dwww Home | Manual pages | Find package

passt(1)                    General Commands Manual                    passt(1)

NAME
       passt - Unprivileged user-mode network connectivity for virtual machines
       pasta  -  Unprivileged  user-mode network connectivity for network name-
       spaces

SYNOPSIS
       passt [OPTION]...
       pasta [OPTION]... [COMMAND [ARG]...]
       pasta [OPTION]... PID
       pasta [OPTION]... --netns [PATH|NAME]

DESCRIPTION
   passt
       passt (Plug A Simple Socket Transport) provides full, quasi-native  net-
       work connectivity to virtual machines in user-mode without requiring any
       capabilities or privileges.

       The  data plane implements a translation layer between a Layer-2 virtual
       network interface and native Layer-4 (TCP, UDP,  ping)  sockets  on  the
       host,  giving  the  illusion  that application processes residing on the
       guest are running on the local host, from a networking perspective.

       Built-in ARP, DHCP, NDP, and DHCPv6 implementations are designed to pro-
       vide the guest with a network configuration that tightly  resembles  the
       host  native  configuration.  With  the  default options, guest and host
       share IP addresses, routes, and port bindings.

       Port forwarding and translation allow networking services running in the
       guest to be reachable from both local and remote hosts.

       Unlike slirp4netns(1), passt doesn't implement a full TCP stack: the TCP
       translation layer has no stateful data buffering  and  operates  by  re-
       flecting  one  peer's  observed  parameters (congestion window size, ac-
       knowledged data, etc.)  to the corresponding peer.

       Currently, the only supported hypervisor is qemu(1), connecting to passt
       by means of a UNIX domain socket. This is supported starting  from  qemu
       7.2. For older qemu versions, see the qrap(1) wrapper.

   pasta
       pasta  (Pack A Subtle Tap Abstraction) provides equivalent functionality
       to network namespaces, as the one offered by passt for virtual machines.

       If PID or --netns are given, pasta associates to an  existing  user  and
       network namespace. Otherwise, pasta creates a new user and network name-
       space,  and spawns the given command or a default shell within this con-
       text. A tap device within the network namespace is  created  to  provide
       network connectivity.

       For  local TCP and UDP traffic only, pasta also implements a bypass path
       directly mapping Layer-4 sockets between init and target namespaces, for
       performance reasons.

OPTIONS
       Unless otherwise noted below, if conflicting  or  multiple  options  are
       given, the last one takes effect.

       -d, --debug
              Be verbose, don't log to the system logger.

       --trace
              Be extra verbose, show single packets. Implies --debug.

       -q, --quiet
              Don't print informational messages.

       -f, --foreground
              Don't  run  in  background.  This implies that the process is not
              moved to a detached PID namespace after starting, because the PID
              itself cannot change.  Default is to fork into background.

       -e, --stderr (DEPRECATED)
              This option has no effect, and is  maintained  for  compatibility
              purposes only.

              Note that this configuration option is deprecated and will be re-
              moved in a future version.

       -l, --log-file PATH
              Log to file PATH, and not to the system logger.

              Specifying  this  option multiple times does not lead to multiple
              log files: the last given option takes effect.

       --log-size SIZE
              Limit log file size to SIZE bytes. When the  log  file  is  full,
              make  room for new entries by removing old ones at the beginning.
              This limit is mandatory.  Default is 1048576 (1 MiB).

       --runas UID|UID:GID|LOGIN|LOGIN:GROUP
              Attempt to change to given UID and corresponding group if UID  is
              given,  or to given UID and given GID if both are given. Alterna-
              tively, login name, or login name and group name can  be  passed.
              This  requires  privileges  (either  initial  effective  UID 0 or
              CAP_SETUID capability) to work.  Default is to change to user no-
              body if started as root.

       -h, --help
              Display a help message and exit.

       --version
              Show version and exit.

       -p, --pcap file
              Capture tap-facing (that is, guest-side or  namespace-side)  net-
              work packets to file in pcap format.

              Specifying  this  option multiple times does not lead to multiple
              capture files: the last given option takes effect.

       -P, --pid file
              Write own PID to file once initialisation is done, before forking
              to background (if configured to do so).

       -m, --mtu mtu
              Assign mtu via DHCP (option 26) and NDP (option type 5).  A  zero
              value  disables  assignment.   By  default, the advertised MTU is
              65520 bytes, that is, the maximum 802.3 MTU minus the length of a
              802.3 header, rounded to 32 bits (IPv4 words).

       -a, --address addr
              Assign IPv4 addr via DHCP (yiaddr), or addr via DHCPv6 (option 5)
              and an addr-based prefix via  NDP  Router  Advertisement  (option
              type 3) for an IPv6 addr.  This option can be specified zero (for
              defaults)  to  two  times (once for IPv4, once for IPv6).  By de-
              fault, assigned IPv4 and IPv6 addresses are taken from  the  host
              interfaces  with  the first default route, if any, for the corre-
              sponding IP version. If no default routes are available and there
              is any interface with any route for a given IP version, the first
              of these interfaces will be chosen instead. If no such  interface
              exists,  the link-local address 169.254.2.1 is assigned for IPv4,
              and no additional address will be assigned for IPv6.

       -n, --netmask mask
              Assign IPv4 netmask mask, expressed as dot-decimal or  number  of
              bits, via DHCP (option 1).  By default, the netmask associated to
              the host address matching the assigned one is used. If there's no
              matching address on the host, the netmask is determined according
              to the CIDR block of the assigned address (RFC 4632).

       -M, --mac-addr addr
              Use source MAC address addr when communicating to the guest or to
              the  target  namespace.   Default is the locally administered MAC
              addresses 9a:55:9a:55:9a:55.

       -g, --gateway addr
              Assign IPv4 addr as default gateway via DHCP (option 3), or  IPv6
              addr  as source for NDP Router Advertisement and DHCPv6 messages.
              This option can be specified zero (for  defaults)  to  two  times
              (once  for IPv4, once for IPv6).  By default, IPv4 and IPv6 gate-
              ways are taken from the host interface  with  the  first  default
              route,  if  any, for the corresponding IP version. If the default
              route is a multipath one, the gateway is the first nexthop router
              returned by the kernel which has the highest weight in the set of
              paths. If no default routes are available and there is  just  one
              interface  with any route, that interface will be chosen instead.
              If no such interface exists, the link-local  address  169.254.2.2
              is  used for IPv4, and the link-local address fe80::1 is used for
              IPv6.

              Note: these addresses are also used as source address for packets
              directed to the guest or to the target namespace having  a  loop-
              back  or  local source address, to allow mapping of local traffic
              to guest and target namespace. See the NOTES below for  more  de-
              tails about this mechanism.

       -i, --interface name
              Use  host interface name to derive addresses and routes.  Default
              is to use the interfaces specified by --outbound-if4  and  --out-
              bound-if6, for IPv4 and IPv6 addresses and routes, respectively.

              If  no interfaces are given, the interface with the first default
              routes for each IP version is selected. If no default routes  are
              available  and  there  is just one interface with any route, that
              interface will be chosen instead. If no  such  interface  exists,
              host interfaces will be ignored for the purposes of assigning ad-
              dresses  and  routes,  and  link-local addresses will be used in-
              stead.

       -o, --outbound addr
              Use an IPv4 addr as source address for IPv4 outbound TCP  connec-
              tions,  UDP  flows, ICMP requests, or an IPv6 addr for IPv6 ones,
              by binding outbound sockets to it.  This option can be  specified
              zero  (for defaults) to two times (once for IPv4, once for IPv6).
              By default, the source address is selected by the routing tables.

       --outbound-if4 name
              Bind IPv4 outbound sockets to host interface  name,  and,  unless
              another  interface is specified via -i, --interface, use this in-
              terface to derive IPv4 addresses and routes.

              By default, the interface given by the default route is selected.
              If no default routes are available and there is just  one  inter-
              face with any route, that interface will be chosen instead. If no
              such  interface exists, outbound sockets will not be bound to any
              specific interface.

       --outbound-if6 name
              Bind IPv6 outbound sockets to host interface  name,  and,  unless
              another  interface is specified via -i, --interface, use this in-
              terface to derive IPv6 addresses and routes.

              By default, the interface given by the default route is selected.
              If no default routes are available and there is just  one  inter-
              face with any route, that interface will be chosen instead. If no
              such  interface exists, outbound sockets will not be bound to any
              specific interface.

       -D, --dns addr
              Instruct the guest (via DHCP, DHVPv6 or NDP) to use addr (IPv4 or
              IPv6) as a nameserver, as configured (see options  --no-dhcp-dns,
              --dhcp-dns)  instead  of reading addresses from /etc/resolv.conf.
              This option can be specified multiple times.  Specifying -D  none
              disables  usage  of  DNS  addresses altogether.  Unlike addresses
              from /etc/resolv.conf, addr is given to the guest without  remap-
              ping.  For example --dns 127.0.0.1 will instruct the guest to use
              itself as nameserver, not the host.

       --dns-forward addr
              Map  addr  (IPv4  or IPv6) as seen from guest or namespace to the
              nameserver (with  corresponding  IP  version)  specified  by  the
              --dns-host  option.  Maps  only UDP and TCP traffic to port 53 or
              port 853.  Replies are translated back with  a  reverse  mapping.
              This  option  can  be specified zero to two times (once for IPv4,
              once for IPv6).

       --dns-host addr
              Configure the host nameserver which guest or namespace queries to
              the --dns-forward address will be redirected to. This option  can
              be  specified  zero  to two times (once for IPv4, once for IPv6).
              By  default,  the  first  nameserver  from  the  host's  /etc/re-
              solv.conf.

       -S, --search list
              Use  space-separated list for DHCP, DHCPv6, and NDP purposes, in-
              stead of reading entries from /etc/resolv.conf. See options --no-
              dhcp-search and --dhcp-search. --search none disables the DNS do-
              main search list altogether (if  you  need  to  search  a  domain
              called "none" you can use --search none.).

       --no-dhcp-dns
              In  passt mode, do not assign IPv4 addresses via DHCP (option 23)
              or IPv6 addresses via NDP Router Advertisement (option  type  25)
              and  DHCPv6  (option  23)  as DNS resolvers.  By default, all the
              configured addresses are passed.

       --dhcp-dns
              In pasta mode, assign IPv4 addresses via DHCP (option 23) or IPv6
              addresses via NDP  Router  Advertisement  (option  type  25)  and
              DHCPv6  (option 23) as DNS resolvers.  By default, configured ad-
              dresses, if any, are not passed.

       --no-dhcp-search
              In passt mode, do not send the DNS domain search  list  addresses
              via  DHCP (option 119), via NDP Router Advertisement (option type
              31) and DHCPv6 (option 24).  By default, the  DNS  domain  search
              list resulting from configuration is passed.

       --dhcp-search
              In pasta mode, send the DNS domain search list addresses via DHCP
              (option  119),  via NDP Router Advertisement (option type 31) and
              DHCPv6 (option 24).  By default, the DNS domain search  list  re-
              sulting from configuration is not passed.

       --no-tcp
              Disable  the TCP protocol handler. No TCP connections will be ac-
              cepted host-side, and TCP packets coming  from  guest  or  target
              namespace will be silently dropped.

       --no-udp
              Disable  the UDP protocol handler. No UDP traffic coming from the
              host side will be forwarded, and UDP packets coming from guest or
              target namespace will be silently dropped.

       --no-icmp
              Disable the ICMP/ICMPv6 echo handler. ICMP and  ICMPv6  echo  re-
              quests  coming  from  guest  or target namespace will be silently
              dropped.

       --no-dhcp
              Disable the DHCP server. DHCP client requests coming  from  guest
              or target namespace will be silently dropped. Implied if there is
              no gateway on the selected IPv4 default route.

       --no-ndp
              Disable  NDP  responses. NDP messages coming from guest or target
              namespace will be ignored.

       --no-dhcpv6
              Disable the DHCPv6 server. DHCPv6  client  requests  coming  from
              guest or target namespace will be silently dropped.

       --no-ra
              Disable  Router  Advertisements. Router Solicitations coming from
              guest or target namespace will be ignored.

       --freebind
              Allow any binding address to be specified for -t and -u  options.
              Usually  binding addresses must be addresses currently configured
              on the host.  With --freebind, the IP_FREEBIND  or  IPV6_FREEBIND
              socket  option  is enabled allowing any address to be used.  This
              is typically used to bind addresses which might be configured  on
              the  host  in  future, at which point the forwarding will immedi-
              ately start operating.

       --map-host-loopback addr
              Translate addr to refer to the host. Packets from  the  guest  to
              addr  will  be  redirected to the host.  On the host such packets
              will appear to have both source and destination of  127.0.0.1  or
              ::1.

              If  addr  is 'none', no address is mapped (this implies --no-map-
              gw).  Only one IPv4 and one IPv6 address can  be  translated,  if
              the  option  is  specified multiple times, the last one takes ef-
              fect.

              Default is to translate the guest's default gateway address,  un-
              less --no-map-gw is given, in which case no address is mapped.

       --no-map-gw
              Don't  remap  TCP connections and untracked UDP traffic, with the
              gateway address as destination, to the host. Implied if there  is
              no  gateway  on the selected default route, or if there is no de-
              fault route, for any of the enabled address families.

       --map-guest-addr addr
              Translate addr in the guest to be equal to the  guest's  assigned
              address  on  the  host.   That is, packets from the guest to addr
              will be redirected to the address assigned to the guest with  -a,
              or  by  default the host's global address.  This allows the guest
              to access services available on the host's global  address,  even
              though its own address shadows that of the host.

              If  addr  is 'none', no address is mapped.  Only one IPv4 and one
              IPv6 address can be translated, and if the  option  is  specified
              multiple times, the last one for each address type takes effect.

              By default, mapping happens as described for the --map-host-loop-
              back option.

       -4, --ipv4-only
              Enable IPv4-only operation. IPv6 traffic will be ignored.  By de-
              fault,  IPv6  operation  is  enabled  as long as at least an IPv6
              route and an interface address are configured on a given host in-
              terface.

       -6, --ipv6-only
              Enable IPv6-only operation. IPv4 traffic will be ignored.  By de-
              fault, IPv4 operation is enabled as long  as  at  least  an  IPv4
              route and an interface address are configured on a given host in-
              terface.

       -H, --hostname name
              Hostname  to configure the client with.  Send name as DHCP option
              12 (hostname).

       --fqdn name
              FQDN to configure the client with.  Send  name  as  Client  FQDN:
              DHCP option 81 and DHCPv6 option 39.

   passt-only options
       -s, --socket-path, --socket path
              Path for UNIX domain socket used by qemu(1) or qrap(1) to connect
              to  passt.  Default is to probe a free socket, not accepting con-
              nections,     starting      from      /tmp/passt_1.socket      to
              /tmp/passt_64.socket.

       --vhost-user
              Enable  vhost-user.  The vhost-user command socket is provided by
              --socket.

       --print-capabilities
              Print back-end capabilities in JSON format, only  meaningful  for
              vhost-user mode.

       --repair-path path
              Path for UNIX domain socket used by the passt-repair(1) helper to
              connect  to  passt in order to set or clear the TCP_REPAIR option
              on sockets, during migration. --repair-path  none  disables  this
              interface (if you need to specify a socket path called "none" you
              can prefix the path by ./).

              Default,  for --vhost-user mode only, is to append .repair to the
              path chosen for the hypervisor UNIX domain socket. No  socket  is
              created if not in --vhost-user mode.

       -F, --fd FD
              Pass  a pre-opened, connected socket to passt. Usually the socket
              is opened in the parent process and passt inherits it when run as
              a child. This allows the parent process to open sockets using an-
              other address family or requiring special privileges.

              This option implies the behaviour described for  --one-off,  once
              this socket is closed.

       -1, --one-off
              Quit after handling a single client connection, that is, once the
              client closes the socket, or once we get a socket error.

       -t, --tcp-ports spec
              Configure TCP port forwarding to guest. spec can be one of:

              none   Don't forward any ports

              all    Forward  all unbound, non-ephemeral ports, as permitted by
                     current capabilities.  For low (< 1024) ports, see  NOTES.
                     No  failures are reported for unavailable ports, unless no
                     ports could be forwarded at all.

              ports  A comma-separated list of ports, optionally ranged with -,
                     and, optionally, with target ports after :, if  they  dif-
                     fer. Specific addresses can be bound as well, separated by
                     /,  and  also, since Linux 5.7, limited to specific inter-
                     faces, prefixed by %. Within given ranges, selected  ports
                     and  ranges can be excluded by an additional specification
                     prefixed by ~.

                     Specifying excluded ranges only  implies  that  all  other
                     ports  are  forwarded.  In  this case, no failures are re-
                     ported for unavailable ports, unless  no  ports  could  be
                     forwarded at all.

                     Examples:

                     -t 22  Forward local port 22 to port 22 on the guest

                     -t 22:23
                            Forward local port 22 to port 23 on the guest

                     -t 22,25
                            Forward local ports 22 and 25 to ports 22 and 25 on
                            the guest

                     -t 22-80
                            Forward  local  ports  between  22 and 80 to corre-
                            sponding ports on the guest

                     -t 22-80:32-90
                            Forward local ports between 22 and 80 to ports  be-
                            tween 32 and 90 on the guest

                     -t 192.0.2.1/22
                            Forward  local port 22, bound to 192.0.2.1, to port
                            22 on the guest

                     -t 192.0.2.1%eth0/22
                            Forward local port 22, bound to 192.0.2.1  and  in-
                            terface eth0, to port 22

                     -t %eth0/22
                            Forward  local port 22, bound to any address on in-
                            terface eth0, to port 22

                     -t 2000-5000,~3000-3010
                            Forward local ports between 2000 and  5000,  except
                            for those between 3000 and 3010

                     -t 192.0.2.1/20-30,~25
                            For  the local address 192.0.2.1, forward ports be-
                            tween 20 and 24 and between 26 and 30

                     -t ~20000-20010
                            Forward all ports to  the  guest,  except  for  the
                            range from 20000 to 20010

              Default is none.

       -u, --udp-ports spec
              Configure  UDP port forwarding to guest. spec is as described for
              TCP above.

              Note: unless overridden, UDP ports with numbers corresponding  to
              forwarded  TCP  port numbers are forwarded too, without, however,
              any port translation. IPv6 bound ports  are  also  forwarded  for
              IPv4.

              Default is none.

   pasta-only options
       -I, --ns-ifname name
              Name  of tap interface to be created in target namespace.  By de-
              fault, the same interface name as the external,  routable  inter-
              face is used.  If no such interface exists, the name tap0 will be
              used instead.

       -t, --tcp-ports spec
              Configure TCP port forwarding to namespace. spec can be one of:

              none   Don't forward any ports

              auto   Dynamically forward ports bound in the namespace. The list
                     of  ports is periodically derived (every second) from lis-
                     tening   sockets    reported    by    /proc/net/tcp    and
                     /proc/net/tcp6, see proc(5).

              ports  A comma-separated list of ports, optionally ranged with -,
                     and,  optionally,  with target ports after :, if they dif-
                     fer. Specific addresses can be bound as well, separated by
                     /, and also, since Linux 5.7, limited to  specific  inter-
                     faces,  prefixed by %. Within given ranges, selected ports
                     and ranges can be excluded by an additional  specification
                     prefixed by ~.

                     Specifying  excluded  ranges  only  implies that all other
                     ports are forwarded. In this case,  no  failures  are  re-
                     ported  for  unavailable  ports,  unless no ports could be
                     forwarded at all.

                     Examples:

                     -t 22  Forward local port 22 to 22 in the target namespace

                     -t 22:23
                            Forward local port 22 to  port  23  in  the  target
                            namespace

                     -t 22,25
                            Forward local ports 22 and 25 to ports 22 and 25 in
                            the target namespace

                     -t 22-80
                            Forward  local  ports  between  22 and 80 to corre-
                            sponding ports in the target namespace

                     -t 22-80:32-90
                            Forward local ports between 22 and 80 to ports  be-
                            tween 32 and 90 in the target namespace

                     -t 192.0.2.1/22
                            Forward  local port 22, bound to 192.0.2.1, to port
                            22 in the target namespace

                     -t 192.0.2.1%eth0/22
                            Forward local port 22, bound to 192.0.2.1  and  in-
                            terface eth0, to port 22

                     -t %eth0/22
                            Forward  local port 22, bound to any address on in-
                            terface eth0, to port 22

                     -t 2000-5000,~3000-3010
                            Forward local ports between 2000 and  5000,  except
                            for those between 3000 and 3010

                     -t 192.0.2.1/20-30,~25
                            For  the local address 192.0.2.1, forward ports be-
                            tween 20 and 24 and between 26 and 30

                     -t ~20000-20010
                            Forward all ports  to  the  namespace,  except  for
                            those between 20000 and 20010

              IPv6 bound ports are also forwarded for IPv4.

              Default is auto.

       -u, --udp-ports spec
              Configure  UDP port forwarding to namespace. spec is as described
              for TCP above, and the list of ports is  derived  from  listening
              sockets   reported   by  /proc/net/udp  and  /proc/net/udp6,  see
              proc(5).

              Note: unless overridden, UDP ports with numbers corresponding  to
              forwarded  TCP  port numbers are forwarded too, without, however,
              any port translation.

              IPv6 bound ports are also forwarded for IPv4.

              Default is auto.

       -T, --tcp-ns spec
              Configure TCP port forwarding from target namespace to init name-
              space.  spec is as described above for TCP.

              Default is auto.

       -U, --udp-ns spec
              Configure UDP port forwarding from target namespace to init name-
              space.  spec is as described above for UDP.

              Default is auto.

       --host-lo-to-ns-lo
              If specified, connections forwarded  with  -t  and  -u  from  the
              host's  loopback  address  will appear on the loopback address in
              the guest as well.  Without this option  such  forwarded  packets
              will appear to come from the guest's public address.

       --userns spec
              Target  user namespace to join, as a path. If PID is given, with-
              out this option, the user namespace will be the one of the corre-
              sponding process.

       --netns spec
              Target network namespace to join, as a path or a name.  A name is
              treated  as  with  ip-netns(8)  as  equivalent  to  a   path   in
              /run/netns.

              This option can't be specified with a PID.

       --netns-only
              Join  only  a target network namespace, not a user namespace, and
              don't create one for sandboxing purposes either. This is  implied
              if PATH or NAME are given without --userns.

       --no-netns-quit
              Do not exit once the target namespace reference is removed.

              Without  this  option, pasta will terminate if the target network
              namespace is bound to the  filesystem,  and  the  given  path  is
              deleted,  or  if the target network namespace is represented by a
              procfs entry, and that entry is deleted,  representing  the  fact
              that a process with the given PID terminated.

       --config-net
              Configure  networking  in  the  namespace:  set  up addresses and
              routes as configured or sourced from the host, and bring  up  the
              tap interface.

       --no-copy-routes (DEPRECATED)
              With  --config-net,  do not copy all the routes associated to the
              interface we derive addresses and routes from: set  up  only  the
              default  gateway. Implied by -g, --gateway, for the corresponding
              IP version only.

              Default is to copy all the routing entries from the interface  in
              the outer namespace to the target namespace, translating the out-
              put  interface  attribute  to the outbound interface in the name-
              space.

              Note that this configuration option is deprecated and will be re-
              moved in a future version. It is not expected to be of  any  use,
              and  it  simply  reflects a legacy behaviour. If you have any use
              for this, refer to REPORTING BUGS below.

       --no-copy-addrs (DEPRECATED)
              With --config-net, do not copy all the  addresses  associated  to
              the  interface we derive addresses and routes from: set up a sin-
              gle one. Implied by -a, --address, for the corresponding IP  ver-
              sion only.

              Default is to copy all the addresses, except for link-local ones,
              from  the  interface from the outer namespace to the target name-
              space.

              Note that this configuration option is deprecated and will be re-
              moved in a future version. It is not expected to be of  any  use,
              and  it  simply  reflects a legacy behaviour. If you have any use
              for this, refer to REPORTING BUGS below.

       --ns-mac-addr addr
              Configure MAC address addr on the tap interface in the namespace.

              Default is to let the tap driver build  a  pseudorandom  hardware
              address.

       --no-splice
              Disable  the bypass path for inbound, local traffic. See the sec-
              tion Handling of local traffic in pasta in the NOTES for more de-
              tails.

EXAMPLES
   pasta
       Create and use a new, connected, user and network namespace
              $ iperf3 -s -D
              $ ./pasta
              Outbound interface: eth0, namespace interface: eth0
              ARP:
                  address: 28:16:ad:39:a9:ea
              DHCP:
                  assign: 192.168.1.118
                  mask: 255.255.255.0
                  router: 192.168.1.1
              NDP/DHCPv6:
                  assign: 2a02:6d40:3ca5:2001:b81d:fa4a:8cdd:cf17
                  router: fe80::62e3:27ff:fe33:2b01
              #
              # dhclient -4 --no-pid
              # dhclient -6 --no-pid
              # ip address show
              1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
                  link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
                  inet 127.0.0.1/8 scope host lo
                     valid_lft forever preferred_lft forever
                  inet6 ::1/128 scope host
                     valid_lft forever preferred_lft forever
              2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 65520 qdisc pfifo_fast state UNKNOWN group default qlen 1000
                  link/ether 5e:90:02:eb:b0:2a brd ff:ff:ff:ff:ff:ff
                  inet 192.168.1.118/24 brd 192.168.1.255 scope global eth0
                     valid_lft forever preferred_lft forever
                  inet6 2a02:6d40:3ca5:2001:b81d:fa4a:8cdd:cf17/128 scope global
                     valid_lft forever preferred_lft forever
                  inet6 2a02:6d40:3ca5:2001:5c90:2ff:feeb:b02a/64 scope global dynamic mngtmpaddr
                     valid_lft 3591sec preferred_lft 3591sec
                  inet6 fe80::5c90:2ff:feeb:b02a/64 scope link
                     valid_lft forever preferred_lft forever
              # ip route show
              default via 192.168.1.1 dev eth0
              192.168.1.0/24 dev eth0 proto kernel scope link src 192.168.1.118
              # ip -6 route show
              2a02:6d40:3ca5:2001:b81d:fa4a:8cdd:cf17 dev eth0 proto kernel metric 256 pref medium
              2a02:6d40:3ca5:2001::/64 dev eth0 proto kernel metric 256 expires 3584sec pref medium
              fe80::/64 dev eth0 proto kernel metric 256 pref medium
              default via fe80::62e3:27ff:fe33:2b01 dev eth0 proto ra metric 1024 expires 3584sec pref medium
              # iperf3 -c 127.0.0.1 -t1
              Connecting to host 127.0.0.1, port 5201
              [  5] local 127.0.0.1 port 51938 connected to 127.0.0.1 port 5201
              [ ID] Interval           Transfer     Bitrate         Retr  Cwnd
              [  5]   0.00-1.00   sec  4.46 GBytes  38.3 Gbits/sec    0   3.93 MBytes
              - - - - - - - - - - - - - - - - - - - - - - - - -
              [ ID] Interval           Transfer     Bitrate         Retr
              [  5]   0.00-1.00   sec  4.46 GBytes  38.3 Gbits/sec    0             sender
              [  5]   0.00-1.41   sec  4.45 GBytes  27.1 Gbits/sec                  receiver

              iperf Done.
              # iperf3 -c ::1 -t1
              Connecting to host ::1, port 5201
              [  5] local ::1 port 50108 connected to ::1 port 5201
              [ ID] Interval           Transfer     Bitrate         Retr  Cwnd
              [  5]   0.00-1.00   sec  4.35 GBytes  37.4 Gbits/sec    0   4.99 MBytes
              - - - - - - - - - - - - - - - - - - - - - - - - -
              [ ID] Interval           Transfer     Bitrate         Retr
              [  5]   0.00-1.00   sec  4.35 GBytes  37.4 Gbits/sec    0             sender
              [  5]   0.00-1.41   sec  4.35 GBytes  26.4 Gbits/sec                  receiver

              iperf Done.
              # ping -c1 -4 spaghetti.pizza
              PING spaghetti.pizza (172.67.192.217) 56(84) bytes of data.
              64 bytes from 172.67.192.217: icmp_seq=1 ttl=255 time=37.3 ms

              --- spaghetti.pizza ping statistics ---
              1 packets transmitted, 1 received, 0% packet loss, time 0ms
              # ping -c1 -6 spaghetti.pizza
              PING spaghetti.pizza(2606:4700:3034::6815:147a (2606:4700:3034::6815:147a)) 56 data bytes
              64 bytes from 2606:4700:3034::6815:147a: icmp_seq=1 ttl=255 time=35.6 ms

              --- spaghetti.pizza ping statistics ---
              1 packets transmitted, 1 received, 0% packet loss, time 0ms
              rtt min/avg/max/mdev = 35.605/35.605/35.605/0.000 ms
              # logout
              $

       Connect an existing user and network namespace
              $ unshare -rUn
              # echo $$
              2446678

                   [From another terminal]
              $ ./pasta 2446678
              Outbound interface: eth0, namespace interface: eth0
              ARP:
                  address: 28:16:ad:39:a9:ea
              DHCP:
                  assign: 192.168.1.118
                  mask: 255.255.255.0
                  router: 192.168.1.1
              NDP/DHCPv6:
                  assign: 2a02:6d40:3ca5:2001:b81d:fa4a:8cdd:cf17
                  router: fe80::62e3:27ff:fe33:2b01

                   [Back to the original terminal]
              # dhclient -4 --no-pid
              # dhclient -6 --no-pid
              # ip address show
              1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
                  link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
                  inet 127.0.0.1/8 scope host lo
                     valid_lft forever preferred_lft forever
                  inet6 ::1/128 scope host
                     valid_lft forever preferred_lft forever
              2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 65520 qdisc pfifo_fast state UNKNOWN group default qlen 1000
                  link/ether fa:c1:2a:27:92:a9 brd ff:ff:ff:ff:ff:ff
                  inet 192.168.1.118/24 brd 192.168.1.255 scope global eth0
                     valid_lft forever preferred_lft forever
                  inet6 2a02:6d40:3ca5:2001:b81d:fa4a:8cdd:cf17/128 scope global
                     valid_lft forever preferred_lft forever
                  inet6 2a02:6d40:3ca5:2001:f8c1:2aff:fe27:92a9/64 scope global dynamic mngtmpaddr
                     valid_lft 3594sec preferred_lft 3594sec
                  inet6 fe80::f8c1:2aff:fe27:92a9/64 scope link
                     valid_lft forever preferred_lft forever

   passt
       Start and connect a guest with basic port forwarding
              $ ./passt -f -t 2222:22
              Outbound interface: eth0
              ARP:
                  address: 28:16:ad:39:a9:ea
              DHCP:
                  assign: 192.168.1.118
                  mask: 255.255.255.0
                  router: 192.168.1.1
                      search:
                          redhat.com
              NDP/DHCPv6:
                  assign: 2a02:6d40:3ca5:2001:b81d:fa4a:8cdd:cf17
                  router: fe80::62e3:27ff:fe33:2b01
                      search:
                          redhat.com
              UNIX domain socket bound at /tmp/passt_1.socket

              You can now start qrap:
                  ./qrap 5 qemu-system-x86_64 ... -net socket,fd=5 -net nic,model=virtio
              or directly qemu, patched with:
                  qemu/0001-net-Allow-also-UNIX-domain-sockets-to-be-used-as-net.patch
              as follows:
                  qemu-system-x86_64 ... -net socket,connect=/tmp/passt_1.socket -net nic,model=virtio

                   [From another terminal]
              $ ./qrap 5 qemu-system-x86_64 test.qcow2 -m 1024 -display none -nodefaults -nographic -net socket,fd=5 -net nic,model=virtio
              Connected to /tmp/passt_1.socket

                   [Back to the original terminal]
              passt: DHCP: ack to request
              passt:     from 52:54:00:12:34:56
              passt: NDP: received NS, sending NA
              passt: NDP: received RS, sending RA
              passt: DHCPv6: received SOLICIT, sending ADVERTISE
              passt: NDP: received NS, sending NA
              passt: DHCPv6: received REQUEST/RENEW/CONFIRM, sending REPLY
              passt: NDP: received NS, sending NA

                   [From yet another terminal]
              $ ssh -p 2222 root@localhost
              root@localhost's password:
                   [...]
              # ip address show
              1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
                  link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
                  inet 127.0.0.1/8 scope host lo
                     valid_lft forever preferred_lft forever
                  inet6 ::1/128 scope host
                     valid_lft forever preferred_lft forever
              2: ens2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 65520 qdisc pfifo_fast state UP group default qlen 1000
                  link/ether 52:54:00:12:34:56 brd ff:ff:ff:ff:ff:ff
                  inet 192.168.1.118/24 brd 192.168.1.255 scope global noprefixroute ens2
                     valid_lft forever preferred_lft forever
                  inet6 2a02:6d40:3ca5:2001:b81d:fa4a:8cdd:cf17/128 scope global noprefixroute
                     valid_lft forever preferred_lft forever
                  inet6 2a02:6d40:3ca5:2001:b019:9ae2:a2fe:e6b4/64 scope global dynamic noprefixroute
                     valid_lft 3588sec preferred_lft 3588sec
                  inet6 fe80::1f98:d09f:9309:9e77/64 scope link noprefixroute
                     valid_lft forever preferred_lft forever

NOTES
   Handling of traffic with loopback destination and source addresses
       Both passt  and  pasta  can  bind  on  ports  with  a  loopback  address
       (127.0.0.0/8  or ::1), depending on the configuration. Loopback destina-
       tion or source addresses need to be changed before packets are delivered
       to the guest or target namespace:  most  operating  systems  would  drop
       packets received with loopback addresses on non-loopback interfaces, and
       it  would  also be impossible for guest or target namespace to route an-
       swers back.

       For convenience, the source address on these packets  is  translated  to
       the  address  specified by the --map-host-loopback option (with some ex-
       ceptions in pasta mode, see next section below).  If not specified  this
       defaults,  somewhat  arbitrarily, to the address of default IPv4 or IPv6
       gateway (if any) -- this is known to be an existing,  valid  address  on
       the  same subnet.  If --no-map-gw or --map-host-loopback none are speci-
       fied this translation is disabled and packets  with  loopback  addresses
       are simply dropped.

       Loopback  destination  addresses are translated to the observed external
       address of the guest or target namespace. For IPv6, the  observed  link-
       local  address  is  used if the translated source address is link-local,
       otherwise the observed global address is used. For both IPv4  and  IPv6,
       if  no  addresses  have  been seen yet, the configured addresses will be
       used instead.

       For example, if passt or pasta receive a connection from 127.0.0.1, with
       destination 127.0.0.10, and the default IPv4 gateway is 192.0.2.1, while
       the last observed source address from guest or namespace  is  192.0.2.2,
       this will be translated to a connection from 192.0.2.1 to 192.0.2.2.

       Similarly, for traffic coming from guest or namespace, packets with des-
       tination  address  corresponding to the --map-host-loopback address will
       have their destination address translated to a loopback address.

       As an exception, traffic identified as DNS, originally directed  to  the
       --map-host-loopback  address, if this address matches a resolver address
       on the host, is not translated to loopback, but rather  handled  in  the
       same way as if specified as --dns-forward address, if no such option was
       given.   In the common case where the host gateway also acts a resolver,
       this avoids that the host mapping shadows the gateway/resolver itself.

   Handling of local traffic in pasta
       Depending on the configuration, pasta can bind to  local  ports  in  the
       init  namespace,  in  the target namespace, or both, and forward connec-
       tions and packets to corresponding ports in the other namespace.

       To avoid unnecessary overhead, these connections  and  packets  are  not
       forwarded  through  the tap device connecting the namespaces: pasta cre-
       ates a socket in the destination namespace, with matching Layer-4 proto-
       col, and uses it to forward local data. For TCP, data is  forwarded  be-
       tween the originating socket and the new socket using the splice(2) sys-
       tem  call,  and  for  UDP,  a pair of recvmmsg(2) and sendmmsg(2) system
       calls deals with packet transfers.

       Because it's not possible to bind sockets to foreign addresses, this by-
       pass only applies to local connections and traffic.  It also means  that
       the  address  translation differs slightly from passt mode.  Connections
       from loopback to loopback on the host will appear to come from the  tar-
       get namespace's public address within the guest, unless --host-lo-to-ns-
       lo is specified, in which case they will appear to come from loopback in
       the namespace as well.  The latter behaviour used to be the default, but
       is  usually  undesirable,  since it can unintentionally expose namespace
       local services to the host.

   Binding to low numbered ports (well-known or system ports, up to 1023)
       If the port forwarding configuration requires binding to ports with num-
       bers lower than 1024, passt and pasta will try to bind to them, but will
       fail, unless, either:

       • the sys.net.ipv4.ip_unprivileged_port_start sysctl is set to the  num-
         ber of the lowest port passt and pasta need. For example, as root:

              sysctl -w net.ipv4.ip_unprivileged_port_start=443

         Note:  this is the recommended way of enabling passt and pasta to bind
         to ports with numbers below 1024.

       • or the CAP_NET_BIND_SERVICE Linux  capability  is  granted,  see  ser-
         vices(5) and capabilities(7).

         This  is, in general, not the recommended way, because passt and pasta
         might be used as vector to effectively use this  capability  from  an-
         other process.

         However,  if  your  environment  is  sufficiently controlled by an LSM
         (Linux Security Module) such as AppArmor, SELinux,  Smack  or  TOMOYO,
         and  no  other processes can interact in such a way in virtue of this,
         granting this capability to passt and pasta only can effectively  pre-
         vent other processes from utilising it.

         Note that this will not work for automatic detection and forwarding of
         ports  with  pasta,  because  pasta will relinquish this capability at
         runtime.

         To grant this capability, you can issue, as root:

              for p in $(which passt passt.avx2); do
                   setcap 'cap_net_bind_service=+ep' "${p}"
              done

   ICMP/ICMPv6 Echo sockets
       ICMP and ICMPv6 Echo requests coming from guest or target namespace  are
       handled  using  so-called "ping" sockets, introduced in Linux 2.6.30. To
       preserve the original identifier (see RFC 792, page 14,  for  ICMP,  and
       RFC  4443,  section  4.1, for ICMPv6), passt and pasta try to bind these
       sockets using the observed source identifier as "port"  --  that  corre-
       sponds to Echo identifiers for "ping" sockets.

       As  bind(2)  failures  were  seen  with particularly restrictive SELinux
       policies, a fall-back mechanism maps different identifiers to  different
       sockets,  and identifiers in replies will be mapped back to the original
       identifier of the request. However, if bind(2) fails and  the  fall-back
       mechanism  is  used, echo requests will be forwarded with different, al-
       beit unique, identifiers.

       For ICMP and ICMPv6 Echo requests to work, the ping_group_range  parame-
       ter needs to include the PID of passt or pasta, see icmp(7).

   pasta and loopback interface
       As  pasta  connects  to  an existing namespace, or once it creates a new
       namespace, it will also ensure  that  the  loopback  interface,  lo,  is
       brought  up.  This is needed to bind ports using the loopback address in
       the namespace.

   TCP sending window and TCP_INFO before Linux 5.3
       To synchronise the TCP sending window from host Layer-4 sockets  to  the
       TCP  parameters  announced  in TCP segments sent over the Layer-2 inter-
       face, passt and pasta routinely query the size  of  the  sending  window
       seen by the kernel on the corresponding socket using the TCP_INFO socket
       option,  see  tcp(7).  Before Linux 5.3, i.e. before Linux kernel commit
       8f7baad7f035 ("tcp:  Add  snd_wnd  to  TCP_INFO"),  the  sending  window
       (snd_wnd field) is not available.

       If  the sending window cannot be queried, it will always be announced as
       the current sending buffer size to guest or target namespace. This might
       affect throughput of TCP connections.

   Local mode for disconnected setups
       If passt and pasta fail to find a host interface with a  configured  ad-
       dress,  other than loopback addresses, they will, obviously, not attempt
       to source addresses or routes from the host.

       In this case, unless configured otherwise, they  will  assign  the  IPv4
       link-local  address 169.254.2.1 to the guest or target namespace, and no
       IPv6 address. The notion of the guest or target namespace  IPv6  address
       is derived from the first link-local address observed.

       Default  gateways will be assigned as the link-local address 169.254.2.2
       for IPv4, and as the link-local address fe80::1 for IPv6.

LIMITATIONS
       Currently, IGMP/MLD proxying (RFC 4605) and support for SCTP (RFC  4960)
       are not implemented.

       TCP  Selective  Acknowledgment (RFC 2018), as well as Protection Against
       Wrapped Sequences (PAWS) and Round-Trip Time  Measurement  (RTTM),  both
       described by RFC 7232, are currently not implemented.

AUTHORS
       Stefano  Brivio  <sbrivio@redhat.com>,  David Gibson <david@gibson.drop-
       bear.id.au>.

REPORTING BUGS
       Please report issues on the bug tracker at https://passt.top/passt/bugs,
       or  send  a  message  to  the  passt-user@passt.top  mailing  list,  see
       https://passt.top/passt/lists.

COPYRIGHT
       Copyright (c) 2020-2022 Red Hat GmbH.

       passt and pasta are free software: you can redistribute them and/or mod-
       ify  them under the terms of the GNU General Public License as published
       by the Free Software Foundation, either version 2 of the License, or (at
       your option) any later version.

SEE ALSO
       namespaces(7), qemu(1), qrap(1), slirp4netns(1).

       High-level documentation is available at https://passt.top/passt/about/.

                                                                       passt(1)

Generated by dwww version 1.16 on Tue Dec 16 05:55:43 CET 2025.