netbird/gvisor
0
0
Fork 0
mirror of https://github.com/netbirdio/gvisor.git synced 2026-05-22 17:12:49 -07:00
Code Issues Packages Projects Releases Wiki Activity

Returns ENOBUFS when writing to a veth device whose buffer is full.

Browse Source 
A task is stuck when it keeps sendMsg w/o any recvMsg  until  the number of
message exceeds the buffer limit. The change also significantly bumps the
limit from 64 to 1000 for a larger buffer.

Reported-by: syzbot+8788f5c1ba95e12ab07d@syzkaller.appspotmail.com
Reported-by: syzbot+67759fceac0c30ece991@syzkaller.appspotmail.com
PiperOrigin-RevId: 704904404
This commit is contained in:
Jing Chen
2024-12-10 17:35:54 -08:00
committed by gVisor bot
parent d9fa8c2e8a
commit c88ffa321e
5 changed files with 74 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files
pkg/sentry/socket/netstack/stack.go
+1 -1
View File
@@ -298,7 +298,7 @@ func (s *Stack) newVeth(ctx context.Context, linkAttrs map[uint16]nlmsg.BytesVie
}
}
}
ep, peerEP := veth.NewPair(defaultMTU)
ep, peerEP := veth.NewPair(defaultMTU, veth.DefaultBacklogSize)
id := s.Stack.NextNICID()
peerID := peerStack.Stack.NextNICID()
if ifname == "" {
pkg/tcpip/link/veth/BUILD
+1
View File
@@ -46,6 +46,7 @@ go_test(
deps = [
"//pkg/buffer",
"//pkg/refs",
"//pkg/sync",
"//pkg/tcpip",
"//pkg/tcpip/header",
"//pkg/tcpip/link/ethernet",
pkg/tcpip/link/veth/veth.go
+11 -6
View File
@@ -21,6 +21,9 @@ import (
"gvisor.dev/gvisor/pkg/tcpip/stack"
)
// DefaultBacklogSize is the default size of a veth device's buffer.
const DefaultBacklogSize = 1000
var _ stack.LinkEndpoint = (*Endpoint)(nil)
var _ stack.GSOEndpoint = (*Endpoint)(nil)
@@ -62,8 +65,6 @@ type vethPacket struct {
pkt *stack.PacketBuffer
}
const backlogQueueSize = 64
// Endpoint is link layer endpoint that redirects packets to a pair veth endpoint.
//
// +stateify savable
@@ -84,7 +85,7 @@ type Endpoint struct {
}
// NewPair creates a new veth pair.
func NewPair(mtu uint32) (*Endpoint, *Endpoint) {
func NewPair(mtu, backlogQueueSize uint32) (*Endpoint, *Endpoint) {
veth := veth{
backlogQueue: make(chan vethPacket, backlogQueueSize),
mtu: mtu,
@@ -219,14 +220,18 @@ func (e *Endpoint) WritePackets(pkts stack.PacketBufferList) (int, tcpip.Error)
Payload: payload.DeepClone(),
})
payload.Release()
(e.veth.backlogQueue) <- vethPacket{
select {
case (e.veth.backlogQueue) <- vethPacket{
e: e.peer,
protocol: pkt.NetworkProtocolNumber,
pkt: newPkt,
}:
n++
default:
newPkt.DecRef()
return n, &tcpip.ErrNoBufferSpace{}
}
n++
}
return n, nil
}
pkg/tcpip/link/veth/veth_test.go
+59 -5
View File
@@ -21,6 +21,7 @@ import (
"gvisor.dev/gvisor/pkg/buffer"
"gvisor.dev/gvisor/pkg/refs"
"gvisor.dev/gvisor/pkg/sync"
"gvisor.dev/gvisor/pkg/tcpip"
"gvisor.dev/gvisor/pkg/tcpip/header"
"gvisor.dev/gvisor/pkg/tcpip/link/ethernet"
@@ -30,7 +31,7 @@ import (
func TestSetLinkAddress(t *testing.T) {
addrs := []tcpip.LinkAddress{"abc", "def"}
e, e2 := veth.NewPair(1500)
e, e2 := veth.NewPair(1500, veth.DefaultBacklogSize)
defer e.Close()
defer e2.Close()
for _, addr := range addrs {
@@ -44,11 +45,13 @@ func TestSetLinkAddress(t *testing.T) {
type testNetworkDispatcher struct {
ch chan *stack.PacketBuffer
wg *sync.WaitGroup
}
func (d *testNetworkDispatcher) DeliverNetworkPacket(_ tcpip.NetworkProtocolNumber, pkt *stack.PacketBuffer) {
pkt.IncRef()
d.ch <- pkt
d.wg.Wait()
}
func (*testNetworkDispatcher) DeliverLinkPacket(tcpip.NetworkProtocolNumber, *stack.PacketBuffer) {
@@ -64,7 +67,7 @@ func TestWritePacket(t *testing.T) {
nicID = 5
)
veth1, veth2 := veth.NewPair(1500)
veth1, veth2 := veth.NewPair(1500, veth.DefaultBacklogSize)
veth1.SetLinkAddress(localLinkAddr)
veth2.SetLinkAddress(remoteLinkAddr)
@@ -73,7 +76,8 @@ func TestWritePacket(t *testing.T) {
t.Fatalf("s.CreateNIC(%d, _): %s", nicID, err)
}
sink := &testNetworkDispatcher{ch: make(chan *stack.PacketBuffer, 1)}
var wg sync.WaitGroup
sink := &testNetworkDispatcher{ch: make(chan *stack.PacketBuffer, 1), wg: &wg}
veth2Ethernet := ethernet.New(veth2)
veth2Ethernet.Attach(sink)
@@ -98,13 +102,63 @@ func TestWritePacket(t *testing.T) {
}
}
func TestVethOverflows(t *testing.T) {
const (
localLinkAddr = tcpip.LinkAddress("\x02\x02\x03\x04\x05\x06")
remoteLinkAddr = tcpip.LinkAddress("\x02\x02\x03\x04\x05\x07")
netProto = 55
nicID = 5
)
backlogSize := uint32(1)
veth1, veth2 := veth.NewPair(1500, backlogSize)
veth1.SetLinkAddress(localLinkAddr)
veth2.SetLinkAddress(remoteLinkAddr)
s := stack.New(stack.Options{})
if err := s.CreateNIC(nicID, ethernet.New(veth1)); err != nil {
t.Fatalf("s.CreateNIC(%d, _): %s", nicID, err)
}
var wg sync.WaitGroup
wg.Add(1)
// Use a unbuffered channel in the dispatcher so that the received packet won't be processed.
sink := &testNetworkDispatcher{ch: make(chan *stack.PacketBuffer), wg: &wg}
veth2Ethernet := ethernet.New(veth2)
veth2Ethernet.Attach(sink)
// Send 3 packets, the first packet will be blocked at the sink's channel, the
// second packet will be waiting at the veth device's backlog queue, the third
// packet will be rejected since it overflows the backlog queue.
if err := s.WritePacketToRemote(nicID, remoteLinkAddr, netProto, buffer.Buffer{}); err != nil {
t.Fatalf("s.WritePacketToRemote(%d, %s, _): %s", nicID, remoteLinkAddr, err)
}
select {
case pkt := <-sink.ch:
defer pkt.DecRef()
if err := s.WritePacketToRemote(nicID, remoteLinkAddr, netProto, buffer.Buffer{}); err != nil {
t.Fatalf("s.WritePacketToRemote(%d, %s, _): %s", nicID, remoteLinkAddr, err)
}
if err := s.WritePacketToRemote(nicID, remoteLinkAddr, netProto, buffer.Buffer{}); err == nil {
t.Fatalf("s.WritePacketToRemote(%d, %s, _) got: %v, want: %v", nicID, remoteLinkAddr, err, &tcpip.ErrNoBufferSpace{})
}
wg.Done()
}
pkt := <-sink.ch
if pkt == nil {
t.Fatal("expected to read a packet")
}
pkt.DecRef()
}
func TestDestroyDevices(t *testing.T) {
const (
vethFirstID = 5
vethSecondID = 6
)
veth1, veth2 := veth.NewPair(1500)
veth1, veth2 := veth.NewPair(1500, veth.DefaultBacklogSize)
s1 := stack.New(stack.Options{})
if err := s1.CreateNIC(vethFirstID, ethernet.New(veth1)); err != nil {
@@ -129,7 +183,7 @@ func TestDestroyDevices(t *testing.T) {
func TestMTU(t *testing.T) {
mtus := []uint32{100, 200}
e, e2 := veth.NewPair(1500)
e, e2 := veth.NewPair(1500, veth.DefaultBacklogSize)
defer e.Close()
defer e2.Close()
for _, mtu := range mtus {
pkg/tcpip/stack/bridge_test.go
+2 -2
View File
@@ -116,7 +116,7 @@ func TestWritePacketBetweenDevices(t *testing.T) {
)
// Creates a pair of veth devices which will be attached to different
// network stacks.
veth1, veth2 := veth.NewPair(1500)
veth1, veth2 := veth.NewPair(1500, veth.DefaultBacklogSize)
veth1.SetLinkAddress(vethLinkAddr1)
veth2.SetLinkAddress(vethLinkAddr2)
ch1 := channel.New(1, header.EthernetMinimumSize, channelLinkAddr1)
@@ -199,7 +199,7 @@ func TestBridgeFDB(t *testing.T) {
vethID2 = 8
bridgeID = 9
)
veth1, veth2 := veth.NewPair(1500)
veth1, veth2 := veth.NewPair(1500, veth.DefaultBacklogSize)
ch := channel.New(1, header.EthernetMinimumSize, channelLinkAddr)
bridgeEndpoint := stack.NewBridgeEndpoint(1500)
bridgeEndpoint.SetLinkAddress(bridgeLinkAddr)