error {
return pkgHandle.SetNetNsIdByFd(fd, nsid)
}
// getNetNsId requests the netnsid for a given type-val pair
// type should be either NETNSA_PID or NETNSA_FD
func (h *Handle) getNetNsId(attrType int, val uint32) (int, error) {
req := h.newNetlinkRequest(unix.RTM_GETNSID, unix.NLM_F_REQUEST)
rtgen := nl.NewRtGenMsg()
req.AddData(rtgen)
b := make([]byte, 4, 4)
native.PutUint32(b, val)
attr := nl.NewRtAttr(attrType, b)
req.AddData(attr)
msgs, err := req.Execute(unix.NETLINK_ROUTE, unix.RTM_NEWNSID)
if err != nil {
return 0, err
}
for _, m := range msgs {
msg := nl.DeserializeRtGenMsg(m)
attrs, err := nl.ParseRouteAttr(m[msg.Len():])
if err != nil {
return 0, err
}
for _, attr := range attrs {
switch attr.Attr.Type {
case NETNSA_NSID:
return int(int32(native.Uint32(attr.Value))), nil
}
}
}
return 0, fmt.Errorf("unexpected empty result")
}
// setNetNsId sets the netnsid for a given type-val pair
// type should be either NETNSA_PID or NETNSA_FD
// The ID can only be set for namespaces without an ID already set
func (h *Handle) setNetNsId(attrType int, val uint32, newnsid uint32) error {
req := h.newNetlinkRequest(unix.RTM_NEWNSID, unix.NLM_F_REQUEST|unix.NLM_F_ACK)
rtgen := nl.NewRtGenMsg()
req.AddData(rtgen)
b := make([]byte, 4, 4)
native.PutUint32(b, val)
attr := nl.NewRtAttr(attrType, b)
req.AddData(attr)
b1 := make([]byte, 4, 4)
native.PutUint32(b1, newnsid)
attr1 := nl.NewRtAttr(NETNSA_NSID, b1)
req.AddData(attr1)
_, err := req.Execute(unix.NETLINK_ROUTE, unix.RTM_NEWNSID)
return err
}

View File

@@ -0,0 +1,19 @@
// +build !linux
package netlink
func GetNetNsIdByPid(pid int) (int, error) {
return 0, ErrNotImplemented
}
func SetNetNsIdByPid(pid, nsid int) error {
return ErrNotImplemented
}
func GetNetNsIdByFd(fd int) (int, error) {
return 0, ErrNotImplemented
}
func SetNetNsIdByFd(fd, nsid int) error {
return ErrNotImplemented
}

View File

@@ -54,24 +54,18 @@ func (msg *IfAddrmsg) Len() int {
// __u32 tstamp; /* updated timestamp, hundredths of seconds */
// };
const IFA_CACHEINFO = 6
const SizeofIfaCacheInfo = 0x10
type IfaCacheInfo struct {
IfaPrefered uint32
IfaValid uint32
Cstamp uint32
Tstamp uint32
unix.IfaCacheinfo
}
func (msg *IfaCacheInfo) Len() int {
return SizeofIfaCacheInfo
return unix.SizeofIfaCacheinfo
}
func DeserializeIfaCacheInfo(b []byte) *IfaCacheInfo {
return (*IfaCacheInfo)(unsafe.Pointer(&b[0:SizeofIfaCacheInfo][0]))
return (*IfaCacheInfo)(unsafe.Pointer(&b[0:unix.SizeofIfaCacheinfo][0]))
}
func (msg *IfaCacheInfo) Serialize() []byte {
return (*(*[SizeofIfaCacheInfo]byte)(unsafe.Pointer(msg)))[:]
return (*(*[unix.SizeofIfaCacheinfo]byte)(unsafe.Pointer(msg)))[:]
}

View File

@@ -11,8 +11,8 @@ const (
/* Bridge Flags */
const (
BRIDGE_FLAGS_MASTER = iota /* Bridge command to/from master */
BRIDGE_FLAGS_SELF /* Bridge command to/from lowerdev */
BRIDGE_FLAGS_MASTER = iota + 1 /* Bridge command to/from master */
BRIDGE_FLAGS_SELF /* Bridge command to/from lowerdev */
)
/* Bridge management nested attributes

View File

@@ -40,9 +40,10 @@ const (
NFNETLINK_V0 = 0
)
// #define NLA_F_NESTED (1 << 15)
const (
NLA_F_NESTED = (1 << 15)
NLA_F_NESTED uint16 = (1 << 15) // #define NLA_F_NESTED (1 << 15)
NLA_F_NET_BYTEORDER uint16 = (1 << 14) // #define NLA_F_NESTED (1 << 14)
NLA_TYPE_MASK = ^(NLA_F_NESTED | NLA_F_NET_BYTEORDER)
)
// enum ctattr_type {
@@ -76,12 +77,17 @@ const (
// __CTA_MAX
// };
const (
CTA_TUPLE_ORIG = 1
CTA_TUPLE_REPLY = 2
CTA_STATUS = 3
CTA_TIMEOUT = 7
CTA_MARK = 8
CTA_PROTOINFO = 4
CTA_TUPLE_ORIG = 1
CTA_TUPLE_REPLY = 2
CTA_STATUS = 3
CTA_PROTOINFO = 4
CTA_TIMEOUT = 7
CTA_MARK = 8
CTA_COUNTERS_ORIG = 9
CTA_COUNTERS_REPLY = 10
CTA_USE = 11
CTA_ID = 12
CTA_TIMESTAMP = 20
)
// enum ctattr_tuple {
@@ -163,6 +169,29 @@ const (
CTA_PROTOINFO_TCP_FLAGS_REPLY = 5
)
// enum ctattr_counters {
// CTA_COUNTERS_UNSPEC,
// CTA_COUNTERS_PACKETS, /* 64bit counters */
// CTA_COUNTERS_BYTES, /* 64bit counters */
// CTA_COUNTERS32_PACKETS, /* old 32bit counters, unused */
// CTA_COUNTERS32_BYTES, /* old 32bit counters, unused */
// CTA_COUNTERS_PAD,
// __CTA_COUNTERS_M
// };
// #define CTA_COUNTERS_MAX (__CTA_COUNTERS_MAX - 1)
const (
CTA_COUNTERS_PACKETS = 1
CTA_COUNTERS_BYTES = 2
)
// enum CTA TIMESTAMP TLVs
// CTA_TIMESTAMP_START /* 64bit value */
// CTA_TIMESTAMP_STOP /* 64bit value */
const (
CTA_TIMESTAMP_START = 1
CTA_TIMESTAMP_STOP = 2
)
// /* General form of address family dependent message.
// */
// struct nfgenmsg {

View File

@@ -0,0 +1,63 @@
package nl
// All the following constants are coming from:
// https://github.com/torvalds/linux/blob/master/include/uapi/linux/devlink.h
const (
GENL_DEVLINK_VERSION = 1
GENL_DEVLINK_NAME = "devlink"
)
const (
DEVLINK_CMD_GET = 1
DEVLINK_CMD_PORT_GET = 5
DEVLINK_CMD_ESWITCH_GET = 29
DEVLINK_CMD_ESWITCH_SET = 30
)
const (
DEVLINK_ATTR_BUS_NAME = 1
DEVLINK_ATTR_DEV_NAME = 2
DEVLINK_ATTR_PORT_INDEX = 3
DEVLINK_ATTR_PORT_TYPE = 4
DEVLINK_ATTR_PORT_NETDEV_IFINDEX = 6
DEVLINK_ATTR_PORT_NETDEV_NAME = 7
DEVLINK_ATTR_PORT_IBDEV_NAME = 8
DEVLINK_ATTR_ESWITCH_MODE = 25
DEVLINK_ATTR_ESWITCH_INLINE_MODE = 26
DEVLINK_ATTR_ESWITCH_ENCAP_MODE = 62
DEVLINK_ATTR_PORT_FLAVOUR = 77
)
const (
DEVLINK_ESWITCH_MODE_LEGACY = 0
DEVLINK_ESWITCH_MODE_SWITCHDEV = 1
)
const (
DEVLINK_ESWITCH_INLINE_MODE_NONE = 0
DEVLINK_ESWITCH_INLINE_MODE_LINK = 1
DEVLINK_ESWITCH_INLINE_MODE_NETWORK = 2
DEVLINK_ESWITCH_INLINE_MODE_TRANSPORT = 3
)
const (
DEVLINK_ESWITCH_ENCAP_MODE_NONE = 0
DEVLINK_ESWITCH_ENCAP_MODE_BASIC = 1
)
const (
DEVLINK_PORT_FLAVOUR_PHYSICAL = 0
DEVLINK_PORT_FLAVOUR_CPU = 1
DEVLINK_PORT_FLAVOUR_DSA = 2
DEVLINK_PORT_FLAVOUR_PCI_PF = 3
DEVLINK_PORT_FLAVOUR_PCI_VF = 4
DEVLINK_PORT_FLAVOUR_VIRTUAL = 5
)
const (
DEVLINK_PORT_TYPE_NOTSET = 0
DEVLINK_PORT_TYPE_AUTO = 1
DEVLINK_PORT_TYPE_ETH = 2
DEVLINK_PORT_TYPE_IB = 3
)

222
vendor/github.com/vishvananda/netlink/nl/ipset_linux.go generated vendored Normal file
View File

@@ -0,0 +1,222 @@
package nl
import (
"strconv"
"golang.org/x/sys/unix"
)
const (
/* The protocol version */
IPSET_PROTOCOL = 6
/* The max length of strings including NUL: set and type identifiers */
IPSET_MAXNAMELEN = 32
/* The maximum permissible comment length we will accept over netlink */
IPSET_MAX_COMMENT_SIZE = 255
)
const (
_ = iota
IPSET_CMD_PROTOCOL /* 1: Return protocol version */
IPSET_CMD_CREATE /* 2: Create a new (empty) set */
IPSET_CMD_DESTROY /* 3: Destroy a (empty) set */
IPSET_CMD_FLUSH /* 4: Remove all elements from a set */
IPSET_CMD_RENAME /* 5: Rename a set */
IPSET_CMD_SWAP /* 6: Swap two sets */
IPSET_CMD_LIST /* 7: List sets */
IPSET_CMD_SAVE /* 8: Save sets */
IPSET_CMD_ADD /* 9: Add an element to a set */
IPSET_CMD_DEL /* 10: Delete an element from a set */
IPSET_CMD_TEST /* 11: Test an element in a set */
IPSET_CMD_HEADER /* 12: Get set header data only */
IPSET_CMD_TYPE /* 13: Get set type */
)
/* Attributes at command level */
const (
_ = iota
IPSET_ATTR_PROTOCOL /* 1: Protocol version */
IPSET_ATTR_SETNAME /* 2: Name of the set */
IPSET_ATTR_TYPENAME /* 3: Typename */
IPSET_ATTR_REVISION /* 4: Settype revision */
IPSET_ATTR_FAMILY /* 5: Settype family */
IPSET_ATTR_FLAGS /* 6: Flags at command level */
IPSET_ATTR_DATA /* 7: Nested attributes */
IPSET_ATTR_ADT /* 8: Multiple data containers */
IPSET_ATTR_LINENO /* 9: Restore lineno */
IPSET_ATTR_PROTOCOL_MIN /* 10: Minimal supported version number */
IPSET_ATTR_SETNAME2 = IPSET_ATTR_TYPENAME /* Setname at rename/swap */
IPSET_ATTR_REVISION_MIN = IPSET_ATTR_PROTOCOL_MIN /* type rev min */
)
/* CADT specific attributes */
const (
IPSET_ATTR_IP = 1
IPSET_ATTR_IP_FROM = 1
IPSET_ATTR_IP_TO = 2
IPSET_ATTR_CIDR = 3
IPSET_ATTR_PORT = 4
IPSET_ATTR_PORT_FROM = 4
IPSET_ATTR_PORT_TO = 5
IPSET_ATTR_TIMEOUT = 6
IPSET_ATTR_PROTO = 7
IPSET_ATTR_CADT_FLAGS = 8
IPSET_ATTR_CADT_LINENO = IPSET_ATTR_LINENO /* 9 */
IPSET_ATTR_MARK = 10
IPSET_ATTR_MARKMASK = 11
/* Reserve empty slots */
IPSET_ATTR_CADT_MAX = 16
/* Create-only specific attributes */
IPSET_ATTR_GC = 3 + iota
IPSET_ATTR_HASHSIZE
IPSET_ATTR_MAXELEM
IPSET_ATTR_NETMASK
IPSET_ATTR_PROBES
IPSET_ATTR_RESIZE
IPSET_ATTR_SIZE
/* Kernel-only */
IPSET_ATTR_ELEMENTS
IPSET_ATTR_REFERENCES
IPSET_ATTR_MEMSIZE
SET_ATTR_CREATE_MAX
)
/* ADT specific attributes */
const (
IPSET_ATTR_ETHER = IPSET_ATTR_CADT_MAX + iota + 1
IPSET_ATTR_NAME
IPSET_ATTR_NAMEREF
IPSET_ATTR_IP2
IPSET_ATTR_CIDR2
IPSET_ATTR_IP2_TO
IPSET_ATTR_IFACE
IPSET_ATTR_BYTES
IPSET_ATTR_PACKETS
IPSET_ATTR_COMMENT
IPSET_ATTR_SKBMARK
IPSET_ATTR_SKBPRIO
IPSET_ATTR_SKBQUEUE
)
/* Flags at CADT attribute level, upper half of cmdattrs */
const (
IPSET_FLAG_BIT_BEFORE = 0
IPSET_FLAG_BEFORE = (1 << IPSET_FLAG_BIT_BEFORE)
IPSET_FLAG_BIT_PHYSDEV = 1
IPSET_FLAG_PHYSDEV = (1 << IPSET_FLAG_BIT_PHYSDEV)
IPSET_FLAG_BIT_NOMATCH = 2
IPSET_FLAG_NOMATCH = (1 << IPSET_FLAG_BIT_NOMATCH)
IPSET_FLAG_BIT_WITH_COUNTERS = 3
IPSET_FLAG_WITH_COUNTERS = (1 << IPSET_FLAG_BIT_WITH_COUNTERS)
IPSET_FLAG_BIT_WITH_COMMENT = 4
IPSET_FLAG_WITH_COMMENT = (1 << IPSET_FLAG_BIT_WITH_COMMENT)
IPSET_FLAG_BIT_WITH_FORCEADD = 5
IPSET_FLAG_WITH_FORCEADD = (1 << IPSET_FLAG_BIT_WITH_FORCEADD)
IPSET_FLAG_BIT_WITH_SKBINFO = 6
IPSET_FLAG_WITH_SKBINFO = (1 << IPSET_FLAG_BIT_WITH_SKBINFO)
IPSET_FLAG_CADT_MAX = 15
)
const (
IPSET_ERR_PRIVATE = 4096 + iota
IPSET_ERR_PROTOCOL
IPSET_ERR_FIND_TYPE
IPSET_ERR_MAX_SETS
IPSET_ERR_BUSY
IPSET_ERR_EXIST_SETNAME2
IPSET_ERR_TYPE_MISMATCH
IPSET_ERR_EXIST
IPSET_ERR_INVALID_CIDR
IPSET_ERR_INVALID_NETMASK
IPSET_ERR_INVALID_FAMILY
IPSET_ERR_TIMEOUT
IPSET_ERR_REFERENCED
IPSET_ERR_IPADDR_IPV4
IPSET_ERR_IPADDR_IPV6
IPSET_ERR_COUNTER
IPSET_ERR_COMMENT
IPSET_ERR_INVALID_MARKMASK
IPSET_ERR_SKBINFO
/* Type specific error codes */
IPSET_ERR_TYPE_SPECIFIC = 4352
)
type IPSetError uintptr
func (e IPSetError) Error() string {
switch int(e) {
case IPSET_ERR_PRIVATE:
return "private"
case IPSET_ERR_PROTOCOL:
return "invalid protocol"
case IPSET_ERR_FIND_TYPE:
return "invalid type"
case IPSET_ERR_MAX_SETS:
return "max sets reached"
case IPSET_ERR_BUSY:
return "busy"
case IPSET_ERR_EXIST_SETNAME2:
return "exist_setname2"
case IPSET_ERR_TYPE_MISMATCH:
return "type mismatch"
case IPSET_ERR_EXIST:
return "exist"
case IPSET_ERR_INVALID_CIDR:
return "invalid cidr"
case IPSET_ERR_INVALID_NETMASK:
return "invalid netmask"
case IPSET_ERR_INVALID_FAMILY:
return "invalid family"
case IPSET_ERR_TIMEOUT:
return "timeout"
case IPSET_ERR_REFERENCED:
return "referenced"
case IPSET_ERR_IPADDR_IPV4:
return "invalid ipv4 address"
case IPSET_ERR_IPADDR_IPV6:
return "invalid ipv6 address"
case IPSET_ERR_COUNTER:
return "invalid counter"
case IPSET_ERR_COMMENT:
return "invalid comment"
case IPSET_ERR_INVALID_MARKMASK:
return "invalid markmask"
case IPSET_ERR_SKBINFO:
return "skbinfo"
default:
return "errno " + strconv.Itoa(int(e))
}
}
func GetIpsetFlags(cmd int) int {
switch cmd {
case IPSET_CMD_CREATE:
return unix.NLM_F_REQUEST | unix.NLM_F_ACK | unix.NLM_F_CREATE
case IPSET_CMD_DESTROY,
IPSET_CMD_FLUSH,
IPSET_CMD_RENAME,
IPSET_CMD_SWAP,
IPSET_CMD_TEST:
return unix.NLM_F_REQUEST | unix.NLM_F_ACK
case IPSET_CMD_LIST,
IPSET_CMD_SAVE:
return unix.NLM_F_REQUEST | unix.NLM_F_ACK | unix.NLM_F_ROOT | unix.NLM_F_MATCH | unix.NLM_F_DUMP
case IPSET_CMD_ADD,
IPSET_CMD_DEL:
return unix.NLM_F_REQUEST | unix.NLM_F_ACK
case IPSET_CMD_HEADER,
IPSET_CMD_TYPE,
IPSET_CMD_PROTOCOL:
return unix.NLM_F_REQUEST
default:
return 0
}
}

View File

@@ -1,6 +1,8 @@
package nl
import (
"bytes"
"encoding/binary"
"unsafe"
)
@@ -13,7 +15,9 @@ const (
IFLA_INFO_KIND
IFLA_INFO_DATA
IFLA_INFO_XSTATS
IFLA_INFO_MAX = IFLA_INFO_XSTATS
IFLA_INFO_SLAVE_KIND
IFLA_INFO_SLAVE_DATA
IFLA_INFO_MAX = IFLA_INFO_SLAVE_DATA
)
const (
@@ -87,7 +91,8 @@ const (
const (
IFLA_IPVLAN_UNSPEC = iota
IFLA_IPVLAN_MODE
IFLA_IPVLAN_MAX = IFLA_IPVLAN_MODE
IFLA_IPVLAN_FLAG
IFLA_IPVLAN_MAX = IFLA_IPVLAN_FLAG
)
const (
@@ -164,6 +169,24 @@ const (
IFLA_BOND_SLAVE_PERM_HWADDR
IFLA_BOND_SLAVE_QUEUE_ID
IFLA_BOND_SLAVE_AD_AGGREGATOR_ID
IFLA_BOND_SLAVE_AD_ACTOR_OPER_PORT_STATE
IFLA_BOND_SLAVE_AD_PARTNER_OPER_PORT_STATE
)
const (
IFLA_GENEVE_UNSPEC = iota
IFLA_GENEVE_ID // vni
IFLA_GENEVE_REMOTE
IFLA_GENEVE_TTL
IFLA_GENEVE_TOS
IFLA_GENEVE_PORT // destination port
IFLA_GENEVE_COLLECT_METADATA
IFLA_GENEVE_REMOTE6
IFLA_GENEVE_UDP_CSUM
IFLA_GENEVE_UDP_ZERO_CSUM6_TX
IFLA_GENEVE_UDP_ZERO_CSUM6_RX
IFLA_GENEVE_LABEL
IFLA_GENEVE_MAX = IFLA_GENEVE_LABEL
)
const (
@@ -217,9 +240,11 @@ const (
IFLA_VF_RSS_QUERY_EN /* RSS Redirection Table and Hash Key query
* on/off switch
*/
IFLA_VF_STATS /* network device statistics */
IFLA_VF_TRUST /* Trust state of VF */
IFLA_VF_MAX = IFLA_VF_TRUST
IFLA_VF_STATS /* network device statistics */
IFLA_VF_TRUST /* Trust state of VF */
IFLA_VF_IB_NODE_GUID /* VF Infiniband node GUID */
IFLA_VF_IB_PORT_GUID /* VF Infiniband port GUID */
IFLA_VF_MAX = IFLA_VF_IB_PORT_GUID
)
const (
@@ -236,7 +261,9 @@ const (
IFLA_VF_STATS_TX_BYTES
IFLA_VF_STATS_BROADCAST
IFLA_VF_STATS_MULTICAST
IFLA_VF_STATS_MAX = IFLA_VF_STATS_MULTICAST
IFLA_VF_STATS_RX_DROPPED
IFLA_VF_STATS_TX_DROPPED
IFLA_VF_STATS_MAX = IFLA_VF_STATS_TX_DROPPED
)
const (
@@ -248,6 +275,7 @@ const (
SizeofVfLinkState = 0x08
SizeofVfRssQueryEn = 0x08
SizeofVfTrust = 0x08
SizeofVfGUID = 0x10
)
// struct ifla_vf_mac {
@@ -318,6 +346,59 @@ func (msg *VfTxRate) Serialize() []byte {
return (*(*[SizeofVfTxRate]byte)(unsafe.Pointer(msg)))[:]
}
//struct ifla_vf_stats {
// __u64 rx_packets;
// __u64 tx_packets;
// __u64 rx_bytes;
// __u64 tx_bytes;
// __u64 broadcast;
// __u64 multicast;
//};
type VfStats struct {
RxPackets uint64
TxPackets uint64
RxBytes uint64
TxBytes uint64
Multicast uint64
Broadcast uint64
RxDropped uint64
TxDropped uint64
}
func DeserializeVfStats(b []byte) VfStats {
var vfstat VfStats
stats, err := ParseRouteAttr(b)
if err != nil {
return vfstat
}
var valueVar uint64
for _, stat := range stats {
if err := binary.Read(bytes.NewBuffer(stat.Value), NativeEndian(), &valueVar); err != nil {
break
}
switch stat.Attr.Type {
case IFLA_VF_STATS_RX_PACKETS:
vfstat.RxPackets = valueVar
case IFLA_VF_STATS_TX_PACKETS:
vfstat.TxPackets = valueVar
case IFLA_VF_STATS_RX_BYTES:
vfstat.RxBytes = valueVar
case IFLA_VF_STATS_TX_BYTES:
vfstat.TxBytes = valueVar
case IFLA_VF_STATS_MULTICAST:
vfstat.Multicast = valueVar
case IFLA_VF_STATS_BROADCAST:
vfstat.Broadcast = valueVar
case IFLA_VF_STATS_RX_DROPPED:
vfstat.RxDropped = valueVar
case IFLA_VF_STATS_TX_DROPPED:
vfstat.TxDropped = valueVar
}
}
return vfstat
}
// struct ifla_vf_rate {
// __u32 vf;
// __u32 min_tx_rate; /* Min Bandwidth in Mbps */
@@ -430,6 +511,30 @@ func (msg *VfTrust) Serialize() []byte {
return (*(*[SizeofVfTrust]byte)(unsafe.Pointer(msg)))[:]
}
// struct ifla_vf_guid {
// __u32 vf;
// __u32 rsvd;
// __u64 guid;
// };
type VfGUID struct {
Vf uint32
Rsvd uint32
GUID uint64
}
func (msg *VfGUID) Len() int {
return SizeofVfGUID
}
func DeserializeVfGUID(b []byte) *VfGUID {
return (*VfGUID)(unsafe.Pointer(&b[0:SizeofVfGUID][0]))
}
func (msg *VfGUID) Serialize() []byte {
return (*(*[SizeofVfGUID]byte)(unsafe.Pointer(msg)))[:]
}
const (
XDP_FLAGS_UPDATE_IF_NOEXIST = 1 << iota
XDP_FLAGS_SKB_MODE
@@ -446,6 +551,14 @@ const (
IFLA_XDP_MAX = IFLA_XDP_PROG_ID
)
// XDP program attach mode (used as dump value for IFLA_XDP_ATTACHED)
const (
XDP_ATTACHED_NONE = iota
XDP_ATTACHED_DRV
XDP_ATTACHED_SKB
XDP_ATTACHED_HW
)
const (
IFLA_IPTUN_UNSPEC = iota
IFLA_IPTUN_LINK
@@ -546,3 +659,53 @@ const (
GTP_ROLE_GGSN = iota
GTP_ROLE_SGSN
)
const (
IFLA_XFRM_UNSPEC = iota
IFLA_XFRM_LINK
IFLA_XFRM_IF_ID
IFLA_XFRM_MAX = iota - 1
)
const (
IFLA_TUN_UNSPEC = iota
IFLA_TUN_OWNER
IFLA_TUN_GROUP
IFLA_TUN_TYPE
IFLA_TUN_PI
IFLA_TUN_VNET_HDR
IFLA_TUN_PERSIST
IFLA_TUN_MULTI_QUEUE
IFLA_TUN_NUM_QUEUES
IFLA_TUN_NUM_DISABLED_QUEUES
IFLA_TUN_MAX = IFLA_TUN_NUM_DISABLED_QUEUES
)
const (
IFLA_IPOIB_UNSPEC = iota
IFLA_IPOIB_PKEY
IFLA_IPOIB_MODE
IFLA_IPOIB_UMCAST
IFLA_IPOIB_MAX = IFLA_IPOIB_UMCAST
)
const (
IFLA_CAN_UNSPEC = iota
IFLA_CAN_BITTIMING
IFLA_CAN_BITTIMING_CONST
IFLA_CAN_CLOCK
IFLA_CAN_STATE
IFLA_CAN_CTRLMODE
IFLA_CAN_RESTART_MS
IFLA_CAN_RESTART
IFLA_CAN_BERR_COUNTER
IFLA_CAN_DATA_BITTIMING
IFLA_CAN_DATA_BITTIMING_CONST
IFLA_CAN_TERMINATION
IFLA_CAN_TERMINATION_CONST
IFLA_CAN_BITRATE_CONST
IFLA_CAN_DATA_BITRATE_CONST
IFLA_CAN_BITRATE_MAX
IFLA_CAN_MAX = IFLA_CAN_BITRATE_MAX
)

View File

@@ -21,7 +21,13 @@ const (
FAMILY_ALL = unix.AF_UNSPEC
FAMILY_V4 = unix.AF_INET
FAMILY_V6 = unix.AF_INET6
FAMILY_MPLS = AF_MPLS
FAMILY_MPLS = unix.AF_MPLS
// Arbitrary set value (greater than default 4k) to allow receiving
// from kernel more verbose messages e.g. for statistics,
// tc rules or filters, or other more memory requiring data.
RECEIVE_BUFFER_SIZE = 65536
// Kernel netlink pid
PidKernel uint32 = 0
)
// SupportedNlFamilies contains the list of netlink families this netlink package supports
@@ -29,6 +35,9 @@ var SupportedNlFamilies = []int{unix.NETLINK_ROUTE, unix.NETLINK_XFRM, unix.NETL
var nextSeqNr uint32
// Default netlink socket timeout, 60s
var SocketTimeoutTv = unix.Timeval{Sec: 60, Usec: 0}
// GetIPFamily returns the family type of a net.IP.
func GetIPFamily(ip net.IP) int {
if len(ip) <= net.IPv4len {
@@ -42,7 +51,7 @@ func GetIPFamily(ip net.IP) int {
var nativeEndian binary.ByteOrder
// Get native endianness for the system
// NativeEndian gets native endianness for the system
func NativeEndian() binary.ByteOrder {
if nativeEndian == nil {
var x uint32 = 0x01020304
@@ -253,6 +262,29 @@ func NewIfInfomsgChild(parent *RtAttr, family int) *IfInfomsg {
return msg
}
type Uint32Attribute struct {
Type uint16
Value uint32
}
func (a *Uint32Attribute) Serialize() []byte {
native := NativeEndian()
buf := make([]byte, rtaAlignOf(8))
native.PutUint16(buf[0:2], 8)
native.PutUint16(buf[2:4], a.Type)
if a.Type&NLA_F_NET_BYTEORDER != 0 {
binary.BigEndian.PutUint32(buf[4:], a.Value)
} else {
native.PutUint32(buf[4:], a.Value)
}
return buf
}
func (a *Uint32Attribute) Len() int {
return 8
}
// Extend RtAttr to handle data and children
type RtAttr struct {
unix.RtAttr
@@ -271,15 +303,22 @@ func NewRtAttr(attrType int, data []byte) *RtAttr {
}
}
// Create a new RtAttr obj anc add it as a child of an existing object
// NewRtAttrChild adds an RtAttr as a child to the parent and returns the new attribute
//
// Deprecated: Use AddRtAttr() on the parent object
func NewRtAttrChild(parent *RtAttr, attrType int, data []byte) *RtAttr {
return parent.AddRtAttr(attrType, data)
}
// AddRtAttr adds an RtAttr as a child and returns the new attribute
func (a *RtAttr) AddRtAttr(attrType int, data []byte) *RtAttr {
attr := NewRtAttr(attrType, data)
parent.children = append(parent.children, attr)
a.children = append(a.children, attr)
return attr
}
// AddChild adds an existing RtAttr as a child.
func (a *RtAttr) AddChild(attr *RtAttr) {
// AddChild adds an existing NetlinkRequestData as a child.
func (a *RtAttr) AddChild(attr NetlinkRequestData) {
a.children = append(a.children, attr)
}
@@ -360,16 +399,12 @@ func (req *NetlinkRequest) Serialize() []byte {
}
func (req *NetlinkRequest) AddData(data NetlinkRequestData) {
if data != nil {
req.Data = append(req.Data, data)
}
req.Data = append(req.Data, data)
}
// AddRawData adds raw bytes to the end of the NetlinkRequest object during serialization
func (req *NetlinkRequest) AddRawData(data []byte) {
if data != nil {
req.RawData = append(req.RawData, data...)
}
req.RawData = append(req.RawData, data...)
}
// Execute the request against a the given sockType.
@@ -394,6 +429,14 @@ func (req *NetlinkRequest) Execute(sockType int, resType uint16) ([][]byte, erro
if err != nil {
return nil, err
}
if err := s.SetSendTimeout(&SocketTimeoutTv); err != nil {
return nil, err
}
if err := s.SetReceiveTimeout(&SocketTimeoutTv); err != nil {
return nil, err
}
defer s.Close()
} else {
s.Lock()
@@ -413,10 +456,13 @@ func (req *NetlinkRequest) Execute(sockType int, resType uint16) ([][]byte, erro
done:
for {
msgs, err := s.Receive()
msgs, from, err := s.Receive()
if err != nil {
return nil, err
}
if from.Pid != PidKernel {
return nil, fmt.Errorf("Wrong sender portid %d, expected %d", from.Pid, PidKernel)
}
for _, m := range msgs {
if m.Header.Seq != req.Seq {
if sharedSocket {
@@ -425,12 +471,9 @@ done:
return nil, fmt.Errorf("Wrong Seq nr %d, expected %d", m.Header.Seq, req.Seq)
}
if m.Header.Pid != pid {
return nil, fmt.Errorf("Wrong pid %d, expected %d", m.Header.Pid, pid)
continue
}
if m.Header.Type == unix.NLMSG_DONE {
break done
}
if m.Header.Type == unix.NLMSG_ERROR {
if m.Header.Type == unix.NLMSG_DONE || m.Header.Type == unix.NLMSG_ERROR {
native := NativeEndian()
error := int32(native.Uint32(m.Data[0:4]))
if error == 0 {
@@ -610,21 +653,31 @@ func (s *NetlinkSocket) Send(request *NetlinkRequest) error {
return nil
}
func (s *NetlinkSocket) Receive() ([]syscall.NetlinkMessage, error) {
func (s *NetlinkSocket) Receive() ([]syscall.NetlinkMessage, *unix.SockaddrNetlink, error) {
fd := int(atomic.LoadInt32(&s.fd))
if fd < 0 {
return nil, fmt.Errorf("Receive called on a closed socket")
return nil, nil, fmt.Errorf("Receive called on a closed socket")
}
rb := make([]byte, unix.Getpagesize())
nr, _, err := unix.Recvfrom(fd, rb, 0)
var fromAddr *unix.SockaddrNetlink
var rb [RECEIVE_BUFFER_SIZE]byte
nr, from, err := unix.Recvfrom(fd, rb[:], 0)
if err != nil {
return nil, err
return nil, nil, err
}
fromAddr, ok := from.(*unix.SockaddrNetlink)
if !ok {
return nil, nil, fmt.Errorf("Error converting to netlink sockaddr")
}
if nr < unix.NLMSG_HDRLEN {
return nil, fmt.Errorf("Got short response from netlink")
return nil, nil, fmt.Errorf("Got short response from netlink")
}
rb = rb[:nr]
return syscall.ParseNetlinkMessage(rb)
rb2 := make([]byte, nr)
copy(rb2, rb[:nr])
nl, err := syscall.ParseNetlinkMessage(rb2)
if err != nil {
return nil, nil, err
}
return nl, fromAddr, nil
}
// SetSendTimeout allows to set a send timeout on the socket

View File

@@ -0,0 +1,79 @@
package nl
import (
"encoding/binary"
"fmt"
"log"
)
type Attribute struct {
Type uint16
Value []byte
}
func ParseAttributes(data []byte) <-chan Attribute {
native := NativeEndian()
result := make(chan Attribute)
go func() {
i := 0
for i+4 < len(data) {
length := int(native.Uint16(data[i : i+2]))
attrType := native.Uint16(data[i+2 : i+4])
if length < 4 {
log.Printf("attribute 0x%02x has invalid length of %d bytes", attrType, length)
break
}
if len(data) < i+length {
log.Printf("attribute 0x%02x of length %d is truncated, only %d bytes remaining", attrType, length, len(data)-i)
break
}
result <- Attribute{
Type: attrType,
Value: data[i+4 : i+length],
}
i += rtaAlignOf(length)
}
close(result)
}()
return result
}
func PrintAttributes(data []byte) {
printAttributes(data, 0)
}
func printAttributes(data []byte, level int) {
for attr := range ParseAttributes(data) {
for i := 0; i < level; i++ {
print("> ")
}
nested := attr.Type&NLA_F_NESTED != 0
fmt.Printf("type=%d nested=%v len=%v %v\n", attr.Type&NLA_TYPE_MASK, nested, len(attr.Value), attr.Value)
if nested {
printAttributes(attr.Value, level+1)
}
}
}
// Uint32 returns the uint32 value respecting the NET_BYTEORDER flag
func (attr *Attribute) Uint32() uint32 {
if attr.Type&NLA_F_NET_BYTEORDER != 0 {
return binary.BigEndian.Uint32(attr.Value)
} else {
return NativeEndian().Uint32(attr.Value)
}
}
// Uint64 returns the uint64 value respecting the NET_BYTEORDER flag
func (attr *Attribute) Uint64() uint64 {
if attr.Type&NLA_F_NET_BYTEORDER != 0 {
return binary.BigEndian.Uint64(attr.Value)
} else {
return NativeEndian().Uint64(attr.Value)
}
}

View File

@@ -0,0 +1,35 @@
package nl
const (
RDMA_NL_GET_CLIENT_SHIFT = 10
)
const (
RDMA_NL_NLDEV = 5
)
const (
RDMA_NLDEV_CMD_GET = 1
RDMA_NLDEV_CMD_SET = 2
RDMA_NLDEV_CMD_SYS_GET = 6
RDMA_NLDEV_CMD_SYS_SET = 7
)
const (
RDMA_NLDEV_ATTR_DEV_INDEX = 1
RDMA_NLDEV_ATTR_DEV_NAME = 2
RDMA_NLDEV_ATTR_PORT_INDEX = 3
RDMA_NLDEV_ATTR_CAP_FLAGS = 4
RDMA_NLDEV_ATTR_FW_VERSION = 5
RDMA_NLDEV_ATTR_NODE_GUID = 6
RDMA_NLDEV_ATTR_SYS_IMAGE_GUID = 7
RDMA_NLDEV_ATTR_SUBNET_PREFIX = 8
RDMA_NLDEV_ATTR_LID = 9
RDMA_NLDEV_ATTR_SM_LID = 10
RDMA_NLDEV_ATTR_LMC = 11
RDMA_NLDEV_ATTR_PORT_STATE = 12
RDMA_NLDEV_ATTR_PORT_PHYS_STATE = 13
RDMA_NLDEV_ATTR_DEV_NODE_TYPE = 14
RDMA_NLDEV_SYS_ATTR_NETNS_MODE = 66
RDMA_NLDEV_NET_NS_FD = 68
)

View File

@@ -79,3 +79,29 @@ func (msg *RtNexthop) Serialize() []byte {
}
return buf
}
type RtGenMsg struct {
unix.RtGenmsg
}
func NewRtGenMsg() *RtGenMsg {
return &RtGenMsg{
RtGenmsg: unix.RtGenmsg{
Family: unix.AF_UNSPEC,
},
}
}
func (msg *RtGenMsg) Len() int {
return rtaAlignOf(unix.SizeofRtGenmsg)
}
func DeserializeRtGenMsg(b []byte) *RtGenMsg {
return &RtGenMsg{RtGenmsg: unix.RtGenmsg{Family: b[0]}}
}
func (msg *RtGenMsg) Serialize() []byte {
out := make([]byte, msg.Len())
out[0] = msg.Family
return out
}

View File

@@ -99,6 +99,49 @@ func DecodeSEG6Encap(buf []byte) (int, []net.IP, error) {
return mode, srh.Segments, nil
}
func DecodeSEG6Srh(buf []byte) ([]net.IP, error) {
native := NativeEndian()
srh := IPv6SrHdr{
nextHdr: buf[0],
hdrLen: buf[1],
routingType: buf[2],
segmentsLeft: buf[3],
firstSegment: buf[4],
flags: buf[5],
reserved: native.Uint16(buf[6:8]),
}
buf = buf[8:]
if len(buf)%16 != 0 {
err := fmt.Errorf("DecodeSEG6Srh: error parsing Segment List (buf len: %d)", len(buf))
return nil, err
}
for len(buf) > 0 {
srh.Segments = append(srh.Segments, net.IP(buf[:16]))
buf = buf[16:]
}
return srh.Segments, nil
}
func EncodeSEG6Srh(segments []net.IP) ([]byte, error) {
nsegs := len(segments) // nsegs: number of segments
if nsegs == 0 {
return nil, errors.New("EncodeSEG6Srh: No Segments")
}
b := make([]byte, 8, 8+len(segments)*16)
native := NativeEndian()
b[0] = 0 // srh.nextHdr (0 when calling netlink)
b[1] = uint8(16 * nsegs >> 3) // srh.hdrLen (in 8-octets unit)
b[2] = IPV6_SRCRT_TYPE_4 // srh.routingType (assigned by IANA)
b[3] = uint8(nsegs - 1) // srh.segmentsLeft
b[4] = uint8(nsegs - 1) // srh.firstSegment
b[5] = 0 // srh.flags (SR6_FLAG1_HMAC for srh_hmac)
// srh.reserved: Defined as "Tag" in draft-ietf-6man-segment-routing-header-07
native.PutUint16(b[6:], 0) // srh.reserved
for _, netIP := range segments {
b = append(b, netIP...) // srh.Segments
}
return b, nil
}
// Helper functions
func SEG6EncapModeString(mode int) string {
switch mode {

View File

@@ -0,0 +1,76 @@
package nl
import ()
// seg6local parameters
const (
SEG6_LOCAL_UNSPEC = iota
SEG6_LOCAL_ACTION
SEG6_LOCAL_SRH
SEG6_LOCAL_TABLE
SEG6_LOCAL_NH4
SEG6_LOCAL_NH6
SEG6_LOCAL_IIF
SEG6_LOCAL_OIF
__SEG6_LOCAL_MAX
)
const (
SEG6_LOCAL_MAX = __SEG6_LOCAL_MAX
)
// seg6local actions
const (
SEG6_LOCAL_ACTION_END = iota + 1 // 1
SEG6_LOCAL_ACTION_END_X // 2
SEG6_LOCAL_ACTION_END_T // 3
SEG6_LOCAL_ACTION_END_DX2 // 4
SEG6_LOCAL_ACTION_END_DX6 // 5
SEG6_LOCAL_ACTION_END_DX4 // 6
SEG6_LOCAL_ACTION_END_DT6 // 7
SEG6_LOCAL_ACTION_END_DT4 // 8
SEG6_LOCAL_ACTION_END_B6 // 9
SEG6_LOCAL_ACTION_END_B6_ENCAPS // 10
SEG6_LOCAL_ACTION_END_BM // 11
SEG6_LOCAL_ACTION_END_S // 12
SEG6_LOCAL_ACTION_END_AS // 13
SEG6_LOCAL_ACTION_END_AM // 14
__SEG6_LOCAL_ACTION_MAX
)
const (
SEG6_LOCAL_ACTION_MAX = __SEG6_LOCAL_ACTION_MAX - 1
)
// Helper functions
func SEG6LocalActionString(action int) string {
switch action {
case SEG6_LOCAL_ACTION_END:
return "End"
case SEG6_LOCAL_ACTION_END_X:
return "End.X"
case SEG6_LOCAL_ACTION_END_T:
return "End.T"
case SEG6_LOCAL_ACTION_END_DX2:
return "End.DX2"
case SEG6_LOCAL_ACTION_END_DX6:
return "End.DX6"
case SEG6_LOCAL_ACTION_END_DX4:
return "End.DX4"
case SEG6_LOCAL_ACTION_END_DT6:
return "End.DT6"
case SEG6_LOCAL_ACTION_END_DT4:
return "End.DT4"
case SEG6_LOCAL_ACTION_END_B6:
return "End.B6"
case SEG6_LOCAL_ACTION_END_B6_ENCAPS:
return "End.B6.Encaps"
case SEG6_LOCAL_ACTION_END_BM:
return "End.BM"
case SEG6_LOCAL_ACTION_END_S:
return "End.S"
case SEG6_LOCAL_ACTION_END_AS:
return "End.AS"
case SEG6_LOCAL_ACTION_END_AM:
return "End.AM"
}
return "unknown"
}

View File

@@ -21,6 +21,13 @@ const (
FRA_TABLE /* Extended table id */
FRA_FWMASK /* mask for netfilter mark */
FRA_OIFNAME
FRA_PAD
FRA_L3MDEV /* iif or oif is l3mdev goto its table */
FRA_UID_RANGE /* UID range */
FRA_PROTOCOL /* Originator of the rule */
FRA_IP_PROTO /* ip proto */
FRA_SPORT_RANGE /* sport */
FRA_DPORT_RANGE /* dport */
)
// ip rule netlink request types
@@ -42,16 +49,6 @@ const (
TCPDIAG_NOCOOKIE = 0xFFFFFFFF /* TCPDIAG_NOCOOKIE in net/ipv4/tcp_diag.h*/
)
const (
AF_MPLS = 28
)
const (
RTA_NEWDST = 0x13
RTA_ENCAP_TYPE = 0x15
RTA_ENCAP = 0x16
)
// RTA_ENCAP subtype
const (
MPLS_IPTUNNEL_UNSPEC = iota
@@ -67,6 +64,7 @@ const (
LWTUNNEL_ENCAP_IP6
LWTUNNEL_ENCAP_SEG6
LWTUNNEL_ENCAP_BPF
LWTUNNEL_ENCAP_SEG6_LOCAL
)
// routing header types

View File

@@ -1,6 +1,7 @@
package nl
import (
"encoding/binary"
"unsafe"
)
@@ -64,6 +65,15 @@ const (
TCA_PRIO_MAX = TCA_PRIO_MQ
)
const (
TCA_STATS_UNSPEC = iota
TCA_STATS_BASIC
TCA_STATS_RATE_EST
TCA_STATS_QUEUE
TCA_STATS_APP
TCA_STATS_MAX = TCA_STATS_APP
)
const (
SizeofTcMsg = 0x14
SizeofTcActionMsg = 0x04
@@ -79,8 +89,14 @@ const (
SizeofTcU32Key = 0x10
SizeofTcU32Sel = 0x10 // without keys
SizeofTcGen = 0x14
SizeofTcConnmark = SizeofTcGen + 0x04
SizeofTcMirred = SizeofTcGen + 0x08
SizeofTcTunnelKey = SizeofTcGen + 0x04
SizeofTcSkbEdit = SizeofTcGen
SizeofTcPolice = 2*SizeofTcRateSpec + 0x20
SizeofTcSfqQopt = 0x0b
SizeofTcSfqRedStats = 0x18
SizeofTcSfqQoptV1 = SizeofTcSfqQopt + SizeofTcSfqRedStats + 0x1c
)
// struct tcmsg {
@@ -412,6 +428,57 @@ func (x *TcHtbGlob) Serialize() []byte {
return (*(*[SizeofTcHtbGlob]byte)(unsafe.Pointer(x)))[:]
}
// HFSC
type Curve struct {
m1 uint32
d uint32
m2 uint32
}
type HfscCopt struct {
Rsc Curve
Fsc Curve
Usc Curve
}
func (c *Curve) Attrs() (uint32, uint32, uint32) {
return c.m1, c.d, c.m2
}
func (c *Curve) Set(m1 uint32, d uint32, m2 uint32) {
c.m1 = m1
c.d = d
c.m2 = m2
}
func DeserializeHfscCurve(b []byte) *Curve {
return &Curve{
m1: binary.LittleEndian.Uint32(b[0:4]),
d: binary.LittleEndian.Uint32(b[4:8]),
m2: binary.LittleEndian.Uint32(b[8:12]),
}
}
func SerializeHfscCurve(c *Curve) (b []byte) {
t := make([]byte, binary.MaxVarintLen32)
binary.LittleEndian.PutUint32(t, c.m1)
b = append(b, t[:4]...)
binary.LittleEndian.PutUint32(t, c.d)
b = append(b, t[:4]...)
binary.LittleEndian.PutUint32(t, c.m2)
b = append(b, t[:4]...)
return b
}
type TcHfscOpt struct {
Defcls uint16
}
func (x *TcHfscOpt) Serialize() []byte {
return (*(*[2]byte)(unsafe.Pointer(x)))[:]
}
const (
TCA_U32_UNSPEC = iota
TCA_U32_CLASSID
@@ -586,11 +653,47 @@ const (
TCA_BPF_FD
TCA_BPF_NAME
TCA_BPF_FLAGS
TCA_BPF_MAX = TCA_BPF_FLAGS
TCA_BPF_FLAGS_GEN
TCA_BPF_TAG
TCA_BPF_ID
TCA_BPF_MAX = TCA_BPF_ID
)
type TcBpf TcGen
const (
TCA_ACT_CONNMARK = 14
)
const (
TCA_CONNMARK_UNSPEC = iota
TCA_CONNMARK_PARMS
TCA_CONNMARK_TM
TCA_CONNMARK_MAX = TCA_CONNMARK_TM
)
// struct tc_connmark {
// tc_gen;
// __u16 zone;
// };
type TcConnmark struct {
TcGen
Zone uint16
}
func (msg *TcConnmark) Len() int {
return SizeofTcConnmark
}
func DeserializeTcConnmark(b []byte) *TcConnmark {
return (*TcConnmark)(unsafe.Pointer(&b[0:SizeofTcConnmark][0]))
}
func (x *TcConnmark) Serialize() []byte {
return (*(*[SizeofTcConnmark]byte)(unsafe.Pointer(x)))[:]
}
const (
TCA_ACT_MIRRED = 8
)
@@ -626,6 +729,69 @@ func (x *TcMirred) Serialize() []byte {
return (*(*[SizeofTcMirred]byte)(unsafe.Pointer(x)))[:]
}
const (
TCA_TUNNEL_KEY_UNSPEC = iota
TCA_TUNNEL_KEY_TM
TCA_TUNNEL_KEY_PARMS
TCA_TUNNEL_KEY_ENC_IPV4_SRC
TCA_TUNNEL_KEY_ENC_IPV4_DST
TCA_TUNNEL_KEY_ENC_IPV6_SRC
TCA_TUNNEL_KEY_ENC_IPV6_DST
TCA_TUNNEL_KEY_ENC_KEY_ID
TCA_TUNNEL_KEY_PAD
TCA_TUNNEL_KEY_ENC_DST_PORT
TCA_TUNNEL_KEY_NO_CSUM
TCA_TUNNEL_KEY_ENC_OPTS
TCA_TUNNEL_KEY_ENC_TOS
TCA_TUNNEL_KEY_ENC_TTL
TCA_TUNNEL_KEY_MAX
)
type TcTunnelKey struct {
TcGen
Action int32
}
func (x *TcTunnelKey) Len() int {
return SizeofTcTunnelKey
}
func DeserializeTunnelKey(b []byte) *TcTunnelKey {
return (*TcTunnelKey)(unsafe.Pointer(&b[0:SizeofTcTunnelKey][0]))
}
func (x *TcTunnelKey) Serialize() []byte {
return (*(*[SizeofTcTunnelKey]byte)(unsafe.Pointer(x)))[:]
}
const (
TCA_SKBEDIT_UNSPEC = iota
TCA_SKBEDIT_TM
TCA_SKBEDIT_PARMS
TCA_SKBEDIT_PRIORITY
TCA_SKBEDIT_QUEUE_MAPPING
TCA_SKBEDIT_MARK
TCA_SKBEDIT_PAD
TCA_SKBEDIT_PTYPE
TCA_SKBEDIT_MAX = TCA_SKBEDIT_MARK
)
type TcSkbEdit struct {
TcGen
}
func (x *TcSkbEdit) Len() int {
return SizeofTcSkbEdit
}
func DeserializeSkbEdit(b []byte) *TcSkbEdit {
return (*TcSkbEdit)(unsafe.Pointer(&b[0:SizeofTcSkbEdit][0]))
}
func (x *TcSkbEdit) Serialize() []byte {
return (*(*[SizeofTcSkbEdit]byte)(unsafe.Pointer(x)))[:]
}
// struct tc_police {
// __u32 index;
// int action;
@@ -708,3 +874,110 @@ const (
TCA_FQ_CODEL_DROP_BATCH_SIZE
TCA_FQ_CODEL_MEMORY_LIMIT
)
const (
TCA_HFSC_UNSPEC = iota
TCA_HFSC_RSC
TCA_HFSC_FSC
TCA_HFSC_USC
)
// struct tc_sfq_qopt {
// unsigned quantum; /* Bytes per round allocated to flow */
// int perturb_period; /* Period of hash perturbation */
// __u32 limit; /* Maximal packets in queue */
// unsigned divisor; /* Hash divisor */
// unsigned flows; /* Maximal number of flows */
// };
type TcSfqQopt struct {
Quantum uint8
Perturb int32
Limit uint32
Divisor uint8
Flows uint8
}
func (x *TcSfqQopt) Len() int {
return SizeofTcSfqQopt
}
func DeserializeTcSfqQopt(b []byte) *TcSfqQopt {
return (*TcSfqQopt)(unsafe.Pointer(&b[0:SizeofTcSfqQopt][0]))
}
func (x *TcSfqQopt) Serialize() []byte {
return (*(*[SizeofTcSfqQopt]byte)(unsafe.Pointer(x)))[:]
}
// struct tc_sfqred_stats {
// __u32 prob_drop; /* Early drops, below max threshold */
// __u32 forced_drop; /* Early drops, after max threshold */
// __u32 prob_mark; /* Marked packets, below max threshold */
// __u32 forced_mark; /* Marked packets, after max threshold */
// __u32 prob_mark_head; /* Marked packets, below max threshold */
// __u32 forced_mark_head;/* Marked packets, after max threshold */
// };
type TcSfqRedStats struct {
ProbDrop uint32
ForcedDrop uint32
ProbMark uint32
ForcedMark uint32
ProbMarkHead uint32
ForcedMarkHead uint32
}
func (x *TcSfqRedStats) Len() int {
return SizeofTcSfqRedStats
}
func DeserializeTcSfqRedStats(b []byte) *TcSfqRedStats {
return (*TcSfqRedStats)(unsafe.Pointer(&b[0:SizeofTcSfqRedStats][0]))
}
func (x *TcSfqRedStats) Serialize() []byte {
return (*(*[SizeofTcSfqRedStats]byte)(unsafe.Pointer(x)))[:]
}
// struct tc_sfq_qopt_v1 {
// struct tc_sfq_qopt v0;
// unsigned int depth; /* max number of packets per flow */
// unsigned int headdrop;
// /* SFQRED parameters */
// __u32 limit; /* HARD maximal flow queue length (bytes) */
// __u32 qth_min; /* Min average length threshold (bytes) */
// __u32 qth_max; /* Max average length threshold (bytes) */
// unsigned char Wlog; /* log(W) */
// unsigned char Plog; /* log(P_max/(qth_max-qth_min)) */
// unsigned char Scell_log; /* cell size for idle damping */
// unsigned char flags;
// __u32 max_P; /* probability, high resolution */
// /* SFQRED stats */
// struct tc_sfqred_stats stats;
// };
type TcSfqQoptV1 struct {
TcSfqQopt
Depth uint32
HeadDrop uint32
Limit uint32
QthMin uint32
QthMax uint32
Wlog byte
Plog byte
ScellLog byte
Flags byte
MaxP uint32
TcSfqRedStats
}
func (x *TcSfqQoptV1) Len() int {
return SizeofTcSfqQoptV1
}
func DeserializeTcSfqQoptV1(b []byte) *TcSfqQoptV1 {
return (*TcSfqQoptV1)(unsafe.Pointer(&b[0:SizeofTcSfqQoptV1][0]))
}
func (x *TcSfqQoptV1) Serialize() []byte {
return (*(*[SizeofTcSfqQoptV1]byte)(unsafe.Pointer(x)))[:]
}

View File

@@ -50,34 +50,44 @@ const (
// Attribute types
const (
/* Netlink message attributes. */
XFRMA_UNSPEC = 0x00
XFRMA_ALG_AUTH = 0x01 /* struct xfrm_algo */
XFRMA_ALG_CRYPT = 0x02 /* struct xfrm_algo */
XFRMA_ALG_COMP = 0x03 /* struct xfrm_algo */
XFRMA_ENCAP = 0x04 /* struct xfrm_algo + struct xfrm_encap_tmpl */
XFRMA_TMPL = 0x05 /* 1 or more struct xfrm_user_tmpl */
XFRMA_SA = 0x06 /* struct xfrm_usersa_info */
XFRMA_POLICY = 0x07 /* struct xfrm_userpolicy_info */
XFRMA_SEC_CTX = 0x08 /* struct xfrm_sec_ctx */
XFRMA_LTIME_VAL = 0x09
XFRMA_REPLAY_VAL = 0x0a
XFRMA_REPLAY_THRESH = 0x0b
XFRMA_ETIMER_THRESH = 0x0c
XFRMA_SRCADDR = 0x0d /* xfrm_address_t */
XFRMA_COADDR = 0x0e /* xfrm_address_t */
XFRMA_LASTUSED = 0x0f /* unsigned long */
XFRMA_POLICY_TYPE = 0x10 /* struct xfrm_userpolicy_type */
XFRMA_MIGRATE = 0x11
XFRMA_ALG_AEAD = 0x12 /* struct xfrm_algo_aead */
XFRMA_KMADDRESS = 0x13 /* struct xfrm_user_kmaddress */
XFRMA_ALG_AUTH_TRUNC = 0x14 /* struct xfrm_algo_auth */
XFRMA_MARK = 0x15 /* struct xfrm_mark */
XFRMA_TFCPAD = 0x16 /* __u32 */
XFRMA_REPLAY_ESN_VAL = 0x17 /* struct xfrm_replay_esn */
XFRMA_SA_EXTRA_FLAGS = 0x18 /* __u32 */
XFRMA_MAX = 0x18
XFRMA_UNSPEC = iota
XFRMA_ALG_AUTH /* struct xfrm_algo */
XFRMA_ALG_CRYPT /* struct xfrm_algo */
XFRMA_ALG_COMP /* struct xfrm_algo */
XFRMA_ENCAP /* struct xfrm_algo + struct xfrm_encap_tmpl */
XFRMA_TMPL /* 1 or more struct xfrm_user_tmpl */
XFRMA_SA /* struct xfrm_usersa_info */
XFRMA_POLICY /* struct xfrm_userpolicy_info */
XFRMA_SEC_CTX /* struct xfrm_sec_ctx */
XFRMA_LTIME_VAL
XFRMA_REPLAY_VAL
XFRMA_REPLAY_THRESH
XFRMA_ETIMER_THRESH
XFRMA_SRCADDR /* xfrm_address_t */
XFRMA_COADDR /* xfrm_address_t */
XFRMA_LASTUSED /* unsigned long */
XFRMA_POLICY_TYPE /* struct xfrm_userpolicy_type */
XFRMA_MIGRATE
XFRMA_ALG_AEAD /* struct xfrm_algo_aead */
XFRMA_KMADDRESS /* struct xfrm_user_kmaddress */
XFRMA_ALG_AUTH_TRUNC /* struct xfrm_algo_auth */
XFRMA_MARK /* struct xfrm_mark */
XFRMA_TFCPAD /* __u32 */
XFRMA_REPLAY_ESN_VAL /* struct xfrm_replay_esn */
XFRMA_SA_EXTRA_FLAGS /* __u32 */
XFRMA_PROTO /* __u8 */
XFRMA_ADDRESS_FILTER /* struct xfrm_address_filter */
XFRMA_PAD
XFRMA_OFFLOAD_DEV /* struct xfrm_state_offload */
XFRMA_SET_MARK /* __u32 */
XFRMA_SET_MARK_MASK /* __u32 */
XFRMA_IF_ID /* __u32 */
XFRMA_MAX = iota - 1
)
const XFRMA_OUTPUT_MARK = XFRMA_SET_MARK
const (
SizeofXfrmAddress = 0x10
SizeofXfrmSelector = 0x38

View File

@@ -13,7 +13,7 @@ const (
SizeofXfrmAlgoAuth = 0x48
SizeofXfrmAlgoAEAD = 0x48
SizeofXfrmEncapTmpl = 0x18
SizeofXfrmUsersaFlush = 0x8
SizeofXfrmUsersaFlush = 0x1
SizeofXfrmReplayStateEsn = 0x18
)

View File

@@ -18,6 +18,10 @@ type Protinfo struct {
// String returns a list of enabled flags
func (prot *Protinfo) String() string {
if prot == nil {
return "<nil>"
}
var boolStrings []string
if prot.Hairpin {
boolStrings = append(boolStrings, "Hairpin")

View File

@@ -41,7 +41,7 @@ func (h *Handle) LinkGetProtinfo(link Link) (Protinfo, error) {
if err != nil {
return pi, err
}
pi = *parseProtinfo(infos)
pi = parseProtinfo(infos)
return pi, nil
}
@@ -49,8 +49,7 @@ func (h *Handle) LinkGetProtinfo(link Link) (Protinfo, error) {
return pi, fmt.Errorf("Device with index %d not found", base.Index)
}
func parseProtinfo(infos []syscall.NetlinkRouteAttr) *Protinfo {
var pi Protinfo
func parseProtinfo(infos []syscall.NetlinkRouteAttr) (pi Protinfo) {
for _, info := range infos {
switch info.Attr.Type {
case nl.IFLA_BRPORT_MODE:
@@ -71,5 +70,5 @@ func parseProtinfo(infos []syscall.NetlinkRouteAttr) *Protinfo {
pi.ProxyArpWiFi = byteToBool(info.Value[0])
}
}
return &pi
return
}

View File

@@ -176,6 +176,13 @@ type Netem struct {
CorruptCorr uint32
}
func (netem *Netem) String() string {
return fmt.Sprintf(
"{Latency: %v, Limit: %v, Loss: %v, Gap: %v, Duplicate: %v, Jitter: %v}",
netem.Latency, netem.Limit, netem.Loss, netem.Gap, netem.Duplicate, netem.Jitter,
)
}
func (qdisc *Netem) Attrs() *QdiscAttrs {
return &qdisc.QdiscAttrs
}
@@ -231,6 +238,33 @@ func (qdisc *GenericQdisc) Type() string {
return qdisc.QdiscType
}
type Hfsc struct {
QdiscAttrs
Defcls uint16
}
func NewHfsc(attrs QdiscAttrs) *Hfsc {
return &Hfsc{
QdiscAttrs: attrs,
Defcls: 1,
}
}
func (hfsc *Hfsc) Attrs() *QdiscAttrs {
return &hfsc.QdiscAttrs
}
func (hfsc *Hfsc) Type() string {
return "hfsc"
}
func (hfsc *Hfsc) String() string {
return fmt.Sprintf(
"{%v -- default: %d}",
hfsc.Attrs(), hfsc.Defcls,
)
}
// Fq is a classless packet scheduler meant to be mostly used for locally generated traffic.
type Fq struct {
QdiscAttrs
@@ -249,6 +283,13 @@ type Fq struct {
LowRateThreshold uint32
}
func (fq *Fq) String() string {
return fmt.Sprintf(
"{PacketLimit: %v, FlowPacketLimit: %v, Quantum: %v, InitialQuantum: %v, Pacing: %v, FlowDefaultRate: %v, FlowMaxRate: %v, Buckets: %v, FlowRefillDelay: %v, LowRateThreshold: %v}",
fq.PacketLimit, fq.FlowPacketLimit, fq.Quantum, fq.InitialQuantum, fq.Pacing, fq.FlowDefaultRate, fq.FlowMaxRate, fq.Buckets, fq.FlowRefillDelay, fq.LowRateThreshold,
)
}
func NewFq(attrs QdiscAttrs) *Fq {
return &Fq{
QdiscAttrs: attrs,
@@ -267,13 +308,22 @@ func (qdisc *Fq) Type() string {
// FQ_Codel (Fair Queuing Controlled Delay) is queuing discipline that combines Fair Queuing with the CoDel AQM scheme.
type FqCodel struct {
QdiscAttrs
Target uint32
Limit uint32
Interval uint32
ECN uint32
Flows uint32
Quantum uint32
// There are some more attributes here, but support for them seems not ubiquitous
Target uint32
Limit uint32
Interval uint32
ECN uint32
Flows uint32
Quantum uint32
CEThreshold uint32
DropBatchSize uint32
MemoryLimit uint32
}
func (fqcodel *FqCodel) String() string {
return fmt.Sprintf(
"{%v -- Target: %v, Limit: %v, Interval: %v, ECM: %v, Flows: %v, Quantum: %v}",
fqcodel.Attrs(), fqcodel.Target, fqcodel.Limit, fqcodel.Interval, fqcodel.ECN, fqcodel.Flows, fqcodel.Quantum,
)
}
func NewFqCodel(attrs QdiscAttrs) *FqCodel {
@@ -290,3 +340,27 @@ func (qdisc *FqCodel) Attrs() *QdiscAttrs {
func (qdisc *FqCodel) Type() string {
return "fq_codel"
}
type Sfq struct {
QdiscAttrs
// TODO: Only the simplified options for SFQ are handled here. Support for the extended one can be added later.
Quantum uint8
Perturb uint8
Limit uint32
Divisor uint8
}
func (sfq *Sfq) String() string {
return fmt.Sprintf(
"{%v -- Quantum: %v, Perturb: %v, Limit: %v, Divisor: %v}",
sfq.Attrs(), sfq.Quantum, sfq.Perturb, sfq.Limit, sfq.Divisor,
)
}
func (qdisc *Sfq) Attrs() *QdiscAttrs {
return &qdisc.QdiscAttrs
}
func (qdisc *Sfq) Type() string {
return "sfq"
}

View File

@@ -175,15 +175,15 @@ func qdiscPayload(req *nl.NetlinkRequest, qdisc Qdisc) error {
opt.Peakrate.Rate = uint32(qdisc.Peakrate)
opt.Limit = qdisc.Limit
opt.Buffer = qdisc.Buffer
nl.NewRtAttrChild(options, nl.TCA_TBF_PARMS, opt.Serialize())
options.AddRtAttr(nl.TCA_TBF_PARMS, opt.Serialize())
if qdisc.Rate >= uint64(1<<32) {
nl.NewRtAttrChild(options, nl.TCA_TBF_RATE64, nl.Uint64Attr(qdisc.Rate))
options.AddRtAttr(nl.TCA_TBF_RATE64, nl.Uint64Attr(qdisc.Rate))
}
if qdisc.Peakrate >= uint64(1<<32) {
nl.NewRtAttrChild(options, nl.TCA_TBF_PRATE64, nl.Uint64Attr(qdisc.Peakrate))
options.AddRtAttr(nl.TCA_TBF_PRATE64, nl.Uint64Attr(qdisc.Peakrate))
}
if qdisc.Peakrate > 0 {
nl.NewRtAttrChild(options, nl.TCA_TBF_PBURST, nl.Uint32Attr(qdisc.Minburst))
options.AddRtAttr(nl.TCA_TBF_PBURST, nl.Uint32Attr(qdisc.Minburst))
}
case *Htb:
opt := nl.TcHtbGlob{}
@@ -193,8 +193,12 @@ func qdiscPayload(req *nl.NetlinkRequest, qdisc Qdisc) error {
// TODO: Handle Debug properly. For now default to 0
opt.Debug = qdisc.Debug
opt.DirectPkts = qdisc.DirectPkts
nl.NewRtAttrChild(options, nl.TCA_HTB_INIT, opt.Serialize())
// nl.NewRtAttrChild(options, nl.TCA_HTB_DIRECT_QLEN, opt.Serialize())
options.AddRtAttr(nl.TCA_HTB_INIT, opt.Serialize())
// options.AddRtAttr(nl.TCA_HTB_DIRECT_QLEN, opt.Serialize())
case *Hfsc:
opt := nl.TcHfscOpt{}
opt.Defcls = qdisc.Defcls
options = nl.NewRtAttr(nl.TCA_OPTIONS, opt.Serialize())
case *Netem:
opt := nl.TcNetemQopt{}
opt.Latency = qdisc.Latency
@@ -211,21 +215,21 @@ func qdiscPayload(req *nl.NetlinkRequest, qdisc Qdisc) error {
corr.DupCorr = qdisc.DuplicateCorr
if corr.DelayCorr > 0 || corr.LossCorr > 0 || corr.DupCorr > 0 {
nl.NewRtAttrChild(options, nl.TCA_NETEM_CORR, corr.Serialize())
options.AddRtAttr(nl.TCA_NETEM_CORR, corr.Serialize())
}
// Corruption
corruption := nl.TcNetemCorrupt{}
corruption.Probability = qdisc.CorruptProb
corruption.Correlation = qdisc.CorruptCorr
if corruption.Probability > 0 {
nl.NewRtAttrChild(options, nl.TCA_NETEM_CORRUPT, corruption.Serialize())
options.AddRtAttr(nl.TCA_NETEM_CORRUPT, corruption.Serialize())
}
// Reorder
reorder := nl.TcNetemReorder{}
reorder.Probability = qdisc.ReorderProb
reorder.Correlation = qdisc.ReorderCorr
if reorder.Probability > 0 {
nl.NewRtAttrChild(options, nl.TCA_NETEM_REORDER, reorder.Serialize())
options.AddRtAttr(nl.TCA_NETEM_REORDER, reorder.Serialize())
}
case *Ingress:
// ingress filters must use the proper handle
@@ -233,50 +237,70 @@ func qdiscPayload(req *nl.NetlinkRequest, qdisc Qdisc) error {
return fmt.Errorf("Ingress filters must set Parent to HANDLE_INGRESS")
}
case *FqCodel:
nl.NewRtAttrChild(options, nl.TCA_FQ_CODEL_ECN, nl.Uint32Attr((uint32(qdisc.ECN))))
options.AddRtAttr(nl.TCA_FQ_CODEL_ECN, nl.Uint32Attr((uint32(qdisc.ECN))))
if qdisc.Limit > 0 {
nl.NewRtAttrChild(options, nl.TCA_FQ_CODEL_LIMIT, nl.Uint32Attr((uint32(qdisc.Limit))))
options.AddRtAttr(nl.TCA_FQ_CODEL_LIMIT, nl.Uint32Attr((uint32(qdisc.Limit))))
}
if qdisc.Interval > 0 {
nl.NewRtAttrChild(options, nl.TCA_FQ_CODEL_INTERVAL, nl.Uint32Attr((uint32(qdisc.Interval))))
options.AddRtAttr(nl.TCA_FQ_CODEL_INTERVAL, nl.Uint32Attr((uint32(qdisc.Interval))))
}
if qdisc.Flows > 0 {
nl.NewRtAttrChild(options, nl.TCA_FQ_CODEL_FLOWS, nl.Uint32Attr((uint32(qdisc.Flows))))
options.AddRtAttr(nl.TCA_FQ_CODEL_FLOWS, nl.Uint32Attr((uint32(qdisc.Flows))))
}
if qdisc.Quantum > 0 {
nl.NewRtAttrChild(options, nl.TCA_FQ_CODEL_QUANTUM, nl.Uint32Attr((uint32(qdisc.Quantum))))
options.AddRtAttr(nl.TCA_FQ_CODEL_QUANTUM, nl.Uint32Attr((uint32(qdisc.Quantum))))
}
if qdisc.CEThreshold > 0 {
options.AddRtAttr(nl.TCA_FQ_CODEL_CE_THRESHOLD, nl.Uint32Attr(qdisc.CEThreshold))
}
if qdisc.DropBatchSize > 0 {
options.AddRtAttr(nl.TCA_FQ_CODEL_DROP_BATCH_SIZE, nl.Uint32Attr(qdisc.DropBatchSize))
}
if qdisc.MemoryLimit > 0 {
options.AddRtAttr(nl.TCA_FQ_CODEL_MEMORY_LIMIT, nl.Uint32Attr(qdisc.MemoryLimit))
}
case *Fq:
nl.NewRtAttrChild(options, nl.TCA_FQ_RATE_ENABLE, nl.Uint32Attr((uint32(qdisc.Pacing))))
options.AddRtAttr(nl.TCA_FQ_RATE_ENABLE, nl.Uint32Attr((uint32(qdisc.Pacing))))
if qdisc.Buckets > 0 {
nl.NewRtAttrChild(options, nl.TCA_FQ_BUCKETS_LOG, nl.Uint32Attr((uint32(qdisc.Buckets))))
options.AddRtAttr(nl.TCA_FQ_BUCKETS_LOG, nl.Uint32Attr((uint32(qdisc.Buckets))))
}
if qdisc.LowRateThreshold > 0 {
nl.NewRtAttrChild(options, nl.TCA_FQ_LOW_RATE_THRESHOLD, nl.Uint32Attr((uint32(qdisc.LowRateThreshold))))
options.AddRtAttr(nl.TCA_FQ_LOW_RATE_THRESHOLD, nl.Uint32Attr((uint32(qdisc.LowRateThreshold))))
}
if qdisc.Quantum > 0 {
nl.NewRtAttrChild(options, nl.TCA_FQ_QUANTUM, nl.Uint32Attr((uint32(qdisc.Quantum))))
options.AddRtAttr(nl.TCA_FQ_QUANTUM, nl.Uint32Attr((uint32(qdisc.Quantum))))
}
if qdisc.InitialQuantum > 0 {
nl.NewRtAttrChild(options, nl.TCA_FQ_INITIAL_QUANTUM, nl.Uint32Attr((uint32(qdisc.InitialQuantum))))
options.AddRtAttr(nl.TCA_FQ_INITIAL_QUANTUM, nl.Uint32Attr((uint32(qdisc.InitialQuantum))))
}
if qdisc.FlowRefillDelay > 0 {
nl.NewRtAttrChild(options, nl.TCA_FQ_FLOW_REFILL_DELAY, nl.Uint32Attr((uint32(qdisc.FlowRefillDelay))))
options.AddRtAttr(nl.TCA_FQ_FLOW_REFILL_DELAY, nl.Uint32Attr((uint32(qdisc.FlowRefillDelay))))
}
if qdisc.FlowPacketLimit > 0 {
nl.NewRtAttrChild(options, nl.TCA_FQ_FLOW_PLIMIT, nl.Uint32Attr((uint32(qdisc.FlowPacketLimit))))
options.AddRtAttr(nl.TCA_FQ_FLOW_PLIMIT, nl.Uint32Attr((uint32(qdisc.FlowPacketLimit))))
}
if qdisc.FlowMaxRate > 0 {
nl.NewRtAttrChild(options, nl.TCA_FQ_FLOW_MAX_RATE, nl.Uint32Attr((uint32(qdisc.FlowMaxRate))))
options.AddRtAttr(nl.TCA_FQ_FLOW_MAX_RATE, nl.Uint32Attr((uint32(qdisc.FlowMaxRate))))
}
if qdisc.FlowDefaultRate > 0 {
nl.NewRtAttrChild(options, nl.TCA_FQ_FLOW_DEFAULT_RATE, nl.Uint32Attr((uint32(qdisc.FlowDefaultRate))))
options.AddRtAttr(nl.TCA_FQ_FLOW_DEFAULT_RATE, nl.Uint32Attr((uint32(qdisc.FlowDefaultRate))))
}
case *Sfq:
opt := nl.TcSfqQoptV1{}
opt.TcSfqQopt.Quantum = qdisc.Quantum
opt.TcSfqQopt.Perturb = int32(qdisc.Perturb)
opt.TcSfqQopt.Limit = qdisc.Limit
opt.TcSfqQopt.Divisor = qdisc.Divisor
options = nl.NewRtAttr(nl.TCA_OPTIONS, opt.Serialize())
default:
options = nil
}
req.AddData(options)
if options != nil {
req.AddData(options)
}
return nil
}
@@ -348,10 +372,14 @@ func (h *Handle) QdiscList(link Link) ([]Qdisc, error) {
qdisc = &Htb{}
case "fq":
qdisc = &Fq{}
case "hfsc":
qdisc = &Hfsc{}
case "fq_codel":
qdisc = &FqCodel{}
case "netem":
qdisc = &Netem{}
case "sfq":
qdisc = &Sfq{}
default:
qdisc = &GenericQdisc{QdiscType: qdiscType}
}
@@ -375,6 +403,10 @@ func (h *Handle) QdiscList(link Link) ([]Qdisc, error) {
if err := parseTbfData(qdisc, data); err != nil {
return nil, err
}
case "hfsc":
if err := parseHfscData(qdisc, attr.Value); err != nil {
return nil, err
}
case "htb":
data, err := nl.ParseRouteAttr(attr.Value)
if err != nil {
@@ -403,6 +435,10 @@ func (h *Handle) QdiscList(link Link) ([]Qdisc, error) {
if err := parseNetemData(qdisc, attr.Value); err != nil {
return nil, err
}
case "sfq":
if err := parseSfqData(qdisc, attr.Value); err != nil {
return nil, err
}
// no options for ingress
}
@@ -469,11 +505,24 @@ func parseFqCodelData(qdisc Qdisc, data []syscall.NetlinkRouteAttr) error {
fqCodel.Flows = native.Uint32(datum.Value)
case nl.TCA_FQ_CODEL_QUANTUM:
fqCodel.Quantum = native.Uint32(datum.Value)
case nl.TCA_FQ_CODEL_CE_THRESHOLD:
fqCodel.CEThreshold = native.Uint32(datum.Value)
case nl.TCA_FQ_CODEL_DROP_BATCH_SIZE:
fqCodel.DropBatchSize = native.Uint32(datum.Value)
case nl.TCA_FQ_CODEL_MEMORY_LIMIT:
fqCodel.MemoryLimit = native.Uint32(datum.Value)
}
}
return nil
}
func parseHfscData(qdisc Qdisc, data []byte) error {
Hfsc := qdisc.(*Hfsc)
native = nl.NativeEndian()
Hfsc.Defcls = native.Uint16(data)
return nil
}
func parseFqData(qdisc Qdisc, data []syscall.NetlinkRouteAttr) error {
native = nl.NativeEndian()
fq := qdisc.(*Fq)
@@ -561,6 +610,17 @@ func parseTbfData(qdisc Qdisc, data []syscall.NetlinkRouteAttr) error {
return nil
}
func parseSfqData(qdisc Qdisc, value []byte) error {
sfq := qdisc.(*Sfq)
opt := nl.DeserializeTcSfqQoptV1(value)
sfq.Quantum = opt.TcSfqQopt.Quantum
sfq.Perturb = uint8(opt.TcSfqQopt.Perturb)
sfq.Limit = opt.TcSfqQopt.Limit
sfq.Divisor = opt.TcSfqQopt.Divisor
return nil
}
const (
TIME_UNITS_PER_SEC = 1000000
)
@@ -577,10 +637,10 @@ func initClock() {
return
}
parts := strings.Split(strings.TrimSpace(string(data)), " ")
if len(parts) < 3 {
if len(parts) < 4 {
return
}
var vals [3]uint64
var vals [4]uint64
for i := range vals {
val, err := strconv.ParseUint(parts[i], 16, 32)
if err != nil {
@@ -594,7 +654,12 @@ func initClock() {
}
clockFactor = float64(vals[2]) / TIME_UNITS_PER_SEC
tickInUsec = float64(vals[0]) / float64(vals[1]) * clockFactor
hz = float64(vals[0])
if vals[2] == 1000000 {
// ref https://git.kernel.org/pub/scm/network/iproute2/iproute2.git/tree/lib/utils.c#n963
hz = float64(vals[3])
} else {
hz = 100
}
}
func TickInUsec() float64 {
@@ -642,6 +707,7 @@ func latency(rate uint64, limit, buffer uint32) float64 {
return TIME_UNITS_PER_SEC*(float64(limit)/float64(rate)) - float64(tick2Time(buffer))
}
func Xmittime(rate uint64, size uint32) float64 {
return TickInUsec() * TIME_UNITS_PER_SEC * (float64(size) / float64(rate))
func Xmittime(rate uint64, size uint32) uint32 {
// https://git.kernel.org/pub/scm/network/iproute2/iproute2.git/tree/tc/tc_core.c#n62
return time2Tick(uint32(TIME_UNITS_PER_SEC * (float64(size) / float64(rate))))
}

View File

@@ -0,0 +1,280 @@
package netlink
import (
"bytes"
"encoding/binary"
"fmt"
"net"
"github.com/vishvananda/netlink/nl"
"golang.org/x/sys/unix"
)
// LinkAttrs represents data shared by most link types
type RdmaLinkAttrs struct {
Index uint32
Name string
FirmwareVersion string
NodeGuid string
SysImageGuid string
}
// Link represents a rdma device from netlink.
type RdmaLink struct {
Attrs RdmaLinkAttrs
}
func getProtoField(clientType int, op int) int {
return ((clientType << nl.RDMA_NL_GET_CLIENT_SHIFT) | op)
}
func uint64ToGuidString(guid uint64) string {
//Convert to byte array
sysGuidBytes := new(bytes.Buffer)
binary.Write(sysGuidBytes, binary.LittleEndian, guid)
//Convert to HardwareAddr
sysGuidNet := net.HardwareAddr(sysGuidBytes.Bytes())
//Get the String
return sysGuidNet.String()
}
func executeOneGetRdmaLink(data []byte) (*RdmaLink, error) {
link := RdmaLink{}
reader := bytes.NewReader(data)
for reader.Len() >= 4 {
_, attrType, len, value := parseNfAttrTLV(reader)
switch attrType {
case nl.RDMA_NLDEV_ATTR_DEV_INDEX:
var Index uint32
r := bytes.NewReader(value)
binary.Read(r, nl.NativeEndian(), &Index)
link.Attrs.Index = Index
case nl.RDMA_NLDEV_ATTR_DEV_NAME:
link.Attrs.Name = string(value[0 : len-1])
case nl.RDMA_NLDEV_ATTR_FW_VERSION:
link.Attrs.FirmwareVersion = string(value[0 : len-1])
case nl.RDMA_NLDEV_ATTR_NODE_GUID:
var guid uint64
r := bytes.NewReader(value)
binary.Read(r, nl.NativeEndian(), &guid)
link.Attrs.NodeGuid = uint64ToGuidString(guid)
case nl.RDMA_NLDEV_ATTR_SYS_IMAGE_GUID:
var sysGuid uint64
r := bytes.NewReader(value)
binary.Read(r, nl.NativeEndian(), &sysGuid)
link.Attrs.SysImageGuid = uint64ToGuidString(sysGuid)
}
if (len % 4) != 0 {
// Skip pad bytes
reader.Seek(int64(4-(len%4)), seekCurrent)
}
}
return &link, nil
}
func execRdmaSetLink(req *nl.NetlinkRequest) error {
_, err := req.Execute(unix.NETLINK_RDMA, 0)
return err
}
// RdmaLinkList gets a list of RDMA link devices.
// Equivalent to: `rdma dev show`
func RdmaLinkList() ([]*RdmaLink, error) {
return pkgHandle.RdmaLinkList()
}
// RdmaLinkList gets a list of RDMA link devices.
// Equivalent to: `rdma dev show`
func (h *Handle) RdmaLinkList() ([]*RdmaLink, error) {
proto := getProtoField(nl.RDMA_NL_NLDEV, nl.RDMA_NLDEV_CMD_GET)
req := h.newNetlinkRequest(proto, unix.NLM_F_ACK|unix.NLM_F_DUMP)
msgs, err := req.Execute(unix.NETLINK_RDMA, 0)
if err != nil {
return nil, err
}
var res []*RdmaLink
for _, m := range msgs {
link, err := executeOneGetRdmaLink(m)
if err != nil {
return nil, err
}
res = append(res, link)
}
return res, nil
}
// RdmaLinkByName finds a link by name and returns a pointer to the object if
// found and nil error, otherwise returns error code.
func RdmaLinkByName(name string) (*RdmaLink, error) {
return pkgHandle.RdmaLinkByName(name)
}
// RdmaLinkByName finds a link by name and returns a pointer to the object if
// found and nil error, otherwise returns error code.
func (h *Handle) RdmaLinkByName(name string) (*RdmaLink, error) {
links, err := h.RdmaLinkList()
if err != nil {
return nil, err
}
for _, link := range links {
if link.Attrs.Name == name {
return link, nil
}
}
return nil, fmt.Errorf("Rdma device %v not found", name)
}
// RdmaLinkSetName sets the name of the rdma link device. Return nil on success
// or error otherwise.
// Equivalent to: `rdma dev set $old_devname name $name`
func RdmaLinkSetName(link *RdmaLink, name string) error {
return pkgHandle.RdmaLinkSetName(link, name)
}
// RdmaLinkSetName sets the name of the rdma link device. Return nil on success
// or error otherwise.
// Equivalent to: `rdma dev set $old_devname name $name`
func (h *Handle) RdmaLinkSetName(link *RdmaLink, name string) error {
proto := getProtoField(nl.RDMA_NL_NLDEV, nl.RDMA_NLDEV_CMD_SET)
req := h.newNetlinkRequest(proto, unix.NLM_F_ACK)
b := make([]byte, 4)
native.PutUint32(b, uint32(link.Attrs.Index))
data := nl.NewRtAttr(nl.RDMA_NLDEV_ATTR_DEV_INDEX, b)
req.AddData(data)
b = make([]byte, len(name)+1)
copy(b, name)
data = nl.NewRtAttr(nl.RDMA_NLDEV_ATTR_DEV_NAME, b)
req.AddData(data)
return execRdmaSetLink(req)
}
func netnsModeToString(mode uint8) string {
switch mode {
case 0:
return "exclusive"
case 1:
return "shared"
default:
return "unknown"
}
}
func executeOneGetRdmaNetnsMode(data []byte) (string, error) {
reader := bytes.NewReader(data)
for reader.Len() >= 4 {
_, attrType, len, value := parseNfAttrTLV(reader)
switch attrType {
case nl.RDMA_NLDEV_SYS_ATTR_NETNS_MODE:
var mode uint8
r := bytes.NewReader(value)
binary.Read(r, nl.NativeEndian(), &mode)
return netnsModeToString(mode), nil
}
if (len % 4) != 0 {
// Skip pad bytes
reader.Seek(int64(4-(len%4)), seekCurrent)
}
}
return "", fmt.Errorf("Invalid netns mode")
}
// RdmaSystemGetNetnsMode gets the net namespace mode for RDMA subsystem
// Returns mode string and error status as nil on success or returns error
// otherwise.
// Equivalent to: `rdma system show netns'
func RdmaSystemGetNetnsMode() (string, error) {
return pkgHandle.RdmaSystemGetNetnsMode()
}
// RdmaSystemGetNetnsMode gets the net namespace mode for RDMA subsystem
// Returns mode string and error status as nil on success or returns error
// otherwise.
// Equivalent to: `rdma system show netns'
func (h *Handle) RdmaSystemGetNetnsMode() (string, error) {
proto := getProtoField(nl.RDMA_NL_NLDEV, nl.RDMA_NLDEV_CMD_SYS_GET)
req := h.newNetlinkRequest(proto, unix.NLM_F_ACK)
msgs, err := req.Execute(unix.NETLINK_RDMA, 0)
if err != nil {
return "", err
}
if len(msgs) == 0 {
return "", fmt.Errorf("No valid response from kernel")
}
return executeOneGetRdmaNetnsMode(msgs[0])
}
func netnsModeStringToUint8(mode string) (uint8, error) {
switch mode {
case "exclusive":
return 0, nil
case "shared":
return 1, nil
default:
return 0, fmt.Errorf("Invalid mode; %q", mode)
}
}
// RdmaSystemSetNetnsMode sets the net namespace mode for RDMA subsystem
// Returns nil on success or appropriate error code.
// Equivalent to: `rdma system set netns { shared | exclusive }'
func RdmaSystemSetNetnsMode(NewMode string) error {
return pkgHandle.RdmaSystemSetNetnsMode(NewMode)
}
// RdmaSystemSetNetnsMode sets the net namespace mode for RDMA subsystem
// Returns nil on success or appropriate error code.
// Equivalent to: `rdma system set netns { shared | exclusive }'
func (h *Handle) RdmaSystemSetNetnsMode(NewMode string) error {
value, err := netnsModeStringToUint8(NewMode)
if err != nil {
return err
}
proto := getProtoField(nl.RDMA_NL_NLDEV, nl.RDMA_NLDEV_CMD_SYS_SET)
req := h.newNetlinkRequest(proto, unix.NLM_F_ACK)
data := nl.NewRtAttr(nl.RDMA_NLDEV_SYS_ATTR_NETNS_MODE, []byte{value})
req.AddData(data)
_, err = req.Execute(unix.NETLINK_RDMA, 0)
return err
}
// RdmaLinkSetNsFd puts the RDMA device into a new network namespace. The
// fd must be an open file descriptor to a network namespace.
// Similar to: `rdma dev set $dev netns $ns`
func RdmaLinkSetNsFd(link *RdmaLink, fd uint32) error {
return pkgHandle.RdmaLinkSetNsFd(link, fd)
}
// RdmaLinkSetNsFd puts the RDMA device into a new network namespace. The
// fd must be an open file descriptor to a network namespace.
// Similar to: `rdma dev set $dev netns $ns`
func (h *Handle) RdmaLinkSetNsFd(link *RdmaLink, fd uint32) error {
proto := getProtoField(nl.RDMA_NL_NLDEV, nl.RDMA_NLDEV_CMD_SET)
req := h.newNetlinkRequest(proto, unix.NLM_F_ACK)
data := nl.NewRtAttr(nl.RDMA_NLDEV_ATTR_DEV_INDEX,
nl.Uint32Attr(link.Attrs.Index))
req.AddData(data)
data = nl.NewRtAttr(nl.RDMA_NLDEV_NET_NS_FD, nl.Uint32Attr(fd))
req.AddData(data)
return execRdmaSetLink(req)
}

View File

@@ -27,26 +27,44 @@ type Encap interface {
Equal(Encap) bool
}
//Protocol describe what was the originator of the route
type RouteProtocol int
// Route represents a netlink route.
type Route struct {
LinkIndex int
ILinkIndex int
Scope Scope
Dst *net.IPNet
Src net.IP
Gw net.IP
MultiPath []*NexthopInfo
Protocol int
Priority int
Table int
Type int
Tos int
Flags int
MPLSDst *int
NewDst Destination
Encap Encap
MTU int
AdvMSS int
LinkIndex int
ILinkIndex int
Scope Scope
Dst *net.IPNet
Src net.IP
Gw net.IP
MultiPath []*NexthopInfo
Protocol RouteProtocol
Priority int
Table int
Type int
Tos int
Flags int
MPLSDst *int
NewDst Destination
Encap Encap
Via Destination
MTU int
Window int
Rtt int
RttVar int
Ssthresh int
Cwnd int
AdvMSS int
Reordering int
Hoplimit int
InitCwnd int
Features int
RtoMin int
InitRwnd int
QuickACK int
Congctl string
FastOpenNoCookie int
}
func (r Route) String() string {
@@ -65,6 +83,9 @@ func (r Route) String() string {
if r.Encap != nil {
elems = append(elems, fmt.Sprintf("Encap: %s", r.Encap))
}
if r.Via != nil {
elems = append(elems, fmt.Sprintf("Via: %s", r.Via))
}
elems = append(elems, fmt.Sprintf("Src: %s", r.Src))
if len(r.MultiPath) > 0 {
elems = append(elems, fmt.Sprintf("Gw: %s", r.MultiPath))
@@ -89,9 +110,11 @@ func (r Route) Equal(x Route) bool {
r.Table == x.Table &&
r.Type == x.Type &&
r.Tos == x.Tos &&
r.Hoplimit == x.Hoplimit &&
r.Flags == x.Flags &&
(r.MPLSDst == x.MPLSDst || (r.MPLSDst != nil && x.MPLSDst != nil && *r.MPLSDst == *x.MPLSDst)) &&
(r.NewDst == x.NewDst || (r.NewDst != nil && r.NewDst.Equal(x.NewDst))) &&
(r.Via == x.Via || (r.Via != nil && r.Via.Equal(x.Via))) &&
(r.Encap == x.Encap || (r.Encap != nil && r.Encap.Equal(x.Encap)))
}
@@ -121,6 +144,7 @@ type NexthopInfo struct {
Flags int
NewDst Destination
Encap Encap
Via Destination
}
func (n *NexthopInfo) String() string {
@@ -132,6 +156,9 @@ func (n *NexthopInfo) String() string {
if n.Encap != nil {
elems = append(elems, fmt.Sprintf("Encap: %s", n.Encap))
}
if n.Via != nil {
elems = append(elems, fmt.Sprintf("Via: %s", n.Via))
}
elems = append(elems, fmt.Sprintf("Weight: %d", n.Hops+1))
elems = append(elems, fmt.Sprintf("Gw: %s", n.Gw))
elems = append(elems, fmt.Sprintf("Flags: %s", n.ListFlags()))

View File

@@ -1,8 +1,11 @@
package netlink
import (
"bytes"
"encoding/binary"
"fmt"
"net"
"strconv"
"strings"
"syscall"
@@ -21,6 +24,23 @@ const (
SCOPE_NOWHERE Scope = unix.RT_SCOPE_NOWHERE
)
func (s Scope) String() string {
switch s {
case SCOPE_UNIVERSE:
return "universe"
case SCOPE_SITE:
return "site"
case SCOPE_LINK:
return "link"
case SCOPE_HOST:
return "host"
case SCOPE_NOWHERE:
return "nowhere"
default:
return "unknown"
}
}
const (
RT_FILTER_PROTOCOL uint64 = 1 << (1 + iota)
RT_FILTER_SCOPE
@@ -32,6 +52,10 @@ const (
RT_FILTER_SRC
RT_FILTER_GW
RT_FILTER_TABLE
RT_FILTER_HOPLIMIT
RT_FILTER_PRIORITY
RT_FILTER_MARK
RT_FILTER_MASK
)
const (
@@ -207,6 +231,7 @@ func (e *SEG6Encap) Decode(buf []byte) error {
}
buf = buf[:l] // make sure buf size upper limit is Length
typ := native.Uint16(buf[2:])
// LWTUNNEL_ENCAP_SEG6 has only one attr type SEG6_IPTUNNEL_SRH
if typ != nl.SEG6_IPTUNNEL_SRH {
return fmt.Errorf("unknown SEG6 Type: %d", typ)
}
@@ -259,6 +284,244 @@ func (e *SEG6Encap) Equal(x Encap) bool {
return true
}
// SEG6LocalEncap definitions
type SEG6LocalEncap struct {
Flags [nl.SEG6_LOCAL_MAX]bool
Action int
Segments []net.IP // from SRH in seg6_local_lwt
Table int // table id for End.T and End.DT6
InAddr net.IP
In6Addr net.IP
Iif int
Oif int
}
func (e *SEG6LocalEncap) Type() int {
return nl.LWTUNNEL_ENCAP_SEG6_LOCAL
}
func (e *SEG6LocalEncap) Decode(buf []byte) error {
attrs, err := nl.ParseRouteAttr(buf)
if err != nil {
return err
}
native := nl.NativeEndian()
for _, attr := range attrs {
switch attr.Attr.Type {
case nl.SEG6_LOCAL_ACTION:
e.Action = int(native.Uint32(attr.Value[0:4]))
e.Flags[nl.SEG6_LOCAL_ACTION] = true
case nl.SEG6_LOCAL_SRH:
e.Segments, err = nl.DecodeSEG6Srh(attr.Value[:])
e.Flags[nl.SEG6_LOCAL_SRH] = true
case nl.SEG6_LOCAL_TABLE:
e.Table = int(native.Uint32(attr.Value[0:4]))
e.Flags[nl.SEG6_LOCAL_TABLE] = true
case nl.SEG6_LOCAL_NH4:
e.InAddr = net.IP(attr.Value[0:4])
e.Flags[nl.SEG6_LOCAL_NH4] = true
case nl.SEG6_LOCAL_NH6:
e.In6Addr = net.IP(attr.Value[0:16])
e.Flags[nl.SEG6_LOCAL_NH6] = true
case nl.SEG6_LOCAL_IIF:
e.Iif = int(native.Uint32(attr.Value[0:4]))
e.Flags[nl.SEG6_LOCAL_IIF] = true
case nl.SEG6_LOCAL_OIF:
e.Oif = int(native.Uint32(attr.Value[0:4]))
e.Flags[nl.SEG6_LOCAL_OIF] = true
}
}
return err
}
func (e *SEG6LocalEncap) Encode() ([]byte, error) {
var err error
native := nl.NativeEndian()
res := make([]byte, 8)
native.PutUint16(res, 8) // length
native.PutUint16(res[2:], nl.SEG6_LOCAL_ACTION)
native.PutUint32(res[4:], uint32(e.Action))
if e.Flags[nl.SEG6_LOCAL_SRH] {
srh, err := nl.EncodeSEG6Srh(e.Segments)
if err != nil {
return nil, err
}
attr := make([]byte, 4)
native.PutUint16(attr, uint16(len(srh)+4))
native.PutUint16(attr[2:], nl.SEG6_LOCAL_SRH)
attr = append(attr, srh...)
res = append(res, attr...)
}
if e.Flags[nl.SEG6_LOCAL_TABLE] {
attr := make([]byte, 8)
native.PutUint16(attr, 8)
native.PutUint16(attr[2:], nl.SEG6_LOCAL_TABLE)
native.PutUint32(attr[4:], uint32(e.Table))
res = append(res, attr...)
}
if e.Flags[nl.SEG6_LOCAL_NH4] {
attr := make([]byte, 4)
native.PutUint16(attr, 8)
native.PutUint16(attr[2:], nl.SEG6_LOCAL_NH4)
ipv4 := e.InAddr.To4()
if ipv4 == nil {
err = fmt.Errorf("SEG6_LOCAL_NH4 has invalid IPv4 address")
return nil, err
}
attr = append(attr, ipv4...)
res = append(res, attr...)
}
if e.Flags[nl.SEG6_LOCAL_NH6] {
attr := make([]byte, 4)
native.PutUint16(attr, 20)
native.PutUint16(attr[2:], nl.SEG6_LOCAL_NH6)
attr = append(attr, e.In6Addr...)
res = append(res, attr...)
}
if e.Flags[nl.SEG6_LOCAL_IIF] {
attr := make([]byte, 8)
native.PutUint16(attr, 8)
native.PutUint16(attr[2:], nl.SEG6_LOCAL_IIF)
native.PutUint32(attr[4:], uint32(e.Iif))
res = append(res, attr...)
}
if e.Flags[nl.SEG6_LOCAL_OIF] {
attr := make([]byte, 8)
native.PutUint16(attr, 8)
native.PutUint16(attr[2:], nl.SEG6_LOCAL_OIF)
native.PutUint32(attr[4:], uint32(e.Oif))
res = append(res, attr...)
}
return res, err
}
func (e *SEG6LocalEncap) String() string {
strs := make([]string, 0, nl.SEG6_LOCAL_MAX)
strs = append(strs, fmt.Sprintf("action %s", nl.SEG6LocalActionString(e.Action)))
if e.Flags[nl.SEG6_LOCAL_TABLE] {
strs = append(strs, fmt.Sprintf("table %d", e.Table))
}
if e.Flags[nl.SEG6_LOCAL_NH4] {
strs = append(strs, fmt.Sprintf("nh4 %s", e.InAddr))
}
if e.Flags[nl.SEG6_LOCAL_NH6] {
strs = append(strs, fmt.Sprintf("nh6 %s", e.In6Addr))
}
if e.Flags[nl.SEG6_LOCAL_IIF] {
link, err := LinkByIndex(e.Iif)
if err != nil {
strs = append(strs, fmt.Sprintf("iif %d", e.Iif))
} else {
strs = append(strs, fmt.Sprintf("iif %s", link.Attrs().Name))
}
}
if e.Flags[nl.SEG6_LOCAL_OIF] {
link, err := LinkByIndex(e.Oif)
if err != nil {
strs = append(strs, fmt.Sprintf("oif %d", e.Oif))
} else {
strs = append(strs, fmt.Sprintf("oif %s", link.Attrs().Name))
}
}
if e.Flags[nl.SEG6_LOCAL_SRH] {
segs := make([]string, 0, len(e.Segments))
//append segment backwards (from n to 0) since seg#0 is the last segment.
for i := len(e.Segments); i > 0; i-- {
segs = append(segs, fmt.Sprintf("%s", e.Segments[i-1]))
}
strs = append(strs, fmt.Sprintf("segs %d [ %s ]", len(e.Segments), strings.Join(segs, " ")))
}
return strings.Join(strs, " ")
}
func (e *SEG6LocalEncap) Equal(x Encap) bool {
o, ok := x.(*SEG6LocalEncap)
if !ok {
return false
}
if e == o {
return true
}
if e == nil || o == nil {
return false
}
// compare all arrays first
for i := range e.Flags {
if e.Flags[i] != o.Flags[i] {
return false
}
}
if len(e.Segments) != len(o.Segments) {
return false
}
for i := range e.Segments {
if !e.Segments[i].Equal(o.Segments[i]) {
return false
}
}
// compare values
if !e.InAddr.Equal(o.InAddr) || !e.In6Addr.Equal(o.In6Addr) {
return false
}
if e.Action != o.Action || e.Table != o.Table || e.Iif != o.Iif || e.Oif != o.Oif {
return false
}
return true
}
type Via struct {
AddrFamily int
Addr net.IP
}
func (v *Via) Equal(x Destination) bool {
o, ok := x.(*Via)
if !ok {
return false
}
if v.AddrFamily == x.Family() && v.Addr.Equal(o.Addr) {
return true
}
return false
}
func (v *Via) String() string {
return fmt.Sprintf("Family: %d, Address: %s", v.AddrFamily, v.Addr.String())
}
func (v *Via) Family() int {
return v.AddrFamily
}
func (v *Via) Encode() ([]byte, error) {
buf := &bytes.Buffer{}
err := binary.Write(buf, native, uint16(v.AddrFamily))
if err != nil {
return nil, err
}
err = binary.Write(buf, native, v.Addr)
if err != nil {
return nil, err
}
return buf.Bytes(), nil
}
func (v *Via) Decode(b []byte) error {
native := nl.NativeEndian()
if len(b) < 6 {
return fmt.Errorf("decoding failed: buffer too small (%d bytes)", len(b))
}
v.AddrFamily = int(native.Uint16(b[0:2]))
if v.AddrFamily == nl.FAMILY_V4 {
v.Addr = net.IP(b[2:6])
return nil
} else if v.AddrFamily == nl.FAMILY_V6 {
if len(b) < 18 {
return fmt.Errorf("decoding failed: buffer too small (%d bytes)", len(b))
}
v.Addr = net.IP(b[2:])
return nil
}
return fmt.Errorf("decoding failed: address family %d unknown", v.AddrFamily)
}
// RouteAdd will add a route to the system.
// Equivalent to: `ip route add $route`
func RouteAdd(route *Route) error {
@@ -273,6 +536,32 @@ func (h *Handle) RouteAdd(route *Route) error {
return h.routeHandle(route, req, nl.NewRtMsg())
}
// RouteAppend will append a route to the system.
// Equivalent to: `ip route append $route`
func RouteAppend(route *Route) error {
return pkgHandle.RouteAppend(route)
}
// RouteAppend will append a route to the system.
// Equivalent to: `ip route append $route`
func (h *Handle) RouteAppend(route *Route) error {
flags := unix.NLM_F_CREATE | unix.NLM_F_APPEND | unix.NLM_F_ACK
req := h.newNetlinkRequest(unix.RTM_NEWROUTE, flags)
return h.routeHandle(route, req, nl.NewRtMsg())
}
// RouteAddEcmp will add a route to the system.
func RouteAddEcmp(route *Route) error {
return pkgHandle.RouteAddEcmp(route)
}
// RouteAddEcmp will add a route to the system.
func (h *Handle) RouteAddEcmp(route *Route) error {
flags := unix.NLM_F_CREATE | unix.NLM_F_ACK
req := h.newNetlinkRequest(unix.RTM_NEWROUTE, flags)
return h.routeHandle(route, req, nl.NewRtMsg())
}
// RouteReplace will add a route to the system.
// Equivalent to: `ip route replace $route`
func RouteReplace(route *Route) error {
@@ -335,18 +624,18 @@ func (h *Handle) routeHandle(route *Route, req *nl.NetlinkRequest, msg *nl.RtMsg
if err != nil {
return err
}
rtAttrs = append(rtAttrs, nl.NewRtAttr(nl.RTA_NEWDST, buf))
rtAttrs = append(rtAttrs, nl.NewRtAttr(unix.RTA_NEWDST, buf))
}
if route.Encap != nil {
buf := make([]byte, 2)
native.PutUint16(buf, uint16(route.Encap.Type()))
rtAttrs = append(rtAttrs, nl.NewRtAttr(nl.RTA_ENCAP_TYPE, buf))
rtAttrs = append(rtAttrs, nl.NewRtAttr(unix.RTA_ENCAP_TYPE, buf))
buf, err := route.Encap.Encode()
if err != nil {
return err
}
rtAttrs = append(rtAttrs, nl.NewRtAttr(nl.RTA_ENCAP, buf))
rtAttrs = append(rtAttrs, nl.NewRtAttr(unix.RTA_ENCAP, buf))
}
if route.Src != nil {
@@ -380,6 +669,14 @@ func (h *Handle) routeHandle(route *Route, req *nl.NetlinkRequest, msg *nl.RtMsg
rtAttrs = append(rtAttrs, nl.NewRtAttr(unix.RTA_GATEWAY, gwData))
}
if route.Via != nil {
buf, err := route.Via.Encode()
if err != nil {
return fmt.Errorf("failed to encode RTA_VIA: %v", err)
}
rtAttrs = append(rtAttrs, nl.NewRtAttr(unix.RTA_VIA, buf))
}
if len(route.MultiPath) > 0 {
buf := []byte{}
for _, nh := range route.MultiPath {
@@ -410,17 +707,24 @@ func (h *Handle) routeHandle(route *Route, req *nl.NetlinkRequest, msg *nl.RtMsg
if err != nil {
return err
}
children = append(children, nl.NewRtAttr(nl.RTA_NEWDST, buf))
children = append(children, nl.NewRtAttr(unix.RTA_NEWDST, buf))
}
if nh.Encap != nil {
buf := make([]byte, 2)
native.PutUint16(buf, uint16(nh.Encap.Type()))
rtAttrs = append(rtAttrs, nl.NewRtAttr(nl.RTA_ENCAP_TYPE, buf))
children = append(children, nl.NewRtAttr(unix.RTA_ENCAP_TYPE, buf))
buf, err := nh.Encap.Encode()
if err != nil {
return err
}
children = append(children, nl.NewRtAttr(nl.RTA_ENCAP, buf))
children = append(children, nl.NewRtAttr(unix.RTA_ENCAP, buf))
}
if nh.Via != nil {
buf, err := nh.Via.Encode()
if err != nil {
return err
}
children = append(children, nl.NewRtAttr(unix.RTA_VIA, buf))
}
rtnh.Children = children
buf = append(buf, rtnh.Serialize()...)
@@ -455,15 +759,70 @@ func (h *Handle) routeHandle(route *Route, req *nl.NetlinkRequest, msg *nl.RtMsg
}
var metrics []*nl.RtAttr
// TODO: support other rta_metric values
if route.MTU > 0 {
b := nl.Uint32Attr(uint32(route.MTU))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_MTU, b))
}
if route.Window > 0 {
b := nl.Uint32Attr(uint32(route.Window))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_WINDOW, b))
}
if route.Rtt > 0 {
b := nl.Uint32Attr(uint32(route.Rtt))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_RTT, b))
}
if route.RttVar > 0 {
b := nl.Uint32Attr(uint32(route.RttVar))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_RTTVAR, b))
}
if route.Ssthresh > 0 {
b := nl.Uint32Attr(uint32(route.Ssthresh))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_SSTHRESH, b))
}
if route.Cwnd > 0 {
b := nl.Uint32Attr(uint32(route.Cwnd))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_CWND, b))
}
if route.AdvMSS > 0 {
b := nl.Uint32Attr(uint32(route.AdvMSS))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_ADVMSS, b))
}
if route.Reordering > 0 {
b := nl.Uint32Attr(uint32(route.Reordering))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_REORDERING, b))
}
if route.Hoplimit > 0 {
b := nl.Uint32Attr(uint32(route.Hoplimit))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_HOPLIMIT, b))
}
if route.InitCwnd > 0 {
b := nl.Uint32Attr(uint32(route.InitCwnd))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_INITCWND, b))
}
if route.Features > 0 {
b := nl.Uint32Attr(uint32(route.Features))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_FEATURES, b))
}
if route.RtoMin > 0 {
b := nl.Uint32Attr(uint32(route.RtoMin))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_RTO_MIN, b))
}
if route.InitRwnd > 0 {
b := nl.Uint32Attr(uint32(route.InitRwnd))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_INITRWND, b))
}
if route.QuickACK > 0 {
b := nl.Uint32Attr(uint32(route.QuickACK))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_QUICKACK, b))
}
if route.Congctl != "" {
b := nl.ZeroTerminated(route.Congctl)
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_CC_ALGO, b))
}
if route.FastOpenNoCookie > 0 {
b := nl.Uint32Attr(uint32(route.FastOpenNoCookie))
metrics = append(metrics, nl.NewRtAttr(unix.RTAX_FASTOPEN_NO_COOKIE, b))
}
if metrics != nil {
attr := nl.NewRtAttr(unix.RTA_METRICS, nil)
@@ -574,6 +933,8 @@ func (h *Handle) RouteListFiltered(family int, filter *Route, filterMask uint64)
continue
}
}
case filterMask&RT_FILTER_HOPLIMIT != 0 && route.Hoplimit != filter.Hoplimit:
continue
}
}
res = append(res, route)
@@ -590,7 +951,7 @@ func deserializeRoute(m []byte) (Route, error) {
}
route := Route{
Scope: Scope(msg.Scope),
Protocol: int(msg.Protocol),
Protocol: RouteProtocol(int(msg.Protocol)),
Table: int(msg.Table),
Type: int(msg.Type),
Tos: int(msg.Tos),
@@ -649,7 +1010,7 @@ func deserializeRoute(m []byte) (Route, error) {
switch attr.Attr.Type {
case unix.RTA_GATEWAY:
info.Gw = net.IP(attr.Value)
case nl.RTA_NEWDST:
case unix.RTA_NEWDST:
var d Destination
switch msg.Family {
case nl.FAMILY_MPLS:
@@ -659,10 +1020,16 @@ func deserializeRoute(m []byte) (Route, error) {
return nil, nil, err
}
info.NewDst = d
case nl.RTA_ENCAP_TYPE:
case unix.RTA_ENCAP_TYPE:
encapType = attr
case nl.RTA_ENCAP:
case unix.RTA_ENCAP:
encap = attr
case unix.RTA_VIA:
d := &Via{}
if err := d.Decode(attr.Value); err != nil {
return nil, nil, err
}
info.Via = d
}
}
@@ -690,7 +1057,7 @@ func deserializeRoute(m []byte) (Route, error) {
route.MultiPath = append(route.MultiPath, info)
rest = buf
}
case nl.RTA_NEWDST:
case unix.RTA_NEWDST:
var d Destination
switch msg.Family {
case nl.FAMILY_MPLS:
@@ -700,9 +1067,15 @@ func deserializeRoute(m []byte) (Route, error) {
return route, err
}
route.NewDst = d
case nl.RTA_ENCAP_TYPE:
case unix.RTA_VIA:
v := &Via{}
if err := v.Decode(attr.Value); err != nil {
return route, err
}
route.Via = v
case unix.RTA_ENCAP_TYPE:
encapType = attr
case nl.RTA_ENCAP:
case unix.RTA_ENCAP:
encap = attr
case unix.RTA_METRICS:
metrics, err := nl.ParseRouteAttr(attr.Value)
@@ -713,8 +1086,36 @@ func deserializeRoute(m []byte) (Route, error) {
switch metric.Attr.Type {
case unix.RTAX_MTU:
route.MTU = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_WINDOW:
route.Window = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_RTT:
route.Rtt = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_RTTVAR:
route.RttVar = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_SSTHRESH:
route.Ssthresh = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_CWND:
route.Cwnd = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_ADVMSS:
route.AdvMSS = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_REORDERING:
route.Reordering = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_HOPLIMIT:
route.Hoplimit = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_INITCWND:
route.InitCwnd = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_FEATURES:
route.Features = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_RTO_MIN:
route.RtoMin = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_INITRWND:
route.InitRwnd = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_QUICKACK:
route.QuickACK = int(native.Uint32(metric.Value[0:4]))
case unix.RTAX_CC_ALGO:
route.Congctl = nl.BytesToString(metric.Value)
case unix.RTAX_FASTOPEN_NO_COOKIE:
route.FastOpenNoCookie = int(native.Uint32(metric.Value[0:4]))
}
}
}
@@ -734,6 +1135,11 @@ func deserializeRoute(m []byte) (Route, error) {
if err := e.Decode(encap.Value); err != nil {
return route, err
}
case nl.LWTUNNEL_ENCAP_SEG6_LOCAL:
e = &SEG6LocalEncap{}
if err := e.Decode(encap.Value); err != nil {
return route, err
}
}
route.Encap = e
}
@@ -741,15 +1147,28 @@ func deserializeRoute(m []byte) (Route, error) {
return route, nil
}
// RouteGetOptions contains a set of options to use with
// RouteGetWithOptions
type RouteGetOptions struct {
VrfName string
SrcAddr net.IP
}
// RouteGetWithOptions gets a route to a specific destination from the host system.
// Equivalent to: 'ip route get <> vrf <VrfName>'.
func RouteGetWithOptions(destination net.IP, options *RouteGetOptions) ([]Route, error) {
return pkgHandle.RouteGetWithOptions(destination, options)
}
// RouteGet gets a route to a specific destination from the host system.
// Equivalent to: 'ip route get'.
func RouteGet(destination net.IP) ([]Route, error) {
return pkgHandle.RouteGet(destination)
}
// RouteGet gets a route to a specific destination from the host system.
// Equivalent to: 'ip route get'.
func (h *Handle) RouteGet(destination net.IP) ([]Route, error) {
// RouteGetWithOptions gets a route to a specific destination from the host system.
// Equivalent to: 'ip route get <> vrf <VrfName>'.
func (h *Handle) RouteGetWithOptions(destination net.IP, options *RouteGetOptions) ([]Route, error) {
req := h.newNetlinkRequest(unix.RTM_GETROUTE, unix.NLM_F_REQUEST)
family := nl.GetIPFamily(destination)
var destinationData []byte
@@ -764,11 +1183,42 @@ func (h *Handle) RouteGet(destination net.IP) ([]Route, error) {
msg := &nl.RtMsg{}
msg.Family = uint8(family)
msg.Dst_len = bitlen
if options != nil && options.SrcAddr != nil {
msg.Src_len = bitlen
}
msg.Flags = unix.RTM_F_LOOKUP_TABLE
req.AddData(msg)
rtaDst := nl.NewRtAttr(unix.RTA_DST, destinationData)
req.AddData(rtaDst)
if options != nil {
if options.VrfName != "" {
link, err := LinkByName(options.VrfName)
if err != nil {
return nil, err
}
var (
b = make([]byte, 4)
native = nl.NativeEndian()
)
native.PutUint32(b, uint32(link.Attrs().Index))
req.AddData(nl.NewRtAttr(unix.RTA_OIF, b))
}
if options.SrcAddr != nil {
var srcAddr []byte
if family == FAMILY_V4 {
srcAddr = options.SrcAddr.To4()
} else {
srcAddr = options.SrcAddr.To16()
}
req.AddData(nl.NewRtAttr(unix.RTA_SRC, srcAddr))
}
}
msgs, err := req.Execute(unix.NETLINK_ROUTE, unix.RTM_NEWROUTE)
if err != nil {
return nil, err
@@ -783,7 +1233,12 @@ func (h *Handle) RouteGet(destination net.IP) ([]Route, error) {
res = append(res, route)
}
return res, nil
}
// RouteGet gets a route to a specific destination from the host system.
// Equivalent to: 'ip route get'.
func (h *Handle) RouteGet(destination net.IP) ([]Route, error) {
return h.RouteGetWithOptions(destination, nil)
}
// RouteSubscribe takes a chan down which notifications will be sent
@@ -840,13 +1295,19 @@ func routeSubscribeAt(newNs, curNs netns.NsHandle, ch chan<- RouteUpdate, done <
go func() {
defer close(ch)
for {
msgs, err := s.Receive()
msgs, from, err := s.Receive()
if err != nil {
if cberr != nil {
cberr(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
@@ -876,3 +1337,54 @@ func routeSubscribeAt(newNs, curNs netns.NsHandle, ch chan<- RouteUpdate, done <
return nil
}
func (p RouteProtocol) String() string {
switch int(p) {
case unix.RTPROT_BABEL:
return "babel"
case unix.RTPROT_BGP:
return "bgp"
case unix.RTPROT_BIRD:
return "bird"
case unix.RTPROT_BOOT:
return "boot"
case unix.RTPROT_DHCP:
return "dhcp"
case unix.RTPROT_DNROUTED:
return "dnrouted"
case unix.RTPROT_EIGRP:
return "eigrp"
case unix.RTPROT_GATED:
return "gated"
case unix.RTPROT_ISIS:
return "isis"
//case unix.RTPROT_KEEPALIVED:
// return "keepalived"
case unix.RTPROT_KERNEL:
return "kernel"
case unix.RTPROT_MROUTED:
return "mrouted"
case unix.RTPROT_MRT:
return "mrt"
case unix.RTPROT_NTK:
return "ntk"
case unix.RTPROT_OSPF:
return "ospf"
case unix.RTPROT_RA:
return "ra"
case unix.RTPROT_REDIRECT:
return "redirect"
case unix.RTPROT_RIP:
return "rip"
case unix.RTPROT_STATIC:
return "static"
case unix.RTPROT_UNSPEC:
return "unspec"
case unix.RTPROT_XORP:
return "xorp"
case unix.RTPROT_ZEBRA:
return "zebra"
default:
return strconv.Itoa(int(p))
}
}

View File

@@ -2,6 +2,8 @@
package netlink
import "strconv"
func (r *Route) ListFlags() []string {
return []string{}
}
@@ -9,3 +11,11 @@ func (r *Route) ListFlags() []string {
func (n *NexthopInfo) ListFlags() []string {
return []string{}
}
func (s Scope) String() string {
return "unknown"
}
func (p RouteProtocol) String() string {
return strconv.Itoa(int(p))
}

View File

@@ -12,6 +12,7 @@ type Rule struct {
Table int
Mark int
Mask int
Tos uint
TunID uint
Goto int
Src *net.IPNet
@@ -22,6 +23,8 @@ type Rule struct {
SuppressIfgroup int
SuppressPrefixlen int
Invert bool
Dport *RulePortRange
Sport *RulePortRange
}
func (r Rule) String() string {
@@ -40,3 +43,14 @@ func NewRule() *Rule {
Flow: -1,
}
}
// NewRulePortRange creates rule sport/dport range.
func NewRulePortRange(start, end uint16) *RulePortRange {
return &RulePortRange{Start: start, End: end}
}
// RulePortRange represents rule sport/dport range.
type RulePortRange struct {
Start uint16
End uint16
}

View File

@@ -1,6 +1,7 @@
package netlink
import (
"bytes"
"fmt"
"net"
@@ -55,6 +56,9 @@ func ruleHandle(rule *Rule, req *nl.NetlinkRequest) error {
if rule.Table >= 0 && rule.Table < 256 {
msg.Table = uint8(rule.Table)
}
if rule.Tos != 0 {
msg.Tos = uint8(rule.Tos)
}
var dstFamily uint8
var rtAttrs []*nl.RtAttr
@@ -138,18 +142,28 @@ func ruleHandle(rule *Rule, req *nl.NetlinkRequest) error {
}
}
if rule.IifName != "" {
req.AddData(nl.NewRtAttr(nl.FRA_IIFNAME, []byte(rule.IifName)))
req.AddData(nl.NewRtAttr(nl.FRA_IIFNAME, []byte(rule.IifName+"\x00")))
}
if rule.OifName != "" {
req.AddData(nl.NewRtAttr(nl.FRA_OIFNAME, []byte(rule.OifName)))
req.AddData(nl.NewRtAttr(nl.FRA_OIFNAME, []byte(rule.OifName+"\x00")))
}
if rule.Goto >= 0 {
msg.Type = nl.FR_ACT_NOP
msg.Type = nl.FR_ACT_GOTO
b := make([]byte, 4)
native.PutUint32(b, uint32(rule.Goto))
req.AddData(nl.NewRtAttr(nl.FRA_GOTO, b))
}
if rule.Dport != nil {
b := rule.Dport.toRtAttrData()
req.AddData(nl.NewRtAttr(nl.FRA_DPORT_RANGE, b))
}
if rule.Sport != nil {
b := rule.Sport.toRtAttrData()
req.AddData(nl.NewRtAttr(nl.FRA_SPORT_RANGE, b))
}
_, err := req.Execute(unix.NETLINK_ROUTE, 0)
return err
}
@@ -163,6 +177,19 @@ func RuleList(family int) ([]Rule, error) {
// RuleList lists rules in the system.
// Equivalent to: ip rule list
func (h *Handle) RuleList(family int) ([]Rule, error) {
return h.RuleListFiltered(family, nil, 0)
}
// RuleListFiltered gets a list of rules in the system filtered by the
// specified rule template `filter`.
// Equivalent to: ip rule list
func RuleListFiltered(family int, filter *Rule, filterMask uint64) ([]Rule, error) {
return pkgHandle.RuleListFiltered(family, filter, filterMask)
}
// RuleListFiltered lists rules in the system.
// Equivalent to: ip rule list
func (h *Handle) RuleListFiltered(family int, filter *Rule, filterMask uint64) ([]Rule, error) {
req := h.newNetlinkRequest(unix.RTM_GETRULE, unix.NLM_F_DUMP|unix.NLM_F_REQUEST)
msg := nl.NewIfInfomsg(family)
req.AddData(msg)
@@ -184,6 +211,7 @@ func (h *Handle) RuleList(family int) ([]Rule, error) {
rule := NewRule()
rule.Invert = msg.Flags&FibRuleInvert > 0
rule.Tos = uint(msg.Tos)
for j := range attrs {
switch attrs[j].Attr.Type {
@@ -225,10 +253,44 @@ func (h *Handle) RuleList(family int) ([]Rule, error) {
rule.Goto = int(native.Uint32(attrs[j].Value[0:4]))
case nl.FRA_PRIORITY:
rule.Priority = int(native.Uint32(attrs[j].Value[0:4]))
case nl.FRA_DPORT_RANGE:
rule.Dport = NewRulePortRange(native.Uint16(attrs[j].Value[0:2]), native.Uint16(attrs[j].Value[2:4]))
case nl.FRA_SPORT_RANGE:
rule.Sport = NewRulePortRange(native.Uint16(attrs[j].Value[0:2]), native.Uint16(attrs[j].Value[2:4]))
}
}
if filter != nil {
switch {
case filterMask&RT_FILTER_SRC != 0 &&
(rule.Src == nil || rule.Src.String() != filter.Src.String()):
continue
case filterMask&RT_FILTER_DST != 0 &&
(rule.Dst == nil || rule.Dst.String() != filter.Dst.String()):
continue
case filterMask&RT_FILTER_TABLE != 0 &&
filter.Table != unix.RT_TABLE_UNSPEC && rule.Table != filter.Table:
continue
case filterMask&RT_FILTER_TOS != 0 && rule.Tos != filter.Tos:
continue
case filterMask&RT_FILTER_PRIORITY != 0 && rule.Priority != filter.Priority:
continue
case filterMask&RT_FILTER_MARK != 0 && rule.Mark != filter.Mark:
continue
case filterMask&RT_FILTER_MASK != 0 && rule.Mask != filter.Mask:
continue
}
}
res = append(res, *rule)
}
return res, nil
}
func (pr *RulePortRange) toRtAttrData() []byte {
b := [][]byte{make([]byte, 2), make([]byte, 2)}
native.PutUint16(b[0], pr.Start)
native.PutUint16(b[1], pr.End)
return bytes.Join(b, []byte{})
}

View File

@@ -4,6 +4,7 @@ import (
"errors"
"fmt"
"net"
"syscall"
"github.com/vishvananda/netlink/nl"
"golang.org/x/sys/unix"
@@ -49,10 +50,15 @@ func (r *socketRequest) Serialize() []byte {
native.PutUint32(b.Next(4), r.States)
networkOrder.PutUint16(b.Next(2), r.ID.SourcePort)
networkOrder.PutUint16(b.Next(2), r.ID.DestinationPort)
copy(b.Next(4), r.ID.Source.To4())
b.Next(12)
copy(b.Next(4), r.ID.Destination.To4())
b.Next(12)
if r.Family == unix.AF_INET6 {
copy(b.Next(16), r.ID.Source)
copy(b.Next(16), r.ID.Destination)
} else {
copy(b.Next(4), r.ID.Source.To4())
b.Next(12)
copy(b.Next(4), r.ID.Destination.To4())
b.Next(12)
}
native.PutUint32(b.Next(4), r.ID.Interface)
native.PutUint32(b.Next(4), r.ID.Cookie[0])
native.PutUint32(b.Next(4), r.ID.Cookie[1])
@@ -89,10 +95,15 @@ func (s *Socket) deserialize(b []byte) error {
s.Retrans = rb.Read()
s.ID.SourcePort = networkOrder.Uint16(rb.Next(2))
s.ID.DestinationPort = networkOrder.Uint16(rb.Next(2))
s.ID.Source = net.IPv4(rb.Read(), rb.Read(), rb.Read(), rb.Read())
rb.Next(12)
s.ID.Destination = net.IPv4(rb.Read(), rb.Read(), rb.Read(), rb.Read())
rb.Next(12)
if s.Family == unix.AF_INET6 {
s.ID.Source = net.IP(rb.Next(16))
s.ID.Destination = net.IP(rb.Next(16))
} else {
s.ID.Source = net.IPv4(rb.Read(), rb.Read(), rb.Read(), rb.Read())
rb.Next(12)
s.ID.Destination = net.IPv4(rb.Read(), rb.Read(), rb.Read(), rb.Read())
rb.Next(12)
}
s.ID.Interface = native.Uint32(rb.Next(4))
s.ID.Cookie[0] = native.Uint32(rb.Next(4))
s.ID.Cookie[1] = native.Uint32(rb.Next(4))
@@ -141,10 +152,13 @@ func SocketGet(local, remote net.Addr) (*Socket, error) {
},
})
s.Send(req)
msgs, err := s.Receive()
msgs, from, err := s.Receive()
if err != nil {
return nil, err
}
if from.Pid != nl.PidKernel {
return nil, fmt.Errorf("Wrong sender portid %d, expected %d", from.Pid, nl.PidKernel)
}
if len(msgs) == 0 {
return nil, errors.New("no message nor error from netlink")
}
@@ -157,3 +171,91 @@ func SocketGet(local, remote net.Addr) (*Socket, error) {
}
return sock, nil
}
// SocketDiagTCPInfo requests INET_DIAG_INFO for TCP protocol for specified family type.
func SocketDiagTCPInfo(family uint8) ([]*InetDiagTCPInfoResp, error) {
s, err := nl.Subscribe(unix.NETLINK_INET_DIAG)
if err != nil {
return nil, err
}
defer s.Close()
req := nl.NewNetlinkRequest(nl.SOCK_DIAG_BY_FAMILY, unix.NLM_F_DUMP)
req.AddData(&socketRequest{
Family: family,
Protocol: unix.IPPROTO_TCP,
Ext: (1 << (INET_DIAG_VEGASINFO - 1)) | (1 << (INET_DIAG_INFO - 1)),
States: uint32(0xfff), // All TCP states
})
s.Send(req)
var result []*InetDiagTCPInfoResp
loop:
for {
msgs, from, err := s.Receive()
if err != nil {
return nil, err
}
if from.Pid != nl.PidKernel {
return nil, fmt.Errorf("Wrong sender portid %d, expected %d", from.Pid, nl.PidKernel)
}
if len(msgs) == 0 {
return nil, errors.New("no message nor error from netlink")
}
for _, m := range msgs {
switch m.Header.Type {
case unix.NLMSG_DONE:
break loop
case unix.NLMSG_ERROR:
native := nl.NativeEndian()
error := int32(native.Uint32(m.Data[0:4]))
return nil, syscall.Errno(-error)
}
sockInfo := &Socket{}
if err := sockInfo.deserialize(m.Data); err != nil {
return nil, err
}
attrs, err := nl.ParseRouteAttr(m.Data[sizeofSocket:])
if err != nil {
return nil, err
}
res, err := attrsToInetDiagTCPInfoResp(attrs, sockInfo)
if err != nil {
return nil, err
}
result = append(result, res)
}
}
return result, nil
}
func attrsToInetDiagTCPInfoResp(attrs []syscall.NetlinkRouteAttr, sockInfo *Socket) (*InetDiagTCPInfoResp, error) {
var tcpInfo *TCPInfo
var tcpBBRInfo *TCPBBRInfo
for _, a := range attrs {
if a.Attr.Type == INET_DIAG_INFO {
tcpInfo = &TCPInfo{}
if err := tcpInfo.deserialize(a.Value); err != nil {
return nil, err
}
continue
}
if a.Attr.Type == INET_DIAG_BBRINFO {
tcpBBRInfo = &TCPBBRInfo{}
if err := tcpBBRInfo.deserialize(a.Value); err != nil {
return nil, err
}
continue
}
}
return &InetDiagTCPInfoResp{
InetDiagMsg: sockInfo,
TCPInfo: tcpInfo,
TCPBBRInfo: tcpBBRInfo,
}, nil
}

84
vendor/github.com/vishvananda/netlink/tcp.go generated vendored Normal file
View File

@@ -0,0 +1,84 @@
package netlink
// TCP States
const (
TCP_ESTABLISHED = iota + 0x01
TCP_SYN_SENT
TCP_SYN_RECV
TCP_FIN_WAIT1
TCP_FIN_WAIT2
TCP_TIME_WAIT
TCP_CLOSE
TCP_CLOSE_WAIT
TCP_LAST_ACK
TCP_LISTEN
TCP_CLOSING
TCP_NEW_SYN_REC
TCP_MAX_STATES
)
type TCPInfo struct {
State uint8
Ca_state uint8
Retransmits uint8
Probes uint8
Backoff uint8
Options uint8
Snd_wscale uint8 // no uint4
Rcv_wscale uint8
Delivery_rate_app_limited uint8
Fastopen_client_fail uint8
Rto uint32
Ato uint32
Snd_mss uint32
Rcv_mss uint32
Unacked uint32
Sacked uint32
Lost uint32
Retrans uint32
Fackets uint32
Last_data_sent uint32
Last_ack_sent uint32
Last_data_recv uint32
Last_ack_recv uint32
Pmtu uint32
Rcv_ssthresh uint32
Rtt uint32
Rttvar uint32
Snd_ssthresh uint32
Snd_cwnd uint32
Advmss uint32
Reordering uint32
Rcv_rtt uint32
Rcv_space uint32
Total_retrans uint32
Pacing_rate uint64
Max_pacing_rate uint64
Bytes_acked uint64 /* RFC4898 tcpEStatsAppHCThruOctetsAcked */
Bytes_received uint64 /* RFC4898 tcpEStatsAppHCThruOctetsReceived */
Segs_out uint32 /* RFC4898 tcpEStatsPerfSegsOut */
Segs_in uint32 /* RFC4898 tcpEStatsPerfSegsIn */
Notsent_bytes uint32
Min_rtt uint32
Data_segs_in uint32 /* RFC4898 tcpEStatsDataSegsIn */
Data_segs_out uint32 /* RFC4898 tcpEStatsDataSegsOut */
Delivery_rate uint64
Busy_time uint64 /* Time (usec) busy sending data */
Rwnd_limited uint64 /* Time (usec) limited by receive window */
Sndbuf_limited uint64 /* Time (usec) limited by send buffer */
Delivered uint32
Delivered_ce uint32
Bytes_sent uint64 /* RFC4898 tcpEStatsPerfHCDataOctetsOut */
Bytes_retrans uint64 /* RFC4898 tcpEStatsPerfOctetsRetrans */
Dsack_dups uint32 /* RFC4898 tcpEStatsStackDSACKDups */
Reord_seen uint32 /* reordering events seen */
Rcv_ooopack uint32 /* Out-of-order packets received */
Snd_wnd uint32 /* peer's advertised receive window after * scaling (bytes) */
}
type TCPBBRInfo struct {
BBRBW uint64
BBRMinRTT uint32
BBRPacingGain uint32
BBRCwndGain uint32
}

353
vendor/github.com/vishvananda/netlink/tcp_linux.go generated vendored Normal file
View File

@@ -0,0 +1,353 @@
package netlink
import (
"bytes"
"errors"
"io"
)
const (
tcpBBRInfoLen = 20
)
func checkDeserErr(err error) error {
if err == io.EOF {
return nil
}
return err
}
func (t *TCPInfo) deserialize(b []byte) error {
var err error
rb := bytes.NewBuffer(b)
t.State, err = rb.ReadByte()
if err != nil {
return checkDeserErr(err)
}
t.Ca_state, err = rb.ReadByte()
if err != nil {
return checkDeserErr(err)
}
t.Retransmits, err = rb.ReadByte()
if err != nil {
return checkDeserErr(err)
}
t.Probes, err = rb.ReadByte()
if err != nil {
return checkDeserErr(err)
}
t.Backoff, err = rb.ReadByte()
if err != nil {
return checkDeserErr(err)
}
t.Options, err = rb.ReadByte()
if err != nil {
return checkDeserErr(err)
}
scales, err := rb.ReadByte()
if err != nil {
return checkDeserErr(err)
}
t.Snd_wscale = scales >> 4 // first 4 bits
t.Rcv_wscale = scales & 0xf // last 4 bits
rateLimAndFastOpen, err := rb.ReadByte()
if err != nil {
return checkDeserErr(err)
}
t.Delivery_rate_app_limited = rateLimAndFastOpen >> 7 // get first bit
t.Fastopen_client_fail = rateLimAndFastOpen >> 5 & 3 // get next two bits
next := rb.Next(4)
if len(next) == 0 {
return nil
}
t.Rto = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Ato = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Snd_mss = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Rcv_mss = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Unacked = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Sacked = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Lost = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Retrans = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Fackets = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Last_data_sent = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Last_ack_sent = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Last_data_recv = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Last_ack_recv = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Pmtu = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Rcv_ssthresh = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Rtt = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Rttvar = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Snd_ssthresh = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Snd_cwnd = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Advmss = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Reordering = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Rcv_rtt = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Rcv_space = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Total_retrans = native.Uint32(next)
next = rb.Next(8)
if len(next) == 0 {
return nil
}
t.Pacing_rate = native.Uint64(next)
next = rb.Next(8)
if len(next) == 0 {
return nil
}
t.Max_pacing_rate = native.Uint64(next)
next = rb.Next(8)
if len(next) == 0 {
return nil
}
t.Bytes_acked = native.Uint64(next)
next = rb.Next(8)
if len(next) == 0 {
return nil
}
t.Bytes_received = native.Uint64(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Segs_out = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Segs_in = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Notsent_bytes = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Min_rtt = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Data_segs_in = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Data_segs_out = native.Uint32(next)
next = rb.Next(8)
if len(next) == 0 {
return nil
}
t.Delivery_rate = native.Uint64(next)
next = rb.Next(8)
if len(next) == 0 {
return nil
}
t.Busy_time = native.Uint64(next)
next = rb.Next(8)
if len(next) == 0 {
return nil
}
t.Rwnd_limited = native.Uint64(next)
next = rb.Next(8)
if len(next) == 0 {
return nil
}
t.Sndbuf_limited = native.Uint64(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Delivered = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Delivered_ce = native.Uint32(next)
next = rb.Next(8)
if len(next) == 0 {
return nil
}
t.Bytes_sent = native.Uint64(next)
next = rb.Next(8)
if len(next) == 0 {
return nil
}
t.Bytes_retrans = native.Uint64(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Dsack_dups = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Reord_seen = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Rcv_ooopack = native.Uint32(next)
next = rb.Next(4)
if len(next) == 0 {
return nil
}
t.Snd_wnd = native.Uint32(next)
return nil
}
func (t *TCPBBRInfo) deserialize(b []byte) error {
if len(b) != tcpBBRInfoLen {
return errors.New("Invalid length")
}
rb := bytes.NewBuffer(b)
t.BBRBW = native.Uint64(rb.Next(8))
t.BBRMinRTT = native.Uint32(rb.Next(4))
t.BBRPacingGain = native.Uint32(rb.Next(4))
t.BBRCwndGain = native.Uint32(rb.Next(4))
return nil
}

View File

@@ -54,11 +54,15 @@ func XfrmMonitor(ch chan<- XfrmMsg, done <-chan struct{}, errorChan chan<- error
go func() {
defer close(ch)
for {
msgs, err := s.Receive()
msgs, from, err := s.Receive()
if err != nil {
errorChan <- err
return
}
if from.Pid != nl.PidKernel {
errorChan <- fmt.Errorf("Wrong sender portid %d, expected %d", from.Pid, nl.PidKernel)
return
}
for _, m := range msgs {
switch m.Header.Type {
case nl.XFRM_MSG_EXPIRE:

View File

@@ -35,16 +35,36 @@ func (d Dir) String() string {
return fmt.Sprintf("socket %d", d-XFRM_SOCKET_IN)
}
// PolicyAction is an enum representing an ipsec policy action.
type PolicyAction uint8
const (
XFRM_POLICY_ALLOW PolicyAction = 0
XFRM_POLICY_BLOCK PolicyAction = 1
)
func (a PolicyAction) String() string {
switch a {
case XFRM_POLICY_ALLOW:
return "allow"
case XFRM_POLICY_BLOCK:
return "block"
default:
return fmt.Sprintf("action %d", a)
}
}
// XfrmPolicyTmpl encapsulates a rule for the base addresses of an ipsec
// policy. These rules are matched with XfrmState to determine encryption
// and authentication algorithms.
type XfrmPolicyTmpl struct {
Dst net.IP
Src net.IP
Proto Proto
Mode Mode
Spi int
Reqid int
Dst net.IP
Src net.IP
Proto Proto
Mode Mode
Spi int
Reqid int
Optional int
}
func (t XfrmPolicyTmpl) String() string {
@@ -64,11 +84,14 @@ type XfrmPolicy struct {
Dir Dir
Priority int
Index int
Action PolicyAction
Ifindex int
Ifid int
Mark *XfrmMark
Tmpls []XfrmPolicyTmpl
}
func (p XfrmPolicy) String() string {
return fmt.Sprintf("{Dst: %v, Src: %v, Proto: %s, DstPort: %d, SrcPort: %d, Dir: %s, Priority: %d, Index: %d, Mark: %s, Tmpls: %s}",
p.Dst, p.Src, p.Proto, p.DstPort, p.SrcPort, p.Dir, p.Priority, p.Index, p.Mark, p.Tmpls)
return fmt.Sprintf("{Dst: %v, Src: %v, Proto: %s, DstPort: %d, SrcPort: %d, Dir: %s, Priority: %d, Index: %d, Action: %s, Ifindex: %d, Ifid: %d, Mark: %s, Tmpls: %s}",
p.Dst, p.Src, p.Proto, p.DstPort, p.SrcPort, p.Dir, p.Priority, p.Index, p.Action, p.Ifindex, p.Ifid, p.Mark, p.Tmpls)
}

View File

@@ -27,6 +27,7 @@ func selFromPolicy(sel *nl.XfrmSelector, policy *XfrmPolicy) {
if sel.Sport != 0 {
sel.SportMask = ^uint16(0)
}
sel.Ifindex = int32(policy.Ifindex)
}
// XfrmPolicyAdd will add an xfrm policy to the system.
@@ -61,6 +62,7 @@ func (h *Handle) xfrmPolicyAddOrUpdate(policy *XfrmPolicy, nlProto int) error {
msg.Priority = uint32(policy.Priority)
msg.Index = uint32(policy.Index)
msg.Dir = uint8(policy.Dir)
msg.Action = uint8(policy.Action)
msg.Lft.SoftByteLimit = nl.XFRM_INF
msg.Lft.HardByteLimit = nl.XFRM_INF
msg.Lft.SoftPacketLimit = nl.XFRM_INF
@@ -77,6 +79,7 @@ func (h *Handle) xfrmPolicyAddOrUpdate(policy *XfrmPolicy, nlProto int) error {
userTmpl.XfrmId.Spi = nl.Swap32(uint32(tmpl.Spi))
userTmpl.Mode = uint8(tmpl.Mode)
userTmpl.Reqid = uint32(tmpl.Reqid)
userTmpl.Optional = uint8(tmpl.Optional)
userTmpl.Aalgos = ^uint32(0)
userTmpl.Ealgos = ^uint32(0)
userTmpl.Calgos = ^uint32(0)
@@ -90,6 +93,9 @@ func (h *Handle) xfrmPolicyAddOrUpdate(policy *XfrmPolicy, nlProto int) error {
req.AddData(out)
}
ifId := nl.NewRtAttr(nl.XFRMA_IF_ID, nl.Uint32Attr(uint32(policy.Ifid)))
req.AddData(ifId)
_, err := req.Execute(unix.NETLINK_XFRM, 0)
return err
}
@@ -183,6 +189,9 @@ func (h *Handle) xfrmPolicyGetOrDelete(policy *XfrmPolicy, nlProto int) (*XfrmPo
req.AddData(out)
}
ifId := nl.NewRtAttr(nl.XFRMA_IF_ID, nl.Uint32Attr(uint32(policy.Ifid)))
req.AddData(ifId)
resType := nl.XFRM_MSG_NEWPOLICY
if nlProto == nl.XFRM_MSG_DELPOLICY {
resType = 0
@@ -197,12 +206,7 @@ func (h *Handle) xfrmPolicyGetOrDelete(policy *XfrmPolicy, nlProto int) (*XfrmPo
return nil, err
}
p, err := parseXfrmPolicy(msgs[0], FAMILY_ALL)
if err != nil {
return nil, err
}
return p, nil
return parseXfrmPolicy(msgs[0], FAMILY_ALL)
}
func parseXfrmPolicy(m []byte, family int) (*XfrmPolicy, error) {
@@ -220,9 +224,11 @@ func parseXfrmPolicy(m []byte, family int) (*XfrmPolicy, error) {
policy.Proto = Proto(msg.Sel.Proto)
policy.DstPort = int(nl.Swap16(msg.Sel.Dport))
policy.SrcPort = int(nl.Swap16(msg.Sel.Sport))
policy.Ifindex = int(msg.Sel.Ifindex)
policy.Priority = int(msg.Priority)
policy.Index = int(msg.Index)
policy.Dir = Dir(msg.Dir)
policy.Action = PolicyAction(msg.Action)
attrs, err := nl.ParseRouteAttr(m[msg.Len():])
if err != nil {
@@ -242,6 +248,7 @@ func parseXfrmPolicy(m []byte, family int) (*XfrmPolicy, error) {
resTmpl.Mode = Mode(tmpl.Mode)
resTmpl.Spi = int(nl.Swap32(tmpl.XfrmId.Spi))
resTmpl.Reqid = int(tmpl.Reqid)
resTmpl.Optional = int(tmpl.Optional)
policy.Tmpls = append(policy.Tmpls, resTmpl)
}
case nl.XFRMA_MARK:
@@ -249,6 +256,8 @@ func parseXfrmPolicy(m []byte, family int) (*XfrmPolicy, error) {
policy.Mark = new(XfrmMark)
policy.Mark.Value = mark.Value
policy.Mark.Mask = mark.Mask
case nl.XFRMA_IF_ID:
policy.Ifid = int(native.Uint32(attr.Value))
}
}

View File

@@ -94,6 +94,8 @@ type XfrmState struct {
Limits XfrmStateLimits
Statistics XfrmStateStats
Mark *XfrmMark
OutputMark *XfrmMark
Ifid int
Auth *XfrmStateAlgo
Crypt *XfrmStateAlgo
Aead *XfrmStateAlgo
@@ -102,8 +104,8 @@ type XfrmState struct {
}
func (sa XfrmState) String() string {
return fmt.Sprintf("Dst: %v, Src: %v, Proto: %s, Mode: %s, SPI: 0x%x, ReqID: 0x%x, ReplayWindow: %d, Mark: %v, Auth: %v, Crypt: %v, Aead: %v, Encap: %v, ESN: %t",
sa.Dst, sa.Src, sa.Proto, sa.Mode, sa.Spi, sa.Reqid, sa.ReplayWindow, sa.Mark, sa.Auth, sa.Crypt, sa.Aead, sa.Encap, sa.ESN)
return fmt.Sprintf("Dst: %v, Src: %v, Proto: %s, Mode: %s, SPI: 0x%x, ReqID: 0x%x, ReplayWindow: %d, Mark: %v, OutputMark: %v, Ifid: %d, Auth: %v, Crypt: %v, Aead: %v, Encap: %v, ESN: %t",
sa.Dst, sa.Src, sa.Proto, sa.Mode, sa.Spi, sa.Reqid, sa.ReplayWindow, sa.Mark, sa.OutputMark, sa.Ifid, sa.Auth, sa.Crypt, sa.Aead, sa.Encap, sa.ESN)
}
func (sa XfrmState) Print(stats bool) string {
if !stats {

View File

@@ -158,6 +158,17 @@ func (h *Handle) xfrmStateAddOrUpdate(state *XfrmState, nlProto int) error {
out := nl.NewRtAttr(nl.XFRMA_REPLAY_ESN_VAL, writeReplayEsn(state.ReplayWindow))
req.AddData(out)
}
if state.OutputMark != nil {
out := nl.NewRtAttr(nl.XFRMA_SET_MARK, nl.Uint32Attr(state.OutputMark.Value))
req.AddData(out)
if state.OutputMark.Mask != 0 {
out = nl.NewRtAttr(nl.XFRMA_SET_MARK_MASK, nl.Uint32Attr(state.OutputMark.Mask))
req.AddData(out)
}
}
ifId := nl.NewRtAttr(nl.XFRMA_IF_ID, nl.Uint32Attr(uint32(state.Ifid)))
req.AddData(ifId)
_, err := req.Execute(unix.NETLINK_XFRM, 0)
return err
@@ -184,12 +195,7 @@ func (h *Handle) xfrmStateAllocSpi(state *XfrmState) (*XfrmState, error) {
return nil, err
}
s, err := parseXfrmState(msgs[0], FAMILY_ALL)
if err != nil {
return nil, err
}
return s, err
return parseXfrmState(msgs[0], FAMILY_ALL)
}
// XfrmStateDel will delete an xfrm state from the system. Note that
@@ -275,6 +281,9 @@ func (h *Handle) xfrmStateGetOrDelete(state *XfrmState, nlProto int) (*XfrmState
req.AddData(out)
}
ifId := nl.NewRtAttr(nl.XFRMA_IF_ID, nl.Uint32Attr(uint32(state.Ifid)))
req.AddData(ifId)
resType := nl.XFRM_MSG_NEWSA
if nlProto == nl.XFRM_MSG_DELSA {
resType = 0
@@ -372,6 +381,21 @@ func parseXfrmState(m []byte, family int) (*XfrmState, error) {
state.Mark = new(XfrmMark)
state.Mark.Value = mark.Value
state.Mark.Mask = mark.Mask
case nl.XFRMA_SET_MARK:
if state.OutputMark == nil {
state.OutputMark = new(XfrmMark)
}
state.OutputMark.Value = native.Uint32(attr.Value)
case nl.XFRMA_SET_MARK_MASK:
if state.OutputMark == nil {
state.OutputMark = new(XfrmMark)
}
state.OutputMark.Mask = native.Uint32(attr.Value)
if state.OutputMark.Mask == 0xffffffff {
state.OutputMark.Mask = 0
}
case nl.XFRMA_IF_ID:
state.Ifid = int(native.Uint32(attr.Value))
}
}
@@ -394,11 +418,7 @@ func (h *Handle) XfrmStateFlush(proto Proto) error {
req.AddData(&nl.XfrmUsersaFlush{Proto: uint8(proto)})
_, err := req.Execute(unix.NETLINK_XFRM, 0)
if err != nil {
return err
}
return nil
return err
}
func limitsToLft(lmts XfrmStateLimits, lft *nl.XfrmLifetimeCfg) {

View File

@@ -37,7 +37,6 @@ func main() {
// Create a new network namespace
newns, _ := netns.New()
netns.Set(newns)
defer newns.Close()
// Do something with the network namespace
@@ -49,3 +48,14 @@ func main() {
}
```
## NOTE
The library can be safely used only with Go >= 1.10 due to [golang/go#20676](https://github.com/golang/go/issues/20676).
After locking a goroutine to its current OS thread with `runtime.LockOSThread()`
and changing its network namespace, any new subsequent goroutine won't be
scheduled on that thread while it's locked. Therefore, the new goroutine
will run in a different namespace leading to unexpected results.
See [here](https://www.weave.works/blog/linux-namespaces-golang-followup) for more details.

View File

@@ -10,7 +10,8 @@ package netns
import (
"fmt"
"syscall"
"golang.org/x/sys/unix"
)
// NsHandle is a handle to a network namespace. It can be cast directly
@@ -24,11 +25,11 @@ func (ns NsHandle) Equal(other NsHandle) bool {
if ns == other {
return true
}
var s1, s2 syscall.Stat_t
if err := syscall.Fstat(int(ns), &s1); err != nil {
var s1, s2 unix.Stat_t
if err := unix.Fstat(int(ns), &s1); err != nil {
return false
}
if err := syscall.Fstat(int(other), &s2); err != nil {
if err := unix.Fstat(int(other), &s2); err != nil {
return false
}
return (s1.Dev == s2.Dev) && (s1.Ino == s2.Ino)
@@ -36,11 +37,11 @@ func (ns NsHandle) Equal(other NsHandle) bool {
// String shows the file descriptor number and its dev and inode.
func (ns NsHandle) String() string {
var s syscall.Stat_t
if ns == -1 {
return "NS(None)"
}
if err := syscall.Fstat(int(ns), &s); err != nil {
var s unix.Stat_t
if err := unix.Fstat(int(ns), &s); err != nil {
return fmt.Sprintf("NS(%d: unknown)", ns)
}
return fmt.Sprintf("NS(%d: %d, %d)", ns, s.Dev, s.Ino)
@@ -49,11 +50,11 @@ func (ns NsHandle) String() string {
// UniqueId returns a string which uniquely identifies the namespace
// associated with the network handle.
func (ns NsHandle) UniqueId() string {
var s syscall.Stat_t
if ns == -1 {
return "NS(none)"
}
if err := syscall.Fstat(int(ns), &s); err != nil {
var s unix.Stat_t
if err := unix.Fstat(int(ns), &s); err != nil {
return "NS(unknown)"
}
return fmt.Sprintf("NS(%d:%d)", s.Dev, s.Ino)
@@ -67,7 +68,7 @@ func (ns NsHandle) IsOpen() bool {
// Close closes the NsHandle and resets its file descriptor to -1.
// It is not safe to use an NsHandle after Close() is called.
func (ns *NsHandle) Close() error {
if err := syscall.Close(int(*ns)); err != nil {
if err := unix.Close(int(*ns)); err != nil {
return err
}
(*ns) = -1

View File

@@ -1,4 +1,4 @@
// +build linux
// +build linux,go1.10
package netns
@@ -6,44 +6,30 @@ import (
"fmt"
"io/ioutil"
"os"
"path"
"path/filepath"
"runtime"
"strconv"
"strings"
"syscall"
)
// SYS_SETNS syscall allows changing the namespace of the current process.
var SYS_SETNS = map[string]uintptr{
"386": 346,
"amd64": 308,
"arm64": 268,
"arm": 375,
"mips": 4344,
"mipsle": 4344,
"ppc64": 350,
"ppc64le": 350,
"s390x": 339,
}[runtime.GOARCH]
"golang.org/x/sys/unix"
)
// Deprecated: use syscall pkg instead (go >= 1.5 needed).
const (
CLONE_NEWUTS = 0x04000000 /* New utsname group? */
CLONE_NEWIPC = 0x08000000 /* New ipcs */
CLONE_NEWUSER = 0x10000000 /* New user namespace */
CLONE_NEWPID = 0x20000000 /* New pid namespace */
CLONE_NEWNET = 0x40000000 /* New network namespace */
CLONE_IO = 0x80000000 /* Get io context */
CLONE_NEWUTS = 0x04000000 /* New utsname group? */
CLONE_NEWIPC = 0x08000000 /* New ipcs */
CLONE_NEWUSER = 0x10000000 /* New user namespace */
CLONE_NEWPID = 0x20000000 /* New pid namespace */
CLONE_NEWNET = 0x40000000 /* New network namespace */
CLONE_IO = 0x80000000 /* Get io context */
bindMountPath = "/run/netns" /* Bind mount path for named netns */
)
// Setns sets namespace using syscall. Note that this should be a method
// in syscall but it has not been added.
func Setns(ns NsHandle, nstype int) (err error) {
_, _, e1 := syscall.Syscall(SYS_SETNS, uintptr(ns), uintptr(nstype), 0)
if e1 != 0 {
err = e1
}
return
return unix.Setns(int(ns), nstype)
}
// Set sets the current network namespace to the namespace represented
@@ -52,23 +38,67 @@ func Set(ns NsHandle) (err error) {
return Setns(ns, CLONE_NEWNET)
}
// New creates a new network namespace and returns a handle to it.
// New creates a new network namespace, sets it as current and returns
// a handle to it.
func New() (ns NsHandle, err error) {
if err := syscall.Unshare(CLONE_NEWNET); err != nil {
if err := unix.Unshare(CLONE_NEWNET); err != nil {
return -1, err
}
return Get()
}
// NewNamed creates a new named network namespace and returns a handle to it
func NewNamed(name string) (NsHandle, error) {
if _, err := os.Stat(bindMountPath); os.IsNotExist(err) {
err = os.MkdirAll(bindMountPath, 0755)
if err != nil {
return None(), err
}
}
newNs, err := New()
if err != nil {
return None(), err
}
namedPath := path.Join(bindMountPath, name)
f, err := os.OpenFile(namedPath, os.O_CREATE|os.O_EXCL, 0444)
if err != nil {
return None(), err
}
f.Close()
nsPath := fmt.Sprintf("/proc/%d/task/%d/ns/net", os.Getpid(), syscall.Gettid())
err = syscall.Mount(nsPath, namedPath, "bind", syscall.MS_BIND, "")
if err != nil {
return None(), err
}
return newNs, nil
}
// DeleteNamed deletes a named network namespace
func DeleteNamed(name string) error {
namedPath := path.Join(bindMountPath, name)
err := syscall.Unmount(namedPath, syscall.MNT_DETACH)
if err != nil {
return err
}
return os.Remove(namedPath)
}
// Get gets a handle to the current threads network namespace.
func Get() (NsHandle, error) {
return GetFromThread(os.Getpid(), syscall.Gettid())
return GetFromThread(os.Getpid(), unix.Gettid())
}
// GetFromPath gets a handle to a network namespace
// identified by the path
func GetFromPath(path string) (NsHandle, error) {
fd, err := syscall.Open(path, syscall.O_RDONLY, 0)
fd, err := unix.Open(path, unix.O_RDONLY|unix.O_CLOEXEC, 0)
if err != nil {
return -1, err
}
@@ -188,12 +218,18 @@ func getPidForContainer(id string) (int, error) {
filepath.Join(cgroupRoot, "system.slice", "docker-"+id+".scope", "tasks"),
// Even more recent docker versions under cgroup/systemd/docker/<id>/
filepath.Join(cgroupRoot, "..", "systemd", "docker", id, "tasks"),
// Kubernetes with docker and CNI is even more different
// Kubernetes with docker and CNI is even more different. Works for BestEffort and Burstable QoS
filepath.Join(cgroupRoot, "..", "systemd", "kubepods", "*", "pod*", id, "tasks"),
// Another flavor of containers location in recent kubernetes 1.11+
filepath.Join(cgroupRoot, cgroupThis, "kubepods.slice", "kubepods-besteffort.slice", "*", "docker-"+id+".scope", "tasks"),
// When runs inside of a container with recent kubernetes 1.11+
filepath.Join(cgroupRoot, "kubepods.slice", "kubepods-besteffort.slice", "*", "docker-"+id+".scope", "tasks"),
// Same as above but for Guaranteed QoS
filepath.Join(cgroupRoot, "..", "systemd", "kubepods", "pod*", id, "tasks"),
// Another flavor of containers location in recent kubernetes 1.11+. Works for BestEffort and Burstable QoS
filepath.Join(cgroupRoot, cgroupThis, "kubepods.slice", "*.slice", "*", "docker-"+id+".scope", "tasks"),
// Same as above but for Guaranteed QoS
filepath.Join(cgroupRoot, cgroupThis, "kubepods.slice", "*", "docker-"+id+".scope", "tasks"),
// When runs inside of a container with recent kubernetes 1.11+. Works for BestEffort and Burstable QoS
filepath.Join(cgroupRoot, "kubepods.slice", "*.slice", "*", "docker-"+id+".scope", "tasks"),
// Same as above but for Guaranteed QoS
filepath.Join(cgroupRoot, "kubepods.slice", "*", "docker-"+id+".scope", "tasks"),
}
var filename string