netlink.AddrSubscribeWithOptions() was causing systemd-resolved not to obtain DNS servers, and therefore the system was unable to resolve domain names. This ssue started happening on Fedora 42, with systemd-resolved 257 and kernels 6.14.x. Using netlink.AddrSubscribe() solves this issue for now. Closes #1343
162 lines
3.9 KiB
Go
162 lines
3.9 KiB
Go
package ebpf
|
|
|
|
import (
|
|
"syscall"
|
|
"time"
|
|
|
|
"github.com/evilsocket/opensnitch/daemon/core"
|
|
"github.com/evilsocket/opensnitch/daemon/log"
|
|
daemonNetlink "github.com/evilsocket/opensnitch/daemon/netlink"
|
|
"github.com/vishvananda/netlink"
|
|
)
|
|
|
|
// we need to manually remove old connections from a bpf map
|
|
// since when a bpf map is full it doesn't allow any more insertions
|
|
func monitorMaps() {
|
|
for {
|
|
select {
|
|
case <-ctxTasks.Done():
|
|
goto Exit
|
|
default:
|
|
time.Sleep(time.Second * 5)
|
|
for name := range ebpfMaps {
|
|
// using a pointer to the map doesn't delete the items.
|
|
// bpftool still counts them.
|
|
if items := getItems(name, name == "tcp6" || name == "udp6"); items > 500 {
|
|
deleted := deleteOldItems(name, name == "tcp6" || name == "udp6", items/2)
|
|
log.Debug("[ebpf] old items deleted: %d", deleted)
|
|
}
|
|
}
|
|
}
|
|
}
|
|
Exit:
|
|
}
|
|
|
|
func monitorCache() {
|
|
for {
|
|
select {
|
|
case <-ctxTasks.Done():
|
|
goto Exit
|
|
case <-ebpfCacheTicker.C:
|
|
ebpfCache.DeleteOldItems()
|
|
}
|
|
}
|
|
Exit:
|
|
}
|
|
|
|
// maintain a list of this machine's local addresses
|
|
func monitorLocalAddresses() {
|
|
newAddrChan := make(chan netlink.AddrUpdate)
|
|
done := make(chan struct{})
|
|
defer close(done)
|
|
|
|
addrs := daemonNetlink.GetLocalAddrs()
|
|
if addrs != nil {
|
|
lock.Lock()
|
|
localAddresses = addrs
|
|
lock.Unlock()
|
|
log.Debug("local Addrs: %v", localAddresses)
|
|
}
|
|
|
|
netlink.AddrSubscribe(newAddrChan, done)
|
|
|
|
for {
|
|
select {
|
|
case <-ctxTasks.Done():
|
|
done <- struct{}{}
|
|
goto Exit
|
|
case addr := <-newAddrChan:
|
|
if addr.NewAddr && !findAddressInLocalAddresses(addr.LinkAddress.IP) {
|
|
log.Debug("local addr added: %+v\n", addr)
|
|
lock.Lock()
|
|
|
|
localAddresses[addr.LinkAddress.IP.String()] = daemonNetlink.AddrUpdateToAddr(&addr)
|
|
|
|
lock.Unlock()
|
|
} else if !addr.NewAddr {
|
|
log.Debug("local addr removed: %+v\n", addr)
|
|
lock.Lock()
|
|
delete(localAddresses, addr.LinkAddress.IP.String())
|
|
lock.Unlock()
|
|
}
|
|
}
|
|
}
|
|
Exit:
|
|
log.Debug("monitorLocalAddresses exited")
|
|
}
|
|
|
|
// monitorAlreadyEstablished makes sure that when an already-established connection is closed
|
|
// it will be removed from alreadyEstablished. If we don't do this and keep the alreadyEstablished entry forever,
|
|
// then after the genuine process quits,a malicious process may reuse PID-srcPort-srcIP-dstPort-dstIP
|
|
func monitorAlreadyEstablished() {
|
|
tcperr := 0
|
|
errLimitExceeded := func() bool {
|
|
if tcperr > 100 {
|
|
log.Debug("monitorAlreadyEstablished() generated too much errors")
|
|
return true
|
|
}
|
|
tcperr++
|
|
|
|
return false
|
|
}
|
|
|
|
for {
|
|
select {
|
|
case <-ctxTasks.Done():
|
|
goto Exit
|
|
default:
|
|
time.Sleep(time.Second * 2)
|
|
socketListTCP, err := daemonNetlink.SocketsDump(uint8(syscall.AF_INET), uint8(syscall.IPPROTO_TCP))
|
|
if err != nil {
|
|
log.Debug("monitorAlreadyEstablished(), error dumping TCP sockets via netlink (%d): %s", tcperr, err)
|
|
if errLimitExceeded() {
|
|
goto Exit
|
|
}
|
|
|
|
continue
|
|
}
|
|
alreadyEstablished.Lock()
|
|
for aesock := range alreadyEstablished.TCP {
|
|
found := false
|
|
for _, sock := range socketListTCP {
|
|
if daemonNetlink.SocketsAreEqual(aesock, sock) {
|
|
found = true
|
|
break
|
|
}
|
|
}
|
|
if !found {
|
|
delete(alreadyEstablished.TCP, aesock)
|
|
}
|
|
}
|
|
alreadyEstablished.Unlock()
|
|
|
|
if core.IPv6Enabled {
|
|
socketListTCPv6, err := daemonNetlink.SocketsDump(uint8(syscall.AF_INET6), uint8(syscall.IPPROTO_TCP))
|
|
if err != nil {
|
|
if errLimitExceeded() {
|
|
goto Exit
|
|
}
|
|
log.Debug("monitorAlreadyEstablished(), error dumping TCPv6 sockets via netlink (%d): %s", tcperr, err)
|
|
|
|
continue
|
|
}
|
|
alreadyEstablished.Lock()
|
|
for aesock := range alreadyEstablished.TCPv6 {
|
|
found := false
|
|
for _, sock := range socketListTCPv6 {
|
|
if daemonNetlink.SocketsAreEqual(aesock, sock) {
|
|
found = true
|
|
break
|
|
}
|
|
}
|
|
if !found {
|
|
delete(alreadyEstablished.TCPv6, aesock)
|
|
}
|
|
}
|
|
alreadyEstablished.Unlock()
|
|
}
|
|
}
|
|
}
|
|
Exit:
|
|
log.Debug("monitorAlreadyEstablished exited")
|
|
}
|